Gourlay, T.P. & Lilienthal, T. 2002 Dynamic stability of ships in waves.

Proc. Pacific 2002 International Maritime Conference, Sydney, Jan 2002.

Dynamic Stability of Ships in Waves

Tim Gourlay & Tim Lilienthal
Australian Maritime College, Tasmania

ABSTRACT

A method is proposed for evaluating the overall dynamic stability of an intact vessel
in a seaway. We use an existing ship motions program to study the motion of a vessel
with a certain loading condition, speed and heading, in given wave conditions. A
deterministic method is discussed for looking at the stability of a vessel over a wide
range of these parameters. This is done with a view to giv ing operators advice on the
safest headings and speeds to adopt in extreme conditions, as well as gauging the
overall safety of a particular vessel. It is hoped that eventually such dynamic stability
analysis can be used to modify the present IMO stability criteria for ships.

INTRODUCTION

At present, safety regulations with regard to ship capsize are based primarily on static
stability concepts. The efficient calculation of GZ curves that is possible nowadays
permits quick and accurate determination of a ship’s static stability. Although it is
well accepted that capsizing of ships is a dynamic phenomenon, static stability
considerations are still used almost exclusively to gauge the propensity of a ship to
capsize in waves.

This means, for example, that some vessels are at risk of dynamic capsize despite
having favourable GZ curves, whereas others are perhaps being over-penalized due to
the nature of their GZ curve, despite having good dynamic stability.

Seakeeping is now a well-developed field, and basic techniques for simulating the
motion of a ship in both regular and irregular waves are well-known (see e.g. [1,2]).
Newton’s 2nd Law is used to predict the time rate of change of each of the six degrees
of freedom (roll, pitch, yaw, heave, surge, sway), using the nett forces and moments
acting on the vessel at each point in time. The hydrodynamic forces and moments are
caused by the accelerated motion of the vessel through the water, as well as the action
of the wave, and may involve significant cross-coupling. Still, if the time-varying
forces and moments on the vessel can be accurately modelled, the computational
solution of the resulting ship motions is straightforward.

Such time -domain simulations may be used to predict the “capsize” of a vessel, and
thereby used to assess ship safety. The capsize of a vessel is normally defined by a
predetermined roll angle, such as the angle at which downflooding occurs, or the
angle at which the GZ-curve becomes negative. By definition, capsize involves large
roll angles and hence highly nonlinear behaviour. Therefore, basic linear seakeeping
theory has given way to more accurate modelling of the hydrodynamic forces for
predicting capsize, involving significant cross-coupling and nonlinear terms (see e.g.
[3,4,5])
One problem with using dynamic analysis to gauge the stability of a vessel is the large
number of parameters involved. Static stability analysis can be condensed into a
single GZ-curve for each vessel and loading condition, whereas dynamic stability is
also a function of ship speed, heading angle and wave conditions. Despite continuing
improvements in the accuracy of ship motions programs, a reliable method for
condensing all the relevant data into an overall safety assessment remains an elusive
goal.

In this paper we discuss the use of both irregular and regular waves to assess dynamic
stability, and propose a method whereby the dynamic stability of ships in waves can
be compared through the complete spectrum of relevant parameters. For our results
the dynamic stability code FREDYN has been used (courtesy of CRNav and the
Australian Defence Force), although the methods are equally applicable to the use of
any dynamic stability code.

DYNAMIC ANALYSIS IN IRREGULAR WAVES

Clearly the ideal way to approach dynamic stability is to run the ship motion
simulation in an irregular seaway, which is as close as possible to the actual seaways
the ship is likely to encounter.

In irregular waves, a wave spectrum is chosen to correspond to a likely extreme sea

state for a given region, if the operating region of the ship is fixed, or otherwise a
representative extreme sea state. Therefore the variables are the type of spectrum
used, as well as the governing parameters for that spectrum, such as significant wave
height and zero-crossing period.

This method can produce two types of outcome: the average time taken to capsize, or
the probability of exceeding a certain roll angle in a specified time.

Time to capsize

For a given speed and heading, simulations are performed up until the ship’s roll
angle reaches the predetermined capsize angle. The time to capsize depends strongly
on the initial wave conditions; hence many runs must be performed in order to gain a
statistically significant “average time to capsize” for that particular seaway, speed and
heading.

Probability of exceeding a given roll angle

The other way to numerically model capsize is to perform simulations over a shorter
time, say one hour, and then fit a distribution to the probability of exceeding a given
roll angle in a specified time. This approach was pioneered by McTaggart ([6], see [7]
for a comprehensive overview). The extrapolation to large roll angles is done by
assuming a Gumbel distribution of the data, resulting in a “probability of capsize in a
specified time” for that particular seaway, speed and heading.

