Sei sulla pagina 1di 80

HSE

Health & Safety


Executive
Collision resistance of
structures to side impact
MSL Engineering Limited
5-7 High Street
Sunninghill
Ascot
Berkshire
SL5 9NQ
HSE BOOKS
EXECUTIVE SUMMARY
A study on behalf of the UK Health and Safety Executive, Offshore Safety Division, has
been undertaken by MSL Engineering Limited to address the subject of side shell collision
resistance of FPSO units (Floating, Production, Storage and Off-loading units) and other
ship-shaped vessels.
Alternative offshore installation concepts are playing an increasingly important role in UK
Continental Shelf deep waters. Ship shape structures including FPSOs are being developed to
exploit North Sea resources. In light of the popularity of FPSO type units, the need to
demonstrate their structural integrity and safety requires that they be subject to an assessment
for collision resistance.
The objeCtives of this study were as follows:
To identify areas of uncertainty where the various codes differ and to identify areas of
uncertainty in the methods to assess side shell collision.
To assist HSE to develop good practice guidelines based on the findings of the study.
To achieve the objectives a review of Classification Rules and codes of practice specific to
floating production systems, selected FPSO safety cases and literature relating to the
methods of collision assessment was undertaken. Sensitivity studies were also performed on:
i) a typical tanker vessel identified from information in the public domain, and
ii) two selected FPSO type units selected from information provided by the HSE.
The overall conclusion of the study are as follows:
The codified guidance and criteria for collision can be improved by incorporating some of
the methods addressed in this report. Therefore the designer would gain a greater insight into
the mechanics of collision. The sensitivity analyses have revealed the importance of varying
either the dimensions of scantling parameters and /or the impact location of striking vessel tO
significantly increase the ability of the struck vessel to absorb energy. Guidance in this area
to indicate which scantling elements/ impact location would influence the strength of the ship
against collision would be desirable. It should also be mentioned that if the strength of a ship
is found not to be sufficient to withstand collision, then the assumption of the infinitely stiff
attacking bow should be abandoned, and the shared energy approach should be used. This
allows for energy to be absorbed in the attacking bow as well as the struck side.
The methods for analysing ship collision above can give different values for energy
absorption as identified from work published in the literature and demonstrated in the
sensitivity studies performed. Further validation using experimental work should be
undertaken in this area accompanied by hand calculations, where reliable, and numerical
simulations (such as finite element analysis) to produce a more precise general method.
ii
CONTENTS
EXECUTIVE SUMMARY
CONTENTS
I.
2.
3.
4.
5.
6.
INTRODUCTION
1.1 General
1.2 Collision Mechanics
1.2.1 External collision mechanics
1.2.2 Internal Collision Mechanics
REVlEW OF CODES AND STANDARDS
2.1 Development of UK Criteria
2.2 Det Norske Veritas (DNV)
2.3 Norwegian Petroleum Directorate (NPD)
2.4 NORSOK
2.5 Lloyds Register of Shipping
2.6 Germanischer Lloyd
2.7 International Maritime Organisation (IM:O)
2.8 International Standards Organisation (ISO)
2.9 Summary
REVlEW OF SUBMITTED SAFETY CASES
LITERATURE REVIEW
4.1 General
4.2 Methods
4.2.1 The Minorsky method
4.2.2 The Rosenblatt method
4.2.3 Kinkead's substantiation method
4.2.4 Japanese Studies
4.2.5 Towards fmite element analysis- the work of Pedersen
COMPARISON OF METHODS
5.1 General
5.2 Published experimental comparisons
5 .2.1 The Stem test
5.2.2 The Bulb test
5.2.3 Results from stem and bulb tests
5.3 International Experimental Studies
PARAMETRIC STUDIES
6.1 General
6.2 NKK assessments
6.3 Pedersen assessments
iii
ii
iii
1
2
2
3
7
8
9
10
10
13
13
15
16
16
20
23
23
23
23
25
26
27
29
37
37
37
37
38
38
41
44
44
44
46
7.
6.4 MSL Assessments 48
6.4.1 Parametric study to assess the energy absorption capabilities of
a typical tanker from information in the public domain 48
6.4.2 Parametric study to assess the energy absorption capabilities of
two typical FPSO units from information provided by HSE 50
CONCLUSIONS
7.1 Review of Classification Society Rules and Standards
7.2 Review of Safety cases
7.3 Methods of analysis
7.4 Parametric Studies
7.5 Overall Conclusions
69
69
69
69
71
72
REfERENCES 73
iv
1. INTRODUCTION
1.1 General
Collisions involving ship shaped structures have both safety and environmental concerns.
However, guidelines have tended to concentrate on the environmental hazards of pollution
rather than safety.
There has been a growing interest in reducing the risk of cargo spillage due to accidents
involving oil tankers and other ships, which carry polluting and or/hazardous cargo. As a
result regulations for tanker design have been strengthened. For example, the Exxon Valdez
which grounded in 1989 resulted in the passage of the US Oil and Pollution Act of 1990
which became law in August 1990(
0
, requiring that all tanker vessels must have double hulls
by the year 2015. Later the Marine Environment Protection Committee (MEPC) of the
International Maritime Organisation (IM:O) agreed that oil tankers must have a double hull
structure or an equivalent alternative approved by the MEPC, and added amendments to
Annex I to MARPOL 73/78 in March 199i
2
l. Since then, the Braer grounded and broke up
near the Shetland islands in the UK in January 1993, which was followed by the Maersk
Navigator collision accident near Sumutra.
With increasing awareness of the environmental damage caused by pollution and the
legislation to prevent it, designers of Floating Production Units are required to produce
designs and modifications that can effectively reduce the adverse effects of collision. The
double hull concept is one of the effective ways for oil pollution prevention in collisions.
However, the background to how the double side width or the double bottom height is
determined is not clear and a rational procedure for its determination is needed. Many of the
studies on the strength of ships during collisions have been done primarily to evaluate the
safety of nuclear ships involving major collisions in contrast to the case of minor collisions
involving ship type structures with double hulls. The emphasis of this study therefore
naturally tends towards to examining the collision resistance of double-hulled structures.
The process involved in collision is complicated. The analysis of a ship-ship collision is
usually separated into two classes. These are External mechanics and Internal mechanics.
External mechanics is concerned with the rigid body motion of the two ships involved in the
collision. The two ships are hence referred to as the struck and striking ships. Thus, external
mechanics deals with the kinetic energy of the struck and striking ships, including an
allowance for the added masses of water.
Internal mechanics is concerned with the response of the struck and the striking vessels to
collision. This specifically means the relationship between force and deformation of the two
vessels.
The methods used to assess collision are often simplified due to its complexity. Complexities
often occur with the transformation of the energy prior to collision to rotational, translation
and strain _energy after collision. This can be simplifieR by considering the time duration of
impact. When the duration ofimpact is short compared to the natural periods of vibration of
the system then the collision event is essentially dynamic. However, if the duration of impact
is long compared to the natural periods, as is normally the case, the impact can be treated as
quasi-static. Complexities also occur with internal mechanics when the distribution of strain
energy between the struck and the striking vessel is to be determined. This is dependent on
the relative stiffness of the attacking bow or stem to a struck side and this includes a number
of issues such as shape of bow, relative draughts of the two vessels and structural
arrangements.
In this study the scope will cover quasi-static collisions, since these are assumed to be the
most probable impacts to produce significant deformation. The attacking bow will be taken
as infinitely stiff, so that the damage to the FPSO side-shell is assessed conservatively.
1.2 Collision Mechanics
1.2.1 External collision mechanics
Collisions as shown in Figure 1.1 are conveniently described in terms of energy. This
involves the transformation of kinetic energy into elastic and plastic strain energy and
equivalently to elastic and plastic deformations. Collisions are by their very nature dynamic
events and can be described by the general differential equations of motion. However if the
impact duration is long compared to the natural period of vibration, the collision event can be
described quasi-statically by using the conservation of momentum principle (although this is
not applicable to all possible collisions involving FPSOs).
The conservation of momentum yields:
where:
Ma is the displacement mass of the striking vessel;
MA is the displacement mass of the struck vessel;
V a s the velocity of the striking vessel;
VA is the velocity of the struck vessel;
U is the velocity after collision
Om is the added mass coefficient
At the end of the impact, the translational kinetic energy is:
Kinetic energy = 0.5. (M
8
+ MA +Om). U
2
,
(1)

(2)
when the two vessels move as one body after collision i.e. the coefficient of restitution is
zero. The direction of collisions normal to the struck ship centre-line is of importance, in this
direction VA = 0 and the velocity component of the striking ship is V Bsin6; where 9 is the
angle of encounter. Rewriting (I) gives:
u
MB. VB .sine
MB+MA+Om'
(3)
for the component of U in the direction of collision, which is normal to that of the struck
ship. Therefore the lost kinetic energy:
E; 0.5.M
8
. (Vasin9)
2
- 0.5. (M
8
+ (MA+ Om)).
MB
2
.(VB .sin9)
2
[M, + (M, .om)]'
Where i denotes internal energy, since the lost kinetic energy is conserved in this form.
2
(4)
If MA >> Ma and that the striking ship is almost totally stopped by the platform, thus the
second part of the equation is negligible yielding:
(5)
The above assumes that the centre of masses for the struck and striking vessel have
coincident line of actions producing no rotation.
However for an FPSO, the colliding object could also be such that the masses have a similar
order of magnitude; thus equation (4) would be applicable

The case of a relatively stiff bow colliding with an FPSO is shown graphically in Figure 1.2.
1.2.2 Internal Collision Mechanics
When a ship's bow has collided with the side of a vessel, there is a global and local response
to impact. In side-ship collisions, the deformations can be quite large and as a result the
structural members in the struck ship can experience failure modes such as yielding in
bending, crushing and rupture. The damage process can be different depending on the
collision situation, as depicted in Figure 1.3.
The global response involves the bending of the struck vessel longitudinally, with the struck
side experiencing compressive stresses. These stresses are small compared to the stresses
induced in the local impact zone. In this area the lateral component of the collision force
produces bending and shear action in the stiffened side shell. This has two phases: first
elastic and then plastic. Beyond this point the side shell and supporting stiffeners
momentarily unload and act effectively as a mechanism, deforming laterally towards the axis
of the ship, until sufficient geometry is achieved to resist the load axially. Thus the side shell
resists the load as a tensile membrane up to the point of rupture.
This sequence of elastic bending, plastic bending and rotation and membrane action
describes the collapse phenomena of the stiffened side shell. Collapse progresses beyond the
side shell with the failure of the supporting web frames and intermediate, sub-deck plating.
The web frames are vertical plates consisting of a series of stiffened panels connected to the
side shells, the bottom and top deck of the ship, and the inner hull for a doubled-hull vessel.
For the hull to be breached, any horizontal sub-deck plating between the web frames and
shell will also need to fail. Finally, the worst failure case is when the striking ship has
penetrated to the centre of the struck ship, thus failure of the deck levels has been incurred.
For a double hull vessel with stiff web frames, it can be assumed that the outer and inner hull
side shells act independently when resisting impact. There is assumed to be little or no
composite action between the inner and outer shells. For an FPSO the relevant failure criteria
would not be the extreme case of penetration to the ship's centre. Instead, the rupture and
breaching of the cargo tanks should be taken as the ultimate limit state.
The above, however, only describes the collapse behaviour of the struck ship. Complexities
in idealisation of collapse occur when the both the struck and the striking vessel are
considered. The bow of the striking ship also absorbs energy from the impact, and how this is
shared between the two vessels must be known or at least assumed. Also how energy is
distributed amongst the side shell, web frame, and decks of the substructure must be
investigated. The first problem that has faced the investigators of collisions, has been the
rupture behaviour of the side shell, specifically the rupture strain.
3
When investigating collision the designer needs guidance in relation to the internal
mechanics of collapse. Two solutions must be found:
What is the energy absorbed as deformation for a given collapse situation; and thus the
maximum energy allowed if some limit to deformation for the struck ship is given?
How is the energy of collision is shared among the constituent elements of the ship(s) so that
the individual integrity of the elements can be established?
4
Figure 1.1
Figure 1.2
Ship to ship collision
Indentations
Impact force ship and
"""' coalesce
Force deflection relationship for stiff bow collision into a side
shell
5
~ outer plate
,.------
f---web
~
----'--
~ - - ~
i n ~ plate
impact on web frames
(b) crushing of the web
r- (_ tl
liiV
(c) rupture of outer plate (d) rupture of inner plate
and crushing of side webs
outer plate
vi
vweb
I
r /1
I
i n ~ plate
(a) initial stage
(b) rupture of the outer
plate
impact between web frames
~ t s --
" ~ v
::u iC
(c) crushing of the webs
(d) rupture of inner
plate
Figure 1.3 Stages in the collapse mechanism for a side shell
6
2. REVIEW OF CODES AND STANDARDS
The first point of reference for the designer is often the codes of practice produced by the
various Regulatory Authorities and/or Classification Societies. In this section, the relevant
sections of codes will be reviewed. Eight codes have been reviewed as follows:
DEniHSE<
3
J
Det norske Veritas <
4

