PHD

University of Southampton Research Repository
ePrints Soton
Copyright and Moral Rights for this thesis are retained by the author and/or other
copyright owners. A copy can be downloaded for personal non-commercial
research or study, without prior permission or charge. This thesis cannot be
reproduced or quoted extensively from without first obtaining permission in writing
from the copyright holder/s. The content must not be changed in any way or sold
commercially in any format or medium without the formal permission of the
copyright holders.
When referring to this work, full bibliographic details including the author, title,
awarding institution and date of the thesis must be given e.g.
AUTHOR (year of submission) "Full thesis title", University of Southampton, name
of the University School or Department, PhD Thesis, pagination
http://eprints.soton.ac.uk
Damage Stability of Ships as a Safety Criterion for

Optimisation Tools
Thesis for the degree of Doctor of Philosophy
Deniz Saydan
Faculty of Engineering, Science & Mathematics
School of Engineering Sciences
UNIVERSITY OF SOUTHAMPTON
FACULTY OF ENGINEERING, SCIENCE & MATHEMATICS
School of Engineering Sciences
Damage Stability of Ships as a Safety Criterion for Optimisation Tools
by
Deniz Saydan
Thesis for the degree of Doctor of Philosophy
July 2006
dedicated to
my beloved mother, father and sister
UNIVERSITY OF SOUTHAMPTON
ABSTRACT
FACULTY OF ENGINEERING, SCIENCE & MATHEMATICS
SCHOOL OF ENGINEERING SCIENCES
Doctor of Philosophy
DAMAGE STABILITY OF SHIPS AS A SAFETY CRITERION FOR
OPTIMISATION TOOLS
by Deniz Saydan
A literature overview of past optimisation studies revealed that whilst satisfaction of

intact stability requirements has been built into existing alternative hull form
optimisation packages, seeking improved hydrodynamic hull forms in terms of
seakeeping, calm water resistance and added resistance, damage stability is not an
automated feature. Within the context of the hydrodynamic hull form optimisation
techniques their application to novel hull forms would only permit use of deterministic
damage stability analysis and as this is not straight-forward damage is applied after the
hull is optimised. The damage must be relevant to ship type and applied in appropriate
locations with sensible extents of damage. To fulfil this need both the Marine Accident
Investigation Branch (MAIB) damage data base and a damage data base generated by
Lutzen (2002) are interrogated and findings are reported.
The hydrodynamic analysis of the optimised hull and basis hull for the intact and
damage cases is thereafter carried out using a three-dimensional singularity distribution
method. The relative vertical motion responses of both intact and damaged hull forms
are determined with greater structural cross-coupling than is usually applied in the
solution of the equations of ship motions. This has necessitated the development of a
novel approach to implement the calculation of the pure and product moment of inertias
for the intact and damaged hull forms to facilitate meaningful comparison of intact and
damaged ship motions. The processes are equally applicable to any kind of ship.
CONTENTS
ACKNOWLEDGEMENTS ......................................................................................XIII
NOMENCLATURE .................................................................................................. XIV
1.
2.
INTRODUCTION .................................................................................................. 1
1.1
The Ship Design Concept ................................................................................. 1
1.2
Optimisation Tools ........................................................................................... 4
1.3
A Possible Shortcoming of Existing Optimisation Tools................................. 4
1.4
Aims and Outline of the Research.................................................................... 5
THE OPTIMISATION PROCESS....................................................................... 8

2.1
Background....................................................................................................... 9
2.1.1
A Specific Motion and Resistance Driven Optimisation Research
Programme ............................................................................................................... 9
2.1.2
3.
2.2
Discussion on Past Optimisation Studies ....................................................... 20
2.3
Statement of the Optimisation Process........................................................... 21
2.4
Summary......................................................................................................... 22
ANALYSIS OF DAMAGE DATA...................................................................... 23

3.1
Damage Statistics ........................................................................................... 23
3.1.1
4.
Ship Design Optimisation with Different Design Drivers ..................... 14
Summary of Damage Statistics .............................................................. 59
THE DEVELOPMENT OF SHIP REGULATIONS AND STABILITY
ANALYSIS .................................................................................................................... 61
4.1
Development of Safety ................................................................................... 61
4.2
The Deterministic and Probabilistic Approaches to Ship Stability Analysis . 66
4.2.1
The Deterministic Approach to Ship Stability Analysis ........................ 67
Lost Buoyancy Method ....................................................................................... 69

I
Added Weight Method ........................................................................................ 70

4.2.2
The Probabilistic Approach to Ship Stability Analysis.......................... 70
4.2.3
Discussion on Current Stability Methods............................................... 75
4.3
Damage Stability Suite ................................................................................... 76
4.3.1
Independent Damage Stability Analysis................................................. 77
4.3.2
Presentation of the Damaged Hull Forms .............................................. 79
4.4
5.
Summary......................................................................................................... 80
SEAKEEPING ...................................................................................................... 82
5.1
General Review .............................................................................................. 82
5.2
Frequency and Time Domain Approach ........................................................ 88
5.3
Justification of Selected Approach ................................................................. 89
5.4
Motion Analysis ............................................................................................. 90
5.4.1
Forces and Moments Acting on a Floating Ship in Waves .................... 90
Forces and Moments Acting on an Intact Ship................................................... 90

Forces and Moments Acting on a Damaged Ship .............................................. 91
5.4.2
The Equations of Motion........................................................................ 91
Equations of Motion for Intact Hull Forms........................................................ 92

Equations of Motion for Damaged Hull Forms ................................................. 97
5.4.3
A Novel Method for Determining the Products of Inertia for Intact and
Damaged Ship ...................................................................................................... 101

The Intact Ship.................................................................................................. 101
The Damaged Ship ........................................................................................... 109
6.
5.4.4
Determination of Motion Characteristics ............................................. 112
5.4.5
Discussion............................................................................................. 113
SELECTION AND FUNDAMENTAL ANALYSIS OF BASIS SHIP .......... 115

6.1
Choice of Hull Form..................................................................................... 115
6.2
Public Domain Particulars for Derbyshire.................................................... 116
6.3
Optimisation Objective Function and Design Parameters............................ 120
6.3.1
Choice of the Objective Function......................................................... 121

II
7.
6.4
IMO Intact Stability Requirements and Related Analysis ........................... 125
6.5
Floodable Length Curve and Bulkhead Locations ....................................... 129
6.6
Damage Properties of the Derbyshire........................................................... 132
6.7
The Damage Analysis................................................................................... 134
6.8
Orientation of Damaged Hull Forms ............................................................ 134
6.9
Summary....................................................................................................... 140
HYDRODYNAMIC CHARACTERISTICS OF INTACT AND DAMAGED
SHIPS........................................................................................................................... 141
7.1
Calculation of Hydrodynamic Coefficients and Excitation Forces .............. 142
7.2
Validation of Hydrodynamic Analysis of Intact and Damaged Forms of
Derbyshire ................................................................................................................ 145
8.
7.3
Discussion of Hydrodynamic Data for Intact and Damaged Ships .............. 155
7.4
Summary....................................................................................................... 174
RELATIVE VERTICAL MOTION ANALYSIS OF INTACT AND
DAMAGED SHIPS .................................................................................................... 175
9.
8.1
Motion Dependent Qualities of Engineering Interest................................... 175
8.2
Selection of Points Investigated for Relative Vertical Displacement........... 176
8.3
Comparison of Relative Vertical Motion for the Selected Points ................ 180
8.4
Summary....................................................................................................... 193
CONCLUSIONS AND FUTURE WORK........................................................ 194

9.1
Key Observation of Completed Research .................................................... 195
9.2
Novel Aspects of Completed Research ........................................................ 199
9.3
Recommendations for Future Work ............................................................. 201
BIBLIOGRAPHY....................................................................................................... 203
APPENDIX A HOOKE JEEVES ALGORITHM............................................. 222
APPENDIX B - BOUNDARY VALUE PROBLEM ............................................... 225
III
APPENDIX C - NUMERICAL APPLICATION OF THE NEW METHOD TO

DETERMINE THE MASS MOMENT OF INERTIA FOR PARENT
DERBYSHIRE AND FOR A TANKER HULL FORM ......................................... 229
APPENDIX D - GENERAL ARRANGEMENT DRAWING OF THE
DERBYSHIRE............................................................................................................ 233
APPENDIX E GENERAL APPRAISAL OF DAMAGE SCENARIOS FOR
PARENT DERBYSHIRE .......................................................................................... 235
APPENDIX F - ITERATIVE METHOD FOR DETERMINATION OF
EQUILIBRIUM TRIM AND HEEL ANGLES....................................................... 237
APPENDIX G SENSITIVITY ANALYSIS OF THE PURE AND PRODUCT
MOMENT OF INERTIAS TO THE EQUILIBRIUM ANGLES ......................... 239
APPENDIX H HYDRODYNAMIC COEFFICIENT PLOTS FOR THE
INTACT DERBYSHIRE AND FOR THE DAMAGED DERBYSHIRE
(SCENARIO A) .......................................................................................................... 241
APPENDIX I THEORY BEHIND THE RELATIVE VERTICAL MOTION . 248
APPENDIX J RELATIVE VERTICAL MOTION FOR DIFFERENT
SELECTED POINTS ALONG THE SHIP ............................................................. 252
APPENDIX K EXTENSION OF EQUATIONS OF MOTION FOR A SHIP TO
INCLUDE INFLUENCES OF INGRESSED WATER WITHIN THE DAMAGED
HOLD(S) ..................................................................................................................... 288
IV
LIST OF TABLES
Table 3.1: Damage dimensions .................................................................................... 59
Table 6.1: Main particulars of the Derbyshire taken from the Department of Transport
(1989) ................................................................................................................. 116
Table 6.2: The four masses and their longitudinal positions for parent
Derbyshire .......................................................................................................... 119
Table 6.3: The horizontal and vertical positions of the four masses for parent
Derbyshire .......................................................................................................... 119
Table 6.4: Parameters for the parent and optimised hull forms ................................. 124
Table 6.5: Intact stability curve areas calculated using Optistanbul suite .................. 127
Table 6.6: Intact stability curve areas calculated using Wolfson unit hydrostatics and
stability code ....................................................................................................... 127
Table 6.7: Calculated IMO intact stability curve properties from Optistanbul
suite ..................................................................................................................... 128
Table 6.8: Calculated IMO intact stability curve properties from Wolfson unit
hydrostatics and stability code ............................................................................ 128
Table 6.9: The four masses and their longitudinal positions for optimised
Derbyshire .......................................................................................................... 130
Table 6.10: The horizontal and vertical positions of the four masses for optimised
Derbyshire .......................................................................................................... 131
Table 6.11: Damage scenarios .................................................................................... 133
Table 6.12: Parallel sinkage, trim and heel angles for damage Scenario A ............... 135
Table 6.13: Parallel sinkage, trim and heel angles for damage Scenario B ............... 135
Table 6.14: Parallel sinkage, trim and heel angles for damage Scenario C ............... 135
Table 6.15: Parallel sinkage, trim and heel angles for damage Scenario D ............... 135
Table 6.16: Position of the centre of gravity for different damage scenarios for the
parent Derbyshire ............................................................................................... 137
optimised Derbyshire .......................................................................................... 137
Table 6.18: Position of the point masses for damage scenario A for the parent
Derbyshire .......................................................................................................... 137
Table 6.19: Position of the point masses for damage scenario B for the parent
Derbyshire .......................................................................................................... 138
Table 6.20: Position of the point masses for damage scenario C for the parent
Derbyshire .......................................................................................................... 138
Table 6.21: Position of the point masses for damage scenario D for the parent
Derbyshire .......................................................................................................... 138
Table 6.22: Position of the point masses for damage scenario A for the optimised
Derbyshire .......................................................................................................... 138
Table 6.23: Position of the point masses for damage scenario B for the optimised
Derbyshire .......................................................................................................... 139
Table 6.24: Position of the point masses for damage scenario C for the optimised
Derbyshire .......................................................................................................... 139
Table 6.25: Position of the point masses for damage scenario D for the optimised
Derbyshire .......................................................................................................... 139
Table 7.1: Hydrostatic coefficients for the intact case for the parent Derbyshire ...... 171
Table 7.2: Hydrostatic coefficients for the damage case scenario A for the parent
Derbyshire .......................................................................................................... 172
Table 7.3: Hydrostatic coefficients for the damage case scenario B for the parent
Derbyshire .......................................................................................................... 172
Table 7.4: Hydrostatic coefficients for the damage case scenario C for the parent
Derbyshire .......................................................................................................... 172
Table 7.5: Hydrostatic coefficients for the damage case scenario D for the parent
Derbyshire .......................................................................................................... 172
Table 7.6: Hydrostatic coefficients for the intact case for the optimised
Derbyshire .......................................................................................................... 173
VI
Table 7.7: Hydrostatic coefficients for the damage case scenario A for the optimised
Derbyshire .......................................................................................................... 173
Table 7.8: Hydrostatic coefficients for the damage case scenario B for the optimised
Derbyshire .......................................................................................................... 173
Table 7.9: Hydrostatic coefficients for the damage case scenario C for the optimised
Derbyshire .......................................................................................................... 173
Table 7.10: Hydrostatic coefficients for the damage case scenario D for the optimised
Derbyshire .......................................................................................................... 174
Table 8.1: Location of the points of interest on intact hull forms .............................. 179
Table 8.2: Location of the points of interest for the damaged parent Derbyshire hull
form .................................................................................................................... 179
Table 8.3: Location of the equivalent points of interest for the damaged optimised
Derbyshire hull form .......................................................................................... 179
VII
LIST OF FIGURES
Figure 3.1: Relative occurrence of different classifications of accidents for UK
registered merchant vessels .................................................................................. 25
registered fishing boats .......................................................................................... 27
Figure 3.3: Damage database (Lutzen (2002)) .............................................................. 29
Figure 3.4: Two main accidents from damage database (Lutzen (2002)) ..................... 31
Figure 3.5: The representation of the location of the damage ....................................... 32
Figure 3.6: Location of damage in collision incidents for different ship types ............. 34
Figure 3.7: Location of damage in grounding incidents for different ship types ......... 36
Figure 3.8: Damage properties in collision incidents for general cargo ships for the
3L/4-FP region ...................................................................................................... 40
Figure 3.9: Damage properties in collision incidents for tankers for the 3L/4-FP
region .................................................................................................................... 43
Figure 3.10: Damage properties in collision incidents for containerships for the L/4amidships region ................................................................................................... 46
Figure 3.11: Damage properties in collision incidents for containerships for the 3L/4-FP
region ..................................................................................................................... 49
Figure 3.12: Damage properties in grounding incidents for general cargo ships for the
amidships-3L/4 region .......................................................................................... 52
Figure 3.13: Damage properties in grounding incidents for tankers for the 3L/4-FP
region ..................................................................................................................... 55
Figure 3.14: Damage properties in grounding incidents for bulk carriers for the 3L/4-FP
region .................................................................................................................... 58
Figure 4.1: The influence of permeability on the floodable length curve for a
cargo/passenger vessel of 134m in length ............................................................ 69
Figure 4.2: Definition of the water head, h (IMO (2002a)) ....................................... 74
Figure 4.3: Vertical position of damage ....................................................................... 78
VIII
Figure 5.1: The motion-hydrodynamic Cartesian coordinate reference system ............ 93

Figure 5.2: Total mass of the ship represented as four equivalent point masses ........ 103
Figure 5.3: Description of the trim angle and parallel sinkage .................................. 110
Figure 5.4: Description of the heel angle ................................................................... 110
Figure 6.1: The parent hull form ................................................................................ 117
Figure 6.2: The modified longitudinal mass distribution of the Derbyshire .............. 118
Figure 6.3: The variation of the objective function and the calm water resistance with
iteration number .................................................................................................. 122
Figure 6.4: The parent (continuous) and optimised (dashed) hull forms ................... 123
Figure 6.5: Relative bow motion responses for the parent and optimised Derbyshire
from Optistanbul Suite ....................................................................................... 124
Figure 6.6: The intact stability curves for the parent and for its optimised
configuration ....................................................................................................... 127
Figure 6.7: The presentation of the four point masses for the parent and optimised hull
forms ................................................................................................................... 130
Figure 6.8: The floodable length curves for the parent and optimised hull forms for
different levels of permeability .......................................................................... 132
Figure 7.1: Discretisation of the intact Derbyshire .................................................... 146
Figure 7.2: Discretisation of the intact optimised Derbyshire ..................................... 147
Figure 7.3: Conditioning number for the intact Derbyshire ....................................... 148
Figure 7.4: Sway-roll and roll-sway added mass and fluid damping coefficients for the
intact Derbyshire ................................................................................................ 149
Figure 7.5: Surge wave excitation forces for the intact Derbyshire (in head seas) .... 150
Figure 7.6: Heave wave excitation forces for the intact Derbyshire (in head seas) ... 150
Figure 7.7: Pitch wave excitation moments for the Intact Derbyshire (in head seas) 151
Figure 7.8: Discretisation of the damaged Derbyshire scenario A. ............................ 152
Figure 7.9: Conditioning number for the damaged Derbyshire (scenario A) ............. 152
Figure 7.10: Sway-roll and roll-sway added mass and fluid damping coefficient for the
damaged Derbyshire (scenario A) ...................................................................... 153
IX
Figure 7.11: Surge wave excitation forces for the damaged Derbyshire scenario A (in
head seas) ............................................................................................................. 154
Figure 7.12: Heave wave excitation forces for the damaged Derbyshire scenario A (in
head seas) ............................................................................................................ 154
Figure 7.13: Pitch wave excitation moments for the damaged Derbyshire scenario A (in
head seas) ............................................................................................................ 155
Figure 7.14: Pure surge added mass coefficient for intact and damaged hull
forms ................................................................................................................... 156
Figure 7.15: Pure sway added mass coefficient for intact and damaged hull
forms .................................................................................................................... 157
Figure 7.16: Pure heave added mass coefficient for intact and damaged hull
forms .................................................................................................................... 157
Figure 7.17: Pure roll added mass coefficient for intact and damaged hull forms ...... 158
Figure 7.18: Pure pitch added mass coefficient for intact and damaged hull
forms .................................................................................................................... 158
Figure 7.19: Pure yaw added mass coefficient for intact and damaged hull forms ..... 159
Figure 7.20: Pure surge fluid damping coefficient for intact and damaged hull
forms ................................................................................................................... 159
Figure 7.21: Pure sway fluid damping coefficient for intact and damaged hull
forms ................................................................................................................... 160
Figure 7.22: Pure heave fluid damping coefficient for intact and damaged hull
forms .................................................................................................................... 160
Figure 7.23: Pure roll fluid damping coefficient for intact and damaged hull
forms .................................................................................................................... 161
Figure 7.24: Pure pitch fluid damping coefficient for intact and damaged hull
forms .................................................................................................................... 161
Figure 7.25: Pure yaw fluid damping coefficient for intact and damaged hull
forms .................................................................................................................... 162
Figure 7.26: Heave induced surge added mass for the intact and damaged hull
forms .................................................................................................................... 164
X
Figure 7.27: Pitch induced surge added mass for the intact and damaged hull
forms ................................................................................................................... 164
Figure 7.28: Roll induced sway added mass for the intact and damaged hull
forms .................................................................................................................... 165
Figure 7.29: Yaw induced sway added mass for the intact and damaged hull
forms .................................................................................................................... 165
Figure 7.30: Pitch induced heave added mass for the intact and damaged hull
forms ................................................................................................................... 166
Figure 7.31: Yaw induced roll added mass for the intact and damaged hull
forms ................................................................................................................... 166
Figure 7.32: Heave induced surge fluid damping for the intact and damaged hull
forms .................................................................................................................... 167
Figure 7.33: Pitch induced surge fluid damping for the intact and damaged hull
forms ................................................................................................................... 167
Figure 7.34: Roll induced sway fluid damping for the intact and damaged hull
forms ................................................................................................................... 168
Figure 7.35: Yaw induced sway fluid damping for the intact and damaged hull
forms ................................................................................................................... 168
Figure 7.36: Pitch induced heave fluid damping for the intact and damaged hull
forms ................................................................................................................... 169
Figure 7.37: Yaw induced roll fluid damping for the intact and damaged hull
forms ................................................................................................................... 169
Figure 8.1: Relative positions of the points of interest for the parent Derbyshire for
investigating vertical displacement .................................................................... 176
Figure 8.2: Heave motion for the intact and damaged parent and optimised hull forms of
Derbyshire ........................................................................................................... 181
Figure 8.3: Roll motion for the intact and damaged parent and optimised hull forms of
Derbyshire .......................................................................................................... 181
Figure 8.4: Pitch motion for the intact and damaged parent and optimised hull forms of
Derbyshire ........................................................................................................... 182
XI
Figure 8.5: Relative vertical motion responses for the intact and damaged hull forms for
point A (in head seas) .......................................................................................... 183
point B (in head seas) ......................................................................................... 184
point G (in head seas) ......................................................................................... 184
point C (in head seas) .......................................................................................... 187
point F (in head seas) .......................................................................................... 187
Figure 8.10: Relative vertical motion responses for the intact and damaged hull forms
for point D (in head seas) ................................................................................... 188
for point E (in head seas) .................................................................................... 189
Figure 8.12: Definition of wave headings .................................................................. 190
Figure 9.1: An overview of the dynamic analysis procedure ..................................... 195
XII
ACKNOWLEDGEMENTS
I would like to thank to all of the following people without whom this research could
not have been undertaken.
In particular, I would like to thank to Dr. . Belik, Professor P. Temarel and Professor
P.A. Wilson, without their presence such a research could not have started. It was a
great pleasure to work under the supervision of Professor G.E. Hearn. I am really
grateful for his support and encouragement. In addition his and Jenny Hearns excellent
hospitability is greatly appreciated. My thanks are also directed to the members of the
Department of Ship Science, staff and students.
The opportunity to meet Dr. Ltzen arose at just the right time. I greatly appreciate the
discussions with her during this research.
I would like to acknowledge zgr for his invaluable help, support and encouragement.
Last, but certainly not least, I am grateful to my late Grandfather and Grandmother,
Mum and Dad, my sister, zgrs family and my friends at the University for always
being there for me spiritually and physically.
XIII
NOMENCLATURE
At the end of variable definitions a list of acronyms used within the text are defined.
Distribution
density
of
damage
location
along
ships
length
awpc
Waterplane area of compartment, that is, area between selected

bulkheads
Attained survival probability index
[ A]
Linear motions coefficient matrix including hydrodynamic,

hydrostatic and mass distribution influences
Akj
Added mass matrix for intact ship. Elements Akj denote added
mass coefficient for hydrodynamic reaction in kth direction
arising from motion in the jth direction. The pure added mass
coefficient for j, k = 1,2,3 is given in kg, whereas the pure added
mass coefficient for j, k = 4,5,6 is presented in kg m2. Finally,
the cross-coupled added mass coefficients are given in kg m
Akj
Added mass matrix of damaged ship. Akj is value of Akj for

damaged ship
Amax
Required subdivision index associated with probabilistic

approach to damaged stability of a ship
Awps
Waterplane area of intact ship
Awps
Waterplane area of damaged ship
AP , FP
Aft and forward perpendiculars of ship
AP , AS
Subscripts indicating quantity associated with aft portion of ship

either to port or starboard
Either penetration of damage (m) or base points for the objective

function in Hooke-Jeeves optimisation method
XIV
Breadth (or moulded beam) of ship
Bkj
Fluid damping matrix of intact ship. Bkj denotes fluid damping

coefficient for hydrodynamic reaction in kth direction arising
from motion in the jth direction. The pure fluid damping
coefficient for j, k = 1,2,3 is given in kg rad/s, whereas the pure
fluid damping coefficient for j, k = 4,5,6 is presented in kg m2
rad/s. Finally, the cross-coupled fluid damping coefficients are
given in kg m rad/s
Bkj
Fluid damping matrix of damaged ship. Bkj is value of Bkj for

damaged ship
BWL
Breadth of ship at still waterline
CB
Block coefficient
Ckj
Hydrostatic restoring matrix of intact ship. Ckj is hydrostatic

coefficient for restoration in kth direction arising from motion in
the jth direction. The pure hydrostatic coefficient for j, k = 3 is
given in kg/s2, whereas the pure hydrostatic coefficient for j, k
= 4,5 is presented in kg m2/s2. Finally, the cross-coupled
hydrostatic coefficients are given in kg m/s2
Ckj
Hydrostatic restoring matrix of damaged ship. Ckj is value of

Ckj for damaged ship
CN
Conditioning
number
of
hydrodynamic
fluid
interaction influence matrix of ship being analysed

CP
Prismatic coefficient
CWP
Waterplane area coefficient
dm
Elemental constituent mass for ship
dm
Elemental constituent mass of ingressed water
Depth of intact ship

XV
structure
Factor of subdivision, assumed distribution function of damage

location along ships length and wave exciting forces and
moments matrix.
Fk
Time dependent wave exciting force/moment in kth direction

for intact ship. Fk used to form kth row of F. The wave exciting
force for j, k = 1,2,3 is given in kg m/s2, whereas the wave
exciting moment for j, k = 4,5,6 is presented in kg m2/s2
Fk
Corresponding value of Fk for damaged ship
FP , FS
Subscripts indicating quantity associated with forward portion of

ship either to port or starboard
Acceleration due to gravity
g ( x)
Objective function in Hooke-Jeeves optimisation method
GM
Metacentric height
GZ
Righting lever
GZ max
Maximum righting lever
Either elevation of water on deck, measured with respect to

undisturbed free surface or height of damage (m) presented as a
proportion of the ship depth D
H ( )
Motion transfer function
HS
Catamaran demi-hull separation parameter
HJ
Step length in Hooke-Jeeves optimisation method
i, j, k subscripts used singly or in pairs to denote a member of a group of values.

I XX , IYY , I ZZ
or
Pure moments of inertia of intact ship
I 55 , I 66 , I 44
, IYY
, I ZZ
I XX
or
Pure moments of inertia of damaged ship
, I 66
, I 44
I 55
XVI
I XY , I ZX , IYZ
or
Products of inertia for intact ship
I 56 , I 54 , I 46
, I ZX
, IYZ
I XY
or
Products of inertia for damaged ship
, I 54
, I 46
I 56
J max
Maximum non-dimensional damage length
k XX , kYY , k ZZ
Radius of gyration with respect to X, Y and Z axis respectively
Extent of damage (m)
l p , lr
Levers for pitch and roll motions
Length of ship between perpendiculars
LOA
Overall length of ship
LS
Moulded length of ship
LWL
Length of vessel at still water waterline
LCB
Longitudinal centre of buoyancy
LCF
Longitudinal centre of flotation
LCG
Longitudinal centre of gravity
mw
Total mass of ingressed water
Ship mass
M AP , M AS , M FP , M FS Representative point masses for intact ship

, M AS , M FP
, M FS
Representative point masses for damaged ship made up from
M AP
intact ship point masses and ingressed water
M kj
Generalised mass matrix of intact vessel
M kj
Generalised mass matrix of damaged vessel
Ml
Mass of liquid contained within a compartment of the intact ship
XVII
M XYZ
Earth-fixed right-handed Cartesian frame of axes (X, Y, Z) with

its origin, M, located at the intersection of the undistributed freesurface, the mid-ship section and the longitudinal central plane
of symmetry of the ship in its intact position of static
equilibrium.
Iteration number
pi
Probability of damage of either the ith ship compartment or ith

group of ship compartments
Required subdivision index, often used in stability regulations as

being an equivalent alternative designation for Amax
R ( )
Response spectrum
RVM ( z, x )
Amplitude of vertical relative displacement
si
Probability of survival of either the ith ship compartment or ith

group of ship compartments
S ( )
Wave spectrum
SV ( z, x )
Amplitude of vertical displacement
SW
Wetted surface of ship
Time
Draught of ship
Service speed of ship
wlost
Structural mass loss due to ship damage
Variables
for
the
objective
function
in
Hooke-Jeeves
optimisation method
x1 , x2
Distance from the aft terminal of LS to the foremost/ aftermost

portion of the aft/forward end of the compartment being
considered
XVIII
XP , XS
Transverse distance of port and starboard specified point

masses with respect to MXYZ
YAP , YAS , YFP , YFS
Vertical distance of aft and forward located port and starboard

point masses
YB
Vertical centre of buoyancy for intact ship
YB
Vertical centre of buoyancy for damaged ship
YSINK
Measure of extent of parallel sinkage
Vertical position of damage (m)
Z , X ,Y
Distance of elemental ship mass coordinates with respect to the

MXYZ reference system
Z , X , Y1
New location of X , Y , Z after ship experiences parallel

sinkage
ZA , ZF
Longitudinal distance of aft and forward located point masses
Z A , X P , YAP
Location of M AP for the damaged ship
Z A , X S , YAS
Location of M AS for the damaged ship
ZC , X C , YC
Centre of flotation for intact ship
ZC , X C , YC
Centre of flotation for damaged ship
Z F , X P , YFP
Location of M FP for the damaged ship
Z F , X S , YFS
Location of M FS for the damaged ship
Z G , X G , YG
Centre of gravity for intact ship
Z G , X G , YG
Centre of gravity for damaged ship
Z m , X m , Ym
Location of elemental constituent mass for intact ship
Z m , X m , Ym
Location of elemental constituent mass for damaged ship
Z w , X w , Yw
Location of elemental mass of water
Z1 , X , Y2
New location of X , Y , Z after ship heels
Z1 , X 1 , Y3
New location of X , Y , Z after ship trims

XIX
Angle to determine the position of bulkheads
Displaced weight for intact ship
Displaced weight for damaged ship
( z , x, t )
Complex wave amplitude
Six degree of freedom motion displacement column matrix
Time dependent motion response of intact ship in jth degree of

freedom. The motion response amplitude for j=1,2,3 is given in
metres, whereas the motion response amplitude for j=4,5,6 is
presented in radians.
Time dependent motion response of damaged ship in jth degree

of freedom
Velocity in jth degree of freedom for intact ship
Velocity in jth degree of freedom for damaged ship
Acceleration in jth degree of freedom for intact ship
Acceleration in jth degree of freedom for damaged ship
ja
Complex amplitude of jth degree of freedom motion response of

intact ship
ja
Complex amplitude of jth degree of freedom motion response of

damaged ship
ji , jr
Imaginary and real parts of ja
ji , jr
Imaginary and real parts of ja
Generalised direction cosine in the kth direction
Heel angle
Volumetric permeability of the damaged compartment
Surface permeability of the damaged compartment
vc
Compartment volume
XX
vi
Intact volume of ith compartment or ith grouping of

compartments
v jn
Normal velocity component of point on wetted surface of ship

due to oscillation in the jth degree of freedom
vlost
Lost buoyancy volume
Associated material density
Fluid density
Sea water density
Either trim angle or time independent velocity potential
Diffraction velocity potential
Radiation velocity potential associated with motion in jth degree

of freedom
ji , jr
Imaginary and real parts of jth radiation velocity potential
Velocity potential of incident wave
Time dependent velocity potential defined
Wave frequency
Displaced volume for intact ship
Displaced volume for damaged ship
CEM
Concept exploration model
CFD
Computational fluid dynamics
HARDER
Harmonization of rules and design rationale
IMCO
Inter-governmental maritime consultative organisation
IMO
International maritime organisation
ITTC
International towing tank conference
MAIB
Marine accident investigation branch
MARPOL
The international convention for the prevention of pollution

from ships
O.B.O
Oil-bulk-ore
XXI
RAO
Response amplitude operator
RBM
Relative bow motion
RO - RO
Roll on roll off
SEM
Static equivalent method
SOLAS
Safety of life at sea
UK
United Kingdom
XXII
1.
INTRODUCTION
This dissertation addresses the impact of damage upon the hydrodynamic and
hydrostatic characteristics and the motion responses of a ship hull form optimised for
seakeeping with no increase in calm water resistance. To appreciate how optimisation
may assist the ship design process, a short overview of ship design approaches is
provided in the next section. This is followed by a brief overview of what optimisation
can achieve and this provides the identification of a possible shortcoming in existing
optimisation tools for improved hydrodynamic performance. It is this possible defect
and its consequences that drive the research.
1.1 The Ship Design Concept

Ship design is an interesting and demanding task. It provides the best opportunity of
combining theoretical analysis, scientific knowledge of hydrostatics, hydrodynamics,
dynamics, materials and structures, with the historical experience that include
perceived beneficial ratios of geometric dimensions and form factors.
Depending on the ship type a volumetric or deadweight design will be undertaken to

identify the principal hull form characteristics for the required cargo capacity, ship
speed and range. Volumetric design is used when the ship must provide a specific
amount of cargo space. This approach is applied to passenger ships, containerships and
most naval ships. Deadweight design is based on equating the sum of the lightship
component masses and the cargo masses to the ship displacement. It is used mainly in
the design of tankers and bulk carriers.
An important step in the earlier stages of design is the selection of principal ship
dimension related parameters such as displacement, length, beam, draught, depth and
block coefficient. At this stage of design few naval architecture tools can be applied.
1
The preliminary design stage is generally carried out in three ways. The first method is
known as the basis ship approach. This technique, which uses a particular existing hull
form to lead the families of designs. This approach was very useful when the number of
new ships being constructed was high, so that design can be optimised by small
improvements from ship to ship and class to class. However, when the number of new
ships constructed decreases, such as at the present time and single vessel classes become
more common this method looses its effectiveness.
Another design method is to use the trend curves. This method is based on the plots of
the data gathered from existing ships (see, for example, Watson (1962) and Watson &
Gilfillan (1977)). This technique provides an interval of values for specific ship sizing
parameters to be used. Consequently, the quality of new designs based on this approach
depends on both the relevance and inappropriateness of previous designs included in the
associated database. This aspect of design is also discussed by Sarioz (1993).
Furthermore, this method limits the capability of the designer to create novel designs, as
the new designs based on this method will have similar features.
The third option is the parametric survey (see, for example, Murphy et al. (1965) and
Gilfillan (1969)). This approach looks for the most advantageous combination of overall
ship dimensions and form parameters within defined limits of the design study. This
method tries to find the optimum hull form for the prescribed criteria but it does not
look at different alternatives.
All these methods provide an initial estimate of the indicated principal hull form
parameters. After the basic ship parameters are defined the resulting design must be
examined to check its technical and economic feasibility in terms of intact stability,
cargo carrying capacity and freeboard & power requirements and if it achieves the goals
necessary for the intended service. Consequently, the design is developed iteratively.
The indicated technical analyses are usually carried out when the design geometric
characteristics are well established. Therefore, some required changes may not be
accommodated due to the level of commitment to the design and the perceived costs in
design re-development. The failure to reflect the results of the most accurate analysis in
the design process limits the usefulness of empiricism and experience based design
methods. An alternative approach is to apply rational computer-based searches or
optimisation tools in the earlier design stages to improve the measures of merit for
seakeeping, resistance, structural and economic qualities.
The complexity of the design relies on different features of the technical design, the
construction of the ship and its operation. As the ship dimensions have a direct effect on
material and construction cost the ship-builder requires a design with minimum
dimensions to achieve the prerequisites of the ship owner. At the same time the operator
of the ship restricts the design for the minimum operational and capital costs.
Because of these different objectives and its multidisciplinary nature the ship design
process is developed iteratively. This iterative process can be expressed schematically
as a design spiral (see, for example, Andrews (1981)). At the design spiral early
estimates of the ship sizing parameters are made. As the design proceeds along the
spiral these estimates are altered and developed as different levels of analysis of
differing complexity are applied. This continuous evolution of the design is the result of
the feedback provided at subsequent steps and increased information about the
sensitivities of the design. The process continues until convergence and a final
economic and technical feasible design is produced.
The outcome of this process represents a compromise of choices as influenced by; the
route of the ship, the physicals limitations associated with water depth and canal
dimensions, the complicated environment the ship must survive, the type of cargo
carried, port operating restrictions and the minimum requirements of Classification
Authorities on intact and damage stability.
3
The internal subdivision of the ship by means of transverse and/or longitudinal

watertight bulkheads or by horizontal subdivisions like double bottoms in commercial
ships might add to the cost of the ship. This unavoidably involves a compromise
between safety and cost. The more severe the standard adopted for subdivision and
stability, the greater the probability that capital and operating costs will increase and the
economic viability of the ship may be compromised. Therefore, the stability
requirements are usually satisfied at the minimum level required by the Classification
Authorities.
1.2 Optimisation Tools

A design based optimisation tool generally determines a technically more advantageous
set of design parameters. The parameters may be related through specific mathematical
functions or defined by some empirical formulae. These parameters will be subject to
physical, technical, legal and economic restrictions. If more than one combination of the
design variables satisfies all these conditions, the algorithm within the optimisation
tools determines that combination of design variables that optimises the hull form for
some measure of merit specified by the designer (see, for example, Schneekluth &
Bertram (1998)). This ability to iterate several hull form parameter changes at the
design stage encourages the designer to carry out alternative modifications of the design
parameters in a way that was not previously possible.
1.3 A Possible Shortcoming of Existing Optimisation Tools

Seakeeping and resistance qualities of a ship have significant impact on the operability
of the ship and hence upon the economic viability of the design. Consequently most
ship design optimisation tools addressing ship hydrodynamic aspects consider
seakeeping or resistance or both as primary tasks whereas intact stability is usually
treated as a design constraint to be fulfilled (see, for example, Sarioz (1993) and Keane
4
et al. (1991)). However, it is not evident that damage stability is addressed at all in such
optimisation tools. Therefore, it could be the case that the computer based optimisation
programs, which are used for improving the behaviour of the non-damaged hull form
in terms of seakeeping, resistance, structural strength and economy may lead to ship
designs that are less survivable when damaged. Essentially this is the question
addressed in this research programme (see Saydan and Hearn (2004)).
1.4 Aims and Outline of the Research