Although irregular wave analysis is considered the optimum method for modelling the
motions of a ship in a real seaway, its use in assessing ship stability has some serious
complications:
• If using the time to capsize, large simulation times are needed for each run, and
many runs (we use 50) are needed to find the average time to capsize. If using the
probability of exceeding a given roll angle, the results only loosely follow a
Gumbel distribution, and the extrapolation to larger roll angles remains
questionable. Even with the extrapolation the computations are very lengthy.
• The question remains what constitutes an acceptable time to capsize, or acceptable
probability of capsize for a given situation.
• Validation of the dynamic stability codes with model experiments is very difficult
for irregular waves.
• The physics of the capsize phenomenon can be difficult to understand in irregular
waves.

Note that at this stage only long-crested irregular waves can be accurately modelled,
which are themselves an approximation to the short-crested waves in a true seaway.

DYNAMIC ANALYSIS IN REGULAR WAVES

It has been shown (e.g. [8]) that ship behaviour leading to capsize in irregular waves
is similar to that in the dominant regular waves. Therefore, a simplified approach is to
conduct stability assessment in regular waves, which are determined by a wave height
and length. In order to represent an extreme sea state, a characteristic dominant wave
height and length can be chosen, or a wide range of component waves can be
considered individually.

This approach is deterministic rather than probabilistic, in that each simulation results
in a definitive “capsize” or “no capsize”. If the ship has not capsized in a moderate
amount of time in the regular sea, it will never capsize, so only relatively short runs
are needed.

Because of the comparative simplicity of a regular sea analysis, another relevant

parameter can be brought into the analysis: the vertical centre of gra vity height (or
KG value – the height of the centre of gravity above the keel line). Running the
simulations over a range of possible KG values gives an understanding of how
different loading conditions affect the stability of the vessel, and also allows a
“limiting KG” value to be determined, above which the ship will capsize in the given
seaway [9]. This limiting KG can then be compared to the actual KG of the ship, as a
measure of its dynamic stability.
 7

6
KG (m)

3
0 10 20 30 40 50 60 70 80 90 100

Maximum roll angle (°)

Figure 1: Maximum roll angle in regular waves, with increasing KG

An example of this regular wave analysis is given in figures 1 and 2, which show the
maximum roll angle reached in regular waves of height h=15.5 metres and length
λ=217 metres for a certain ship. In figure 1 a wide range of KG values are plotted,
and we see that the dependence on loading is complicated. In this case capsize occurs
in bands of KG corresponding to resonance situations. These “capsize bands” were
not present over all headings and speeds, but they do provide a complication to the
concept of limiting KG values.

7.0

simulation with 800s ramp simulation with 3500s ramp

6.5
KG (m)

6.0

5.5
0 10 20 30 40 50 60 70 80 90 100

Maximum roll angle (°)

Figure 2: Maximum roll angle with different ramp times.

In figure 2 we see the effect of the ramp time, which is the time from when the
program starts in calm water until the target wave height is reached. We notice that a
long ramp time is required so that the ship does not capsize artificially on the ramp.
Subtleties such as this and the capsize bands tend to be obscured in an irregular wave
analysis.

A PROPOSED IRREGULAR / REGULAR ANALYSIS

As we have discussed, both irregular and regular waves have their advantages and
disadvantages for assessing dynamic stability. We are currently developing a method
that we hope will combine the strengths of each of the methods.

Correlation between irregular and regular waves

The approach is to correlate the capsize risk of a ship in irregular waves with its
propensity to capsize in regular waves. To do this we consider a spectrum of regular
waves (i.e. the full range of wavelengths) which aims to model the most dangerous
components of the irregular spectrum.

The correlation between irregular and regular waves results in a certain wave height
for each wavelength, to use in the regular analysis. An example of this is given in
figure 3.

30

25

20
Wave height (m)

15

10

0
0 100 200 300 400 500 600 700

Wave length (m)

Figure 3: Wave height h as a function of wavelength λ for regular wave analysis

The curve of wave height versus length is designed to model the most dangerous
elements of an irregular sea, by choosing the larger wave heights that might occur at
each wavelength. Even in an irregular sea the wave steepness h/λ generally has an
upper limit, and it is this limit that we are modelling in a broad sense. All smaller
wave steepness values do not represent the largest wave height for that particular
wavelength, and therefore are not as dangerous as the chosen curve, so need not be
input.

In order to correlate the results with irregular waves, we choose a constant wave
steepness h/λ for the regular waves. We must also ca p the wave height at a realistic
value; this might be for example H1 / 100 , the average height of the highest 1% of
waves in the chosen irregular spectrum.

The regular wave simulation can now be run over the entire range of wavelengths and
corresponding wave heights to find the maximum roll angle reached by the vessel in
each case. Again, we go one step further and consider a large range of loading
conditions, in order to determine the limiting KG value both at each wavelength and
over the whole spectrum.

We must now correlate the boundary between capsize and no capsize in regular waves
(which we measure in terms of the limiting KG value), with an “acceptable risk” or
“acceptable time to capsize” in irregular waves. At present we nominally use a 10-
hour average time to capsize, which gives us our corresponding limiting KG value in
the irregular sea (see figure 4).