51
Norwegian Petroleum Directorate (NPD)<
6
J
NORSOK r
71
Lloyd's Register of Shipping <
81
Germanischer Lloyd <
91
International Maritime Organisation (IMO)
001
International Standards Organisation (ISO)
011
The codes from the verifying authorities follow approximately the same approach; that being
the assessment of impact energy to be absorbed by the struck Floating Production, Storage
and Offloading vessel (FPSO). The collisions investigated generally make two types of
probable striking objects explicit:
Collision with supply vessel
Collision with tanker vessel.
The collision into the side shell of a ship, i.e. FPSO or tanker, can be put into the category of
"Accidental Load" as described by the codes. Thus the definition of striking object can
encompass:
vessels in senrice to and from the installation
tankers loading at the field
ships and fishing vessels passing the installation
floating installations such as flotels
icebergs or ice.
Probabilistic analysis such as Risk Analysis can be used to assess which or how many of
these cases are applicable to collision into the side shell of FPSO hulls. This naturally
depends on location, environmental, installation collision statistics. Most codes recommend
that some form of risk analysis be performed. This must rely on historical data in which the
possible striking objects can be categorised. Three broad categories can be used and assessed
for probability of occurrence:
Vessels having business with the installation and which may berth beside it (supply
vessels, diving support vessels, tankers etc.)
Vessels on passage through the area (merchant vessels, ferries, other offshore vessels in
transit)
Vessels which have infringed on the exclusion zone (fishing vessels, pleasure craft etc.)
From all these vessel types, there is the fact that collisions are likely to be due to both vessels
drifting and vessels under power. Collision with supply boats has been investigated as the
most probabilistic event. Thus, the codes are explicit with this type of collision. Significant
7
studies were carried out by Det norske Veritas <
12
) and are reiterated by the Acts and
Regulations from the Norwegian Petroleum Directorate (NPD) <
6
l.
Several studies were undertaken by the UK Department of Energy in the 1980's and by the
UK HSE Offshore Safety Division in the 1990's. Studies, such as those undertaken by Lloyds
Register of Shipping (l
3
J and JP Kenny (l
4
l, compiled statistical data and guidelines for
collisions based on actual collisions, in an attempt to reconcile some of the differences
between the Department of Energy and Dn V approaches. Germanischer Lloyd investigated
tanker to tanker collision <
9
l. However, the philosophy behind the guidelines for collisions
and the appreciation of external collision mechanics remain generally the same.
A summary of the codes recommendations is given below:
2.1 Development of UK Criteria
Previously, before issue of the 4th Edition, Department of Energy Guidance Notes<
3
l many
offshore structures were only designed to be able to resist a boat impact of 0.5 MJ. The
earlier editions stated that where fendering was fitted the fenders were to be designed for 0.5
MJ. This specifically applied only to the design offenders. However, where fenders were not
fitted, then the installation itself was required to be designed to meet the 0.5 MJ. The
Norwegian requirements at this time were based on DNV rules which required consideration
of a vessel of 5000 Tonnes travelling at 2m/sec when it impacts the structure. This gives an
impact energy of 11-14 MJ depending on the way in which the vessel strikes the platform.
The above estimates were based on an estimate of the typical displacement of supply vessels
and an assessment of the velocity at which a vessel, which has lost power, might be propelled
into the structure by wave action. However, the DNV rules allowed the energy of impact to
be shared between the installation and the impacting vessel. The Rules allowed assumptions
to be made about the stiffness of the vessel and the proportion of energy absorbed in damage
to the vessel structure. It is understood that in some cases the energy absorbed by the
installation would be only 1 MJ.
Although the difference in impact energy between the UK and Norway was not as high as
might first appear, there was still a disparity. Also the value of0.5 MJ looked rather low in
the light of some vessel impacts that had already occurred. As a result of this the UK
Department of Energy commissioned Lloyds Register of Shipping
03
) to investigate the
impact energy absorbed in damage to the structure for a number of boat impact accidents.
This was based on the actual recorded damage to ten structures as a result of vessel impacts.
As a result of the Lloyds work the DEn commissioned J P Kenny <
14
) to perform a further
more wide ranging review and the result of this review was the "Accidental Loads Section"
in the 4th Edition of the Guidance notes (3). The required level of impact energy being,
effectively, the same as the level of energy to be considered in the Norwegian Sector.
However in order to provide a more consistent standard the energy to be absorbed in the
structure was recommended to be a minimum of 4 MJ unless a study of the collision hazards
and consequences specific to the installation demonstrates that a lower value is appropriate.
A lower value may be appropriate for a number of reasons:
i. There is a restriction on visiting vessel size.
ii. There is a restriction on significant wave height for vessel operations.
iii. The stiffness of the impacted part of the installation is large compared to the vessel.
iv. The installation could be small occasionally unmanned and infrequently supplied.
v. Specific analysis shows it is inappropriate.
8
If the vessel size is restricted then the Guidance Notes allowed the applied energy to be
absorbed by the installation to be reduced to
E = 0.5 +m' (4.2xlo'- 5.6xw"m) Ml (6)
Reducing the allowable vessel displacement from 5000 tonnes to 3500 tonnes for instance
reduces the impact energy requirement from 4 MJ to 2.8 MJ
Recommendations are also given in the Guidance Notes for (iiiiv). Recommendations in the
Guidance for (iii), which states that in cases where the stiffuess of the impacted part of the
installation is very large in comparison to the vessel, analysis of impact energy is
inappropriate, since almost no damage occurs in the struck installation.
2.2 Det Norske Veritas (DNV)
With respect to the "Rules for the Classification of Ships" <
4
l in Appendix A clause A3. 7.1, it
is stated that for collisions, risk analysis would be necessary ''for example, in connection
with collisions and the operation of ballast system". However probable scenarios are not
discussed any further but assistance is given on the bolUldaries of associated risk with the
statement: ''the most unlikely accidental effects may be disregarded if annual probability
does not exceed I0
4
order of magnitude".
Parallel to these statements on risk Section A clause A3.7.2 gives a simple deterministic
approach as an alternative: "The hull side is to be able to absorb the energy created as a
result of a collision with a supply vessel, without resulting in the rupture of crude oil tanks.
The kinetic energy is not to be taken less than II MJ for bow or stern collision ... ". This
corresponds to a 5000 tonne displacement supply vessel colliding with the installation at a
speed of 2 mJs and an added mass component of 10% for the supply vessel. The code does
state though, that a reduced energy may be used but "subject to approval". For this, the
calculated energy should use the equation below <
4
l:
E = Y:z (m +a) y2 {kJ), (7)
Where
m displacement of the striking vessel in tonnes
a added mass of the vessel, normally assumed as 0.4m for sideways collision and O.lm
for bow and stern collision
v impact speed in m/s.
If the impact speed is caused by the wave induced motion, the striking vessel speed is given
as <
4
l
V = 0.5. Hs (m/s); (8)
where Hs is the significant wave height in m for operation close to the illlit.
Collision is to be considered for all elements of the illlit. The vertical extent of the collision
zone is to be based on the depth and draught of the impacting vessel and the relative motion
between the striking vessel and struck illlit.
The code further states that the vessel should be designed to avoid progressive collapse and
therefore should be able to withstand 1 oo.year environmental conditions while being in a
9
damaged state due to impact (clause A3.7.4). This means that.the hull girders in the grillage
must resist 100-year recurrence environmental loading while in a damaged state for the
Progressive Limit State (PLS) condition. The first element to consider in the failure criteria is
the shell plate and/or stiffener and the fmal and most important element is the girder of the
hull framing which ensures the global integrity of the vessel. DnV states that when checking
structural integrity for this PLS condition the load and material factors should be taken as
unity.
The criteria for failure mentioned in the code, (clause A.3.7.2) is such that the hull side
resists the impact energy "without resulting in the rupture of crude oil tanks". For some
vessels this could be interpreted as rupture of the hull side shell, since for these vessel the
tanks and the hull shell are one, and the same. However, in the DNV code relating to general
oil carriers <sl, section C202, specific recommendations are made for the hull to be doubled;
with two skins. The recommended distance between the inner and outer hull is a function of
the dead-weight of the tanker such that for tankers above 30,000 tons dead-weight the
minimum distance should be 2.0 metres.
No mention is made of any methods to evaluate the internal mechanics of collision
specifically for FPSO or tanker collision, and how the energy is to be absorbed by the hull of
the struck vessel. There are however general guidelines for stiffened plating and general ship
structural integrity that can be applied in lieu of any detailed analysis, but these take no
account of interaction of collapse mechanisms in impact. As such, considerations such as, for
example, glancing impact, from another vessel almost parallel to the struck vessel or from an
iceberg is not discussed in this code.
2.3 Norwegian Petroleum Directorate (NPD)
Rules established by the Norwegian Petroleum Directorate <
6
l refer explicitly to the work
prepared by Det Norske Veritas.
In summary NPD recommends exactly the same as DNV <
4
l. In the absence or risk
assessment analysis, the impact load on a installation should be considered from supply
ships. These should be " ... no less than 5000 tons and speed not less than 2 m!s. A
hydrodynamic additional mass of 40 %for sideways and 10% for bow and stem impact can
be assumed " This produces energies of 14 MJ for sideways impact and 11 MJ for bow or
stem impact.
As with DNV no mention is made of internal collision mechanics specifically concerned with
FPSO or tanker collision.
2.4NORSOK
Of all the codes and standards, NORSOK' s "Design of Steel Structures N-003" standard <
7
> is
the most detailed in its explanation of collision. The code has provisions for both external
and internal mechanics as well as strain criteria for rupture. It has a separate annex devoted to
Ship collision: Annex A - "Design Against Accidental Actions".
General precepts
In this annex (Section A.3) there is a detailed explanation of
External Collision Mechanics
10
Force deformation relationships
NORSOK is specific in its recommendations for ship collisions. The code states:
"It is convenient to consider the strain energy dissipation in the installation to take part on
three different levels:
Local cross-section
Component/sub-structure
Total system"
The code recognises that on a local level in the area of impact, plastic behaviour dominates.
For the local impact area the code defines three types of strain energy dissipation idealisation
to be used in design:
"Strength design implies that the installation is strong enough to resist the collision force
with minor deformation, so that the ship is forced to deform and dissipate the major part of
the energy.
Ductility design implies that the installation undergoes large, plastic deformations and
dissipates the major part of the collision energy.
Shared energy design implies that both the installation and ship contribute significantly to
energy dissipation. " (7)
For strength design, the struck ship is effectively assumed to be rigid so that it is subject to
the maximwn force, as determined by the load-deformation characteristic and initial kinetic
energy of the striking ship. For ductility design the striking ship is assumed to be rigid so
that now all the energy has to be absorbed by the struck ship, the resulting deformation being
determined from the struck ship load-deformation response. For shared energy design,
energy is absorbed by both vessels according to their relative response curves (see Figure
1.2).
External Collision Mechanics
The code (clause A.3.1) gives an equation for the kinetic energy absorbed as strain energy in
collision and this corresponds with equation (4) shown in Section 1.2.1.
E,
where,
m, is the mass of ship
a, is the added mass coefficient for the striking ship
Vs is the velocity of the striking ship
m; is the mass of the installation
a; is the added mass coefficient of the installation
v; is the velocity of installation
II
(9)
Internal Collision Mechanics
The codes' provision for ductility design has obvious application to the analysis of collision
into the side shell of an FPSO. However, the code provides force-deflection curves for
strength design thus assuming that the struck vessel experiences the majority, if not all, of the
plastic strain deformation. The curves shown in clauses A.3-4 and A.3-5(
7
l give the force-
deflection relationship for collision with a 5000 tonne supply boat and a 125,000 tonne
tanker respectively. The collision with a supply boat curves are based on collision with an
infinitely rigid plane wall or cylinder. The collision with the tanker is based on collision with
a relatively stiff installation i.e. the stem of an FPSO.
Therefore, the force-deflection curves given by NORSOK are not strictly applicable to the
case of collision with the side shell of an FPSO. The curves are based on the strength design.
whereas when the stern of an impacting vessel collides with the side of an FPSO, shared or
ductile design would be the most appropriate when the masses of the struck and striking ship
are of similar order. The code actually states the following if the strength design
requirements are not complied with (A.3-5): "If this condition is not met interaction between
the bow and the impacted structure shall be taken into consideration. Non-linear finite
element methods or simplified plastic analysis techniques of members subjected to axial
crushing may be employed.". The force-deformation curves are shown in Figure 2.1.
When considering ductility requirements the NORSOK code gives explanation to the
"simplified plastic analysis techniques". Figure A.3-8 in the code a curve shows the plastic
force-deformation relationship for a stiffened plate. The curve is specifically for the case
where the adjacent panels to the impacted zone provide some axial flexibility in the impacted
plate supports. This is in fact the case of a stiffened plate hulL From this curve is obtained the
strain energy (the area underneath). The parameters needed to obtain the deformation and
. energy are:
The axial spring stiffness ofthe plate, c (kN/m)
The plastic collapse resistance in bending of the member, Ro (kN)
The total resistance to lateral load, R (kN).
Guidance in establishing the value, R, is provided in section A.3.10.2 of the code where the
buckling resistance of stiffened plates is defined (see Figure 2.1).
However for structures other than the stiffened plate hull no clear explanation is given for the
mechanism of collapse post-rupture, i.e. the sequence of events that define the collapse
phenomenon, and the resistances specific to decks and web frames.
Material Properties
The strain that produces rupture is given in the code for various steel types. These are used to
obtain the limit for deformation of the stiffened plate (equation A.3.22 < ~ i.e. the lateral
deformation to produce rupture. A summary of the recommended rupture strains is presented
in Table 2.1.
12
Table 2.1
NORSOK: Rupture strain and steel grade
Steel grade e.
S235 20%
S355 1 5 ~
S460 IOo/o
Energy Requirements
The energy requirements are given in Annex L of the code (special design provision for ship
shaped units). The specific energy absorption requirements of collision comply with those of
the other Norwegian codes. It also uses the case of the 5000 tonne displacement supply boat
travelling at a speed of 2 m/s; the added mass coefficients are also 10 ~ and 4 0 ~ for bow
and stem impacts respectively. The code also states that the zone of collision, which depends
on the relative draughts and motions of the attending units should be considered.
Summary
The NORSOK code is very comprehensive. The fact that it does not explain the analysis of
collision in intricate detail is not a handicap. Most codes are there to give guidance on design
only. The forces and stresses for a given situation are to be determined by the designer. Thus
the behaviour and forces generated in collapse are assumed kno'Wll and assessed by the
designer. The force-deformation curves are given to show that they are material dependent
(the resistance is dependent on the critical strain) and as such are part of design guidance.
Thus, once the behaviour of the impacted vessel is assumed and the sequence of events that
describe collision are ascertained, then the NORSOK code gives good guidance for design.
Note other elements of the ship, which are subject to collision forces, can be designed in
accordance with Section 6.6 of the code- Design of Plated Structures.
2.5 Lloyds Register of Shipping
Lloyds also adopt the deterministic approach and repeat the recommendations of the
Norwegian codes- specifically DNV, in the document "Rules and Regulations for the
Classification of Floating Offshore Insta/lations'"-
8
). The source for this code and the
Norwegian codes are the statistical data and recommendations commissioned by the
Department of Energy in 1985 (IJJ.
2.6 Germanischer Lloyd
General precepts
A different approach is taken with this classification authority. Germanischer Lloyd<"l are
more specific in their recommendations. Their basis for collision resistance assessment is the
definition of a critical situation.
The critical situations are:
tearing of cargo tanks with subsequent leakage of oil, chemicals, etc.
water ingress into dry cargo holds during carriage of particularly valuable or dangerous
cargo.
tearing of fuel tanks with subsequent leakage of fuel oil.
13
However Germanischer Lloyd specify a "Critical Collision" speed at which the bow of the
ramming ship just touches the wall of the tank for the struck ship. Therefore this code
concentrates on critical deformations for the most probabilistic collision type between two
vessels, and thus the calculation of"Critical Energies" absorbed and "Critical Speeds" which
result in the deformations. The critical speed is given as:
2.75 (10)
Ec, deformation energy, once critical speed has been reached, in [kJ]
m
1
mass of the striking ship, including 10% hydrodynamic added mass, in [t]
m
2
mass of struck ship, including 40% hydrodynamic mass, in [t]
Vcr critical speed in knots.
The recommendations differentiate between bow shapes and draught levels for calculating
mean critical values. In doing this Germanischer Lloyd have moved away from the
deterministic approach of specifying a particular energy that should be resisted, but rather
towards the probabilistic approach in which the "Critical Energy" and ''Critical Speed" are
specific to the vessels under investigation. The aim of this approach is to standardise
collision cases referenced to the critical speed.
In the classification scheme there are two types of ship, these are given below together with
their critical conditions:
Unstrengthened ship (single-hull ship): side shell ruptures
Strengthened ship (double-hull ship): bow of ramming ship touches cargo walls
This differentiation is not made by other codes.
Germanischer Lloyd uses the value of "Critical Speed" to classify strengthened ships in a
scale of collision resistante with the "COLL" notation, ranging from 1 through to 6, with
increasing resistance as given in Table 2.2. This gives the advantage of comparing and
assessing the collision resistance of FPSOs.
External Collision Mechanics
The code makes no recommendation here. This is because of Germanischer Lloyd's
emphasis on a critical situation. Thus, the code concentrates on internal mechanics - the
energy and consequently the speed to produce the critical situation.
Internal Collision Mechanics
The code states that the calculation of the energy to produce deformations can be based on
the Minorsky method for high-energy collisions. For low energy collisions, i.e. collisions that
produce minor penetrations, the codes states that the Minorsky method does not give
sufficiently accurate results. Therefore for these collisions the code recommends the use of
the ultimate load method " ... which takes into account the ultimate loads of the bow and side
structures hitting each other in the area calculated, and their interactions. The computations
of the ultimate loads are to based on the assumption of ideal elastic plastic material
behaviour. "<
9
>. Thus the energy that is calculated, from either Minorsky or by elastic-plastic
methods, is used to determine Vcr in equation (10).
14
However, for classification purposes, it is not just one absolute value of energy that is to be
calculated. The code specifies that mean critical energies be determined. These mean values
are to be determined from four different collision cases, which are related to 4 separate
draught level differences (see Figure 2.3) between the two colliding vessels (i.e. clause sfl_
Each of these four draught levels are calculated for two different bow shapes for the striking
ship: raked bow with bulb and raked bow without bulb (see Figure 2.3). The four mean
energies for each bow shape (eight in total) that result in this investigation are then used to
establish a minimum Vcr to then produce the COLL classification as shown in Table 2.2.
Table 2.2
Germanischer Lloyd classification
"COLL"-notation
1.0
2 1.5
3 2.5
4 4.0
5 5.5
6 7.0
v"
is the mean critical speed
For the Germanischer Lloyd methods, a range of critical energies are calculated which are
dependent on the bow shapes of the struck and striking vessels. A mean of these values is
then determined, to give the mean critical speed Ver to produce the critical situation.
Germanischer Lloyd has developed a program to be used in their ship structure design
program- POSEIDON, investigating ship collision. Development of this method has come
from full-scale, experimental tanker-tanker, side shell collision tests (
9
,
9
a>_ This employs a
modified substantiation method similar to the Kinkead method outlined in Section 4.2.3 in
this report.
Material properties
For the Ultimate load method mentioned above the material is assumed to have a limit stress
equal to the mean of the yield stress and the tensile stress. The rupture strain is not dependent
on the material and is given a value of5%.
Summary
The emphasis of Germanischer Lloyd's work is the classification of vessels on the basis of
strength. Therefore collision resistance i.e. the critical situation, is calculated specifically for
a given ship. From this critical situation are derived energy and velocity, which are used to
classify the impacted vessel. This is different from the other codes that specify an extreme
impact energy that a given vessel has to resist.
2.71nternational Maritime Organisation (IMO)
The International Maritime Organisation (IM0)
00
> is specific in its criteria for damage, as
regards the positioning of cargo tanks. In the IMO codes, dangerous cargo is recommended
to be located at a distance from the side shell of0.76m (minimum). Rupture or damage is to
be limited to 20% of the ship's breadth with a maximum value of 11.5m.
!5
In 1992 the Marine Environment Protection Committee (MEPC) of IMD agreed that oil
tankers must have a double hulled structure or equivalent alternative approved by the MEPC.
The minimum hull depth required being 2 m, which is similar to the recommendation in
DNV(
5
l.
However there is no mention of an energy or velocity that the vessel should resist, to provide
for strength against collision.
2.81nternational Standards Organisation (ISO)
Section 3.3 of the current draft ISO code for Floating systems (Ill covers the structural design
of conventional ship-shaped floating production systems (i.e. floating production unit,
floating storage units, floating production, storage and offloading installations). The code
emphasises the need that structural design shall follow the rules and standards of a
Recognised Classification Society (RCS), the units flag state (when applicable) and the
sovereign state in whose waters the unit 'Nill operate. Section 3.3.3.3 deals with the minimum
impact energy levels that must be satisfied. The relevant clause states that "minimum impact
energy levels shall comply with requirements of aRCS. Collision shall be considered for all
elements of the unit, which may be impacted by sideways, bow or stem collisions. The
vertical extent of the collision zone shall be based on the depth and draught of attending
vessels and the relative motions between the attending vessels and the unit".
Section 1.5 .5 of the code specifically emphasises the need for the structure to be designed so
as to behave in a ductile manner to absorb energy caused by accidental loads. Measures to
obtain structural ductility are given including the avoidance of energy absorption in slender
struts with a limited degree of post buckling reserve strength for the unit.
There is no explicit mention of criteria for failure ofFPSOs. As with some of the other codes
there is no mention of how the energy should be absorbed (internal collision mechanics). As
yet, it would appear that the current ISO recommendations rely heavily on the requirements
of aRCS.
2.9Summary
A tabulated summary of the codes reviewed in this section is given in Table 2.3.
16
Table2.3
Summary of codes reviewed
Ref. Title Relevant Summary
No. Sections
4 Rules for the Classification of Part 5. Ship must resist rupture of cargo
Ships- Oil Storage and Chapter 9 tanks; 11 MJ sideways collision;
Production Vessels 14 MJ stem or bow collision; 40%
Det Norske Veritas 1996 added mass for struck vessel; 10%
for striking.
5 Rules for the Classification of Part 5. Minimum and maximum
Ships- Oil Carriers Chapter 3 dimensions for the protection of
Det Norske Veritas 1996 cargo tanks.
6 Norwegian Petroleum Section 6, Similar to Det Norske V eritas.
Directorate- Acts, Vol. 2
Regulations and Provisions
for the Petroleum Activity
Norwegian Petroleum
Directorate 1993
7 NORSOK Standard Annex 3 As in Det Norske Veritas, but has
detailed explanation of external
and internal collision mechanics of
collision- Force-deformation
curves.
8 Rules and Regulations for the Part 4: As in Det Norske Veritas.
Classification of Floating Chapter 3
Offshore Installations.
Lloyds Register of Shigging
1999
9 Rules for Classification and Ship Definition of critical speed and
Construction Technology method to classify ships in order
Gerrnanischer Lloyd Part 1 of collision resistance.
10 International code for the Sections Gives minimum dimensions for
construction and equipment 2.5. 2.6 the location of cargo to prevent
of ships carrying chemical in damage; 2m double hull depth
bulk recommended.
IMO
11 ISO Draft As in Det Norske Veritas;
recognises the need to investigate
different draughts and bow shapes.
17
0 C.& 1.5-
Df:Jonna!iOfl tml
2
Indentation tm\
c
~
c

,g
~
~
"
:!