The absence of damage stability considerations in an optimisation routine that addresses
hydrodynamic performance subject to intact stability may be viewed as incomplete. To
determine whether this is the case one must either include damage stability in the
optimisation process and examine its impact on the computer programme outcomes, or,
undertake hydrodynamic and motion analyses of an assumed non-optimised hull form
and its optimised hull form to try and demonstrate a shortcoming in the optimisation
process. At this stage there is no need to comment on the pros and cons of either
possible approach since it is necessary to investigate all aspects of the posed problem
and to identify a possible feasible way forward. Certainly it would be extremely
perverse if one were to find that a damaged optimised hull form was less survivable
than the associated damaged conceptual form.
Improving hydrodynamic performance requires appropriate modification of the

geometric form of the basis hull. Modifying the hull form is not an easy task without an
appropriate wealth of design/operational experience related to the ship type addressed.
Hence a systematic process to try and identify the beneficial changes is required. This
could be undertaken by using design charts, that is, plots of ship behaviour
characteristics as a function of different design pairings. These pairings are designated
primary or secondary parameter changes (L and B/T form the first set and CWP, LCB
and LCF the second set) when creating new/novel designs.
5
Having perceived which changes are beneficial in terms of the vessel designed, the
continuous iteration of the selected parameters should be undertaken by using a suitable
algorithm in order to produce new hull forms. This could be achieved using so called
inverse analysis whereby hull form shapes have preferable seakeeping and resistance
responses (compared to the initial designed ship) are sought by seeking out beneficial
geometric changes. In Chapter 2 this inverse analysis is explained and the approach of
different researchers is reviewed. Various researchers have used different objective
functions (or drivers). The Hooke-Jeeves optimisation process (see, for example, Hooke
& Jeeves (1961), Kowalik & Osborne (1968), Aoki (1971) and Walsh (1975)) is used to
provide an optimised hull form.
Prior to attributing damage to any ship (optimised or otherwise) it is necessary to

understand what damage is most likely together with most likely location and the
expected extent of the damage. These aspects are discussed and appropriate statistical
results presented in Chapter 3.
Chapter 4 reviews the development of ship regulations and stability analysis and
identifies why certain developments have taken place over the past 76 years. Two
possible methods of examining the stability of a ship when it is damaged are discussed.
The first one is deterministic and is mandatory, whereas the second one is probabilistic
approach and is used when the deterministic method is found unsatisfactory. An intense
analysis of the two different stability methods provides their advantages and
disadvantages for the intended damage research. Selection of the deterministic approach
is justified in the care of general optimisation used with novel hull forms.
The seakeeping analysis of intact and damaged ships is, in principle, the same but the
data required to carry out motion analysis of damaged ships is not always available. In
Chapter 5 frequency versus time domain analysis together with the presentation of the
required generalised motion equations is provided. A novel method of providing
products of inertia is also presented.
6
The hydrodynamic analysis of the optimised hull and basis hull for the intact and
damage cases is carried out using the Matthew Diffraction Suite (Hearn (1978)). In
Chapter 6 the Hooke-Jeeves optimisation process as built into the Optistanbul Suite
(Sarioz (1993)) is applied to the selected Derbyshire hull form. The reasons for its
selection and the objective function and constraints used are explained. Validation of
the intact stability of both basis and optimised hull forms are provided together with
approval of the bulkhead division in each case. The damage statistics of Chapter 3 are
used to determine the orientation of the parent and optimised Derbyshire hull form for
four distinct damage scenarios.
Chapter 7 then undertakes a three-dimensional hydrodynamic analysis of all selected ten

hull forms and justifies the quality of the results prior to undertaken relative vertical
motion response calculations in Chapter 8.
Chapter 9 provides a closure to the thesis with general comments and observations,
conclusions and an indication of the future research that could prove beneficial.
2.
THE OPTIMISATION PROCESS
The improvement of ship design using optimisation techniques requires selection of an

appropriate optimisation tool, the selection of an objective function that reflects the
improvements sought and appropriate supporting analysis to undertake evaluation of the
parameters (variables) included in the selected objective function. The process is
essentially iterative with the initial independent parameters (variables) being gradually
modified so as to generate improvements in the selected metric or metrics of interest.
Optimisation can be misleading if one aspect of the design is improved with little or no
consideration of other equally important aspects. The quality of the outcomes can also
depend upon the ability of the analysis tools to properly reflect the impact of changes in
the governing design parameters selected. The governing parameters selected, if
inappropriate may limit the capability of the optimisation process to identify a true
global optimal solution.
Thus optimisation tools may provide a large number of alternative designs within a
relatively short time with better characteristics than the parent design. In many cases
important constraints will have influenced the outcome. In this chapter previously
applied optimisation tools, applied in the context of improved hydrodynamic and
motion characteristics, will be reviewed and assessed in terms of their strengths and
weaknesses and their potential to include damage stability assessments.
Since this PhD study will use tools generated initially at the University of Newcastle the
first part of the review will summarise the research carried out at the University of
Newcastle. Thereafter many other examples of analysis used in other aspects of ship
design optimisation are presented and discussed.
2.1 Background
2.1.1 A Specific Motion and Resistance Driven Optimisation Research Programme
In the late 1980s, Hearn et al. (1988) developed a Frank close-fit (see Frank (1967))
velocity potential based strip theory for in-house seakeeping analysis within the design
offices of British Shipbuilders. Since there was little familiarity (within the design
offices) with the mathematical formulation and solution of either the ship motions & the
dynamic shear force/bending moment characteristics or the associated fluid structure
interaction analysis the computer system developed Lynette Suite (Hearn et al. (1988))
required many self correcting and automatic recovery processes. Because the designers
in British Shipbuilders would not agree to geometric data being provided in a preferred
form, automatic spline fitting through arbitrary defined waterplanes and transverse
sections was necessary to generate required geometric data at the usual 21 stations. This
in turn also required automatic discretisation of the stations to allow the necessary
hydrodynamic analysis. Availability of robust friendly software with error detection and
recovery was thought by the authors to be a necessary tool to allow improved initial
designs within British Shipbuilders. It was soon evident that the designers could not
control the development of the hull forms using fully automated analysis. A search
methodology was required to indicate what level of changes was needed to achieve the
desired performance improvements. Thus the need for seeking an optimisation process
was industry based not research driven within the academic environment.
An approach developed by Hearn et al. (1990, 1991 and 1992), Sarioz et al. (1992) and
Hearn et al. (1994 and 1995a) addressed the search methodology for improved
seakeeping with consideration of frictional resistance, wave making resistance, added
resistance and satisfaction of IMO intact stability requirements. Their initial approach
was designated the Forward technique because it simply looked at straightforward use
of the analysis tools, whilst investigating which geometric hull design parameters were
9
influential in affecting the design and the sensitivity of the different hydrodynamic
characteristics to different level of changes in different hull form parameters. Having
gained some insight through the Forward technique, a fully automated optimisation
approach the Inverse technique was developed since it used the analysis to identify the
hull form design parameter values. Both techniques seek cause and effect information
regarding the hydrodynamic characteristics dependence upon hull form parameters.
The forward technique simply analyses the required engineering responses for defined
variations of the selected hull form parameters. The literature search undertaken by
Hearn et al. (1990) indicated the dependence of ship motions upon the secondary
parameters of CWP, LCB and LCF. Consequently, these secondary parameters were
investigated on the basis they influenced seakeeping and related quantities, but were not
present in standard empirical resistance calculation formulae. Hence it was argued that
seakeeping
could
be
influenced
without
significantly
influencing
resistance
characteristic. This information is used to generate so called design charts that provide
a three-dimensional graphical representation of the cause and effect relationship for
different engineering responses as a function of different pairs of hull form parameters.
In theory one could use the resulting surface plots to try and identify beneficial changes
of the initial hull. Secondary parameters are investigated using the Lackenby
transformation (see Lackenby (1950)) to change the sectional area curve (to change
LCB) and the waterline curve (to change LCF and CWP). Each parameter can be
modified without influencing the other two parameters. A year later, Hearn et al. (1991)
included the effect of altering the primary parameters of L and B/T. The primary
parameters are modified by using linear distortion methods in which the displacement
of the ship and its block coefficient remain fixed. Thus any proportional change in ship
length is balanced by appropriate adjustments of the value of the product of beam and
draught subject to B/T remaining invariant.
The inverse method systematically modifies all permitted hull form parameters to
identify a more beneficial hull form. That is, those combinations of hull form parameter
10
changes that provide improved engineering responses per se. In each step a hull form is
generated and analysed. Initially, the Landweber-Macagno three parameter conformal
mapping procedure (see Landweber & Macagno (1959)) was used to develop each
modified transverse section. The hydrodynamic characteristics of these sections are then
determined using a Frank close-fit method. A year later, the generation of alternative
hull forms is produced using the Lackenby transformation by Hearn et al. (1991).
In the inverse approach the optimisation technique of Hooke-Jeeves (see Appendix A)

was employed. The responses investigated were seakeeping responses selected from a
menu of possibilities, the wave making resistance and the frictional resistance subject to
the intact stability characteristics complying with the IMO requirements (outlined in
Section 6.4 of this dissertation). The objective function was a linear combination of the
selected hydrodynamic responses, each scaled with respect to the response of the parent
(initial) design. Initially the objective function has a value of unity. The parameter
changes allowed were specified as part of the input data on the basis it should be the
designer who stipulates no-change or specifies permissible levels of change for each
parameter. Similarly the selection of those responses that are considered more critical
for the efficient operation of the ship operating on a specific route should be specified
by the designer prior to applying either the forward or inverse analyses.
Ship responses in random waves are modelled by Hearn et al. (1990 and 1991) using the
linear spectral analysis (see, for example, Lloyd (1998) and Saydan (1999)), that is,
irregular waves are assumed to correspond to the superposition of regular waves having
different wave lengths, wave amplitudes and directions. Having selected a wave
spectrum, S ( ) , representation of the operational area of the sea and calculated the
motion transfer functions, H ( ) , of the ship (and its variants resulting from the
optimisation search) the response spectrum or spectra can be calculated using
R ( ) = H ( ) S ( ) .
2
11
The term H ( )
is often referred to as the response amplitude operator (RAO).
Various statistical parameters of the response may be determined such as probability of

slamming, acceleration thresholds being exceeded, deck wetnesses et cetera.
The area under the response spectrum defines the variance of the ship responses to the
selected sea-state. Firstly this area is minimised for all sea states. In practice, Hearn et
al. (1992) and Sarioz et al. (1992) found it more suitable to reduce the peak value of the
response amplitude operator of selected response(s). This approach required less
computational effort and produced the same optimised hull form parameter values given
by the initial computationally intensive approach. The reduction of peak values leads to
a general reduction in the response amplitude operator amplitudes and the response
motion spectrum values are automatically decreased for all sea-states.
Sarioz et al. (1992) found that application of other researchers optimisation techniques
such as that of Lloyd (1991) based on optimising in a specific sea-state often meant the
optimised form was better in the selected sea-state but not all sea-states. Hearn et al.
(1990, 1991 and 1992) and Sarioz et al. (1992)s approach does not exhibit this
characteristic. Their optimisation method produced new hull forms which showed better
hydrodynamic characteristics for all sea-states. In these mono-hull studies strip theory
(see, for example, Salvesen et al. (1970)) is implemented for the seakeeping analysis;
whereas a selection of resistance prediction techniques were included such as the
method of Holtrop & Mennen (see Holtrop & Mennen (1982) and Holtrop (1984))
together with thin-ship wave-making resistance of Michell (1898) and International
Towing Tank Conference (ITTC) 1957 frictional resistance correlation curve, see, for
example, Hearn & Wright (1999). Intact stability is usually checked against the
International Maritime Organisations criteria contained within IMO A-749 (see Section
6.4). Manoeuvring analysis was added to the forward technique by Furukawa & Hearn
(2000). It is based on IMO manoeuvring requirements see, IMO (1993). Hearn et al.
(2000) added manoeuvring analysis to the inverse part of the optimisation process.
12
The mono-hull studies by Hearn et al. (1990, 1991 and 1992) and Sarioz et al. (1992)
were restricted to vertical motions in head waves such as heave, pitch, relative bow
motion (RBM) and slamming. This was because the authors, like Lloyd (1991),
believed that if a ship is optimised for vertical motions in head seas, it will generally
show better characteristics in other motions. In the study of multi-hull vessels by Hearn
et al. (1994, 1995a and 1995b) the same motion characteristics are analysed in head seas
with the inclusion of roll in bow and quartering seas; whereas again only the vertical
motions are analysed in head waves in Hearn et al. (1995c) and Hearn & Wright (1997,
1998a and 1998b).
For the twin-hull vessels the demi-hull separation parameter HS is added to the
primary parameters because the graphs of hydrodynamic coefficients of the twin hulls
plotted by Hearn et al. (1994) were significantly different for variations of this demihull separation parameter. The wave-making resistance is calculated in these twin-hull
studies using a modified Michell thin-ship theory (see Lunde (1951)) so that interaction
between demi-hulls is included as necessary, this is dependent upon demi-hull
separation and forward speed of advance (see, for example, Hearn & Wright (1997) and
Tuck (1987)). The Michell thin-ship theory is chosen because it has the advantage of
consistency of treatment regarding seakeeping and wave resistance in the sense that
both analysis methods use actual hull form shape rather than global hull form
parameters. In order to facilitate a wider range of hull parameter combinations, the
secondary parameters were extended to include CP in Hearn & Wright (1998a and
1999). CP does not directly influence the optimisation results, but facilitates seakeeping
and resistance conflict resolution through greater variation of the other parameters with
practical ship forms maintained.
Significantly different behaviour of twin-hull hydrodynamic coefficients compared to

mono-hull hydrodynamic coefficients is outlined by Hearn et al. (1994). This
necessitated preparing a significantly larger database of catamaran hydrodynamic
coefficients, compared to the mono-hull database (see Hearn et al. (1994 and 1995a)),
13
for efficient and accurate motion analysis without real-time calculation of required
hydrodynamic coefficients using Frank close-fit in-line. However, once the database is
provided the hydrodynamic coefficients for several alternative hull forms are generated
within a small amount of time. Then as in mono-hull studies, linear distortion methods
and the Landweber-Macagno three parameter conformal mapping procedure or
Lackenby transformation techniques are used in order to generate new hull forms.
For mono-hulls, in order to minimise a selected response of the ship, standard

optimisation techniques such as Hooke-Jeeves can be used. This is possible because the
objective function associated with the selected response of the ship has only one
minimum. However, the hydrodynamic coefficients of twin hulls plotted for heave
motion by Hearn et al. (1994) for two different non-dimensional hull separation
coefficients show the highly nonlinear variation of the catamaran responses with hull
geometry variations. Thus, whilst linear analyses were used in the calculation it was
found that the responses were very nonlinear with respect to the hull form design
parameter changes and so the Hooke-Jeeves approach was not appropriate. So a new
evolutionary programming based search strategy was proposed by Hearn et al. (1995a
and 1995b) to cope with the nonlinear nature of the catamaran objective function. This
Genetic Algorithm (see Hearn et al. (1995a and 1995b) and Bertram (2003)) made it
possible to converge to the global minimum despite the design charts being full of local
minima.
2.1.2 Ship Design Optimisation with Different Design Drivers
Other than the particular ship design optimisation research programme just reviewed,
one finds many other good examples of analysis being coupled with optimisation
techniques. One of the earliest procedures to identify the principal ship dimensions in
concept or preliminary design was that of Watson (1962) and Watson & Gilfillan
(1977). In this technique experience in previous designs is stored in a series of graphs
and presented as a ratio of ship hull form parameters such as L/B, B/D and D/T. Here
14
design process starts by the assumption of three ship lengths, then by using the plots and
relations provided in Watson (1962) and Watson & Gilfillan (1977) it is possible to
determine beam, draught and block coefficient and hence obtain a displacement. Once
the lightship weight for each ship is calculated through the plots and relations of Watson
(1962) and Watson & Gilfillan (1977), it is subtracted from the displacement to obtain
deadweight values for three different ships. The three deadweight values plotted versus
length gives the possibility to identify the required ship length for the deadweight
specified at the beginning of the design. This procedure used historical data. Thus it is
possible that new designs generated, using this data, may be influenced by both relevant
and inappropriate previous designs. Another disadvantage of the method lies in the fact
that the use of historical data limits the capability of the designer to create novel
designs, as the new designs will have features similar to those of the database.
Over the period 1965-1985 several attempts were undertaken to solve the problem of
identifying ship sizing parameters by computer algorithms. They were developed by
Murphy et al. (1965), Mandel & Leopold (1966), Gilfillan (1969), Nowacki et al.
(1970), Fisher (1972) and Lyon & Mistree (1985). There were two basic optimisation
methods used in these cited papers.
The first approach is of Murphy et al. (1965) and Gilfillan (1969). In the work of
Murphy et al. (1965) the ship dimensions affecting the size and cost of the ship are
varied over a finite range of step sizes. The size of this multi-dimensional method is
determined by the number of variables together with the step size and the permissible
range of each variable chosen. Whilst the design charts readily allow identification of
suitable hull form parameters the minimum cost of building and operating such a ship
for a year is less readily identified. To seek minimum cost a great deal of graphical data
has to be systematically searched. This approach is thus relatively cumbersome, since
data has to be graphically manipulated before the desired result is obtained. Gilfillan
(1969) used a procedure in which the vessel length is increased in steps until the
deadweight satisfies the owners requirements, while all other design variables (B, CB
15
and T) were expressed as functions of ship length. Then all calculations regarding the
power, weight, stability and cost were carried out by the computer code for each trial
length. Again the desired design was not available directly, but it was selected by
examination of the output of the computer code.
However, in the second method mathematical programming is employed to achieve

optimum solutions. The final result arrived at the computer algorithm is the desired
result sought by the designer, whereas in the first approach the output from graphical
manipulation or the computer code does not lead directly to an optimum. There is
always some further refinement of the process. In the computer oriented optimisation
techniques either direct or random search techniques (see, for example, Fisher (1972)
and Mandel & Leopold (1966)) or nonlinear programming (see, for example, Nowacki
et al. (1970) and Lyon & Mistree (1985)) are applied.
The preliminary ship design stage has frequently been viewed as an economic
optimisation problem with the physical, technical and legal aspects treated as
restrictions or constraints. Consequently, the design process becomes a multi-criteria
optimisation problem as demonstrated by Sen (1992) and Ravn (2002). In the earlier
cited papers, Murphy et al. (1965) used only the ship construction cost in their
economic criteria; whereas Fisher et al. (1972) considers it better to render some
approximation to all factors rather than disregard some of them entirely. Therefore
Fisher et al. (1972) include impact of taxes, bank interest and borrowed capital to
owners capital ratio. The optimisation problem is that of seeking minimum cost of each
ton of cargo each year as a function of fleet size.
Liu et al. (1981) stresses the advantage of applying optimisation tools in the preliminary
design stage by two facts. Firstly, use of these tools in the early stages of design offers
the designer large potential savings in initial ship structural cost. Secondly, early
analysis of the design parameters improves the quality of the detailed design depending
on the inputs of the preliminary design.
16
An investigation of the seakeeping performance of the British Ship Research

Association merchant series is carried out by Wilson (1986). By his method it is
possible to consider large number of different designs at early design stage at a small
cost penalty which is only a function of the speed of the processor in the computer used.
The need of a philosophy, which uses recent advances in computer graphics to
understand the nature of the design process and to create radically new ship design
synthesis, is extensively discussed in Andrews (1981, 1986 and 2003).
Different optimisation techniques (random and direct) are compared by Keane et al.
(1991) for the minimum resistance of a frigate of 3300 tonnes displacement. Such an
optimisation process requires generation of the mathematical hull forms see Keane
(1988).
Doctors & Day (1995) introduced the genetic algorithm into their research to improve
catamaran ferries in terms of wave resistance so as to reduce the erosions of river banks
by wash. Later, their analysis was extended to include also vertical acceleration in head
seas (Day & Doctors (1997)). The wave resistance is calculated by either Holtrop &
Mennen in Keane et al. (1991) or Michell thin-ship theory in Doctors & Day (1995) and
Day & Doctors (1997). The first method has the advantage of covering a wide range of
ship types of varying sizes; whereas the thin-ship theory of Michell is more sensitive to
the change of hull form parameters, since it uses the actual hull form shape rather than
global hull form parameters to determine the wave resistance. The Michell thin-ship
theory has also been used extensively by Tuck & Lazauskas (1998) for predicting the
wave resistance for multi-hulls.
Work on resistance of high speed displacement catamarans has also been accomplished
by Insel & Molland (1992), Molland et al. (1996) and Molland & Lee (1997). Insel &
Molland (1992) presented the resistance experiments on NPL series of models with
changes in length displacement ratio. This work is extended by Molland et al. (1996) by
including beam draught ratio. A further extension is carried out by Molland & Lee
17
(1997) by investigating the influence of prismatic coefficient on catamaran resistance.

Whilst all these works have been performed in calm water, Molland et al. (2001)
examined the performance characteristics of catamarans in head and oblique waves.
Finally, Ghani (2003) investigated the influence of bulbous bows on the high speed
displacement catamaran performance in shallow water condition.
Keane & Robinson (1999) suggested that research in the conceptual design of ships
should not focus only on dealing with different hull form innovation, but also should
seek to find better search techniques in optimisation. Different optimisation techniques
in the sense of Keane & Robinson (1999) are provided by Schneekluth & Bertram
(1998) and Holden et al. (2002).
In automatic optimisation designer interaction is not needed (see, for example, Janson &
Larsson (1996)). However, such an approach risks the optimum design not being
practical. Therefore an interactive optimisation that does not leave the designer out of
the design process, but supports him with his decision making is more preferred.
However, the designer should be clear with respect to what his/her objectives are as the
computer program cannot automatically perform the optimisation without a clear
framework.
Concept exploration models (CEM) are an alternative to the automatic optimisation. In

this method a pool of candidate solutions is generated by varying design variables. Each
of these solutions is evaluated and the optimum is selected between them. An
application of the CEM for small warship design is provided by Eames & Drummond
(1977); whereas the development of a model for the conceptual exploration of
alternative high-speed ferry types is presented by Molland & Karayannis (1997). In the
optimisation process known as optimisation shells (see, for example, Schneekluth &
Bertram (1998)) the designer provides all necessary knowledge in the form of
relationships and then the shell checks if all the given relationships are (a) necessary and
(b) sufficient to solve the problem.
18
In the use of expert systems (see, for example, Dai et al. (1994) and Bertram (2003)) a
knowledge database is built up from the knowledge of experts within a particular
specialisation. In this technique the designer may specify the appropriate inputs but the
outputs are purely a function of the inbuilt expertise, there is usually little scope for the
designer to strongly influence the process outputs. This method has the disadvantage of
producing similar ships to those of an experienced designer. The need for a full
automatic or an interactive optimisation procedure is considered by Schneekluth &
Bertram (1998), Liu et al. (1981) and Keane & Robinson (1999). It is suggested that an
efficient optimisation technique coupled with the interaction of the ship designer is
more suitable for the optimisation process. In the papers of Hearn et al. (1990, 1991 and
1992) it is also assumed that constraints on parameter variation and objective function
context are the responsibility of the designer. That is, the final range of ship dimensions
should be selected by a naval architect who understands the relationships between
different design parameters and the information presented by optimisation tools.
Therefore, an optimisation procedure should not absolve the designer of his
responsibility, but assist him with his decisions.
Usually, the safety of a ship is regarded as a design constraint. Whilst it should be

fulfilled it does not directly drive the design optimisation process. Attempts to include
the safety of a ship in the optimisation process are achieved by Sen et al. (1997) and
Cramer & Tellkamp (2002 and 2003). Sen et al. (1997) investigated safety issues in the
context of lifeboat design. Different safety analysis methods are discussed prior to their
choice of Cause-Consequence analysis. This method considers every dangerous event
and then traces all possible earlier events that could lead to its occurrence. After that, it
examines all the consequences of the given event. Cramer & Tellkamp (2002 and 2003)
followed an equivalent safety approach and showed a qualitative comparison of
different ship designs with respect to their probability of survival in severe seas. While
their method is a step forward towards the introduction of safety into the optimisation
process, the approach does not include a mathematical optimisation technique to permit
19
variation of the design parameters systematically and this omits the evaluation of the
effect of changes of design parameters on the hydrodynamic responses.
Pawlowski et al. (2004) proposed a way to use the floodable length curve as a function
of s, conditional survival probability of a ship following the flooding of each
compartment in turn. This method presents a novel approach for effective subdivision at
the early stages of the design. However, their technique suffers from some numerical
instability that presents different solutions for the same initial conditions and case.
2.2 Discussion on Past Optimisation Studies

The discussed optimisation tools could be considered incomplete, from a safety point of
view, as they are only concerned about the intact stability of ships and they do not
address damage stability. To assess the significance, or otherwise, of this omission it
was decided to carry out the hydrodynamic analysis and hence dynamic motion
responses of hull forms in the intact and damage case for the ship as designed and
when optimised, to try and understand the influence of damage upon optimised hull
forms. In particular, the work carried out at the University of Newcastle has
demonstrated that for different ship types (trawlers, containerships, warships and
catamarans) it is possible to improve seakeeping, wave-making resistance and added
resistance whilst satisfying the IMO intact stability conditions imposed. Since no
apparent obvious disadvantage (in terms of the metrics used) exist with the intact ship it
might be the case (however perverse) that the disadvantages show themselves when
studying the damaged ship.
To produce alternative hull forms from a basis hull form to facilitate investigation of the
influence of damage on the hydrodynamic and motion characteristics of the optimised
and non-optimised hull forms the Hooke-Jeeves method is used. This method is selected
as it is fast at finding an optimum value of the selected objective function see, for
20
example, Keane et al. (1991). In the next section the optimisation method applied in this
research programme will be described in greater detail.
2.3 Statement of the Optimisation Process

A two-dimensional hydrodynamic analysis (strip theory) within the optimisation
process provides the opportunity to analyse the effect of several hull form parameter
changes on seakeeping within a very short amount of time. However, the physics of the
damage cannot be captured well within such a two-dimensional approach. Hence, the
possibility of developing a modified two-dimensional optimisation process, in which
damage stability is simultaneously investigated with seakeeping and resistance, was
rejected. Inclusion of full three-dimensional hydrodynamic analysis within the
optimisation procedure is not considered a practical option because of the computational
designs of three dimensional fluid-structure interaction analyses. The discretisation of
the original structure and its variants would require a very robust automated process.
This process would require a large amount of cross checking that the representation of
the current ship and the variant ship were consistently well modelled in terms of
geometry and discretisation, as this influences the quality of fluid-structure interaction
results (equality of cross-terms, well conditioned influence matrices and satisfaction of
Haskind equality conditions). Whilst this might not involve a large amount of
processing unit time the development of the associated logic could prove to be an
unwanted digression. Any evaluations included in the optimisation process must be
reliably undertaken for the thousands of alternative hull forms involved in the search for
an optimal hull form. Clearly, a compromise is required. It was therefore decided to
optimise a selected hull form using the two-dimensional hydrodynamic analysis based
optimisation process and then to create the extent of damage in the original and in the
optimised hull forms and reanalyse as three dimensional structures.
21
The Optistanbul Suite (see Sarioz (1993)) is used in this study in order to optimise the
selected basis hull form (see Section 6.2). This software represents an extension of the
method originally developed at the University of Newcastle upon Tyne. As before, a
database of hydrodynamic coefficients (added mass and fluid damping) for a suitable
representative set of transverse sections is used for the hydrodynamic analysis. The
optimisation of a hull form is provided using the Hooke-Jeeves algorithm. The primary
parameters within the optimisation process are modified by using linear distortion
methods and the secondary parameters are varied by using Lackenby transformation to
change the sectional area curve or the waterline curve. The generation of alternative hull
forms is produced by linear distortion or, in the case of Sarioz (1993) by the Lackenby
transformation. Having edited the software to provide some of the omitted main and
print algorithms, optimisation of the basis hull form was undertaken for an objective
function based on minimisation of the peak relative bow motion (RBM) in head seas
subject to the constraint that calm water resistance is not increased. RBM is selected as
an objective function since it includes the amplitude and phase information of both the
heave and pitch motions, which in turn are coupled to surge for the intact ship and
coupled to all other motions in the case of a damaged ship.
2.4 Summary
Having reviewed optimisation process in a specific hydrodynamic/motion context and
also reviewed more general aspects of optimisation, it is noted that safety aspects rarely
provide the driver of the process, but may occur in the context of a constraint.
Therefore, a way ahead has been suggested that permits a suitable combination of twodimensional
hydrodynamic
optimisation
and
three-dimensional
post
damage
hydrodynamic analysis.
Prior to present such a procedure one must examine damage statistics and identify a
representative set of damaged conditions. This aspect is addressed in the next chapter.
22
3.
ANALYSIS OF DAMAGE DATA
In this chapter the damage statistics are investigated in order to provide a general
appreciation of those accidents that have led to capsize or to the loss of a ship. These
damage statistics are prepared by the Marine Accident Investigation Branch (MAIB).
When analysing the damage statistics, this organisation treats fishing boats and
merchant vessels as two distinct industries. Hence, these two vessel types will be
examined separately. In addition to the damage statistics prepared by MAIB, the
Lutzen 2002 damage database is analysed to provide a general understanding of the
damage location and the extent of the damage. These studies will influence the selection
of the ship to be analysed and the location and extent of damage to be investigated on
such a ship.
3.1 Damage Statistics

The MAIB investigates marine accidents involving UK registered ships in world-wide
waters and non-UK registered ships in UK territorial waters. The gathered information
is published as a collection of short reports and accidents statistics only for the UK
registered ships. The statistics extracted are presented here as pie-charts for merchant
vessels in Figures 3.1(a), 3.1(b) & 3.1(c) and for fishing boats in Figures 3.2(a), 3.2(b)
& 3.2(c) respectively.
The types of accidents identified in the merchant ship pie-charts (Figures 3.1(a), 3.1(b)
and 3.1(c)) are: foundering/flooding, grounding, collision/contact, fire/explosion,
capsizing/listing, heavy weather damage, machinery damage, etc.
23
OTHER
4.2 %
HEAVY
WEATHER
DAMAGE
4.9 %
FOUNDERING /
FLOODING
2.9 %
GROUNDING
14.7 %
MACHINERY
DAMAGE
23.4 %
FIRE /
EXPLOSION
9.8 %
CAPSIZING /
LISTING
0.5 %
COLLISION /
CONTACT
39.6 %
(a) 1992-2000 (MAIB (2000))
FOUNDERING /
FLOODING
1.6 %
OTHER
31.9 %
HEAVY
WEATHER
DAMAGE
2.1 %
MACHINERY
DAMAGE
23.5 %
GROUNDING
7.9 %
COLLISION /
CONTACT
21.5 %
CAPSIZING /
LISTING
0.4 %
FIRE /
EXPLOSION
11.1 %
(b) 1994-2001 (MAIB (2001))
24
CAPSIZING/
LISTING COLLISION /
CONTACT
0.4 %
OTHER
22.3 %
FIRE /
30.3 %
EXPLOSION
11.7 %
MACHINERY
DAMAGE
23.5 %
HEAVY
WEATHER
DAMAGE
2.2 %
FOUNDERING /
FLOODING
1.5 %
GROUNDING
8.1 %
(c) 1994-2002 (MAIB (2002))

registered merchant vessels.
Figure 3.1(a) indicates that the most common types of merchant vessel accidents are
collision/contact, machinery damage and grounding (in that order). On the other hand,
according to Figures 3.1(b) and 3.1(c) merchant vessels experience a higher percentage
of machinery damage than collision/contact damage. Machinery damage and
capsizing/listing provide the same overall percentage (23.9%) in the accident statistics
belonging to the three different periods; even though contributions from other types of
accident are quite distinct in Figures 3.1(a) and 3.1(b) or 3.1(c). There is generally a
decrease in the occurrence of the accidents in the 1994-2001 with 1623 accidents and
1994-2002 with 1745 accidents compared to those of 1992-2000 with 1042 accidents.
The increase in the occurrence of accident statistics belonging to different years takes
place in the fire/explosion and other categories of the damage statistic pie-charts.
The increase in the other category (4.2% to 31.9% to 30.3%) of the pie-charts and the
decreases in the remaining classification of accidents, excluding the fire/explosion
category, occurs due to the 1999 changes in accident reporting and investigation
regulations (see MAIB (1999)). In particular, these new regulations changed what were
25
previously known as dangerous occurrences to accidents. These new types of

accidents, which are included in the other category of the pie-chart, have decreased
percentages associated with other categories.
MISSING
VESSELS
0.1 %
OTHER
0.5 %
HEAVY
WEATHER
DAMAGE
0.6 %
FOUNDERING /
FLOODING
14.1 %
GROUNDING
10.4 %
COLLISION /
CONTACT
6.1 %
MACHINERY
DAMAGE
63.6 %
FIRE /
EXPLOSION
3%
CAPSIZING /
LISTING
1.6 %
(a) 1995-2000 (MAIB (2000))
MISSING
VESSELS
0.1 %
OTHER
2.7 %
HEAVY
WEATHER
DAMAGE
0.6 %
MACHINERY
DAMAGE
61.3 %
FIRE /
EXPLOSION
3.5 %
FOUNDERING /
FLOODING
14 %
GROUNDING
10 %
COLLISION /
CONTACT
6%
CAPSIZING /
LISTING
1.8 %
(b) 1994-2001 (MAIB (2001))
26
MISSING
VESSELS
0.1 %
HEAVY
WEATHER
DAMAGE
0.6 %
MACHINERY
DAMAGE
61.4 %
OTHER
2.6 %
FOUNDERING /
FLOODING
14.1 %
GROUNDING
9.9 %
COLLISION /
CONTACT
6%
FIRE /
EXPLOSION
3.5 %
CAPSIZING /
LISTING
1.8 %
(c) 1994-2002 (MAIB (2002))

registered fishing boats.
The damage statistics for fishing boat accidents (see Figures 3.2(a), 3.2.(b) and 3.2(c))
are classified according to: foundering/flooding, grounding, collision/contact,
fire/explosion, capsizing/listing, heavy weather damage, machinery damage, missing
vessels, etc. The accident statistics presented in Figure 3.2(a) are for the period 19952000 and not the period 1992-2000 used in Figure 3.1(a), simply because MAIB (2000)
fishing boat data was only available for this period.
The percentages of occurrences of the different categories of accident in Figures 3.2(a),

3.2(b) and 3.2(c) are almost invariant and certainly consistent over the different time
periods for each category. The only real difference arises in the other category of the
damage statistics. According to Figures 3.2(a), 3.2(b) and 3.2(c) 85.3% to 88.1% of
reported
fishing
boats
incidents
are
associated
with
machinery
damage,
foundering/flooding and grounding. Here foundering/flooding indicates water

accumulation in the holds because of heavy waves.
27
The significance of machinery damage versus collision/contact changes from a ratio of

approximately 23.5: 21.5 (or 23.4: 39.6 in Figure 3.1(a)) in representing 45% (or 63% in
Figure 3.1(a)) of the statistics in Figures 3.1(a) to 3.1(b) or 3.1(c) to representing a total
of 67.3% (or 69.7 in Figure 3.2(a)) of the statistics in a ratio of approximate in 61.3 to 6
(or 63.6 to 6.1 in Figure 3.2(a)) in Figures 3.2(a), 3.2(b) and 3.2(c). Within this research
programme, damage means the hull form damage. Hence, the machinery damage and
foundering/flooding are not considered in this study, as well as those denoted as other
in the pie-charts due to lack of information.
The MAIB statistics reviewed suggest that from a structural damage perspective fishing
boats are not the primary concern regarding ship type selection. For fishing boats
machinery damage is particularly significant and therefore fishing boats will not be
researched further in this thesis. Merchant vessels are more likely to benefit from an
analysis of their responses and the influence of damage on their responses.
Ideally the MAIB accident statistics would also indicate the extent of damage, areas
most vulnerable to damage along the length of the ship, the height of damage and the
position of damage. Since the MAIB statistics analysed do not give such guidance,
another damage database designated Lutzen 2002 is investigated.
The damage database Lutzen 2002 is a large collection of accident data. It includes
2946 reports addressing seven different categories of accident for different ship types
over the period 1935 and 1999. The sought damage related data has been extracted and
processed to allow creation of appropriate pie-charts and the formation of appropriate
conclusions.
Figures 3.3(a) and 3.3(b) indicate the relative percentages of each of nine ship types that
have been involved in accidents and the relative percentage occurrence of each of seven
accident categories.
28
RO-RO+
FERRY+ CAR
FERRY
3.7 %
PASSENGER+
PASS/CARGO
3.4 %
OTHER
1.7 %
FISHING
VESSEL
2.3 %
TANKER
22.2 %
CONTAINER
SHIP
5.8 %
SERVICE
SHIP+
SPECIALIZED
3.4 %
BULK
CARRIER
5.7 %
GENERAL
CARGO
51.8 %
(a) Ship types
HEAVY
WEATHER
DAMAGE
0.3 %
OTHER
4.7 %
LOSS
0.2 %
CAPSIZING
0.1 %
COLLISION /
CONTACT
62.9 %
GROUNDING
31.6 %
FIRE /
EXPLOSION
0.2 %
(b) Different casualties
Figure 3.3: Damage database (Lutzen (2002)).
Figure 3.3(a) indicates that general cargo ships, tankers, containerships and bulk carriers
account for approximately 85.5% of all accidents reported.
29
Figure 3.3(b) indicates that collision/contact and grounding in the approximate ratio of
2:1 account for 94.5% of all accidents. The same conclusions are also withdrawn by
Fach (2004) as a result of the analysis of structural damages on high speed crafts. Here
grounding also includes collision to a rock at the bottom of the sea.
There are differences in the percentages of accident types in Figures 3.1(a), 3.1(b),
3.1(c) and Figure 3.3(b). This is due to the fact that MAIB accidents statistics present
accidents that involved UK registered ships in world-wide waters and non-UK
registered ships in UK territorial waters during the periods 1992-2000, 1994-2001 and
1994-2002, whereas the damage statistics included in Figure 3.3(b) are investigated by
Det Norske Veritas, Lloyds Register of Ship Repair Statistics, Hellenic Register of
Shipping, DSRK (former East German authorities), IMO and Germanisher Lloyd over
the significantly longer period of 1935-1999. To identify which class of merchant ship
is more often the subject of collision/contact and grounding the statistics associated with
Figures 3.3(a) and 3.3(b) were further analysed to produce the pie-charts of Figures
3.4(a) and 3.4(b).
30
TANKER
21.7 %
SERVICE
SHIP +
SPECIALISED
3.2 %
RORO+FERRY+
CARFERRY
3.6 %
PASSENGER+
PASS/CARGO
3.5 %
BULK
CARRIER
3.9 %
CONTAINER
SHIP
7%
FISHING
VESSEL
2.5 %
GENERAL
CARGO
52.6 %
OTHER
2%
(a) Collision/contact
PASSENGER+
PASS/CARGO
3.3 %
RORO+FERRY+
CARFERRY
4.3 %
SERVICE
SHIP+
SPECIALIZED
4.1 %
TANKER
22.7 %
BULK
CARRIER
6.2 %
OTHER
0.9 %
GENERAL
CARGO
52.3 %
CONTAINER
SHIP
FISHING
4%
VESSEL
2.2 %
(b) Grounding
Figure 3.4: Two main accidents from damage database (Lutzen (2002)).
According to Figure 3.4(a) and Figure 3.4(b) general cargo ships and tankers account
for 74.3% to 75% of the collision/contact and grounding accidents with other ship types
such as containership, bulk carrier and ferries accounting for single figure amount.
31
The high percentage of accidents associated with general cargo ships is discussed by
Spouge (2003). Spouge (2003) tried to identify the cause of this problem. He analysed
several factors such as ship age, ship size, flag of registration, ship quality, ship design,
domestic operations, coastal operations and classification society approval. Finally, he
concluded that high loss rate on general cargo ships might be attributed to either poor
quality of ship operation or this ship type might be more vulnerable to flooding.
Having identified the types of merchant ships that collide or experience grounding most
often, data related to damage location is determined next by further analysis of the
Lutzen 2002 source.
For collision incidents the damage location is examined for general cargo ships
(52.6%), tankers (21.7%) and containerships (7%); whereas for grounding the damage
location is considered for general cargo ships (52.3%), tankers (22.7%) and bulk
carriers (6.2%). In order to specify the location of the damage, the ship hull is divided
longitudinally into six regions. These regions are shown in Figure 3.5. The longitudinal
location of damage for general cargo ships, tankers, containerships and bulk carriers for
collision and grounding problems are presented in pie-chart form in Figures 3.6(a),
3.6(b), 3.6(c) and Figures 3.7(a), 3.7(b) and 3.7(c) respectively.
Stern AP-L/4
L/4-Amidships Amidships-3L/4
AP
3L/4-FP Stem
FP
Figure 3.5: The representation of the location of the damage.
32
STERN
1.8 %
AP-L/4
5%
UNDETERMINED
45.9 %
L/4AMIDSHIPS
11.1 %
AMIDSHIPS3L/4
9.8 %
3L/4-FP
21.7 %
STEM
4.7 %
(a) General cargo ships
STERN
1.5 %
UNDETERMINED
34 %
AP-L/4
9.7 %
L/4AMIDSHIPS
13.2 %
AMIDSHIPS3L/4
10 %
STEM
3%
3L/4-FP
28.6 %
(b) Tankers
33
STERN
6.2 %
AP-L/4
7.8 %
L/4AMIDSHIPS
8.5 %
AMIDSHIPS3L/4
0.8 %
UNDETERMINED
61.2 %
3L/4-FP
8.5 %
STEM
7%
(c) Containerships
Figure 3.6: Location of damage in collision incidents for different ship types.
34
AP-L/4
2.1 %
STERN
0.2 %
L/4AMIDSHIPS
7.2 %
AMIDSHIPS3L/4
9.7 %
3L/4-FP
8.6 %
UNDETERMINED
71.6 %
STEM
0.6 %
(a) General cargo ships
AP-L/4
4.3 %
L/4AMIDSHIPS
6.1 %
AMIDSHIPS3L/4
11.8 %
UNDETERMINED
49.8 %
3L/4-FP
25.6 %
STEM
2.4 %
(b) Tankers
35
AP-L/4
5.2 %
UNDETERMINED
55.2 %
L/4AMIDSHIPS
10.3 %
AMIDSHIPS3L/4
13.8 %
3L/4-FP
15.5 %
(c) Bulk carriers
Figure 3.7: Location of damage in grounding incidents for different ship types.
Figures 3.6(a), 3.6(b) & 3.6(c) and Figures 3.7(a), 3.7(b) & 3.7(c) indicate that the most
vulnerable part of the ship in a collision scenario is the 3L/4-FP region for general cargo
ships and tankers and L/4-amidships and 3L/4-FP regions for containerships. In a
grounding scenario, when information is reported, mostly likely location of damage is
in the amidships-3L/4 region for general cargo ships and then 3L/4-FP region for
tankers and bulk carriers.
Figures 3.6(a), 3.6(b) and 3.6(c) show that data is available for 54.1% of general cargo
ships, 66% of tankers and 38.8% of containerships for collision incidents. For
grounding incidents data is reported for 28.4% of general cargo ships, 50.2% of tankers
and 44.8% of bulk carriers. These different percentages of known data result in different
degrees of confidence in each ship type for each accident classification.
Having indicated the most likely location of collision and grounding damage the next
task is to present the extent of the data. The damage data available within the Lutzen
2002 database will be extracted and presented in non-dimensional form. In particular,
36
the longitudinal extent of the damage l will be sealed using ship length between
perpendiculars L, the penetration of the damage b is sealed by ship beam B, the
vertical position of the lowest point of damage Y (measured from the baseline of the
ship) and the height of the damage h are both sealed by ship depth D.
The analysed data is now presented in Figures 3.8 to 3.14. In each figure there are five
pie-charts designated (a) to (e) to denote non-dimensional extent of damage l/L, height
of damage h/D, damage penetration b/B, vertical position of the lowest point of the
damage Y/D and damage location in terms of port and starboard. Figures 3.8 & 3.9
provides collision damage details for the longitudinal position designated 3L/4-FP in
Figure 3.5 for general cargo ships and tankers respectively. This is the most likely
location for these ships as 3.6(a) & (b) indicated. Figures 3.10 & 3.11 provides the
corresponding collision data for containerships at two equally likely longitudinal
position designated L/4-amidships and 3L/4-FP in Figure 3.5 and confirmed in Figure
3.6(c).
Figures 3.12, 3.13 & 3.14 provide the corresponding data for grounding damage in the
most likely regions amidships-3L/4 for general cargo ships and 3L/4-FP for tankers and
bulk carriers respectively as indicated in Figure 3.7.
37
UNDETERMINED
19.4 %
0 < l / L 0.02
14.7 %
l / L > 0.3
0.5 %
0.02 < l / L
0.04
12.8 %
0.1 < l / L 0.3