16

14

12
Average Time (hr)

10

KG (m)

Figure 4: Time to capsize in irregular seas (average of 50 simulations,

H 1 / 3 = 11 .5 metres, T1 = 11.5 seconds)

In this way, correlating the limiting KG values in the regular and irregular sea allows
the regular wave steepness to be determined, which will ensure that our regular wave
results correlate with the irregular wave results.

Preliminary calculations suggest that the wave steepness values obtained are close to
those observed in an irregular sea, around 1/14, and that performing the correlations
over different cases will yield a sensible average to use.
Regular wav e simulations

Having chosen this wave steepness, dynamic stability assessments can be performed
with far less computational effort in regular seas, with subtleties such as the effect on
capsize of loading, speed and heading readily understood.

Figure 5 s hows an example of limiting KG values (marking the boundary between

capsize/no capsize of the ship) for a range of wavelength ratios (λ/L, where L is the
shiplength) and heading angles (away from astern seas). We see that for this case the
longer wavelengths, with larger wave heights, tend to be the most dangerous. It
should be noted that this is not true for all headings and speeds; often the most
dangerous waves are the smaller ones whose length is close to the length of the ship.

We also notice that the most dangerous heading angle is around 50 degrees (away
from astern seas) for most wavelengths at this speed.

KG (m)

0.60

1.00

Wave length ratio 1.40

80 90
1.80 60 70
40 50
20 30
10 Heading angle (°)

Figure 5: Limiting KG contours as a function of wavelength and heading angle for a

given speed

We can now define the overall limiting KG value for each heading and speed by
picking off the smallest limiting KG value over all wavelengths. Doing this allows a
polar plot to be generated of limiting KG values as a function of speed and heading
angle.
Using the results

Areas of small limiting KG on the polar plot are the most dangerous heading/speed
combinations, while areas of high limiting KG are the safest. This information can be
passed onto ship captains - which headings and speeds are best to adopt or avoid in
extreme seas.

These limiting KG values can also be used to gain an overall limiting KG value that is
deemed safe for the survival of the ship in these extreme conditions. For example, this
might be the smallest limiting KG on the whole polar plot (so that the vessel should
be able to survive all headings and speeds), or it may not include heading/speed
combinations which are known to be dangerous and would be avoided by ship
captains.

Finally, the chosen overall limiting KG value can be compared with current IMO
criteria to gauge the effect of wave dynamics on the vessel’s stability. Comparing this
between different vessels shows which are at risk of dynamic capsize despite having
favourable GZ curves, or which are being over-penalized due to the nature of their GZ
curve, despite having good dynamic stability. This may be used to modify IMO
stability requirements in future.

CONCLUSIONS

We have outlined a method for assessing the overall dynamic stability of a ship in
waves. The method is based on regular wave simulations, over a range of wavelengths
whose heights are chosen to correlate with an irregular sea. The advantages of this
method over the full probabilistic method are greatly decreased computing time, a
simplified analysis and increased understanding of the essential capsize phenomena.

Acknowledgement

The authors wish to acknowledge the help of the Australian Defence Force in
providing funding for this research.

REFERENCES

1. Lloyd, A.R.J.M., “Seakeeping: ship behaviour in rough weather”, 1989,

Chichester: Ellis Horwood.
2. Bhattacharya, R., “Dynamics of Marine Vehicles”, 1972, Ed. M.E.
McCormick, Wiley.
3. Hamamoto, M. & Munif, A., 1998, “A mathematical model to describe ship
motion leading to capsize in waves”, Journal of The Society of Naval
Architects of Japan, Vol. 184.
4. Umeda, N., 1999, “Nonlinear dynamics of ship capsize due to broaching in
following and quartering seas”, Journal of Marine Science and Technology,
Vol. 4, No. 1., pp. 16-26.
5. Spyrou, K.J., 1996, “Dynamic instability in following seas: The behaviour of a
ship during broaching”, Journal of Ship Research, SNAME, Vol. 40, No. 1,
pp. 46-59.
6. McTaggart, K.A., 1999, “Ship capsize risk in a seaway using time domain
simulations and fitted Gumbel distributions”, Proceedings of the 18th
International Conference on Offshore Mechanics and Arctic Engineering, St.
John’s, Newfoundland, July 1999.
7. McTaggart, K.A, & de Kat, J.O., 2000, “Capsize risk of intact frigates in
irregular seas”, SNAME Transactions, Vol. 108, pp. 147-177.
8. Takaishi, Y., 1982, “Consideration on the dangerous situations leading to
capsize of ships in waves”, Proceedings of the 2nd International Conference on
Stability of Ships and Ocean Vehicles, Tokyo, Japan, October 1982.
9. Renilson, M.R., “Assessment of ship stability using dynamics” , Proceedings
of the 20th International Conference on Offshore Mechanics and Arctic
Engineering, Rio de Janeiro, June 2001.