Professional Documents
Culture Documents
STII Ship Dynamics in Waves
STII Ship Dynamics in Waves
Rostock
2012
2
Contents
3
4 Ship oscillations in small head waves 57
4.1 Exciting forces and ship oscillations . . . . . . . . . . . . . . . 57
4.2 Estimations of slamming and deck flooding . . . . . . . . . . . 61
4
List of Tables
5
6
List of Figures
3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2 Illustration of hydrostatic force . . . . . . . . . . . . . . . . . 39
3.3 Illustration of hydrostatic moment . . . . . . . . . . . . . . . . 39
3.4 Ship as linear system . . . . . . . . . . . . . . . . . . . . . . . 46
3.5 Response function versus referred frequency . . . . . . . . . . 47
3.6 Phase displacement versus referred frequency . . . . . . . . . . 48
3.7 Ship oscillations in resonance case . . . . . . . . . . . . . . . . 48
3.8 Oscillation of a raft with a big metacentric height . . . . . . . 48
3.9 Illustration of the frame in beam waves . . . . . . . . . . . . . 49
3.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.11 Reduction coefficient of the heave oscillations . . . . . . . . . 53
3.12 Sea classification . . . . . . . . . . . . . . . . . . . . . . . . . 54
7
4.2 Position of ship at different time instants in a head wave . . . 61
4.3 Curves y = ±zmax and y = z(x) . . . . . . . . . . . . . . . . . 62
4.4 Sample for a real ship . . . . . . . . . . . . . . . . . . . . . . . 63
8
8.9 Offshore Patrol Vessel [1] . . . . . . . . . . . . . . . . . . . . . 94
8.10 Passenger Ship [1] . . . . . . . . . . . . . . . . . . . . . . . . . 94
8.11 Reduction of the roll amplitude using the Flume tank designed
by Hoppe Marine (taken from [1]). . . . . . . . . . . . . . . . 95
8.12 Active stabilizing system . . . . . . . . . . . . . . . . . . . . . 97
8.13 Bilge keel [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
8.14 Main parameters of bilge keels. . . . . . . . . . . . . . . . . . 98
8.15 Schematic representation of active fins. . . . . . . . . . . . . . 99
8.16 Active rudder roll. . . . . . . . . . . . . . . . . . . . . . . . . 99
8.17 Schematic representation of the pitch stabilizing system . . . . 100
8.18 Reduction of pitch and heave oscillations using the stabilizing
system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9
10
Chapter 1
• hydrostatic coupling
• hydrodynamic coupling
• gyroscopics coupling
Hydrodynamic coupling can be illustrated when the ship is moving with ac-
celeration in transverse direction (sway motion). Since the ship is asymmetric
with respect to the midships, such a motion is conducted with appearance
11
of the yaw moment. Therefore, the sway and yaw oscillations are hydrody-
namically coupled.
In this chapter we consider the ship oscillations with small amplitude. For
such oscillations the coupling mentioned above can be neglected.
12
Figure 1.2: Displacement of the center of effort due to change of the ship
draught (from [3])
the theoretical basis for the determination of wave induced forces, since the
most contribution to these forces is caused by processes properly described
by inviscid flow models.
6
X
ϕ(x, y, z, t) = Uj ϕj (x, y, z) + AϕA (x, y, z) cos ωt =
j=1
6
!
X (1.4)
= ζj0 ωϕj (x, y, z) + AϕA (x, y, z) cos ωt
j=1
where
13
• ϕj (x, y, z) is the velocity potential of the ship oscillation in j-th motion
with the unit amplitude ζj0 = 1 in the absence of incident waves,
• ϕA (x, y, z) is the potential taking the incident waves and their interac-
The potentials must satisfy the Laplace equation ∆ϕj = 0, ∆ϕA = 0 and
appropriate boundary conditions. The boundary conditions to be imposed
on the ship surface are the no penetration conditions (see also formulae 3.18
in the Chapter 3 [7]):
where n is the normal vector to the ship surface, directed into the body,
(x, y, z) are the coordinates of a point on the ship surface. The r.h.s. of the
conditions (1.5) is the normal components of the ship local velocities caused
by particular oscillating motions.
ϕA = ϕ∞ + ϕp (1.7)
14
which ϕ∞ is the potential of incident waves not perturbed by the ship pres-
ence and ϕp is the perturbation potential describing the interaction between
the incident waves and the ship. The potential of regular waves ϕ∞ is known
(see Chapter 6 in [7]). The boundary condition for ϕp on the ship surface is
∂ϕp ∂ϕ∞
=− (1.8)
∂n ∂n
Away from the ship the radiation potentials ϕj and the diffraction perturba-
tion potential ϕp decay, i.e. ϕp −−−→ 0, ϕj −−−→ 0.
r→∞ r→∞
On the free surface the linearized mixed boundary condition (see formula (6.17)
in [7]) reads
∂ 2ϕ ∂ϕ
2
+g = 0 on z = 0. (1.9)
∂t ∂z
Substituting (1.4) in (1.9) yields for ϕj (x, y, z) and ϕA (x, y, z):
ω2 ∂ϕj
− ϕj + = 0 on z = 0.
g ∂z
(1.10)
ω2 ∂ϕA
− ϕA + = 0 on z = 0.
L
g ∂z
The force and the moment on the ship are determined by integrating the
pressure over the wetted ship surface. The pressure can be found from the
Bernoulli equation written in the general form:
ρu2 ∂ϕ
p+ + ρgz + ρ = C(t) (1.11)
2 ∂t
Here the potential is the potential of the perturbed motion. The constant C(t)
which is the same for the whole flow domain is calculated from the condition
that the pressure on the free surface far from the ship is constant and equal
to the atmospheric pressure:
pa = C(t) (1.12)
15
Substituting (1.12) in (1.11) gives:
ρu2 ∂ϕ
p − pa = − − ρgz − ρ (1.12a)
2 ∂t
Remembering that the ship speed is zero and perturbation velocities as well
as the velocities caused by incident waves are small we neglect the first term
in (1.12a):
∂ϕ
p − pa = −ρ + gz (1.13)
∂t
Together with (1.4) it gives
6
!
X
p − pa = ρ ζj0 ωϕj (x, y, z) + AϕA (x, y, z) ω sin ωt − ρgz (1.14)
j=1
The forces and the moment are then calculated by integration of p − pa over
the wetted ship area
Z Z
~ ~
F = (p − pa)~ndS, M = (p − pa)(~r × ~n)dS. (1.15)
S S
The normal vector direction in (1.15) is into the body. The vertical ordi-
nate z of any point on the wetted area can be represented as the difference
between the submergence under unperturbed free surface ζ and free surface
elevation ζ0 . Substituting (1.14) in (1.15) one obtains
Z Z
F~ = −ρg ~nζdS + ρg ~nζ0 dS
S S
6
X Z
+ρ ζj0 ω 2 sin ωt ~nϕj dS+ (1.16)
j=1 S
Z
+ ρ(Aω sin ωt) ~n(ϕ∞ + ϕp )dS
S
16
• the hydrostatic component (the first term) acting on the ship oscillating
on the unperturbed free surface (in calm water),
• the damping and the added mass component (the third term) and
The third term describes the force acting on the ship oscillating in calm water.
The last term arises due to incident waves acting on the ship. Within the
linear theory keeping only the terms proportional to the amplitude A and
L
neglecting small terms of higher orders proportional to ∼ An , n > 1 one can
L
show that the integration in the last term can be done over the wetted area
corresponding to the equilibrium state. Thus, the last term describes the
force induced by waves on the ship at rest.
6
X Z
F~2 = ρ ζj0 ω 2 sin ωt ~nϕj dS (1.17)
j=1 S
6 Z 6
X X dUj
F2i = ρ ζj0 ω 2 sin ωt ni ϕj dS = − cji (1.18)
j=1 j=1
dt
S
1
cji = µji − λji
ω
The term − ω1 λji has been introduced to take the fact into account that the
force due to influence of the free surface depends not only on ω 2 but also
on ω. The force is then
17
6 6 6
X 1 dUj X X π
F2i = − µji − λji =− µji aj − λji Uj ts = t −
j=1
ω dt j=1 j=1
2ω
(1.19)
As seen from (1.19) the first component of the force is proportional to the
acceleration aj whereas the second one is proportional to the velocity Uj .
The first component is called the added mass component, whereas the sec-
ond one- the damping component.
Using the Green’s theorem, Haskind derived the following symmetry condi-
tions for the case zero forward speed:
(
µji = µij
cji = cij ⇒
λji = λij
Similarly to the added mass one can introduce the damping coefficients. The
full mechanic energy in the progressive wave with the amplitude A is (see
chapter 6.4 in [7])
1
E = ρgA2 (1.20)
2
per wave length.
The energy transported by waves through sides 1 and 2 (see Fig. 1.3) per
time unit is
1
δE = 2 ρgA2 U (1.21)
2
18
Figure 1.3: Illustration to derivation of damping coefficient
where λij is the coefficient of damping in i-th direction when the ship oscil-
lates in j-th motion. Uj2 is the time averaged square of the ship oscillations
2
(ωζ 0 )
speed. Obviously, Uj2 = 2j and
2
ωζj0
δE = λij (1.23)
2
The group velocity (see formulae (6.39) and (6.40) in [7]):
r
c 1 g g
U= = = (1.24)
2 2 k 2ω
since kg = ω 2 (see formula (6.21) in [7]). Equating (1.21) and (1.23) one
obtains with account for (1.24)
2
ωζj0
ρgA2 U = λij
2
⇓ (1.25)
0 2 2
ωζj ρg 2
g A
ρgA2 = λij ⇒ λij =
2ω 2 ω3 ζj0
19
The damping coefficient λij depends on the square of the ratio of the wave
amplitude to the ship oscillation amplitude causing the wave.