12.3 %
0.08 < l / L 0.1
11.9 %
0.06 < l / L
0.08
14.2 %
0.04 < l / L
0.06
14.2 %
(a) The extent of the damage l/L
0 < h / D 0.2
12.8 %
UNDETERMINED
27 %
0.2 < h / D 0.4

6.2 %
0.4 < h / D 0.6
8%
h / D > 1.4
4.3 %
0.6 < h / D 0.8

14.7 %
1.2 < h / D 1.4

7.6 %
1 < h / D 1.2
6.6 %
0.8 < h / D 1
12.8 %
(b) The height of the damage h/D
38
0 < b / B 0.2
24.2 %
0.2 < b / B 0.4

10.4 %
UNDETERMINED
58.7 %
0.4 < b / B 0.6

5.7 %
0.6 < b / B 0.8
0.5 %
0.8 < b / B 1
0.5 %
(c) The penetration of the damage b/B
UNDETERMINED
8%
0 < Y / D 0.2
15.2 %
0.2 < Y / D 0.4

13.3 %
Y/D>1
34.6 %
0.8 < Y / D 1
6.6 %
0.4 < Y / D 0.6

15.2 %
0.6 < Y / D 0.8
7.1 %
(d) The vertical position of the damage Y/D
39
PORT
11.8 %
PORT &
STARBOARD
4.3 %
STARBOARD
7.1 %
UNDETERMINED
76.8 %
(e) The location of the damage
Figure 3.8: Damage properties in collision incidents for general cargo ships for the
3L/4-FP region.
40
UNDETERMINED
15.7 %
0.1 < l / L 0.3
6%
0 < l / L 0.02
15.7 %
0.08 < l / L 0.1

7%
0.02 < l / L
0.04
29.6 %
0.06 < l / L
0.08
13 %
0.04 < l / L
0.06
13 %
0 < h / D 0.2
14.8 %
UNDETERMINED
26.1 %
0.2 < h / D 0.4

11.3 %
0.4 < h / D 0.6
4.4 %
h/D>1
11.3 %
0.6 < h / D 0.8

10.4 %
0.8 < h / D 1
21.7 %
41
0 < b / B 0.2
29.6 %
UNDETERMINED
51.3 %
0.2 < b / B 0.4

7.8 %
0.8 < b / B 1
1.7 %
0.4 < b / B 0.6

9.6 %
0 < Y / D 0.2
16.5 %
UNDETERMINED
33.9 %
0.2 < Y / D 0.4

11.3 %
Y/D>1
5.2 %
0.8 < Y / D 1
9.6 %
0.4 < Y / D 0.6

15.7 %
0.6 < Y / D 0.8
7.8 %
42
PORT
1.7 %
STARBOARD
2.6 %
UNDETERMINED
95.7 %
Figure 3.9: Damage properties in collision incidents for tankers for the 3L/4-FP region.
43
UNDETERMINED
9.1 %
0.1 < l / L 0.2

27.3 %
0 < l / L 0.1
63.6 %
0 < h / D 0.2
27.3 %
UNDETERMINED
36.4 %
0.8 < h / D 1
9%
0.2 < h / D 0.4

27.3 %
44
0 < b / B 0.01
27.3 %
UNDETERMINED
63.6 %
b / B > 0.01
9.1 %
0 < Y / D 0.2
18.2 %
UNDETERMINED
27.3 %
0.6 < Y / D 0.8

18.2 %
0.8 < Y / D 1
36.3 %
45
STARBOARD
27.3 %
PORT
72.7 %
Figure 3.10: Damage properties in collision incidents for containerships for the L/4amidships region.
46
l / L > 0.1
9.1 %
0 < l / L 0.1
90.9 %
UNDETERMINED
18.2 %
0 < h / D 0.2
36.4 %
h/D>1
18.2 %
0.6 < h / D 0.8
9%
0.2 < h / D 0.4

18.2 %
47
0 < b / B 0.01
18.2 %
UNDETERMINED
81.8 %
0 < Y / D 0.2
27.3 %
0.8 < Y / D 1
45.5 %
0.2 < Y / D 0.4

9%
0.6 < Y / D 0.8
18.2 %
48
UNDETERMINED
9.1 %
PORT &
STARBOARD
9.1 %
PORT
36.3 %
STARBOARD
45.5 %
Figure 3.11: Damage properties in collision incidents for containerships for the 3L/4-FP
region.
49
UNDETERMINED
19.2 %
0 < l / L 0.2
31.9 %
0.8 < l / L 1
8.5 %
0.6 < l / L 0.8

10.6 %
0.2 < l / L 0.4

17 %
0.4 < l / L 0.6

12.8 %
0 < h / D 0.2
23.4 %
0.2 < h / D 0.4
8.5 %
0.4 < h / D 0.6
2.1 %
UNDETERMINED
61.8 %
h/D>1
2.1 %
0.8 < h / D 1
2.1 %
50
0 < b / B 0.2
25.5 %
0.4 < b / B 0.6

2.1 %
0.8 < b / B 1
4.3 %
UNDETERMINED
68.1 %
0 < Y / D 0.2
31.9 %
UNDETERMINED
63.8 %
0.2 < Y / D 0.4

4.3 %
51
PORT
6.4 %
STARBOARD
4.3 %
PORT &
STARBOARD
17 %
UNDETERMINED
72.3 %
Figure 3.12: Damage properties in grounding incidents for general cargo ships for the
amidships-3L/4 region.
52
0.3 < l / L 0.4

5.6 %
l / L > 0.4
1.9 %
0.2 < l / L 0.3

3.7 %
UNDETERMINED
5.4%
0 < l / L 0.02
20.4 %
0.02 < l / L
0.04
13 %
0.1 < l / L 0.2

25.9 %
0.04 < l / L
0.06
7.4 %
0.06 < l / L
0.08 < l / L 0.1

7.4 %
0.08
9.3 %
UNDETERMINED
14.8 %
0 < h / D 0.2
42.6 %
h/D>1
9.3 %
0.8 < h / D 1
3.7 %
0.4 < h / D 0.6
7.4 %
0.2 < h / D 0.4
22.2 %
53
UNDETERMINED
24 %
0 < b / B 0.02
40.7 %
0.1 < b / B 0.2

13 %
0.08 < b / B 0.1

1.9 %
0.04 < b / B
0.06
5.6 %
0.02 < b / B
0.04
14.8 %

UNDETERMINED
0.6 < Y / D 0.8 16.7 %
3.7 %
0.4 < Y / D 0.6

1.9 %
0 < Y / D 0.2
74 %
0.2 < Y / D 0.4

3.7 %
54
STARBOARD
3.7 %
PORT &
STARBOARD
1.9 %
UNDETERMINED
94.4 %

Figure 3.13: Damage properties in grounding incidents for tankers for the 3L/4-FP
region.
55
0.1 < l / L 0.3

11.2 %
0.08 < l / L 0.1
11.1 %
0.06 < l / L
0.08
11.1 %
0.04 < l / L
0.06
22.2 %
0 < l / L 0.02
22.2 %
0.02 < l / L
0.04
22.2 %
UNDETERMINED
33.4 %
0 < h / D 0.2
33.3 %
0.2 < h / D 0.4

33.3 %
56
UNDETERMINED
33.3 %
0 < b / B 0.2
55.6 %
0.4 < b / B 0.6

11.1 %
0 < Y / D 0.2
22.2 %
0.2 < Y / D 0.4

11.1 %
UNDETERMINED
55.6 %
0.6 < Y / D 0.8

11.1 %
57
STARBOARD
22.2 %
UNDETERMINED
77.8 %
Figure 3.14: Damage properties in grounding incidents for bulk carriers for the 3L/4-FP
region.
Having provided the detailed results of the analysis of the Lutzen 2002 database the
next task is summarise the essential characteristics of the data captured in Figure 3.8
through 3.14 and to indicate (albeit simplistically) a measure of confidence in the
identified data characteristics.
58
Table 3.1: Damage dimensions.
Collision
(i) l/L0.1
Extension
(ii) l/L0.1
(iii) l/L0.1
(i) 0.6<h/D0.8
Height
(ii) 0.8<h/D1
(iii) h/D0.4 or h/D0.2
(i) b/B0.2
Penetration
(ii) b/B0.2
(iii) b/B0.01
(i) Y/D>1
Vertical Position (ii) Y/D0.2
(iii) 0.8<Y/D1
(i) Port
Port and Starboard (ii) Starboard
(iii) Port/starboard
Notations Associated (i) General cargo ship
with Statistics (ii) Tanker
(iii) Containership
Reported
Damage Dimensions
80.6%
84.3%
90.9% or 100%
73%
73.9%
63.6% or 81.8%
41.3%
48.7%
36.4% or 18.2%
92%
66.1%
72.7% or 100%
23.2%
4.3%
100% or 90.9%
Grounding
(i) l/L0.2
(ii) l/L0.1
(iii) l/L0.1
(i) h/D0.2
(ii) h/D0.2
(iii) h/D0.4
(i) b/B0.2
(ii) b/B0.02
(iii) b/B0.2
(i) Y/D0.2
(ii) Y/D0.2
(iii) Y/D0.2
(i) Port/starboard
(ii) Starboard
(iii) Starboard
(i) General cargo ship
(ii) Tanker
(iii) Bulk carrier
80.8%
94.6%
100%
38.2%
85.2%
66.6%
31.9%
76%
66.7%
36.2%
83.3%
44.4%
27.7%
5.6%
22.2%
In Table 3.1 limits are placed on each of the damage dimensions of extension, height,
penetration and vertical position for both collision data and grounding data. The
percentage of accidents providing data for each ship type are specified as a simply
measure of how confident or otherwise one may be in the data extracted. Since for
container collision data there were two equally likely longitudinal locations (as defined
in Figure 3.5) the first and second entries (when there is a difference) correspond to L/4amidships and 3L/4-FP respectively.
3.1.1 Summary of Damage Statistics
Relevant accident statistics for different ship types have been extracted and processed
from the records of the MAIB and the Classification Society contributions to the
Lutzen 2002 database. The extracted statistics indicate which ship type is most
vulnerable to different forms of accident and provided some indication of the likely
59
characteristics of the incurred damage. Some measure of confidence in the extracted

statistics is provided by noting the percentage of each accident type for which reported
details exist.
In the next chapter how accidents at sea have influenced the historical development of
various design operation related regulations and stability analysis itself will be
reviewed.
60
4.
THE
DEVELOPMENT
OF
SHIP
REGULATIONS
AND
STABILITY ANALYSIS
This chapter represents an overview of the most relevant rules, regulations and stability
methods applicable to practical ship design.
4.1 Development of Safety

Lessons learned from past accidents have helped designers to design safer ships. In the
past the new knowledge (or understanding) gained was manifested in the form of
recommendations governing appropriate proportions and ship hull dimensions that
would provide improved stability and seakeeping. However, whilst this unwritten
insight was passed from generation to generation it was insufficient to significantly
improve the safety of ships.
There was thus a necessity to develop regulations. In the second half of the eighteenth
century, the first attempt was made by Lloyds Register of Shipping. It required that the
magnitude of freeboard should be two to three inches per foot of the height of the hold
(see, for example, Kobylinski & Kastner (2003)). The British Merchant Shipping Act of
1854 (see, for example, Kobylinski & Kastner (2003)) demanded that passenger ships
have to be fitted with a collision bulkhead and two bulkheads around the machinery
space. Towards the end of nineteenth century, Samuel Plimsoll, a British lawyer,
succeeded in introducing legislation on freeboard requirements. According to this all
British ships had to have a load line mark indicating the deepest permissible draught.
A couple of years later, the Committee of the British Board of Trade proposed a twocompartment standard for passenger ships (see Section 4.2.1). These proposals were
never implemented, since they were considered unnecessarily severe by the maritime
world.
61
After the Titanic disaster of April 1912 with 1513 lives lost, international organisations
tried to improve maritime safety. The first international convention for the Safety of
Life at Sea (SOLAS) was held in 1914, but the recommendations agreed did not enter
into force because of the outbreak of the First World War.
In 1929, another international Safety of Life at Sea conference was convened and the
recommendations designated the Criterion of Service and the Subdivision Standard
(see, for example, Turan (1993) and Kristensen (2002)). These parameters express
respectively the extent to which a ship is passenger carrying and the permissible
separation of bulkheads. The parameters are discussed further and are defined in Section
4.2.1. The major thrust of the 1929 Convention was directed towards passenger ships.
Consequently ships engaged in the carriage of passengers were subjected to much
stricter regulations.
In the 1948 SOLAS conference additional important decisions were agreed (see, for
example, Turan (1993) and Kristensen (2002)). In particular, there was the introduction
of stability standards and guidance regarding permissible extent of damage. The damage
length was formally defined as ( 3% L + 3.05 ) m and the unlimited vertical extent of
damage from the tank top was introduced. The Margin Line, which is 76mm below
the upper side of the bulkhead deck, was introduced and defines the maximum level of
ship immersion in its final damaged condition. At the United Nations hosted 1948
Geneva conference the organisation currently known as the International Maritime
Organisation (IMO) was established. The original name, the Inter-Governmental
Maritime Consultative Organisation (IMCO), was changed to IMO in 1982.
After the capsizing of Andrea Doria in 1956 with 52 lives lost, a passenger liner built
under the 1948 convention, it became clear that the 1948 stability standard exhibited
some shortcomings in its practical application. Another convention was required to
identify additional necessary changes. In 1960, the SOLAS sub-committee on stability,
subdivision and load lines was charged with the task of investigating the stability and
62
subdivision of ships (see, for example, Kobylinski & Kastner (2003)). The subcommittee analysed ship accidents and the behaviour of ships when damage occurred
and proposed new ideas to deal with the subdivision of ships.
The 1960 convention decided that the damage could be unlimited vertically from the
baseline of the ship and recommended that IMO should develop intact stability
standards for passenger ships, cargo ships and fishing vessels (see, for example,
Francescutto (2004)).
These deterministic analyses on subdivision and stability of ships were updated in 1974
(see, for example, Turan (1993) and Kristensen (2002)) and designated the probabilistic
approach (IMO (1971)). The probabilistic approach (explained in detail in Section
4.2.2) was accepted as an alternative rather than a replacement of the existing
deterministic stability criteria because of their perceived complexity.
In connection with the development of the first probabilistic rules two sets of model
experiments were carried out to examine and systematically analyse for the first time
the actual capsize mechanisms for damaged ships. One was in the United Kingdom (see
Bird & Browne (1974)) and the other one was in the United States (see Middleton &
Numata (1970)). This was a very important step in developing international stability
standards. Furthermore, these investigations were consistent with the recommendation
for a scientific approach for the development of safety standards.
In 1966, the IMO conveyed the International Convention on Load Lines. This
convention recommended some limitations on the draught to which a ship may be
loaded in the form of freeboard requirements (see, for example, Friis et al. (2002)). In
1969, the International Convention on Tonnage Measurements of Ships (see, for
example, Friis et al. (2002)) was adopted. Whilst this does not influence stability
directly it represents an example of other ship design improvements being sought.
63
In 1967, the Torrey Canyon tanker disaster resulted in 120,000 tonnes of oil being split
into the sea between England and France killing most of the marine life. Consequently,
the important problem of pollution was added to the responsibilities of IMO. In 1973,
IMO introduced the International Convention for the Prevention of Pollution from
Ships. Later this document was modified by the Protocol of 1978 (MARPOL 73/78),
see, for example, IMO (1997).
Between 1978 and 1979 the Safeship project (see Bird & Morrall (1986)) was
formulated by the United Kingdom Intact Stability Working Group. The aim of this
project was to develop simple stability criteria applicable to passenger and cargo ships
less than 100m in length in the short term. The long term aim was to extend the existing
knowledge of large amplitude rolling motions and capsize mechanisms to establish
better design criteria and regulations.
Brook (1988) discussed the incompetence of the current stability criteria, since an
analysis of casualty statistics showed several vessels satisfying the existing criteria were
lost as a consequence of capsize. Brook (1988) emphasized the need to take into
account environmental effects when analysing the intact stability of a vessel.
In 1987, the Ro-Ro passenger car ferry Herald of Free Enterprise capsized and
foundered just outside Zeebrugge Harbour, Belgium, with the loss of 193 lives. This
ship had been designed according to the 1960 SOLAS convention. This accident
demonstrated that the existing stability criteria were still unsatisfactory and also
emphasized the necessity for more realistic rules to improve the safety of ships. After
the accident, new amendments expanded the existing approach by introducing three
changes namely; a value of 15 for a minimum range of positive residual righting arm,
a value of 0.015 m-rad. for the area under the righting curve of the ship in the final
damaged condition and a maximum GZ value of 0.1 m. These amendments, made in
1988 and known as SOLAS 90, increased the damage stability standards of all new
64
passenger ships. In 1992 similar requirements were applied to all existing passenger
ships.
In 1990, residual stability standards concerning righting levers, the area under the GZ
curve and limiting angles of inclination were introduced for both cargo and passenger
ships. These requirements came into force at the beginning of 1992 (see, for example,
Turan (1993), Turan & Vassalos (1994) and Kristensen (2002)).
The Ro-Ro passenger ferry Estonia sank in 1994 with 852 lives lost as a result of large
amounts of accumulated water on the cargo decks. This water collection reduced the
stability. Following the Estonia disaster, a panel of experts met in December 1994 to
identify ways of improving the maritime safety of Ro-Ro passenger ships. The
conference agreed that all existing Ro-Ro passenger ships must deal with the damage
stability requirements of SOLAS 90 (see, for example, Kristensen (2002)).
The tragic accidents of the Herald of Free Enterprise and Estonia emphasized the
magnitude of the problem presented when water enters the deck of ships with large
undivided spaces (such as Ro-Ro vessels). Therefore, the Joint Northwest European
project (see Vassalos et al. (1996)) was established to address the development and
validation of numerical tools for assessing the damage survivability of passenger/Ro-Ro
vessels. This research resulted in the development of the static equivalent method
(SEM) for Ro-Ro ships (see, for example, Pawlowski (1999) and Tagg & Tuzcu
(2003)). This method statically calculates the volume of water that will reduce the
damage GZ curve to exactly zero (see Tagg & Tuzcu (2003)). From this neutral stability
position, it follows that any lesser amounts of water will imply survivability of the ship
and further additions of water will cause ship capsizing.
The Regional Agreements Concerning Specific Stability Requirements for Ro-Ro

Passenger Ships (otherwise known as the Stockholm Agreement (see, for example,
Schroter & Juhl (2002)) is considered to provide an alternative to the deterministic
65
approach of stability analysis. This model test method requires that at least five
experiments for each peak period should be carried out and be documented by means of
a written report and a video recording of the experiments. The ship model should
exhibit the same outer and internal configuration as the original ship and it should be
placed in beam seas with the damage hole facing the oncoming waves. More details can
be found in IMO (1997).
The development of new regulations and the revision of SOLAS still continue. In recent
years much work is being undertaken within the umbrella of the Harder Project (see,
IMO (2002a, 2002b) and Rusas (2002)). The aim of the HArmonization of Rules and
DEsign Rationale (HARDER) project is to harmonise the stability demands on
passenger and cargo ships greater than 80m in length. The final outcome of the
HARDER project can be found in Tagg & Tuzcu (2003). The NEREUS project
commenced in 2000 has the aim of developing design tools and methodologies to
improve Ro-Ro damage resistance against capsize. Some of the validation work carried
out within NEREUS can be found in Woodburn et al. (2002).
Stability analysis has evolved over a considerable time with different accidents acting as
catalysts for further development. Whilst a probabilistic approach exists as an
alternative to the deterministic approach the designer may select the approach adopted.
In the next section both approaches are examined together with their advantages and
disadvantages.
4.2 The Deterministic and Probabilistic Approaches to Ship Stability

Analysis
The deterministic approach to ship stability is mandatory, whereas the probabilistic
method is accepted as an alternative approach when the deterministic method is found
to be unsatisfactory. Each approach reviewed in turn in the next sections.
66
4.2.1 The Deterministic Approach to Ship Stability Analysis
The deterministic approach offers the possibility to account for the floodable length and
for the disposition of transverse watertight bulkheads. The floodable length varies with
longitudinal position on the ship. It is the maximum length of a compartment, which if it
is flooded, will permit the vessel to float at a waterline that is below or touches the
Margin Line. The Margin Line is a fair curve drawn 76mm below the bulkhead deck,
which is the uppermost deck to which watertight bulkheads extend.
Ships are subdivided according to their Criterion of Service and their Factor of
Subdivision (or Subdivision Standard). The Criterion of Service is a numeral that
expresses the degree to which a ship is a passenger carrying ship. For example, a
numeral of 23 corresponds to a ship primarily engaged in carrying cargo with a small
number of passengers, whereas a numeral of 123 applies to a ship engaged, mainly, in
the carriage of passengers (see, for example, Lewis (1988)). This assignment of the
numeral depends on the ship length, the number of passengers, the total volume of the
ship below the margin line, the volume of the machinery space and the volume of the
accommodation spaces below the margin line. The lower the value of the numeral
assigned the further apart the watertight bulkheads may be spaced.
The Factor of Subdivision, designated F, depends on ship length and the designated
criterion of service. The Factor of Subdivision establishes the permissible length
between watertight bulkheads. It is expressed as a percentage varying from 30 to 100
percent. A factor of F=0.3 means the bulkheads may be spaced a separation distance
only equal to 30 percent of the floodable length. If the factor of subdivision is greater
than 0.5, the ship must satisfy the one compartment standard (any one compartment
can be flooded without the ship sinking), if F is between 0.33 and 0.5 then the vessel
should satisfy the two compartment standard and finally if F is smaller than 0.33 the
floating structure should meet the three compartment standard.
67
So when the floodable length at each location along the ship length is calculated (see,
for example, Lewis (1988)); the permissible compartment length at each point is
obtained from the multiplication of the floodable length and the factor of subdivision. In
this way, a permissible length curve along the ship is derived.
However, when a ship is damaged and some compartments become open to sea, the
seawater cannot fill such volumes totally as some space will already be occupied by
internal ship arrangements e.g. bulkheads, main engines, auxiliaries, pumps, cargo et
cetera. So before the floodable length can be calculated, definite values of the
permeabilities of the spaces involved must be determined. That is, the fraction of
floodable volume in a compartment. Thus the floodable length curve has to be
determined for different levels of permeabilities. Figure 4.1 indicates the influence of
permeability levels on the floodable length curve.
Once the floodable length curves have been determined the level of stability at the
appropriate level of impermeability can be examined. For each compartment a triangle
is constructed on the floodable length curve with the base line defined by the end points
of the compartment and the apex constructed from the intersection of the other two sides
each of which makes an angle = tan 1 ( 2 ) with the baseline, as illustrated in Figure
4.1. For an acceptable level of stability at the selected level of permeability the triangle
apex must lie below the corresponding floodable length curve. Conversely the locations
of transverse watertight bulkheads are determined by ensuing each constructed triangle
meets the requirement indicated.
68
50
Floodable
Length
40
30
= 0.63
20
10
= 0.85
0
0
AP
10
15
20
25
30
35
Longitudinal Position (m)
40
45
FP
50
Figure 4.1: The influence of permeability on the floodable length curve for a
cargo/passenger vessel of 134m in length.
More detailed information on the Floodable Length and Factor of Subdivision is

available in the SOLAS 60 related documentation (IMCO (1960)).
In classical naval architecture, there are two principal deterministic methods to examine
the stability of damaged ships: the Lost Buoyancy and the Added Weight techniques.
Therefore as the positions of bulkheads and compartments are assigned in the design
process one may use either the Lost Buoyancy or the Added Weight method to
understand the behaviour of the damaged ship. Each method is summarised next with
further discussions of the two procedures provided in Section 4.2.3.
Lost Buoyancy Method
The lost buoyancy method considers that the damaged compartments are open to the sea
and the ship has lost buoyancy in these compartments. In order to compensate for the
lost volume and its moments the vessel will undertake parallel sinkage, trim and heel.
69
The amount of parallel sinkage can be calculated from the division of the volume of the
flooded compartment (or in other words lost buoyancy) by the new waterplane area of
the damaged ship. The heel angle and trim are determined by equating the moments of
the lost buoyancy with respect to new LCF to the moments of the buoyancy gain
accompanying parallel sinkage, trim and heel attributed to the remaining non-damaged
compartments of the ship.
Added Weight Method
In the added weight method the mass of water that has entered the damaged
compartment(s) is treated as an additional weight on board. The additional weight
changes the displacement weight and the longitudinal, vertical and horizontal positions
of the centre of gravity. Hence this additional weight causes parallel sinkage, trim and
heel of the ship. The parallel sinkage is calculated from the division of the volume of
the water entrapped in the damaged compartment(s) by the waterplane area of the intact
ship. The heel angle and trim are obtained by equating the moments of the added weight
respect to LCF to the moments of the new displacement weight of the damaged ship.
Some further discussion about these two methods is presented in Section 4.2.3.
4.2.2 The Probabilistic Approach to Ship Stability Analysis
The probabilistic approach to stability analysis was initially developed in 1973. To

estimate the probabilities of different damage stability related events available accident
records are used. It is the known occurrence of such damage stability related events that
governs the concept of stability in this procedure.
In the probabilistic approach the survival probability of a damaged ship is defined

through the attained survival probability index, A . To determine this probability, one
70
must first estimate the conditional survival probabilities, pi , assigned to either the ith
damaged compartment or the ith specific grouping of adjacent damaged compartments.
This, in turn, requires estimation of the probability of survival, si , after flooding the ith
identified area or grouping of areas. Thus A is defined as
A = pi si .
(4.1)
i =1
Clearly i represents each compartment or group of compartments under consideration

and N is the total number of individual and groups of compartments considered.
pi for each single compartment considered is determined according to MCA (1999) as
follows,
if the compartment considered extends over LS which is the moulded length of

the ship
pi = 1
(4.2)
if the aft limit of the compartment considered coincides with the aft terminal of
LS
pi = F + 0.5ap + q
(4.3)
if the forward limit of the compartment considered coincides with the forward
terminal of LS
pi = 1 F + 0.5ap
71
(4.4)
if both ends of the compartment considered are within the aft and forward
terminals of LS
pi = ap .
(4.5)
The assumed distribution function of damage location along the ships length is defined
as
F = 0.4 + 0.25 E (1.2 + a ) .
(4.6)
The assumed distribution density of damage location along the ships length is
a = 1.2 + 0.8 E , which should be not more than 1.2
(4.7)
E = E1 + E2 1
(4.8)
E1 =
x1
Ls
(4.9)
E2 =
x2
Ls
(4.10)
Here x1 and x2 stand for the distance from the aft terminal of LS to the
foremost/aftermost portion of the aft/forward end of the compartment considered.
p = F1 J max
F1 = y 2
y3
1
if y <1, otherwise F1 = y
3
3
72
(4.11)
(4.12)
y=
(4.13)
J max
J = E2 E1
(4.14)
The maximum non-dimensional damage length J max is given by
J max =
48
, which should be not more than 0.24
LS
q = 0.4 F2 ( J max )
F2 =
(4.15)
(4.16)
y3 y 4
y3 y 1
if y <1, otherwise F2 =
+
3 12
2 3 12
(4.17)
More details and discussions about pi can be found in IMO (2002b).
si can be determined for each single compartment considered according to MCA (1999)
as follows,
si = C 0.5GZ max range
C = 1 if 25, C = 0 if >30, otherwise C =
(4.18)
30
.
5
(4.19)
Here GZ max stands for the maximum positive righting lever which should not be more
than 0.1m, range is range of positive righting levers beyond the angle of equilibrium
which should not be more than 20 and is the final equilibrium angle of heel.
73
The factor si can be calculated by using either a GZ based formulation (see Equation
(4.18)) or the static equivalent method (summarised in Section 4.1). According to the
static equivalent method, the factor si depends on the water head h on the vehicle
deck above the sea level, as shown in Figure 4.2, at the critical heel of equilibrium. The
critical volume of water is the volume of water necessary to reduce the GZ curve of the
ship in the damaged condition to neutral stability (see IMO (2002a)). Different
formulations for the determination of the factor si for Ro-Ro and other types of ships
are discussed by Vassalos et al. (1996), Pawlowski (1999) and IMO (2002a).
Figure 4.2: Definition of the water head, h (IMO (2002a)).
A second index of subdivision, is the required subdivision index designated either by

R or Amax . The calculation of the index, R , for a cargo ship (which by definition
includes tankers, general cargo ships, containerships and bulk carriers) of length greater
than 100m is given in Equation 4.20.
R = ( 0.002 + 0.0009 LS ) .
1/ 3
Equation (4.20) is different for cargo and passenger ships.
74
(4.20)
The degree of subdivision of a ship is considered sufficient if the stability of the ship in
a damaged condition meets the requirements of the SOLAS 90 criteria (see, for
example, Turan (1993)) and the attained survival probability index (A) is not less than
the required subdivision index ( R or Amax ).
4.2.3 Discussion on Current Stability Methods
Having outlined two possible distinct methods of determining of the stability of a

damaged ship, it is appropriate to discuss the marriage of the hull form optimisation
tools discussed in Chapter 2 and the stability analysis choices.
The probabilistic concept of survival, based on actual accident reports, the studying of
damage stability related random events and associated probabilities, seems more
realistic than the deterministic approach which is based on numerical evaluation of fixed
predefined parameters.
However, the aim of this research is to examine the capability of optimised and nonoptimised ships to withstand damage. Given the optimisation tools are to be applied in
the earlier stages of the design to produce new improved hull forms from an initial
design, the probabilistic approach is likely to be inappropriate when the initial design is
particularly innovative; since the probabilistic approach requires accident statistics of
existing operational ships. For less novel hull forms the designer always has the choice
of allowing the probabilistic approach to influence the initial design to be optimised.
Within this research programme the deterministic approach is therefore considered
sufficient to ensure that modifications of the basis hull during the optimisation are
constrained to satisfy the deterministic constraints with limited changes of hull form
design parameters.
75
The lost buoyancy approach appears quite realistic as the hydrostatic calculations are
carried out using the residual waterplane area, that is, the original intact waterplane
area minus the waterplane areas no longer contributing because of the ship damage.
However, the added weight method considers the weight of water entrapped in the
damaged compartments and so permits calculation of parallel sinkage either with heel or
trim changes. Consequently, both approaches are adopted in combination when damage
occurs below and at the still waterline.
4.3 Damage Stability Suite

Section 4.2.3 concluded that the probabilistic method could be beneficial in influencing
the initial design to be optimised when well established operational statistics existed, it
was not appropriate for novel designs. It has also been noted that the hull form
optimisation procedures were initially developed to allow better first level iterations of
design by companies without specific ship type design capability techniques.
The reported applications of such optimisation processes indicate that whereas intact
stability is used as a constraint and the hull forms generated exhibit improved overall
(frictional & wave-making) resistances, seakeeping and added resistance qualities
stability of damaged hull forms has not been addressed. Including damage stability
within the optimisation process was rejected on the basis that the structure is perceived
as a series of two-dimensional transverse sections and this approach does not readily
capture the three-dimensional aspects of damage stability (as discussed in Section 2.3).
Hence it has been decided that the deterministic damage stability method will be applied
when both the initial design and optimised design have been damaged in equivalent
ways. Thereafter, the three-dimensional nature of the seakeeping motions can also be
captured for both the initial and optimised hulls in their intact and damaged states.
76
4.3.1 Independent Damage Stability Analysis
Damage stability is an integral part of computerised stability analysis packages.