"
-" z
&,4rJ

e

...
'"
.::o
~ o

"
Figure 2.1
2.5 3 3.5

. ..
'
'
I
. (?:
'
i
l--<
v>
ic/
f.
[,9
.
\
b ; ~
!!
:. \
, .... :
DeformatiOn tm1
NORSOK force--deformation curves
18
4
.,

'
'
4,5
3,5
'
3
'
!
C,5
D
D
'
0,5
E
0
0
"'
" 0
'
i
'
'
I
j
I
ILl
I /
v
c=L
'
/
fl;,/
oy
'
//
f//f''/
v
~ r:--::..:::: ~ ' '
(,
.
'
' ' .
,.
.
.
. '
.
: :; ; : ' '; l : ~ : : . .
.
I
/
)/j
/
/
/1
)
/ ~ ; / -
/'
.
/
.
.
'
'
'
' In
i i
--Effiling & membrtllle
- Mem':lra= an!y
f (rollisJOnio.id)
~ . i A!"1
K ~ - K
1,5 2
''
3 3,5 4
Defcwme.l.lon W
Figure 2.2
NORSOK force deformation curves for stiffened plates
E
oo_
~
1,1 m
B=28,5m
Figure 2.3
Germanischer Lloyd's two bow shapes: bulbous and raked
19
3. REVIEW OF SUBMITTED SAFETY CASES
A review of eight safety cases, comprising 12 collision scenarios, was undertaken. The
purpose of the review was to:
Identify current operator philosophies for collision assessment
Identify methods and procedures used
Gather views on perceived accuracy of estimates.
It was apparent from the review undertaken that most submissions recognised the potential
hazard associated with collisions. Impacts resulting from either shuttle tank operations,
supply and passing vessels were noted as being of potential concern.
Operations involving shuttle tankers (i.e. loading/offloading) immediately prior to
connection and during departure were identified as representing the greatest risk. However,
due to the operational procedures adopted prior to shuttle tank approach, collisions were
unlikely to be of major concern. Drifting was identified as uhlikely to cause a collision, as
the shuttle tanker is always located downwind of the FPSO. The most likely scenario
identified was engine failure close to installation but when speed is low (i.e. 0.25mls -0.5m/s)
such that any damage would likely to be minor.
However, side shell collisions resulting from supply vessels and passing ships were treated
with more concern particularly high-energy collisions, which could lead to the rupturing of
the cargo tank, resulting in cargo spillage, potential of possible flash fire due to release of
hydrocarbons and mooring failure.
It was generally recognised that significant damage would result from large merchant
vessels. However, the energy absorbed from impacts resulting from supply vessels up to
5000 tonnes at 2m/s (i.e. 14 MJ), as specified in class rules such as Lloyds and DNV rules,
was assessed to result in only minor damage in most cases. Typical ranges of vessel sizes,
impact speeds and energy to be considered, identified from cases examined are given in
Table 3.1.
Table 3.1
Range of impact scenarios considered in safety case review
Vessel Type Collision Type Displacement Velocity Energy
Tonnes m/s MJ
Supply Alongside 2000 0.73 0.5
Supply Manoeuvring 2000 1.29 1.6
Supply Drifting 2000 1.82 3.3
Shuttle Tanker Approach 80000 0.5 10
Merchant Drifting 1600 1.5 1.8
Merchant Passing 1600 6 28.8
Merchant Drifting 250000 1.5 281
Merchant Passing 250000 6 4500
20
A review of the safety cases also identified a range of different damage criteria resulting
from different collision scenarios. Typical features identified were as follows:
i. Minor damage limited to denting. Total kinetic energy generally in the order of 5 MJ
(including added mass) estimated to penetrate the plating of a wing tank if FPSO is hit
between two frames.
ii. Penetration of wing ballast tanks only (i.e. cargo tanks are intact). Energy generally
estimated to penetrate wing ballast of the order of 15 MJ.
111. Penetration of wing tank and cargo tank- Ignition and fire possible. For collisions leading
to penetration of the centre tank estimates uncertain although initial estimates of 150 MJ
noted.
iv. Global Collapse- Penetration of wing tanks and cargo followed by fire with severe heat
loads. Energy greater than 150 MJ.
v. Mooring Failure. Estimates of collision energy to cause catastrophic mooring failure
were around 200 MJ.
Using the above criteria, the extent of damage and energy estimated for each of the safety
cases was undertaken and the results are presented in Table 3.2. From the review there was
limited information provided on the methods used to estimate the extent of damage and
resulting energy absorbed. However, it was noted that in cases involving high-energy
collisions, detailed finite element analyses had been undertaken.
From Table 3.2 it can be observed that in some cases a range of energy absorbed had been
provided. For example for Case B the energy for both minor/moderate and severe damage
varies by a factor of3 (i.e. 36 MJ compared with 12 MJ and 50 MJ compared with 150 MJ).
The reasons given for this were attributed to studies on dynamic ship collisions which had
shown that 33% of the impact energy would be absorbed by the FPSO with the remaining of
the energy being absorbed by the impacting vessel through structural collapse of its bow
area. However, the source of these studies could not be identified from the documents
reviewed.
The range in energy absorbed for minor/severe damage for Safety Case F was associated
with different assessments undertaken. However details of the different assessments could
not be identified in any detail although the range of energies suggest that some account of the
energy to be absorbed between FPSO and colliding vessel was undertaken.
It can be seen from Table 3.1 that collisions involving supply vessel and shuttle tankers have
potential for some local damage (i.e. energies up to 10 MJ) but are unlikely to be severe.
Passing vessels have the potential for severe/catastrophic damage.
It would appear also from Table 3.2 that collisions involving energies greater than 150 MJ
could result in global collapse. However it should be noted that the collision energies given
in Table 3.1 do not take account of the added mass which depends on the type of impact (i.e.
broadside/stem). Further more the energy to be absorbed does not distinguish between that to
be absorbed by the FPSO and that by the colliding vessel. In reality the energy absorbed will
depend on the relative stiffness and motions of the FPSO and colliding vessels.
21
Table 3.2
Range of collision energies for minor and severe damage
Safety Minor/Moderate Severe Damage
Case DamageMJ
MJ
A 15 50
B 12 50
B 36 150
c 36 150
D 200
E 10 280
F 6-24 110-122
F 15 96
G 15 96
H 15 110
In some safety cases the need for mitigating the risks associated with high-energy collisions
were addressed by means of the following:
i. early warning radar/navigation systems
ii. use of thrusters
nt. QRA.
In the case of QRA the risks associated with impacts from various vessels in the vicinity of
the FPSO (i.e. shipping routes) were evaluated using industry-recognised models such as
Crash and Collide.
22
4. LITERATURE REVIEW
4.1 General
MSL undertook a review of a number of technical papers to identify and collate available
methods and data for the assessment of side shell collision resistance. These consisted of
published papers in the public domain concerned with the collision of ships and FPSO type
vessels. Where possible papers were categorised into the following areas:
Papers involving detailed methods of collision analysis
Papers involving numerical analyses
Papers involving experimental work and simulations
A list of papers that were identified from the literature reviewed are shown in Tables 4.1 to
4.3. It should be noted that a number of other papers (not shown in the tables) were identified
as being possibly relevant to the current study. However, unfortunately English versions of
these papers could not be obtained or translated within the timescale of this study.
From the review of the papers given in Tables 4.1 to 4.3 a number of methods were identified
as follows:
The Minorsky method (including modified versions)
The Rosenblatt!McDermott method
The Kinkead substantiation method
The Japanese Collision studies
The work of Pre ben Temdrup Pedersen
4.2Methods
The attempt to find a universal method for the calculation of collision resistance of ships has
its origins in the pioneering work ofV.U. Minorsky. This empirical method was an attempt
to give an estimate of the amount of energy absorbed in a collision as a function of
penetration. This method was developed and expanded upon by Woisin and Haywood. The
Minorsky method was found to be satisfactory for large penetration impacts (high-energy
collisions) but for small penetration impacts the results were found to be inaccurate.
Rosenblatt and Kinkead developed ultimate load methods to assess the energy needed to
produce collapse in the impact zone. To further assess the progressive nature of collapse
Hisashi Ito and his colleagues at the Mitsubishi Corporation carried out a number of
experiments. These experiments were used to develop matrix methods of analysis and hence
non-linear finite element simulations. Pedersen used experiments to also develop computer
simulations with the aim being to establish optimum designs and to investigate how the local
impact interacted with the ship's structure globally
4.2.1 The Minorsky method
In 1959 V.U. Minorsky published a method based on assessment of actual data in a paper
entitled "An Analysis of Ship Collisions with Reference to Protection of Nuclear Power
Plants .CIS)_ This study of26 collisions found a method of determining a relationship benveen
the kinetic energy lost in collision and the volume of indentation for the side of the ship. The
basis of the method is equating the kinetic energy of the striking vessel (resolved
perpendicular to the centreline of the struck ship) and the plastic work done in crushing
23
transverse frames and deck plating. As previously shown, from external mechanics we have
the principle of momentum (see equation (1 )).
In equation (4), from the momentum and energy equations we have an expression for the
kinetic energy stored as strain energy in the attacking and impacted vessels.
For impacts of short duration, a hull vibrating transversely in deep water has an added mass
of 0.4 M
3
_ This simplifies the expression for lost kinetic energy to:
(
MA .M, J ( . e)'
. VB .sm
1.43MB + M , ~ .
(II)
This is the kinetic energy that produces the indentations in both the struck and the striking
vessel. What Minorsky showed was that by comparing this energy, Er with the sum of
indentation volumes for bulkheads, stiffeners and main structural members for both vessels
(the Resistance factor, Rr); the following relationship could be established:
E, = (47 R,+ 33) (MJ), (12)
Where
Er is the absorbed energy (MJ)
Rr is the resistance factor (crushed volume of steel) (m\
This relationship is illustrated in Figure 4.1.
The structural elements investigated comprised the decks, flats and transverse bulkheads for
the struck vessel, as well as the longitudinal bulkheads and a portion Of the side shell for the
striking vessel. Therefore the first part of equation (12) refers to the crushing of elements in
the direction of collision. These are elements such as transverse bulkheads and horizontal
deck plating. The second term relates to the out-of-plane deformation of the hull sides which
causes elongation and membrane tension. However, Minorsky ignored the effect of the side
shell for the struck vessel; because at the high velocity impacts he was investigating, this
gave little or no resistance due to rupture (high-energy impacts). Thus at low energy impacts,
where the side shell provides significant resistance, the energy bore no correlation to
indentation volumes. This limits the use of Minorsky's theory to high speed or high-energy
impacts.
Minorsky's work was based on nuclear powered ships and as such ships of a certain
configuration type. To account for LNG carriers-FPSO type vessels, specifically the rib
stiffening on their outer and inner side shells, Haywood modified the original Minorsky
formula in 1971 ( ~ note on collision estimates for LNG carriers') <
16
>:
E, = (48.3R, + 182.2) (MJ) (13)
Where
Er is the absorbed energy (MJ)
Rr is the crushed steel volume (m).
In Haywood's interpretation of the fonnula, the first term of the equation represents the
amount of energy needed to travel through the ship and to fail and crush the ship elements
beyond the hull. The second term is the energy needed to break and rupture the side shell and
its stiffeners.
24
The original Minorsky formula has also been modified since this time to include the effect of
the ruptured shell plate for the struck ship by G. Woisin
7
l_ Woisin's modified Minorsky
formula is given below.
E, = (47R,+ 0.49I(h. t')). (MJ)
Where
Er is the absorbed energy (MJ)
Rr is the crushed steel volume (m
3
)
h is the height of the rupture of the side shell for the struck vessel (m)
is the side shell thickness (em).
(14)
Thus, the emphasis for all these formulas is on a direct relationship between absorbed energy
and deformation. As long as the material has not exceeded its strain capacity through
elongation, the relationship is assumed to hold true. With known characteristics of a ship's
scantlings (i.e. the deformation behaviour and collapse mechanisms) and safe limit of
penetration, the impact energy can be determined.
However, these empirical formulae for energy absorption were based on ships with hulls
strengthened with anti-collision barriers to protect nuclear power generators. The presence of
this form of strengthening is absent for most FPSO and tanker type structures. Tills means
that lower energies than those investigated by Minorsky could produce significant
penetrations. Therefore what can be termed the "Post-Minorsky" methods were developed in
the following decades specifically for oil carriers, and would be applicable to both high and
low energy impacts.
4.2.2 The Rosenblatt method
The Rosenblatt/McDermott method was developed to cover the shortfall in Minorsky's
theory, ie. response to minor collision in which the cargo tank remains intact. The method
was extensively presented in the 1974 paper "Tanker Structural Analysis for Minor
Coilisions"<
18
>. The emphasis in this method is to determine the ultimate strength of the hull
prior to rupture.
The Rosenblatt.l?vfcDermott method is a two dimensional method applicable to single and
double hull ships subjected to minor collisions. The assumption, when using this method, is
that the elastic energy is negligible compared to the plastic deformations at impact.
The part of the side shell involved in collision is idealised as a series of independent
(parallel) tee beam units spanning longitudinally between transverse bulkheads or web
frames, and restrained at these connections. Each tee stiffener comprises a portion of the hull
side shell and a longitudinal stiffener. The aim is to fmd an ultimate load to produce rupture.
The method is analytical and assumes that the tee behaves as a beam in plastic bending.
When the tees buckle, the beam units behave as mechanisms, rotating about the plastic hinge
until sufficient geometry is reached to resist impact axially. Thus, the side shell resists the
load as a tensile membrane. The supporting transverse bulkheads (web frames) resist the
impact by acting as plates in compression and shear, and consequently fail by this load path.
With this method the added resistance provided by the structure by membrane action is
asswned. Hence the strength of the material in the longitudinal direction of the side shell
provides resistance.
25
If the characteristics such as ultimate strain {t:) and the plastic bending moment capacity are
1m own, then the ultimate load capacity of the hull can be determined.
The Rosenblatt theory is essentially concerned with progressive collapse. The strength of the
web frames and deck must be lmown. If they are less than the force induced by membrane
tension on the hull up to rupture then the collapse model extends horizontally so that
equilibrium is maintained. Thus the mechanism of collapse is directly related to the forces
induced in the hull just prior to rupture.
However although this method takes into account the elongation of the hull shell in the
longitudinal direction, elongation in the vertical direction is not accounted for. This method
can therefore overestimate the response to impact or underestimate it, since the true three-
dimensional response to impact is ignored. Figure 4.2 provides a flow chart of the various
stages in the collapse mechanism assumed.
4.2.3 Kinkead's substantiation method
The method described by A.N. Kinkead in his 1979 paper, ':A Method of Analysing Cargo
Protection Afforded by Ship structures in Collision and its Application to an LNG
Carrier'>(
19
J considers collision as a three-dimensional process. He used the example of a
LNG carrier with 12,200 dwt, and a loaded displacement of24,075 tons. This was subjected
to hypothetical impacts from six striking ships ranging frorh 4,860 to 121,400 tons
displacement, and impact bow locations ranging from 17.2- 37.4 ft in bow height position
above water level respectively, as shown Figure 4.3.
In this method the damage to ship is divided into its constituent parts. The method does not
use force equilibrium to develpp the direction for the mechanism of collapse. Rather the
extent of a particular collision damage is either assumed or 1m own. The elements of the ships
damaged in the collision are ~ e n assessed separately to determine the energy required to
produce damage or failure.
This method therefore has a direct link with Minorsky in that the damage is either assumed
or lmown at the outset. Indeed, Kinkead shows that with high energy impacts his energy
values correlate with those using Haywood's modified Minorsky formula. However, unlike
Minorsky the damage involved does not depend on volumes, but rather the known plastic and
buckling formulas to produce a given deformation. Thus where Minorsky would establish the
crushed volume of each element and multiply it by a coefficient to establish the energy,
Kinkead establishes the energy needed to rupture, bend, shear, tear or buckle each element.
The sum of these energies is then the total energy absorbed in impact. The mechanisms of the
energy absorption are as follows:
The membrane energy of the outer and inner hull shells until rupture
The energy produced by the rupture in the vertical direction of the outer and inner hull
