i
TECHÑISCRE Uf4VER$JThT
taboratojum voor
Scheep3hydromhi
Archíef
CRITERIA FOR SURVIVAL IN DAMAGED
cXèDêjN28CD
Deift015 788873 Faxt 015-78183S
D. Vassalos, M. Pawlowski" and O. Turan
Department of Ship and Marine Technology, University of Strathclyde, Glasgow, UK Visiting Research Fellow, Gdansk Technical University, Poland
SUMMARY
Buildinc on earlier developments under the UK RoRo Research Programme, the focus of the effort during the North West
European Joint R&D Project has been on improving and validating comprehensively an exisring theoretical model,
capaOle of predicting the capsizal resistance of a damaged vessel in realistic environmental conditions wniisr accounting for progressive floocing. Having successfully fulfilled this task, the Strathc!yde University Ship Stability (SUSS) Research Group has subsequently undertaken a systematic parametric investigation aiming to identity the key factors influencing the vessel's ability to survive a said damage as well as representative parameters characterising uniquely this ability for any vessel type. form, size and compartmentation as a function of sea state with the view to using these as a basis tor
developing rational survival criteria. In their quest to meeting these goals, the Strathclyde team achieved a major
breakthrough which is likely to serve as a key contribution to new 1MO regulations, for assessing the damage survivablilty of passenger/RoRo vessais, and to facilitate the harmonisation of probabilistic stability standards.
AUTHORS' BOGRAPHES
Dracos Vassalos is a Professor in the Department of Ship and Marine Technology at the University of Strathclyde and
the leader of the Stability Research Group, con'iorising some 12 researchers and research students dealing with all
aspects of stability of ships and ocean vehicles. He has been involved with stability research for over 18 years and has
served as a Principal Investigator in a numoer of major research projects including: SAFESHIP, PRESS, UK RoRo
Research Programme, Stability of High Speed Twin-Hull Craft. MOD Trimaran Programme ana the Joint North West
European Project. Professor Vassaios is the overall co-ordinator of a major EU initiative on the safety of RoRo vessels involving S Universities and over 30 companies and research institutes from 11 European Nations an a member of the UK delegation to 1MO for ship stability.
Maciej Pawlowski is a visiting Research Feilow in the Department of Ship and Marine Technology at the University of
Strathclyde since July 1995 involved the development of probabilistic damage stability cnteria within the North West
European R&D Project. Maciej Pawlowski is a Professor in the Department of Ocean Engineering and Shipbuilding at
the Technical University of Gdansk. specialising on ship statics and, in particular. the probabilistic assessment of ship safety in damaged condition. In connection with the latter Professor Pawiowski has since 1972 been very active in the
SLF Sub-Committee at 1MO where he has made significant contributions to the prooabilistic subdivision regulations. Osman Turan is a Research Fellow in the Department of Ship and Marine Technology at the University of Strathclyde and a key member of the Stability Research Group. Dr. Turan is involved with theoretical, experimental and rule development
work concerning RoRo vessels as well as general stability problems of ships and advanced marine vehicles. After
completing his PhD at Strathclvde on the "Dynamic Stability Assessment of Damaged Passenger Ships Using a Time
Simulation Aoroach" and working as a research assistant on damage stability related projects he took up a lectureship post at Yildiz Technical University in Turkey. He rejoined the Strathciyoe team in February 1995.
1. INTRODUCTION
The limited understanding of the complex dynamic
behaviour of a damaged vessel and the progression of flood water through the ship in a random sea state has,
to date, resulted in
approaches for assessing the
damage survivability of ships that
rely mainly on
hydrostatic properties.Furthermore, in case of serious flooding of ships with
large undivided deck spaces. such as RoRo vessels, the
loss could be catastrophic as a result of rapid capsize,
rendering evacuation of passengers and crew im practical, with disastrous (unacceptable) consecuences. Trie tragic accidents of the Herald of Free
Enterprise and more recently
ofEstonia were the
strongest indicators yet of the existing gaps in assessing damage survivability in a flooded condition particularly for this class of vessels.
Following the tragic capsize of the Herald of Free
Enterprise in March 1987 and realising thisunsatisfactory state of affairs, the UK Department of
Transport initiated anextensive RoRo Research
Programme, comprising a number of studies
intodamage survivability of these ships, including model tests n waves, [1] to [4] and numerical simulations, [5],
[6.
The Strathclyde team was charged with the responsibility of developina and validating a theoretical capsize model
which could predict the minimum stability needed by a
damaged vessel to resist capsizing in a given sea state. This responsibility was successfully discharged in 1992,
[1. With the Estonia tragedy shaking once more the
foundations of shipping, it forced the profession to seek soiutions urgently and, in attempting to do so, to use the
right expertise and experience to
provide the rightcapsize safety has suddenly reached the deserved and
lonc overdue intensity. The Nordic countries reacted
quickly in undertaking this responsibility leading to a
wider-based project, known as the North West European Research arid Development Project (Joint R&D Project), aiming to ensure "passenger survival" irrespective of thedegree of damage and severity of sea state RoRo
vessels are likely to encounter.
Deriving from the success demonstrated by the SUSS Group during theUK RoRo research and taken into consideration that the
UK had already commissioned Phase lIA of the RoRo Survivability Research. focusing on the mathematical model developed at the University of Strathclyde, the
Group was invited to participate in this Joint R&D Project.