However, it was found to be more convenient and faster to use the alternative
equilibrium search code developed by Saydan (2004). The developed algorithm initially
determines the intact hydrostatic properties of the vessel and then undertakes a damage
stability analysis of the ship according to the specified extent of damage, see Section
6.6.
When a ship is damaged it may lose some of its structural mass. Assuming l , b and h
denote the length, penetration and height of the damage and is the associated
material density the structural mass loss is estimated using
wlost = l b h .
(4.21)
With structural damage there can be the accompanying water ingress and resulting
changes in trim and heel together with parallel sinkage. These aspects are
considered next.
The volume loss, vlost , within the damaged compartment is determined from knowledge
of:
the original compartment volume vc
the volume of compartment not affected by ingressed water (e.g. closed tanks)
vi
the total mass of liquid inside the compartment prior to damage occurring M l
77
the density of the fluid l and the permeability of the compartment c
That is
M
vlost = ( vc vi ) c l .
l
(4.22)
The parallel sinkage is estimated taking into account the vertical position and the height
of damage. If the damaged area straddles the intact still waterline (case A of Figure 4.3)
then the parallel sinkage is given by Equation (4.23a), whereas if the intact ship
waterplane area and its properties are conserved (case B of Figure 4.3), the parallel
sinkage is determined from Equation (4.23b);
YSINK =
vlost
Awps 2 awpc
(4.23a)
vlost
Awps
(4.23b)
YSINK =
Case A
Case B
Figure 4.3: Vertical position of damage.
78
In Equations (4.23a) & (4.23b) Awps denotes the waterplane area of the intact ship. In
Equation (4.23a) 2 represents the surface permeability of the damaged compartment
and awpc is the waterplane area of the damaged compartment.
From Euler angle descriptions of rotations one is aware that changing the order of
rotations from say followed by (about two different axes) to by leads to
different final positions of the structure. To determine the trim and heel angles of the
damaged ship the trim and heel angles are estimated in three related ways. In the first
option the hull form is trimmed first and then it is heeled, in the second option the hull
form is heeled first and then it is trimmed and in the third option the equilibrium angles
are determined by averaging the angles obtained from the first and second methods. The
acceptability of these averaged trim and heel angle is then interrogated to establish that
the implied damaged ship orientation is a position of equilibrium. If this is not the case
using the current predicted ship orientation the whole process is repeated until static
equilibrium is established to determine the heel and trim angles sought. As indicated
earlier (Section 4.2.3) the damaged ship equilibrium position is investigated using a
combination of the lost buoyancy and added weight methods (see, for example, Lewis
(1988)).
4.3.2 Presentation of the Damaged Hull Forms
Downward movement, trim and heel will change the location of the hull wetted surface
data points along the structure. These new positions must be determined to define the
wetted surface of the damaged hull form. The necessary transformation will be achieved
using vertical translation and rotation in accordance with appropriate Euler angles. The
transformation of the points defining the hull form with Euler angles are presented
below for a structure which trims first and then heels.
As a first step, the downward movement is applied to yield;

79
Z Z 0
X = X 0

Y1 Y YSINK
(4.24)
whereas the trim and heel will lead to;
Z1 cos
X = 0

Y2 sin
0 sin Z
1
0 X
0 cos Y1
(4.25)
and
0
Z1 1
X = 0 cos
1
Y3 0 sin
0 Z1
sin X
cos Y2
(4.26)
Here and denote the trim and heel angles and the parallel sinkage is denoted by
YSINK . The reference axis system is shown in Figure 5.1.
These new data values are used in the Matthew Diffraction Suite (see Hearn (1978)) to
evaluate the hydrodynamic coefficients and wave excitation forces of the damaged hull
forms. This data is presented in Chapter 7 and Appendix H.
4.4 Summary
Having indicated how ship regulations and stability analysis has changed to overcome
difficulties highlighted by various public attention noted accidents, a method of
investigating the impact of damage upon the initial and optimal identified hull forms has
80
been proffered. Since investigating the impact of damage includes reassessing the
motions, seakeeping aspects are discussed next.
81
5.
SEAKEEPING
Researchers, designers and operators are interested in the dynamic behaviour of ships in
waves because the seakeeping capability of a ship may determine the operational and
hence the financial success of the design.
This chapter starts with a general review of seakeeping of both intact and damaged
ships. Having formulated the general six degrees of freedom steady state equations of
motion of an intact ship a novel technique is presented for determining the crossproducts of inertia of a damaged ship.
5.1 General Review

Sea waves are irregular. Within a linear hydrodynamic theory a representation of
irregular waves can be achieved by assuming that sea waves are a consequence of the
linear superposition of progressing regular and harmonic waves of different wave
amplitude, circular frequency and relative phasing. Furthermore, if the characteristic
linear responses of a ship to progressing regular and harmonic wave components are
known, then the response of that ship in an irregular seaway can be determined. Spectral
analysis applied to linear dynamic systems provides the link between responses in
regular waves and responses in a seaway or irregular waves (see, for example, Lloyd
(1998) and Saydan (1999)).
The fluid-structure interaction associated with ship motions in regular waves can be
formulated using potential flow theory. In this case viscous effects are neglected, the
fluid is assumed incompressible and the fluid motion is assumed irrotational. With the
additional assumption of small unsteady motions of the ship and of the surrounding
fluid, linear superposition can be applied. Another consequence of linearisation is that
82
the wave structure interaction problem can be decomposed into diffraction and radiation
problems (see, for example, Newman (1978) and Lewis (1989)).
In the diffraction problem regular harmonic incident waves act upon the otherwise
fixed ship. Linearisation of the fluid-structure interaction problem means that the
resulting velocity potential consists of two components, the velocity potential of the
undisturbed incident wave system and the velocity potential representing the diffraction
or scattering of the incident waves by the fixed ship. The hydrodynamic forces
(moments) resulting from the incident and diffracted waves are called the excitation
forces (moments) (see, for example, Newman (1978) and Lewis (1989)).
The radiation forces and moments acting on the body are a consequence of the structure
being forced to oscillate in otherwise still calm water at a frequency corresponding to
each incident wave frequency (for zero forward speed), or, incident encounter frequency
(for forward speed case) in each of the six possible degrees of freedom in turn. This
approach is adopted because linearisation implies that the forces (moments) required to
induce motion in any degree of freedom are the same as those resulting from forced
oscillation at the same amplitude and frequency. Since the forces (moments) are
dependent upon the (as yet) unknown amplitudes of motion the radiation hydrodynamic
forces are recast into an equivalent two coefficient based expression. The two
coefficients, referred to as added mass and fluid damping, are essentially the force
(moment) in-phase with the acceleration and velocity of the structure respectively.
Hence both amplitude and phase information is encoded in these coefficients (see, for
example, Newman (1978) and Odabasi & Hearn (1978)).
The excitation forces (moments) and the radiation forces (moments) are determined
from integration of the associated dynamic pressures over the wetted surface of the
structure (see Section 7.1 for required integral relationships). Bernoullis equation links
pressure and the corresponding velocity potentials (see, for example Bertram (2000)). In
order to obtain the unknown diffraction and radiation velocity potentials, boundary
83
integral methods may be used (see, for example Odabasi & Hearn (1978)). Whether 3D
methods or 2D based strip theory are applied appropriate boundary conditions must be
satisfied. These boundary conditions (see Appendix B) are satisfied at the fluid and
body boundaries. The fluid boundaries are the free surface, the sea-bed and the so-called
radiation surface linking the free surface and the sea-bed in the form of a vertical
cylinder to form a closed solution domain. The body boundary is the wetted surface of
the ship or floating structure. Once the velocity potentials are evaluated, the
hydrodynamic pressure can be derived from the linearised form of the Bernoulli
pressure equation; the hydrodynamic forces can be determined from integration of the
incident and diffraction or the radiation pressure over the wetted surface of the floating
structure (as indicated earlier).
For a structure damaged at or below the waterline the boundary element methods may
be extended to include internal wetted surfaces. In this case the appropriate boundary
condition is one of a non-permeable structure, that is, normal velocity of the fluid
matches the normal component of the structural velocity consistent with the radiation or
diffraction problem being addressed. In the case of ingressed fluid there will be internal
free-surfaces and if necessary these may be modelled as structurally massless plates
with appropriate degrees of freedom (for equations of motions see Appendix K). Such
an approach is not implemented here.
If the attitude of the structure is such that the deck is exposed to waves then this surface
may also be included in a manner analogous to other rigid surfaces. That is, the water in
contact with the deck is not considered to be entrapped on deck but to satisfy the usual
continuity of normal velocity of fluid at deck.
For completeness, the problem of entrapped fluid is reviewed next.
Huang & Hsiung (1997) analysed the dynamic behaviour of the ships using a timedomain approach when a large amount of water is trapped on the deck due to spillage of
84
water or roll motion. Later Grochowalski et al. (1998) and Huang et al. (1998)
considered the same time-domain problem with the addition of water shipping on and
off the deck. The water on deck usually causes the progressive flooding of the spaces
inside the hull and this may ultimately result in the loss of the ship. This problem is
highlighted by Borlase (2003).
When the ship hull is damaged, water may flow into the damaged compartment(s). The
water trapped within damaged compartment(s) may change the behaviour of the ship.
Subramanian & Kastner (2000) examined the behaviour of a Ro-Ro ship in beam seas
when water enters on deck. The model vessel was analysed in different conditions: open
deck condition, a solid weight on deck (equivalent to the weight of 0.5 m depth of
water) and 0.58 m depth of water on deck. It was found that if water is entrapped on the
cargo deck it causes two superposed frequencies of oscillation: one is due to the
frequency of oscillation of the liquid mass inside the ship and the other is due to the
natural rolling frequency of the ship.
Turan (1993) and Turan & Vassalos (1994) carried out a number of different damage
scenarios to investigate the dynamic behaviour of a damaged car/passenger vessel using
a time simulation approach. Their mathematical model included coupled sway-heaveroll motions in beam seas. Their motion equations take into account the amount of water
entering and leaving a damaged compartment and also variations in the ship mass.
However, in their research there is no comparison between the computed and
experimentally measured results for these damage scenarios.
Chan et al. (2002) analysed motions of a Ro-Ro ship in stern quartering waves in intact
and damaged conditions using a time domain method. Their theoretical predictions and
experimental measurements showed good agreement except in the roll-resonant region.
The authors attributed this discrepancy to the strong coupling between all modes of
motion for large amplitude responses in the roll-resonant region.
85
Umeda et al. (2004) suggested that the centre of harmonic motion of a damaged ship
changes gradually with the increased accumulation of water inside the hull. Turan &
Vassalos (1994) calculated the radiation and diffraction forces of the damaged ship
without taking into account the change of the centre of harmonic motion. Four years
later, Vassalos et al. (1998) calculated radiation and diffraction forces taking into
account the change of centre of harmonic motion with the sinkage, heel and trim.
Traditionally, in the damaged ship dynamic calculations the free surface of the
floodwater within the flooded compartment(s) is assumed to be horizontal (see, for
example, IMO (1997), Vassalos et al. (1996) and Palazzi & Kat (2004)). In the
probabilistic A-265 criteria the effect of floodwater is based on empirical data derived
from model tests (see IMO (1971)). Woodburn et al. (2002) questioned these methods
as they do not necessarily represent a realistic scenario when the ship undergoes large
amplitudes of motion. Woodburn et al. (2002) used in their research a coupled model
that consists of a dynamic model for the ship motions and a computational fluid
dynamics (CFD) model for floodwater dynamics. Their computed results show
satisfactory correlation with experimentally measured data for the mass of floodwater
inside the vessel.
For the examination of the behaviour of a wall-sided, damaged Ro-Ro passenger ship in
beam waves Hasegawa et al. (2000) assumed that water ingress and water egress were
functions of water level inside and outside of the damaged compartment. Veer & Kat
(2000) examined theoretically and experimentally the progressive flooding and sloshing
in compartments. Large flow obstructions such as main engine were presented in the
geometry of the model and small flow obstructions were modelled by changing only the
permeability of the compartment.
Ikeda & Ma (2000) carried out an experimental study on the large dynamic roll motion
of a passenger ferry at intermediate stages of flooding due to sudden ingress of water
through a damage opening. It was realised that the intact stability of the vessel, the
86
arrangement of obstacles, the damaged area and the location of the damage opening
affected the roll motion characteristics at intermediate stages.
If as a result of the frequency-domain analysis of the behaviour of the damaged and

intact ship, the impact of the damage on the responses of the ship is found to be
significant then these effects may have to be taken into account in a refined analysis.
With this possibility in mind the problem of sloshing is reviewed next.
Faltinsen (1978) considered two-dimensional analysis of a rectangular tank forced to

oscillate harmonically with small amplitudes of sway. Later, Rognebakke & Faltinsen
(2001) undertook two-dimensional experiments investigating the effects of sloshing on
ship motions for a box shaped ship section excited in a regular beam sea. The ship
section analysed contained two tanks and it was forced to sway. The change in sway
motion of the model with wave frequency, for different filling levels of the two tanks
was presented. Kim (2001) worked on the coupled analysis of ship motions and
sloshing flows in beam seas.
The problem of ship motions in waves is relatively complicated; therefore some

simplifications may be appropriate. The transverse dimensions of the structure may be
assumed to be small compared to its length. This geometric simplification of the
structure leads to strip theory. Detailed reviews of the strip theory, covering its
historical and mathematical development may be found in Salvesen et al. (1970),
Odabasi & Hearn (1978) and Newman (1978).
For the analysis of ship motions, a representation of the wetted surface is necessary.
This may be numerical using boundary elements or analytical using conformal mapping
techniques. The latter technique is used to transform the sections of the ship into a
circle, for which the form of the potential is known and hence the added mass and
damping properties of a ship section can be determined. Mathematical details related to
the conformal mapping techniques may be found in Kerczek & Tuck (1969) and
87
Hoffman & Zielinski (1977). A multi-parameter conformal mapping technique is

presented by Westlake & Wilson (2000) and Westlake et al. (2000). This technique
provides the mapping of asymmetric sections and sections with large bulbous into a
circle. The conformal mapping technique could prove useful when dealing with the
transverse sections of damaged ships regarding hydrodynamic aspects. However, since
strip theory description of hull form is not useful for damage stability this approach has
not been considered for implementation in this research programme.
5.2 Frequency and Time Domain Approach

The frequency-domain method is a standard approach in seakeeping analysis. The
frequency-domain approach provides steady state amplitudes under small perturbation
assumptions. The steady state approach implies that the dynamic motions are linear and
the loads acting on the structure oscillate harmonically with the same frequency as the
waves exciting the structure. It is a useful preliminary design tool as it uses significantly
less computational time than the time-domain approach.
An alternative way of dealing with the damaged and intact structure is to use a timedomain approach. Damage and the ingress of water into the compartments are
continuous phenomena and they may be analysed in detail in time steps (see, for
example, Chan et al. (2002) and Turan (1993)). In the frequency-domain method this
time-dependence is not modelled.
In this study the question of whether optimised hulls are more vulnerable to damage
than non-optimised hulls is to be addressed. Since the damage stability analysis outlined
in Chapter 4 seeks a new static equilibrium state when ship is damaged, the frequencydomain approach is useful in providing steady state motions about this heeled and
trimmed ship state. The time-domain approach would be considered useful when trying
to understand how different the behaviour is in detail.
88
5.3 Justification of Selected Approach

Different theoretical and experimental investigations of intact and damaged ships have
been reviewed. Frequency-domain approaches are simpler to apply in practice but yield
steady state characteristics, whereas time-domain and experimental studies may yield
details of how a steady state may or may not be achieved. Since the literature has not (so
far) yielded any studies related to the impact of damage on initial conceived hull forms
and subsequent optimised hull forms it is considered more prudent to follow the
frequency-domain approach since, as Section 5.4 will reveal, some fundamental
difficulties have to be overcome if the motions of the damaged ship are to be modelled
so as to make comparison with the intact ship meaningful.
The aim of this research is to analyse the capability of optimised and non-optimised
ships to withstand damage. As this will be analysed in the earlier stages of the design
process when different alternative hull forms are considered the selected analysis
method for predicting the hydrodynamic characteristics and hence motion responses of
the intact and damaged structure should be relatively fast. Thus a wealth of experience
may be gained regarding the damage stability of a new hull form design. Hence a
frequency-domain approach will be considered in this study. The steady state situation
for both intact and damaged structure will be analysed.
The water on deck problem is not analysed in this research, since the magnitude of the
orientation of the vessel for the most probable damage to be expected does not expose
the deck of the vessel to the free-surface of the water. When the selected methods are
applied to the selected ship it will be found that sloshing within the ship is not an issue
and the changes in behaviour are relatively subtle.
89
5.4 Motion Analysis

The motion responses of the damaged hull forms must be determined with greater
structural (mechanical) cross-coupling than is usually adhered to the solution of the
equations of motions of intact ships. This has necessitated the development of a novel
approach to implement the calculation of the pure and product moments of inertia for
the intact and damaged hull forms to facilitate meaningful comparisons of intact and
damaged ship motions. This novel method is presented in Section 5.4.3.
The responses of a damaged structure cannot be evaluated with the two-dimensional

strip theory approach, since the physics of the damage will not be captured well. Hence,
the hydrodynamic and the motion analysis are carried out in three-dimensions.
5.4.1 Forces and Moments Acting on a Floating Ship in Waves
Forces and Moments Acting on an Intact Ship
The presence of the intact structure in wind-generated progressive regular & harmonic
waves leads to the generation of diffracted waves by the hull. The incident and
diffracted waves are responsible for the loading of the structure through the resulting
wave excitation forces and moments.
As a consequence of the excitation the ship will respond and this will lead to the
generation of ship radiation or reaction forces and moments due to the radiation waves
generated. Furthermore, there will be restoring forces and moments of an Archimedean
hydrostatic nature as well as the inertial forces and moments. All depend upon the shape
and orientation of the ship.
90
Forces and Moments Acting on a Damaged Ship
When a ship is damaged, the water that flows into the damaged compartments causes
static and dynamic effects. Static effects are due to the mass of water that is entrapped
in the damaged compartment and so lead to changes in the mass distribution, centre of
gravity and product of inertia of the vessel. The hydrodynamic effects of water
entrapped within damaged compartment(s) are ignored because this would necessitate
the modelling of the internal wetted surfaces in a manner consistent with conservation
of total displaced ship volume.
When a ship is damaged; the wave excitation and the radiation forces and moments
change due to the changes in the extent and orientation of the wetted surface area. The
Archimedean restoring forces are altered due to the change in waterplane area, centre of
gravity, centre of flotation, centre of buoyancy and displacement of the ship.
Furthermore, the inertial characteristics of the hull form will change because the mass
distribution for the damaged ship is different to the intact ship by virtue of mass
orientation changes.
All these changes must be present in the equations of motion of the damaged ship.
5.4.2 The Equations of Motion
The equations of motion of an intact ship are generally simplified due to the portstarboard geometric symmetry of the ship. Since these simplifications will not generally
exist when a ship is damaged the equations of motion presented next assume their most
general form.
91
Equations of Motion for Intact Hull Forms
Having linearised the fluid-structure interaction analysis it is consistent to linearise the

governing equations of motion. With the motion responses being harmonic, the six
coupled second-order ordinary differential equations can be reduced to six coupled
simultaneous algebraic equations. More details of the theory can be found in Salvesen et
al. (1970), Newman (1978), Lewis (1989) and Faltinsen (1990).
When determining the heel and trim of the damaged ship the longitudinal centre of the
intact waterplane is used as the origin of the right-handed Cartesian reference system.
Here the origin of the motion reference system must be coincident with the earth fixed
hydrodynamic reference system M so that the motions and the forces and moments are
described with respect to the same system. Since the motion/fluid-structure interaction
system is generally chosen as a convenient point in the undisturbed free-surface it will
not be coincident with the centre of gravity. Consequently, mechanical cross-coupling
terms originating from the inertial forces through the centre of gravity will couple the
translational and rotational degrees of freedom. Any lack of mass distribution symmetry
will lead to non-zero products of inertia and this will lead to mechanical coupling
between the rotational degrees of freedom.
Without being too specific at this point let M denote the origin with positive X to port,
positive Y directed upwards and positive Z is directed toward the bow of the ship. M
and its relationship with the longitudinal centre of the intact waterplane will be specified
more definitively when applying the formulated equations of relative motion in Chapter
8.
The translatory displacements of surge (1 ), sway ( 2 ) and heave (3 ) of the ship are
thus in the positive Z, X and Y directions. The angular displacements of roll ( 4 ), pitch
(5 ) and yaw (6 ) are right handed rotations about the Z, X and Y axes respectively as
illustrated in Figure 5.1.
92
BOW
PORT
4
1
3
5
2
STARBOARD
STERN
Figure 5.1: The motion-hydrodynamic Cartesian coordinate reference system.
The six linear coupled differential equations of motion for an intact hull can be written
as:
( M
6
j =1
kj
+ Akj ) j + Bkj j + Ckj j = Fk : k = 1,...., 6
(5.1)
The generalised mass matrix M kj elements are defined in Equation (5.2). The
elements of Akj and Bkj are the added mass and fluid damping coefficients for
reactions in the kth direction arising from the ship motion in the jth direction. The
restoration matrix Ckj
elements are the hydrostatic restoring force/moment
coefficients for reactions in the kth direction arising from the ship motion in the jth
direction. This is defined in Equation (5.3). The time-dependent wave excitation
complex amplitudes are denoted by F1 , F2 and F3 denoting surge, sway and heave
93
wave exciting forces and F4 , F5 and F6 corresponding to roll, pitch and yaw wave
exciting moments. is the wave circular frequency and j is the jth time-dependent
displacement of the ship. Finally, j is the acceleration and j is the velocity of the
ship in the jth degree of freedom. Clearly subscripts j and k assume the values 1 to 6.
M
0
M kj = 0
MYG
MX G
MYG
MYG
MX G
MZG
MYG
MX G
MZG
MZG
( X + Y ) dm X Z dm
X Z dm
( Z + Y ) dm
Y Z dm
X Y dm
(
2
m m
MZG
Ym Z m dm
X mYm dm
X m 2 + Z m 2 dm
MX G
(5.2)
Clearly, irrespective of the geometric properties of the ship the generalised mass matrix
M kj is symmetric.
0
0
0
Ckj =
0
0
0
0
0
0
s gAwps
0
0 s gAwps X C
0 s gAwps ZC
0
0
0
0
s gAwps X C
g YG + s gYB + s gI XX
s gI XZ
0
0
0
s gAwps ZC
s gI XZ
g YG + s gYB + s gI ZZ
0
0
0
0
0
0
0
(5.3)
The restoring coefficient matrix Ckj is also symmetric irrespective of shape of

structure.
94
In the generalised mass matrix formulation M kj ; M is the mass of the ship,
( X G , YG , ZG )
are the centre of gravity coordinates of the ship. The notation used
assumes the centre of gravity coordinates of the intact ship are positive, but there is no
loss of generality as the actual values (negative or positive) are provided once an
application is to be undertaken. ( X m , Ym , Z m ) denotes the coordinates of the elemental
ship mass dm .
Within the stiffness matrix Ckj ; ( X C , YC , Z C ) is the centre of flotation coordinates, g
is the acceleration due to gravity and s is the density of sea water. is the displaced
mass of the ship, is the displaced volume of the ship and YB is the vertical coordinate
of the centre of buoyancy. The moments of inertia I XX , I ZZ and I XZ included in the
stiffness matrix Ckj are related to the moments of inertia in the generalised mass
matrix M kj as follows:
I 44 I ZZ = ( X m 2 + Ym 2 ) dm
(5.4)
I 55 I XX = ( Z m 2 + Ym 2 ) dm
(5.5)
I 66 IYY = X m 2 + Z m 2 dm
(5.6)
I 45 I ZX = X m Z m dm = I XZ I 54
(5.7)
I 46 I ZY = Ym Z m dm = IYZ I 64
(5.8)
I 56 I XY = X mYm dm = IYX I 65
(5.9)
95
The displacement of the ship and the wave excitation loads can respectively be written
in the form
j = ja e it ( jr + i ji ) e it
(5.10)
Fk = Fka e it ( Fkr + iFki ) e it
(5.11)
because there is a phase shift between the incident waves and the ship reactions and
excitation loads.
The equations of motion for translations can be written as follows:
M1 = F1 ( A1 j j + B1 j j ) ( MYG5 + MX G6 )
(5.12)
M 2 = F2 ( A2 j j + B2 j j ) ( MYG4 MZ G6 )
(5.13)
j =1
j =1
M3 = F3 ( A3 j j + B3 j j )
6
j =1
C3j j ( MX G 4 + MZ G5 )
(5.14)
j=3,4,5
In the case of no symmetry of mass distribution the rotational equations of motion are
written:
I 44 4 + I 455 + I 466 = F4 ( A4 j j + B4 j j )
6
j =1
( MYG 2 MX G3 )
96
j =3,4,5
C4 j j
(5.15)
I 544 + I 555 + I 566 = F5 ( A5 j j + B5 j j )

6
j =1
j =3,4,5
C5 j j
( MYG1 + MZ G3 )
I 64 4 + I 655 + I 666 = F6 ( A6 j j + B6 j j ) ( MX G1 MZ G2 )
(5.16)
(5.17)
j =1
Equations of Motion for Damaged Hull Forms
The responses of a ship in waves are expected to change when damage occurs for a
number of reasons. There will be coupling between the rotational and translational
degrees of freedom whenever the damage leads to a modified orientation of the ship as a
result of heel and trim. That is, any geometric symmetry present in the intact ship will
no longer exist in the damaged ship. Furthermore, when a hull form is damaged, its
centre of gravity and its mass distribution may change due to the water entering via the
damaged area or due to structural weight loss. Damage may destroy the original
waterplane area and this will cause buoyancy loss.
In order to recover the buoyancy loss and/or the additional weight (due to the water
flooded into the damaged compartment) and/or structural weight loss, the vessel will in
general undergo parallel sinkage, trim and/or heel.
The six linear coupled differential equations of motion for a damaged hull can be
written as;
( M + A ) + B + C = F
6
j =1
kj
kj
kj
kj
97
: k = 1,...., 6
(5.18)
where Fk are the time-dependent wave exciting forces and moments for the damaged
ship, j is the time-dependent displacement of the damaged ship and hence j and j
are the acceleration and the velocity of the damaged ship in the jth degree of freedom.
Clearly new values of the hydrodynamic characteristics for the damaged ship Akj ,
Bkj and Fk must be calculated. The contents of the generalised mass matrix M kj
will change due to extra and differently evaluated terms for the damaged case. Water
ingress into the damaged ship will have the following effects:
Change of the original total mass of the ship due to both ingressed mass of water
and structural mass loss.
The centre of gravity of the damaged ship is likely to be in a different position to

that of the intact ship.
The structural mass distribution within the ship will lead to different moment of
inertia properties due to orientation of intact ship and its original mass and mass
gains/losses cited.
The hydrostatic restoring coefficients Ckj will change due to new centre of flotation
for damaged ship, modified still water waterplane area and the change in centre of
buoyancy because of different transverse sectional area distribution as orientation of the
ship changes.
The generalised mass matrix M kj and the stiffness matrix Ckj are given by:
98
M + mw
0
M kj =
0
( M + mw ) YG
( M + mw ) X G
( X
2
m
+Ym2 dm+ Xw2 +Yw2 dm
Xm Zm dm XwZwdm
YmZm dm YwZwdm
0
0
0
Ckj =
0
0
M + mw
M + mw
( M + mw ) YG
0
( M + mw ) Z G
( M + mw ) X G
( M + mw ) ZG
0
( M + mw ) YG
( M + mw ) YG
( M + mw ) XG
( M + mw ) XG
( M + mw ) ZG
0
( M + mw ) ZG
Xm Zm dm XwZwdm
( Z
2
m
+Ym2 dm+ Zw2 +Yw2 dm
Xm Ymdm XwYwdm
YmZm dm YwZwdm
Xm Ymdm XwYwdm
Xm 2 + Zm 2 dm+ Xw2 + Zw2 dm
(5.19)
0
0
0
0
0
0
0
0
XC
ZC
s gAwps
s gAwps
0 s gAwps
XC gYG + s gYB + s gI XX
s gI XZ
0 s gAwps
ZC
s gI XZ
gYG + s gYB + s gIZZ
0 s gAwps
0
0
0
0
0
0
0
0
0
0
(5.20)
The hydrodynamic and hydrostatic quantities used for the intact and damaged ships
must be calculated to reflect the geometry of the ship. In the damaged case there is a
total lack of geometric symmetry. Within Equations (5.18), (5.19) & (5.20) the
superscript prime indicates that values to be used relate to the damaged ship. The
additional notation, associated with these equations, that needs to be defined are as
follows:
99
mw is either mass of ingressed water in the damaged compartment(s), or

structural mass loss or a combination of both
dm is the elemental mass of ingressed water, or structural loss or a combination
of both at location ( X w , Yw , Z w ) .
The basic displacement and wave excitation relationships and the equations of motion
of the damaged ship can be written in an analogue from the intact ship equations:
j = ja e it ( jr + i ji ) e it
(5.21)
Fk = Fka e it ( Fkr + iFki ) e it
(5.22)
( M + mw )1 = F1 ( A1j j + B1j j ) ( ( M + mw ) YG5 + ( M + mw ) X G 6 )

6
j =1
(5.23)
( M + mw )2 = F2 ( A2 j j + B2 j j ) ( ( M + mw ) YG4 ( M + mw ) ZG 6 )
6
j =1
(5.24)
( M + mw )3 = F3 ( A3 j j + B3 j j )
6
j =1
C
3j
j=3,4,5
( ( M + mw ) X G 4 + ( M + mw ) Z G 5 )
4 + I 45
5 + I 46
6 = F4 ( A4 j j + B4 j j )
I 44
6
j =1
j =3,4,5
C4 j j
( ( M + mw )YG2 ( M + mw )X G 3 )
100
(5.25)
(5.26)
4 + I 55
5 + I 56
6 = F5 ( A5 j j + B5 j j )
I 54
6
j =1
j =3,4,5
C5 j j
(5.27)
( (M + mw )YG1 + (M + mw )Z G 3 )
4 + I 65
5 + I 66
6 = F6 ( A6 j j + B6 j j )
I 64
6
j =1
(5.28)
( (M + mw )X G 1 (M + mw )Z G 2 )
Having highlighted in this section that ship motions will change with damage, it is now
necessary to consider in greater detail the evaluation of the pure moments and products
of inertia for the damaged case.
It is quite common in the motion studies of intact ships to exploit the port-starboard
geometric symmetry that exists and the further simply the analysis by assuming (often
without any real justification, other than difficulty of specification) that the products of
inertia are zero. However, even if this is the case the products of inertia are unlikely to
be zero in the case of the damaged ship; hence their values must be estimated in some
consistent fashion with the physics that persist in the intact ship case. The challenge is
tackled next in the following section.
5.4.3 A Novel Method for Determining the Products of Inertia for Intact and
Damaged Ship
The Intact Ship
The mass moment of inertia calculations for the intact ship are usually undertaken
subject to the implicit assumption of port-starboard and/or fore-aft symmetry of the ship
mass. This results in zero valued products of inertia of the ship. However, this is not the
case for the damaged hull form, even if the ship in its intact form displayed such mass
101
symmetry. The water flooding into the damaged compartment(s), the structural loss
occurred in the damaged part of the ship and the change of the distances of the unit
masses forming the actual mass of the ship will affect the mass products of inertia.
Consequently, the assumed symmetry of the mass with respect to the MZ and MX axes
defined in Figure 5.2(a) will no longer exist.
To assist with the task of determining how the products of inertia have changed from
their zero intact ship values to their non-zero damaged ship values, the actual total mass
of the ship is regarded as being composed of the four representative point masses MAP,
MFP, MAS and MFS defined in Figure 5.2(a) with respect to the earth-fixed right-handed
coordinate system MXYZ.
ZA
MAP
ZF
MFP
XP
MAS
XS
(a) Plan view
102
MFS
XS
XP
X
YAS, YFS
YAP, YFP
MAS + MFS
MAP + MFP
(b) Transverse view
ZA
ZF
YAP, YAS
YFP, YFS
MAP + MAS
MFP + MFS
(c) Side view
Figure 5.2: Total mass of the ship represented as four equivalent point masses.
Irrespective of the geometry and thus the condition of the ship (intact or damaged), the
product of inertias are symmetric; that is, I 45 = I 54 , I 56 = I 65 and I 46 = I 64 (see, for
103
example, Housner and Hudson (1966)). Using the point mass equivalence illustrated in
Figure 5.2(a) it follows that;
I 45 = M AP Z A X P + M FP Z F X P + M AS Z A X S + M FS Z F X S .
(5.29)
The right hand side of the Equation (5.29) can be equated to zero only if:
M FP = M FS
and
M AP = M AS
(5.30)
subject to X S = X P .
That is, irrespective of the relationship between Z A and Z F , port-starboard symmetry

of mass forward and aft of the midship section implies that I 45 = I 54 = 0 .
Similarly, on appealing to Figure 5.2(b), it may be argued that:

I 56 = M AS YAS X S + M FS YFS X S + M APYAP X P + M FPYFP X P .
(5.31)
The port-starboard symmetry assumptions of Equation (5.30) for the intact ship mass
distribution imply that:
I 56 = M AS X S (YAS YAP ) + M FS X S (YFS YFP ) .
104
(5.32)
Clearly, M AS and M FS are positive quantities, X S is non-zero and port-starboard

symmetry of the mass also means that YAS = YAP and YFS = YFP . That is, port-starboard
symmetry of the mass implies that I 45 = I 54 = 0 and I 56 = I 65 = 0 .
Thus assuming that port-starboard geometric symmetry implies port-starboard mass

symmetry, one can readily appreciate that four of the cross-products are zero valued.
This is a common assumption when solving the equations of motion of an intact floating
structure.
Next, the implications of assuming I 46 = I 64 = 0 are considered with the aid of Figure
5.2(c). The products of inertia I 46 may be written in the form:
I 46 = M AP Z AYAP + M AS Z AYAS + M FP Z F YFP + M FS Z F YFS .
(5.33)
By writing M A for ( M AP + M AS ) and M F for ( M FP + M FS ) it follows that:

I 46 = M A Z AYAS + M F Z F YFS .
(5.34)
If I 46 is to be assumed zero for the intact case, Equation (5.34) suggests two
possibilities. Either the ship may have a fore-aft mass symmetry, which is not the case
for most of the ships, or there should be a relationship between YAS and YFS to make the
equation equal to zero. However, since the aft mass ( M A located at Z A ) and the
forward mass ( M F located at Z F ) must be in equilibrium about the centre of gravity, it
follows that M A Z A + M F Z F = 0 since Z A and Z F are of opposite sign. Hence from
Equation (5.34) it follows that YAS = YFS .
105
With M equal to the total mass and kYY denoting the radius of gyration (see, for
example, Peach & Brook (1987)), I 66 is usually expressed as:
2
I 66 = M kYY
subject to kYY lying in the range 0.2 to 0.25L ,
(5.35)
where L is the length between perpendiculars of the ship.
I 66 may also be written, upon appealing to Figure 5.2(a), in the form:
I 66 = M AP Z A2 + X P2 + M FP Z F2 + X P2 + M AS Z A2 + X S2
+ M FS Z F2 + X S2
(5.36)
Upon adopting notation of Equation (5.30) for summing constituent masses Equation
(5.36) may be written as:
I 66 = M A Z A2 + X P2 + M F Z F2 + X P2 .
(5.37)
Equating Equations (5.35) and (5.37) and assuming the veracity of Equation (5.35) then
2
M A Z A2 + X P2 + M F Z F2 + X P2 = ( M A + M F ) kYY
.
(5.38)
Since the details of the longitudinal mass distribution is usually more readily available
than the transverse distribution of the mass, it is reasonable to assume that Z A , Z F , M A
and M F are known. Hence, Equation (5.38) can be used to determine X P .
Similarly, I 44 can be approximated by the formula (see, for example, Peach & Brook
(1987));
106
2
I 44 = M k ZZ
with k ZZ assuming values in the range 0.35 to 0.41 B ,
(5.39)
where B is the beam of the ship.
I 44 can also be written, upon appealing to Figure 5.2(b), as:
2
I 44 = M AS X S2 + YAS2 + M FS X S2 + YFS2 + M AP X P2 + YAP
2
.
+ M FP X P2 + YFP
(5.40)
Again using Equation (5.30), with YAS = YAP and YFS = YFP and equating Equations
(5.39) and (5.40) leads to:
2
M A X S2 + YAS2 + M F X S2 + YFS2 = M k ZZ
.
(5.41)
With X S assumed known then

2
M AYAS2 + M F YFS2 = M k ZZ
MX S2 .
(5.42)
Thus both equations (5.34) and (5.42) provide a relationship between YAS and YFS .
I 55 can also approximated by the formula (see, for example, Peach & Brook (1987)):
2
I 55 = M k XX
with k XX in the range 0.2 to 0.25 L .
Finally, I 55 can also be written in the form:
107
(5.43)
2
2
I 55 = M AP Z A2 + YAP
+ M AS Z A2 + YAS2 + M FP Z F2 + YFP
+ M FS Z F2 + YFS2 .
(5.44)
Again using Equation (5.30) with the equivalence of Equations (5.43) and (5.44) yields
the expression:
2
M A Z A2 + YAS2 + M F Z F2 + YFS2 = M k XX
.
(5.45)
Since Z A , Z F , M A and M F are known from a knowledge of the longitudinal mass

distribution (whether it is from actual data or an assumed Coffin diagram), Equation
(5.45) provides a third relationship between YAS and YFS . Hence, from equations (5.34)
and (5.42) or (5.45), YAS and YFS can be evaluated as there are two unknowns and three
independent equations.
In particular, from Equation (5.45) it follows that:
2
M AYAS2 + M F YFS2 = M kYY
M A Z A2 + M F Z F2 .
(5.46)
Having already deduced from Equation (5.34) the condition for I 46 = 0 :
M Z
YAS = F F
M AZ A
YFS = YFS .
(5.47)
YAS and YFS can be readily determined from Equation (5.46). Hence the total intact ship
mass may be readily equated to four point masses with determinable locations
consistent with the practice of assuming zero intact ship cross-products. Applications of
these equations are provided in Appendix C for the parent Derbyshire and a tanker hull
form.
108
The Damaged Ship
Having indicated how the four equivalent point masses are to be positioned such that
moments of inertias are assigned in a conventional manner and the cross-products are
zero the next step is to determine the position of these equivalent point masses when the
ship is in its damaged position and to identify any mass difference that might exist
between the intact ship and the damaged ship. This information will allow the non-zero
cross-products to be determined for the damaged ship.
In Section 4.3 a method was described to determine the orientation of the damaged ship
once the consequences of the damage were known in terms of water ingress and
structural mass loss. The assignment of the location of the damage and its extent will
clearly influence the subsequent hydrodynamic and motion analysis of the damaged
ship and thus any conclusions drawn. Assignment of damage position and extent will be
based upon statistical damage data presented and discussed in Chapter 3. By using the
outcome of Chapter 3, the likely changes in mass between the intact and damaged ship
can be considered in this section. The damage for the selected ship is most likely to
occur in the forward area of the ship (see Section 3.1) and given the mass distribution
assumptions inherent in the intact ship analysis one may choose either the port or
starboard side.
With this assumption it follows that the damaged ship masses satisfy
= M FP , M AS
= M AS .
= M FS + mw , M AP = M AP , M FP
M FS
(5.48)
The location of the four damaged distinct point masses change as the ship experiences
parallel sinkage ( YSINK ), trim ( ) and heel ( ) as illustrated in Figures 5.3 & 5.4. The
original coordinates for M AP were Z A , X P and YAP and after the ship orientation
change they become:
109
Z A = Z A cos sin (YAP YSINK )
(5.49)
X P = Z A sin sin + X P cos + sin cos (YAP YSINK )
(5.50)
= Z A cos sin X P sin + cos cos (YAP YSINK ) .

YAP
(5.51)
and
LCF
WL2
WL1
YSINK
WL
Figure 5.3: Description of the trim angle and parallel sinkage.
WL3
WL2
Figure 5.4: Description of the heel angle.

110
In the Figures 5.3 & 5.4 WL, WL1, WL2 and WL3 stand for the waterlines for the initial
intact ship position, after parallel sinkage, after parallel sinkage & ship trimming and
after parallel sinkage & ship trimming & ship heeling respectively.
Similarly the coordinates for M AS in the intact condition were Z A , X S and YAS and
they become:
Z A = Z A cos sin (YAS YSINK )
(5.52)
X S = Z A sin sin + X S cos + sin cos (YAS YSINK )
(5.53)
= Z A cos sin X S sin + cos cos (YAS YSINK ) .

