RINA91, International Conference on Report 0889-P, September 1991,

Ro-Ro Safety and Vulnerability the Way Delft University of Technology,

Ahead, Heathrow Penta Hotel, London, Ship Hydromechanics Laboratory,
17 - 19 April 1991, Vol. I. Mekelweg 2, 2628 CD Delft,
The Netherlands.
Reprinted: 22-03-2001
Website: www.shipmotions.nl

Roll Motions of Ships due to Sudden Water Ingress,

Calculations and Experiments
A.W. Vredeveldt (TNO) and J.M.J. Journe (DUT)

SUMMARY

Roll-on Roll-off vessels appear to be sensitive to rapid capsizing due to sudden ingress of
water. Rapid capsizing is caused by a drastic decrease of static stability properties due to
free surfaces, as well as by inertia effects with regard to the roll motion. In this report it is
shown that the dynamic behaviour of the ship due to sudden ingress of water cannot be
neglected. To this end a calculation model was developed which was verified by ingress
tests. For these tests a wing tank cross-duct configuration was chosen. The results are
presented in this report.

1. INTRODUCTION one exception, which refers to the design

of cross-ducts. When a cross-duct is
The safety of ships in general has to be applied a flow calculation is carried out
assessed by judging several aspects. The in order to establish the minimum
stability of both the intact ship and the required cross sectional area. Such
damaged ship is of major importance. calculations usually assume a zero heel
The past has shown the vulnerability of angle during the period of flooding.
Ro/Ro vessels with regard to stability;
see references [l], [2], [3] and [4]. The roll motion of a ship can be
described by a second order differential
The stability of ships is presently equation where the angle of roll is used
determined by applying quasi-static as the independent variable. On the basis
methods. Inertia effects with regard to of this equation, it is possible to define a
the (roll) motions of the ship are natural roll period TR. The moment of
neglected. Usually flow calculations are heel is determined by both the weight of
not carried out either, although there is water, which has flowed into the ship,

1
and the distance of the centre of gravity
to the axis of rotation. The flow of the The aim of the work reported here was
water into the ship can be described with to determine whether dynamic
a first order differential equation. From considerations are required while
this equation a time constant TF can be judging the damage stability of ships
determined. during water ingress. In order to reach
this goal a fairly straightforward
Figure 1 shows the solution of a second approach was used. The next paragraph
order differential equation with two describes this approach.
different right hand terms. The time
constants of the response and the load Then on ship motions as well as on flow
terms are indicated. From the figure, it off fluids is fairly well developed.
can be seen that, when a heeling moment Therefore it is feasible to develop a
is applied to the ship a dynamic calculation method which can describe
amplification is to be expected when this both the roll motion of a ship and the
moment increases from 0 to its flow of fluid. The method is to be
maximum value within the roll period verified with ingress experiments on a
TR. When the moment increases much simple floating structure in laboratory
slower - i.e. from 0 to its maximum conditions, including the effect of roll
value during several times TR - then a motions. Once the calculation method is
much smaller dynamic amplification can verified, a systematic parameter study
be expected. can be carried out on actual ships. On
the basis of the parameter study,
Table 1 shows an estimate of both conclusions may be drawn with regard to
constants for an arbitrary chosen ferry the need to include dynamic
and for a damage area with a probability considerations while judging damage
of occurrence of 50 % (reference [10]). stability. Figure 2 shows this philosophy
schematically. This paper covers the first
four tasks, up to and including the
verification.

Table 1 Estimate of Time Constants TF

(Water Ingress Through a Damage Area) and 3. THE CALCULATION METHOD
TR (Roll Period) for an Arbitrary Chosen
Case

3.1. EQUATION OF MOTION

From this table it can be seen that the
time constant related to the inflow of Both the vertical immersion and the trim
water may equal almost half of the roll due to the water ingress are determined
period. Therefore a dynamic quasi statically, by interpolation based
amplification is to be expected. This on the ships hydrostatic data. No
consideration was the main motive to dynamics are taken into account.
carry out this project.
The roll motion however is determined
2. PHILOSOPHY including dynamic effects. To this end
the following equation of motion with

2
one degree of freedom is used. Coupling motions as presented in Chapter 5 were
effects with sway and yaw motions are based on the latter approach.
neglected as yet.