The intention was, following a process of
furtherdevelopment and validation of the existing simulation
program, to apply it to different vessel types, forms, sizes
and compartmentation and to representative damage scenarios and environments with a view to verifying its
general applicability to assessing the capsize safety of a
damaged ship in a given sea state and to using it as a basis for developing new survival criteria for ships in
damaged condition.
This work formed Task 5 of the
ProjectThe focus in this paoer will be on the development of new survival criteria. Full details of ail the work undertaken at Strathclyde under the Joint R&D Project can be found in
(81.
2. BACKGROUND
2.1 THE ORIGINAL SURVIVAL FACTOR
At about the same time as the 1974 SOLAS Convention was introduced, the International Maritime Organisation
(1MO), published Resolution A265 (VIII), [9]. These
regulations used a probabilistic approach to assessing
damage location and extent drawing upon statistical data to derive estimates for the likelihood of particular damage
cases.
The method consists of the calculation of an Attained
Index of Subdivision, A, for the ship which must be
greater than or equal to a Required Subdivision Index, R, which isa function of ship length, passenger/crew
numbers and lifeboat capacity, Index "A" is in turn a
function of three different probabilities, "a", "p" and "s". Factor "a" accounts for the probability of damage as related to the position of the compartment in the ship's
length: "p"
reflects the effectof variation on the
longitudinal
extent of damage: "s" represents the
probability of survival given
the damage under
consideration.
The total attained index is the sum of the products of "ai",
"p«' and "s" for each of
the compartment and
compartment groups. i. within the ship. lt is the survival factor s, where attention was focused in Task 5.
lt is true to state that the ship damage stability problem has not received much research attention in the past
mainly because a meaningful treatment of it, particularly one involving progressive flooding in a random seaway,
was perceived to be too difficult an undertaking by
theoretical/numerical means.
For this reason an
experimental approach was adopted aiming to establish
a simplified relationship berveen environmental and
stability-related parameters for a damaged ship and
hence determine capsizal resistance in a given sea. On the basis of limited model tests carried out separately inthe United Kingdom, (10], and the USA, [11] such a
relationship was established between critical metacentnc heights (GM ) - limit of capsizal resistance- and vessel freeboard in a given critical sea state, characterised by the significant wave height. (H0 ). From the results of these tests it was decided to use flooded metaceritricheight (GM) and effective freeboard (Fe) rather than the
restoring (GZ ) curve ana GZ curve characteristics to
judge capsizal resistance. Suoolernentary model tests
have shown that for a given freeboard in any given sea state the critical GM is proportional to the beam of the
vessel, (B). Consequently, the following relationship was invoked:
(H3) cntc = f (GM1" FelS) (1)
Deriving from the above, the probability "s" that a snip with a given value of the damage stability parameter
(GM1" F018) will survive damage in a given sea state will
be equal to the probability of not exceeding (H0 )cntj Therefore, the probability "s" can be derived from the
significant wave height distribution relevant to the area of the particular accident in conjunction with the relationship indicated in (1), i.e.:
0=F[H0(GM+" Fe/B)] (2)
A graph of this probability as a function of the damage
stability parameter is shown in Figure 1 below, obtained by using data given in [9] & [10].
0.8 3.6
0.4
02
001 02 003 0.04 (m
Figure 1: The probability of collision survival s
a: based on model tests and sea state distribution(10],
(11];
b: based on a comparative method [121; C: approximation adopted by 1MO
A reasonable approximation of this function is s = t7 (GMf Fe/B) 1/6 = (x/x1 ) 1/6 (3)
for S C (0,1), where x is the damage stability parameter in metres and x1 (= 0.0416 m) is a vaiue of x yielding
s =
1. 1MO approximates this probabilitywith a
considerable underestimation, as shown in Figure 1,
using the equation:
- C (0,1). If s is less than 0.6, which happens with
.o15 m, then 1MO requires s to be taken as zero [9],
rflectiflg the lack of confidence
inthe quality
offor small s factors. In this range. a good
approx mat 0fl can be obtained for the s factor by usingbear interpolation between s = 0.6 for x = 0.015 m and
s = i for x = 0.04 m.
UK RaRo RESEARCH PROGRAMME
A new attempt to generalise the results of model
exoeriments and numerical simulations of the extensive RoRo research programme undertaken in the UK in the wake of the Herald of Free Enterprise disaster has been proposed in the form of a relationship between the ratioHJF
and the non-dimensional flooded metacentric
height OM , defined as follows:
GMfC5Tf
B2 (5)
where, L. f ) volume of displacement of the vessel together with the flooded water block coefficient of the vessel to the damaged waterline
T >
draught of the vessel in the flooded
conditionL5 z) length between perpendiculars
with the other parameters as defined earlier. The
non-dimensional flooded metacentnc height
is definedbasicaily as the ratio GM1 / BM, where BM. is the
metacentric radius in the flooded condition. However, thistype of boundary survivability curve could not be
generaiised either as GMf and Fe do not reflect, for
example, the effect of different structural arrangements on the vehicle deck that can be used for enhancing the
damage survivability of RoRo vessels. For this purpose, the definition of the effective freeboard would have to be much more intricate.
Damage stability is a complex problem involving a large number of variables and with results only from one or two
given shapes it is difficult to identity the significant
Parameters governing the problem.