The damping coefficients of different frames are shown in Fig. 1.4, 1.5, 1.6, 1.7
and 1.8 taken from [3]. Solid lines show results obtained from the potential
theory. Generally, the results show the applicability of the potential theory
for calculation of damping coefficients. The accuracy of prediction is not
satisfactory for the box B/T=8 in heave and B/T=2 in sway because of the
flow separation at corners which has a sufficient impact on hydrodynamics
in these two cases. The agreement for λ44 is not satisfactory (see Fig. 1.8)
because of dominating role of the viscosity for this type of damping. For the
semi circle frame the damping coefficient in roll is zero λ44 = 0 within the
inviscid theory. One hundred per cent of the roll damping is due to viscosity.
Usually λ44 are determined using viscous flow models. It is remarkable,
that the damping coefficients depend on the frequency and amplitude (see
Fig. 1.8).
Figure 1.4: Added mass and damping coefficient of the semi circle frame at
heave oscillations. Here A is the frame area.
20
Figure 1.5: Added mass and damping coefficient of the box frame at heave
oscillations. Here A is the frame area.
Figure 1.6: Added mass and damping coefficient of the semi circle frame at
sway oscillations. Here A is the frame area.
Figure 1.7: Added mass and damping coefficient of the box frame at sway
oscillations. Here A is the frame area.
21
Figure 1.8: Added mass and damping coefficient of the box frame at roll
(heel) oscillations. Here A is the frame area.
The difference with the case of the motion in unlimited space is the presence
of the free surface. The added mass µij have to be calculated with account
for the free surface effect. For their determination the panel methods can be
used. The problem is sufficiently simplified in two limiting cases ω → 0 and
ω → ∞. The boundary condition (1.10) can be written in the form:
∂ϕj
∂z
= 0 for ω → 0,
on z = 0. (1.27)
ϕj = 0 for ω → ∞.
The conventional mirroring method can be used for the case ω → 0 (Fig. 1.9).
The mirroring frame is moving in the same direction for surge, sway and yaw.
For the heave, roll and pitch the fictitious frame is moving in the opposite di-
rection. At the free surface, these tricks make the normal components of the
∂ϕ
total velocity induced by the actual and the fictitious frames zero, i.e. ∂zj = 0
on z = 0.
22
Figure 1.9: Mirroring for the case ω → 0
by fictitious frames.
The added mass of slender body can be found by integration of added mass
of ship frames along the ship length
The added mass of different frames are shown in Fig. 1.4, 1.5, 1.6, 1.7 and 1.8
taken from [3]. Like in case of damping coefficients the results of the potential
theory are not acceptable for roll added mass because of dominating role of
the viscosity. As seen from Fig. 1.4 - Fig. 1.8 the added mass depend on the
frequency ω.
23
1.4 Hydrostatic component
Let us the ship is in the equilibrium state. The ship weight is counterbalanced
by the hydrostatic lift. Due to small heave motion the equilibrium is violated
and an additional hydrostatic force appears. The vertical component of this
additional hydrostatic force can be calculated analytically from (1.16) for the
case of small heave motion
Z Z
∆Fζ = −ρg cos(nz)zdS − −ρg cos(nz)zdS = −ρgAW P ζ
S T +ζ S T
(1.29)
where ζ is the increment of the ship draught, AW P is the waterplane area
and T is the ship draught in the equilibrium state. The roll and pitch hy-
drostatic moments for small change of the roll and pitch angles are
24
R
waves (the second term in (1.16)). The second component ρωA sin ωt ~nϕP dS
S
takes the diffraction effect (the contribution of the scattering potential ϕp to
pressure distribution) into account. As shown by Peters and Stokes the
Froude Krylov force is a dominating part of the wave induced forces for os-
cillations of slender ships in directions j=1 (surge), 3 (heave) and 5 (pitch).
The weight is not present in the second equation of the system (1.32) be-
cause it is counterbalanced by the hydrostatic force at rest. The additional
hydrostatic force −ρgAW P ζ is the difference between the weight and the full
hydrostatic force. The system (1.32) is written in the principle axes coordi-
nate system [6].
The Green’s formula for two functions Φ and Ψ satisfying the Laplace equa-
tion is
25
Z Z
∂Φ ∂Ψ
Ψ −Φ dS = 0, (1.33)
∂n ∂n
Sw
where Sw is the flow boundary (wetted ship surface plus the area away from
the ship, see the sample in Chapter/Section 3.2). Particularly, the rela-
tion (1.33) can be applied to radiation potentials ϕj . Since the potential ϕp
satisfies the Laplace equation and the same boundary conditions as the ra-
diation potentials ϕj , the Green’s formula (1.33) can also be applied to ϕj
and ϕp
Z Z
∂ϕj ∂ϕp
ϕp − ϕj dS = 0 (1.34)
∂n ∂n
Sw
~ = ρω
R
where X ~n (ϕ∞ + ϕp ) dS. Taking (1.5), (3.34) and (1.8) into account
S
we get
Z Z
∂ϕj
Xj = ρω (ϕ∞ + ϕp ) dS (1.35)
∂n
Sw
∂ϕj R R ∂ϕj
ϕj ∂ϕ + ϕj ∂ϕ
R R R R p p
ϕp ∂n
dS = ∂n
dS ⇒ Xj = ρω ϕ∞ ∂n ∂n
dS
Sw Sw Sw
∂ϕp
∂n
= − ∂ϕ
∂n
∞
χ = 90◦
⇓
(1.36)
Z Z
∂ϕj ∂ϕ∞
Xj = ρω ϕ∞ − ϕj dS (1.37)
∂n ∂n
Sw
The formula (1.37) is the Haskind’s relation. As seen the wave induced force
can be calculated through the radiation and free wave potentials avoiding
the determination of the diffraction potential ϕp .
26
The calculation of the integral (1.37) is a complicated problem because the
incident waves don’t decay away from the ship and the integral (1.37) should
be calculated over both the surface far from the ship and the ship wetted
surface. Note that the potential ϕ∞ does not decay away from the ship. The
method of the stationary phase [8] allows one to come to the following force
expression using the Haskind’s relation (1.37):
Z2π
k
Fi,per = Bii ζ˙i , where Bii = |Xi (χ)|2 dχ
8πρg(c/2)
0
Here c is the phase wave velocity (celerity) and χ is the course angle.
Z Z Z Z
∂ϕj ∂ϕ∞ ∂ϕj
Xj = ρω ϕ∞ − ϕj dS ≈ ρω ϕ∞ dS
∂n ∂n ∂n
Sw Sw
Z Z
∂ϕ3 ∂B ∂B
= cos(n, z) = ⇒ X3 ≈ 2ρωω ϕ∞ dS
∂n ∂z ∂z
Sw
The coefficient 2 arises due to the integration over two boards y = +B(x, z)
and y = −B(x, z). Using the potential of an Airy wave (see formulae (6.18)
in ([7]) estimated at y = 0 one can find the potential ϕ∞ :
Ag kz
ϕ= e sin(ky − ωt) ⇒ ϕ∞ = −gekz /ω
ω
For the case of a vertical cylinder for which the vertical force does not depend
on the wave course angle χ the damping coefficient B33 takes a very simple
form [8]:
0 2
Z
2ρg kz ∂B
B33 = e dz (1.38)
(c/2)
∂z
−T
1.8 Exercises
1. Schwimmender Balken
27
(a) Ist die Schwimmlage des Balkens stabil?
(b) Bleibt die Schwimmlage stabil, wenn die Höhe h des Balkens 0, 15 m
beträgt?
(c) Ab welcher Balkenhöhe wird die Schwimmlage instabil?
1
2. Berechnen Sie die maximale elektrische Leistung einer Turbine, die
die ganze mechanische Energie einer Welle umwandelt. Die Welle hat
eine Höhe von 1 m, eine Länge von 100 m und eine Breite von 1 km in
Richtung des Wellenkammes.
3. Hinter dem bildet sich ein stationäres Wellensystem.
Wie groß ist die Geschwindigkeit der Querwellen, wenn die Schiffs-
geschwindigkeit 10 m/sek beträgt?
4. Welche Länge haben die Querwellen hinter einem Schiff, das sich mit
der Geschwindigkeit von 10 m/sek bewegt?
Wie groß ist die Wellenfrequenz?
5. Die Querwelle hinter einem Schiff hat die Amplitude 1 m. Schätzen Sie
den Widerstand des Schiffes!
Benutzen Sie Fig. 1.11!
6. ([8])Das Modell eines Schiffes wird in einer sehr breiten Schlepprinne
mit der Geschwindigkeit 1 m/s 100 geschleppt. Die Modellänge beträgt 5 m.
1
taken from [8]
28
Figure 1.11: Bild aus Buch von Newman
Nach 100 Metern wird das Modell gestoppt. Wie viel Querwellen
befinden sich in der Schlepprinne, wenn die Reflektion von Schlepprin-
nenseiten nicht auftritt?
Hinweis: Das Schiff wird als Superposition von zwei Punktstörungen
betrachtet: Bug und Heck. Der Bug erzeugt die Welle. Das Heck
erzeugt die Welle. Gesamtes Wellenbild wird als Summe betrachtet:
Benutzen Sie die Formel (6.21) und die Aufgaben 2 und 3.