YAS
(5.54)
That is, Equations (5.49) to (5.51) with X P and YAP replaced by X S and YAS
respectively.
The coordinates ( Z F , X P , YFP ) for M FP in the intact condition become:
Z F = Z F cos sin (YFP YSINK )
(5.55)
X P = Z F sin sin + X P cos + sin cos (YFP YSINK )
(5.56)
= Z F cos sin X P sin + cos cos (YFP YSINK )

YFP
(5.57)
As expected X P of Equation (5.56) is identical in form with X P of Equation (5.50)

with Z A replaced with Z F and YAP replaced with YFP . Finally for M FS the original
coordinates ( Z F , X S , YFS ) are transformed to:
111
Z F = Z F cos sin (YFS YSINK ) Z F
(5.58)
X S = Z F sin sin + X S cos + sin cos (YFS YSINK )
(5.59)
which is similar to X S with Z A replaced with Z F and YAS replaced with YFS .
YFS = Z F cos sin X S sin + cos cos (YFS YSINK ) .
(5.60)
With the equivalent point masses and their new positions known the damaged ships
pure and product moment of inertias can be readily determined.
5.4.4 Determination of Motion Characteristics
The hydrodynamic coefficients, the hydrostatic restoring coefficients and the wave
excitation forces and moments for the intact ship and damaged ship are evaluated using
the Matthew Diffraction Suite. The motion characteristics for the intact and damaged
ship for its basis hull form and optimised hull forms are determined using the Motion
code of the author (see Saydan (2003)).
The Motion code solves the linear equations of motions in their most general form. That
is, a set of six linear coupled equations symbolised by:
[ A] = F ,
(5.61)
where [ A] includes the generalised masses, the added masses, the fluid damping and
hydrostatic restoring coefficients indicated in Equation (5.1) and specified in Equation
112
(5.2) & (5.3). is the displacement vector for the six degrees of freedom (surge, sway,
heave, roll, pitch and yaw) and F contains the wave exciting forces and moments.
The linear algebraic simultaneous equation formulated may be solved analytically by

using Cramers rule (see Greenberg (1998)) to derive mathematical expressions for each
degree of freedom. Alternatively one may use classical numerical methods such as the
Gaussian Elimination method (see Greenberg (1998)) whereby [ A] is reduced to upper
triangle form and the complex amplitudes for each degree of freedom determined from
back substitution (see Greenberg (1998)). Equivalently [ A] can be reduced to the unit
matrix to provide the motion amplitudes directly.
To cross check the solutions determined in the intact and damaged case the numerically
solved solutions and the original coefficient matrix [ A] are multiplied to cross check
equality of F so calculated and F provided originally. Furthermore, since for the
intact ship (surge, heave and pitch) and (sway, roll and yaw) are independent Cramers
rule can be used for each three degrees of freedom system to compare with numerically
derived solutions.
5.4.5 Discussion
The forces acting on an intact ship and a damaged ship have been discussed in Section
5.4.1 together with the changes in the governing equations of motion for an intact and a
damaged structure (see Section 5.4.2). The equations of motion are written in their most
general form for both cases.
The variations in the equations of motion due to the damage (discussed in Section 5.4.2)
are valid for damage cases below and at the waterline. For damage cases analysed in
this study, the damage is modelled as a change in mass distribution in the damaged
113
region. This causes some changes in the added mass coefficients, the damping
coefficients and the wave excitation forces since the damage changes the orientation of
the wetted surface area of the ship. The hydrostatic restoring coefficients vary as the
damage changes the displacement weight and the centre of gravity of the ship.
The hydrodynamic coefficients addressing the intact and damaged original ship hull
form and its optimised intact and damaged hull forms are presented in Chapter 7 and
Appendix H. The relative vertical motion responses for the original intact and damaged
ship and for its optimised intact and damaged hull form are plotted in Chapter 8 and
Appendix J.
Selection of an appropriate candidate merchant ship for investigating the impact of

damage on its responses in its as designated and optimised form must take into
account the extent of knowledge required to undertake the optimisation and
subsequently the detailed stability (intact and damaged), hydrodynamic and motion
response analyses addressed in the following chapters.
The following chapter provides the ship selection rationale and the application of the
optimisation process and other indicated analyses.
114
6.
SELECTION AND FUNDAMENTAL ANALYSIS OF BASIS
SHIP
The optimisation of a basis hull form may be undertaken using the secondary
parameters of CWP, LCB and LCF, the primary parameters of L and B/T or both. Ship
trims with respect to LCF. Once damaged the ship LCB will change in accordance with
the extent and location of the damage and the amount of water ingressed. To permit
sensible comparisons of the motions of alternative intact and damaged ships the
secondary parameters will not be modified in the optimisation undertaken.
Consequently, motions can be defined with respect to a common origin located at the
intact LCF and motions differences of the damaged ships will not be complicated by
any combination of LCF and LCB changes that might take place in secondary parameter
optimisation. With CB and intact displacement remaining unchanged L and B/T will be
modified to improve intact ship motions subject to the constraint of no increase in calm
water resistance. Hence there are no changes in the engine requirements, structural mass
distribution of the alternative designs or in engine room bulkheads. The difference will
be a consequence of damage and optimisation for seakeeping only.
In the following sections details of the selected hull form, its optimisation and the IMO
stability requirements addressed will be considered together with details related to mass
distribution and damage scenarios to be investigated.
6.1 Choice of Hull Form

The collisions and groundings statistics of Figure 3.4(a) & (b) clearly indicate that
general cargo ships and tankers should be the principal ship types investigated (having
average percentages of 52.45% and 22.2%). Thereafter containerships and bulk carriers
exhibit comparable average statistics (5.5% against 5.05%). However, such a clear and
115
obvious selection was not possible due to the detailed mass distribution data required to
determine cross-products of inertia. Furthermore, when the bulk carrier Liverpool
Bridge (re-named Derbyshire in 1978) was selected based on the in-depth
investigations available in the literature (Department of Transport (1989), Bishop et al.
(1991), Vassalos et al. (2001) and Paik & Faulkner (2003)), it was discovered that
whilst some very detailed structural aspects were available they were insufficient to
produce accurate cross-products of inertia for the intact ship. Consequently, it was
necessary to develop the novel cross-products of inertia estimation scheme of Chapter 5
for the damaged ship, whilst assuming the zero valued inertia product for the intact ship.
This choice is not ideal in the face of the statistics presented in Chapter 3, but it simply
reflects the fact that ship data sufficient to undertake really detailed analysis is not
readily released in the public domain or under the umbrella of confidentiality within a
university based research project.
6.2 Public Domain Particulars for Derbyshire

The principal Derbyshire hull form parameters are presented in Table 6.1. A general
arrangement drawing of the Derbyshire is given in Appendix D (drawing is taken from
the Department of Transport (1989)).
Table 6.1: Main particulars of the Derbyshire taken from the Department of Transport
(1989).
LOA
Length Overall (m)
Length Between Perpendiculars (m)
L
Longitudinal Centre of Gravity (from AP) (m)LCG
Breadth (m)
B
Draught (m)
T
Depth (m)
D
Service Speed (knots)
V
116
293.25
281.94
144.4906
44.196
17.035
24.994
15.5
The body plan for the Derbyshire is provided in Figure 6.1.
Figure 6.1: The parent hull form.
The longitudinal mass distribution given in Bishop et al. (1991) leads to a longitudinal
centre of gravity (LCG) of 147.6 metres measured from the aft perpendicular. Since this
does not coincide with the longitudinal centre of buoyancy (LCB), but in fact implies a
bow down trim ballast water and fuel are added to the wing tanks in holds 6, 7, 8, 9 and
in the after peak tank to establish a level trim state. To keep total mass constant water
and fuel were removed from forward tanks and added to cited aft tanks to achieve
required level trim. The resulting longitudinal distribution of mass for the Derbyshire
now assumes the form provided in Figure 6.2.
117
1400
Mass / Length (tonnes / m)
1200
1000
800
600
400
200
0
0
0.2
0.4
0.6
0.8
1.2
x/L
Figure 6.2: The modified longitudinal mass distribution of the Derbyshire.
The mass distribution is very peaky and quite unlike the often assumed coffin diagrams.
For accurate calculation of the products of inertia, corresponding sectional distributions
are required at a representative number of stations. Since such details cannot be
provided at arbitrary selected ship stations the equivalent four-point masses and their
location are determined assuming intact products of inertia are zero with respect to
centre of gravity. However, from the longitudinal distribution of the mass one may
readily estimate the pitch associated radius of gyration.
Summing the forward and aft masses separately, and their first moments with respect to
the longitudinal centre of gravity, the four equivalent point masses M AP , M FP , M AS
and M FS and their longitudinal positions can be readily determined to yield the details
presented in Table 6.2.
118
Table 6.2: The four masses and their longitudinal positions for parent Derbyshire.
MA
94686.6 (t)
MF
104793.6 (t)
ZA from LCG 64.1 (m)

ZF from LCG 57.9 (m)
MAP = MAS
47343.3 (t)
MFP = MFS
52396.8 (t)
From Equation (5.38) the coordinate X P can be determined assuming kYY = 0.225 L .
Hence from Equation (5.30) it is clear that X S = X P . With X S known Equation (5.42)
can be used to determine YAS = YFS , subject to the assumption that k ZZ = 0.41B together
with the deduction made earlier from Equation (5.34). Finally, to check consistency of
assumed radii of gyration YAS = YFS is re-determined using Equation (5.45) with
k XX = 0.2163L .
The results of applying the indicated procedure just outlined are presented in Table 6.3.
Clearly YAS and YFS are consistent.
Table 6.3: The horizontal and vertical positions of the four masses for parent
Derbyshire.
Equation Quantity Determined

XP =17.795 (m)
5.38
XS =-17.795 (m)
5.30
Radii of Gyration
kYY = 0.225
kYY = 0.225
5.42
YAS = YFS = 3.41579 (m) kZZ =0.410
5.45
YAS = YFS = 3.41579 (m) kXX = 0.2163
Details of the intermediate calculations are presented for completeness in Appendix C.

The process is in someway artificial as the need to present the total mass as four
119
equivalent points is only carried out to allow motions of damaged ship to be facilitated
in a manner consistent with the intact motion analysis. To achieve moments of inertia
and the consistent coordinate points
( XP, XS )
and
(YAS , YFS )
some iteration is
necessary. One is not guarantying a unique set of answers and, furthermore, the radius
of gyration values used to solve equations indicated in Table 6.3 will seem a little lower
than expected for pitch and yaw values used in general motion calculations. The process
presented is a compromise to overcome a difficulty that could be readily resolved if a
full description of the mass distribution were readily available.
6.3 Optimisation Objective Function and Design Parameters

Having provided the basis known details and body plan for the intact Derbyshire the
next task is to identify an optimised hull form of Derbyshire.
Prior to presenting the results of the optimisation process it is worth noting, if only for
completeness and to demonstrate how this approach sits with other investigated
optimisations, to briefly indicate some of the public literature investigations essentially
based on modifications of the primary hull form characteristics.
Lewis (1959) achieved reduction in the pitching amplitude and vertical bow
acceleration by selecting a longer hull. Vossers et al. (1960) showed that heave, roll,
pitch and RBM amplitudes are sensitive to changes of L/T. Abkowitz et al. (1966)
concluded that the reduction of the draught of the ship results in better vertical motion
responses. Robson (1988) emphasized that vertical seakeeping performances could be
improved by increasing LWL/BWL and BWL/T ratios.
The effects of primary and secondary parameters on vertical motions and added
resistance characteristics are presented in a tabular form in Sarioz (1993) for monohulls
and in Wright (2004) for catamarans.
120
Generation of alternative hull forms in this case is undertaken using linear distortion
methods.
Prior to initiating the optimisation procedure, preliminary studies are undertaken to

estimate the maximum practical percentage changes permissible in the primary
parameters. The changes that led to impractical transverse sections were identified
together with the upper and lower bounds on the changes in L and B/T permitted in the
optimisation process, corresponding to 10% of the intact Derbyshire values. These
limits are consistent with those used and discussed by Sarioz (1993). Other constraints
imposed were fixed CB , fixed displacement and fixed depth.
6.3.1 Choice of the Objective Function
The objective function selected is the peak relative bow motion in head seas in the
Optistanbul Suite (see Hearn et al. (1992) and Sarioz et al. (1992)) subject to the
constraint that calm water resistance is not increased. The relative bow motion is chosen
since it includes the amplitude and phase information of the heave and pitch motions.
As the Hooke-Jeeves algorithm proceeds, the hull form is modified and the resulting
hydrostatic properties, resistance (wave and frictional) and relative bow motion values
are determined. In this case the process converged after 52 iterations. According to
Figure 10.4 of Sarioz (1993) this iteration number is satisfactory for the convergence of
the optimisation variables. The calm water resistance values together with the selected
objective function magnitudes are plotted against iteration number in Figure 6.3.
121
1.2
RBM / RBM PARENT

0.8
RESISTANCE / RESISTANCE
PARENT
0.6
0
10
15
20
25
30
35
40
45
50
55
Number of Iterations
Figure 6.3: The variation of the objective function and the calm water resistance with
iteration number.
Figure 6.3 indicates that the objective function (RBM of alternative designs nondimensionalised with respect to RBM of intact parent Derbyshire) fluctuates as the
optimisation process proceeds. However, not all these small values of the objective
function are acceptable since they do not satisfy either the constraint of intact stability
requirements or the constraint that calm water resistance is not increased. The apparent
solutions determined just after 15 iterations were unacceptable due to the failure to
satisfy the intact stability requirements.
The body hull forms for the parent hull and for the optimised hull form are presented in
Figure 6.4.
122
Figure 6.4: The parent (continuous) and optimised (dashed) hull forms.
It is clear from Figure 6.4 that as the ship is optimised the section curves generally
expand. Fore body is now more round-formed than the original one. The lines of the aft
body are more V-shaped than the original ones. Maximum draught is now decreased
over the fore and aft body. Since the volume was constant during the optimisation this
means that the volume lost due to the decrease of draught is balanced by the expansion
of the sections half beam distributions along the hull. The depth was one of the fixed
parameters in the optimisation process. Therefore, in the case of the optimised ship, a
decrease in the ships draught resulted in an increase in the freeboard height.
The hull form parameters for the parent (intact Derbyshire) hull form and for the
optimum design are presented in Table 6.4. Comparisons of the relative bow motion
values for these alternative hulls are provided in Figure 6.5.
123
Table 6.4: Parameters for the parent and optimised hull forms.
Geometric Parameters Parent Form Optimised Form

Length (m)
281.94
310.134
Beam (m)
44.196
44.196
Draught (m)
17.97
16.34
Depth (m)
24.994
24.994
Displacement (t)
199480
199480
RBM (m)
2.534
2.382
Resistance (KN)
14617
14580
Service Speed (knots)
15.5
15.5
RBM PARENT
RBM OPTIMISED
2.5
1.5
0.5
0
0
0.5
1.5
/L
2.5
3.5
Figure 6.5: Relative bow motion responses for the parent and optimised Derbyshire
from Optistanbul Suite.
Whereas the peak RBM has been reduced by 6% and the resistance has decreased by
0.25% the improved RBM and calm water resistance in the optimised hull form
corresponds to a +10% change in L and +9.46% change in B/T. The draught for the
parent hull form in Table 6.4 is slightly different to the original value quoted in Table
124
6.1. This change reflects the mass distribution consistent with change of location of
ballast water and fuel to establish zero trim angle in Section 6.2.
6.4 IMO Intact Stability Requirements and Related Analysis

The Derbyshire was a B-60 designed ship, namely a ship of type B with a 60%
reduction in the allowable freeboard. In the damage case the requirement was onecompartment damage that resulted in no unprotected or protected openings being
compromised rather than any requirement concerning the margin line (or line of
protected openings, which is applicable to passenger and special purpose ships).
As already indicated in Chapter 2 the optimised hull form has been checked for some
satisfaction of the IMO intact stability rules. The actual full IMO (A-749 criteria)
requirements are as follows:
The area under the righting lever curve (GZ curve) up to 30 angle of heel
should be equal to/or greater than 0.055 m.rad.
The area under the GZ curve between 30 and 40 should be equal to/or greater
than 0.03 m.rad.
The area under the GZ curve up to 40 should be equal to/or greater than 0.09
m.rad.
Initial GM should be equal to/or greater than 0.15m.
The maximum righting arm should occur at an angle of heel preferably

exceeding 30 but not less than 30 .
125
The righting lever GZ should be at least 0.2m at an angle of heel equal to/or
greater than 30 .
The first three requirements of A-749 criteria are included in the optimisation process as
an intact stability check for the optimised hull form. To cross check the intact stability
curves of the parent and optimised hull forms two different computer algorithms are
applied, the Wolfson unit hydrostatics and stability program (Wolfson (2001a)) and
the Optistanbul suite. The resulting intact stability curves are illustrated in Figure 6.6.
The Optistanbul suite only provides GZ curve up to 40 since this is the largest angle
considered in the first three requirements of the IMO criteria and within an optimisation
process one undertakes sufficient calculation not more than necessary. Furthermore the
ships within the Optistanbul suite are defined using just 21 two-dimensional transverse
sections, whereas the Wolfson unit hydrostatics and stability program has a threedimensional description of the entire hull. Within the Optistanbul suite constrained
cubic spline interpolation (see Kruger (2004)) is applied and within the Wolfson unit
hydrostatics and stability program traditional cubic spline technique is used (see
ShipShape (1992)). Thus in viewing Figure 6.6 one is not comparing like with like in
terms of the geometric description or the calculation performed. Here the alternative
Wolfson unit hydrostatics and stability program is utilised to simply demonstrate that
the intact Derbyshire (parent) hull form complies with all the requirements of the IMO
criteria cited.
Details related to areas under the intact stability curve for both two hulls are presented
in Tables 6.5 and 6.6 together with the minimum IMO requirements. Clearly, both hulls
far exceed the limits stipulated by IMO irrespective of means of calculation of
quantities required.
126
2.5
Parent Optistanbul
Optimum Optistanbul
Parent Wolfson
Optimum Wolfson
GZ (m)
1.5
0.5
0
0
10
20
30
40
50
60
70
80
90
100
-0.5
-1
-1.5
-2
Angle of Inclination (degrees)
Figure 6.6: The intact stability curves for the parent and for its optimised configuration.
Table 6.5: Intact stability curve areas calculated using Optistanbul suite.
Area under GZ Curve (m.rad)

Up to 30 Degrees
Up to 40 Degrees
Between 30 and 40 Degrees
Parent Form
0.528
0.796
0.268
Optimised Form
0.537
0.792
0.255
Required Condition
0.055
0.03
0.090
Table 6.6: Intact stability curve areas calculated using Wolfson unit hydrostatics and
stability code.
Area under GZ Curve (m.rad)

Up to 30 Degrees
Up to 40 Degrees
Between 30 and 40 Degrees
Parent Form
0.532
0.833
0.301
Optimised Form
0.554
0.921
0.367
Required Condition
0.055
0.03
0.090
Prior to cross checking fulfilment of the last three IMO intact stability requirements it is
worth noting that the initial GM value is calculated automatically in the Optistanbul
suite and determined manually from the Wolfson unit hydrostatics and stability code
GZ curve. The heel angle at which the maximum GZ value occurs and the maximum
127
GZ value must be provided manually from the generated GZ curves. The calculated
values of the initial GM value, the locations and values of the maximum GZ are
provided in Tables 6.7 and 6.8.
Table 6.7: Calculated IMO intact stability curve properties from Optistanbul suite.
Hull Forms
Parent Hull Form
RBM Optimised Hull Form
Required Condition
Initial GM (m) Angle of the Maximum GZ (degrees)

3.95
25
3.98
25
0.15
30
Maximum GZ (m)
1.687
1.721
0.2
Table 6.8: Calculated IMO intact stability curve properties from Wolfson unit
hydrostatics and stability code.
Hull Forms
Parent Hull Form
RBM Optimised Hull Form
Required Condition
Initial GM (m) Angle of the Maximum GZ (degrees)

4
30
4.2
40
0.15
30
Maximum GZ (m)
1.753
2.095
0.2
Although the two different codes yield different values for the magnitude and location
of maximum GZs and for initial GMs, according to the cited IMO criteria, both hull
forms significantly exceed the minimum limits for initial GM and for the maximum
value of GZ. Table 6.8 is sufficient to demonstrate that the intact Derbyshire hull form
and its optimised hull form fully comply with the IMO criteria. Further development of
the Optistanbul suite might address the complete rather than a subset of the IMO
requirements.
The equilibrium search code developed could be improved to determine the critical KG,
namely the maximum value at which all the IMO stability requirements are satisfied for
a particular condition. This would provide an additional single measure to compare the
performance of the optimised hull against the basis design.
128
6.5 Floodable Length Curve and Bulkhead Locations

Whereas longitudinal bulkheads are generally omitted in the design of O.B.O (Oil-BulkOre) ships a large number of transverse bulkheads are used in their design. The number
of necessary transverse watertight bulkheads depends upon the ship dimensions and is
determined according to classification regulations (see, for example, Friis et al. (2002),
Lloyds Register (2003), American Bureau of Shipping (1992) and Bureau Veritas
(1986)). However, there is no limitation in the total number of bulkheads for the
Derbyshire or other ships longer than 198m.
According to the cited regulations ships other than passenger ships should have a
collision bulkhead, an after peak bulkhead and a watertight bulkhead at each end of the
machinery space. There are limits only on the location of the collision bulkhead from
the fore perpendicular. This distance has to be more than 10m or 0.05 of the 96% of the
waterline length of vessel and this waterline length should be measured at 85% of least
moulded depth. This distance should also be less than 0.08 of the 96% of the waterline
length of vessel. Due to lack of technical information this distance is taken as 96% of
the length between perpendiculars of the vessel. Therefore, the position of the collision
bulkhead for the Derbyshire should be between 10m and 21.65m. For both the parent
and optimised Derbyshire this condition is satisfied as the position of the collision
bulkhead stands in the region of 10 to 11 metres from the fore perpendicular in these
two ships.
The transverse bulkheads of the intact Derbyshire are maintained and within the
optimised form of the Derbyshire the positions of the bulkheads are proportionally
changed in accordance to the increased length of the optimised ship. Since LCF, LCB
and CWP have not be modified and hence LCG cannot be modified, if level trim is to be
maintained in the optimised hull form, then some adjustment of the equivalent point
masses and the relationship between parent and optimised hull form geometry requires
some clarification. In the Figure 6.7 the relative position of the alternative geometric
129
hull form relative to the fixed Z-X datum plane together with hull form parameter
changes noted in the preceding subsections are captured systematically.
MAP MP
MFP MFP
Parent Hull
LCF
MAS
MS
MFS
MFS
Optimised Hull
Figure 6.7: The presentation of the four point masses for the parent and optimised hull
forms.
, M FP
, M AS , M FS
, Z A and Z F are found in exactly the same way as M AP , M FP ,
M AP
M AS , M FS , Z A and Z F with mass distribution of the intact Derbyshire stretched to new
length.
For the optimised ship the point mass and inertia radii data corresponding to Tables 6.2
and 6.3 are provided in Tables 6.9 and 6.10 respectively.
Table 6.9: The four masses and their longitudinal positions for optimised Derbyshire.
MA
106709.6 (t)
MF
92770.8 (t)
ZA from LCG 61.9 (m)

ZF from LCG 71.2 (m)
MAP = MAS
53354.8 (t)
MFP = MFS
46385.4 (t)
130
Table 6.10: The horizontal and vertical positions of the four masses for optimised
Derbyshire.
Equation Quantity Determined

XP =17.945 (m)
5.38
XS =-17.945 (m)
5.30
Radii of Gyration
kYY = 0.2217
kYY = 0.2217
5.42
YAS = YFS = 1.7868 (m)
kZZ =0.41
5.45
YAS = YFS = 1.7868 (m)
kXX = 0.2142
The movement of the bulkheads in the optimised hull form, as a result of the ship
lengthening, now needs to be justified in terms of their acceptability through
consideration of the floodable length curves estimated using the Wolfson unit floodable
length software (Wolfson (2001b). Using a typical permeability coefficient of 0.63 in
the cargo regions and a permeability of 0.85 in the engine rooms floodable length
curves for parent and optimised hulls are presented in Figure 6.8. In Figure 6.8 the dashdot broken line indicates optimised hull form and the continuous line represents the
parent hull form.
As initially discussed in Section 4.2.1 and illustrated in Figure 4.1, bulkhead divisions
of the original and optimised Derbyshire hull forms are acceptable at the different
permeabilities investigated as the clearance between the floodable length curves and the
constructed compartment triangles is more than acceptable.
131
160
140
_____ Original Bulkhead

New Position of Bulkheads
=0.63 Optimised
120
80
60
=0.85 Optimised
40
=0.63 Parent
=0.85 Parent
20
0
Figure 6.8: The floodable length curves for the parent and optimised hull forms for
different levels of permeability.
6.6 Damage Properties of the Derbyshire

As indicated in Section 3.1 bulk carriers are more vulnerable to grounding damage. In
this case the damage position is generally in the region designated 3L/4-FP (see Figure
3.5). The extent of the damage is readily identified from Figure 3.14(a), that is, with
equal possibilities there are three possible length scales:
0 < l / L 0.02
0.02 < l / L 0.04
0.04 < l / L 0.06
132
For the damage height there are two possibilities (see Figure 3.14(b)) with the same
probability of occurrence, namely
0 < h / D 0.2
0.2 < h / D 0.4 .
The most likely damage penetration is between 0 < b / B 0.2 (see Figure 3.14(c)) and
the vertical position of the damage is most likely to satisfy 0 < Y / D 0.2 (see Figure
3.14(d)). The damage is to be created on the starboard side of the ship (see Figure
3.14(e)).
Using the most likely damage characteristics four distinct scenarios will be considered
as defined in Table 6.11.
Table 6.11: Damage scenarios.
Scenario
A
B
C
D
l/L
0.02
0.06
0.06
0.06
h/D
0.2
0.2
0.4
0.4
b/B
0.2
0.2
0.2
0.2
Y/D
0.2
0.2
0.2
0.2
Hold
2
2
1&2
2&3
The possibility of three holds being simultaneously damaged is unlikely using the
statistics available to this study. Whilst more than two holds damaged is unlikely to
appreciate the impact of more damaged holds on the attitude of the ship results are
provided in Appendix E. However, only one or two hold damaged will be considered
here and in subsequent chapters.
The fact that the scenarios suggested in Table 6.11 are based on grounding data it is
perhaps surprising that Y/D the vertical position of the lowest point of the damage has a
133
value of 0.2. However, these are the findings of all reported accidents over the period
1935 and 1999.
From Figure 3.7(c) one notes that the next probable regions of damage after the selected
3L/4-FP region are the regions L/4-amidships and amidships-3L/4. To confirm that the
selected 3L/4-FP region is the more important region, holds in these other cited regions
were damaged and found to yield insignificant trimming moments and hence will not be
considered further. Also an analysis of bulk carriers in collision incidents indicated that
the most likely location of damage stands in the region L/4-amidships. As discussed
earlier this results in an insignificant trimming moment.
Having indicated why the selected scenarios are worth of further in-depth analysis the
next task is to provide details of attitude changes in both the original and optimised
Derbyshire hull forms as each scenario is implemented.
6.7 The Damage Analysis

Each of the damaged ship scenarios presented in Table 6.11 results in a change in heel
and trim attitude of the ship. The different changes in trim and heel angles lead to
different equilibrium positions of the damaged hull. The parallel sinkage of the ship is
calculated first. Then the trim angle is determined before the heel angle is predicted.
Justification of the equilibrium angles found is presented in Appendix F.
6.8 Orientation of Damaged Hull Forms

The damages created in the hull form of the Derbyshire and in its optimised hull form
are relatively minor and are therefore modelled as a change in mass distribution in the
damaged region(s). Since there are no longitudinal bulkheads in the Derbyshire (apart
from the engine room) water entering the damage area creates only trim. Therefore only
134
structural loss to the damage area causes heel. Parallel sinkage, heel and trim following
the damage in the hull form of the Derbyshire and of its optimised design are presented
in Tables 6.12, 6.13, 6.14 and 6.15 for the damaged scenarios A, B, C and D as defined
in Table 6.11.
Table 6.12: Parallel sinkage, trim and heel angles for damage Scenario A.
Attitude of the Damaged Structure Parent Form Optimised Form

Parallel Sinkage (m)
1.037
0.939
Trim Angle (degrees)
0.964
0.796
Heel Angle (degrees)
1.224
1.223
Table 6.13: Parallel sinkage, trim and heel angles for damage Scenario B.

1.243
1.130
1.192
1.000
1.287
1.265
Table 6.14: Parallel sinkage, trim and heel angles for damage Scenario C.

1.992
1.811
1.875
1.551
2.000
1.500
Table 6.15: Parallel sinkage, trim and heel angles for damage Scenario D.

2.035
1.851
1.470
1.216
2.000
1.500
135
The reason for increased parallel sinkage and increasing trim and heel angles as
scenario A passes to scenario B and then to scenario C is the increase in the dimensions
of the damaged region. In scenarios C and D the heel angles are the same as the damage
properties are identical. However, there is a decrease in the trim angle in D as the
distance of the damaged holds from the LCF is reduced. The increased parallel sinkage
for damage scenario D is due to the larger breadth of hold 3 relative to hold 1.
The optimised ship has a longer length and a higher beam to draught ratio than the
parent hull form. Consequently the parallel sinkage, trim and heel angles are quite
sensibly smaller due to the changes in the hull form dimensions.
As indicated in Section 6.5 (Figure 6.7) the equivalent point masses for the Derbyshire
and its optimised hull form are quite distinct (see Appendix C for parent Derbyshire
derivation and Table 6.9 for corresponding values of the optimised hull form). To
determine the motions of the two hull forms for the cases of interest the location of the
centre of gravity and the new location of the point masses are required to provide
structural cross-terms and products of inertia of generalised mass matrix. The centre of
gravity details are provided in Tables 6.16 and 6.17 for the parent and optimised hull
forms. The corresponding pure moments and cross-products of inertia data is provided
in Appendix G. The new location of the point masses are provided in Tables 6.18-6.25.
136
parent Derbyshire.
Cases
Intact
Scenario A
Scenario B
Scenario C
Scenario D
X
0.0000
0.0961
0.1059
0.1917
0.1925
Y
-3.4158
-4.4958
-4.7118
-5.4883
-5.5137
Z
2.6527
2.5775
2.5553
2.4744
2.5120
optimised Derbyshire.
Cases
Intact
Scenario A
Scenario B
Scenario C
Scenario D
X
0.0000
0.0590
0.0654
0.0960
0.0967
Y
-1.7868
-2.7618
-2.9619
-3.6670
-3.6920
Z
2.6527
2.6146
2.6014
2.5544
2.5749
Table 6.18: Position of the point masses for damage scenario A for the parent
Derbyshire.
XP,YAP,ZA
17.8640
-3.0375
-61.5135
XP,YFP,ZF
17.9078
-5.0895
60.4692
XS,YAS,ZA
-17.7179
-3.7977
-61.5135
XS,YFS,ZF
-17.6741
-5.8498
60.4692
137
Table 6.19: Position of the point masses for damage scenario B for the parent
Derbyshire.
XP,YAP,ZA
17.8664
-2.9790
-61.5309
XP,YFP,ZF
17.9234
-5.5163
60.4427
XS,YAS,ZA
-17.7146
-3.7783
-61.5309
XS,YFS,ZF
-17.6576
-6.3156
60.4427
Table 6.20: Position of the point masses for damage scenario C for the parent
Derbyshire.
XP,YAP,ZA
17.9026
-2.7713
-61.5913
XP,YFP,ZF
18.0419
-6.7606
60.3433
XS,YAS,ZA
-17.6657
-4.0134
-61.5913
XS,YFS,ZF
-17.5264
-8.0027
60.3433
Table 6.21: Position of the point masses for damage scenario D for the parent
Derbyshire.
XP,YAP,ZA
17.9193
-3.2493
-61.5669
XP,YFP,ZF
18.0285
-6.3771
60.3929
XS,YAS,ZA
-17.6490
-4.4913
-61.5669
XS,YFS,ZF
-17.5398
-7.6192
60.3929
Table 6.22: Position of the point masses for damage scenario A for the optimised
Derbyshire.
XP,YAP,ZA
17.9815
-1.5190
-59.2795
XP,YFP,ZF
18.0210
-3.3677
73.8077
XS,YAS,ZA
-17.9003
-2.2850
-59.2795
XS,YFS,ZF
-17.8608
-4.1337
73.8077
138
Table 6.23: Position of the point masses for damage scenario B for the optimised
Derbyshire.
XP,YAP,ZA
17.9822
-1.4857
-59.2892
XP,YFP,ZF
18.0335
-3.8081
73.7905
XS,YAS,ZA
-17.8991
-2.2781
-59.2892
XS,YFS,ZF
-17.8478
-4.6004
73.7905
Table 6.24: Position of the point masses for damage scenario C for the optimised
Derbyshire.
XP,YAP,ZA
17.9910
-1.5224
-59.3230
XP,YFP,ZF
18.0853
-5.1238
73.7283
XS,YAS,ZA
-17.8867
-2.4619
-59.3230
XS,YFS,ZF
-17.7924
-6.0633
73.7283
Table 6.25: Position of the point masses for damage scenario D for the optimised
Derbyshire.
XP,YAP,ZA
18.0011
-1.9091
-59.3112
XP,YFP,ZF
18.0751
-4.7327
73.7589
XS,YAS,ZA
-17.8766
-2.8486
-59.3112
XS,YFS,ZF
-17.8026
-5.6722
73.7589
The coordinates in Tables 6.16-6.25 are given respect to LCF, intact waterline and
centreline for the parent and optimised Derbyshire.
139
6.9 Summary
In this chapter, the intact stability of the Derbyshire and of its optimised configuration is
investigated. This forms the static part of the analysis. The dynamic part of the research
is presented in the next chapters. Therefore, the hydrodynamic characteristics and the
relative vertical motion response analysis of the parent and optimised hull forms when
intact and damaged are presented in Chapters 7 & 8.
140
7.
HYDRODYNAMIC CHARACTERISTICS OF INTACT AND
DAMAGED SHIPS
The previous chapters established:
How the bulk carrier Derbyshire might be damaged using analysed damage
statistics.
The most likely damage scenarios would not lead to ship loss.
Required intact stability criteria were satisfied for parent and optimised hull
forms.
Determined the orientation and mass-inertia characteristics of original and

optimised intact and damaged hull forms.
The next key step is to undertake the hydrodynamic analysis of each hull form for each
damage scenario in order that the subsequent motion analyses may be undertaken.
Whilst forward speed was included in the optimisation process, it does not necessarily
have to be included in the damaged ship motion analysis. The forward speed of the ship
is not addressed simply because it is assumed that once the ship is damaged it will either
stop to permit passengers and the crew to abandon the ship (if necessary), or, to assess
extent of damage possible low speed advance to harbour. In the latter case forward
speed effect has to be examined. Here the fluid-structure interaction analysis is
undertaken for zero forward speed using the Matthew Diffraction Suite.
The wetted surface boundary elements over the damage area are not removed since the
automatic quantity data checking of the Matthew Diffraction Suite is so through that it
141
would refuse to proceed to the interaction analysis on the basis that the hole in one side
leads to inconsistent estimate of volume. To model the hole it would be necessary to
model the internal wetted volumes and this was considered too complex as a first step in
analysing the damaged ship.
In the following sections the method of calculation will be briefly described together
with an assessment of the quality of the calculations completed prior to discussing the
predictions completed for both intact and damaged parent and optimised hull forms.
7.1 Calculation of Hydrodynamic Coefficients and Excitation Forces

The general fluid-structure interaction problem can be formulated as an elliptic partial
differential equation subject to linear free surface boundary conditions expressing
continuity of pressure and velocity across the free surface. On the structure itself and on
the stationary seabed continuity of normal velocity across these boundaries is required.
In the far-field the radiation and diffraction waves generated are required to be outgoing
cylindrical waves. Since the partial differential equation formulation requires modelling
the fluid domain explicitly alternative Fredholm integral equation formulations have
been developed that only requires solution of the unknown velocity potential on the
wetted surface of the structure (see, for example, Hearn (1991)). This greatly reduces
the computational demands of solving free surface based fluid-structure interaction
problems. The boundary elements (small subdivisions of the wetted surface of a
structure) may be triangular, quadrilateral or polygonal with any mix required to
represent the geometrical form being permitted. The selection of the boundary elements
is generally subjective and hence the quality of the solutions produced for the radiation
potential, j , and the diffraction potentials for different wave frequencies and wave
headings must be checked.
142
In this study the numerical stability of the equations solved is monitored through the
conditioning number provided by the Matthew Diffraction Suite. The quality of
modelling the fluid structure interaction is judged through the quality of the cross-terms
of the reactive coefficients ( Akj = Ajk and Bkj = B jk for zero forward speed) and the
equality of the wave excitation forces and moments calculated directly and calculated
indirectly using the relationship of Haskind (1954).
Without proof or detailed explanations the basic relationships required to indicate how
each hydrodynamic quantity is determined is presented next.
Assuming
( Z , X , Y , t ) = ( Z , X , Y ) e it
(7.1)
and the incident wave potential is described by
W ( Z , X , Y ) =
ag cosh k ( y + d ) ik ( Z cos + X sin )

e
,
cosh(kd )
(7.2)
where is the wave heading, is the wave frequency, a is the wave amplitude and
d is the water depth. Specification of is discussed in Chapter 8.
The wave excitation force/moment in the kth direction is given by

Fk ( , ) = i
+ D ) nk dS .
(7.3)
SW
Alternatively the Haskind relationship uses
Fk ( , ) =
e-it
aj
dS
W
j
n
n
SW
143
(7.4)
where nk is the generalised direction cosine associated with the wave excitation force
(k=1, 2, 3) or moment (k=4, 5, 6), SW is the wetted surface area of the vessel, D is the
diffraction velocity potential and t denotes time.
The hydrodynamic reactive coefficients of added mass and fluid damping are
determined from the radiation velocity potentials j associated with the structure
executed the jth mode of motion. Since the reactive loads will not be in phase with the
incident wave the velocity potential j is complex to model the phase difference. In
particular:
j = ( jr + i ji ) eit
(7.5)
with jr and ji denoting the real and imaginary parts of the jth velocity potential.
The added mass and fluid damping coefficients are essentially resolved components of
the reactive force (moment) in-phase with the structures acceleration and velocity
respectively. These are calculated using the standard relationships:
Akj =
ja
jI
nk dS
(7.6)
nk dS .
(7.7)
SW
and
Bkj =
ja
jR
SW
Irrespective of the geometric characteristics of the ship, independently of whether it is

intact or damaged the zero forward speed hydrodynamic coefficients Akj and Bkj may
144
be mathematically shown to be symmetric. Part of the analysis quality check is to

establish if the calculated reactive coefficients satisfy
Akj = Ajk
(7.8)
Bkj = B jk for all j and k .
(7.9)
and
A mathematical proof of equations (7.8) and (7.9) is provided by Odabasi and Hearn
(1978). The details are not reported here. When the vessel has a forward speed, the
cross-coupling terms must then satisfy the Timman-Newman relationship (see Timman
and Newman (1962)).
7.2 Validation of Hydrodynamic Analysis of Intact and Damaged