shells
The energy to produce the rupture of the inner and outer side shells in the horizontal
direction.
The en_ergy to wedge through a deck(s) (if applicable)
The energy to bend the side shell plastically as the striking ship advances through the
struck ship.
The energy to crush the web frame plates (transverse bulkheads)
26
However these mechanisms or collapse events may or may not be dependent on each other.
This is the major difference between this method and that of the Rosenblatt method, which
shows progressive collapse of dependent units. Thus with the Kinkead method the separation
of the collapse process into discrete events could lead to over or under estimation of the
energy involved in collapse. This method is applicable for analyses involving raked stem
impacting ships rather than the more recently developed and quite prevalent bulbous type
bow. The decision by Kinkead to concentrate on raked stems was arrived at by consultation
at the time with two leading authorities and in addition examining the literature and test data.
In the latter Akita (lO) had concluded that the bulbous bow structure is soft in comparison with
a conventional ship stem and this seems to have been substantiated by tests both in Japan and
in Germany. Of the former, DNV (lll calculated collision damage for both types of bow in
relationship to an LNG carrier and found that the most serious damage to the hull side is
given by the conventional bow.
4.2.4 Japanese Studies
The Japanese studies identified refer to the early work of Ando and Arita (
22
l, and later by
Hisashi Ito (
23
"
27
>. The analytical methods developed by the Japanese are derived from
experiments from which simplified methods of analysis were then developed and used to
derive numerical energy absorption characteristics for different tanker like vessels.
4.2.4.1 Ando and Arila
Just as Kinkead reinterpreted and built upon the work of Minorsky, Ando and Arita built on
and improved the work of Rosenblatt. The Rosenblatt method had assumed that resistance to
impact was mainly longitudinal, along the axis of the ship. However Ando and Arita aimed
to quantify the three-dimensional behaviour whilst using the progressive collapse method
proposed by Rosenblatt.
Ando and Arita published their work in "A Study on the Strength of Double-Hull Structures
in Collision" (1976) <
22
>. This energy method was developed for floating type double hulled
oil storage tanks. The experiments described were carried out on scale models with the two
hulls separated by rectangular plates producing a square pattern in plan. This arrangement
was considered to be sufficient since the experiment was to investigate minor collisions
where the resistance to impact is provided by the hull fabric. In the experiment, the attacking
bow was assumed rigid, in order to better correlate energy absorption and deformation. From
the experimental results, it conceived the destruction mechanism of a double -hulled
structure into three modes:
Stretching of side shell like a plastic membrane which is supported by girders and webs
Buckling of girders and webs whose reaction forces are kept constant
Additional membrane forces of girders and webs which are col1apsed
The experiment showed that by recognising that the impact load was absorbed by the plate;
in both horizontal and vertical directions, the side shell absorbed a large amount of energy
prior to rupture. Figure 4.4 shows a typical stress pattern and developed deformed shape of
shell during one of the tests undertaken (i.e. after breaking of outer plate). In the method
developed- from the experiments, the shape of the deformed shell could be approximated by
triangular or quadrilateral segments as shown in Figure 4.4. The equilibrium condition was
obtained by minimising the total potential energy of the system. The damage develops
outward when the girders and webs which support the plastic membrane of the side-shell
buckle. This method therefore accounted for the damage developed in both the longitudinal
27
and vertical directions without a restriction of main direction of structure, such as in the
Rosenblatt Method. However, the test model was a simplified version of the actual condition
of ships. The double-hulled structures which can be treated using this method are limited to
the simplest form where girders and webs have the same crushing strength and the spacing of
members are square. Moreover, the colliding point on the side structure is limited to
locations, which are structurally symmetrical, and the loading shape is modelled by a straight
line. Furthermore the side shell idealised in the models do not account for the stiffening
arrangements that occur in practice.
4.2.4.2 Mitsubishi (the NKK experiments)
The work carried out by Mitsubishi (Hisashi Ito et al.) was developed to take account of
more complicated conditions and improved upon the work of Ando and Arita This work was
published by the Society for Naval Architects in Japan in 1984 <
23
l, 1985 <
24
l and 1986<
25
). The
validity of the method was confirmed by many experiments using a wide variety of a series
of models of double-hulled structures. It improved on the Ando and Arita method <
22
l
specifically by replicating the complexity of real ships scantlings in the test model-stiffeners,
plates, voids, welds, etc. The main advancements that were developed are listed below as
follows:
applicability to more complex ship structure arrangements- detail modelling of a
structure on a stiffener level
elimination of the restriction of structures and damaged shapes (i.e. orthogonal treatment
of side shell; applicability to asymmetric structures; automatic damage development in a
weaker direction)
arbitrariness of the shape of a stem.
Through these modifications the damage simulation could be undertaken without unrealistic
assumptions. Thus, the test data was not based on an over-simplified idealisation. In addition
to this an attacking ship model with a bulbous bow was also used to impact onto the test
model. Therefore, nvo types of impacts were looked at specifically<
23
l collision with the stem
of the bow and collision with the bulb of the bow. The double-hulled structure models
consisted of the side shell, the inner hull, the upper deck, the bilge shell, the transverse webs,
the side stringers and the stiffeners. The scantlings of the members were approximately
proportioned to those of a 100,000 DWT type double- hulled structure (i.e. 1110 scale). The
impact location was either the centre of the transverse webs or the centre of the side stringers.
Further details relating to the failure mechanisms, load-deformation and energy-deformation
characteristics are presented in Section 5.0 of this report. From the results from these tests a
numerical method known as the NKK method was developed that had as its basis the
following equation:
is the total potential energy
is the work done by the membrane forces
is the work done by the webs and girders supporting the hull
is the work done by external forces.
28
(15)
This equation is applied to nodes at various points along the side shell to develop a matrix
method of analysis. The main conclusions from the NKK experiments <
23
) were:
The side shell, which is impacted by a bow, can be considered as a membrane resisting
the collision by tension. The web frames supporting the hull can be considered as
buckling members which support the plastic membrane, in the case of a minor collision
The inner hull of a double hull ship hardly deforms until it is actually struck by the bow.
A simplified numerical method can be developed to relate the amount of destruction to
the destruction load, during impact.
A series of further static destruction tests <
24
'
25
) covering a wide range of scantling and
strength distributions were investigated, which included the double sides and double bottoms
of ships. Data was obtained from 10 ships, which consisted of four bulk carriers, two open
bulk carriers, an ore/bulk carrier, a general cargo ship, a heavy cargo ship and a gas carrier.
The range of capacities varied from 9,300 DWT- 190,000 DWT. The validity of the analysis
was confirmed by comparing the calculated results with the experimental results. Finally, the
effect of changing the design parameters (i.e. the side shell thickness, the transverse web
thickness and the double hull breadth) on the collision strength was examined. This was done
by using the analysis method, where a double-hulled ship of 100,000 tons displacement was
used as a standard ship type; and was carried out using the side shell thickness, the transverse
web thickness and the double hull breadth as parameters. It was found that a ship can be
significantly strengthened and be able to absorb more energy by increasing the transverse
web thickness. Further details relating to these parametric studies are presented in Section 6.0
More recently in 1994 Hisashi Ito <
27
) has extended his work and undertaken assessments
using the method developed to demonstrate the collision resistance of a 290,000 DWT type
double-hulled tanker designed by NKK compared to a single hull tanker of the same size. A
number of different collision scenarios were examined (i.e. different impact locations). The
results from these studies indicated that the strength of a double-hulled design by NKK was
remarkably high compared to a single hull tanker. Further details of these comparisons are
presented in Section 6.0
It should be noted that the analysis method developed is based on the following assumptions:
1. The stem of the striking ship is rigid
u. One ship strikes the other at a right angle
iii. The behaviour of structures can be treated quasi-statically.
The authors indicate that although these assumptions are in some sense still open to
discussion, they are generally accepted in research in this field. The analysis method is also
considered to be appropriate until the stem of the colliding ship reaches the inner hull of the
ship that is hit. The experiments were however, replicating the local response of the side shell
to impact. The effect on the global ship structure was not considered. Thus, the so-called
"coupling effect" between global and local failures was ignored.
4.2.5 Towards finite element analysis- the work of Pedersen
Numerical methods have been developed for general simulation of ship/ship collisions. The
finite element method is one of the powerful calculation tools, which has developed
significantly. However a weak feature of the finite element method is that it requires
enormous modelling efforts and computing times for large ship structures. Therefore, almost
all efforts in the development of new calculation methods have been focused on reducing
29
modelling and computing times. Pettersen <
281
developed a simplified nonlinear fmite
element method in which yielding and crushing are carefully considered. Similarly Egge and
Bockenhauer <
291
of Germanischer Lloyd presented detailed FEM procedures for analyses of
sidecollisions which include rupture of side shells. Since these methods are based on the
finiteelement approach, modelling and computing times are still so large that these
procedures are not suitable for design. A significant amount of work in this area to reduce the
modelling and computing times was developed by Pedersen <
30
'
31
J. This method concentrates
on reducing the number of degrees of freedom. Modelling the object structure with very
largesized structural units significantly reduces the number of degrees of freedom. The
idealised units in such a process can then be used to model the actual non linear behaviour of
larger structural units.
The main features of the model are described. \\Then a ship collides with another ship, the
side shell plates of the struck ship will be subject to the action of membrane tension in the
shell plating. To determine these forces, the outer and inner side shell plates are modelled by
membranetension, triangular/rectangular plate units with a stiffness matrix formulated by
considering rupture. Transverse webs and side stringers, which connect the outer and the
inner hulls, are modelled by rectangular plate units which take into acCOWlt yielding,
crushing and rupture. The effect of stiffeners on stiffness and strength is modelled by means
of an equivalent plate thiclmess approach. Coupling effects between local and global failure
modes of the structure are also considered. The striking ship is modelled as a rigid body, but
the effect of the shape of the striking ship is considered by means of an technique based on
the use of gap/contact elements. The method is basically formulated as a steadystate loading
condition. Dynamic effects are considered simply by including the influence of strainrate
sensitivity in the material model. The stiffness matrices of all structural Wlits are formulated
in closed forms to achieve further reduction of computing times. The procedw-e has been
verified by a comparison of numerical solutions with experimental results from physical
model tests.
The method also provides an important aid in design enabling sensitivity analyses to be
performed on the collision resistance to changes in dimensions of certain scantlings such as
double-side width, the outer side shell thiclmess, the innerside shell thiclmess and the side
stringer/transverse web thiclmess. Sensitivity results from the work undertaken by Pedersen
are presented in Section 6.0.
30
Table 4.1
Literature reviewed: those detailing methods of collision analysis
Ref. Author Title Summary
No.
15 V.U Minorsky An analysis of ship collision Outline ofMinorsky's statistical
with reference to protection of method of determining energy
Nuclear Powered Plants absorption
19 A.N. Kinkead A Method for Analysing Cargo Outline of the substantiation
Protection Afforded by Ship method and compares results with
Structures in Collision and its Minorsky for large deformation
Application to an LNG carrier impacts
18 J. McDermott et al. Tanker Structural Analysis for Outline ofRosenblatt method:
Minor Collisions energy absorption in ben<ling ,
membrane and buckling
32 K. Reckling Mechanics of Minor ship Explanation of stiffened plate
collisions failure
29 A. Bockenhauer Assessment of CQllision An explanation of the
Resistance of Ships for Germanischer Lloyd method
Classification purposes
33 P. Sen eta! Integrated Collision Analysis Establishes an all-embracing
for Ships method for external and internal
collision mechanics
Table4.2
Literature reviewed: Numerical analysis
Ref. Author Title Summary
No.
43 J. Paik et a1 Residual strength Assessment A numerical analysis on the
of Ships after Collision and residual strength of bull girders
Groun<ling and web frames
44 P. Sen Collision Resistance Investigation into designs that
provide increased resistance
45 G. Woisin Analysis of the Collision Numerical analysis for protrusion
bernreen Rigid bulb and Side of a bulbous bow
shell Panel
46 H. Zhu etal Rigid Plastic Dynamic Analysis Virtual work methods to establish
of Ship Collisions impact deformations
31
Table4.3
Literature reviewed: Experiments and numerical simulations
Ref. Author Title Summary
No.
22 N. Ando, K. Arita A study on the strength of A simple model representing a
Double hull Structures in double hull is subjected to impact
C6llision loads.
23 Hisashi Ito et al A Simplified method to A test model is used to develop a
Analyse strength of Double simple numerical method to
hulled tankers in Collision determine energy absorption
26 Hisashi Ito et at A Comparative study of The different methods are used to
Simplified Methods to Analyse analyse a test model and results are
strength of Double hulled compared.
tankers in Collision
24,25& Hisasbi Ito et al Collision Resistance of an NKK Experiments and sensitivity studies
27 Double-Hull tanker on a model of a double hull tanker
30,31 P. Pedersen On Design of Double Hull Experiments and numerical
Tankers Against Collisions analysis to obtain optimum sizing
for ship scantlings.
34 K. Hagiwara A Proposed Method of Simple impact tests on models to
Predicting Ship Collision fmd patterns in the data.
Damage
35 0. Kitamura Comparative Study on Experimental and finite element
Collision Resistance of Side analysis.
Structure
36 0. Kitamura et al A Study on the improved Model test used to develop fmite
Tanker structure against element models
collision and grounding
damage
37 0. Kitamura A Study on the Numerical simulations
Crashworthiness of Double
Side Structure ofVLCC
38 K. Arita A Study on the Strength of Experiments on shell elements, and
Shells in Relation to Ship doubled side shell models
Collision
39 M. Mizukami et al Collision Simulation of a Finite element simulations.
Double Hulled Structure with
Unidirectional Girder system
40 J. Kim et at Behaviour of Double Hull Simulations to compare the impact
VLCCs in Collision response of various vessels
41 A. Vredevelt Full Scale Ship Collision Tests Experiments on inland waterway
vessels. Numerical simulations to
aide analysis
42 J. Daidola Tanker Structure Behaviour Finite element analysis using the
During Collision and Rosenblatt method.
Grounding
32
where Br =the absorbed energy (MJ)
RT =the resistance factor (m
3
)
Area of Low energy points
Energy absorbed, Er
Figure 4.1
Minorsky's energy relationship
33
w.ob tr=es 11r0
Wti:l tr- or swab hulkheaclll c:onroect. the
two bulb but are not signi.fi=ntly
to bath hu.ll..s, ;;,:. the
by {lu.y be sli;ISIIred . .,....y !raal)
deck, uoi e<:: the allip
the 4aclc a:'<1 the s
""-
bottom.
l
r='":
reenter !feb ....
Mdbran_T.,....ic:n !'hue l'ha.se
Iif-r
4 Spael!!a @ t.a '" I.i
f:= . .r -"''::j

""""" ..

-
1
outer

ll h . ;ll -fl
- Mll.lpe
'"''

with web' &:
_fi Wcl;> n:-...l

{
_'j
for sinql.oo hull.
::.:.P.n-2a1Ld- Bulkhead
--....' .
iShip,
inner bull. each hull aep.ar<rteoly
J
ut td th. both, bull.s de1'0Diill c"'-P. y "<-tnner
...Usoa. Spall. of e,.c=. '"11:1:! Rul.l
un i.s '1:!14 b<:ltw&om
!'!J.; T2
r1.r-
hull:headll: (rat:ber
u two web--t'raae
T2 . Y"l'1
DOes the eetr.tor web
Inner lllllfJ
yield or budl:le beo!ore
After raptur.. 'o ot' either hlll.l1
Outer bu.ll, ana-
lyze inner hull ao Yes. the aJ!.a.lyod
with web frames. to aecount: Pu at
oater

.. n.