An extensive Investigation can, of course, be undertaken by usingnumerical simulations, provided the latter are of the same
accuracy as the model tests. Full benefit from these
results will not be derived, however, until a consistent theory is developed which will enable the results to be generalised to other ship forms, sizes, and subdivision arrangements, thus leading to a definition of a unique
boundary survivability curve valid for
ail vessels, asexplained in the following.
This goal is essentially synonymous with the
development of rational survival criteria.
BOUNDARY
SURV1VABL1TYCURVE
lt is generally acknowledged that the critical sea state a
Ship with a given loading condition and compartment
3
flooaed can withstand, cannot be determined uniquely.
This is
not because of inaccuracies
in the modelexperiments or numerical simulations, nor because of limited duration of test runs, but simply because of the
random nature of the critical sea state, characterised by a certain distribution. Hence, any boundary survivability curve is a fuzzy curve, indicative only of the mean values
of the critical significant wave heights, characterising such sea states. Therefore, to define the distribution of
critical sea states (and its mean or median value), a
given flooding scenario will need to be repeated many times with different initial conditions. A boundary curve
cannot be defined by just one irial". This dependence of capsize boundaries on initial conditions is characteristic
of the non-linear nature of the problem at hand but has only recently been appreciated. The probability s that a ship with a given loading condition and compartment flooded will not capsize after damage is equal to the mean probabiiity that the critical significant wave height
related to this case is not exceeded:
s = [HF(HS) (H5) dH5 (7)
f(H5 ) z) probability density function of
critical sea states for a ship with a given loading condition and compartment flooded
The range of change of F(H5) for moderate and higher
critical sea states is small, as can be seen in [9], whereas
for low critical sea states (when damaged stability is deficient) the range of variation of critical sea states is
narrow. Therefore, by virtue of the mean value theorem, equation (7) yields:
SF[(Hs)me1
(8)This implies that,
inpractice, the s factor can be
calculated as if the critical sea states were of binary
nature, cut oft at the mean value. As can be seen, of prime importance for the determination of the s factor, and thus for the whole method, is the basic expression for the s factor, given by equation (8), that requires the
critical sea state (Hs)mean to be known tor each damage case, denoted in the following simply by H5. However, to obtain this basic factor, it will be necessary to determine
H5 without the
aidof model tests
or numerical
simulations, as these are not suited and intended for
routine applications. For this purpose, a simple but
meaningful procedure must be developed which can be verified by previous and new results, preferably based onthe theory underlying the numerical simulations, [5].
Before proceeding with the description
of such a
procedure, a
brief outline of the development and
validation of the numerical simulation program will first be given, highlighting some of the deficiencies still in need of further attention and development.
3. MATHEMATICAL MODELLING
To study effectively damage survivabiiity one needs to put together a highly non-linear six-dearees-of-freecom
where F (H5 z) cumulative distribution
function
of sea states at the
seakeeping model (that allows the vessel to dritt as well
as changes in
its mass, centre of mass and mean
attitude relative to the mean waterplane with time - the
same dependence of environmental excitation and
hydrodynamic reaction forces on the changingunderwater volume of the vessel must also be catered for), a water ingress model (that allows for
multiple-compartment flooding in the presence of oscillatory flows in extreme wave conditions and at times of shear flows),
a sloshing model (that allows for random inflow and
outflow through multiple openings), together with their
interaction.
An attempt in
this direction represents ongoing efforts among the SUSS Group [13] This is a top-down approach to solving this problem. The use ofsuch a
tool" inits complete form, however, for
parametric investigations and routine design applications is outwith reach at present.
The state-of-the art,
currently represented by the
numerical model developed at Strathclyde, relates to a three-degree-of-freedom non-linear seakeeping model (allowing for change of vessel mass and heave and roil
restoring that take
into account ship motions,
trim,sinkage and heel) and a hydraulic water ingress model (allowing for water inflow and outflow and associated
gravitational forces in a semi-empirical manner,
applicable to multiple-compartment flooding and to any vessel subdivision and deck arrangement). Flood water s assumed to move in phase with the ship roll motion with an instantaneous free-surface parailel to the mean waterplane. This is the model developed during the UK
RoRo research following a bottom-up approach and
formed also the basis for the Joint R&D Project. The
numerical model has undergone vigorous validation by means of supporting model experiments using three
different RoRo vessels tested
by three
differentorganisations (British Maritime Technology Ltd. [7], and Danish Maritime Institute and M.ARINTEK, [8]), providing
strong evidence that this relatively simpler model can
describe adequately the behaviour of a damaged vessel
subjected to large scale flooding, thus allowing for
determination of a capsizal boundary with
sufficient engineering accuracy. Typical comparisons betweentheoretical and experimental results are shown in Figure 2, taken from [8]. In this respect, this model serves an
extremely useful purpose.
lt must be understood. however, that damaged vessels
could in theory capsize in a mode still dominated by
vessel and flood water dynamics and in some cases in a very unpredictable manner, characteristic of highly non-linear systems. Furthermore, the emphasis in studyingdamage survivability need not focus necessarily
indetermining the capsizal boundary. For example, the
actual time taken for the vessel to capsize (survival
time), knowledge of which is absolutely vital to ensure
successful mustering and evacuation of passengers and
crew, requires a more accurate modelling of the whole
dynamic process.