Forms of Derbyshire
The wetted surface discretisation must satisfy two distinct aspects. Initially the
definition of the wetted surface should represent the actual geometric form so that
wetted surface and volumetric properties are consistent with hull form characteristics of
original hull. Next the distribution of boundary elements should provide sufficient
definition of the normal velocity components at their centroid to allow good modelling
of the fluid structure interaction problem being investigated.
In this particular case several different forms of the basis hull form and optimised hull
forms are to be analysed. Whereas the intact hull forms exhibit port-starboard symmetry
the damaged hulls do not exhibit such geometric symmetry. The maximum number of
boundary elements available is 1500 for the case of no planes of symmetry, so that
intact and damaged hull forms are modelled consistently.
145
For the intact Derbyshire and the intact optimised hull form of Derbyshire a total of
1314 boundary elements were used; 1294 quadrilateral and 20 triangular elements. The
discretisation for each case is pictorially presented in Figures 7.1 and 7.2.
Figure 7.1: Discretisation of the intact Derbyshire.
146
Figure 7.2: Discretisation of the intact optimised Derbyshire.
As discussed earlier a measure of the numerical stability of the calculations performed

to complete the hydrodynamic calculations is the so-called conditioning number of the
equations solved. This number ideally should have the value of unity. This is a
necessary rather than a sufficient condition. It cannot be sufficient because if one
used considerably fewer boundary elements to model the fluid-structure interaction the
equations solved could be very stable, but the quality of hydrodynamic predictions
could be totally unacceptable because the distribution of the boundary conditions is
insufficient to capture the true nature of the interaction.
147
1.25
CN
0.75
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Frequency (rad / sec)
Figure 7.3: Conditioning number for the intact Derbyshire.
In Figure 7.3 the calculated conditioning number values of the hydrodynamic analysis
for the intact Derbyshire is presented. Clearly the values are very close to the ideal value
of unity and so it is worthwhile moving on to the next step of accessing the equality of
the hydrodynamic cross-coupling reactive coefficients.
The port-starboard geometric symmetry of the intact case means that only the sway-rollyaw and surge-heave-pitch added mass and fluid damping reactive coefficients are
coupled. For the same reason only the heave-pitch hydrostatic restoring coefficients are
coupled. The non-zero horizontal and vertical motions associated cross-coupling
hydrodynamic terms for the intact Derbyshire are presented in Appendix H through
Figures H.1-H.6 respectively. Only sway-roll and roll-sway added mass and fluid
damping coefficients plot is presented in Figure 7.4 (also in Figure H.4).
148
5.0E+04
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
B24
-5.0E+04
B42
-1.0E+05
A24
-1.5E+05
A42
-2.0E+05
Figure 7.4: Sway-roll and roll-sway added mass and fluid damping coefficients for the
intact Derbyshire.
The largest difference for the cross-coupling reactive hydrodynamic coefficients is

found in Figure 7.4 for the sway-roll and roll-sway coefficients. The very close
agreement of all the other cross-terms implies that all hydrodynamic calculations are
stable and further improvement would require more circumferential boundary elements
so as to model more precisely the roll and sway fluid-structure interactions. The number
of circumferential strips of boundary elements is of the order of 75. Hence for symmetry
of the boundary elements (port-starboard) 150 facets would be required to improve the
sway-roll modelling. Since this would conflict with the initial plan to model the internal
free-surfaces for the damaged ships (subject to 1500 upper limit) further refinement was
not undertaken.
The next quality check is the prediction of wave excitation loads using direct calculation
based on Equation (7.3) and the Haskind relationship of Equation (7.4). The surge,
heave and pitch wave excitation loads for the intact Derbyshire calculated using
Matthew Diffraction Suite for head seas are compared in Figures 7.5 to 7.7.
149
1.2E+04
Haskind
Diffraction
1.0E+04
8.0E+03
6.0E+03
4.0E+03
2.0E+03
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.5: Surge wave excitation forces for the intact Derbyshire (in head seas).
1.E+05
9.E+04
8.E+04
Haskind
Diffraction
7.E+04
6.E+04
5.E+04
4.E+04
3.E+04
2.E+04
1.E+04
0.E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.6: Heave wave excitation forces for the intact Derbyshire (in head seas).
150
4.5E+06
4.0E+06
Haskind
Diffraction
3.5E+06
3.0E+06
2.5E+06
2.0E+06
1.5E+06
1.0E+06
5.0E+05
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.7: Pitch wave excitation moments for the Intact Derbyshire (in head seas).
The plots of the head sea wave excitation forces and moments for the intact Derbyshire
associated with surge, heave and pitch that are calculated by using these alternative
methods match very well. This set of results thus suggests that the modelling of surge,
heave and pitch radiation problems is consistent with the modelling of the diffraction
problems.
To demonstrate that the quality of the coupled hydrodynamic coefficients and the
numerical stability of the equations processed is also maintained in the damaged
scenarios the damage scenario A is addressed. The discretisation for the damage
scenario A is pictorially presented in Figures 7.8. In this case the parent Derbyshire hull
form is damaged in the region 3L/4-FP as specified in Figure 3.7 (c) previously. An
indication of the quality of the calculated conditioning number is provided in Figure 7.9.
The vertical plane and horizontal plane based cross-coupling terms are provided in
Figures H.7 to H.9 and Figures H.10 to H.12 respectively. As noted with the intact
parent Derbyshire form it is only the sway-roll cross-terms that exhibit any real
difference as shown in Figure 7.10 or in Figure H.10.
151
Figure 7.8: Discretisation of the damaged Derbyshire scenario A.
CN
1.25
0.75
0
0.1
0.2
0.3
0.4
0.5
0.6
Figure 7.9: Conditioning number for the damaged Derbyshire (scenario A).
152
0.7
1.5E+05
A24
1.0E+05
A42
5.0E+04
B24
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
B42
-5.0E+04
Figure 7.10: Sway-roll and roll-sway added mass and fluid damping coefficient for the
damaged Derbyshire (scenario A).
For completeness the surge, heave and pitch wave excitation loads for the damaged
Derbyshire (scenario A) calculated using Matthew Diffraction Suite for head seas are
compared in Figures 7.11 to 7.13.
153
1.4E+04
Haskind
Diffraction
1.2E+04
1.0E+04
8.0E+03
6.0E+03
4.0E+03
2.0E+03
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.11: Surge wave excitation forces for the damaged Derbyshire scenario A (in
head seas).
1.E+05
9.E+04
8.E+04
Haskind
Diffraction
7.E+04
6.E+04
5.E+04
4.E+04
3.E+04
2.E+04
1.E+04
0.E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.12: Heave wave excitation forces for the damaged Derbyshire scenario A (in
head seas).
154
4.5E+06
4.0E+06
Haskind
Diffraction
3.5E+06
3.0E+06
2.5E+06
2.0E+06
1.5E+06
1.0E+06
5.0E+05
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.13: Pitch wave excitation moments for the damaged Derbyshire scenario A (in
head seas).
Having established numerical stability of the fluid-structure interaction analysis, good

equality of the majority of hydrodynamic cross-coupling coefficients and very good
agreement of wave exciting loads calculated directly and indirectly (Haskind) the
hydrodynamic data based on the discretisation predicted in Figures 7.1, 7.2 and 7.8 (and
those not shown) are sufficient to undertake the motion analyses for each of the ten
scenarios.
7.3 Discussion of Hydrodynamic Data for Intact and Damaged Ships

Given that four damage scenarios are to be investigated for each of the alternative intact
Derbyshire hull forms the influence of all damage scenarios upon the hydrodynamic
coefficients is presented next. The variation of the pure hydrodynamic added mass and
fluid damping coefficients for the parent and optimised Derbyshire hull form for the
155
intact and damaged conditions is provided through Figures 7.14 to 7.19 and Figures
7.20 to 7.25 respectively.
There is no a priori theory that allows objective judgement of the correctness, or

otherwise of these calculated pure hydrodynamic reactive coefficients. However, certain
trends are expected in terms of the frequency dependence. As frequency tends to zero
the fluid damping should tend to zero and if provided frequencies investigated are high
enough added mass should tend to a constant (see, for example, Lewis (1989)).
1.9E+04
Intact Parent
Damaged Scenario A Parent
Damaged Scenario B Parent
Damaged Scenario C Parent
Damaged Scenario D Parent
Intact Optimised
Damaged Scenario A Optimised
Damaged Scenario B Optimised
Damaged Scenario C Optimised
Damaged Scenario D Optimised
A11
1.7E+04
1.5E+04
1.3E+04
1.1E+04
9.0E+03
7.0E+03
5.0E+03
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.14: Pure surge added mass coefficient for intact and damaged hull forms.
156
3.5E+05
Intact Parent
Intact Optimised
A22
3.0E+05
2.5E+05
2.0E+05
1.5E+05
1.0E+05
5.0E+04
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.15: Pure sway added mass coefficient for intact and damaged hull forms.
5.0E+05
Intact Parent
Intact Optimised
A33
4.5E+05
4.0E+05
3.5E+05
3.0E+05
2.5E+05
2.0E+05
1.5E+05
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.16: Pure heave added mass coefficient for intact and damaged hull forms.
157
1.10E+07
Intact Parent
Intact Optimised
A44
1.05E+07
1.00E+07
9.50E+06
9.00E+06
8.50E+06
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.17: Pure roll added mass coefficient for intact and damaged hull forms.
Intact Parent
Intact Optimised
A55
3.0E+09
2.5E+09
2.0E+09
1.5E+09
1.0E+09
5.0E+08
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.18: Pure pitch added mass coefficient for intact and damaged hull forms.
158
Intact Parent
Intact Optimised
A66
2.1E+09
1.9E+09
1.7E+09
1.5E+09
1.3E+09
1.1E+09
8.5E+08
6.5E+08
4.5E+08
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.19: Pure yaw added mass coefficient for intact and damaged hull forms.
6.E+03
Intact Parent
Intact Optimised
B11
5.E+03
4.E+03
3.E+03
2.E+03
1.E+03
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.20: Pure surge fluid damping coefficient for intact and damaged hull forms.
159
1.4E+05
B22
1.2E+05
Intact Parent
Intact Optimised
1.0E+05
8.0E+04
6.0E+04
4.0E+04
2.0E+04
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.21: Pure sway fluid damping coefficient for intact and damaged hull forms.
Intact Parent
Intact Optimised
9.E+04
B33
8.E+04
7.E+04
6.E+04
5.E+04
4.E+04
3.E+04
2.E+04
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.22: Pure heave fluid damping coefficient for intact and damaged hull forms.
160
7.E+05
Intact Parent
Intact Optimised
B44
6.E+05
5.E+05
4.E+05
3.E+05
2.E+05
1.E+05
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.23: Pure roll fluid damping coefficient for intact and damaged hull forms.
6.E+08
Intact Parent
Intact Optimised
B55
5.E+08
4.E+08
3.E+08
2.E+08
1.E+08
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.24: Pure pitch fluid damping coefficient for intact and damaged hull forms.
161
8.E+08
Intact Parent
Intact Optimised
B66
7.E+08
6.E+08
5.E+08
4.E+08
3.E+08
2.E+08
1.E+08
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.25: Pure yaw fluid damping coefficient for intact and damaged hull forms.
Since relative vertical motion (see Chapter 8) is determined from the heave, roll and
pitch motions the pure hydrodynamic reactive coefficients associated with these degrees
of freedom are commented first.
Figures 7.16 & 7.22 indicate that it is the pure heave fluid damping that is sensitive to
the optimisation and damage changes. The heave added mass increases due to the
optimisation but is generally insensitive to change associated with damage. Vassalos &
Jasionowski (2002) observed similar insensitivities in heave added mass in their
investigation.
For pure roll, Figures 7.17 & 7.23 demonstrate that roll added inertia is more sensitive
to hull form geometry changes than wave frequency, whereas fluid damping is sensitive
to both geometry and incident wave frequency variation. The damping almost
monotonically increases after an initial slow variation with wave frequency for each
geometric form. In the Vassalos & Jasionowski (2002) study the roll hydrodynamic
coefficients (unlike heave) also showed some dependency on the damaged vessels
162
attitude. In this study the pure roll (as well as pure surge, sway and yaw) added and
fluid damping coefficients increase for the optimised hull as one considers damage case
A and then B, and then damage case D with further increases for case C. Here the
differences in parallel sinkage, trim and heel (although small) are apparently sufficient
to generate a measurable difference.
For pure pitch, Figures 7.18 & 7.24, the pitch fluid damping (like heave) increase when
the structure is optimised with maximum increases for intact case. The increases in pure
pitch added inertia are more significant than the heave added mass changes.
When the structure is optimised B/T increases from 2.45 to 2.69. As Lewis (1989)
points out a structure with the same cross sectional area but of beamier form will
generate more fluid disturbance than a narrow section. Hence it is the B/T increases that
it is thought to increase the pure vertical motion added mass hydrodynamic coefficients
as a result of optimisation. B/T increases is also considered the underpinning reason for
reduced surge and sway added mass coefficients as the structure is optimised (Newman
(1977)).
The sensitivities of the hydrodynamic yaw coefficients, Figures 7.19 & 7.25, are quite
worked due to optimisation and different damage scenarios.
The cross-coupled added mass and fluid damping coefficients for the intact and
damaged parent and optimised Derbyshire hull forms are plotted in Figures 7.26 to 7.31
and Figures 7.32 to 7.37 respectively to indicate the sensitivity of the various
coefficients to optimisation and damage. The hydrodynamic coefficients plotted
correspond to those that are non-zero in the intact case.
163
9.E+03
Intact Parent
Intact Optimised
A13
8.E+03
7.E+03
6.E+03
5.E+03
4.E+03
3.E+03
2.E+03
1.E+03
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.26: Heave induced surge added mass for the intact and damaged hull forms.
4.0E+06
Intact Parent
Intact Optimised
A15
3.5E+06
3.0E+06
2.5E+06
2.0E+06
1.5E+06
1.0E+06
5.0E+05
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.27: Pitch induced surge added mass for the intact and damaged hull forms.
164
2.E+05
A24
1.E+05
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Intact Parent
Intact Optimised
-1.E+05
-2.E+05
-3.E+05
-4.E+05
Figure 7.28: Roll induced sway added mass for the intact and damaged hull forms.
1.E+07
A26
Intact Parent
Intact Optimised
8.E+06
6.E+06
4.E+06
2.E+06
0.E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-2.E+06
Figure 7.29: Yaw induced sway added mass for the intact and damaged hull forms.
165
2.0E+06
A35
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-2.0E+06
-4.0E+06
Intact Parent
Intact Optimised
-6.0E+06
-8.0E+06
-1.0E+07
-1.2E+07
Figure 7.30: Pitch induced heave added mass for the intact and damaged hull forms.
3.5E+07
Intact Parent
Intact Optimised
A46
3.0E+07
2.5E+07
2.0E+07
1.5E+07
1.0E+07
5.0E+06
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.31: Yaw induced roll added mass for the intact and damaged hull forms.
166
1.6E+03
Intact Parent
Intact Optimised
B13
1.4E+03
1.2E+03
1.0E+03
8.0E+02
6.0E+02
4.0E+02
2.0E+02
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-2.0E+02
Figure 7.32: Heave induced surge fluid damping for the intact and damaged hull forms.
1.2E+06
B15
1.0E+06
8.0E+05
Intact Parent
Intact Optimised
6.0E+05
4.0E+05
2.0E+05
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 7.33: Pitch induced surge fluid damping for the intact and damaged hull forms.
167
4.0E+04
B24
2.0E+04
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-2.0E+04
Intact Parent
Intact Optimised
-4.0E+04
-6.0E+04
-8.0E+04
-1.0E+05
-1.2E+05
-1.4E+05
-1.6E+05
Figure 7.34: Roll induced sway fluid damping for the intact and damaged hull forms.
3.5E+06
B26
Intact Parent
Intact Optimised
3.0E+06
2.5E+06
2.0E+06
1.5E+06
1.0E+06
5.0E+05
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-5.0E+05
Figure 7.35: Yaw induced sway fluid damping for the intact and damaged hull forms.
168
1.0E+06
B35
5.0E+05
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
-5.0E+05
0.8
0.9
Intact Parent
Intact Optimised
-1.0E+06
-1.5E+06
-2.0E+06
Figure 7.36: Pitch induced heave fluid damping for the intact and damaged hull forms.
1.4E+07
B46
1.2E+07
1.0E+07
Intact Parent
Intact Optimised
8.0E+06
6.0E+06
4.0E+06
2.0E+06
0.0E+00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-2.0E+06
Figure 7.37: Yaw induced roll fluid damping for the intact and damaged hull forms.
Figures 7.26 & 7.27 provide the heave and pitch induced surge added mass and Figures
7.32 & 7.33 present the corresponding fluid damping. The heave induced surge
169
coefficients are significantly more sensitive to optimisation and damage changes than
the pitch induced surge coefficients. The intact and damaged heave induced surge added
mass and fluid damping coefficients are quite distinct with the intact values lower.
The roll and yaw induced sway added mass and fluid damping coefficients of Figures
7.28 & 7.29 and Figures 7.34 & 7.35 are well spread out. Apart from the largest
predicted values for the parent Derbyshire hull form for damage scenario A the roll
induced sway results partitioned into the lowest set of values for the damaged nonoptimised hull form, the intact parent and the damaged optimised scenario D values and
then the intact optimised and the damaged optimised scenarios A, B & C hull form
values. For the yaw induced sway coefficients there is a different but common pattern.
Starting with the lowest values for the intact parent Derbyshire hull form the next set are
the damaged parent hull form followed by the optimised intact and damaged hull form
values.
The pitch induced heave coefficient values in Figures 7.30 & 7.36 are quite distinct
again. In each case there are three distinct groupings of results. The lowest values are
associated with the intact optimised hull form. Next the damaged optimised hull form
values form a narrow tight band of variation with the third set form by the parent hull
form both intact and damaged.
Finally, there are the yaw induced roll coefficients of Figures 7.31 & 7.37 which divide
into the intact ship values and the damaged ship values. For the added mass there are
five pairs of curves. The lowest values are associated with the intact parent and
optimised hull forms, the second pair is formed of the parent and optimised hull form
values of Scenario A, the third pair is formed of the parent and optimised hull form
values of Scenario B, the fourth pair is the parent and optimised hull form values of
Scenario D and finally the fifth pair is the parent and optimised hull form values of
Scenario C. For the fluid damping curves the order of the grouping is similar with few
exceptions.
170
In general terms it appears that:
The shape of the added mass curves is similar in trend for intact and damaged
hull forms.
Apart from a few exceptions the fluid damping coefficients increase as a

consequence of optimisation.
For added mass the heave induced surge value decrease as result of optimisation,
whereas they increase for the pitch-surge, yaw-sway and pitch-heave added
masses.
Generally added mass coefficient increases with damage are associated with
moving from scenario A to scenario B, then B to D and finally D to C.
Prior to undertaking the formulation and solution of the relative vertical motion set out
in Chapter 8 and Appendix I completeness requires provision of the various hydrostatic
restoration coefficients. Data for each of the 10 scenarios to be investigated are provided
in Tables 7.1 to 7.10.
Table 7.1: Hydrostatic coefficients for the intact case for the parent Derbyshire.
Hydrostatic Coefficients Heave

Roll
Pitch
5
5.27x105
1.16x10 0.00
Heave
1.36x107 0.00
Roll
0.00
7.09x108
5.27x105 0.00
Pitch
171
Table 7.2: Hydrostatic coefficients for the damage case scenario A for the parent
Derbyshire.

1.16x105
Heave
Roll
Pitch
4
-1.17x10 2.24x105
Roll
-1.17x104 1.22x107
Pitch
2.24x105
-9.27x104
-9.43x104 7.10x108
Table 7.3: Hydrostatic coefficients for the damage case scenario B for the parent
Derbyshire.

1.16x105
Heave
Roll
Pitch
4
-1.56x10 2.32x105
Roll
-1.56x104 1.73x107
Pitch
2.32x105
-1.05x105
-9.27x104 7.09x108
Table 7.4: Hydrostatic coefficients for the damage case scenario C for the parent
Derbyshire.

1.16x105
Heave
Roll
Pitch
4
-2.75x10 8.24x103
Roll
-2.75x104 1.97x107
Pitch
8.24x103
-1.13x105
-1.20x105 7.23x108
Table 7.5: Hydrostatic coefficients for the damage case scenario D for the parent
Derbyshire.

1.17x105
Heave
Roll
Pitch
4
-2.71x10 1.34x105
Roll
-2.71x104 2.00x107
Pitch
1.34x105
172
-8.12x104
-9.21x104 7.29x108
Table 7.6: Hydrostatic coefficients for the intact case for the optimised Derbyshire.

1.27x105
Heave
Roll
Pitch
0.00
-3.00x106
Roll
Pitch
-3.00x106
0.00
1.05x107 0.00
1.01x109
0.00
Table 7.7: Hydrostatic coefficients for the damage case scenario A for the optimised
Derbyshire.

1.28x105
Heave
Roll
Pitch
4
-1.19x10 -1.55x106
Roll
-1.19x104 1.28x107
6.73x104
Pitch
-1.55x106 6.17x104
9.69x108
Table 7.8: Hydrostatic coefficients for the damage case scenario B for the optimised
Derbyshire.

1.28x105
Heave
Roll
Pitch
4
-1.22x10 -1.63x106
Roll
-1.22x104 1.33x107
4.81x104
Pitch
-1.63x106 4.13x104
9.74x108
Table 7.9: Hydrostatic coefficients for the damage case scenario C for the optimised
Derbyshire.

1.28x105
Heave
Roll
Pitch
4
-1.70x10 -1.81x106
Roll
-1.70x104 1.51x107
1.36x105
Pitch
-1.81x106 1.27x105
9.88x108
173
Table 7.10: Hydrostatic coefficients for the damage case scenario D for the optimised
Derbyshire.

1.29x105
Heave
Roll
Pitch
4
-1.70x10 -1.68x106
Roll
-1.70x104 1.54x107
1.39x105
Pitch
-1.68x106 1.28x105
9.90x108
As expected cross-coupling of the restoration terms, in particular for the heave-roll and
roll-pitch hydrostatic coefficients occurs when the structure is damaged (see Tables 7.27.5 and Tables 7.7-7.10) and does not occur when the structure is intact. Irrespective of
geometric characteristics the cross-terms should be symmetric. In general the crossterms agree quite well, but significant figures are being lost for those terms one or two
magnitudes smaller.
While the hydrostatic restoration coefficients in pure heave do not vary so much the
other coefficients change in magnitude from one case to another.
7.4 Summary
Having argued that the numerical stability of the fluid-structure interaction is more than
satisfactory and noted that for some degrees of freedom hydrodynamic coefficients are
slightly sensitive to geometric and damage changes the next stage is to provide relative
motion responses to ascertain whether the optimised damaged ships have less pleasing
motion characteristics than the parent damaged hull forms.
174
8.
RELATIVE VERTICAL MOTION ANALYSIS OF INTACT AND
DAMAGED SHIPS
With the hydrodynamic, hydrostatic and mass-inertia details for all ten scenarios
available the motion equations formulated in Chapter 5 may be solved. The components
of the motion amplitudes in each case for each degree of freedom have real and
imaginary parts. Whilst some typical transfer functions are presented it is more
meaningful to consider vector relationships such as relative vertical motion or
accelerations as they combine both the amplitude and phase data determined. The
transfer functions merely define individual amplitude envelops, whereas the other cited
qualities provide measures of what a human being or machinery will actual experience
on board of the ship.
The significance of the scalar quantity of Equation (7.2) will be highlighted prior to
presentation of the motion results. The determination of the different motion qualities
and the selection of their location are addressed next.
8.1 Motion Dependent Qualities of Engineering Interest

Each degree of freedom is described as
j = ja e it ( jr + i ji ) e it ,
(8.1)
with subscripts r and i denoting real and imaginary parts.
To determine the resultant vertical motion at some generic point (z,x,y) defined relative
to origin located at LCF of ship, the positive sign convention indicated in Figure 5.1 is
175
used. That is, heave is positive upwards, pitch is positive bow down and roll is positive
to starboard. The derivation of the Equation (8.2) is given in Appendix I.
The positive sign convention defined in Appendix I implies that

SV = SVr + iSVi = (3r + i3i ) l p sgn( z ) (5r + i5i ) + lr sgn( x) ( 4r + i4i ) ,
(8.2)
where (z,x) is projected coordinates of selected point in the z-x plane.
8.2 Selection of Points Investigated for Relative Vertical Displacement

When the ship is damaged the ships crew and passengers (if appropriate) will move to
the deck and muster at the appropriate lifeboat stations. Therefore, in general the points
of interest for determining and comparing the relative vertical motion lie on the
perimeter of the deck. In this study the points designated A, B, C, D, E, F & G of Figure
8.1 will be investigated.
x
D
B
A
AP
E
L/4
F
L/4
G
L/4
L/4
FP
Figure 8.1: Relative positions of the points of interest for the parent Derbyshire for
investigating vertical displacement.
176
Solution of the diffraction problem and hence specification of wave excitation loads
requires explicit specification of the incident wave velocity potential. In Chapter 7 the
incident wave velocity potential was defined as
W ( z, x, y ) =
ag cosh k ( y + d ) ik ( z cos + x sin )

e
.
cosh(kd )
(8.3)
The parameter is set to different values within different computer programmes.

Consequently it is a constant source of concern when comparing independently
determined motion details, since influences the associated phase relationship
between the ship motions and the incident wave.
Irrespective of which free-surface boundary condition is selected it may be readily

shown that the incident wave free surface profile associated with the defined velocity
potential is given by
( z , x, t ) = i aeik ( z cos + x sin ) e it = a ( z , x ) e it
(8.4)
where defines the wave direction.
As a consequence of the fluid-structure interactions the actual resulting wave profile

should include the contributions of each of the radiation wave systems generated and
the diffraction wave system for the wave direction of interest. This is not, however,
normal practice in much of ship motion calculations and hence (in common with
observed practice) the free-surface profile will be simply equated to the incident freesurface profile.
In the Matthew Diffraction Suite, used in the earlier three-dimensional hydrodynamic

analysis, = 1 . In the Optistanbul Suite a different description for is used and hence
177
there will not be a one-to-one correspondent because of different initial phase selection
and differences of two-dimensional versus three-dimensional calculations.
The time-independent relative vertical displacement is here defined as

RVM ( z , x ) = SV ( z , x ) a ( z , x ) .
(8.5)
Therefore at location (z,x) it follows that

RVM ( z , x ) = (3r + i3i ) l p sgn( z ) (5r + i5i ) + lr sgn( x) (4r + i4i )
+ a sin ( kz cos + kx sin ) i cos ( kz cos + kx sin ) .
(8.6)
Furthermore, without loss of generality, a = 1 is assumed, because of the linearity of the

motion equations and the linearity of the fluid-structure interaction analysis used.
Setting a = 1 in Equation (8.6) is also equivalent to treating RVM ( z , x ) as a nondimensional relative vertical motion, that is, RVM ( z , x ) / a .
The changes in the coordinates of the points A to G, originally defined on the parent
hull form of the Derbyshire, are initially a consequence of the hull optimisation
undertaken. Thereafter the points take up new positions as a consequence of damage
scenarios A to D being applied to the parent and optimised Derbyshire hull forms. The
equivalent positions of the selected points are defined in Tables 8.1 to 8.3.
178
Table 8.1: Location of the points of interest on intact hull forms.
Point
A
B
C
D
E
F
G
Intact Derbyshire
(140.102,0.000,0.000)
(69.617,22.098,7.024)
(-0.868,22.098,7.024)
(-71.353,22.098,7.024)
(-71.353,-22.098,7.024)
(-0.868,-22.098,7.024)
(69.617,-22.098,7.024)
Intact Derbyshire Optimised

(168.296,0.000,0.000)
(90.763,21.995,8.654)
(13.229,21.995,8.654)
(-64.304,21.995,8.654)
(-64.304,-21.995,8.654)
(13.229,-21.995,8.654)
(90.763,-21.995,8.654)
Table 8.2: Location of the points of interest for the damaged parent Derbyshire hull
form.
Point Case A
A (140.065,0.073,-3.393)
B (69.708,21.990,5.286)
C (-0.767,21.965,6.471)
D (-71.242,21.939,7.657)
E (-71.242,-22.247,6.713)
F (-0.767,-22.221,5.527)
G (69.708,-22.196,4.342)
Case B
(140.046,0.093,-4.156)
(69.722,21.995,4.827)
(-0.747,21.962,6.293)
(-71.217,21.929,7.759)
(-71.217,-22.256,6.766)
(-0.747,-22.223,5.300)
(69.722,-22.190,3.834)
Case C
(139.962,0.230,-6.571)
(69.745,21.989,3.521)
(-0.703,21.908,5.826)
(-71.150,21.828,8.131)
(-71.150,-22.342,6.588)
(-0.703,-22.261,4.283)
(69.745,-22.181,1.979)
Case D
(140.004,0.196,-5.625)
(69.722,21.973,3.971)
(-0.740,21.910,5.778)
(-71.201,21.847,7.585)
(-71.201,-22.323,6.042)
(-0.740,-22.259,4.235)
(69.722,-22.196,2.428)
Table 8.3: Location of the equivalent points of interest for the damaged optimised
Derbyshire hull form.
Point Case A
A (168.267,0.070,-3.276)
B (90.861,21.852,6.921)
C (13.335,21.829,7.998)
D (-64.191,21.806,9.075)
E (-64.191,-22.174,8.136)
F (13.335,-22.151,7.059)
G (90.861,-22.128,5.982)
Case B
(168.251,0.090,-4.066)
(90.880,21.859,6.423)
(13.358,21.829,7.776)
(-64.163,21.799,9.129)
(-64.163,-22.181,8.157)
(13.358,-22.151,6.805)
(90.880,-22.121,5.452)
179
Case C
(168.186,0.167,-6.363)
(90.915,21.873,4.958)
(13.410,21.818,7.056)
(-64.096,21.763,9.154)
(-64.096,-22.212,8.002)
(13.410,-22.157,5.904)
(90.915,-22.102,3.807)
Case D
(168.219,0.142,-5.420)
(90.887,21.860,5.449)
(13.371,21.817,7.094)
(-64.146,21.774,8.739)
(-64.146,-22.201,7.588)
(13.371,-22.158,5.943)
(90.887,-22.115,4.298)
In subsequent calculations of RVM using Equation (8.6), with a = 1 , the bow point A
(initially located at the undisturbed free surface) and the deck points B & G, C & F and
D & E will essentially be analysed as four distinct cases. Point A is basically the
selected point considered in the optimisation of the relative bow motion.
8.3 Comparison of Relative Vertical Motion for the Selected Points

Optimisation of the Derbyshire was undertaken subject to improving the relative bow
motion without increasing calm water resistance. That is, the relative vertical motion at
point A, where A is the intersection of the undisturbed free surface and the longitudinal
profile of the parent ship. The other points are located on the first deck, in this case the
bulkhead deck.
The basic heave, roll and pitch motions will change as each damage scenario is
investigated. The heave, roll and pitch transfer functions are presented in Figures 8.2 to
8.4. To compare results for the same relative wavelength the incident wave wavelength
for the parent Derbyshire (intact or damaged) is scaled with respect to the parent hull
length of Table 6.4. Similarly for the optimised Derbyshire (intact or damaged) the
incident wave wavelength is scaled with respect to the optimised hull length provided in
Table 6.4.
180
180,Parent Intact
180,Optimised Intact
0.75
180,Parent Damaged Case A

180,Optimised Damaged Case A
180,Parent Damaged Case B
0.5
180,Optimised Damaged Case B

180,Parent Damaged Case C
180,Optimised Damaged Case C
0.25
180,Parent Damaged Case D

180,Optimised Damaged Case D
0
0
/L
Figure 8.2: Heave motion for the intact and damaged parent and optimised hull forms of
Derbyshire.
0.075
180,Parent Intact
0.05

0.025

0
0
/L
Figure 8.3: Roll motion for the intact and damaged parent and optimised hull forms of
Derbyshire.
181
180,Parent Intact
0.0096
0.0064
0.0032
0
0
/L
Figure 8.4: Pitch motion for the intact and damaged parent and optimised hull forms of
Derbyshire.
It is worth noting that in the heave transfer functions peaks and troughs are associated
with
0.5 and
1 and
1 respectively. The roll motion has some peak responses at
1.5 , whereas pitch peaks generally occur for
1.5 .
For heave, roll and pitch the transfer functions for the parent hull represent an upper
bound for the optimised hull transfer functions. For heave only the inseparable
optimised hull form transfer functions for damage scenarios C & D exceed the parent
values. Generally, for pitch the parent intact hull form exceeds all damage scenarios of
the optimised hull form. The lower boundary pitch transfer function is provided by the
intact optimised hull form.
For head seas ( 180 ) the RVM for point A, for all intact and damaged scenarios, is
presented in Figure 8.5.
182
2.5
RVM / Wave Amplitude
180,Parent Intact
1.5

1

0.5

0
0
/L
point A (in head seas).
As might be expected from the resonant wavelengths observed in the pure motion
transfer functions the relative vertical motion at point A exhibits peaks and troughs for
the optimised hull form at
0.5 and
associated with the parent hull form occur for
1 respectively, whereas the troughs
0.5 . At the longer wavelengths the
RVM values for the damaged scenarios lie between an upper boundary formed by the
intact parent RVM values and the lower boundary formed by the intact optimised RVM
values. For intact parent hull form the resonant RVM values for damage scenarios C &
D exceed the intact values and for damage scenarios A & B the resonant RVM values
lie below the intact RVM values. For the intact optimised hull the resonant RVM values
for all damage scenarios lie below the intact RVM values.
Corresponding results for the deck positions B & G, C & F and D & E are provided in
Figures 8.6 to 8.11 for the same head sea wave heading.
183
180,Parent Intact
1.5

1

0.5
0
/L
point B (in head seas).
2.5
180,Parent Intact

1.5


0.5
0
/L
point G (in head seas).
184
For
0.5 the RVM characteristics for point B (Figure 8.6) are similar to those for
point A (Figure 8.5), but at the troughs for
1 there is greater separation of the low
points for point B. Furthermore for damage scenario A imposed on the parent hull the
behaviour is quite different with a peak and trough at
1 and
1.5 respectively.
Ultimately all damage scenario RVM values are bounded above by the intact parent
values and below by the damaged optimised case C RVM values. For shorter
wavelengths,
< 1 , the optimised RVM values exceed those of the parent RVM
values.
At point G (initially the image point of B for the intact case) strong resonances occur for
the parent hull for damage scenarios A, B and C (Figure 8.7). Whilst damage scenario D
does not lead to a resonance it too exceeds the intact parent RVM values. The peaks and
troughs initially associated with damage scenarios C and D (for point B) for optimised
hull have now significantly flattened out and the other peaks and troughs for the
optimised hull form are less separated.
Since the heave and pitch contributions are essentially the same it appears that the roll
influence (even for a 180 heading) is bringing about these changes. To understand
fully why the significant differences between points B and G exist for the parent hull it
is necessary to examine the intermediate calculations. For damage scenario B
comparison of the intermediate calculations for points B and G yield the following
observations:
All contributions to the real and imaginary parts of RVM are negative for point
G. Thus all contributions reinforce to create the larger RVM values for point G.
185
The dominant contribution in the real part of RVM is the roll contribution and in
the imaginary part of RVM it is the incident wave contribution.
Whilst the roll levers for point B & G are comparable the positive roll
contribution to RVM for point B leads to a significantly smaller real part of
RVM.
The smaller imaginary roll motion component means that it is primarily the
incident wave and roll contributions that control the imaginary part of RVM.
The roll motion is an order of magnitude larger than the pitch motion, whereas
the pitch lever to roll lever is approximately 3 to 1. Therefore, the roll
contribution is more significant than the pitch.
Inspection of the intermediate calculations for damage scenario C, for the parent hull,
indicates that peak value is less than that for case B (at point G) because:
Imaginary heave is now positive and larger.
Imaginary pitch is a little larger.
Imaginary roll is slightly smaller.
Otherwise the observations for case B (apart from above minor changes) apply to case
C.
Figures 8.8 & 8.9 provide corresponding results for the pair of points C & F.
186
180,Parent Intact
1.5

1

0.5
0
/L
point C (in head seas).
2.5
180,Parent Intact
1.5
0.5
0
/L
point F (in head seas).
187
For points C & F there is a very small pitch lever (<1m aft of LCF) for the parent hull,
but the roll levers for C & F are comparable with the roll levers for B & G. Hence for
point C the peaks and troughs are still associated with
0.5 and
1 . However,
the differences are small. For the image point F the sign of the roll lever changes
relative to C (as did G relative to B) and consequently the RVM values for the parent
hull under damage scenarios B & A again loom large.
For point F the difference between intact parent and intact optimised hull forms remains
negligible, but now forms the lower boundary of RVM values, whereas for point C it
formed the upper boundary of RVM for
>1.
Next the location points D & E are addressed. The RVM values are presented in Figures
8.10 & 8.11.
2.5
180,Parent Intact
1.5
0.5
0
/L
for point D (in head seas).
188
2.5
Tail Width
2
180,Parent Intact
1.5
0.5
0
/L
for point E (in head seas).
For points D & E the characteristics of the calculated RVM values change again
because of the change in both the sign and size of the pitch associated levers (compared
to points B & G). Whereas for point B the parent hull only exhibited various troughs for
the intact and damaged hull forms, for point D damage scenarios B & C now exhibit
peaks for
1 , a peak and a trough for
0.5 and 1 respectively for damage
scenario A. Thereafter, the parent RVM values are monotonically increasing.
The peak-trough pattern of the parent hull for damage scenario A for point D is now
adopted by all parent hull form cases apart from damage case B for the image point E.
The peak for damage case B is not quite comparable with the peaks for damage
scenarios C & D for the optimised hull form for
1 . For the slightly increased tail
width for point E (compared to that for points C & F) the intact parent RVM values
form a lower boundary for points D and E, whereas the intact optimised RVM values
189
are centrally located with the tail for point E (exceeded by damage scenarios C & D for
the optimised hull) and form an upper boundary for the tail of point D.
In addition to the results presented in Figures 8.5 to 8.11 additional results for points A,
B, C, D, E, F and G are presented in Appendix J as Figures J.1 to J.10, J.11 to J.20, J.21
to J.30, J.31 to J.40, J.41 to J.50, J.51 to J.60 and J.61 to J.70 respectively for the
different wave headings defined in Figure 8.12.
180
157.5
202.5
z
135
225
115
270
90
x
45
0
Figure 8.12: Definition of wave headings.
Figure 8.5, Figures 8.6 & 8.7, Figures 8.8 & 8.9 and Figures 8.10 & 8.11 are repeated in
Appendix J (to provide complete set of results in one location) as Figure J.7, Figures
J.17 & J.67, Figures J.27 & J.57 and Figures J.37 & J.47 respectively.
From the figures in Appendix J and those cited here it is noted that:
All significant changes take place at the wavelengths

190
< 1.5 .
Peak and troughs generally occur at
For large
= 0.5 ,
= 1 and
= 1.5 .
RVM values are asymptotic to 2.
The variation of the RVM characteristics with wave heading is quite different for each
point. To illustrate this point the variation of RVM with wave heading for point A as
presented in Figures J.1 to J.10 is considered.
For a following sea = 0
the intact parent and intact optimised hull form RVM
values provided an upper and lower boundary for
> 1 . The peaks at
0.5 for the
damaged parent hull exceed the intact parent hull values, which have no peak. The
troughs at
1 for the optimised hull for the damaged scenarios exceed the lower
bound intact optimised RVM values.
For a quarter sea = 45
the intact parent and intact optimised hull form RVM
values continue to provide the upper and lower boundary transfer functions.
For a beam sea = 90
) the RVM values are peakier with the
0.5 peaks, for the
optimised hull, the damaged RVM values exceed the intact values. For the
1 peaks
the intact parent RVM values exceed all the damaged RVM values.
For a beam sea ( = 115
all the peaks occur at
0.5 with parent damaged values
forming the highest peaks and the parent intact value forming an upper boundary to the
optimised damaged peaks. Thereafter the parent intact value forms upper boundary at
191
> 1 . The intact optimised RVM values form a lower boundary for almost all
values.
For a bow sea ( = 135
) troughs return at L 0.5 . One could suggest that the troughs
that did not quite develop at = 45 have developed for = 135 otherwise the curves
are quite similar.
For head seas = 157.5 ,180 and 202.5