' ,,
.,
l!'l'Y't':<. 0 S Pwf/T2
At't:er ,.-,:q;otarc ot' 0#1< huJ.l ,
aoll.lyze the othflr bull pan-
ao
el
ning between bull-
bl.d or bud<lc he:torc ru.pture
bcacf:s, i<]llori:ag web fr'llllle$ ;md
e.ither bu.l
ruptured hl>l.l.
Rlo'ldse the analy11ia to I
<:CO'<lnt fo>: P wf at ""'"' t:er web
-
Figure 4.2
Collapse mechanism for the Rosenblatt/McDermott method
34
Type Displacement, DB (Tons) Bow height above W.L. (ft.)
Dry Cargo 4860 17.2
Dry Cargo 9130 26.8
T2 Tanker 21900 27.2
Tanker 54100 33.0
Tmll<" 87800 37.4
Tanker (Jumboised) 121400 37.4
l M.C CAARIER -M:ETHANE PRJNCJ;'IS
MIOSH!P 'SECTION ( 0,_ ~ H. 071> _,)
i
I
--'
i
'
}--'f< ,,. ,.. ..... (
PART PlAN
0 o
SCALE- ffET
Figure 4.3
Kinkead's model of collision with a 12,200 DWT LNG Carrier
35
a
w.
l a 1on curve Load pene t.
Deformation of plate
(a) 10 tons
after breaking of outer plate
"'
,,,
Deformation of plate
(i.JJ
(b)
Penetration of bow after breaking plate
Figure 4.4
Ando & Arita's test model
36
5. COMPARISON OF METHODS
5.1 General
The various rilethods summarised previously show what seem to be an evolutionary process
in understanding the mechanics behind ship collisions. Minorsky's method was statistical;
this was then developed further into analytical methods for low energy impacts such as the
Rosenblatt and the Kinkead methods. Experiments have been conducted to test and/or justify
many assumptions of behaviour within the analytical methods. These experiments then
formed the basis of refined analytical methods, which were transferred to the computer to
conduct non-linear finite element analysis.
However, all these analyses are based on model idealisations of reality. It is the situation and
type of collision that determines which idealised model and which theory is appropriate.
Minorsky's method is satisfactory for high-energy impacts with large deformation. The
experimental work done by the Japanese largely confirm mechanisms assumed in the
Rosenblatt method (where the side shell provides a lot of resistance). Thus despite the
varying collision strengths that the methods may produce they are all relevant depending on
the collision situation, the scantling of the struck ship and the level of accuracy needed by the
designer.
5.2 Published experimental comparisons
Most of the teclmical papers in the public domain, shown in the references, give brief
summaries on the various methods outlined previously. However published work that
compares results from experimental tests with those obtained from the methods given in
Section 4.0 are either limited or details, such as those from tests undertaken by German,
Japanese and Dutch, remain confidential (private communication with Germanischer Lloyd,
January 2000). However, the work previously highlighted in Section 4.2 by Hisashi Ito et al.,
1984 (Z3) provides information to enable a detailed assessment to be made of the various
methods compared with the experimental results. As described two tests were carried out on
the model, a stem impact and a bulb impact, simulating collision with a 100,000 DWT
double-hulled ship. The energy predicted from each of the following methods, An do & Arita
<
22
), Kinkead
09
), Rosenblatt (IS), Minorsky
05
) and NKK(2
3
) were compared to that obtained
from the experimental model tests. Details relating to the experimental load-deformation
characteristics, failure mechanisms and comparisons between the experimental energy
absorbed with those predicted from the various methods are presented below.
5.2.1 The Stem test
Figure 5.1 shows the load-deformation characteristics obtained from the test, the sequence of
~ a i l u r e (shown by the sequences of numbers), tile state of damage after the test (depicted by
the photograph) and the damage characteristics of the ship model. From Figure 5.1 the
sequence of failure and failure mechanisms observed are as follows:
The stem touches the web (a) at stage 2 and web (b) at stage 3
The webs deform, buckle and further deform as a direct action of the stem.
The deformed area of the shell consisting of one panel surrounded by webs and stringers
at the start grows to three panels by the end of stage 3.
The deformed area surrounding the panels at stage 3 was very small.
37
During stages 4 to 5 cusps appeared on the side shell at each location of the webs (a) &
(b).
Just before stage 6 a large noise was heard and the load instantly decreased.
At this stage the side shell (c) ruptured along the upper deck of stem over an area of more
than one web frame.
From stage 8 to 14, the stem continued to penetrate whilst the structure outside of the
penetration area hardly deformed.
At stage 15 the crack on the side shell changed direction towards (d).
At stage 16 the lower part of the stem touched the side stringer (e) which was crushed as
a direct action of the stem.
At this stage the stem had reached the inner hull and the load was increased substantially.
The inner hull had hardly deformed until stage 15, after stage 15 the deformation clearly
showed a wedge shape reflecting the shape of the stem tip.
5.2.2 The Bulb test
Figure 5.1 shows the load-deformation characteristics obtained from the test, the sequence of
failure (shown by the sequences of numbers), the state' of damage after the test (depicted by
the photograph) and the damage characteristics of the ship model. From Figure 5.1 the
sequence of failure and failure mechanisms observed are as follows:
Only th6 side shell of the area of one web frame spacing by one stringer spacing being
impinged by the bulb directly was dented until reaching stage 4
The webs (a) and (b) buckled at stages 4 and stage 5 respectively.
The deformed area of the side shell spread to the area of tlrree web spacing by one
stringer spacing.
The side stringer (c) just above the loading point was depressed.
At stage 7 the wave pattern deformation shown occurred in the outer shell.
At stage 10 the four corners of the panel carrying the load begatL to collapse.
At stage 12 a loud noise was heard and the load was instantly decreased.
At this stage it was found that the side shell had ruptured with a crack at the loading
point running in a vertical direction.
At stage 14 the indenter touched the side stringer (d).
During stages 15 and 16 the load and deformation did not change, while the penetration
of the indenter actually increased.
The load increased again at stage 16.
The inner hull had hardly deformed until stage 15.
During stage 10 to stage 15 the inner hull deformation increased.
At stage 17 the inner shell deformation clearly showed the bulb shape assuming the
shape of the indenter. At the end of the test the inner hull had not ruptured.
5.2.3 Results from stem and bulb tests
Comparisons between calculated and experimental energy-deformation results in the case of
the stem and bulbous collision are shown in Figure 5.2 and Figure 5.3 respectively. These
figures clearly demonstrate the considerable differences in results obtained from the various
methods chosen. Comparisons between the methods with those obtained from the
experiments are summarised as follows:
38
Minorsky
The estimates in both cases are significantly higher than the experiments, particularly in the
early stages of the collision. In the early stage of the collision the effect of the side shell in
this method is given as a constant value which is therefore dominant in the early stages. This
leads to an excessive overestimation, especially in the case of the stem collision when the
side shell tends to rupture in the early stage. Overall, the tests seem to confirm that the
Minorsky method gives inaccurate results when impact produces small deformations.
Modified Minorsky (Woisin)
The modified Minorsky method by Woisin takes into account the membrane energy
according to the length of side shell rUpture. Two curves are given in Figures 5.2 and the one
in Figure 5.3 and results have been based on calculations with different lengths of rupture as
the definition of the length to be used is not clear in this method. In the case of the stem
collision as shown in Figure 5.2, when the height of rupture is chosen in the vertical
direction, the penetratio11 (i.e. modified Minorsky curve 1 is completely different from that
observed in the experiment. However, when the length chosen is the horizontal direction (i.e.
which was the direction that happened in the experiment), the result (i.e. modified Minorsky
curve 2) gives a reasonable correlation with the experiment. However, in the case of the
bulbous penetration test as shown in Figure 5.3, both results do not correlate with the
experiments. The Woisirr method gives low results when compared with experiment, which
reflects the fact that with small deformations, this method cannot accurately predict the
energy absorption.
Kinkead
For the stem condition as shown in Figure 5.2, at the final stage when the stem reaches the
inner hull the Kinkead method gives a reasonably good estimation, albeit that it over-predicts
the final experimental result. However, the method significantly over-predicts the energy
during the early stages, which :may be as a result of the fact that this method treats the
collapse as a collection of discrete events, which may not be independent. Thus the method
counts the energy which is absorbed by the development of the rupture as well as the energy
which the shell absorbed until its rupture. The Kinkead method results show high-energy
values for small penetrations, but then lower increases as the penetrations increase. This may
well be as a result of the fact that the Kinkead method does not take into account the residual
strength of the transverse bulkheads (web frames) after buckling. Although the method
appears to be applicable for stem analyses only, results using the same method have been
calculated for the bulbous condition. Similar observations are noted albeit in this case the
Kinkead method under-predicts the fmal energy when compared to the experimental result as
shown in Figure 5.3.
Ando &Arita
For the stem condition shown in Figure 5.2 the Ando & Arita method does not correlate with
the experimental energy-deformation response and significantly overpredicts the final
energy. The reason for this can be attributed to the simplification of this method previously
highlighted in Section 42. In the experiment the damage as described above developed on
the stiffener level whilst in the Ando & Arita method damage is taken into account only on
the girder or web level. Therefore as in this case where the damage develops on a more
detailed level than on the girder or web level, this method may gives an excessive over
estimation. However, in the case of the bulb collision as shown in Figure 5.3 this method
gives a gopd correlation with the experimental results because the damage developed in this
39
case on the girder or web level. Generally when this method is applied to an actual ship, a
considerable simplification of the structure and the collision conditions are required.
Rosenblatt
A number of uncertainties and assumptions in applying this method were noted when
predicting the energy-defonnation response (i.e. the strength of a double-hulled structure was
calculated in the same way as for a single longitudinal system tanker for which the method
was based on). It would appear from Figures 5.2 and 5.3 that for both the stem and bulbous
condition that the calculated results correlate reasonably well with the experimental results.
However, notwithstanding the assumptions that were made it should be recognised that
different estimates may well be predicted using this method in cases where the hull ruptures
in the horizontal direction which is not considered in the Rosenblatt method (i.e. longitudinal
direction only). Furthermore it should be noted that although the damage state of the side
shell is 3 dimensional, a-two dimensional state approach like the Rosenblatt method may also
predict a different damage state from the real condition.
NKK
The NKK method appears to give a good estimation of the final and early stage energy-
deformation response in both cases. This may not be so surprising as the method has been
calibrated against the test results. However, the applicability of this method to other collision
have been also been examined by conducting a series of model tests which correspond to a
wide range of double-hulled structures commonly used in ships as mentioned previously in
Section4.2
Summary
From the above it is worth comparing the ranges of fmal energy predicted for both the stem
and bulb cases for each of the methods with those obtained experimentally. The results are
tabulated in Table 5.1 for which the deformation is that of contact with the inner shell. It can
be seen for the stem condition that the range of predicted energies varied from 18 MJ.-46 MJ
compared to 18 MJ that was obtained experimentally. For the bulb condition the range of
predicted energies varied from 13 MJ-46 MJ compared to 50 MJ that was obtained from the
test model. It is worth noting that the final energy absorption was significantly higher in the
case of the bulb test (i.e. 50 MJ versus 17 MJ).
Table 5.1
Results of NKK comparison test
Method of Analysis Stem Collision
MJ
Minorsky 36
Kinkead 23.3
Rosenblatt 22.8
Ando/Arita 45.6
Ito (NKK) 17.8
Woisin(l) 6.7
Woisin (2) 13.3
Experimental 17.8
40
Bulb Collision
MJ
32.7
35.5
43.6
46.4
46.4
12.7
12.7
50
5.31nternational Experimental Studies
As previously described in Section 5.2 a number of experimental studies involving side-shell
collisions remain confidential. One such sttldy, which has had some information provided by
private communication between MSL and Gennanischer Lloyd (Germany) relates to the
work undertaken by the latter, Japan and the Netherlands. This follows on from the ultimate
strength approach developed for determining the collision resistance of ships mentioned
previously in Section 2.6. At present this concept is being enhanced within the scope of a
research project sponsored by the Federal Ministry of Education and Research (BMBF). The
scope of work is to undertake more finite element formulations without causing the
computational effort to rise excessively. In addition, work on developing more precise
fracture criteria are being formulated for the outer and inner shells. To support this analytical
work and to Verity the calculation method, international side-shell collision tests involving
Japanese, German and Dutch participation have recently been undertaken. A total of four
collision tests (two Japanese, one German and one Dutch) with different test sections were
performed during the period 1997 and 1998 under the leadership of Dutch Institute TNO in
Delft on a branch of the river Rhine. Because of the test set up, all test sections were of
similar size approximately 9.0m by 4.5m including the force application structure. Except for
the German model, they were produced in the original size (i.e. to the scale 1:1 ), as the aim
was to study the local strength behaviour of the double side-skin structures of tankers. Only
in the German test was a reduced scale model used, with the purpose of investigating the
structural strength behaviour of the entire double skin between bulkheads, deck and bottom.
The collision tests were performed with two inland waterway tankers. The struck ship had a
length of 80m, width of 9.5m and a. draught of 2.2m. In the midship section, the starboard
side was removed up to half the ship width and replaced by the relevant test section. The
ramming ship had a length of 80m, a width of 8.2m and a draught of 2.2m. It was equipped
with a special bulbous bow arranged above the water surface. The test section for the German
test was derived from the side structure of a 30,000 DWT production tanker. A section 13.5m
long and 13.5m wide with a 2.1m depth of double side-skin was produced to a scale of 1:3,
because of the limited capacity of the test facility. This section was built into the test ship, so
that the forces could be passed. on into the hull in a realistic manner. The collision speed for
the test was estimated via the deformation energy and the external collision mechanics, and
selected so that that the inner skin of the struck ship would be ruptured. The calculations
were performed with the aid of fmite element analyses. Before the test took place, an
orthogonal grid with a spacing of 200 mm was marked on the outer and inner skin, to enable
the reconstruction of the strain history. By measuring the grid intervals in the horizontal,
vertical and diagonal directions, it was possible to determine the plastic strain history. These
measurements served mainly to derive realistic calculation approaches for the breaking
strain. The test was performed with a speed of 5 knots, the bulbous bulb rammed the test
section, first tore open the outer shell, deformed the internal components and then ruptured to
the inner skin. With a penetration depth of about lm, the striking ship came to a standstill.
The information provided by the tests is currently being evaluated and incorporated into the
enhanced semi-empiricaVanalytical calculation method developed by Gennanischer Lloyd.
41


0
.3
o
@
50

"'
i
@
'0
20
>O
100 200 300
Penetration { mm}
cv.n;:e. o the stem
l=t .
The .state of damage to the ship
side model after the .
{Stem t!!;;t)
i ( l i . l I i ! i
...

! .-Ld(.;;..!.. )IW! J l
> ..... I I 1CniGk I

J ; [ JSftingF :'4.11: i 1
I I I I >@o I I o I
...
@-SEC.
Damage characteristics of the ship
side model (stem test)
Figure 5.1
60
so
-;;
""
0
'0
"
0
20
.3
10
" 0
0 100 200 300
PeneTration {mm l
Load-penetration cune of bulb
'"'
The smte oi damage to the ship
side model aftN the experiment
(Bulb test)
@-@sec
Damage cllaractt:ristics of the ship
side model (Buib test)
Ito's {NKK experiments) stem and bulb test
42
20
1
(M)
Figure 5.2
NKK comparative tests: stem test

1
7 01-
-
Buib . l-
"
"'
, f;re=kinG ;:cmtl
of :tre siC;: siuaif
6
J t/j
l
r .
. v !
4 0 - tylinotS<y !
,! I J
7 .
3 or ... ------!
1
:::;
'
! I A.
/present / j
2 ol Kinkead / jsr,o:! \ / .. .; t:
I \/ / Roser:f:iatt
'r / ... . ::...-
,- " . .
10" _/ ... ... .:..-.-- '
l --/. . ,,
j " - /. MoCiiied Mfnctsky