7. Für eine fortschreitende Welle mit der Amplitude 6 m und der Länge 200 m
berechnen Sie die Phasengeschwindigkeit und die maximale Geschwindigkeit
der Wasserteilchen.
In welchen Punkten ist diese Geschwindigkeit maximal?
8. Task:
Develop the theory of vertical oscillations of a very sharp cone with
the draught T=10m and the diameter of 1m in regular and irregular
waves using the Haskind’s relation (1.38). The added mass A33 can be
neglected.
29
30
Chapter 2
2.1 Theory
Let us consider a ship in the equilibrium position at calm water condition.
The ship has zero forward speed. If a perturbation acts on the ship, it
performs oscillating motions in three directions:
• heave,
• roll,
• pitch.
Yaw, surge and sway motions did not arise at calm water conditions. The
reason is the presence of restoring hydrostatic forces in heave, roll and pitch
directions.
ζ̈ + 2νζ ζ̇ + ωζ2 ζ = 0,
ϕ̈ + 2νϕ ϕ̇ + ωϕ2 ϕ = 0, (2.2)
ψ̈ + 2νψ ψ̇ + ωψ2 ψ = 0,
31
where
r r s
ρgAW P ρg∇0 GMγ ρg∇0 GML
ωζ = , ωϕ = , ωψ = (2.4)
m + A33 Ixx + A44 Iyy + A55
The equations (2.2) are fully independent of each other. The solutions of the
equations (2.2) written in the general form:
ξ¨ + 2ν ξ˙ + ω 2 ξ = 0 (2.5)
is given as:
ξ = Cept (2.6)
Substitution of (2.6) into (2.5) yields the algebraic equation
p2 + 2νp + ω 2 = 0 (2.7)
which solution is
√
p1,2 = −ν ± ν 2 − ω2 (2.8)
If the system has no damping the solution is
For real ships the damping coefficient is smaller than the eigenfrequency ν ω
and the equation (2.8) has two solutions:
√
p1 = −ν + i ω 2 − ν 2 = −ν + iω̄,
√ (2.10)
p2 = −ν − i ω 2 − ν 2 = −ν − iω̄,
32
In turn, the solution of the differential equation is
2π 2π
T = =√ (2.13)
ω̄ ω − ν2
2
Since ω ν
2π 2π 2π
T = =√ ≈ (2.14)
ω̄ ω2 − ν 2 ω
From (2.12) and (2.14) we obtain the frequencies for different types of oscil-
lations which are listed in the table below:
ξ(t) e−νt
= −ν(t+T ) = eνT (2.15)
ξ(t + T ) e
The logarithm of the ratio (2.15) is
33
ξ(t) e−νt 2πν
ln = ln −ν(t+T ) = lneνT = (2.16)
ξ(t + T ) e ω̄
The ratio ω̄ν is called as the referred damping factor ν̄. The decay of the oscil-
lation amplitude is equal to this factor multiplied by 2π. Referred damping
factors for different types of oscillation can be found from this definition.
The results obtained under assumption ω ν are listed in the table 2.2.
2.2 Exercise
1. ϕ0 is the roll angle at t = 0. Find the number of periods N of free roll
oscillations necessary to reduce the amplitude oscillations by factor ea .
What is influence of the metacentric height on N?
3. Typical periods of roll and pitch oscillations for different ships are [9]:
Explain why Tϕ > Tψ .
34
Ship Tϕ , sec Tψ , sec
tanker 9 ... 15 7 ... 11
ice breaker 8 ... 12 3 ... 5
trawler 6 ... 8 3 ... 4
big cruise liner 20 ... 28 10 ... 12
container ships (20000 − 30000 t) 16 ... 19 7 ... 9
35
36
Chapter 3
• waves amplitudes related to the wave lengths are small. Wave slope is
small.
From the first two assumptions it follows, that the collective action of waves
on ship can be considered through the superposition principle. Therefore,
the theory can be developed for the interaction of the ship with a single wave
with given length and amplitude. The effects of different waves are then
summed. For the case of small waves the oscillations are decoupled. The
hydrodynamic, hydrostatic and gyroscopic coupling effects are neglected.
The perturbation forces (see the last column in the equation system (1.32))
arise due to wave induced change of the hydrostatic forces and due to hy-
drodynamic effects caused by orbital motion in waves. The orbital motion
causes the hydrodynamic pressure change which results in the wave induced
hydrodynamic forces.
37
centre A on the free surface. When considering the roll and pitch oscillations
in transverse waves it is additionally assumed that the ship draught change ζ
and the ship slope relatively to the free surface are constant along the ship.
The wave ordinate is given by the formula derived for the progressive wave
(see Chapter 6 in [7])
In this section the incident waves are perpendicular to the ship (see Fig.3.1).
The wave propagation direction is in η direction, i.e. χ = η. The waves induce
roll and heave oscillations. The curvature of the free surface is neglected, the
free surface is considered as the plane performing angular oscillations and
translational oscillations in vertical direction.
Figure 3.1:
38
where η = 0 at the point A. Since the wave slope is neglected, the de-
pendence of ζ0 on ζ is not considered. The first part −ρgAW P ζ is the
restoring hydrostatic force which is already present in (1.32). The second
part ρgAW P A sin ωt is the wave induced hydrostatic force.
The relative slope of the ship to the free surface is ϕ − α, where α is the wave
surface slope
39
ω2
dζ0
α= = Ak cos ωt = A cos ωt = αA cos ωt (3.3)
dη η−0 g
αA is the amplitude of the angular water plane oscillations. The hydrostatic
pressure increases linearly in direction perpendicular to the free surface plane.
Therefore, the restoring moment is the same as in the case if the free surface is
horizontal and the ship is inclined at the angle ϕ − α. The restoring moment
is known from the ship hydrostatics
The first part −ρg∇0 GMγ ϕ is the restoring hydrostatic moment, whereas the
second part ρg∇0 GMγ αA cos ωt is the wave induced hydrostatic moment.
ρu2 ∂ϕ
p=− − ρgz − ρ + pa (3.5)
2 ∂t
The constant pressure pa does not need to be considered since being inte-
grated over the ship wetted surface results in zero force and moment. The
first term in (3.5) is neglected within the linear theory under consideration.
The second term results in force and moment considered above in the sec-
tion 3.1. The remaining term punst = −ρ ∂ϕ ∂t
is responsible for hydrodynamic
effects caused by waves. If the interaction between the ship and incident
waves is neglected (Krylov - Froude formalism) the potential can be written
as the potential of uniform unsteady parallel flow:
ϕ = uζ (t)ζ (3.6)
dζ0
uζ (t) = (3.7)
dt
∂u
Since the unsteady pressure punst = −ρ ∂ϕ∂t
= −ρζ ∂tζ is zero at ζ = 0 the
total pressure is equal to the atmospheric pressure p = pa (see (3.5)). The
gradient of the hydrodynamic pressure in vertical direction reads:
40
∂punst ∂ ∂ϕ ∂ ∂ϕ ∂uζ (t)
= −ρ = −ρ = −ρ = −ρζ̈0 = ρω 2 A sin ωt
∂ζ ∂ζ ∂t ∂t ∂ζ ∂t
(3.8)
The unsteady pressure at the point ζ is then:
Zζ Zζ
unst unst ∂punst ∂punst
p (ζ) = p (ζ = 0) + dz = dz
∂z ∂z
0 0
Zζ
I I unst
∂p
dFζdyn = punst cos(nζ)dC = dz cos(nζ)dC =
∂z
0
(3.9)
Zζ
I I
= ρω 2 A sin ωtdz cos(nζ)dC = ρω 2 A sin ωt ζ cos(nζ)dC
0
Being integrated along the ship length this force gives the force acting on the
whole ship length
ZL ZL
Fζdyn = dFζdyn dξ =− ρω 2 A sin ωtAf dξ =
0 0
(3.11)
ZL
= −ω 2 A sin ωt ρAf dξ = −mω 2 A sin ωt = mζ̈0
0
41
I
dMϕdyn = punst (η cos(nζ) − ζ cos(nη))dC =
I (3.12)
2
= ρω A sin ωt ζ(η cos(nζ) − ζ cos(nη))dC
Within the linear theory considering small ship slopes the last integral in (3.12)
is zero, i.e.
ZL
dMϕdyn =0→ Mϕdyn = dMϕdyn = 0 (3.13)
0
The first term is caused by hydrostatic effect, whereas the second one by
hydrodynamic effects. The second term is referred in the literature to as the
Smith effect.