( I xx + a 44 )&& + b44& + c44 = M k 3.3. DAMPING

Equation 1 Because of viscous effects, the

where: determination of the damping coefficient
is mainly based on empirical methods
I xx mass moment of inertia of the [7]. Generally, only a small percentage
may be determined by applying potential
dry ship, flow theory, whereas the larger
a 44 hydrodynamic mass moment of contribution originates from viscous
inertia due to the presence of effects. In the calculation model the
water, damping coefficient b44 is assumed to
&& angular acceleration of roll angle, show a linear relation with the rail
b44 roll damping coefficient, velocity. This assumption is acceptable
& angular velocity of roll motion, since most of the damping originates
from friction forces, which vary in a
c 44 righting mo ment coefficient,
quadratic way with the roll velocity.
angle of roll, Hence the following formula holds:
Mk inclining moment.
b44& = b44 & &
'

Sections 3.2, 3.3 and 3.4 give some

considerations with regard to the left- Equation 2
hand side terms of equation (1). Section
3.5 elaborates on the heeling moment. Here again a roll decay test will yield the
most reliable figures for damping. It
should be noted that the dependency of
3.2. SHIPS MOMENT OF INERTIA the damping coefficient b44 on the roll
velocity disturbs the linearity of the
The mass moment of inertia may be
determined, by extending the ships equation of motion.
mass calculation with a radius for each
mass item. Some approximate methods
3.4. RIGHTING MOMENT
are available as well [5]. The
hydrodynamic mass moment of inertia
may be determined in several ways. The righting moment may be determined
in a rather straightforward manner.
Some simple estimation methods are
available [8]. More elaborate methods
are based on two [61 and three- c 44 =
d
d
(
gGN sin )
dimensional potential flow methods.
However the most accurate way of Equation 3
determining ( I xx + a 44 ) is by performing
a roll decay test. The calculated rail
where:

3
 for both the water flow and the airflow
density of water, through each orifice *).
g acceleration of gravity, However, two adjustments are made:
displaced volume of ship, 1. The variation of the hydrostatic
water pressure to the height of the
GN metacentric above C.o.G. of ship, orifice is taken into account by
roll angle. subdividing the orifice into
horizontal strips. The flow
GN is usually available from contribution of each strip is
hydrostatic calculations. calculated separately followed by a
summation of the contributions of all
It should be noted that the righting strips.
moment coefficient c 44 (in fact a spring 2. In order to account for the pressure
loss through orifices, a pressure loss
constant) depends on the roll angle,
coefficient is introduced.
which disturbs the linearity of the
equation of motion.
Hence the water flow through each flow
strip is determined with the next
formula.
3.5. HEELING MOMENT

2 P
The heeling moment M k may be Qi = A
determined by a summation of the K
heeling moments caused by the weight Equation 5
of the fluid in each compartment.
where:
nc
M k = GViYi Qi volume of water flow,
i =1
P pressure loss through strip,
Equation 4 K pressure loss coefficient,
with: density of water,
A sectional area of flow strip.
Mk heeling moment,
The airflow through each strip is
density of water,
determined with a similar formula. In
g acceleration due to gravity, order to cater for the compressibility of
vi volume of water in each the air the density is not used any more.
damaged compartment i due to The air pressure in each compartment
water ingress, can be derived from the theorem of
yi heeling lover of water volume in Boyle - Gay Lussac. Thus the following
compartment i , formula is derived.
nc number of compartments.
P1 R T
Q1 = 2A
The volume in each compartment is P1 + P2 K
calculated by applying Bernoullis law
Equation 6

4
where: 4. the flow of water between the
compartments and the cross duct.
Q1 volume of airflow,
P1 pressure loss through strip, These aspects called for a fairly large
test model. Moreover the sensitivity of
R specific gas constant of air,
the pressure transducers also imposed
T temperature of air,
lower limits to the size of the test model.
P1 pressure at front of strip, However on the other hand there were
P2 pressure at rear of strip, limitations to the size of the model
K pressure loss coefficient, because of the size of the test basin.
A flow area of flow strip.