0.0 0.5 10 15 zo
PEl'iETR.AT !ON (:l.l)
Figure 5.3
NKK comparative tests: bulb test
43
2 E
6. PARAMETRIC STUDIES
6.1 General
There has been a large volume of experimental work carried out recently on the steps that can
be taken to strengthen ships against collisions. Japanese and Danish researchers, to name just
two, have provided a wealth of information on the subject. The experiments have been
concerned with the validation of methods to account for the effects of varying certain design
parameters such as side shell thickness, the arrangement of stiffeners, the depth of the double
hull and the thickness of the transverse bulkheads (web frames). These have been useful in
that they provide an insight for designers when trying to establish the collision resistance
especially when concerned with the conversion of a tanker to a FPSO.
6.2 NKK assessments
Quantitative assessments were conducted(2
7
) using the NKK method described 1n Section 4 to
assess the collision strength of a 290000 DWT double-hulled tanker designed by NKK. The
energy absorption capacity and the relationship between the critical speed and displacement
of a colliding ship were obtained. The midship section had a 3740 mm double hull depth as
shown in Figure 6.1. A ship with a displacement of 11000 tons was chosen for the colliding
ship. The relative vertical position of both ships and details of the stem assumed for the
colliding ship are shown in Figure 6.1
Furthermore, the effect of the double hull depth related to the 2 metre minimum specified in
the IMO regulations was also evaluated. The NKK studies <
27
) included a comparison between
single and double hull ships and also the effect of varying the depth of the double hull where
three depths were chosen: 2000 mm. 3000 mm and 3740 mm.
In the analysis, critical damage for the double-hulled ship was defined when the stern touches
the inner hull. The critical condition for a single hulled tanker was defined when the stem
ruptures the side shell. '
Table 6.1 compares the total energy absorption capacity obtained for the single and double-
hull structures. It can be seen from Table 6.1 that changing from a single hull to a double hull
(DH)is extremely effective in raising the energy absorption capacity, even with a double hull
depth of just 2m as required by MARPOL. It can also be observed from Table 6.1 that fue
strength during collision varies significantly depending on the collision location (i.e.
collision between webs or on a web). From Table 6.1 and Figure 6.2 and Figure 6.3 it can be
seen that the collision strength is significantly higher for a collision occurring on a web. The
authors attribute this to the large spacing between the transverse webs (i.e. 6000 mm)
compared to the stern size. However, the authors emphasise that this tendency depends on the
geometric relationship between the stem and the transverse web space. For example, the
authors indicate that for a LNG ship with smaller transverse web spacing of 3000 nun the
results are reversed.
44
Table 6.1
Comparison of collision resistance between single and double-hulled tankers (290,000
DWTtype)
Energy Absorption capacity before an oil leak MJ
Double Hull Depth (mm)
Location Single hull
2000 3000
On a web 19.7 49.9 66.6
Between webs 8.34 16.6 29.2
Mean 14.0 33.2 47.9
3740
86.6
48.3
67.4
The critical condition for leaking of cargo oil from a double-hulled tanker during a collision
was also evaluated and a relationship bet\veen the critical speed and displacement of a
colliding ship based on using the Minorsky relationship (equation (11) in Section 4) are
shown in Figure 6.4.
As described previously in Section 4.2 a series of further static destruction tests (
24
"
25
)
covering a wide range of scantling and strength distributions were investigated, which
included the double sides and double bottoms of ships. The effect of changing the design
parameters (i.e. the side shell thickness, the transverse web thickness and the double hull
depth) on the collision strength were examined using the NKK analysis method, where a
double hulled ship of 100,000 tons displacement was used as a standard ship type. The
results from these parametric studies are summarised below:
Increasing side shell thickness:
The resistance offered by the side shell becomes greater in proportion to the side shell
thickness; the plastic membrane force of the side shell is proportional to its thickness.
However, this energy is, at most, 50% of the total.
Increasing the thickness of the side shell plates increases the energy absorption capacity
at the rate of2%/mm and 5%/100 tons (increase in weight).
Increasing the thickness of the transverse webs:
The resistance force increases at an early stage because the ultimate strength of the
transverse web increases.
The side shell rupture is strongly encouraged because the strength of the side shell
decreases in comparison with the ultimate strength of the transverse web.
Increasing the. thickness of the transverse web plates increases the energy absorption
capacity at the rate of7%/mm and 25% /100 tons (increase in weight).
Increasing the double huU depth:
The energy absorption capacity increases steadily because the crushable length of
transverse webs or stringers becomes longer.
Moreover, the number of transverse webs or stringers, which are crushed by the bow
also, increases since the damage area develops.
For the case considered the total reaction force by the transverse webs or stringers
increaSes nearly in proportion to the penetration depth and the absorbed energy increase
in proportion to the square of the penetration depth.
For the ship examined, the reaction forces by the transverse webs are small and the
effects of the double hull breadth are also small. The increase in energy is 18% when the
breadth is cbanged from 2,230 mm (standard) to 2500 mm (adding 270 mm).
45
Increasing breadth of the double hull increase the energy absorption capacity at the rate
of?%/100 mm and 10%/100 tons(increase in weight)
From the investigations above it was found that the most effective method to increase the
collision strength was to increase the transverse web thickness. However, the authors indicate
that a further effective method not examined was to decrease the transverse web spacing. If
the transverse web spacing is decreased, the number of the transverse webs, which are
crushed by the stem, is increased and therefore the absorbed energy is also increased. On the
other hand the authors point out that decreasing the side stringer spacing does not seem to be
effective because the stem of a ship may not crush stringers directly when it collides at the
point just below the side stringer.
6.3 Pedersen assessments
Pedersen (Jo) using the finite element procedure described previously in Section 4.2.5
undertook a series of collision analyses between the bow of a container ship of 40,000 DWT
and the side structure of a double VLCC Tanker of290,000 DWT. It was assumed that the
container ship is pushed into the midship structure of the double hull VLCC at the central,
right-angled position. The maximum speed of the striking ship was 12.9m/s, while the struck
ship is at a standstill position at the beginning of the collision. The container ship had a
bulbous bow which is in contact with the side structure of the struck ship at the beginning of
the collision, but at later stages with larger penetrations the stem (i.e. flare section) comes
into contact with the deck. The depth of the double side is 4m and the depth of the double
bottom is 3m. These distances were chosen to satisfy the IMO requirements for oil pollution
prevention. The frame spacing 'Yas 6m in the side and 3m in the deck bottom with the
collision contact points of the bow being on the side stringers and transverse webs.
Pedersen undertook a series of 15 sensitivity analyses varying a number of design parameters
(i.e. double-side depth, inner and outer side shell thicknesses, and transverse web and stringer
thicknesses). Table 6.2 gives details of the 15 cases examined. It should be noted that case
number 1 refers to the reference condition.
Results from the various analyses are shown in Figures 6.5 to 6.7. Figure 6.5 shows the
collision energy-deformation curve obtained for the reference case. Also shown are the
contributions to the total energy from the flare and bulbous bow. Initially contact is with the
bulbous bow; as penetration increases the side stringers and transverse webs struck by the
bulb of the striking ship are crushed. At approximately 900 mm penetration, the flare comes
into contact with the deck structure and the stiffness increases suddenly. Crushing of the deck
and yielding of the plates continues until the outer shell ruptures and the stiffness drops
suddenly. With further increase in penetration, the irmer shell goes into membrane tension
and as a result the stiffness increases. Finally the inner shell plates rupture when the tensile
strain exceeds the critical rupture strain. The critical strain was taken as 5% which was
considered by Pedersen to be applicable for the steel used in the study (i.e. yield and ultimate
tensile strength of 315 MPa and 441 MPa respectively). Pedersen indicates that the critical
rupture strain is smaller than those measured in material tests (i.e. values ranging between
10%-30%) because of such effects such as out-of plane bending, stress concentrations, initial
cracks, c o ~ o s i o n , impact loading etc.
46
Table 6.2
Details of Pedersen (
301
sensitivity analyses
Double-side Change in Change in Change in Change in Structural
width outer shell inner shell side transverse weight of
Case No thickness thickness stringer web model
thickness thickness
(m) (mm) (mm) (mm) (mm) (ton)
1 4 0 0 0 0 1034.4
2 4 0 0 0 0 1034.4
3 4 0 0 0 0 1034.4
4 2 0 0 0 0 908.4
5 3 0 0 0 0 971.4
6 5 -5 0 0 0 1097.4
7 4 5 0 0 0 974.1
8 4 10 0 0 0 1094.7
9 4 0 0 0 0 1155.0
10 4 0 -5 0 0 974.1
11 4 0 5 0 0 1094.7
12 4 0 10 0 0 1155.0
13 4 0 0 -5 -5 942.7
14 4 0 0 5 5 1126.1
15 4 0 0 10 10 1217.8
Figure 6.6 shows the influence of increasing the double s'ide width on the critical energy
absorption capacity. Three criteria' are considered, namely rupture of the outer hull,
penetration till the original position of inner hull, and rupture of the inner hull. It can be seen
from Figure 6.6 that for the outer hull rupture the effect of increasing the double-side width is
small. However, other criteria are very sensitive to the double-side width and significant
increases in the energy absorption capacity can be observed (i.e. the energy absorption
capabilities are increased by more than 100% when increasing the double-side width from
2m to4m).
Figure 6. 7 shows the results from the sensitivity analyses conducted for all of the design
parameters varied. It can be seen that the most sensitive design parameter with respect to
structural weight is the web thickneSs against rupture of the outer hull, and the double-side
width against rupture of the inner hull.
Pedersen (Jl) extended the work to compare the critical energy and critical velocities of the
above for different arrangements of the side structure of the struck double hull VLCC. Two
types were considered depending on the location of the transverse struts (i.e. located at side
or centre) as shown in Figure 6.8. The resUlts were also compared with those obtained from
the design procedure adopted by Germanischer Lloyd. Figure 6.9 compares the critical
collision energy obtained for the different arrangements as a function of the double-sided
width. It can be seen that the both the Pedersen and Germanischer Lloyd results are in good
agreement for the side structure with the struts at the centre. Figure 6.10 also compares the
critical velocities using the Gennanischer Lloyd relationship given in Section 2.6, equation
(10). Agafu for the side structure with the strut at the centre both results agree well. It was
concluded that if the critical collision speed is specified then the optimum width of double
side can be determined from the design curves such as shown in Figure 6.10.
47
6.4 MSL Assessments
6.4.1 Parametric study to assess the energy absorption capabilities of a
typical tanker from information in the public domain
A study was undertaken by MSL to compare the various methods to assess the absorption
capacity of a typical tanker. The striking ships and struck tanker chosen for the study were
identical to those given in Kinkead's paper (!
9
) and described in Section 4.2.3. The various
methods of ship collision assessment described in Section 4 were used in the study:
i. Kinkead's method
ii. Minorsky's method
iii. Woisin's modified Minorsky method
iv. Haywood's modified Minorsky method
v. The Rosenblatt/McDermott method.
The results obtained with Kinkead's method were similar to those reported by him(l
9
).
Comparisons of the energy absorbed by the struck ship for each of the different methods
applied are given in Table 6.3 and sho'WD. graphically in Figure 6.11.
As this study followed that by Kinkead (l
9
l, the draught levels of the striking ships varied as
shown previously in Figure 4.3. Therefore the impact zone was not the same for each of the
striking vessels and hence the energy absorbed by the struck ship varied depending on the
location of impact.
The energy absorbed also varied significantly depending on the method used and amount of
disrupted volume. It can be seen from Table 6.3 and Figure 6.11 that the range of energies
obtained varied from 36.5 :MJ - 217.4 MJ. The highest energy predictions were generally in
the order of Haywood, Kinkead, Woisin, Rosenblatt and Minorsky respectively. The
Haywood method gave significantly higher predictions than the other methods particularly
for low disrupted volumes.
It can also be observed from Figure 6.11 that the energy predicted using either the Kinkead,
Woisin and Minorsky method varied by more than a factor of2 respectively for the range of
disrupted volume. For predictions based on either the Kinkead or Woisin method the
increases in energy mainly occurred for impacts involving striking ships with high bow
heights above the water line. For the Kinkead method the increases in energy comes from the
buckling of the main deck/hull sub structure and wedge splitting of the main deck/sub decks.
The additional energy predicted using the Woisin method is a result of the increase in contact
height of shell rupturing. Furthermore it can be seen from Table 6.3 and Figure 6.11 that
predictions using the Kinkead method were generally an order of two magnitudes higher than
those obtained uSing the Woisin method, particularly for cases involving high bow height
impacts. This is a result of the significant contribution that the buckling/wedge splitting has
in the Kinkead method compared to the other impact locations in which shell rupturing
tended to play a more significant role in the total energy absorbed.
It can be Seen from Table 6.3 and Figure 6.11 that the Rosenblatt and Kinkead methods
produce results of similar magnitude for two striking ships: Dry Cargo 2 and T2 Tanker. The
values for striking ship Dry Cargo 1 show some difference between the two methods. 'This is
because of the Kinkead recognition of three-dimensional behaviour of the side shell at impact
and also vertical stringers behind the hull shell. However, as the bow height of the striking
48
ships above waterline increases the values diverge. This can be ascribed mainly to two
mechanisms that are not accounted for in the Rosenblatt methods. These are:
i. Wedge splitting of decks
n. The vertical distribution of load on the side shell.
As the bow height increases the probability of the bow tearing into the deck increases.
Rosenblatt recognises buckling and membmne action in a deck, but the action of the bow of
the attacking ship tearing through the ship is not accounted for. As the bow height above
waterline increases it is more probable that the impact load will spread, initially, vertically
towards the deck when the side shell is hit, as well as longitudinally along the length of the
ship. Furthermore the Rosenblatt method considers buckling of web frames only, while
Kinkead considers buckling of the deck as a whole strut unit.
The Minorsky methods produce the lowest values for all the striking ships. By looking at the
damaged volume, it can be seen that low impact/energy collisions do not involve significant
disruption of volume and hence Minorsky's method computes relatively low energy
absorption.
The Haywood method (I?) described previously in Section 4 was adapted from the Minorsky
method to account for different types of stiffening of the side shell for double-hull
arrangements. The predictions using the Haywood method are similar for each impact
location and as stated previously tend to significantly give higher predictions than observed
from other methods. The higher capacities are as a result of the constant term given in the
Haywood method, i.e. a value of 1 8 ~ . 2 MJ, as given in equation (13). However, for the larger
tanker displacements the results are generally in agreement with those obtained from
Kinkead.
As in the case of Kinkead ('
9
), the critical impact velocities of the six striking ships using the
Minorsky relationship (i.e. equation (11 )) have been evaluated. From these values of striking
speed, the relationship between speed and ship displacement is shown in Figure 6.12. As
expected from the variations in energy obtained there is also considerable variation in the
critical speed for different sizes of striking ship. The results given by Kinkead show
reasonable agreement with the Haywood method at the upper ranges of displacements.
However, the results also reveal a critical state for intermediate striking ship sizes. This
minimum safe speed is clearly identified. in the displacement range 15-30,000 tons. The
reason for these differences relates to the vulnerability of the tanker being struck. In the
example tanker used by Kinkead the stiffening was more substantially towards and below the
waterline particularly in its inner hull whose thickness diminishes progressively from the
waterline to the main box girder under the main deck. Therefore, one would expect that the
weakest point for penetration to occur is immediately below this upper box girder. Therefore
the configuration of striking ship giving a critical minimum safe speed is that whose bow on
impact would be closest to the underside of the upper box girder which was for the T2
Tanker. This highlights the importance of considering different impact scenarios when
undertaking collision assessments.
For the Woisin, Minorsky, Rosenblatt methods the critical speeds are similar for all ranges of
striking vessel size. However, the critical speeds are generally much lower than those
obtained from Kinkead, particularly outside the critical intermediate range of striking ship
sizes of 15-30,000 tons described above and significantly lower than predicted using the
Haywood method for all displacement sizes.
49
This variation of energy absorbed as a result of the different location of impact is best
illustrated by referring to Table 6.4. It can be seen from Table 6.4 that as the bow height of
the striking ship above water line varies so does the contnbution of each impacted element to
the total energy. It can be seen that for high bow height impacts the contribution of energy
from buckling plays a more significant role, while shell plate rupturing plays a more
significant role for the lower impact locations. This proves the three dimensional nature of
Kinkead's substantiation method. The position of the striking ship affects the amount of
impact energy absorbed by the web frames in buckling, for instance, or the membrane in
tension. This means that the substantiation method can be a powerful tool in the design of
ship scantlings, since the size of the striking ship and its relative draught has a varying effect
on the main members in the impact zone.
A fiuther separate study was undertaken by MSL to establish the share of energy between
membrane, web frames, decks and impacted sub-structures of the struck ship from the
Kinkead paper(
19
l, with varying side shell thiclmess. The basis for the MSL parametric study
was the struck ship investigated in the Kinkead paper
09
l. Two parameters were investigated.
These were:
i. rupture strain
11. shell thiclmess.
The results are shovm in Table 6.5 and Figure 6.13. The greatest effect on energy absorption
is produced by the change in rupture strain (see Figure '6.13). This is because the shell hull is
no longer acting elastically- there is a non-linear relationship between stress and strain.
The shell thiclmess has an effect on the load carrying capacity of the plate and hence the in-
plane stiffuess. However, the rupture strain has an influence on the lateral deflection of the
plate and hence the geometry. The plate, as a membrane, exhibits an increase in out-of-plane
stiffness as the lateral indentation increases. Since the majority of the load on the plate is
lateral load i.e. out-of-plane; the rupture strain has a greater influence on the energy absorbed
prior to rupture than does thiclmess.
6.4.2 Parametric study to assess the energy absorption capabilities of two
typical FPSO units from information provided by HSE
Two FPSO units identified as type A and type B in this report respectively, covering typical
ranges of scantling parameters (i.e. different ship dimensions, including variations in side
shell wall thiclmesses, web frame spacing, web depth and thiclmesses and hull depths) were
selected from information provided by HSE. Table 6.6 provides information on the variations
in typical scantling parameters normalised with respect to Type B unit
For both type A and B, side shell collision assessments, to evaluate the energy absorption
capabilities using the methods identified previously in Section 4 were undertaken (i.e.
Kinkead, Rosenblatt, Minorsky and Woisin). A vessel similar to the dry cargo vessel with a
displacement of 9130 tons shown in Figure 4.3 was chosen to represent the characteristics of
the striking vessel. A parametric study varying scantling parameters (i.e. shell wall thickness)
was performed by MSL to evaluate the variations in absorption energy capabilities. The
assessments undertaken were as follows:
Impact at mid span between web frames
Impact on web frames
Sensitivity analyses varying the side shell plate thickness and rupture strain of material.
50
Results obtained for a collision between the webs and collision on the web, based on the
actual scantling parameters (referred to as the Base Case; Strain limit 0.1) are given in Table
6. 7 for type A and type B respectively.
It can be seen from Table 6.7 that the collision energy varies depending on the collision
scenario considered, the method used and type of vessel. The ranking of energy absorbed for
the different methods followed a similar pattern to that observed in the assessments
undertaken in Section 6.4.1, with the Kinkead and Rosenblatt methods, Method A and B
respectively predicting the highest energy absorbed and the Minorsky and Woisin methods,
Method C and D respectively predicting the lowest.
It can be observed from Table 6.7 that the range of energy obtained varied between 39 and
135.5 MJ when based on Kinkead (Method A) and Rosenblatt (Method B) for a collision
between webs. It can also be observed from Table 6. 7 that collisions involving impact on the
web resulted in higher energies than when impact occurs between the webs. This illustrates
the importance in considering different collision scenarios when determining the energy
absorption capabilities of the vessel.
Sensitivity analyses as indicated above were performed for each vessel type by varying the
shell wall thickness and material rupture strain. The results from the sensitivity analyses are
presented in Figures 6.14 to 6.19 respectively. Figures 6.14 and Figure 6.15 compare the
energy absorption capabilities for a collision between web and collision on the web
respectively for a rupture strain of 0.05, Figure 6.16 and Figure 6.17 compare the energy
absorption capabilities between web and on web for a rupture strain of 0.1, while Figures
6.18 and Figure 6.19 compare the energy absorption capabilities for collision between web
and on web respectively for a rupture sirain of0.15.
It should be noted that the letter notation adopted for each of the different methods given in
each of the figures was as follows:
Method A - Kinkead
Method B - Rosenblatt
Method C - Minorsky
Method D - Woisin
From Figure 6.14 to Figure 6.19 the following observations can be made as follows:
a) The energy absorbed by both type A and B units is significantly influenced by the
rupture strain.
b) The energy absorbed by both type A and B increases with increasing side shell wall
thickness but is not as pronounced as that obtained by increasing the rupture strain (strain
limits of0.05, 0.1 and 0.15 were used in the assessment).
c) The energy absorbed by both type A and B is higher for collisions involving impact on
the web when compared with a collision between the web.
d) Methods A and B give higher estimates of energy absorption when compared to Methods
CandD.
e) In all_cases the energy absorbed for type B were higher than type A when based on
estimates using either Method A or B respectively. This is reflected in the different
scantling arrangements for each of the vessel types.
The differences in the energy required to produce critical damage can be attributed to the
different collapse mechanisms visualised for each method. It should be noted for the purpose
51
of this study that the assessments carried out on both type A and B have been based on
characteristics of an assumed striking vessel. As demonstrated previously in Section 6.4.1 the
energy absorbed by the struck vessel can be influenced by the characteristics of the striking
vessel.
52
Table6.3
Results of MSL comparison of collision analysis methods
Detail of Striking ships Assessed Calculation Method
volumeRr Height of Height of
Energy absorbed by struck ship
of Outer Inner
Type
Bow height Displacement
disrupted Shell Shell
(metres) (tonnes)
structure
Minorsky's Woisin's Haywood's Kinkead's Rosenblatt's
(m3) (m) (m)
(MJ) (MJ) (MJ) (MJ) (MJ)
Dry Cargo_l 5.24 4860 0.186 11.201 9.677 41.736 54.090 191.192 95.86 68.783
DryCargo_2 8.17 9130 0.106 9.144 4.572 37.996 36.530 187.352 49.82 52.875
T2 Tanker 8.29 21900 0.088 14.478 4.572 37.148 47.625 186.482 59 64.968
Tanker 10.06 54100 0.283 10.763 7.049 46.316 59.548 195.895 137.08 62.336
Tanker 11.04 87800 0.729 10.287. 7.544 67.240 100.754 217.382 204.9 69.945
Tanker 11.04 121400 0.729 10.287 7.544 67.240 100.754 217.382 204.9 69.945
Details of
Outer shell Inner shell
Web details Deck details Longitudinal angle stiffeners
struck ships
thickness thickness
Web Web Deck width Deck Spacing Web Web length Flange
scantlings
spacing thickness thickness thickness length
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)
21.9 14.2 2305.1 12.7 1270 11.2 762 9.5 142.9 101.6
53
Table6.4
Comparison of Absorbed Energies (from Klnkead<
19
>)
Striking Ship
Energy absorbing
4860 Tonnes 9130Tonnes 21900 Totu1es 54100 Tonnes 87800 Totmes
mech\lnisms
Location
*5.24 m *8.17m *8.29m *10.06m *l1.4m
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
asa%of asa%of asa%of asa%of asa%of
Rupture Strain = 0.1
(MJ)
Total
(MJ)
Total
(MJ)
Total
(MJ)
Total
(MJ)
Total
(I) Initial Puncturing Outer shell 4.02 4.19 2.98 5.98 2.98 5.05 2.07 1.51
Inner shell 2.39 2.49 2.51 5.05 2.51 4.26 1:30 0.95
Total (1) 6.41 6.68 5.49 11.03 5.49 9.31 3.37 2.46
(2)Vertical Membrane Outer shell 22.63 23.61 18.70 37.53 28.41 48.15 21.96 16.02 21.05 10.27
Side shell rupture Inner shell 12.76 13.31 5.84 1Li3 6.17 10.46 9.39 6.85 10.08 4.92
Total (2) 35.39 36.92 24.54 49.26 34.58 58.62 31.35 22.87 31.13 15.19
(3) Horizontal rupture Shell/web 23.66 24.68 5.43 10.90 4.13 7.01 4.68 3.41
Total (3) 23.66 24.68 5.43 10.90 4.13 7.01 4.68 3.41
(4) Wedge splitting Deck 12.04 12.56 8.96 17.99 8.96 15.19 25.56 18.64 63.15 30.83
Total (4) 12.04 12.56 8.96 17.99 8.96 15.19 25.56 18.64 63.15 30.83
(5) Plastic bending Hull 1.90 1.98 1.45 2.90 2.69 4.56 1.37 1.00 1.30 0.63
Total (5) 1.90 1.98 1.45 2.90 2.69 4.56 1.37 1.00 1.30 0.63
(6) Buckling 2.08 2.17 1.22 2.46 1.22 2.08 10.21 7.45 28.55 13.94
Impacted sub-structures Frame
panels
(7) Buckling Adjacent 14.38 15.00 2.72 5.47 1.91 3.24 60.55 44.17 80.72 39.41
Deck/hull sub structures Frome
panels
Total 17.18 3.95 7.93 3.14 H2 70.75 51.61 109.28 53.34
(6)&(7)
*Bow height above Total (1)(7) 95.86 100.00 49.82 100.00 59.00 100.00 137.09 100.00 204.85 100.00
waterline
54
Table6.5
MSL Investigation Into energy share variation with side shell thickness (striking ship= 4860 tonnes)
Side shell Thickness
Energy absorbing
Striking ship
t 2t 1.5t l.25t l.lt
mechanisms
location
*5.24 m
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
Energy
(MJ)
asa%of
(MJ)
asa%of
(MJ)
asa%Qf
(MJ)
asa%of
(MJ)
asa%of
Rupture Strain = 0.1 Total Total Total Total Total
(I) Initial puncturing Outer shell 3.90 4.08 7.25 4.51 5.58 4.35 4.74 4.23 4.24 4.14
Inner shell 2.54 2.66 4.53 2.82 3.54 2.76 3.04 2.71 2.74 2.68
Total (I) 6.44 6.74 11.79 7.33 9.11 7.10 7.78 6.95 6.97 6.82
(2) Vertical membrane Outer shell 20.47 21.41 38.07 23.67 29.27 22.82 24.87 22.20 22.23 21.75
Side shell rupture Inner shell 11.52 12.05 20.60 12.81 16.04 12.50 13.78 12.30 12.42 12.15
Total (1) 31.99 33.45 58.67 36.48 45.31 35.32 38.65 34.50 34.65 33.91
(3) Horizontal rupture Shell/web 21.46 22.45 28.39 17.65 24.93 19.43 23.19 20.70 22.15 21.68
Total (3) 21.46 22.45 28.39 17."65 24.93 19.43 23.19 20.70 22.15 21.68
(4) Wedge splitting Deck 11.65 12.19 11.65 7.24 11.65 9.08 11.65 10.40 11.65 11.40
Total(4) 11.65 12.19 11.65 7.24 11.65 9.08 11.65 10.40 11.65 11.40
(5) Plastic bending Hull 1.54 1.62 5.22 3.24 3.11 2.42 2.26 2.02 1.81 1.77
Total(5) 1.54 1.62 5.22 3.24 3.11 2.42 2.26 2.02 1.81 1.77
(6) Buckling impacted Inter-web 2.13 2.23 3.77 2.35 3.00 2.33 2.58 2.30 2.32 2.27
sub-structures frame panels
(7) Buckling Deck/Hull Adjacent 20.37 21.31 41.34 25.70 31.18 24.30 25.91 23.13 22.63 22.15
suO-structures frame panels
Total (6) & 22.51 23.54 45.11 28.05 34.17 26.64 28.49 25.43 24.95 24.42
(7)
*Bow height above Total (1)-(7) 95.59 100.00 160.82 100.00 128.28 100.00 112.02 100.00 102.19 100.00
waterline
55
FPSO
Type
A
B
Table6.6
Comparison of scantling parameters normalised with respect to Type B
Side shell Side shell Web Web Web Depth Impact
thickness thickness Fnune 1hickness Height
Outer
Inn"
Spacing Outer
1.06 1.14 0.79 .96 1.57 1
1 1 1 1 1 1
Table 6.7
Comparison of base case energy absorbed for different collision scenarios
for Type A and B Vessels
Impact
Height
Inn"
0.635
1
Collision Scenario
FPSO MethodAMJ MethodB Methode MethodD
Vessel Type MJ MJ MJ
A 71 39 33 15
Between Webs
B 135 54 37 20
A 91 118 60 69
On Web
B 173 117 64 76
56
A-A Section
Bow shape of tbe strikiag sbip
B! ..
Midsbip sedion of the 290000Dwr type
clooble--buBed taJI.ker used ill tile aaalysis
5hmrillg tbe vertic: loeatio:o of coJ.lisio,q point
Figure 6.1
Typical striking and struck ship details used in the NKK Experiments
57
120
100
2
80
J
60
..
20
0
0 1000
=
g
11
6
2000
Figure 6.2
3000 4000
NKK experiments: Energy penetration curves for each double hull depth (collision on web)
120
100
80
:?
J
..
..
20
~
~
"
~
"