The full Froude Krylov moment contains only the wave induced hydrostatic
component:
42
first components of both forces −Ajj ζ̈j and −Bjj ζ̇j are already represented
by the first and the second columns in the motion equations (1.32). The sec-
ond components Ajj ζ̈jL and Bjj ζ̇jL represent the hydrodynamic forces due
to interaction between the incident waves and floating body. Remembering
that ζ̇jL = ωA cos ωt and ζ̈jL = −ω 2 A sin ωt we obtain the lift force caused
by the interaction between the ship and incident wave:
ω4 ω3
Mϕ2 2
= −A44 ω αA cos ωt − B44 ωαA sin ωt = −A44 A cos ωt − B44 A sin ωt
g g
(3.17)
To calculate the full wave induced force we have to note that the Smith effect
is already represented in the force Ajj ζ̈jL + Bjj ζ̇jL . All hydrodynamic effects
are taken into account. Only the hydrostatic part of the Froude Krylov force
should be added to Ajj jL¨ + Bjj ζ̇j L to get the full wave induced force:
ω2 ω3
Mϕ,per = ρg∇0 GMγ − A44 ω 2 A cos ωt − B44 A sin ωt (3.19)
g g
43
(m + A33 ) ζ̈ − ζ̈0 + B33 ζ̇ − ζ̇0 + ρgAW P (ζ − ζ0 ) = −mζ̈0 (3.20)
(Ixx + A44 ) (ϕ̈ − α̈) + B44 (ϕ̇ − α̇) + ρg∇0 GMγ (ϕ − α) = −Ixx α̈. (3.21)
The solution of both equations can be represented as the sum ζ = ζinh + ζf ree
ϕ = ϕinh + ϕf ree , where ζf ree and ϕf ree are free heave and roll oscillations:
ϕf ree = Ce−νϕ t (cos ω̄ϕ t ± i sin ω̄ϕ t), ζf ree = Ce−νζ t (cos ω̄ζ t ± i sin ω̄ζ t) satis-
fying the homogeneous equations:
When the free oscillations decay ϕf ree , ζf ree −−−→ 0, the solutions of the
t→∞
equation (3.20) and (3.21) tend to the solutions of inhomogeneous equations:
ω2
ϕ̈(r) + 2νϕ ϕ̇(r) + ωϕ2 ϕ(r) = αA cos ωt, (3.23)
1 + kϕ
where kϕ = A44 /Ixx . The solution of (3.23) is seeking in the form
(r)
ϕ(r) = ϕA cos (ωt − δϕ ) (3.24)
Substituting (3.24) into (3.23) and separating terms proportional to cos ωt
and sin ωt gives two equations:
(r) ω2
ϕA (ωϕ2 − ω 2 ) cos δϕ + 2νϕ ω sin δϕ =
αA (3.25)
1 + kϕ
(r)
ϕA −2νϕ ω cos δϕ + (ωϕ2 − ω 2 ) sin δϕ = 0
(3.26)
It follows from (3.25) and (3.26)
44
2 h 2
(r)
ϕA ωϕ2 − ω 2 cos2 δϕ + 4νϕ2 ω 2 sin2 δϕ +
ω4
sin δϕ ωϕ2 2 2
+4νϕ ω −ω cos δϕ = 2 αA
(1 + kϕ )
2 h 2
(r)
ϕA ωϕ2 − ω 2 sin2 δϕ + 4νϕ2 ω 2 cos2 δϕ −
−4νϕ ω cos δϕ ωϕ2 − ω 2 sin δϕ = 0.
(r)
2 h 2 i ω4
ϕA ωϕ2 − ω 2 + 4νϕ2 ω 2 = 2
2 αA
(1 + kϕ )
(r)
allows one to find the ratio ϕA /αA
(r)
ϕA ω̂ϕ2 / (1 + kϕ )
=q , (3.27)
αA 2
1 − ω̂ϕ2 + 4ν̂ϕ2 ω̂ϕ2
where ω̂ϕ = ωωϕ and ν̂ϕ = ωνϕϕ 1 . Eigenfrequency ωϕ and damping coefficient νϕ
are given by formulae (2.3) and (2.4). The phase of the response relative to
that of the input (phase displacement) is found from (3.26):
2ν̂ϕ ω̂ϕ
δϕ = arctan (3.28)
1 − ω̂ϕ2
(r)
ζA ω̂ζ2 /(1 + kζ )
=q (3.29)
A (1 − ω̂ζ2 )2 + 4ν̂ζ2 ω̂ζ2
!
2ν̂ζ ω̂ζ
δζ = arctg (3.30)
1 − ω̂ζ2
ω νζ
with kζ = A33 /m, ω̂ζ = ωζ
and ν̂ζ = ωζ
.
1
since ω ν, ν̂ϕ ≈ ν̄ϕ introduced in table 2.2
45
3.7 Analysis of the formula (3.27)
The formula (3.27) can be rewritten as follows:
(r)
ϕA ϕA − αA ω̂ϕ2 / (1 + kϕ )
= =q
αA αA 2
1 − ω̂ϕ2 + 4ν̂ϕ2 ω̂ϕ2
or
ϕA ω̂ϕ2 / (1 + kϕ )
=1+ q (3.31)
αA 2
1 − ω̂ϕ2 + 4ν̂ϕ2 ω̂ϕ2
The physical meaning of terms in (3.31) is obvious from the following expres-
sion
46
2 /(1+k )
ω̂ϕ ϕ
The enhancement function q
2 goes to zero if referred frequency
(1−ω̂ϕ2 ) +4ν̂ϕ2 ω̂ϕ2
ω̂ 2 /(1+k )
becomes zero. At very large frequencies ω → ∞, q ϕ 2 ϕ → 1/ (1 + kϕ ).
(1−ω̂ϕ2 ) +4ν̂ϕ2 ω̂ϕ2
The enhancement is maximum in the resonance case ω̂ϕ = ωωϕ = √ 1 2 ⇒
1−2ν̂ϕ
ω = √ ωϕ . Strictly speaking the resonance frequency ω = √ ωϕ 2
is not
1−2ν̂ϕ 1−2ν̂ϕ
equal to the eigenfrequency ωϕ , i.e. ω > ωϕ . Since ν̂ϕ is small, this discrep-
(r)
ϕA
ancy can be neglected ω ≈ ωϕ . Typical dependence of the ratio αA
on the
referred frequency is presented in Fig. 3.5.
47
Figure 3.6: Phase displacement versus referred frequency
Similar results are obtained from analysis of the heave oscillations formu-
lae (3.29) and (3.30).
48
3.8 Sway ship oscillations in beam sea
The equation describing the sway oscillations is (see the second equation in
the system (1.32)):
Ag kζ
ϕ= e cos(kη + ωt)
ω
The wave ordinate is
1 ∂ϕ(η, o, t)
ζ=− = A sin(kη + ωt)
g ∂t
The velocity of a fluid particle in η direction is calculated as
∂ϕ
η̇ = = −Aωekζ sin(kη + ωt)
∂η
Acceleration at point η = 0 and ζ = 0 reads
49
The hydrodynamic component of the wave induced force is written in the
similar form as (3.16):
(η − η0 ) = Ceλt (3.40)
which substitution into (3.39) allows one to find λ
B22
λ=− (3.41)
m + A22
The parameter λ is negative. Therefore, (η−η0 ) = Ceλt −−−→ 0 ⇒ η −−−→ η0 .
t→∞ t→∞
As soon the transitional process is finished, the ship oscillates together with
the wave
η = η0 = A sin ωt (3.42)
Ared = κA (3.43)
The ship oscillations at B/Lw ∼ 0(1), T /Lw ∼ 0(1) are smaller than these
at B/Lw ≈ 0, T /Lw ≈ 0 due to two reasons
50
• Hydrostatic force is smaller because the submerged volume is smaller
due to wave surface curvature,
The first reduction factor is mainly due to the finite beam to length ra-
tio B/Lw ∼ 0(1).
Let us consider first the reduction coefficient for the heave oscillations. The
factor κBζ considers the reduction of the hydrostatic force due to the finite
beam to length ratio. To estimate κBζ the fixed ship is considered at the time
instant ωt = π/2 when the wave crest is in the symmetry plane (Fig. 3.10).
Figure 3.10:
R0 = ρgAAwp (3.44)
51
Z ZL/2 B(ξ)/2
Z
Rtrue = ρgA cos kηdξdη = 2ρgA cos kηdηdξ =
Awp −L/2 0
(3.45)
ZL/2
2ρgA kB(ξ)
= sin dξ
k 2
−L/2
kB(ξ) kB (kB)3
sin = − + ...
2 2 48
the final formula for Rtrue takes the form:
ZL/2 ZL/2
(kB)3
2ρgA kB(ξ) 2ρgA
Rtrue = sin dξ ≈ kB/2 − dξ =
k 2 k 48
−L/2 −L/2
(3.46)
ZL/2
ρgAk 2 k2 I
3
= ρgAAwp − B dξ = ρgAAwp 1 − ,
24 2 Awp
−L/2
where
ZL/2 ZL/2
1
Awp = Bdξ, I= B 3 dξ,
12
−L/2 −L/2
The reduction of the hydrostatic force can be taken by the following coeffi-
cient into account:
2
ρgAAwp 1 − k2 AIwp k2 I
κBζ = =1− (3.47)
ρgAAwp 2 Awp
The second reduction factor is mainly due to the finite draught to length ra-
tio T /Lw ∼ 0(1). The factor κT ζ considers the reduction of the hydrodynamic
force due to the finite draught to length ratio. The reduction coefficient is
given here without derivation:
2 3
T χ T χ T
κT ζ = 1 − χ 2π + 2π − 2π (3.48)
Lw 2(2 − χ) Lw 6(3 − 2χ) Lw
52
where χ is the coefficient of the lateral area χ = ALA /(LT ).
The total reduction coefficient κζ is calculated as the product of κBζ and κT ζ
neglecting their mutual influence:
κζ = κBζ κT ζ (3.49)
Lw Lw
The formula (3.49) is valid at > 4,
B
> 8. For heave calculations one
T
can use the formula (3.29) with Aκζ instead of A.
Reduction coefficient of the roll oscillations can be calculated from the ex-
pression gained from regression of experimental data:
"p #1/2
2 BT χrγ /GMγ
κϕ = exp −4.2 (Rω̂ϕ ) , R = χωϕ (3.50)
2πg
53
be made in theory
• the added mass Aij and damping coefficients Bij depend on the en-
counter frequency
v cos ϕwave
ωe = ω − ω 2
g
where ϕwave is encounter angle (Fig. 3.12)
54
The formulae (3.27 - 3.30) can be used also in the case v 6= 0 with the
substitution ωe instead of ω.