4.2. MODEL DESCRIPTION

4. MODEL TESTS
Since wing tank cross-duct
configurations are used frequently; this
4.1. INTRODUCTION configuration was also built into the test
model. The actual dimensions of the
As mentioned in Chapter 2, the tanks and the duct were scaled down
calculation model was verified by from an existing ferry, however some
experiments. These experiments were deviations had to be made because of
necessary mainly because of two practical reasons.
reasons: Figure 3 shows the duct configuration.
1. The calculation model needed to be The hull form was chosen rectangular,
verified. because no favourable resistance
2. Few data were available on pressure characteristics are required and because
loss coefficients. of costs. The actual dimensions are
dictated by the dimensions of the test
Since the trimming motion was expected basin i.e. the width of the towing tank.
to behave quasi statically, this motion Further details are shown in Table 2.
was eliminated by using a test model
which was symmetrical about the
amidships. In fact this assumption
requires an experimental justification as Length, L 3.000 m
well, however this was not carried out in Beam, B 2.100 m
this project. Depth, D 1.250 m
Draught , T 0.625 m
The dynamic behaviour of the test model Length of wing tank, l t 1.000 m
had to be ship-like with regard to the
following aspects: Breadth of wing tank, bt 0.400 m
1. the hydrostatic forces and moments, Height of wing tank, ht 1.250 m
2. the (hydrodynamic) mass moments Length of cross-duct, l d 1.400 m
of inertia and hydrodynamic
Breadth of cross-duct, bd 0.200 m
damping,
3. the flow of water into the damaged Height of cross-duct, h d 0.400 m
compartments and Height of CoG, KG 0.750 m

5
 0.800 m order to obtain a preliminary idea of the
Mass of the model error introduced by this assumption,
(displacement intact), Vi 3937.5 m3 special attention was paid to the flow of
Moment of inertia of dry water into the test model and through the
test model (dry), I xx **) 1970 kgm2 cross-duct.

The following parameters were varied:

size of damage orifice,
Table 2 Characteristics of the Test Pontoon height of orifices between wing tank
and duct, height of orifices in the
4.3 VARIABLE AND MEASURED duct,
PARAMETERS size of the air orifices in the top of
the wing tanks,
Since several parameters will affect the configuration of the cross-duct,
roll motion due to sudden water ingress, height of the centre of gravity, KG .
it was considered to be important to
carry out tests where some of these By carrying out calculations while
parameters were varied. Thus the results varying the parameter values, a
of such measurements could be used to validation of the calculation model could
check the results obtained by be accomplished. Table 3 shows a
calculations. Flow calculations through review of the actual parameter values.
cross ducts assume a stationary flow.
However flow phenomena due to sudden
water ingress are far from stationary. In

Table 3 Review of Parameter Variation (See Figure 4 as well)

Please note that the value of the intact water pressure at the bottom of each
displacement, mass moment of inertia, compartment,
damping and trim ware not varied. air pressure at the top of each
compartment and
The following parameters were water levels in each compartment.
measured:
angle of roll, The actual locations of the gauges are
shown in Figure 3.

6
 This applies in particular to:
mass moment of inertia I xx + a 44 ,
4.4 EQUIPMENT AND TEST SET UP
damping coefficient b44' ,
Data were recorded both in a digital way pressure loss coefficients K .
and an analogue way.
These properties were determined in the
Air pressures and water pressures were case where the independent variables
measured with pressure gauges, which were set as shown below:
were in contact with the top of the tank
via small tubes. Thus the gauges could centre of gravity above base, KG 0.747 m
area of damage orifice, Ai 0.126 m2
remain outside the tank. cross sectional area comp. 1 0.080 m2
cross sectional area comp. 2 0.055 m2
Water levels wore measured with two cross-sectional area comp. 3 0.080 m2
resistance wires per tank as is often used orifice area A, AA 0.027 m2
to measure wave heights at the bow orifice area B, AB 0.027 m2
orifice area C, AC 0.027 m2
during sea keeping experiments. orifice area D, AD 0.027 m2
Unfortunately these level gauges showed area of air hole 1, A1 0.001 m2
a non linear characteristic. area of air hole 2, A2 0.001 m2

The angle of roll was measured with an ( I xx + a 44 )

and b44 were determined
inclinometer, which was very accurate. from the results of a roll decay
experiment. The pressure loss
Water flows and velocities were not coefficients used were determined by
measured. They were derived from the trial and error in such a way that the
water the level measurements. From the calculated roll motion curve and the
level measurements, the increase of measured roll motion curve fitted best.
water volume in each tank could be The height of the centre of gravity, KG ,
determined. The difference between was determined by inclining
volume increases yielded the actual experiments. Once these properties were
water flow through an orifice.
established they were kept constant
throughout all other calculations
The time plots of the flow rates showed covering other damage orifices, other
a rather erratic character. Therefore a
smoothing routine was applied on the cross duct dimensions and another KG -
raw data of the flow rate before they value.
could be used to derive pressure loss
coefficients. Pressure data ware treated
in the same way. 5.1. ROLL MOTION

5. RESULTS OF MEASURE- Table 4 gives a review of some of the

MENTS AND CALCULATIONS characteristics of the roll versus time
curves.
Some of the input data - required to
carry out the calculations - were
determined experimentally.