"'
:
"
0
"'
Q
/
8.34
P e n e ~ o n ~
~
~
'
0 1000 2000 4000
Penetration mm
Figure 6.3
NKK experiments: Energy penetration curves for each double hull depth
(collision between webs)
lOOO 10000 100000 500000
Displacement ton
Figure 6.4
NKK experiments: Critical velocity vs. Displacement for Tankers
58
soar i
--
T
Total -:---!
Component bey flare _L __ j
Compon .. nt butbous bow ' i
I I I
"'
_,

i i
I
' 3
penetrotion(m)
Figure 6.5
Pedersen's energy- penetration relationship
0 Rupture ct inn .. r hull
t> Plm"'tratlon until original
innPr hull po:oiti011
...-., o Ruplur., of >ter
!

2

c
!l.l t---.;..--+
7
L_._
7
"-+--+--c
0
u
""
5
t--;ei>._ --<6.-.--;<P

0 1 2 3 4 5
doubie-side width(m)
The critical energy vs the double-side width
Figure 6.6
Pedersen: The critical collision energy versus the double side width

'
BOO
0 side widjh o hull thickness
4
Wl'b thiCknt'SS 0 Oulfor hull thickness
''" ----:-----:
Rupture ;,.,.,.,,
1---: -.-IJli_.,#L,l'---f>---
!"'t--- ' '
o "'"'') orrginat l
:I:S ; illlll'f huU p<ISiliOI'l .
. 5100 __ _ _J._, __ ! . __ j
Figure 6.7
Pedersen: The critical energy versus the model weight
59
(o) strut ol center (b) strut at sides
Arrangement of struts in double hull VLCC
Figure 6.8
Pedersen: Location of struts in double hull VLCC
..-..soo - _.. str
...., .,..., ...... str
o cer1ler
---- GL
><00
o e
e' '
/
j
l---+-+-h"-'-l--1.1--..J--+---.j
/ _/i
c ,,.
.Q ; ./

0 >
(.) 20()\--l--,---k/''i--"---f-+-1f-+--l
/
0 /
._g 100l-cc ... c-! .. """'+--J--f--\---J--f---l
;;:
o
width of double side (m)
Figure 6.9
Pedersen: The critical collision energy versus the double side width
Figure 6.10
Pedersen: The critical speed versus the double side width
60


i
Comparison of Energy absorbed for different methodS
250.0,-----------------------------,
200.0
150.0
-Haywood

W>isin
JK Ro:senblat method
--Mmorsky
1U 100.0
50.0
Disrupted Volume {m
1
)
Figure 6.11
Comparison of the Energy absorption computed by the different methods
Comparison of critical speeds to cause cargo spillage for different methods
12.00

10.00


8.00
\
------.Haywood
--Kinkead

-*-Woisin
--Rosenblatt
t 6.00
'
l

4.00
0
2.00
0.00
0 20000 40000 60000 60000 100000 120000 140000
DISplacement of striking (Ions)
Figure 6.12
MSL Comparison of critical speeds to cause cargo spillage for different methods
61
Comparison of Energy Absorbed with changes in shell wall thickness
250.00
---

200.00
i
r 100.00

.---'

-
.-----
----
; I
..-
stra<n =0. 1
0.05
' '
150.00
50.00
0.00
25% 75% 100% 125%
lncl"flase In shell wall thicknesses
Figure 6.13
Comparison of Energy Absorbed with changes in shell wall thickness
62
350
300
250
! 200
"' E'

~ 150
100
50
0
0
Comparison of Energy Absorbed (All Methods, Rupture Strain =0.05)
with change in shell wall thickness
Collision Between Webs
25 50 100
% Increase in wall thickness
Figure 6.14
Relative collision resistance: 0.05 strain-collision between web frames
63
liB Method A, Vessel A
Method A, Vessel 8
Method 8, Vessel A
tu Method B, Vessel B
D Method C, Vessel A
Method C, Vessel 8
DMethod D, Vessel A
Method D, Vessel B
350
300
250
,. 200
~
,..
E'

s 150
100
50
0
0
Comparison of Energy Absorbed (All Methods, rupture strain =.05)
with change in shell wall thickness
Collision On Web
25 50 100
% Increase In wall thickness
Figure. 6.15
Relative collision resistance: o.os strain-collision on web frame
64

350
300
250
~ 200
f
Ji 150
100
50
0
0
Comparison of Energy Absorbed {All Methods, rupture strain =.1)
with change in shell wall thickness
Collision Between Webs
25 50 100
% Increase in wall thickness
Figure 6.16
Relative collision resistance: 0.1 strain-collision between web frames
65
111 Method A, Vessel A
Method A, Vessel 8
Method B, Vessel A
m Method 8, Vessel 8
o Method C, Vessel A
Method C, Vessel B
DMethod D, Vessel A
Method D, Vessel B
350
300
250
;:;- 200
~
,.,
~
afi 150
100
50
0
0
Comparison of Energy Absorbed (All Methods, rupture strain =.1)
w i ~ h change in shell wall thickness
Collision On Web

25 50
%Increase In wall thickness
Figure 6.17
Relative collision resistance: 0.1 strain-collision on web frame
66
11 Method A, Vessel A
Method A, Vessel B
1---------jl Method B, Vessel A
El Method B, Vessel B
I ] Method C, Vessel B
100'
350
300
250
:; 200
~
,
E'