55
56
Chapter 4
Within the linear theory of ship oscillation the ship is considered at rest. The
perturbation force acting on the section AB (Fig. 4.1) can be represented as
the sum of
57
• the hydrodynamic force µ33 ζ̈0 (ξ) + λ33 ζ̇0 (ξ)
dFζ,per
= µ33 ζ̈0 (ξ) + λ33 ζ̇0 (ξ) + ρgB(ξ)ζ0 (ξ) (4.4)
dξ
Here we used the principle of relative motion (see section 3.4) for a ship
dFζ,per
frame. Integrating dξ over the ship length we obtain the whole wave
induced force Fζ,per . If the ship is symmetric with respect to the midship
B(ξ) = B(−ξ), Af (ξ) = Af (−ξ), µ33 (ξ) = µ33 (−ξ), λ33 (ξ) = λ33 (−ξ), the
terms with sin(kξ) are neglected and the formula for Fζ,per is simplified to:
Z Z
2
Fζ,per = −ω A µ33 cos kξdξ · sin ωt + ωA λ33 cos kξdξ · cos ωt+
L L
Z Z
ρgB(ξ) − ω 2 µ33 cos kξdξ · sin ωt+
+A ρgB(ξ) cos kξdξ · sin ωt = A
L L
Z
+ ωA λ33 cos kξdξ · cos ωt = Fζ,per sin (ωt − δζ,per )
L
(4.5)
where
v 2 2
u
u Z Z
= At (ρgB(ξ) − ω 2 µ33 ) cos kξdξ + ω 2 λ33 cos kξdξ ,
u
Fζ,per
L L
R
ω λ33 cos kξdξ
L
δζ,per = −arc tan R ;
(ρgB(ξ) − ω 2 µ33 ) cos kξdξ
L
(4.6)
dFζ,per
The wave exciting moment is calculated by multiplication of dξ
with the
arm ξ:
58
Z
dFE,ζ
Mψ,per = − ξ dξ
dξ
ZL
ρgB(ξ) − ω 2 µ33 ξ sin kξ · cos ωt−
=A (4.7)
L
Z
− ωA λ33 ξ sin kξ · sin ωt = Mψ,per sin (ωt − δψ,per )
L
where
v 2 2
u
u Z Z
= At (ρgB(ξ) − ω 2 µ33 ) ξ sin kξdξ + ω 2 λ33 ξ sin kξdξ ,
u
Mψ,per
L L
R 2
(ρgB(ξ) − ω µ33 ) ξ sin kξdξ
Substitution (4.5) and (4.7) in the third and sixth equations of the sys-
tem (1.3.2) gives:
Iyy ψ̈ = −A55 ψ̈ − B55 ψ̇ − ρg∇0 GML ψ + Mψ,per sin (ωt − δψ,per ) (4.10)
(Iyy + A55 ) ψ̈ + B55 ψ̇ + ρg∇0 GML ψ = Mψ,per sin (ωt − δψ,per ) (4.12)
Note that A33 , B33 are the coefficients for the whole ship, whereas µ33 and
λ33 for ship frames.
Dividing both equations by the coefficient of the first term one obtains:
59
ζ̈ + 2νζ ζ̇ + ωζ2 ζ = fζ sin (ωt − δζ,per )
(4.13)
ψ̈ + 2νψ ψ̇ + ωψ2 ψ = fψ sin (ωt − δψ,per )
where
f¯ζ
ζA = q
(ωζ2 − ω 2 )2 + 4νζ2 ω 2
60
2νζ ω
δζper = arctan
ωζ2 − ω 2
Similarly one gets the solution for pitch:
!
fψ 2νψ ω
ψA = q , δψper = arc tan (4.16)
(ωψ2 − ω 2 )2 + 4νψ2 ω 2 ωψ2 − ω 2
61
The local change of the draft is:
A sample of the curve y = z(x) is shown in Fig. 4.3. The maximum draft is
then
q
zmax (x) = f12 (x) + f22 (x) (4.21)
The curve y = +zmax (x) shows the contour of maximum wave elevations
along the ship board in the symmetry plane whereas the curve y = −zmax (x)
the minimum wave elevations. Both curves are symmetric with respect to
the equilibrium water plane.
• Deck flooding takes place if zmax (x) > H, where H is the board height.
• Slamming takes place if zmax (x) > T
There three zones limited by curves y = ±zmax (x) can be distinguished along
the ship board (see Fig. 4.3):
• Allways dry area (white),
• Allways wetted area (red),
• Intermediate area (orange).
A sample of flooding curves for a real ship is given in Fig. 4.4
62
Figure 4.4: Sample for a real ship
63
64
Chapter 5
65
Figure 5.2: Influence of the vertical acceleration on the seasickness depending
on the oscillation period
For the irregular sea state the similar diagram was proposed by Krappinger
(Fig. 5.2) who estimated the percentage of people suffering from the seasick-
ness depending on the root mean square deviation and frequency.
66
Figure 5.3: Number of passengers suffering from seasickness on a cruise liners
depending on vertical accelerations
67
68
Chapter 6
h = 0.17L3/4 Zimmermann,
h = 0.607L1/2 British Lloyd,
h = 0.45L0.6 det Norske Veritas.
69
Figure 6.2: Profile of an irregular wave. (Fig. from [4])
Within the linear theory the irregular waves can be represented as the super-
position of regular waves with different amplitudes, frequencies and course
angles, as shown in Fig. 6.3.
σ 2 = (ζ − ζ0 )2 = Dζ (6.2)
Dζ is the dispersion. Probability P (ζ1 < ζ < ζ2 ) of the event, that the
ordinate lies in the range between ζ1 and ζ2 is then
√
ζ2 /
Zζ2 Zζ2 Z 2Dζ
1 2 /(2σ 2 ) 1 2
P (ζ1 < ζ < ζ2 ) = p.d.f.(ζ)dζ = p e−ζ dζ = √ e−t dt
2πDζ π √
ζ1 ζ1 ζ1 / 2Dζ
(6.3)
The last integral is known as the probability integral
Zx
2 2
ϕ(x) = √ e−t dt (6.4)
π
0
70
Figure 6.3: Representation of irregular wave through the superposition of
regular waves. (Fig. from [4])
The probability P (−∞ < ζ < ζ2 ) = P (ζ < ζ2 ) is the probability of the event
that ζ does not exceed ζ2 :
" !#
1 ζ
P (ζ < ζ2 ) = 1+ϕ p2 (6.7)
2 2Dζ
The probability P (ζ1 < ζ < ∞) = P (ζ1 < ζ) is the probability of the event
that ζ larger than ζ1 :
71
Figure 6.4: p.d.f. of the wave ordinate
" !#
1 ζ
P (ζ1 < ζ) = 1−ϕ p1 (6.8)
2 2Dζ
ζa −ζa2 /(2Dζ )
p.d.f.(ζa ) = e (6.9)
Dζ
The probability that the amplitude is larger than ζ ∗ is
Z∞
ζa −ζa2 /(2Dζ ) ∗2
P (ζa > ζ ∗ ) = e dζa = e−ζ /(2Dζ ) (6.10)
Dζ
ζ∗
When evaluating the wave height an observer determines the middle height
of one third of the highest waves. This height is referred to as the significant
wave height and designated as h1/3 .
72
Dζ = 0.063h21/3 (6.11)
ZZ
ζ(x, y, t) = Real dA(ω, χ)exp[−ik(x cos χ + y sin χ) + iωt + δ(ω, χ)]
(6.12)
2
Here ω is the wave frequency, k is the wave number k = ω /g, χ is the
wave course angle and δ(ω, χ) is the phase angle. The quantity dA(ω, χ, t) is
the function of the amplitude corresponding to the wave propagating at the
course angle χ < χ < χ + ∆χ with the frequency ω < ω < ω + ∆ω. The
mean square elevation is obtained from time averaging the quadrat of the
elevation:
ZT
1
ζ 2 (x, y) = limT →∞ ζ 2 (x, y, t)dt =
T
0
ZZ ZZ
= dA(ω, χ)exp[−ik(x cos χ + y sin χ) + iωt + δ(ω, χ)] dA∗ (ω1 , χ1 )exp [−
ZZ
1
−ik1 (x cos χ1 + y sin χ1 ) − iω1 t − δ(ω, χ)] = dA(ω, χ)dA∗ (ω, χ)
2
(6.13)
Here the superscript ∗ stands for the complex conjugate amplitude function.