7
 orifice height C: 0.225 m
orifice height D: 0.225 m
area airvent SB: 0.002513 m2
area airvent PS: 0.002513 m2

Some results are shown in Figure 10 and

Figure 11.

It was found that the pressure loss

coefficients depended upon the flow
speed. This dependency is not mentioned
in literature.

Table 5 presents both the values of the

lower asymptotes of the K - flow speed
curves and the values as used in the
calculations.
Table 4 Comparison of Calculated and
Measured Characteristics of Roll Versus
Time (M refers to measured and C refers to
calculated).

Figure 4 up to and including Figure 8

present some plots of both the calculated Table 5 Pressure Loss Coefficients, Values as
and the measured roll motions as a Used in Calculations (Air Loss Coefficient
function of time. Moreover Figure 9 from [9])
shows the influence of the area of the
damage orifice area on the maximum The values as used in the calculations,
roll angle. were determined by trial and error until
in the case of run 53 the calculated
maximum roll angle and the leveling
5.2. PRESSURE LOSS time agreed with the measurement.
COEFFICIENTS During all other calculations these
values were kept constant.
From the measurements, pressure loss
coefficients were determined as well.
This was carried out for run 53 only 6. CONCLUSION
where:

KG : 0.750 m 6.1. MATHEMATICAL MODEL

Di : 0.400 m
The mathematical description of the roll
duct height SB: 0.275 m
duct height at CL: 0.400 m and flow phenomena seems appropriate.
duct height PS: 0.275 m Figure 4 to Figure 8 show both
calculated and measured roll response
orifice height A: 0.225 m
orifice height B: 0.225 m curves, while Table 4 gives a review.

8
Please note that the coefficients, as used reported here these motions are
in the calculations were determined in neglected. Some justification may be
such a way that the roll motion of run 53 found in the reasonable agreement
was described best. Subsequent the between calculated and measured
coefficients were kept constant. results. However a systematic
verification should be carried out.
The choice of the pressure loss
6.2. PRESSURE LOSS coefficients is of importance for both the
COEFFICIENTS maximum roll angle and the time
required to level out. Unfortunately the
Pressure loss coefficients, as measured, measured coefficients are not in line
show a strong flow speed dependency. with literature. Moreover measurements
Current literature ([8], [9] and [10]) does show that these coefficients vary with
not mention this dependency. Probably flow speed, which is not reported in
the dependency is due to the fact that literature either. Therefore it is proposed
during the tests the flow of water is far to investigate this matter more
from stationary, which is a requirement extensively. It should be noted that only
for applying the concept of pressure loss one of twelve measuring runs was
coefficients. Moreover inertia effects in analysed with regard to this aspect due
the fluids are neglected. The actual to budget limits.
figures vary within a large range (0.6 -
40.0). The results of the measurements and the
calculations refer to a rectangular
pontoon. It is essential to use the
6.3. OVERSHOOT calculation method on actual ships. Once
this has been done final conclusions may
From the measurements as well as from be drawn with regard to dynamic effects.
the calculations it can be concluded that
there is a significant overshoot in the It seems useful to link the computer
initial roll angle just after the water program as described here with existing
ingress has started. This overshoot is programs on hydrostatic data and cross-
entirely due to inertia effects associated curves. It is also useful to extend the
with the roll motion. Thus it can be current program with subroutines, which
stated that the dynamics of the roll determine added moment of inertia and
motion cannot be neglected when damping or establish a link with
judging the damage stability of ships. A hydrodynamic programs, which can
cross duct does not seem to affect this calculate such parameters.
overshoot significantly (see Table 4).
*) Apart from the air flow calculations,
this approach is taken from the IMCO
6.4. FURTHER RESEARCH recommendations for the determination
of minimum cross-duct areas [101.
During discussions about this project, it
was brought up that sway and yaw
motions could be of significant **) Calculated with: I xx =
1
(
Vi B
2
+D
2
).
12
importance. In the calculation model as