This forms the focus
of a new
research effort, supported yet again by the Manne Safety Agency of the UK Department of Transport, undertaken jointly by the University of Strathciyde and BMT Ltd.
4. DEVELOPMENT OF SURVIVAL CRITERIA 4.1. SENSITIVITY STUDY
In order to identity the most influential parameters for the
stability and survivability of a damaged ship, a series of
parametric studies have been carried out using the
developed numerical simulation program.
For this
purpose a matrix which combines different damaged
freeboards, vehicle deck subdivisions, loading conditions, damage location, vessel size and sea states
has been tested. The parametric investigation began
with a sensitivity study of the two ships used in the Joint R&D Project which were tested extensively for software
verificationlvaíidation purposes. SHIP i was also used
during the UK RoRo Research Programme whilst SHiP 2
is representative of a modem passenger/RoRo Vessel
with improved stability characteristics. Full details of this
investigation can be found in [8]. Some information of
the ships and damaged conditions used in the sensitivity
study are given here, followed by presentation and
discussion of only the results which are directly related to the development of survival criteria.SHIP i drausnt d Disniacernent. Block Coetfc,ent. C5 Intact Freeboard. F 6. IDo I 12.4000 tonnes 0.582 I 1.68 n
Table 1:
Principal Design Particulars of SHiP i
Table 2: Sensitivity Study Test Matrix for SHIP I
As can be seen there are 60 test conditions and for each condition a minimum of four different sea states has been considered.The sea states were tested to Q.25rP
resolution (i.e. sea states were increased progressivelYby 0.25m intervals). Where necessary, several n.in5
were carried out for the same conditions to ensure
statistical consIstency of the results. All condition5
address midship damage and the derived results refer tO 100% SOLAS damage opening, as defined in [7].
Model Scaje 42.05I
Lensth. L 3100mI
Bearn.B 2600mj
Deoth to Buikheao DecK. D 780m I Deoth to Uouer Most Corrnnuous Deck 18 0to I
OPEN DECK CENTRAL
CASING CASINGS
OFEN DECK -TRANS VESSE 8ULKHE..DS KG Oanaed F (mi Damaged F (rai Oarriased F (ml Damaged F 1m) (m l0.2110.571L02 O.2210.37IL0210.2115.37i 1.02 0.2119.5711.02I
9.0
NIX X XIXlX
xixix
XIX
X10.0
lxix
xxIxIxI
XXI
xFxlx
X11.0
IXIX xjxIXlXi XXIX
lxix
X12.0
XIX
X XIXXI XjXI
xxix
XTable 3 Damaged Conditions for SHIP i
Table 5: Sensitivity Study Test Matrix for SHIP
As shown in the table above, the sensitivity study matrix
for SHIP 2 covers 72 different test conditions, for which
the limtting sea states were found. Similar to SHIP 1, sea
states were tested to 0.25m resolution.
Again allconditions refer to midship damage and to 100% SOLAS damage opening unless stated otherwise.
Table 4: Principal Design Particu'ars of SHIP 2
Table 6: Damaged Conditions for SHIP 2
The wave environment used in model tests was also
Used to undertake the numerical simulations. lt is
representative of the North Sea and is modelled by using JONSWAP spectra. Details for this and of the vehicle deck Subdivision arrangements used for the two ships
are given in [8}.
D
4.2 SENSITIVITY OF VESSEL SURVIVAL ON GM
AND OTHER RESiDUAL STABILITY
PARAMETERS
If different subdivisions of the
vehicle deck are
contemplated, then clearly GM1 cannot be considered as
a representative parameter to characterise the damage
suniivabiUty of passenger/RoRo vessels. However, GM
in itself is
the key opening the door to
the most successful characteristic property of a vessel's ability toresist capsize in any condition and environment, namely, the restoring curve. Even if one does not support this
view, any results that this route is likely to yield, offer two distinct advantages: simplicity and applicability. It was
attempted initially, therefore,
to express the survival
factor "s" as a function of residual stability characteristics,
judiciously chosen (e.g. systematic parametric
investigations, regression analyses, experiential
judgement, etc.) to
enable such a factor to be
generalised for application to
all vessel types and
compartnientation.
A first exploration in this direction met with a problem that needed careful thinking. Damage stability calculations for
vessels damaged both above and below the bulkhead
deck would require, according to MO, that the water
level in each damaged compartment open to the sea
must be at the same level as the sea i.e. final equilibrium be reached. However, the O-Z curves derived on the
basis of this approach, simply fail to offer any useful
information.The principal reason lies on the wrong
assumption that water is free-flooding the deck in these calculations. Taking heed from this and from the fact that water on deck is a dominant parameter affecting damage survivability, as earlier experience amply demonstrated,it was decided to attempt to quantify the critical amount
of water on deck as a matter of top priority. The first
important step was to achieve a good understanding of
what is meant by critical amount of water on deck and to develop a practical method of quantifying this.