Rigorous derivation of the formula (6.13) can be found in [10]. Multiply-
ing ζ 2 (x, y) with ρg
ZZ
ρg
2
ρgζ (x, y) = dA(ω, χ)dA∗ (ω, χ) (6.14)
2
and comparing the result with the expression for the energy (6.34) derived
in [7]
ρgA2
E = TF l + Ep = L × 1m
2
73
One can conclude that ρgζ 2 (x, y) is the time averaged energy per surface
unit. Using the representation of the integral
ZZ Z2π Z∞
∗
dA(ω, χ)dA (ω, χ) = 2 Sζ (ω, χ)dωdχ (6.15)
0 0
Z2π
Sζ (ω) = Sζ (ω, χ)dχ (6.16)
0
The spectrum of the wave state Sζ (ω) shows the distribution of the wave en-
ergy on frequencies. The two dimensional spectrum Sζ (ω, χ) can be restored
from the one dimensional one Sζ (ω) using the following simple approxima-
tion:
4
Sζ (ω, χ) = Sζ (ω) cos4 χ
3π
To determine the spectrum, the wave ordinates are measured and represented
in Fourier series. The energy ∆E(ω < ω < ω + ∆ω) is calculated as the
squared wave ordinate for each interval of the frequencies ∆ω. The spectral
density of waves is calculated as
74
One of the most popular wave spectral densities is the spectrum of Pierson
and Moskowitz (PM):
αg 2
g 4
Sζ (ω) = 5 exp −β , (6.19)
ω Uω
Figure 6.5:
Z∞ Z∞
ζa −ζa2 /(2Dζ )
h̄ = 2 ζa p.d.f.(ζa )dζa = 2 ζa e dζa = (2πDζ )1/2 (6.20)
Dζ
0 0
75
R∞
2 ζa p.d.f.(ζa )dζa
ζ1
h1/3 = R∞ (6.21)
p.d.f.(ζa )dζa
ζ1
Z∞
2
p.d.f.(ζa )dζa = 1/3 ⇒ e−ζ1 /(2Dζ ) = 1/3 ⇒ ζ1 =
p
2Dζ ln3 (6.22)
ζ1
However, from point of view of practical applications only the statistical pa-
rameters of oscillations are of importance. To determine them, the ship is
considered as the dynamic system. The seaway is the input which is trans-
formed by the ship into oscillations considered as the output. In the statisti-
cal theory shown, that if the input signal has the Gaussian p.d.f. distribution
76
the output signal has also the Gaussian p.d.f. distribution. With the other
words, the ship oscillation parameters (roll angle, etc) obey the normal Gaus-
sian law whereas the amplitudes of oscillation parameters satisfy the Raleigh
law. The only unknown value in these distributions laws is the dispersion D.
Let us consider the roll oscillations of a ship with the zero forward speed.
As shown in the previous lectures the ratio of the roll oscillations amplitude
(output signal) to the wave slope amplitude (input signal) is given by the
formula
(r)
ϕA ϕA − αA ω̂ 2 /(1 + kϕ ) ϕA
= =p ⇒ =
αA αA 2 2
(1 − ω̂ ) + 4ν̂ ω̂ 2 2 αA
ω̂ 2 /(1 + kϕ )
=p + 1 = Φ(ω)
(1 − ω̂ 2 )2 + 4ν̂ 2 ω̂ 2
where
ω̂ 2 /(1 + kϕ )
Φ(ω) = p +1 (6.26)
(1 − ω̂ 2 )2 + 4ν̂ 2 ω̂ 2
is the so called response function. Since the wave spectral density is propor-
tional to the wave ordinates squared and taking the superposition principle
into account, we obtain the following relation between the spectral density
of the seaway and the spectral density of oscillations
77
Z∞
p
Dϕ̈ = ω 4 Sϕ (ω)dω, σϕ̈ = Dϕ̈ (6.31)
0
The dispersions obtained from (6.29), (6.30) and (6.31) determine fully the
irregular ship oscillations in heavy seaway. Using them the following further
parameters can be calculated
ϕm = σϕ (6.32)
• Number of ship inclinations within the time interval t provided the roll
angle amplitude is larger than ϕ∗ :
2t −0.5(ϕ∗ /σϕ )2
Nϕ∗ = Nt P (ϕA > ϕ∗ ) = e (6.37)
T̄ϕ
The formulae (6.32) - 6.37) are derived under assumption that the oscillations
obey the Gaussian and Raleigh distribution law.
78
Chapter 7
Vm Vs
Fnm = Fns ⇒ q =q
gDm 1/3 gDs 1/3
where D is the ship displacement.
Since the periodic motion are considered, the similarity of Strouhal numbers
should also be satisfied:
ωm Lm ωs Ls
Shm = Shs ⇒ = ,
Vm Vs
where ω is the frequency. Unfortunately the similarity with respect to
Reynolds (viscosity effects) Re = vL
ν
and Weber We = %vδ2 L (spray effects)
numbers are not fulfilled:
Rem 6= Res , W em 6= W es .
Here δ is the surface tension coefficient.
It is recommended to choose the model length from the condition Re ≥ 106 1 .
1
in some literature sources this limit is Re ≥ 3 · 105
79
To reduce the effect of surface tension on roll oscillations the model width
should larger than 0.5 m. If λ is the scale factor:
Ls = λLm
the following relations derived from the similarity conditions are valid
angles ϕs = λ0 ϕm
linear accelerations ξ¨s = λ0 ξ¨m
linear velocities ξ˙s = λ1/2 ξ˙m
periods τs = λ1/2 τm
angular velocities ϕ̇s = λ−1/2 ϕ̇m (7.1)
frequencies ωs = λ−1/2 ωm
angular acceleration ϕ̈s = λ−1 ϕ̈m
displacement, mass, forces Fs = λ3 Fm
inertia moments Is = λ5 Im
• weight G
Gs = λ3 GM
(xgs , zgs ) = (xgm , zgm ) · λ
GMγ,ls = λGMγ,lm
Ixxs = λ5 Ixxm , Iyys = λ5 Iyym
Izzs = λ5 Izzm .
Since the draught and beam are approximately equal B ∼ T one assumes
that Iyy ≈ Izz .
For the determination of Izz the model is hanged out as shown in Fig. 7.1.
The model is oscillating in horizontal plane about the vertical axis as shown in
Fig. 7.1. The period of oscillation τ is measured. The differential equation of
80
Figure 7.1: Determination of the inertia moment Izzm
Gm a2
Mz = − θ
l
The oscillation equation is then
Gm a2 Gm a2
Izzm θ̈ + θ=0 ⇒ θ̈ + θ=0 ⇒ θ̈ + ωθ2 θ = 0
l lIzzm
The period of this oscillation is
r
2π lIzzm
τ= = 2π
ωθ Gm a2
Finally we have the formula for Izzm
Gm a2 τ 2
Izzm =
(2π)2 l
The additional loads are placed on the ship and they are shifted along the
x-axis as long as the condition
Izzs = λ5 Izzm
is fulfilled.
The inertia moment Ixxm and the center of gravity zg is determined using
the setup shown in Fig. 7.3.
To determine zg only one load is used, which causes the model heeling. If ϕ is
the heel angle, the distance a calculated from moment equilibrium equation:
Pl
a · Gm · ϕ = P l ⇒ a =
Gm ϕ
The gravity center ordinate is then:
zg = zn − a
The inertia moment Ixxm is determined when the model is forced to roll with
the period τ of free roll oscillations. The moment Ixxm is then calculated as
τ 2 a
Ixxm = Gm · a(( ) − )
2π g
The similarity conditions
82
Figure 7.3: Determination of Ixxm and zg
Ixxs = λ5 Ixxm
and
zgs = λzgm
are fulfilled by vertical and horizontal shifts of loads P .
The metacentric height GMγ is determined from heel tests (see Fig. 7.4).
Pl
GMγm =
Gm ϕ
ϕ = ϕ0 e−νϕ cos ω1 t,
83
Figure 7.4: Heel test
where ω1 = ωϕ2 − νϕ2 . Since νϕ2 ωϕ2 the frequencies ω1 and ωϕ are ap-
p
proximately equal, i.e. ω1 ≈ ωϕ .
Once the period of free oscillations Tϕ is measured, the frequency ωϕ = T2πϕ
and the sum Ixx + A44 are calculated (see table 2.1):
%g∇0m GMγm
Ixx + A44 = (7.2)
ωϕ2
Since the inertia moment is known, the added mass A44 is found from (7.2).
%g∇0m GMγm
A44 = − Ixx
ωϕ2
The damping factor νϕ is determined from the definition formula (2.16).
ϕ(t) 2πνϕ
ln = ≈ νϕ Tϕ
ϕ(t + Tϕ ) ω̄
where the ϕ(t) and ϕ(t + Tϕ ) are measured.
84
7.1.4 Forced oscillation at calm water conditions
For ships with large damping the free oscillations decay very quickly. It
causes big error in data analysis. For this case and to study the effect of
frequency the method of forced oscillations is applied. The model is forced
to roll using horizontally oscillating load as shown in Fig. 7.5. The load
produces the perturbation moment
Mper = M0 sin ωt
The roll oscillations without incident waves are described by the ordinary
differential equation of the second order:
M0
ϕ̈ + 2νϕ ϕ̇ + ωϕ2 ϕ = ωϕ2 sin ωt
%g∇0m GMγm
which has the solution
ϕ = ϕ0 sin(ωt − εϕ )
where
M0 1
ϕ0 = q (7.3)
%g∇0m GMγm (1 − ω̄ 2 )2 + 4ν̄ 2 ω̄ 2
ϕ
νϕ ω
ν̄ϕ = p 2 , ω̄ =
ωϕ − νϕ2 ωϕ
85
2ν̄ϕ
εϕ = a tan (7.4)
1 − ω̄ 2
The measured quantity is the roll angle ϕ.
The seakeeping tests in regular and irregular waves (Fig. 7.6) are performed
in seakeeping and manoeuvring basin. The seakeeping basins can be open or
closed. In the closed basin the irregular waves are generated using segmented
wave generators consisting of hinged flaps. Each flap is controlled separately
by a driving motor.
The seakeeping basin of MARIN (Fig. 7.7) has dimensions 170 x 40 x 5m.