For a bow sea ( = 225
hull form at
) the trends are quite similar.
the RVM transfer functions have troughs for the optimised
0.5 and peaks for the parent hull form at
For a beam sea = 270
1.
the pattern is analogous to = 90 except the peaks are
generally larger.
For the other points B to G the richness of variation of RVM transfer functions with
wave heading is not diminished. When companying B & G the damaged parent peaks
again arise for reasons similar to these presented when discussing = 180 previously.
Whilst the intact parent and intact optimised RVM values often provide upper and lower
boundaries for RVM values associated with
> 1 and 1.5, it is not the case that they
form the upper or lower boundaries consistently. Hence it is difficult to state that
damage of the optimised hull is less or more beneficial than damage of the parent hull.
192
8.4 Summary
Variation in magnitudes of RVM values for the selected points for different wave
headings are discussed in this chapter. This analysis established:
The results for the three-dimensional analysis of the RBM in head seas (RVM at
point A) for the intact optimised hull and basis hull show that RBM is indeed
better for optimised hull. Hence there is consistency of two-dimensional
optimisation process and three-dimensional hydrodynamic analysis.
All significant changes take place at the wavelengths
Peak and troughs generally occur at
For large
The variation of the RVM characteristics with wave heading is quite different
= 0.5 ,
= 1 and
< 1.5 .
= 1.5 .
RVM values are asymptotic to 2.
for each point.
In the following chapter a summary of the research highlighting the novel aspects of it
is presented.
193
9.
CONCLUSIONS AND FUTURE WORK
Within the literature, optimisation has been presented as a tool developed to assist
designers to test and examine their initial conceptual design by considering variations of
the selected hull form subject to permitted variations being within the control of the
designers. Optimisation is rarely all embracing and hence the development of different
optimisation tools considering different aspects of the complex and multi-tasked design
process has been reviewed within the dissertation.
Whilst satisfaction of intact stability requirements has been built into existing
alternative hull form optimisation packages seeking improved hydrodynamic hull forms
(in terms of seakeeping, calm water resistance and added resistance) damage stability is
not an automated feature. Depending upon knowledge of a particular ship type either a
deterministic or a probabilistic approach to damage stability may be adopted. It was
then argued that within the context of the hydrodynamic hull form optimisation
techniques available their application to novel hull forms would exclude the
probabilistic method. Including the deterministic damage stability analysis in the
optimisation process would require facilitating a full three-dimensional description of
the hull form and its internal subdivisions together with a fast and accurate method of
determining sinkage, trim and heel with automated correct bulkhead subdivision. This
task was considered too involved to include within the optimisation process, since the
steps are quite involved when performing them outside the optimisation process and
also require an objective function review of intermediate calculations performed, to be
sure that damage stability is properly processed as well as required conditions being
fulfilled. This is also the difficulty of building in an automatic selection of the most
likely damage scenarios and hence specifying location and extent of damage. In this
dissertation both MAIB and Lutzen 2002 databases have been accessed to identify the
most likely damage for each ship type in terms of collisions and groundings. Prior to
applying the identified damage extent to the selected bulk carrier Derbyshire the parent
194
hull was optimised for RBM in head seas subject to no increase in calm water resistance
and then the new hydrostatic equilibrium positions determined for each hull form for
each damage scenario. Hence, the fluid-structure interaction analysis for each situation
for a representative selection of wave frequencies and wave directions were undertaken
and the acceptance of the quality of the resulting hydrodynamic characteristics were
justified prior to undertaking general motion analysis and resultant relative vertical
motion at selected corresponding locations on each hull form. The overall approach is
schematically presented in Figure 9.1.
Accident Statistics
Parent Hull Form (Derbyshire)
2-D Optimisation Program
Optimised Intact Hull Form
Parent Intact Hull Form
Damage Introduced
Damage Stability Analysed (Trim, Heel and Parallel Sinkage)
Mass-Inertia Properties Identified for Each case
3-D Hydrodynamic Analysis
Hydrodynamic Properties
3-D Motion Code
Ship Locations of Interest Defined
Motion Responses
Intact
Optimised Hull
Damaged
Optimised Hull
Damaged
Parent Hull
Intact
Parent Hull
Resultant Motion Comparisons Undertaken
Figure 9.1: An overview of the dynamic analysis procedure.
9.1 Key Observation of Completed Research
To compare the motions of a hull in its intact and damage condition, it will be
necessary to determine the products of inertia for at least the damaged ship even
195
if the cross-products of inertia are assumed zero for the intact case. Since no
public domain method was found to complete this task a new method of
producing cross-products of inertia was developed.
The determination of the cross-products of inertia is coupled to the assignment

of the pure moments of inertia. The latter quantity is readily assigned for the
intact case and here, since the trim and heel angles are not excessive, it is
assumed that assignment of the moments of inertia in terms of radius of gyration
can be dealt with similarly for both the intact and damaged ship.
The changes in the amplitudes of the pure and cross-products of inertia for
different damage scenarios are quite sensitive to the position of the damage and
the resulting equilibrium angles.
In the optimisation code the selected objective function is the peak value of the
relative bow motion (RBM) in head seas. Usually, a reduction of the peak value
of the objective function leads to a general reduction in the objective function
amplitudes.
Most of the intact stability properties are improved as the hull is optimised. This
is occurred in this optimisation study where the constraint function was the calm
water resistance and not the intact stability.
The equilibrium angles evaluated for the optimised hull form are smaller than
those associated with the parent hull form. This ought to be advantageous in
terms of passenger and crew comfort.
The damage scenarios carried out in this research represent all the damage cases
possible according to the analysed damage statistics. One-compartment and twocompartment damages are investigated; since three-compartment damage is
196
highly unlikely based on this damage statistics three-compartment damage case

is not analysed in this thesis.
The advantage of using actual collected and recorded ship damage statistics is
that only those parts of the ship that are known to be vulnerable to damage are
investigated.
The damage statistics indicate regions of ship where damage occurs and the
extent of the damage. In generating the damage scenarios one has assumed
(consistent with the statistics) that the damage may affect hold 2 alone and both
hold 1 & hold 2 and hold 2 & hold 3 if damage occurred at interface of two
holds. Since hold 2 contains no cargo it could be completely flooded and this
was identified as the most extreme single hold damaged scenario (case A and
case B). Hold 1 and 3 damaged in isolation was therefore not considered.
Damaging hold 1& hold 2 and hold 2& hold 3 are treated as cases C & D.
Since the grounding damage leads to greater changes in the trimming and
heeling of the damaged ship, compared with data analysed for collisions,
damage imposed on the ship is primarily concerned with the consequences of
the ship grounding.
Fore body of the optimised ship is more round-formed than the original ship.
The lines of the aft body of the optimised ship are more V-shaped than the
original ship. Such a hull form has better RBM characteristics at most headings
and its resistance is decreased.
The numerical modelling of the various ship fluid-structure interactions has been
validated in this thesis through study of conditioning number, direct versus
Haskind excitation loads and symmetry of hydrodynamic reactive cross-terms.
197
The 10% change for L and B/T used leads to practical hull forms that confirms
with intact stability. It is consistent with Hearn et al. (1994), Hearn et al. (1995b)
and Hearn and Wright (1999) who noted that parameter increases beyond this
percentage leads to impractical hull forms.
In this dissertation Derbyshire was selected as a representative bulk carrier

without any intention to understand or proffer reasons for its loss. Furthermore,
it was thought that the official Derbyshire investigation would make available
considerable technical detail in the public domain to allow proper modelling of
the fluid-structure interactions and subsequent motion responses.
The hydrodynamic analysis assumes regular harmonic waves in both the

optimisation process and in the subsequent three-dimensional intact and damage
hull form analysis. Given optimisation of peak value of selected response leads
to general reduction in transfer function then seaway response for optimised hull
will be better than parent hull for all sea-states. Responses in seaways are
therefore not addressed explicitly.
The origin of the coordinate reference system and the point about which motions
are defined is the same point for the parent and optimised hull forms. Hence
meaningful direct comparisons of intact and damaged ships are made possible.
The structural responses of the parent and optimised ships are not examined. In a
damaged state it is assumed that the outer shell of the ship allows water ingress
in accordance with the damage statistics used.
The change of length for the optimised ship necessitated a small change of
bulkhead positions but no increase in the number of bulkheads, so that floodable
length requirements are met by all hull forms analysed.
198
Having summarised the important points of the research completed the next task is to
discuss the novel aspects of the research in greater detail.
9.2 Novel Aspects of Completed Research

The research is concerned with the behaviour of damaged ships and examination of the
possibility that damaged optimised hydrodynamic hull forms exhibit more adverse
seakeeping characteristics than the damaged non-optimised hull form.
The reason for addressing this problem was that different authors have demonstrated
that seakeeping, calm water resistance and added resistance can be improved using
optimisation techniques with automatic fulfilment of those IMO intact stability
requirements included in the process. Since this situation appeared to present a winwin-win situation, with little indication of disadvantages either being discussed or
identified, it would be perverse if damaged optimised hull forms exhibited more
extreme motions than the non-optimised damaged hull form.
To specify the cause and extent of damage any ship type might experience detailed
analysis of damage statistics was considered a necessary first step. The MAIB damage
statistics were found to be lacking in terms of number of accidents reported and the
details recorded. Therefore another data base Lutzen 2002 was thoroughly
interrogated and the damage statistics presented in this dissertation were generated by
the author.
Having identified the ship types most often associated with the most common form of
accidents collisions and groundings a suitable ship was required for investigation.
Whilst a general cargo ship with collision or grounding damage was the obvious
situation to analyse no ship with all the required details for a fluid-structure interaction
199
was available in the public domain. Consequently, Derbyshire a bulk carrier that has
been in the public domain for some time was selected.
Whilst software capable of modelling external and internal fluid-structure interaction

was readily available the subsequent motion analysis required provision of crossproducts of inertia. There is no means of calculation in the public domain other than the
classical text book method of integrating detailed mass distributions. Since the detail
published (even for the Derbyshire) is insufficient a novel method of prediction was
required, developed and implemented. Since the procedure is over specified consistency
checks can be undertaken.
To ensure that the damaged ships represented likely feasible scenarios the damage
statistics were used to generate different damage scenarios. To also ensure that damaged
ships floated in new equilibrium position (rather than sink) floodable length
requirements were fulfilled for each damage scenario. The basis Derbyshire hull and the
optimised Derbyshire hull form fulfilled all IMO intact stability requirements.
As far as the author is aware (as a consequence of searching the literature) this is the
first investigation of either a damaged basis hull or the damaged optimised hull form.
The results for the three-dimensional analysis of the RBM in head seas for the intact
optimised hull and basis hull show that RBM is indeed better for optimised hull. Hence
there is consistency of optimisation process and three-dimensional analysis.
To extend the investigation of the behaviour of the intact and damaged hull forms six
other points on the ship (B to G) were investigated to try and appreciate how heave, roll
and pitch affected relative vertical motion. The results of such analysis were discussed
in Chapter 8. Prior to undertaking such motion studies the quality and appropriateness
of the hydrodynamic fluid-structure interaction analysis was demonstrated.
200
9.3 Recommendations for Future Work
The quality of the external fluid-structure interaction of different intact and

damaged forms of the Derbyshire has been demonstrated. An obvious extensions
is to simultaneously analyse the internal free surfaces (modelled as massless flat
plates with heave, roll and pitch degrees of freedom) and the external fluidstructure interaction. The complexity of the hydrodynamics can be addressed
using the Matthew Diffraction Suite. The increased complexity of the motion
response for scenarios A and B is provided in Appendix K. The extension to
scenarios C and D follows the details provided in Appendix K with two rather
than one massless plate.
The damage statistics clearly relate to ships that survived the incurred damage.
The statistics do not indicate how much more damage the ships could have
experienced and remained afloat. Since the heel and trim angles due to the
damage imposed are small one might argue that damage should be extended
further (despite there being no statistics to support it) to examine how sensitive
the motion responses are to larger heel and trim angles.
Since in the stability analysis the equilibrium angles determined for each of the
four different damage scenarios are small, the question of water on deck does
not arise. If deck wetness did arise it could be readily included in the modelling
of the wetted surfaces. The sloshing of ingressed water in the damaged holds
would not be readily modelled by the massless flat plate model just proposed.
To include sloshing a non-linear free surface model of the internal free surface
would be required to form shape of internal free surface.
For more extreme orientations of the ship analysis of sub-optimised

(intermediate) hull forms should be performed to appreciate impact of different
optimisation levels.
201
Whilst specification of moments of inertia and products of inertia may not be

attractive research topic knowledge of their specification is important when
looking at damaged ships. Hence some further effort could be useful.
Turan (1993) suggests that safety and commercial gain conflict. If practical
procedures could be developed to include damage stability within available
optimisation tools this conflict might be more readily resolved.
In the optimisation process only primary parameters were investigated to permit

LCF to remain invariant and hence make meaningful comparisons of intact and
damaged ships motion responses. The cited optimisation tools indicate that
improvements for secondary parameters are comparable with those for the
primary parameters investigated. Therefore in the future research the secondary
parameters should be investigated in the context of damage stability and motions
of damaged optimised ships.
The equilibrium angles of the damaged ships were small. Consequently their
effect on the longitudinal and transverse mass distribution is neglected.
However, for large equilibrium angles the relationship between the ship mass
distribution and equilibrium angles needs to be investigated.
202
BIBLIOGRAPHY
Abkowitz, M.A., Vassilopoulos, L.A. and Sellars, F.H. (1966). Recent developments in
seakeeping research and its application to design. Trans. SNAME, Vol. 74, pp. 194-259.
American Bureau of Shipping (1992). Rules for Building and Classing Steel Vessels.
Parts 3.3 and 3.12: Hull Construction and Equipment.
Andrews, D. (1981). Creative ship design. Trans. RINA, Vol. 123, pp. 447-471.
Andrews, D. (1986). An integrated approach to ship synthesis. Trans. RINA, Vol. 128,
pp. 73-102.
Andrews, D. (2003). A creative approach to ship architecture. International Journal of

Maritime Engineering, Vol. 145(A3), pp. 69-92.
Aoki, M. (1971). Introduction to Optimization Techniques, Fundamentals and

Applications of Nonlinear Programming. Section 4.7, pp. 152-155. The Macmillan
Company, New York.
Bertram, V. (2000). Practical Ship Hydrodynamics. Section 1.4.1, pp. 9-14.

Butterworth-Heinemann, Oxford.
Bertram, V. (2003). Optimistic, optimization in marine design. In 39th WEGEMT
Summer School, Berlin, Germany, 29-56.
Bird, H. and Browne, R.P. (1974). Damage stability model experiments. Trans. RINA,
Vol. 116, pp. 69-91.
203
Bird, H. and Morrall, A. (1986). Research towards realistic stability criteria. In

Proceedings of the International Conference on the Safeship Project: Ship Stability &
Safety, London, UK.
Bishop, R.E.D., Price, W.G. and Tam, P.K.Y. (1977). The dynamical characteristics of
some dry hulls. Journal of Sound and Vibration, Vol.54(1), pp. 29-38.
Bishop, R.E.D., Price, W.G. and Temarel, P. (1991). A theory on the loss of the MV
Derbyshire. Trans. RINA, Vol.133, pp. 389-453.
Borlase, G.A. (2003). Research opportunities identified during the casualty analysis of
the fishing vessel Arctic Rose. Marine Technology, Vol. 40(4), pp. 270-277.
Brook, A.K. (1988). The effect of motions in assessing intact ship stability.
Transactions of the North East Coast Institution of Engineers and Ship Builders
(NECIES), Vol. 104(3), pp. 81-90.
Bureau Veritas (1986). Rules and Regulations for the Classification of Steel Ships of
More Than 65m in Length. Part II, Section 4.6: Hull Structure, pp. 42-52.
Chan, H.S., Atlar, M. and Incecik, A. (2002). Large-amplitude motion responses of a

Ro-Ro ship to regular oblique waves in intact and damaged conditions. Journal of
Marine Science and Technology, Vol. 7(2), pp. 91-99.
Cramer, H. and Tellkamp, J. (2002). Towards the direct assessment of a ships intact
stability. In Proceedings of the 6th International Ship Stability Workshop, Webb
Institute, pp. 1-7.
Cramer, H. and Tellkamp, J. (2003). Towards safety as performance criteria in ship

design. In Proceedings of the Passenger Ship Safety, London, UK, pp. 17-25.
204
Dai, C., Hambric, S., Mulvihill, L., Tong, S.S. and Powell, D. (1994). A prototype
marine propulsion design tool using artificial intelligence and numerical optimization
techniques. Trans. SNAME, Vol. 102, pp. 57-69.
Day, A.H. and Doctors, L.J. (1997). Design of fast ships for minimal resistance and
motions. In Proceedings of the 6th International Marine Design Conference, IMDC97,
Newcastle, UK, pp. 569-583.
Department of Transport (1989). The Merchant Shipping Act 1984 mv Derbyshire.

Report of Court No.8075.
Doctors, L.J. and Day, A.H. (1995). Hydrodynamically optimal hull forms for river
ferries. In Proceedings of the International Symposium on High Speed Vessels for
Transport and Defence, RINA, London, UK, Paper No: 5, pp.1-21.
Eames, M.C. and Drummond, T.G. (1977). Concept explorationan approach to small
warship design. Trans. RINA, Vol. 119, pp. 29-54.
Fach, K. (2004). Corrosion and damages on high speed craft. In Proceedings of the 4th
International Conference on High-Performance Marine Vehicles, Rome, Italy, pp. 295303.
Faltinsen, O.M. (1978). A numerical nonlinear method of sloshing in tanks with twodimensional flow. Journal of Ship Research, Vol. 22(3), pp. 193-202.
Faltinsen, O.M. (1990). Sea Loads on Ships and Offshore Structures. Chapter 3, pp. 3768. Cambridge University Press, Cambridge.
Fisher, K.W. (1972). Economic optimisation procedures in preliminary ship design

(applied to the Australian ore trade). Trans. RINA, Vol. 114, pp. 293-309.
205
Francescutto, A. (2004). Intact ship stability: the way ahead. Marine Technology, Vol.
41(1), pp. 31-37.
Frank, W. (1967). Oscillations of cylinders in or below the free surface of deep fluids.
NSRDC, Report 2375, DTRC, Bethesda, MD.
Friis, A.M., Andersen, P. and Jensen, J.J. (2002). Ship Design. Part 1, Chapter 6, pp. 7077. Section of Maritime Engineering, Department of Mechanical Engineering,
Technical University of Denmark, Denmark.
Friis, A.M., Andersen, P. and Jensen, J.J. (2002). Ship Design. Chapter 11, pp. 171-176.
Section of Maritime Engineering, Department of Mechanical Engineering, Technical
University of Denmark, Denmark.
Furukawa, Y. and Hearn, G.E. (2000). Objective function development, using design
charts, to reflect ship seakeeping, resistance, and manoeuvrability influences. In
Proceedings of the Design for Excellence Engineering Design Conference, Brunel
University, UK, pp. 571-584.
Ghani, M.P.A. (2003). Design Aspects of Catamarans Operating at High Speed in

Shallow Water. PhD Thesis, The University of Southampton, U.K.
Gilfillan, A.W. (1969). The economic design of bulk cargo carriers. Trans. RINA, Vol.
111, pp. 113-140.
Greenberg, M. D. (1998). Advanced Engineering Mathematics, second edition. Section

8.3, pp. 396-397. Prentice-Hall International, London.
Greenberg, M. D. (1998). Advanced Engineering Mathematics, second edition. Section

10.6.3, pp. 517-518. Prentice-Hall International, London.
206
Grochowalski, S., Hsiung, C.C., Huang, Z.J. and Cong, L.Z. (1998). Theoretical
modeling of ship motions and capsizing in large and steep waves. Trans. SNAME, Vol.
106, pp. 241-267.
Hasegawa, K., Ishibashi, K. and Yasuda, Y. (2000). Modeling and computer animation
of damage stability. In Proceedings of the 7th International Conference on Stability of
Ships and Ocean Vehicles (STAB 2000), Launceston, Tasmania, Australia, pp. 242-249.
Haskind, M.D. (1954). Approximate methods of determination of hydrodynamic

characteristics of ship oscillations. Izvestiya Akad. Naut. SSSR. Otd. Tech. Naut., No.
11.
Hearn, G.E. (1978). Diffraction Suite MATTHEW November 1981 Release.
Hearn, G.E. and Donati, E. (1981). Sea keeping theories-applying some choice.
(NECIES), Vol. 97,pp. 53-72.
Hearn, G.E., Hills, W. and Colton, P. (1988). Incorporating a seakeeping capability in a

computer aided preliminary design system. In Proceedings of the 13th Ship Technology
and Research (Star) Symposium, Pittsburgh, Pennsylvania, pp. 179-194.
Hearn, G.E., Hills, W. and Sarioz, K. (1990). Making seakeeping analysis work for the
designer a new practical approach. In Proceedings of 19th Scientific and
Methodological Seminar on Ship Hydrodynamics , Varna, Bulgaria, Vol. 2, pp. 69-1-698.
Hearn, G.E. (1991). Sea keeping theories: spoilt for choice? Transactions of the North
East Coast Institution of Engineers and Ship Builders (NECIES), Vol. 107,pp. 45-66.
207
Hearn, G.E., Hills, W. and Sarioz, K. (1991). A hydrodynamic design methodology for
conceptual ship design. In Proceedings of 7th International Conference on Computer
Applications in the Automisation of Shipyard Operation and Ship Design ICCAS91,
Brazil, pp. 113-129.
Hearn, G.E., Hills, W. and Sarioz, K. (1992). Practical seakeeping for design: a ship
shape approach. Trans. RINA Spring Meetings, UK, Vol. 134, pp. 225-244.
Hearn, G.E., Wright, P.N.H. and Hills, B. (1994). Seakeeping for design: the demands
of multihulls in comparison to monohulls. In Proceedings of the 8th International
Conference on Computer Applications in Shipbuilding 94, Vol. 2, pp. 11.125-11.144.
Hearn, G.E., Wright, P.N.H. and Hills, B. (1995a). Seakeeping for design: development
and application of an inverse analysis design methodology to multihull forms. In
Proceedings of RINA International Conference on Seakeeping and Weather, London,
UK, Paper No: 13, pp. 1-15.
Hearn, G.E., Wright, P.N.H. and Hills, W. (1995b). Seakeeping for design: balancing
the vertical and horizontal motions of a catamaran. In Proceedings of the 3rd
International Conference on Fast Sea Transportation, FAST95, Lubeck-Travemunde,
Germany, Vol. 1, pp. 205-220.
Hearn, G.E., Wright, P.N.H. and Yaakob, O. (1995c). Seakeeping for design:
identification of hydrodynamically optimal hull forms for large high speed catamarans.
In Proceedings of RINA International Symposium on High Speed Vessels for Transport
and Defence, London, UK, Paper No: 4, pp. 1-15.
Hearn, G.E. and Wright, P.N.H. (1997). Seakeeping for design: optimisation of motion
responses and wave making resistance of catamarans via the application of a genetic
208
algorithm. In Proceedings FAST97, the Fourth International Conference on Fast Sea

Transportation, Sydney, Australia, Vol. 1, pp. 231-240.
Hearn, G.E. and Wright, P.N.H. (1998a). Optimal hull form design for seakeeping and
resistance: a genetic algorithm based inverse method. In Proceedings of Third Osaka
Colloquium on Advanced CFD Applications to Ship Flow and Hull Form Design,
OC98, Osaka, Japan, pp. 499-514.
Hearn, G.E. and Wright, P.N.H. (1998b). Michell wave resistance characteristics
resulting from the optimisation of a catamaran hull form via genetic algorithms. In
Proceedings of the Third Biennial Engineering Mathematics and Applications
Conference and Michell Centenary Symposium, EMAC98, Adelaide, South Australia,
pp. 253-256.
Hearn, G.E. and Wright, P.N.H. (1999). Design for optimal hydrodynamic operation of
large container ships. In Proceedings of the International Conference on Design and
Operation of Container Ships, London, UK, Paper No: 5, pp. 1-13.
Hearn, G.E., Furukawa, Y. and Kijima, K. (2000). Development of a design support

methodology to resolve seakeeping, resistance and manoeuvrability conflicts. Trans. of
the West-Japan Society of Naval Architects, Vol. 100, pp. 1-16.
Hoffman, D. and Zielinski, T. (1977). The use of conformal mapping techniques for
hull surface definition. In Proceedings of the SNAME Computer-Aided Hull Surface
Definition Symposium, Annapolis, USA, pp. 159-174.
Holden, C.M.E., Davies, R. and Keane, A.J. (2002). Optimization methodologies in

conceptual design. In Proceedings of the 9th AIAA/ISSMO Symposium on
Multidisciplinary Analysis and Optimization, Atlanta, USA, pp. 1-11.
209
Holtrop, J. and Mennen, G.G.J. (1982). An approximate power prediction method. In

International Shipbuilding Progress, Vol. 29, pp. 166-170.
Holtrop, J. (1984). A statistical re-analysis of resistance and propulsion data. In

International Shipbuilding Progress, Vol. 31, pp. 272-276.
Hooke, R. and Jeeves, T.A. (1961). Direct search solution of numerical and statistical
problems. Journal of the Association for Computing Machinery, Vol. 8, pp. 212-229.
Housner, G.W. and Hudson, D.E. (1966). Applied Mechanics Dynamics, second
edition. Volume II. Sections 7.3 and 7.4, pp. 188-191. D. Van Nostrand Company, New
Jersey.
Huang, Z. J. and Hsiung, C.C. (1997). Nonlinear shallow-water flow on deck coupled
with ship motion. In Proceedings of the 21st Symposium on Naval Hydrodynamics,
Trondheim, Norway, pp. 220-234.
Huang, J., Cong, L., Grochowalski, S., Hsiung, C.C. (1998). Capsize analysis for ships
with water shipping on and off the deck. In Proceedings of the 22nd Symposium on
Naval Hydrodynamics, Washington, USA, pp. 376-389.
Hudson, D.A. (1999). A Validation Study on Mathematical Models of Speed and

Frequency Dependence in Seakeeping of High Speed Craft. PhD Thesis, The University
of Southampton, U.K.
Ikeda, Y. and Ma, Y. (2000). An experimental study on large roll motion in

intermediate stage of flooding due to sudden ingress water. In Proceedings of the 7th
International Conference on Stability of Ships and Ocean Vehicles, Launceston,
Australia, pp. 270-285.
210
IMCO (1960). SOLAS 1960. International conference on safety of life at sea.
IMO (1971). Resolution A.265 (VIII): Regulation on subdivision and stability of

passenger ships.
IMO (1993). Resolution A.751 (18). Interim standards for ship manoeuvrability.
IMO (1997). SOLAS Consolidated Edition 1997.
IMO (2002a). SLF 45/3/3: Investigations and proposed formulations for the factor s:
the probability of survival after flooding.
IMO (2002b). SLF 45/3/5: Investigations and proposed formulations for the factors p,
r and v: the probability of damage to a particular compartment or compartments.
Insel, M. and Molland, A.F. (1992). An investigation into the resistance components of
high speed displacement catamarans. Trans. RINA, Vol. 134, pp. 1-20.
Janson, C. E. and Larsson, L. (1996). A method for the optimization of ship hulls from a
resistance point of view. In Proceedings of the 21st Symposium on Naval
Hydrodynamics, Trondheim, Norway, pp. 680-696.
Keane, A.J. (1988). A computer based method for hull form concept design:
applications to stability analyses. Trans. RINA, Vol. 130 (A), pp. 61-75.
Keane, A.J., Price, W.G. and Schachter, R.D. (1991). Optimization techniques in ship
concept design. Trans. RINA, Vol. 133, pp. 123-139.
211
Keane, A.J. and Robinson, G.M. (1999). Experiences with optimizers in design. In
Proceedings of the 14th International Symposium on Air Breathing Engines, ISABE 99,
Florence, Italy.
Kerczek, V.C. and Tuck, E.O. (1969). The representation of ship hulls by conformal
mapping functions. In Journal Ship Research, Vol. 13, pp. 284-298.
Kim, Y. (2001). Coupled analysis of ship motions and sloshing flows. In Proceedings of
the 16th International Workshop on Water Waves and Floating Bodies, Hiroshima,
Japan, pp. 1-4.
Kobylinski, L.K. and Kastner, S. (2003). Stability and Safety of Ships. Volume I:
Regulation and Operation. Section 1.1, pp. 3-10. Elsevier Ocean Engineering Book
Series, Oxford.
Kowalik, J. and Osborne, M.R. (1968). Methods for Unconstrained Optimization

Problems. Section 2.4, pp. 21-22. American Elsevier Pub. Co., New York.
Kristensen, H.O. (2002). Design considerations of passenger ships with respect to

damage stability requirements. In DCAMMcourse, Stability of Ships, Lyngby,
Denmark, pp. 1-22.
Kruger, C.J.C. (2004). Constrained cubic spline interpolation. Lecture notes.
Lackenby, H. (1950). On the systematic geometrical variation of ship forms. Trans.

INA, Vol. 92, pp. 289-316.
Landweber, L. and Macagno, M. (1959). Added mass of a three-parameter family of

two-dimensional forces oscillating in a free surface. Journal of Ship Research, pp. 3648.
212
Lewis, E.V. (1959). Increasing the sea speed of merchant ships. Trans. SNAME, Vol.
67, pp. 757-772.
Lewis, E.V. (1988). Principles of Naval Architecture. Volume I: Stability and Strength.
Sections 3.3 and 3.4, pp. 149-176. Sname, New Jersey, USA.
Lewis, E.V. (1988). Principles of Naval Architecture. Volume I: Stability and Strength.
Section 3.6, pp. 178-180. Sname, New Jersey, USA.
Lewis, E.V. (1989). Principles of Naval Architecture. Volume III: Motions in Waves
and Controllability. Section 8.3, pp. 41-83. Sname, New Jersey, USA.
Lewis, E.V. (1989). Principles of Naval Architecture. Volume III: Motions in Waves
and Controllability. Section 8.5, pp. 109-112. Sname, New Jersey, USA.
Liu, D., Hughes, O. and Mahowald, J. (1981). Applications of a computer-aided,

optimal preliminary ship structural design method. Trans. SNAME, Vol. 89, pp. 275294.
Lloyd, A.R.J.M. (1991). The seakeeping design package (SDP). In RINA Spring
Meetings, London, UK and (1992) Trans. RINA, Vol. 134, pp. 161-180.
Lloyd, A.R.J.M. (1998). Seakeeping: Ship Behaviour in Rough Weather. Section 3.3,
pp. 71-77. A.R.J.M. Lloyd, UK.
Lloyd, A.R.J.M. (1998). Seakeeping: Ship Behaviour in Rough Weather. Chapter 8, pp.
152-164. A.R.J.M. Lloyd, UK.
Lloyd, A.R.J.M. (1998). Seakeeping: Ship Behaviour in Rough Weather. Chapter 15,
pp. 303-314. A.R.J.M. Lloyd, UK.
213
Lloyds Register (2003). Rules and regulations for the classification of ships. Part 3,
Section 3.4, pp. 8-10.
Lunde, J.K. (1951). On the linearized theory of wave resistance for displacement ships
in steady and accelerated motion. Trans. SNAME, Vol. 59, pp. 25-85.
Lutzen, M. (2002). Discussions on damage statistics. Private communication.
Lyon, T.D. and Mistree, F. (1985). A computer-based method for the preliminary
design of ships. Journal of Ship Research, Vol. 29, pp. 251-269.
MAIB (1999). Marine Accident Investigation Branch MGN 115 (M+F): Accident
reporting and investigation.
MAIB (2000). Marine Accident Investigation Branch Annual Report 2000.
Mandel, P. and Leopold, R. (1966). Optimization methods applied to ship design.

Trans. SNAME, Vol. 74, pp. 477-521.
MCA (1999). MSN 1715 (M): Subdivision and damage stability of cargo ships of 80m
in length and over.
Michell, J.H. (1898). The wave resistance of a ship, Philosophical Magazine, Series 5,
Vol. 45, pp. 106-123.
214
Middleton, E.H. and Numata, E. (1970). Tests of a damaged stability model in waves.
Trans. SNAME, Vol. 7, pp. 1-14.
Molland, A.F., Wellicome, J.F. and Couser, P.R. (1996). Resistance experiments on a
systematic series of high speed displacement catamaran forms: variation of lengthdisplacement ratio and breadth-draught ratio. Trans. RINA, Vol. 138, pp. 55-71.
Molland, A.F. and Karayannis, T. (1997). Development of a concept exploration and

assessment model for advanced fast marine vehicles. In Proceedings of the Sixth
International Marine Design Conference, Newcastle, UK, Vol.1, pp. 249-265.
Molland, A.F. and Lee, A.R. (1997). An investigation into the effect of prismatic
coefficient on catamaran resistance. Trans. RINA, Vol. 139, pp. 157-165.
Molland, A.F., Wellicome, J.F., Temarel, P., Cic, J. and Taunton, D.J. (2001).
Experimental investigation of the seakeeping characteristics of fast displacement
catamarans in head and oblique seas. Trans. RINA, Vol. 143, pp. 79-98.
Murphy, R.D., Sabat, D.J. and Taylor, R.J. (1965). Least cost ship characteristics by
computer techniques. Marine Technology, Vol. 2, pp. 174-202.
Newman, J.N. (1977). Marine Hydrodynamics. Sections 6.17 and 6.18, pp. 295-307.
The MIT Press, Massachusetts.
Newman, J.N. (1978). The theory of ship motions. Advances in Applied Mechanics,
Vol. 18, pp. 221-283.
Nowacki, H., Brusis, F. and Swift, P.M. (1970). Tanker preliminary designan
optimization problem with constraints. Trans. SNAME, Vol. 78, pp. 357-390.
215
Odabasi, A.Y. and Hearn, G.E. (1978). Sea keeping theories: what is the choice? British
Ship Research Association, pp. 53-84.
Paik, J.K. and Faulkner, D. (2003). Reassessment of the M.V. Derbyshire sinking with
the focus on hull-girder collapse. Marine Technology, Vol. 40(4), pp. 258-269.
Palazzi, L. and Kat, J.D. (2004). Model experiments and simulations of a damaged ship
with air flow taken into account. Marine Technology, Vol. 41(1), pp. 38-44.
Pawlowski, M. (1999). The new damage stability criteria (in English and Polish).
Raport Techniczny Nr.35/99.
Pawlowski, M., Vassalos, D., Cabaj, D., Tuzcu, C. and Konovessis, D. (2004).
Floodable length curves based on probability of survival. In Proceedings of the 4th
International Conference on High-Performance Marine Vehicles, Rome, Italy, pp. 192199.
Peach, R.W. and Brook, A.K. (1987). The radii of gyration of merchant ships.
(NECIES), Berlin, Vol. 103(3), pp. 115-117.
Pelaez, J.G., Papanikolaou, A. and Gonzalez, V. (2000). Numerical and experimental

study on the seakeeping performance of a fast round-bilge monohull. In Proceedings of
the 4th International Osaka Colloquium on Seakeeping Performance of Ships, Osaka,
Japan, pp. 1-10.
Ravn, E.S. (2002). Decision and optimization methods. In DCAMMcourse, Stability of

Ships, Lyngby, Denmark, pp. 1-13.
216
Robson, B.L. (1988). Systematic series of high speed displacement hull forms for naval
combatants. Trans. RINA, Vol. 130, pp. 241-259.
Rognebakke, O.F. and Faltinsen, O.M. (2001). Effect of sloshing on ship motions. In
Proceedings of the 16th International Workshop on Water Waves and Floating Bodies,
Hiroshima, Japan, pp. 1-4.
Rusas, S. (2002). Stability of ships: probability of survival. In DCAMMcourse,

Stability of Ships, Lyngby, Denmark, pp. 1-13.
Salvesen, N., Tuck, E.O. and Faltinsen, O. (1970). Ship motions and sea loads. Trans.
SNAME, Vol. 78, pp. 250-287.
Sarioz, K., Hearn, G.E. and Hills, B. (1992). Practical seakeeping for design: an
optimised approach. In Proceedings of PRADS 92, Vol. 1, pp. 1.233-1.246.
Sarioz, K. (1993). A Hydrodynamic Hull Form Design Procedure in Conceptual and

Preliminary Ship Design. PhD Thesis, The University of Newcastle upon Tyne, U.K.
Saydan, D. (1999). The analysis of irregular waves at time and frequency domain (in
Turkish). Final Year Project, Faculty of Naval Architecture and Ocean Engineering,
Technical University of Istanbul.
Saydan, D. (2003). Manual for Motion Code, School of Engineering Sciences,

University of Southampton, UK.
Saydan, D. and Hearn G.E. (2004). Damage stability as a safety criterion for
optimisation tools. In Proceedings of the 4th International Conference on HighPerformance Marine Vehicles, Rome, Italy, pp. 177-191.
217
Saydan, D. (2004). Manual for Equilibrium Search Code, School of Engineering

Sciences, University of Southampton, UK.
Schey, J.A. (1977). Introduction to Manufacturing Processes. Section 1.4, pp. 8-11.
McGraw-Hill, New York.
Schneekluth, H. and Bertram, V. (1998). Ship Design for Efficiency and Economy.
Section 1.2, pp. 5-13. Butterworth Heinemann, Oxford.
Schneekluth, H. and Bertram, V. (1998). Ship Design for Efficiency and Economy.
Chapter 3, pp. 85-111. Butterworth Heinemann, Oxford.
Schroter, C. and Juhl J.S. (2002). Survivability after damage as observed from model
tests. In DCAMM course, Stability of Ships, Lyngby, Denmark, pp. 1-6.
Sen, P. (1992). Marine design: the multiple criteria approach. Trans. RINA, Vol. 134,
pp. 261-276.
Sen, P., Birmingham, R., Cripps R.M. and Cain, C. (1997). A methodology for the
integration of formal safety analysis into the design of lifeboats. In Proceedings of the
Sixth International Marine Design Conference, Newcastle, UK, Vol. 1, pp. 313-327.
ShipShape (1992). Manual for Shipshape. Wolfson Unit for Marine Technology and
Industrial Aerodynamics (WUMTIA), University of Southampton, UK.
Spouge, J.R. (2003). The safety of general cargo ships. Trans. RINA International
Journal of Maritime Engineering, Vol. 145(A3), pp. 29-40.
218
Subramanian, V.A. and Kastner, S. (2000). Ship behaviour due to dynamic impact of
water on deck. In Proceedings of the 7th International Conference on Stability of Ships
and Ocean Vehicles, pp. 206-225.
Tagg, R. and Tuzcu, C. (2003). A performance-based assessment of the survival of

damaged ships: final outcome of the EU research project HARDER. Marine
Technology, Vol. 40(4), pp. 288-295.
Timman, R. and Newman, J.N. (1962). The coupled damping coefficients of a

symmetric ship. Journal of Ship Research, Vol. 5(4), pp. 1-7.
Tuck, E.O. (1987). Wave resistance of thin ships and catamarans. Report T8701.
Tuck, E.O. and Lazauskas, L. (1998). Optimum hull spacing of a family of multihulls.
Ship Tech. Res., Vol. 45, pp. 180-195.
Turan, O. (1993). Dynamic Stability Assessment of Damaged Passenger Ships Using a

Time Simulation Approach. PhD Thesis, The University of Strathclyde, Glasgow, U.K.
Turan, O. and Vassalos, D. (1994). Dynamic stability assessment of damaged passenger

ships. Trans. RINA, Vol. 136, pp. 79-104.
Umeda, N., Kamo, T. and Ikeda, Y. (2004). Some remarks on theoretical modelling of
damaged stability. Marine Technology, Vol. 41(1), pp. 45-49.
Vassalos, D., Pawlowski, M. and Turan, O. (1996). A theoretical investigation on the
capsizal resistance of passenger/Ro-Ro vessels and proposal of survival criteria. Final
Report, Task 5, The Joint North West European R&D Project, University of
Strathclyde.
219
Vassalos, D., Letizia, L., Shaw, M. and MacPherson, C. (1998). An investigation on the
flooding of damaged Ro-Ro ships. Trans. RINA, Vol. 140, pp. 273-289.
Vassalos, D., Guarin, L. and Jasionowski, A. (2001). Original tests on MV Derbyshire

and the seakeeping investigation programme on the safety of bulk carriers. In
Proceedings of Glasgow Marine Fair and International Workshop on Safety of Bulk
Carriers, Glasgow, Scotland, UK, pp. 1-9.
Vassalos, D. and Jasionowski, A. (2002). Damaged ship hydrodynamics. In

Proceedings of the 6th International Ship Stability Workshop, Webb Institute, pp. 1-13.
Veer, R.V. and Kat, J.O. (2000). Experimental and numerical investigation on
progressive flooding and sloshing in complex compartment geometries. In Proceedings
of the 7th International Conference on Stability of Ships and Ocean Vehicles, pp. 305321.
Vossers, G., Swaan, W.A. and Rijken, H. (1960). Experiments with series 60 models in
waves. Trans. SNAME, Vol. 68, pp. 364-450.
Walsh, G.R. (1975). Methods of Optimization. Section 3.3, pp. 76-79. John Wiley &
Sons, London.
Watson, D.G.M. (1962). Estimating preliminary dimensions in ship design. Trans.