4.3 CRITICAL AMOUNT OF WATER ON
DECK - "THE POINT OF NO-RETURN" The effect of random waves on the rolling motion of a
damaged ship appears to be rather small and for capsize to occur in a "pure" dynamic mode should be regarded as the exception rather than the rule. The main effect of the
waves, therefore, is to exacerbate flooding. In this
respect, the effect of heave motion
in reducing the damaged freeboard is as important as the roil motion.Model experiments and numerical simulations have
clearly demonstrated that the dominant factor determining the behaviour of the vessel is the amount of flood water accumulating on the vehicle deck. In case of large scale flooding,the vessel motions become
subdued with the mean heel angle increasing sIowly untila critical
value isreached beyond which heeling
increases exponentially and the vessel caosizes very
rapidly. In this context, the term "point of no return" is
used as indicative of the fate of the vessel when this critical heel angle is attained. Put differently, the flood water on the vehicle deck increases slowly, deoending on the vessel and environmental conditions, until the
amount accumulated reached a level that cannot be supported by the vessel/environment and the vessel
Modi Scsie j 34 66
130 on
255m
Dectho3uikheadDedD 335m
to Lpoer Most Conunuous Deck 7 10m
575m
DtsvIacement. 5 1200.0 tonnes Block CociScoent. C0 0 612
lrxtactFretboard.F 26m
OPEN DECK OE'TrRAI.CASDG CASC1OSama
OPENDECK"
TRE
KO DamaacsF(mliDan,a2edF(mlI Dom3Funl DooraeaF(mi
nu 11.0 1.0 0.5 0.5 1.0 LS 0.5 LOI LO (LS LO I .0
91)
'XIXIX xIXIX XIXrXIXIXIX
9-.)XX1X X1X1X XIXI X IX!XIXI
0.5
XIXIXIXIXJXI X
Xi X
lxi
XIX
1.5
'(IXIX XIXIXI XXIX Xi XIX
LO
Xl XIX XI XI XI XI XIX
xIx
X13.0 IXIXIX XIXIXIXIXIXIXIXiXI
Damaeco Comparunent Lenonh 1m) ' Daxnagec Freeboard (ml Intact KM (ml Damaged KM (ml 2405 1.5 14.22 I 13.46 3243 1.0 4.22 I 14.48 4025 0 5 1422 14.59 02 1795 13 SSDamaged Intact Damaged Freeboard (rn) KM (m KM 1m)
1395 14.34 0.55
13,95 I 14.74
capsizes very rapidly as a result. The amount of flood water when the point of no-return is reached is the critical amount of water on deck. In relation to this, two points deserve emphasis: This amount is substantially
less than the amount of waterjust before the vessel actually
caosizes but is in excess of the amount required to
statically capsize the ship. In this respect, the energy
input on account of the waves help the vessel sustain a
larger amount of water than what her static restoring
characteristics appear to dictate. Because of the nature of the capsize mode when serious flooding of the vehicle deck takes place, it is not difficuft to estimate the critical
amount of water on deck at the point of no-return from
experimental or numerical simulation records considering
either the flood water
on the vehicle deck or the roil
motion of vessel. This is demonstrated in Figure 3 with
some summary results shown in Figure 5, using SHIP 2. 4.4 STATIC EQUIVALENT METHOD
Careful examination of the numerical simulation results,
such as those shown in Figure 3, led to the following
observations and developments:
The point of no return occurs at a heeling angle
very close to 8m' where the
GZ curve is at
maximum. This angie, in the majority of casesexamined, is less then 10degrees. Reference is
made here to the
GZ curve
calculatedtraditionally, usina the
constant displacementmethod and allowing fer
free-floodingof the
vehicle deck when the deck edge is submerged. s This fact, coupled with observations from physicalmodel experiments and the experience amassed
from studying large numbers of numerical tests led to the development of an "Static Equivalent Method" which allows for the calculation of the
critical amount of water
on deck from statical
stability calculations.
To this end, a floocing
scenario is considered,in which the ship is
damaged only below the vehicle deck but with a certain amount of water on the (undamaged) deck
inside the upper (intact)
part of the ship. The
critical
amount of water on deck
is thendetermined by the amount causing the ship to
assume an angle of loll (angle ofequilibrium) 8e
that equals the angle
8m' determined
previously.Strictly speakinc, the
GZ curve should be
calculated by considering a freely floating ship, toallow for the trimming moment produced by the
amount of water on deck at the point of no-return.
The scenario described above scenario
isdepicted in Figure 4, believed to represent closely
observations of the flooding process near the capsize boundary or when a stationary (steady)
state is reached with the water ori elevated at an average height above the mean water plane, as a result of the wave action and vessel motions.
3
Figure 4: Stability of
a damaged ship with wateraccumulated on deck
(Static Equivalent Method)
In the process of seeking a generalized damaged
stabil-ity cnterion, the quantities considered to describe the
above scenario at the point of no return are h and f as
defined in Figure 4. Due to the dynamic action of waves, the flooded water accumulates on deck causing the ship to heel and trim and, when the deck edge becomes sub-merged, the water continues to elevate above the sea
level until it reached a height h, depending on the sea
state, at which inflow and outflow of water through the
opening is balanced, maintaining the amount of trapped water on the deck constant over some ceriod of time as
if the upper part of the
ship were intact. The ship assumes then a heel angle at which the heeling moment due to the accumulated water on deck is balanced by the restoring moment. This quasi-static heel angle
deter-mines in turn the mean roll angle about which the ship
oscillates - statically, this cannot be greater than the
angle 8max
4.5 A RATIONAL BOUNDARY SURVIVABILITY CUR VE
Deriving from the above and adopting the philosophy outlined in (14], efforts at this stage concentrated on
investigating whether a similar damage stability cnterion (boundary survivability curve), will be valid for any ship of
any size, loading condition, damage scenario and
subdivision arrangement. The relationship presently considered is in the formh/i-I5 = f(fl)
(9)where, h and f are the two quantities derived readily by
the static equivalent method
and H5 is the average
significant wave height characterising a critical sea state,
found with the aid of numerical simulations. Preceding
the examination of uniqueness of equation (9), was
obtaining evidence on the validity of the static equivalent method. which is described next.