The wave generator (Fig. 7.8) produces waves with significant wave height
of 0,45 m and a peak period of 2 sec. The irregular waves have a prescribed
spectrum. The model is either self propelled (free running test) or carried
by the carriage with the speed up to 6m/sec. Model length range is from 2m
to 8m. Additionally to waves the wind is generated by an adjustable 10m
wide platform with electrical fees. Free running tests are performed such
that the model follows an arbitrary pre-defined track through the basin. The
seakeeping tests in open basin are performed under condition that the free
waves, generated naturally, have desirable heights and periods.
86
Figure 7.6: Seakeeping test at MARIN ([5])
One of the most important aims of seakeeping tests in irregular waves is also
the evaluation of slamming (Fig. 7.6) and flooding.
87
The first task within framework of such measurements is the determination
of sea state. The wave heights are measured using bues, hydrostatic pressure
sensors, or by stereo photography from airplanes (see Fig. 7.9).
The ship performs tack motions as long as the one hundred full oscillations
occur (Fig. 7.10).
88
Chapter 8
8.1 Introduction
The ship damping systems are used since the middle of the last century.
The damping devices produce the moment acting against the perturbation
moment and counteract the oscillations.
The damping devices are subdivided into
• damping tanks,
• active fins.
89
8.2 Damping of roll oscillations
8.2.1 Passive systems
The U-tube passive roll stabilization system were proposed by Frahm [12]
in the beginning of the twentieth century. The Frahm’s tanks are passive
means to stabilize the roll oscillations. They don’t need additional energy
consumptions to perform the mission. The stabilizing moment acting against
the perturbing one is created by the
• weight and
• inertia
of the water moving in U- form channel shown in Fig. 8.1. The tubes of the
U-channel are connected by the air channel in the upper side of the device
as shown in Fig. 2. Therefore the water can freely flow from one board to
the other without counteracting effects caused by the air compression.
Figure 8.1: Schematical representation of the Frahm tank of the first type.
The Frahm tanks belong to the resonance adjusted devices. The duct and
air channel are selected to fulfill the following conditions:
ωϕ /ωz ≈ 1 (8.1)
π
δϕ − δz =
2
Where
ωϕ is eigen frequency of the ship
90
Figure 8.2: U-tube passive roll stabilization system manufactured by Hoppe
Marine (http://www.hoppe-marine.com/). Orange tube the are air channel.
Picture is taken from [1].
to the eigen frequency of the ship. As result, the phenomenon of the doubled
resonance occurs at the wave frequency corresponding to the resonance case:
The phase displacement between the perturbing moment oscillation and ship
oscillation for the first resonance is π/2. The ship oscillations are behind the
perturbation moment. The phase displacement between the ship oscillation
and the water oscillation in tank is also π/2. The water oscillation is behind
the ship oscillation. This means that the phase displacement between the
perturbation and the water oscillation is π. We have some sort of anti-phase
motion. The water oscillation acts against the perturbation oscillation re-
sulting in the ship amplitude reduction.
Outside of the resonance zone the stabilization system can increase the roll
oscillations as shown in Fig. 8.3. The reduction of the oscillation amplitude
attained with a well designed Frahm stabilization system can up to 50 per-
cent for the swell. In irregular waves the efficiency of the tanks is lower.
The Frahm tanks of the second type differs from the first type by the absence
of the duct. The water flows into the tubes directly from orifices in the ship
hull (Fig. 8.4).
The efficiency of the Frahm tank of the second type is similar to that of the
first type. The Frahm’s tank of the second type belong also the the resonance
91
Figure 8.3: Response function verus referred frequency without (solid line)
and with (dashed line) stabilization system.
If the ship draught is changed in operations the eigen frequency of the ship
is also changed. The eigen frequency of water oscillations in tank remains
constant. The condition (8.1) is violated. The stabilizing system becomes
non efficient. Therefore, the Frahm system is usually used on ships with
approximately constant displacement and draught.
The Flume tanks shown in Fig. 8.5 and 8.6 are free of this disadvantage. The
water oscillates in a tank which height is larger than the water level. The
water has the free surface in Flum tank. Change of the water level causes
the change of the eigen frequency of the water oscillations in tank. The tank
can be adjusted for each ship draught and loading.
92
Figure 8.5: Free surface Type passive Roll stabilization systems of Flume
ωz is changed in a wide
range due to change of
water level in tank.
Figure 8.6: Free surface Type passive Roll stabilization systems of Flume
Typical arrangement of Flume tanks is shown in Fig. 8.7, 8.8, 8.9 and 8.11
taken from [1].
8.2.2 Theory
The equation of roll oscillations without stabilizing system reads:
93
Figure 8.8: Research Ship [1]
94
Figure 8.11: Reduction of the roll amplitude using the Flume tank designed
by Hoppe Marine (taken from [1]).
95
The solution of this equation is
ϕ = ϕA cos ωt
FT ϕ = FT ϕA cos(ωt − εF T ϕ )
FT ϕA
q
= (BT ϕ − mT ϕ ω 2 )2 + (NT ϕ ω)2
ϕA
NT ϕ ω
εF T ϕ = − arctan
BT ϕ − mtϕ ω 2
ϕA destabilized
N44
ϕA = q
(%g∇0 GML − A44 ω 2 )2 + B44
2 2 2 2
ω + B44 ω
ϕA stabilized
N44
ϕA = q
(%g∇0 GML − A44 ω 2 + BT ϕ − mT ϕ ω 2 )2 + (NT ϕ + B44 )2 ω 2
96
Figure 8.12: Active stabilizing system
Sk rk
p= ; r0 =
LB B
bk
= 0.01...0.02
lk
The ratio of the bilge keel span to its length varies between 0.01 and 0.02.
The enhancement of the damping factor νϕ due to the keel effect is shown
in Fig. 8.14. Bilge keels can reduce the roll amplitude up to forty percent in
swell and up to thirty percent in irregular waves.
97
Figure 8.14: Main parameters of bilge keels.
8.2.5 Rudders
The rudders are very efficient ways to reduce the roll oscillations of fast
passenger and combat ships. The efficiency parameter can attain values of
four and five. They can be either active or passive and are maintained at the
ship bilge. The stabilizing moment arises due to dynamic lift force on rudders
which is proportional to the ship speed squared. Therefore the rudders have
no effect when the ship speed decreases, for instance at zero speed during
anchorage. The actively controlled rudders installed in horizontal plane are
called active fins (Fig. 8.15). They can be retracted during mooring. The
vertical rudder are presented on Fig. 8.16.
All advantages and disadvantages of different stabilizing systems used for roll
damping are summarized in the table 8.1.
98
Figure 8.15: Schematic representation of active fins.
99
Type Costs Conclusion
Active Euro 0,5 - 2,0 Mio. Efficient but only at speed, very high costs
fins Mainly suitable for Passenger Vessels
Rudder Euro 0,2 - 0,4 Mio. Needs well designed Rudder application
roll Did not succeed in Shipbuilding so far
Bilge Shipyard Popular due to low extra costs
keels Little roll reduction, increase resistance
Passive T EUR 50 - 80 Flexible and most cost efficient solution
tanks Requires sophisticated design and system
Figure 8.18: Reduction of pitch and heave oscillations using the stabilizing
system.
!3
Ψ̇ Ψ̇
α = k1 + k3
ωΨ ωΨ
100
Chapter 9
Parametric oscillations
GMγ = GMγ0 .
The motion equation of free roll oscillations with variable metacentric height
is:
B44 ρg∇0 GM 0γ
νϕ = , ωϕ2 =
2(Ixx + A44 ) Ixx + A44 (9.2)
AGM ρg∇0
µϕ =
Ixx + A44
Analyzing (9.2) one can state that change of the metacentric height can cause
the perturbation moment resulting in parametric oscillations.
Physical reason for the appearance of the additional perturbation moment is
101
the effect of the ship submergence change during roll and vertical oscillations.
The change of the roll angle causes the moment:
For parametric oscillation it is necessary that ship is brought from the equi-
librium state by a certain perturbation. The natural reason of such per-
turbation for roll oscillations is beam seaway. However, the parametric roll
oscillation can arise also in head waves. If the ship has a certain roll angle in
head waves, the hydrostatic lift force becomes larger at the wave crest and
smaller at wave valley. This results in a perturbation moment depending on
roll angle.
This kind of parametric oscillations depends on the wave lengths and ship
loading.
102
where
T - ship draught,
U
κ= Awp ·T
- are reduction factors (see 3.9)
U - ship volume,
103
104
Chapter 10
• the motion amplitudes and velocities are small enough to neglect the
nonlinear terms in the free-surface condition and kinematic boundary
condition on the ship frame.
The strongest assumption of the strip theory is the assumption of the two
dimensional character of the flow around each ship frame. The flow around
each ship frame is independent of the flow at other cross sections. The steady
waves are neglected. The waves caused by the ship frame are propagated
only in directions perpendicular to the ship axis (in transversal directions
with respect to ship). The ship does not produce waves in the x -direction.
Each frame is treated hydrodynamically as if it is a segment of an infinitely
long floating cylinder (Fig. 10.1)
105
Figure 10.1: Assumption of the strip theory [1].
Such assumptions are valid for slender bodies with the length to the breadth
ratio larger than at least seven. However, the experience has shown that
strip theory can be applied successfully for floating bodies with a length to
breadth ratio larger than three [1], at least from a practical point of view.
The two-dimensional nature of the problem implies three degrees of freedom
of motion:
• vertical or heave,
• horizontal or sway and
• rotational about a horizontal axis or roll.