IESS, Vol. 105, pp. 1-52.
Watson, D.G.M. and Gilfillan, A.W. (1977). Some ship design methods. Trans. RINA,
Vol. 119, pp. 279-324.
Westlake, P.C. and Wilson, P.A. (2000). A new conformal mapping technique for ship
sections. In International Shipbuilding Progress, Vol. 47(449), pp. 5-22.
220
Westlake, P.C., Wilson, P.A. and Bailey, P.A. (2000). Time domain simulations of ship
motions. Trans. RINA, Vol. 142, pp. 268-288.
Wilson, P.A. (1986). A seakeeping analysis of a family of merchant ships. In

Proceedings of the International Conference on Computer Aided Design, Manufacture
and Operation in the Marine and Offshore Industries, Washington D.C., USA, pp. 237253.
Wolfson (2001a). Wolfson Unit Hydrostatics and Stability Program 2001 Release.
Wolfson (2001b). Wolfson Unit Floodable Length Program 2001 Release.
Woodburn, P., Gallagher, P. and Letezia, L. (2002). Fundamentals of damaged ship

survivability . Trans. RINA, Vol. 144, pp. 1-17.
Wright, P.N.H. (2004). The Preliminary Design of Catamarans for Seakeeping and
Resistance. PhD Thesis, The University of Newcastle upon Tyne, U.K.
221
APPENDIX A HOOKE JEEVES ALGORITHM
222
The Hooke-Jeeves method based optimisation process has two main steps designated
exploratory and pattern. The exploratory move tries to understand the local
behaviour of the objective function. The pattern move seeks to confirm, or otherwise,
the existence of some advantage in adopting the variation of the point suggested by the
exploratory move. The pattern move is always followed by a sequence of exploratory
moves, so that an improved direction of search can be found and another patter move
can be followed in that direction.
Thus if the problem of minimizing g ( x ) is considered, the exploratory move starts with
an initial base point, b1 , from,
b := ( b1 , b2 ....................bn )
(A.1)
HJ := ( HJ1 , HJ 2 ....................HJ n )
(A.2)
together with step lengths,
for the respective variables,

x := ( x1 , x2 ....................xn ) .
(A.3)
After the determination of g ( b1 ) , the evaluation of g ( b1 + HJ1 ) is carried out. If the

g ( b1 + HJ1 ) < g ( b1 ) , the new base point will be b1 + HJ1 instead of b1 ; otherwise
g ( b1 HJ1 ) will be evaluated. If the g ( b1 HJ1 ) < g ( b1 ) , the new base point will be
b1 HJ1 instead of b1 . If none of these moves is a success, the original point will be
retained. The cited procedure has to be carried out for all variables in turn, finally
arriving at a new base point after 2n + 1 function evaluations at most. If b2 = b1 , the step
lengths for each variable has to be halved and the exploratory move has to be repeated.
223
The analysis finishes when the step length is reduced below predetermined limits. If
b2 b1 , a pattern move will be carried out from b2 .
The pattern move starts from the new base point b2 in the direction b2 b1 since
exploratory move indicates a decrease in the value of g ( x) in this direction. Therefore,
a move from b2 to 2b2 b1 will be analysed and new exploratory moves will be
carried out about 2b2 b1 . If the pattern and exploratory moves about 2b2 b1
represents a lowest function value which is lower than g (b2 ) , then a new base point
b3 has been found. In this case, next pattern move will be from b3 to 2b3 b2 .
Otherwise, the pattern move from b2 is terminated and a new exploratory move about
b2 will be initiated.
224
APPENDIX B - BOUNDARY VALUE PROBLEM
225
As it is mentioned earlier in the dissertation the unknown diffraction & radiation

velocity potentials are determined using Greens second identity to convert the partial
differential equation formula presented into an equivalent Fredholm formulation
boundary integral. The fluid and structural boundaries in seakeeping fluid-structure
interaction formulations are the free surface, the wetted surface of the vessel, the seabed and the artificial cylindrical radiation boundary linking sea-bed and free surface.
The boundary condition on this last boundary (which ultimately tends to infinity) is
often referred to as the Sommerfield radiation condition. The radiation condition is
necessary because the governing partial differential equation is Laplaces equation,
which includes the conditions of an incompressible fluid subject to irrotational flow,
and being an elliptic partial differential equation a closed boundary is required.
The Sommerfield radiation condition implies that the energy associated with the
disturbance of the fluid caused by the motion of the vessel must vanish at infinity.
The sea-bed condition assumes that the sea-bed is a rigid horizontal, flat and
impermeable boundary. Since there is no flow across this boundary
W
= 0 and D = 0 on the seabed.

n
n
(B.1)
Here W describes the incident wave potential and D is the diffraction potential,
resulting from the interaction of the incident wave and the structure(s) being
investigated.
On the wetted surface of the structure the fluid velocity normal to the body surface must
be identical to normal velocity of the structure. One can either view this as requiring
continuity of velocity across the wetted surface or impermeability of the wetted surface
and thus
226

(W + D ) = 0 on SW
n
(B.2)
and
j
n
= v jn on SW
(B.3)
Here j is the time-independent radiation potential arising from the structure moving in
the jth mode of motion and v jn denotes the normal velocity component of the structure
moving in the jth mode of motion.
On the free surface continuity of velocity and pressure can be shown to be satisfied
using the linearised composite free surface condition
2
= 0.
Y g
(B.4)
The time independent velocity potential representing either the radiation potential j
or the diffraction potential D with the time dependence satisfying
( Z , X , Y , t ) = ( Z , X , Y ) e it .
(B.5)
The resultant fluid flow is described by
= W + D + j ,
i =1
since the governing equations are linear and superposition may be applied.
227
(B.6)
The details of the theory can be found in many text, see for example Odabasi and Hearn
(1978), Newman (1978) and Hudson (1999).
228
APPENDIX C - NUMERICAL APPLICATION OF THE NEW

METHOD TO DETERMINE THE MASS MOMENT OF INERTIA
FOR PARENT DERBYSHIRE AND FOR A TANKER HULL FORM
229
Here the pure and product moment of inertia equations already provided in Chapter 5
are applied to the parent Derbyshire and a tanker hull form.
Firstly, calculations are carried out for the parent Derbyshire. Hence, the magnitudes of
four point masses and their longitudinal positions already provided in Table 6.2 (with
more decimal points) are used to determine the vertical and horizontal positions of the
four point masses.
From Equation (5.38) the coordinate X P can be determined assuming kYY = 0.225L .
2
M A ( Z A2 + X P2 ) + M F ( Z F2 + X P2 ) = ( M A + M F ) kYY
(5.38)
94686.56 64.056672 + X P2 + 104793.6 57.878592 + X P2 = (199480.16) ( 0.225 281.94)
(C.1)
Therefore, X P equals to 17.79547617 and hence from Equation (5.30) it is clear that
XS = XP .
With X S known, Equation (5.41) can be used to determine YAS = YFS , subject to the
assumption that kZZ = 0.41B together with the deduction made earlier from Equation
(5.34).
2
M A ( X S2 + YAS2 ) + M F ( X S2 + YFS2 ) = M k ZZ
.
(5.41)
94686.56 17.795476172 + YAS2 +104793.6 17.795476172 + YFS2 = (199480.16) ( 0.41 44.196)
(C.2)
Therefore, YAS and YFS equal to 3.41579.

230
Finally, to check consistency of assumed radii of gyration YAS = YFS is re-determined

using Equation (5.45)
2
M A ( Z A2 + YAS2 ) + M F ( Z F2 + YFS2 ) = M k XX
.
(5.45)
94686.56 64.056672 +YAS2 +104793.6 57.878592 +YFS2 = (199480.16) ( 0.216305098 281.94)
(C.3)
This results in YAS = YFS equal to 3.41579.
Second example is a tanker hull form with 216m of length and 32.2m of beam. The
magnitudes of four point masses and their longitudinal positions are provided in Table
C.1. For more details related to the tanker hull form see Bishop et al. (1977).
Table C.1: The four masses and their longitudinal positions for the tanker hull form.
MA
35656.6 tonnes
MF
39255.01 tonnes
ZA from LCG 46.795 metres

ZF from LCG 42.506 metres
MAP = MAS
17828.3 tonnes
MFP = MFS
19627.505 tonnes
Again from Equation (5.38) the coordinate X P can be determined assuming

kYY = 0.215L .
35656.6 46.7952 + X P2 + 39255.01 42.5062 + X P2 = ( 74911.61) ( 0.215 216)
(C.4)
231
Therefore, X P equals to 12.946 and hence from Equation (5.30) it is clear that
XS = XP .
With X S known, Equation (5.41) can be used to determine YAS = YFS , subject to the
assumption that kZZ = 0.41B together with the deduction made earlier from Equation
(5.34).
35656.6 12.9462 + YAS2 + 39255.01 12.9462 + YFS2 = ( 74911.61) ( 0.4132.2)
(C.5)
Therefore, YAS and YFS equal to 2.587.

Finally, to check consistency of assumed radii of gyration YAS = YFS is re-determined
using Equation (5.45)
35656.6 46.7952 +YAS2 + 39255.01 42.5062 +YFS2 = ( 74911.61) ( 0.2068 216)
This results in YAS = YFS equal to 2.587.
232
(C.6)
APPENDIX D - GENERAL ARRANGEMENT DRAWING OF THE

DERBYSHIRE
233
234
APPENDIX E GENERAL APPRAISAL OF DAMAGE SCENARIOS

FOR PARENT DERBYSHIRE
235
In Chapter 6 (Table 6.11) four distinct scenarios were defined based on one hold and
two hold damage. The statistics of Chapter 3 suggest that more than two holds being
simultaneously damaged is unlikely. Whilst this may be the case it was interesting to
consider more damaged holds than is thought feasible to appreciate the attitude of the
hull as a result of water ingress into the parent Derbyshire hull form. Table E.1 indicates
the amount of parallel sinkage and trim angle achieved as the number of damaged
compartments increases. Heel angles were not significant and hence there is no need to
record them.
Table E.1: Different damage scenarios.
Holds Damaged
2
1+2
1+2+Deep Tank
1+2+3+Deep Tank
1+2+3+4+Deep Tank
1+2+3+4+5+Deep Tank
Parallel sinkage (m) Trim (degrees)

1.243
1.192
1.992
1.875
2.6471
2.7321
3.5
3.2
4.3534
3.4795
5.2065
3.556
Draught at AP (m)
16.262
15.319
13.849
13.541
13.7
14.363
Draught at FP (m)
22.128
24.549
27.3
29.3
30.843
31.884
Depending on the number of damaged compartment the magnitude of parallel sinkage

and trim angle vary. This change has also effect on the draught at AP and FP.
236
APPENDIX F - ITERATIVE METHOD FOR DETERMINATION OF

EQUILIBRIUM TRIM AND HEEL ANGLES
237
The determination of the equilibrium trim and heel angles is very important for the
accurate analysis of the static and dynamic properties of the damaged ship. The analysis
method employed is applied to both the parent Derbyshire hull form and the optimised
hull form.
The algorithm employed is summarised as follows:
Determine parallel sinkage in accordance with structural mass losses and mass
gains through water ingress.
Apply 1 of trim (see Figure F.1) and determine whether ship orientation is one
of hydrostatic equilibrium.
Repeat second step until either the equilibrium achieved or trim angle just too
large. In latter case reduce trim angle increment to 0.5 and apply from previous
starting position. Reduce incremental trim angle as necessary until equilibrium
established.
Repeat procedures of step 2 and 3 for heel angle determination for given parallel
sinkage and trim angle.
Note trim and heel angles for equilibrium.
LCF
1
YSINK
Figure F.1: Analysis of the equilibrium angles.

238
WL2
WL1
WL
APPENDIX G SENSITIVITY ANALYSIS OF THE PURE AND

PRODUCT MOMENT OF INERTIAS TO THE EQUILIBRIUM
ANGLES
239
Here, the effect of parallel sinkage, trim and heel angles on the pure and product
moment of inertias is examined for the parent and optimised hull forms (see Table G.1
for the parent hull form and Table G.2. for the optimised hull form).
Table G.1: The pure and product moment of inertias belonging to the parent hull for
different damage scenarios.
PARENT
I44
I45
I46
I55
I56
I66
INTACT
6.55E+07
0.00E+00
4.14E+07
7.42E+08
0.00E+00
8.03E+08
DAMAGED
I44
I45
I46
I55
I56
I66
CASE A
7.05E+07
2.23E+07
5.47E+05
8.54E+08
-3.95E+06
9.18E+08
CASE B
7.04E+07
2.23E+07
3.62E+06
8.53E+08
-3.97E+06
9.18E+08
CASE C
7.52E+07
5.23E+07
3.21E+06
1.03E+09
-6.03E+06
1.10E+09
CASE D
7.57E+07
4.18E+07
-7.33E+05
9.25E+08
-6.81E+06
9.94E+08
Table G.2: The pure and product moment of inertias belonging to the optimised hull for
different damage scenarios.
OPTIMISED
I44
I45
I46
I55
I56
I66
INTACT
6.49E+07
0.00E+00
2.36E+07
8.80E+08
0.00E+00
9.44E+08
DAMAGED
I44
I45
I46
I55
I56
I66
CASE A
7.01E+07
2.81E+07
-1.09E+06
1.06E+09
-3.61E+06
1.12E+09
CASE B
7.00E+07
2.80E+07
2.36E+06
1.06E+09
-3.53E+06
1.12E+09
CASE C
7.50E+07
6.49E+07
4.32E+06
1.33E+09
-4.47E+06
1.40E+09
CASE D
7.54E+07
5.36E+07
-1.26E+06
1.18E+09
-5.21E+06
1.25E+09
The four product of inertias namely I 45 and I 56 are zero valued for the intact case.
However, this is not the case when the ship is damaged. Usually the pure and product
moment of inertias remain the same for damage case A and B, I 46 differs for these two
damage scenarios.
240
APPENDIX H HYDRODYNAMIC COEFFICIENT PLOTS FOR

THE
INTACT
DERBYSHIRE AND
DERBYSHIRE (SCENARIO A)
241
FOR
THE
DAMAGED
3.5E+03
A13
3.0E+03
A31
2.5E+03
2.0E+03
1.5E+03
1.0E+03
B13
5.0E+02
B31
0.0E+00
0
0.1
0.2
-5.0E+02
0.3
0.4
0.5
0.6
0.7
Figure H.1: Surge-heave and heave-surge added mass and fluid damping coefficients for
the intact Derbyshire.
3.5E+06
3.0E+06
A15
A51
2.5E+06
2.0E+06
1.5E+06
B15
1.0E+06
B51
5.0E+05
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure H.2: Surge-pitch and pitch-surge added mass and fluid damping coefficients for
the intact Derbyshire.
242
2.0E+06
1.5E+06
A53
A35
1.0E+06
B35
5.0E+05
B53
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure H.3: Heave-pitch and pitch-heave added mass and fluid damping coefficients of
for the intact Derbyshire.
5.0E+04
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
B24
-5.0E+04
B42
-1.0E+05
A24
-1.5E+05
A42
-2.0E+05
Figure H.4: Sway-roll and roll-sway added mass and fluid damping coefficients for the
intact Derbyshire.
243
8.E+06
7.E+06
A46
6.E+06
A64
5.E+06
4.E+06
3.E+06
B46
2.E+06
B64
1.E+06
0.E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure H.5: Roll-yaw and yaw-roll added mass and fluid damping coefficients for the
intact Derbyshire.
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
B62
-5.0E+04
B26
-1.0E+05
-1.5E+05
-2.0E+05
A26
-2.5E+05
A62
-3.0E+05
Figure H.6: Sway-yaw and yaw-sway added mass and fluid damping coefficients for the
intact Derbyshire.
244
7.E+03
A13
6.E+03
A31
5.E+03
4.E+03
3.E+03
2.E+03
B13
1.E+03
B31
0.E+00
0
0.1
0.2
-1.E+03
0.3
0.4
0.5
0.6
0.7
Figure H.7: Surge-heave and heave-surge added mass and fluid damping coefficient for
the damaged Derbyshire (scenario A).
4.0E+06
3.5E+06
A15
3.0E+06
A51
2.5E+06
2.0E+06
1.5E+06
B15
1.0E+06
B51
5.0E+05
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure H.8: Surge-pitch and pitch-surge added mass and fluid damping coefficient for
245
1.E+06
A35
A53
B35
5.E+05
B53
0.E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
-5.E+05
Figure H.9: Heave-pitch and pitch-heave added mass and fluid damping coefficient for
1.5E+05
A24
1.0E+05
A42
5.0E+04
B24
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
B42
-5.0E+04
Figure H.10: Sway-roll and roll-sway added mass and fluid damping coefficient for the
246
2.0E+07
A64
1.5E+07
A46
1.0E+07
B64
5.0E+06
B46
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
-5.0E+06
Figure H.11: Roll-yaw and yaw-roll added mass and fluid damping coefficient for the
2.5E+06
A62
2.0E+06
A26
1.5E+06
1.0E+06
B62
5.0E+05
B26
0.0E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure H.12: Sway-yaw and yaw-sway added mass and fluid damping coefficient for
247
APPENDIX I THEORY BEHIND THE RELATIVE VERTICAL

MOTION
248
The resultant vertical displacement at (z,x,y) relative to the origin (z0,x0,y0) at LCF is
dependent upon the pitch and roll levers l p and lr defined as
l p =
( z z 0 ) + ( y y0 )
lr =
( x x0 ) + ( y y0 )
(I.1)
and
(I.2)
Depending on the actual location of (z,x,y), relative to (z0,x0,y0), the resultant vertical
motion requires adding/subtracting the vertical resolved components of the pitch and
roll displacements as defined in Figures I.1(a) & I.1(b) to the vertical heave motion.
Pb (zb, xb, yb)

b
lpb
Pa (za, xa, ya)

5
a
y
lpa
b
a
lpbcos b
lpacos a
Figure I.1(a): Pitch and associated vertical displacement.
249
O(z0, x0, y0)
b
4
Rb (zb, xb, yb)

a
lrb
Ra (za, xa, ya)

lra
b
a
O(z0, x0, y0)
x
lrbcos b
lracos a
Figure I.1(b): Roll and associated vertical displacement.
From Figures I.1(a) & I.1(b) the following general observations may be made:
Tangential displacement is lever multiplied by roll or pitch amplitude.
Vertical displacement is tangential displacement multiplied by cosine of angle

subtended by point of interest and horizontal.
Vertical displacement is equivalent to projected lever in z-x plane multiplied by

roll or pitch amplitude.
Projected levers in z-x plane is simply
lp =
( z z0 )
lr =
( x x0 )
(I.3)
and
250
(I.4)
Vertical displacement is explicitly independent of vertical position of point of

interest, but is implicitly dependent on vertical coordinate through variation of
(z,x) on wetted surface. As point of interest moves from undisturbed free surface
to points on selected deck x increases in magnitude and its amplitude becomes
the half beam.
That is, the vertical displacement, SV ( z , x ) , may be written as

SV ( z , x ) = 3 l p sgn( z )5 + lr sgn( x) 4
(I.5)
+1 if z positive
sgn( z ) =
-1 if z negative
(I.6)
+1 if x positive
sgn( x) =
.
-1 if x negative
(I.7)
where
and
251
APPENDIX
RELATIVE
VERTICAL
MOTION
DIFFERENT SELECTED POINTS ALONG THE SHIP
252
FOR
2.5
0,Parent Intact
0,Optimised Intact
1.5


0.5
0
/L
Figure J.1: Relative vertical motion responses for the intact and damaged hull forms for
point A (in following seas).
2.5
45,Parent Intact
45,Optimised Intact
1.5


0.5
0
/L
point A (in quartering seas).
253
2.5
90,Parent Intact
90,Optimised Intact
1.5


0.5
0
/L
point A (in beam seas).
3.5
115,Parent Intact
2.5
1.5
0.5
0
/L
point A (in beam seas).
254
2.5
135,Parent Intact
1.5
0.5
0
/L
point A (in bow seas).
2.5
1.5
157.5,Parent Intact
157.5,Optimised Intact
157.5,Parent Damaged Case A
157.5,Optimised Damaged Case A
157.5,Parent Damaged Case B
157.5,Optimised Damaged Case B
157.5,Parent Damaged Case C
157.5,Optimised Damaged Case C
157.5,Parent Damaged Case D
157.5,Optimised Damaged Case D
0.5
0
0
/L
255
2.5
180,Parent Intact
1.5


0.5

0
0
/L
2.5
202.5,Parent Intact
1.5


0.5

0
0
/L
256

2.5
225,Parent Intact
1.5
0.5
0
0
/L
point A (in bow seas).
2.5
270,Parent Intact
2

1.5


0.5
0
/L
Figure J.10: Relative vertical motion responses for the intact and damaged hull forms
for point A (in beam seas).
257
0,Parent Intact
0,Optimised Intact
1.5

1

0.5
0
/L
for point B (in following seas).
5
45,Parent Intact
45,Optimised Intact

4.5
3.5
2.5
1.5
0.5
0
/L
for point B (in quartering seas).
258
4.5
90,Parent Intact
90,Optimised Intact
3.5
2.5
1.5
0.5
0
/L
for point B (in beam seas).
2.5
115,Parent Intact
1.5
0.5
0
/L
259

1.5
135,Parent Intact
0.5
0
/L
for point B (in bow seas).
1.5
157.5,Parent Intact
0.5
0
/L
for point B (in head seas).
260
1.5
180,Parent Intact
0.5
0
/L
2.5
202.5,Parent Intact
1.5


0.5
0
/L
261
225,Parent Intact
3.5
2.5
1.5
0.5
0
/L
for point B (in bow seas).
3.5
270,Parent Intact
2.5
1.5
0.5
0
/L
262
1.5
0,Parent Intact
0,Optimised Intact
0.5
0
/L
for point C (in following seas).
45,Parent Intact
45,Optimised Intact
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point C (in quartering seas).
263
4.5
90,Parent Intact
90,Optimised Intact
3.5
2.5
1.5
0.5
0
/L
for point C (in beam seas).
3.5
115,Parent Intact
2.5
1.5
0.5
0
/L
264
135,Parent Intact
1.5
0.5
0
/L
for point C (in bow seas).
157.5,Parent Intact
1.5
0.5
0
/L
for point C (in head seas).
265
180,Parent Intact
1.5
0.5
0
/L
3.5
202.5,Parent Intact

2.5

2

1.5
0.5
0
/L
266
225,Parent Intact
4.5
3.5
2.5
1.5
0.5
0
/L
for point C (in bow seas).
270,Parent Intact
3.5
2.5
1.5
0.5
0
/L
267
2.5
0,Parent Intact
0,Optimised Intact
1.5
0.5
0
/L
for point D (in following seas).
6.5
45,Parent Intact
45,Optimised Intact

6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point D (in quartering seas).
268
4.5
90,Parent Intact
90,Optimised Intact

4
3.5
2.5
1.5
0.5
0
/L
for point D (in beam seas).
3.5
115,Parent Intact
2.5
1.5
0.5
0
/L
269
2.5
135,Parent Intact
1.5
0.5
0
/L
for point D (in bow seas).
2.5
157.5,Parent Intact
1.5
0.5
0
/L
270
2.5
180,Parent Intact
1.5
0.5
0
/L
202.5,Parent Intact
3.5


2.5


1.5
0.5
0
/L
271
5.5
225,Parent Intact
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point D (in bow seas).
270,Parent Intact

3.5
2.5
1.5
0.5
0
/L
272
2.5
0,Parent Intact
0,Optimised Intact
1.5
0.5
0
/L
for point E (in following seas).
4.5
45,Parent Intact
45,Optimised Intact
3.5
2.5
1.5
0.5
0
/L
for point E (in quartering seas).
273
90,Parent Intact
90,Optimised Intact

5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
/L
for point E (in beam seas).
5.5
115,Parent Intact
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
274
3.5
135,Parent Intact
2.5
1.5
0.5
0
/L
for point E (in bow seas).
2.5
157.5,Parent Intact
1.5
0.5
0
/L
275
2.5
180,Parent Intact
1.5
0.5
0
/L
3.5
202.5,Parent Intact

2.5

2

1.5
0.5
0
/L
276
225,Parent Intact

4.5
3.5
2.5
1.5
0.5
0
/L
for point E (in bow seas).
6.5
270,Parent Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
277
0,Parent Intact
0,Optimised Intact

1.5
0.5
0
/L
for point F (in following seas).
45,Parent Intact
45,Optimised Intact
4.5
3.5
2.5
1.5
0.5
0
/L
for point F (in quartering seas).
278
6.5
90,Parent Intact
90,Optimised Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point F (in beam seas).
115,Parent Intact
4.5
3.5
2.5
1.5
0.5
0
/L
279
3.5
135,Parent Intact
2.5
1.5
0.5
0
/L
for point F (in bow seas).
2.5
157.5,Parent Intact
1.5
0.5
0
/L
for point F (in head seas).
280
2.5
180,Parent Intact
1.5
0.5
0
/L
202.5,Parent Intact
3.5


2.5


1.5
0.5
0
/L
281
225,Parent Intact
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point F (in bow seas).
6.5
270,Parent Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
282
2.5
0,Parent Intact
0,Optimised Intact
1.5
0.5
0
/L
for point G (in following seas).
45,Parent Intact
45,Optimised Intact
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point G (in quartering seas).
283
6.5
90,Parent Intact
90,Optimised Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point G (in beam seas).
115,Parent Intact
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
/L
284
2.5
135,Parent Intact
1.5
0.5
0
0
/L
for point G (in bow seas).
2.5
157.5,Parent Intact
1.5
0.5
0
/L
for point G (in head seas).
285
2.5
180,Parent Intact
1.5
0.5
0
/L
202.5,Parent Intact
4.5


3.5


2.5

1.5
0.5
0
/L
286
6.5
225,Parent Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
for point G (in bow seas).
6.5
270,Parent Intact
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
/L
287
APPENDIX K EXTENSION OF EQUATIONS OF MOTION FOR A

SHIP TO INCLUDE INFLUENCES OF INGRESSED WATER
WITHIN THE DAMAGED HOLD(S)
288
Within the dissertation the damaged ship has been modelled by considering the external
fluid structure interaction. The ingressed water has influenced the motion dynamics
through the hydrodynamics reactive and excitation loads reflecting the heeled-trimmed
orientation of damaged ship and the mass-inertia matrix changes noting changes in
equivalent point mass model and the inclusion of the products of inertia.
Another aspect of the ingressed water is that as the ship moves so will the internal freesurfaces within any damaged hull form. These free surfaces may be modelled as
structural massless flat plates with heave, roll and pitch degrees of freedom. Here the
hydrodynamic coupling of such a free surface and the ship are provided to indicate how
the nine degrees of freedom system may be modelled. The extension to damage scenario
C with two holds flooded is readily undertaken by extending the ideas encompassed in
the following equations. The distinct hydrostatic restoration coefficients for the
massless plate and the ship together with all the hydrodynamic reactive and active force
and moments may be readily calculated using the Matthew Diffraction Suite.
Equations (K.1) to (K.6) provide the surge, sway, heave, roll, pitch and yaw equations
of motion for the ship. Motion amplitude terms with the superscript FS denote the
motions (to be determined) of the internal free surface. On the hydrodynamic reactive
coefficients superscripts of the form /FS indicate that the loads included are the
hydrodynamic loads experienced by the ship as a consequence of the motions of the
free-surface. Equations (K.7) to (K.9) provide the heave, roll and pitch motion equations
for the internal free surface. In this case the superscript FS/ indicates loads induced on
the free-surface as a result of the motions of the ship. Because the free-surface is
modelled as a massless plate there are no structural cross-coupling terms in the freesurface motion equations.
289
2 ( M + A11 ) i B11 1 + 2 A12 i B12 2 + 2 A13 i B13 3 + 2 A14 i B14 4
+ 2 A15 i B15 + MYG 2 5 + 2 A16 i B16 MX G 2 6 + 2 A13/ FS i B13/ FS 3FS

+ 2 A14/ FS i B14/ FS 4FS + 2 A15/ FS i B15/ FS 5FS = F1
(K.1)
2 A21 i B21 1 + 2 ( M + A22 ) i B22 2 + 2 A23 i B23 3
2
2
2
2
+ A24 i B24 MYG 4 + A25 i B25 5 + A26 i B26 + MZ G 2 6
/ FS
/ FS
/ FS
/ FS
/ FS
/ FS
3FS + 2 A24
4FS + 2 A25
5FS = F2
+ 2 A23
i B23
i B24
i B25
(K.2)
2 A31 i B31 1 + 2 A32 i B32 2 + C33 2 ( M + A33 ) i B33 3
+ C34 2 A34 i B34 + MX G 2 4 + C35 2 A35 i B35 MZ G 2 5

/ FS
/ FS
/ FS
/ FS
3FS + 2 A34
4FS
+ 2 A36 i B36 6 + 2 A33
i B33
i B34
/ FS
/ FS
5FS = F3
+ 2 A35
i B35
(K.3)
2 A41 i B41 1 + 2 A42 i B42 2 MYG 2 + C43 2 A43 i B43 + 2 MX G 3
+ C44 2 ( I 44 + A44 ) i B44 4 + C45 2 A45 i B45 2 I 45 5

/ FS
/ FS
/ FS
/ FS
3FS + 2 A44
4FS
+ 2 A46 i B46 2 I 46 6 + 2 A43
i B43
i B44
/ FS
/ FS
5FS = F4
+ 2 A45
i B45
(K.4)
2 A51 i B51 + M 2YG 1 + 2 A52 i B52 2 + C53 2 A53 i B53 2 MZ G 3
+ C54 2 A54 i B54 2 I 54 4 + C55 2 ( I 55 + A55 ) i B55 5

/ FS
/ FS
/ FS
/ FS
3FS + 2 A54
4FS
+ 2 A56 i B56 2 I 56 6 + 2 A53
i B53
i B54
/ FS
/ FS
5FS = F5
+ 2 A55
i B55
(K.5)
290
2 A61 i B61 2 MX G 1 + 2 A62 i B62 + 2 MZ G 2 + 2 A63 i B63 3
+ 2 A64 i B64 2 I 64 4 + 2 A65 i B65 2 I 65 5 + 2 ( I 66 + A66 ) i B66 6

/ FS
/ FS
/ FS
/ FS
/ FS
/ FS
3FS + 2 A64
4FS + 2 A65
5FS = F6
+ 2 A63
i B63
i B64
i B65
(K.6)
FS
FS
FS
FS
FS
FS
FS
FS
FS
C33
3FS + C34
4FS + C35
5FS
2 A33
i B33
2 A34
i B34
2 A35
i B35
FS /
FS /
FS /
FS /
FS /
FS /
1 + 2 A32
2 + 2 A33
3
+ 2 A31
i B31
i B32
i B33
FS /
FS /
FS /
FS /
FS /
FS /
5 + 2 A36
6 = F3FS
4 + 2 A35
i B36
+ 2 A34
i B34
i B35
(K.7)
FS
FS
FS
FS
FS
FS
FS
FS
FS
C43
3FS + C44
4FS + C45
5FS
2 A43
i B43
2 A44
i B44
2 A45
i B45
FS /
FS /
FS /
FS /
FS /
FS /
1 + 2 A42
2 + 2 A43
3
+ 2 A41
i B41
i B42
i B43
FS /
FS /
FS /
FS /
FS /
FS /
5 + 2 A46
6 = F4FS
4 + 2 A45
i B46
+ 2 A44
i B44
i B45
(K.8)
FS
FS
FS
FS
FS
FS
FS
FS
FS
C53
3FS + C54
4FS + C55
5FS
2 A53
i B53
2 A54
i B54
2 A55
i B55
FS /
FS /
FS /
FS /
FS /
FS /
1 + 2 A52
2 + 2 A53
3
+ 2 A51
i B51
i B52
i B53
FS /
FS /
FS /
FS /
FS /
FS /
4 + 2 A55
5 + 2 A56
6 = F5FS
+ 2 A54
i B54
i B55
i B56
(K.9)
291

PHD

Uploaded by

PHD

Uploaded by

University of Southampton Research Repository

Damage Stability of Ships as a Safety Criterion for

School of Engineering Sciences

Damage Stability of Ships as a Safety Criterion for Optimisation Tools

Thesis for the degree of Doctor of Philosophy

A literature overview of past optimisation studies revealed that whilst satisfaction of

The Ship Design Concept ................................................................................. 1

Optimisation Tools ........................................................................................... 4

A Possible Shortcoming of Existing Optimisation Tools................................. 4

Aims and Outline of the Research.................................................................... 5

THE OPTIMISATION PROCESS....................................................................... 8

A Specific Motion and Resistance Driven Optimisation Research

Discussion on Past Optimisation Studies ....................................................... 20

Statement of the Optimisation Process........................................................... 21

ANALYSIS OF DAMAGE DATA...................................................................... 23

Damage Statistics ........................................................................................... 23

Ship Design Optimisation with Different Design Drivers ..................... 14

Summary of Damage Statistics .............................................................. 59

THE DEVELOPMENT OF SHIP REGULATIONS AND STABILITY

Development of Safety ................................................................................... 61

The Deterministic and Probabilistic Approaches to Ship Stability Analysis . 66

The Deterministic Approach to Ship Stability Analysis ........................ 67

Lost Buoyancy Method ....................................................................................... 69

Added Weight Method ........................................................................................ 70

The Probabilistic Approach to Ship Stability Analysis.......................... 70

Discussion on Current Stability Methods............................................... 75

Damage Stability Suite ................................................................................... 76

Independent Damage Stability Analysis................................................. 77

Presentation of the Damaged Hull Forms .............................................. 79

General Review .............................................................................................. 82

Frequency and Time Domain Approach ........................................................ 88

Justification of Selected Approach ................................................................. 89

Motion Analysis ............................................................................................. 90

Forces and Moments Acting on a Floating Ship in Waves .................... 90

Forces and Moments Acting on an Intact Ship................................................... 90

The Equations of Motion........................................................................ 91

Equations of Motion for Intact Hull Forms........................................................ 92

Damaged Ship ...................................................................................................... 101

Determination of Motion Characteristics ............................................. 112

SELECTION AND FUNDAMENTAL ANALYSIS OF BASIS SHIP .......... 115

Choice of Hull Form..................................................................................... 115

Public Domain Particulars for Derbyshire.................................................... 116

Optimisation Objective Function and Design Parameters............................ 120

Choice of the Objective Function......................................................... 121

IMO Intact Stability Requirements and Related Analysis ........................... 125

Floodable Length Curve and Bulkhead Locations ....................................... 129

Damage Properties of the Derbyshire........................................................... 132

The Damage Analysis................................................................................... 134

Orientation of Damaged Hull Forms ............................................................ 134

HYDRODYNAMIC CHARACTERISTICS OF INTACT AND DAMAGED

Calculation of Hydrodynamic Coefficients and Excitation Forces .............. 142

Validation of Hydrodynamic Analysis of Intact and Damaged Forms of

Derbyshire ................................................................................................................ 145

RELATIVE VERTICAL MOTION ANALYSIS OF INTACT AND

DAMAGED SHIPS .................................................................................................... 175

Motion Dependent Qualities of Engineering Interest................................... 175

Selection of Points Investigated for Relative Vertical Displacement........... 176

CONCLUSIONS AND FUTURE WORK........................................................ 194

Key Observation of Completed Research .................................................... 195

Novel Aspects of Completed Research ........................................................ 199

Recommendations for Future Work ............................................................. 201

APPENDIX C - NUMERICAL APPLICATION OF THE NEW METHOD TO

Figure 5.1: The motion-hydrodynamic Cartesian coordinate reference system ............ 93

Waterplane area of compartment, that is, area between selected

Attained survival probability index

Linear motions coefficient matrix including hydrodynamic,

Added mass matrix of damaged ship. Akj is value of Akj for

Required subdivision index associated with probabilistic

Waterplane area of intact ship

Waterplane area of damaged ship

Aft and forward perpendiculars of ship

Subscripts indicating quantity associated with aft portion of ship

Either penetration of damage (m) or base points for the objective

Breadth (or moulded beam) of ship

Fluid damping matrix of intact ship. Bkj denotes fluid damping

Fluid damping matrix of damaged ship. Bkj is value of Bkj for