(a) Comparison between numerical simulation
and 'static equivalent method"
resultsr 4.
were performed between the static and dynamic
ecicdcn5 of the critical amount of water on deck for
SHIP 2. considering the test matrix of Table 5.
The parameter used for comparason is the critical amount ofwater on the vehicle deck as derived from numerical si1uladofl and the stic equivalent method. Additional
icuIabOflS were also camed out for SHIP 1, considering
o-ity an open deck arrangement. Typical results are
presented in Figure
5, demonstrating the excellent
agreement achieved in all
the cases considered,
parbculariy so for the operational range of KG values. lt
should be noted that, in accordance with the nature of a critical sea state, the critical amount of water on deck is
aise a random quantity.
(b) Uniqueness
To validate the above considerations concerning
uniqueness of the boundary survivability curve, the threemcortant quantities
h, f and critical H3
have been calculated for SHIP i and SHIP 2, for as many as 126cases, including results from two scaled versions of SHIP (scales 1.5 and 2/3). The results of calculations are
lotted in Figures 6 to 9, using H5 instead of H3, where
sr is a modied significant wave height. accounting in a
ay for the relative motion between the vessel and
iaves at the damage opening. The points in the two
gures are differentiated only according to
different amage freeboards, denoted by F. Based on the resultsbtained, the following observations are note worthy:
Contrary to expectations, the immersion of the
deck edge at the damage opening, f, is irrelevant
concerning the capsizing
resistance of the
damaged ship and can be disregarded from
further considerations. This is clearly demonstrated in Figures 6 and 7. The immersionof
the deck edge at
theopening can be
interpreted as a negative freeboard and this is
likely to be the case considering the mode of
capsize described
in the foregoing in thepresence of large scale flooding. When the deck s above the sea level, i.e. freeboard is positive,
the situation is different. Naturally, in this case,
both the flooding process threshold and the
amount of kinetic energy of waves that
isavailable to sustain the water accumulated on the
deck at a given level would be affected by the
freeboard.
The above observation leads directly to the very important conclusion that the elevation of water
on deck above the sea level, h, is the unique
measure of ship survival in the damaged condition(the higher the water elevation at the point of no
return, the higher the sea state needed to elevate
water to this level and the higher the capsizal
resistance of the ship. lt follows, therefore, that
the relationship between h and H3 ¡s unique for a given ship, irrespective of subdivision
arrangements and loading conditions,
as itdepends primarily on the relative motion between the vessel and waves at the opening. lt would be
very surprising, if that were not the case. Other
effects, such as shape and size of the damage or
type of ship, are marginal. The relationship can
7
only be affected by the dynamic properties of the
ship
(primarily heave motion) and sea state
characteristics.s Surprisingly, the size of the ship has little or no
effect on the survival capability. Figure 8 clearly
confirms this conclusion.
The reason for this
derives from the fact that the effect of ship size is
largely accounted for by the modified significant
wave height H3r. Moreover, a change of H5 with
ship size is associated with a change of wave
period T that depends on ship size in such a
manner that the resulting effect is neutralised.The location of the compartment along the ship length has no effect on the survival capability of the ship provided that the static calculations are performed accounting for the trimming moment
due to water elevated on deck.
s The smallest scatter of points in Figures 6 and 7
is obtained for
Hsr- Hs1.3 (10)
s The average value of h /H equals then 0.085 for
both test ships. Hence, the sought boundary
survivability curve may take the form: h = 0.085 H513 (11)
This equation provides on the whole an excellent prediction, with deviations in the majority of cases
less than the sea state resolution used to derive
Hs. This impressive result can be observed in
Figures 8 and 9. Eleven points derived from the
model experiments undertaken during the Joint
R&D Project are also incorporated ¡n the sample data presented in Figure 8 which appear to fit the theoretical prediction very well.
(c) Practical application
This boundary survivability curve in the form of equation (11) could not be simpier for practical
applications.
For this purpose, for a given
compartment group it is necessary to calculate only the measure of damage stability, i.e. the elevation of flooded water on deck h at the point of no return, equalling the angle 6m' The mean
critical sea state is then given as
H3=( h
(12)\0.o851
s Having determined the critical sea state H3 for a given damage, the factor s can be easily obtained
from the sea state distribution occurring at the moment of collision F = F (H3 ), as explained in the foregoing. For purposes of illustration alone, the sea state distribution proposed by 1MO, [9],
could be used, as shown in Figure iO.
lt isnoteworthy that the distribution of sea states at the moment of collision is different from the sea state distribution, obtained from regular weather
statistics. In a large majority of cases, collisions
happen in the proximity of ports,
in confinedin a similar manner to that previously used for the damage stability parameter GMf"FJB. A graph of
this probability is shown in Figure 11, using the
sea state distribution, shown in
Figure 10, inassociation with the boundary survivabiHty curve,
given by equation (12).