The forces δFi acting on each frame are then summed to get the full force:
ZL
Fi = δFi dx
0
Where i is the force direction (2 or 3, sway or heave). The yaw and pitch
moments are calculated by multiplication of corresponding forces with arm x:
ZL ZL
M5 = − xδF3 dx, M6 = xδF2 dx
0 0
106
The rolling moment is just the sum of frame moments
ZL
M4 = δM4 dx
0
The ship frame is forced into harmonic motion ζ̇j cos ωt, j = 2, 3 and 4.
The velocity potential Φj (x, y, t) = Re{ϕj (x, y)exp[−iωt]} should satisfy the
following conditions
∂ 2 Φj ∂ 2 Φj
+ =0 (10.1)
∂x2 ∂z 2
∂ 2 Φj ∂Φj
2
+g =0 (10.2)
∂t ∂z
~n∇Φj = Un (10.3)
where ~n is the unit vector normal to the ship surface and Un is the normal
component of the ship surface velocity. The no penetration condition is
satisfied at the mean (rest) position of the ship frame.
Φj −
z−→
−−∞
→=0 (10.4)
• The radiation condition that the disturbed surface of the fluid takes
the form of regular progressive outgoing gravity waves at large distance
from the ship
Wehausen and Laitone [2] derived the source pulsating under the free surface
which has the potential
107
Z∞ −i·k·(z−ζ̄)
1 e
G(z, ζ) = · Re ln(z − ζ) − ln(z − ζ̄) + 2 · p.v. dk
2·π ν−k
0
− i · Re ei·ν·(z−ζ̄)
z =y+i·z ζ =ξ+i−η ζ̄ = ξ − i · η ν = ω 2 /g
The potential (10.5) satisfies the conditions (10.1), (10.2), (10.4) and the ra-
diation condition. The only condition to be satisfied is the no penetration
condition (10.3).
The sources (10.5) are distributed over the ship frame surface C0 :
Z
−i·ω·t
Φj (y, z, t) = Re Q(s) · G(z, ζ) · e · ds
Co
where C0 is the submerged contour of the cylindrical cross section at its mean
(rest) position and Q(s) represents the complex source density as a function
of the position along C0 .
The intensity Q(s) is determined from the no penetration condition
Z
~
Re (~n · ∇) · Q(s) · G(z, ζ) · ds} = 0
0C
Z (10.6)
~
Im (~n · ∇) · Q(s) · G(z, ζ) · ds = ζ̇m · ω · n(m)
C0
where ζm denotes the amplitude of oscillation and n(m) the direction cosine
of the normal velocity at z on the cylinder. Both ζm and n(m) depend on the
mode of motion.
108
Thus, for i = 1, 2, ..., N :
N
X N
X
(m) (m) (m) (m)
+ Qj · Iij + QN +j · Jij =0
j=1 j=1
N
X N
X
(m) (m) (m) (m) (m)
− Qj · Jij + QN +j · Iij = ω · ζm · ni
j=1 j=1
The hydrodynamic pressure at (yi , zi ) along the cylinder is obtained from the
velocity potential by means of the linearized Bernoulli equation:
(m) ∂Φ(m)
p (yi , zi , ω, t) = −ρ · (yi , zi , ω, t)
∂t
as:
(m) (m)
where pa and pν are the hydrodynamics pressure in-phase with the dis-
placement and in-phase with the velocity, respectively and ρ denotes the
density of the fluid.
N
X
(m) (m) (m)
M (ω) = 2 · pa (yi , zi , ω) · ni · |si |
i=1
N
X
(m) (m) (m)
N (ω) = 2 · pν (yi , zi , ω) · ni · |si |
i=1
109
nonlinear strategy, i.e. the seeking parameters depends on ship kinematic
parameters in non linear manner. The periodic part is considered as small
and found from a linear theory. This formalism is based on Fourier analysis
of unsteady processes. The strip theory described above is the sample of the
frequency domain simulation. Each unsteady quantity q is represented in
form:
q = q 0 + q̂eiωt (10.7)
where
The potential around the ship is represented as the sum of four terms [13]
The terms in the first parenthesis describes the steady flow around the ship
with account for free surface effects. The second parenthesis represent the
periodic flow due to waves. Determination of the potential (−V x + ϕs )
is discussed in the wave resistance potential theory. This problem can be
solved utilizing either a full nonlinear or a linear formalism. Numerical linear
method for two dimensional case is described in Chapter 6.7 ([7]). The
boundary conditions for ϕI are linearized. The following boundary conditions
should be satisfied [13]:
110
• radiation condition.
The Laplace equation and decay condition are automatically satisfied within
the Rankine source method. The unsteady potential is decomposed into
radiation ϕi and diffraction components ϕd :
6
X
ϕI = ϕi ui + ϕd (10.9)
i=1
∂ 2ϕ ∂ζ
2
+g =0 (10.15)
∂t0 ∂t0
111
and substituting (10.14) into (10.15) one obtains
∂ 2ϕ ∂ϕ
2
+g =0 (10.16)
∂t0 ∂z
Using the relation
∂ ∂
= − V~ ∇
∂t0 ∂t
we get
∂ 2ϕ
∂ ∂ϕ ∂ ∂ϕ ~
= = − V ∇ϕ =
∂t20 ∂t0 ∂t0 ∂t0 ∂t
∂ ∂ϕ ~ ∂ϕ
= − V ∇ϕ − V~ ∇ − V~ ∇ϕ =
∂t ∂t ∂t
∂ 2ϕ ∂ϕ ~ ∂ϕ ~ 2 2
= 2 − V~ ∇ −V∇ +V ∇ ϕ=
∂t ∂t ∂t
∂ 2ϕ ∂ϕ
= 2 − 2V~ ∇
∂t ∂t
The mixed boundary condition (10.16) in the ship fixed system reads
∂ 2ϕ ∂ϕ ∂ϕ
2
− 2V~ ∇ +g =0 (10.17)
∂t ∂t ∂z
Substituting ϕ = ϕ0 +ϕ1 , where ϕ0 = −V x+ϕs and ϕ1 = ϕw +ϕI into (10.17)
one obtains
∂ 2 ϕ1 1 1 0
~ ∇ ∂ϕ + g ∂ϕ + g ∂ϕ = 0
− 2V (10.18)
∂t2 ∂t ∂z ∂z
Since ϕ1 = ϕ̂1 eiωt , the Eq. (10.18) takes the form:
1 0
∂ ϕ̂ ∂ϕ −iωt
−ω 2 ϕ̂1 − 2iω V~ ∇ϕ̂1 + g +g e =0
∂z ∂z
0
Within the fully linear formalism developed further in this section ∂ϕ
∂z
=0
at the free surface. Therefore the problem is reduced to the problem with
respect to amplitudes ϕ̂I since ϕw is given
112
∂ ϕ̂I
− ω 2 ϕ̂I − 2iω V~ · ∇ϕ̂I + g = +ω 2 ϕ̂w +
∂z (10.19)
w
∂ ϕ̂
+ 2iω V~ ∇ϕ̂w − g ;
∂z
The no penetration condition on the ship hull is [13]
~ = (~n∇) ∇ϕ0
where m
ϕw = Re −icÂe−kz e−ik(x cos µ−y sin µ) eiωt
~u = (u1 , u2 , u3 )T describes the translations and
~ = (u4 , u5 , u6 )T = (α1 , α2 , α3 )T
α the rotations of the ship.
Z
¨ ¨
~ × x~g = −~
m ~u + α ~
α×G+ p(1) −
S
(10.21)
−% [~u~ag + ~a (~x × ~ag )]) ~ndS
113
m x~g × ~u¨ + I α ~¨ = −x~g × α ~ ×G ~ +
Z (10.22)
p(1) − % [~u~ag + α ~ g )] × (x̄ × ā) dS
+ ~ (~x × α
R R
Ixx = (y 2 + z 2 )dm, Ixz = xzdm
Ixx 0 −Ixz
I= 0 Iyy 0
−Ixz 0 −Izz
~a = (∇ϕ0 ∇)∇ϕ0
a~g = ~a − {0, 0, g}τ
The harmonic pressure p(1) is decomposed into parts due to incident wave,
due to diffraction, and due to radiation.
6
X
(1) w d
p =p +p + pi ui
i=1
i
The components p can be calculated from the linearized Bernoulli equation:
i
i ∂ϕ 0 i
p = −% + ∇ϕ ∇ϕ
∂t
Two momentum vector equations (10.21) and (10.22) form a linear system
of six equations for the six motions ui which can easily be solved.
114
10.2.2 Time domain simulation (TDS)
Time domain simulations are performed to take the nonlinear effect at large
amplitude oscillations into account. Seaway is computed as superposition of
elementary waves using the following rules [13]:
• The wave frequencies ωj are chosen such that the area under the sea
spectrum between ωj and ωj+1 is the same for all j. This results in
constant amplitudes for all elementary waves regardless of frequency.
The frequencies, encounter angles, and phase angles chosen before the simu-
lation must be kept during the whole simulations.
115
116
Bibliography
[5] http://www.marin.nl/web/Facilities-Tools/Basins/Seakeeping-
Manoeuvring-Basin/Seakeeping-and-manoeuvring-basin-pdf.htm.
[10] Price W.G. and Bishop R.E. Probabilistic theory of ship dynamics.
Halsted, London, 1974.
[11] Vugts J.H. The hydrodynamic coefficients for swaying, heaving and
rolling cylinders in a free surface. Int. Shipbuilding Progress, 15:pp. 224–
276, 1968.
[12] Frahm H. Results of trials of the anti-rolling tanks at sea. Trans. of the
Institution of Naval Architects, Vol.53, 1911.
117