9
 6. Journe, J.M.J., Seaway-Delft, User
ACKNOWLEDGEMENT Manual and Theoretical Background
of Release 3.0, Ship
The authors wish to record their Hydrodynamics Laboratory, Delft
appreciation of the work carried out by University of Technology, Report
staff members of the Shiphydro- No. 849, January 1990.
mechanics Laboratory of the Delft 7. lkeda Y. et al, Prediction Method
University of Technology. A special for Ship Rolling, Department of
word of thanks is addressed to Mr. E. Naval Architecture, University of
Vossnack (former head of design Osaka Prefecture, Japan, Report No.
department of Nedlloyd Fleet Services), 00405, 1978.
Mr. H. Vermeer (Netherlands 8. Blevins R.O., Applied Fluid
Directorate General of Shipping and Dynamics Handbook, von Nostrand
Maritime Affairs), Mr. D. Spanjer and Reinhold Company, New York,
Mr. J. Uwland (TNO). Finally, the 1984.
authors like to thank Mr. A. van Strien 9. Ireland N., Damage Stability Model
(Delft University of Technology), who Tests, Project No. 34620, British
has been a great help with the Maritime Technology, May 1988.
performance of the experiments. 10. IMCO, Explanatory Notes to the
Regulations on Subdivision and
Damage Stability of Passenger Ships
REFERENCES as Equivalent to Part B of Chapter 11
of the International Convention for
1. Spouge, J.R., The Technical Safety of life at Sea, 1960, ANNEX
Investigation of the Sinking of the II, STAB XV/11.
RO/RO Ferry EUROPEAN 11. Delft University of Technology,
GATEWAY, RINA Mar., No. 3, Numerical Analysis C1, Lecture
1986. Notes (in Dutch).
2. Boilwood, D.T., Ro/Ro Ship Sur-
vivability; Comments on Damage
Stability Modelling, Ro/Ro-88, NOMENCLATURE
Gothenburg, 7-9 June 1988.
3. Braund, N.A., Damage Stability; A flow area, m2
Research for the Future, Safe a 44 hydrodynamic moment of inertia, kgm2
Ship/Safe Cargo Conference, B beam, m
London, 1978. b44 hydrodynamic damping coeff., Nms/rad
4. Dand, I.W., Hydrodynamic Aspects '
b44 hydrodyn. damping factor, Nms 2 /rad2
of the Sinking of the Ferry HERALD
OF FREE ENTERPRISE, The bd breadth of duct, m
Naval Architect, May 1989. bt breadth of wing tank, m
5. Peach, et al, The Radii of Gyration c 44 spring coefficient, Nm/ rad
of Merchant Ships, North-East
CoG centre of gravity
Coast of Engineers and Shipbuilding D depth, m
Transactions, June 1987, Page 155 - P pressure loss, N/m2
117. Di diameter of damage orifice, m
g acceleration of gravity (9.81), m/s 2

10
GN metacentric height, m Q water flow, m3 /s
( varying with angle of heel) R gas constant, J/(kgK)
ht height of wing tank, m density of water, kg/m3
T draught, m
I xx ridgid moment of inertia of ship, kgm2
ui approximative value by Euler
K pressure loss coefficient, N/m2
integration
KG height of CoG above base, m V volume of displacement, m3
L length of ship, m
v 2i approximative value by Euler
ld length of duct, m
integration
lt length of tank, m vi approximative value by Euler
Mk heeling moment, Nm integration
nc number of compartments yi exact value
&& angular roll acceleration, rad/s 2 hd height of duct, m
& angular roll velocity, rad/s
roll angle, rad

Figure 1 Rough Estimates of Roll response to Sudden Water Ingress

A: TG = 10 s, B: TF = 40 s

11
Figure 2 Schematic Representation of the Philosophy

(* indicates tasks covered in this paper)

12
Figure 3 Cross-Section and Top View of Testmodel

13
 Figure 4 Roll Motion Run 43 (Decreased Duct Area)

Figure 5 Roll Motion Run 51 (Area of Vent. Holes Increased)

14
 Figure 6 Roll Motion Run 53 (Initial Configuration)

Figure 7 Roll Motion Run 59 (Damaged Orifice Decreased)

15
 Figure 8 Roll Motion Run 120 (Height of CoG Increased)

Figure 9 Maximum Roll Angle Versus Area of Damage Orifice

16
Figure 10 Pressure Loss Coefficient as Function of Flow Velocity, Damage Orifice

Figure 11 Pressure Loss Coefficient as Function of Flow Velocity, Orifice C

(see Figure 3)

17