ED 1so
100
50
0
0
Comparison of Energy Absorbed (All Methods, rupture strain =.15)
with change in shell wall thickness
Collision Between Webs
1m Method A, Vessel A
Method A, Vessel 8
1---------11 Method B,Vessel A
25 50
% Increase In wall thickness
Figure 6.18
~ o ~ Method B, Vessel B
D Method C, Vessel A
Method C, Vessel B
I loMethod D, Vessel A
100
Relative collision resistance: 0.15 strain-collision between web frames
67
350
300
250
~ 200
f
.fi 150
100
50
0
0
Comparison of Energy Absorbed (All Methods, rupture strain =.15)
with change in shell wall thickness
Collision On Web
25 50 100
% Increase in wall thickness
Figure 6.19
Relative collision resistance: 0.15 strain-collision on web frame
68
111 Method A, Vessel A
Method A, Vessel B
Method B, Vessel A
1!!.1 Method B, Vessel 8
Ill! Method C, Vessel A
Method C, Vessel B
'Method D, Vessel A
o Method D, Vessel B
7. CONCLUSIONS
7.1 Review of Classification Society Rules and Standards
The majority of the rules from the classification societies follow a deterministic approach for
accidental impact. This is done, by prescribing a particular energy that the struck ship must resist.
Thus the strength of the ship is said to pass or fail on this criteria. The codes are in agreement with
the energy to resist. This is 11 to 14 MJ equivalent to a 5000 tonne displacement supply vessel
travelling at a speed of 2 rnls.
Germanischer Lloyd is the exception to this. Their method is to classify a particular ship in
accordance with the minimum velocity of attacking ship to produce rupture of the struck ship fuel
tanks. There is therefore more detail in the description of the strength of a ship in the Germanischer
Lloyd system.
Guidance in the analysis of collision is only given in detail in one code- NORSOK. This code gives a
comprehensive treatment of collision. A deterministic energy is prescribed along with force-
deflection relationships for stiffened plates to determine the response of the impacted structure.
However, in this code the sequence of the collapse event is not described. This, though, is not the
requirement of the code. A code, in general is to provide guidance in design; not necessarily analysis.
The present draft of the ISO code for Floating Systems appears to rely heavily on the requirements of
a recognised classification society (RCS).
7.2 Review of Safety cases
A limited review of twelve scenarios drawn from submitted safety cases was LUidertaken. All of the
safety cases recognised the need to meet the requirements of classification society rules for resisting
impact from supply vessels (i.e. the 11MJ-14MJ requirement). Most safety cases however also
recognised the potential for major damage resulting from collision, particularly from drifting passing
vessels. It was apparent from the review that shuttle tanker operations were less critical in terms of
damage expected than that resulting from side-shell collision with passing vessels.
A range of vessel sizes and associated energy to be considered for the impact was identified. An
attempt to categorise the extent of damage with energy absorbed was undertaken. For collisions
involving energies up to 36 MJ the review highlighted that this would result in minor or moderate
damage without spillage. However, energies in the range of 150-280 MJ would result in global
collapse and mooring failure.
There was insufficient information contained within the documents reviewed to ascertain what
methods were used to evaluate the extent of damage and corresponding energies absorbed. However
use of detailed finite element analyses were referred to when undertaking high-energy collision
assessments.
7.3 Methods of analysis
The Minorskymethod
This method and its subsequent independent modifications by Woisin and Haywood provide a simple
method to determine the deformation from the volume of material crushed. However, the path of
progressive collapse cannot be determined by this method. The extent of the collapse is either known
or idealised with this method. Therefore, this is a post-damage assessment method.
69
The inappropriateness of this method to small deformation collisions has been mentioned in literature
and investigated. Although Woisin modified the method to take account of side-shell deformation,
the values obtained are very different from those of the other methods at small deformations. With
larger deformations the results converge with the other methods.
The RosenblattffilcDermott method
This method concentrates on the deformation and rupture of the hull, via elastic, plastic and
membrane modes. Collapse is visualised as work done by the collapse load to produce strain energy.
Thus, the loads resulting from the impacted side shell are the same loads that eventually produce the
deformation and failure of all the other structural elements beyond the side shell. However, this
method considers only longitudinal deformation of the side shell and was originally only applicable
to a single hulled ship. For minor collisions, the elongation of the hull shell vertically up the side of
the ship may be taken as negligible. However, this is highly dependent on the location of the impact.
If impact on the side shell is in the proximity of a stiff deck, then the side shell -will carry a
significant amount of energy.transversely (vertically).
The advantage with this method is that the extent of deformation in the impact zone can be predicted.
Each damaged part of the structure is known to have an effect on another part; this damage process
increases with time. The Rosenblatt!McDermott method is thus a progressive collapse analysis.
Kinkead's substantiation method
The Kinkead method calculates the energy to deform and/or fail the elements that comprise the
struck ship's scantlings. The sum of all the separate energies gives the total energy. This treats
collapse as a sum of sub-collapses. All areas of the striking and struck ship affected by the collision
are divided into their structural components, e.g. plate panels, stringers, frames etc and then the
ultimate loads of these components are calculated by ultimate load or buckling theory. However in
doing this the collapse of elements are seen as independent events. This can lead to inaccuracies
since the ship elements subject to collision do not fail independently but are part of a progressive
sequence of failure.
However, the advantage with this method is that it considers the three-dimensional behaviour of hull
shell subject to the impact energy. Thus, elongation of the hull shell occurs vertically as well as
longitudinally along the length of the ship. Because of this, the Kinkead method is sensitive to the
position of the striking ship. The amount of energy to produce rupture in the hull shell is highly
dependent upon the height of the striking point from the nearest deck or subdeck, and the pitch of
the web frames along the ship.
With the Kinkead method, simplification of the ship structure is not required for analysis purposes.
Since the impact is seen as a sum of local impacts on particular components, all of these comporients
should be considered in as much detail as possible to obtain a more accurate result.
The developments of the Japanese experiments
The Japanese studies further developed the RosenblattiMcDermott visualisation of collapse as
progressive sequence of events. Their work was specific to the double hulled structures for which the
Rosenblatt method had not originally catered for. The three-dimensional nature of hull rupture was
investigated, as well as the residual strength of the web frames after buckling. These led to the
development of stiffness equations and methods to predict deformations and forces using ftnite
element methods. This was an evolutionary process. The work of Ando and Arita (IS) was based on an
oversimplifted model of a struck ship, where the structures of the ship was not modelled in great
detail. The work of Ito et al (!
6
) corrects this by introducing some complexity in the model. These
methods are applicable to both large and small deformations.
70
Pedersen
Pedersen used finite element analysis for the purpose of developing guidance in design. He undertook
a number of sensitivity analyses to establish the effect of varying parameters such as:
The inter hull spacing
Shell thickness
Strain rate of the hull
The spacing of the web frames
The web frame web thiclrness
Pedersen found that the resistance to collision was sensitive to the thickness of the web frame.
Pedersen then used his studies to develop curves, which demonstrate the variation of energy
absorption with the various parameters.
He also investigated the relationships between the velocity of the impacting vessel and the absorbed
energy, which depends on draughts and the structural dimensions of the two vessels involved. This
work shows similarity with that undertaken by Germanischer Lloyd.
Summary
The MSL review of the various methods of analysis has developed the following conclusion:
For design purposes the Kinkead substantiation method appears to be the most appropriate method.
To answer the question of how much energy is needed to produce a given deformation, which is the
question posed in the codes, the Kinkead method is simple to use. Its simplicity is a result of the
visualisation of collapse in the substantiation method. By visualising the coiiision process and the
resulting deformation, as a series of discrete events only a minimum knowledge of the mechanics of
collapse and the interdependence of the various structural elements in collapse is required. It is
interesting to note that in the Kinkead method certain mechanisms not mentioned in other methods
are highlighted (i.e. wedge splitting of a deck; and post-rupture plastic bending of the side shell). The
simplicity of the substantiation method lends itself to both hand and computer methods, whilst giving
conservative assessment.
The Japanese work and the work conducted by Pedersen have given insights into the behaviour of the
struck vessel and the progressive nature of collapse. For the designer, their investigations into the
variation of energy absorption with certain structural parameters, is a useful guide for optimisation in
the design process. Therefore finite element analysis can provide more information on the
deformations incurred in collapse and how these are time-history dependent
From the work discussed above, it can be seen that a steel of high quality high strain capacity; thick
side shell and wide web frames spacing, provides the best design. lbis is because the arrangement
allows for the membrane action of the side shell to become active when subjected to out-of-plane
loads. Therefore, the out-of-plane stiffness of the side shell is dependent on the deformed geometry
and thickness of the plate (in a positive sense). Thus as the deflection of the hull increases so does the
stiffuess. As the hull deflects the ability of the plate to resist the collision force increases, and the
high strain capacity of the steel allows rupture to be delayed.
7.4 Parametric Studies
A number of studies involving the sensitivity of double-hull structures to side-shell impact have been
undertaken. Work by Ito et al/NKK and Pedersen have revealed significant enhancements in energy
absorption capabilities resulting from the varying of a number of scantling design parameters. These
include increasing the double-side hull depth, transverse web thickness, side-shell plate thickness and
stringer thickness.
71
Considerable enhancements in energy absorbed were observed when increasing the hull
depth beyond the 2m minimum requirements of MARPOL and more so when compared to a single
hull arrangement. Increasing the transverse web thiclmess was also seen as an effective means of
increasing the energy to be absorbed.
Studies were also conducted on location of impact (i.e. impact on web or in between webs). In the
example given the energy was significantly higher when impact occurred on the web. However for an
alternative arrangement of transverse web spacing and size of impacting stem the reverse was
obtained.
A sensitivity study by MSL using the Kinkead method was also undertaken for a number of different
impacting vessels and impacting bow position. The study highlighted the ability of the struck vessel
to absorb energy depending on vessel size and the location of impact. The energy absorbed varied
between 49 MJ and 205 MJ. Further studies comparing the K.inkead method with other methods (i.e.
Minorsky, Modified Minorsky, Haywood, Rosenblatt)' revealed variations of energy between 36.5 MJ
and 217 MJ. A further study involving the side shell collision resistance of two FPSO vessels
selected from information provided by HSE was performed for a given assumed striking vessel
characteristics. This further sensitivity study demonstrated that the energy absorption capabilities for
each vessel type were again influenced by the different scantling arrangement, impact scenarios (i.e.
impact between webs and on web) and method used. The study also highlighted the potential
enhancements that could be obtained in the side shell energy absorption capabilities of the vessel
when using improved material rupture strain characteristics and/or increasing the side shell wall
thiclmess.
7.5 Overall Conclusions
Side shell collision of ship shaped structures is of concern, because of the increasing volume of ship
traffic carrying petrolewn and its products. Work has been done in this area as detailed in this report.
The codified guidance and criteria for collision could be improved by incorporating some of the
methods discussed above from the technical literature. The designer would gain a greater insight into
the mechanics of collision. Sensitivity analyses have revealed the importance of varying either the
dimensions of scantling parameters and /or the impact location of striking vessel to significantly
increase the ability of the struck vessel to absorb energy. Guidance in this area to indicate which
scantling elements/ impact location would influence the strength of the ship against collision would
be desirable. It should also be mentioned that if the strength of a ship is found not to be sufficient to
withstand collision, then the assumption of the infinitely stiff attacking bow should be abandoned,
and the shared energy approach(7) should be used. This allows for energy to be absorbed in the
attacking bow as well as the struck side.
The methods for analysing ship collision above can give different values for energy absorption as
identified from work published in the literature and demonstrated in the sensitivity studies
performed. Further validation using experimental work should be undertaken in this area
accompanied by hand calculations, where reliable, and numerical methods (such as finite element
analysis) to produce a more precise general method.
72
REFERENCES
1. AMERICAN BUREAU OF SHIPPING
US Oil Pollution Act of 1990, A Summary,
September 1990
2. INTERNATIONAL MARITIME ORGANISATION (IMO)
Marine Environmental Protection Committee (MEPC)- Annex I to MARPOL 73178
1992
3. DEPARTMENTOFENERGY
Guidance on Design Construction and Certification,
Published by HMSO for Department of Energy - now the HSE Offshore Safety Division
1990 (Guidance now withdrawn)
4. DET NORSKE VERITAS
Rules for the Classification of Ships Part 5, Chapter 9- Oil Storage and Production Vessels
1996
5. DET NORSKE VERITAS
Rules for the Classification of Ships Part 5, Chapter 3 - Oil Carriers
1999
6. NORWEGIAN PETROLEUM DIRECTORATE
Norwegian Petroleum Directorate::.Acts, Regulations and Provisions for the Petroleum
Activity
1993
7. NORSOK Standard
Design of Steel Structures- Actions and Action Effects, N-003
Revision 1, 1999
8. LLOYDS REGISTER OF SHIPPING
Rules and Regulations for the Classification of Floating Offshore Installations, Part 4:
Chapter 3.
1999
9.
Rules for Classification and Constroction, 1- Ship Technology Part 1- Seagoing Ships,
Chapter 1 Hull Structures, Section 33
1997
10. INTERNATIONAL MARITIME ORGANISATION
International Code for the Construction and Equipment of Ship Carrying Dangerous
Chemicals in Bulk
1990
II. INTERNATIONAL STANDARDS ORGANISATION (ISO)
Materials, equipment and offshore structures for petroleum and natural gas industries -
Offshore Structures- Floating systems- Part 4: Structures
1999 -DRAFT
73
12. DET NORSKE VERITAS
Impact Loads from Boats
DNV, Technical Note TNA202,1981
13. LLOYDS REGISTER OF SHIPPING
Boat Impact Study, Offshore Technology Report
OTH 85 224
14. J.P. KENNY FOR THE DEPARTMENT OF ENERGY
Protection of Offshore Installations against Impact
OTI 88 535
15. V.U. MINORSKY
An Analysis of Ship Collisions With Reference to Protection of Nuclear Power Plants
Journal of Ship Research, October 1959
16. H.J.HAYWOOD
A note on Collision estimates for LNG carriers
Naval Construction Research Establishment, Dunfennline, February 1971
17. G.WOISIN
Schijjbauliche Forschungsarbeiten fur die Sicherheit, Kernenergiegetriebener Handelsshiffe
Jahrbuch den Schiffbautechnischen Gesellschat 65 Band, 1971
(Springer-Verlag Berlin) 1972
18. J.F. McDERMOTT
Tanker Analysis for Minor Collisions
M. Rosenblatt and Son, Inc. 1974
19. AN.KJNKEAD
A Method for Analysing Cargo Protection Afforded by Ship Structures in Collision and its
Application to LNG Carriers
The Royal Institution of Naval Architects, 1979
20. Y.AKITA
Reply to Discussion in Part J (p404)
Proceedings of the First International Conference on Structural Mechanics In Reactor
Technology, Berlin 1971
21. C. BOE, J.ODLAND
Private Communication (between A.N. Kinkead and J. Odland), DNV
December 1976
22. N. ANDO, K AKITA
A study on the Strength of Double-Hull Structures in Collision
Journal of The Society of Naval Architects of Japan, Vol. 139, Jl.Ule 1976
23. H J S A ~ H J ITO, K. KONDO et a1
A Simplified Method to Analyse the Strength of Double Hull Structures in Collision
The Society of Naval Architects of Japan, Volume 156, November 1984
74
24. illSASHI ITO, K. KONDO eta!
A Simplified Method to Analyse the Strength of Double Hull Strnctures in Collision (2nd
Report)
The Society of Naval Architects of Japan, Volume 158, November 1985
25. illSASHI ITO, K. KONDO eta!
A Simplified Method to Analyse the Strength of Double Hull Structures in Collision (3rd
Report)
The Society of Naval Architects of Japan, Volume 160, November 1986
26. HISASill ITO, K. HAYASill
A Comparative Study of Simplified Methods to Analyse the Strength of Doubled-Hull Tankers
in Collision.
Proceedings of the Fourth (1994) International Offshore and Polar Energy Conference, Japan,
April1994.
27. HISASHI ITO, T. GOTO
Collision Resistance of an NKK Double-Hulled Tanker
NKK Teclmical Review No. 71, 1994
28. P.T PETTERSEN
Analysis and Design of Cellular Structures
Report No UR-79-02, Department of Marine Technology, The Norwegian Institute Of
Tec!mology, 1979
29. M. BOCKENHAUER, E.D. EGGE
Assessment of the Collision Resistance of Ships for Classification Purposes
Gennanischer Lloyd, 1995
30. P. T. PEDERSEN, J. PAIK
Modelling of the Internal Mechanics in Ship Collisions
Ocean Engineering, Vol. 23 No.2, 1996
31. P. T. PEDERSEN, J. PAIK
On Design of Double Hull Tankers against Collisions
The 6th International Symposium on Practical Design of Ships and Mobile Units September
17-22, 1995
32. KARL A. RECKLING
The Mechanics of Minor Ship Collisions
International Journal of Impact Engineering, Volume 1, No. 3, 1983
33. P. SEN, J. A. COCKS
Collision Resistance
University of Newcastle-upon-Tyne, 1989
34. HAGIW ARA
A Proposed method of Predicting Ship Collision Damage
International Journal of Impact Engineering Volume 1, No.3, 1983
35. KITAMURA
Comparative study on Collision Resistance of Side Structure
The Society of Naval Architects of Japan, 1996
75
36. KITAMURA
A study on improved Tanker structure against collision and grounding damage
Nagasaki Research and Development Centre, Mitsubishi Heavy Industries, Ltd, 1998
37. KITAMURA
A Study on the Crashworthiness of a Double Side Structure of VLCC
Nagasaki Research and Development Centre, Mitsubishi Heavy Industries, Ltd, 1997
38. K.ARITA
A Study on the Strength of Shells in Relation to Ship Collision
The Society ofNaval Architects of Japan
39. MIZUKAMI, M. TANIGAWA eta!
Collision Simulation of a Double-Hulled Structure with Uni-Directional Girder System
International Society of Offshore and Polar Engineers, 1996
40. KIM, K. LEE et a!
Behaviour of Double Hull VLCCs in Collision
The Sixth Symposium on Practical Design of Ships, 1995
41. A.VREDEVELDT, L. WEYERS
Full Scale Ship Collision Tests
First Joint Conference on Marine Safety, 1992
42. DAIDOLA
Tanker Structure behaviour during Collision and Grounding
The Society ofNaval Architects and Marine Engineers, New York, 1993
43. J.PAIK,A.THAYAMBALL!
Residual Strength Assessment of Ships after Collision and Grounding
Marine Technology, Volume 35, No.1, 1998
44. SEN, J. COCKS, M. PAWLOWSKI
Integrated Analysis of Ships
University ofNewcastle-upon-Tyne
45. G. WOISIN
Analysis of the Collision between Rigid Bulb and Side shell Panel
Practical Design of Ships and Offshore Structures, 1998
46. ZHU, J. ZHENG, Y. ZHOU
Rigid Plastic Dynamic Analysis of Ship Collisions
Kansai Society ofNaval Architects, 1993
Printed and published by the Health and Safety Executive
co.s 4101
76

Potrebbero piacerti anche