A very good approximation of this function for h up
to 0.6 metres is given by
s {h/0.6+h(0.6-h) (ah2+b hc)]1"4 (14)
for he0, 0.6, with a
= 11.7; b = -16.9; c = 8.40. For h greater than 0.6 m, s = 1.Specific applications ought to consider actual
distribu-tions. characterising the area of operation, as described in paper 6.
5. CONCLUDING REMARKS
Based on the research work described in the foregoing
the following concluding remarks can be made:
A numerical simulation model has been
developed, capable of predicting with good
engineering accuracy, the capsizal resistance of a damaged ship, of any type andcompartrnentation, in a seaway whilst accounting for progressive ffooding.
A comprehensive calibration/validation
programme has ailowed for sufficient confidence
to be built up, rendering the developed model a
valuable design "tool".
A systematic
parametric investigation hasproduced valuable results, showing a clear way
forward in the development of survival criteria for
passenger/RoRo vessels within a probabilistic framework as well as offering sound knowledge
and design ideas for improving their damage
survivability.Finally and, most importantly, a damage stability criterion
has been proposed deriving
from physical considerations, based on modelexperiments and numerical simulations.
ltuniquely represents ships of any form, s;ze and
subdivision arrangement, and can be readily
implemented n practice. This major achievement offers a real opportunity for rationalising damage stability assessment. There is, therefore, hoflethat this fully rationai criterion can be approved by 1MO in a short time scale.
PUCILL, K F and VELSCHOU, S:
"RoRoPassenger Ferries Safety Studies Model Tests
for a Typical Ferry", Proc., RINA and DoT Int.
Symp. on the Safety of RoRo Passenger Ships,
RINA, London, April 1990, paper No. 7.
Dand, I. W: 'Experiments with a Flooded Model
of
a RoRo Passenger Ferry",
Proc., 2ndKummerman mt. Conf. on RoRo Safety
andVulnerability n The Way Ahead, RINA, London,
April 1991, paper No. 11.
[31 VELSCHOU. S and SCHINDLER. M: "RoRo
Passenger Ferry Damage Stability Studies
- A
Continuation of Model Tests for a Typical Ferry", Proc.,
The RINA Symp. on RoRo Ship's
Survivability - Phase 2. RINA, London, Nov. 1994, paper No. 5.DAND, I W: 'Factors Affecting the Capsize of
Damaged RoRo Vessels in Waves", Proc., The
RINA Symp. on RoRo Ship's Survivability* Phase
2, RINA, London, 25 November 1994, paper No.
3.
VASSALOS, D: "Capsizal Resistance Prediction of a Damaged Ship in a Random Sea", Proc., The RINA Symp. on RoRo Ship's Survivability - Phase 2, RINA, London, Nov. 1994, paper No. 2; also in: Trans. RINA, Vol. 138, 1995.
[61
VASSALOS, D and TURAN, O:
"A RealisticApproach to Assessing the Damage Survivability
of Passenger Ships", Trans. SNAME, Vol. 102,
1994.
VASSALOS, D and TURAN. O: "Development of
Survival Criteria for RoRo Passenger Ships - A
Theoretical Approach", Final Report on the RoRo Damage Stability Programme. Phase II,
Department of Ship and Marine Technology,
University of Strathciyde, December 1992. VASSALOS, D, PAWLOWSKI, M and TURAN, O:"A Theoretical
Investigation on the Capsizal
Resistance of Passenger/RoRo Vessels and
Proposal of Survival Criteria", Final Report, Task 5. The North West European R&D Project, March1996.
INTERNATIONAL MARITIME ORGANISATION
(1MO): "Regulation on Subdivision and Stability o"
Passenger Ships (as an Equivalent to Part B of
Chacter II
of the 1974 SOLAS
Convention)",waters and in fog, typically associated with calm
weather. lt is understandable that
in such
6. ACKNOWLEDGEMENTScircumstances sea states are on the wnole lower
than at the
opensea or
under normal operating Thefinancial support of the UK Department of Transportconditions.
As the critical sea state H
is a
and of the Joint R&D Project is gratefully acknowledged.function of h,
the factor s can be directly
We should also like to record our appreciation to ail our
expressed as a composite function of the
colleagues in the Marine Safety Agency and
in themeasure of damage stability, h Project for their help, contribution and support in more ways than one,
s = F [H (h)] (13)
[10]
1MO, London, 1974, 114 pp. This publication
contains IMO resolutions A.265
(VIII), A.266 (VIII), and explanatory notes.61RO, H, and BROWNE, R P: "Damage Stability
Model Experiments", Trans. RINA, Vol.
116.1974, pp. 69-91; also in:
The N.
Architect, October 1974, ibid.MIDOLETON, E H, and NUMATA, E: "Tests of a
Damaged Stability Model in Waves", SNAME
Spring Meeting, April
1970, Washington DC,paper No. 7.
i 2] PAWLOWSKI, M: "Bezpieczenstwo Niezatapiainosciowe Statkow (Safety of Ships n the Damaged Condition)", OSc thesis, Zeszyti
Naukowe "Sudownictwo Okretowe", No. 392142. Journal of Technical University of Gdansk, 1985.
SAFER-EURORO: "An Integrated Approach to
Safer European RoRo Ferry Design",
BRITE-EURAM Clustered Research Programme proposai, April 1996.HUTCHISON, B L: "Water On-Deck Accumulation
Studies by the SNAME Ad Hoc RoRo Safety
Panel", Workshop on Numerical and Physical
Simulation of Ship Capsize in Heavy Seas, Ross Priory, 24-25 July 1995, University of Strathclyde, Glasgow, UK.
=
> o 1 2 5 Intact GM (m) O oFreeboard= 0.50 m
2 305
1.5 2 25 3 ntact GM (ni)Figure 2: SHIW2
- SIDE CAS[NQSI-
Comparison between theoretical and
experimental results
IMARINTEK- CAPSIZE DMARINTE(-NOCAPSIZE ASTRATHCI.YDE -CAPS STRATh1CLY0E- NO CAPSIZE GOMI-NO CAPS OMI-CAPSIZE R oI
G R D Intct GM (ni)Freeboard= 1.0 m
8 7 R IMARINTE<- CAPSIZE6
I
DMAIEK -NO CAPSIZEE o ASATdCLYOE - CAPSIZE O, o ASTRATHCI.YDE - NO CAPSIZE G 2 40Ml- CAPSZE 00Ml-NO CAPSIZE
Freeboard =1.5 m
O 8I
IMAi1NTEX- CAPSZE 7u
D DMARIEX -NO CAPSIZE
R
=4
ASTRAT1-CLYDE -CAPSIZE 2 STRATHCL'YDE - NO CAPSIZE O 4 5 6 5C C o C C o 6000 5000 4000 3000 2000 1000 o -1000 4000 3000 2000 1000 O -1000
Sea State : 3.Om
WATER ON VEHICLE DECK
TIME (nec)
WATER ON VEHICLE DECK
TIME (nec) KG: 12.5m 5000-4000 3000 o 2000 -1000 0
Freeboard : 1.5m Sea State : 2.75m
WATER ON VEHICLE DECK
TIME (nec)
WATER ON VEHICLE DECK
TIME (eec)
KG: 13.Om
Figure 3:
Evaluation of critical amount of water on deck at "the point of no-relulIl''
(SI lIP 2, Open Deck, I)eck Area = 3,000
m2) Freeboard : 1.5m
i )l($)p
$4 -_100 150200_250
3_
g 10Q 2003qQ.4pQ
.600 10000 6000WATER ON VEHICLE DECK
6000
WATER ON VEHICLE DECK
C C 6000 o 4000 o 6000 C C 4000 2000 O -2000 2000 TIME (nec) 100 2 0 400 TIME (nec) Sea State : 4.5m KG : 9.Om
Freeboard : 1.5m Sea State : 4.25m
KG 9,5m
Freeboard : t5m
Sea Stato : 4.75m
KG: 105m
Freeboard 1.5m Sea Stato : 3.5m
KG : 11.5m Freeboard : 1.5m C 2500 2000 1500 1000 500 o -500
co;
2 6cC 02cm
iaco1J
14cc 12ccic
sco
4m
20C o 8.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 -9.0 10.0Freeboard 1.5m
KG (m)Freeboard 1.Om
11.0K(rn)
Freeboard O.5m
140 Dynamic StaticFiure 5:
Critical amount of water on deck for SHIP 2 with operi deck - comparison
between numerical simulation and static equivalent method
12.0 13.0 14.0
Dynamic
Static
Dynamic
0.12 0.10 0.08 0.06 0.04 0.02 0.00 h/Hsr 0.14
Figure 6: Boundary survivability curve for SBIP2 (total 85 cases)
F = 1.5 m, midship and forv.'ard damage, scales: 1 and 1.5
F = 1.0 m, midship and forward damage, scales: 1, 1.5 and 2/3
F = 0.5 m, midship damage, scales: 1, 1.5 and 2/3
Figure 7: Boundary survivability
curve for SI-11F i (all 36 cases)
13 VHs r
o F= 1.5m
F 1.0 mF=0.5
m
D DOiOo1
D D D oo
g
o,Ç!
D o Q O 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.10 0.20 0.30 0.40 0i4 0.12 0.10 0.08 0.06 0.04 0.02 0.00 F = 1.06m
F = 0.56 m F= 0,22 m VHs r1.2
-1.0 0.8 f0.6
0.4
-02
0.0 1.0 0.8 0.6 0.4 0.2 0.0 h O 2 4 6 8 10 Co F1.5m
F=1.Om
F0.5m
-Trend
xD
MARNTEK 12 0 2 4 6 8 10 12Figure 8: Boundary survivability curve for SHIP 2 (total 90 + 11 cases)
Minimum mailtude of water elevation h versus modifed wave height H57.
F = 1.5 m, midship and forward damage, scales: 1, 1.5 and 2/3
F = 1.0 m, midship and forward damage, scales: 1, 1.5 and 2/3
F = 0.5 m, midship damage, scales: 1, 1.5 and 2/3
h
o F=1.06m
F=0.56m
F0.22m
-Trend
Hsr
Fi2ure 9: Boundary survivability curve for SHIP 1 (all 36 cases)
H57-1.0 0.9 0.8 o C, C o 0.5 u,
ö0.4
0.31.0
0.9 0.8 0.7 0.6 0.5 0.4 0,3 O L"0.6
I C I j 0.1 0.2 0.3Sgnñcnt wave height H, metres
1EJetion of water h, meters
0.4 0.5 0.6
Fin-ure 11: The probability of collision survivaJ
- the survival factor s as a fiinction of
water on deck elevation h
15
0.0 1.0 2.0 3.0 4.0