Вы находитесь на странице: 1из 63

ADVANCES IN APPLWD MECHANICS.

VOLUME 18

The Theory of Ship Motions?


J . N . NEWMAN
Department of Ocean Engineering
Massachusetts Institute of Technology
Cambridge. Massachusetts

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
I1. History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
111. The Boundary-Value Problem ......................... 235
A . Exact Formulation . . . . . ......................... 236
B. The Linearized Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 237
C. Linear Decomposition of the Unsteady Potential . . . . . . . . . . . 240
D . Special Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
E. Slender Ships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
IV . Fundamental Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
A. The Two-Dimensional Green Function . . . . . . . . . . . . . . . . . . 245
B. The Three-Dimensional Green Function . . . . . . . . . . . . . . . . . 246
V. Two-Dimensional Bodies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
A . Radiation Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
B. The DiNraction Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 252
C. Applications of Green's Theorem . . . . . . . . . . . . . . . . . . . . . 253
D . LongWavelength Approximations . . . . . . . . . . . . . . . . . . . . 256
VI . Slender-Body Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
A . The Outer Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
B. The Inner Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
C. Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
D. The Inner Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
VII . Slender-Body Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
A. The Outer Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
B. The Inner Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
C. Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
D. The Inner Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
E. The Long-Wavelength Solution . . . . . . . . . . . . . . . . . . . . . . 272
VIII . The Pressure Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
A . Added Mass and Damping . . . . . . . . . . . . . . . . . . . . . . . . . 275
B. The Exciting Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280

t Preparation of this article was supported by the National Science Foundation and by the
Office of Naval Research.
22 1
Copyright 0 1978 by Academic Press. Inc
All rights of reproduction in any form reserved.
ISBN 0-12-002018-1
222 J . N . Newman

I. Introduction

Oceangoing ships are designed to operate in a wave environment that is


frequently uncomfortable, and sometimes hostile. Unsteady motions, and
structural loading of the ship hull, are two of the principal engineering
problems which result. Early research in ship hydrodynamics was devoted
primarily to operations in calm water, but a landmark paper On the Mo-
tions of Ships at Sea by Weinblum and St. Denis (1950) focused attention
on this subject, and extensive work has followed.
Ships generally move with a mean forward velocity, and their oscillatory
motions in waves are superposed upon a steady flow field. The solution of
the steady-state problem is itself of interest, particularly with regard to the
calculation of wave resistance in calm water; a comprehensive survey has
been given by Wehausen (1973). The opposite special case is that of wave
interactions with a vessel which has no mean velocity; this topic is reviewed
by Wehausen (1971), and more recent numerical solutions are described by
Mei (1977). The problem of ship motions in waves can be regarded as a
superposition of these two special cases, but interactions between the steady
and oscillatory flow fields complicate the more general problem. All three
topics are discussed by Ogilvie (1977).
In order to predict its motions in waves, a ship may be regarded as an
unrestrained rigid body with six degrees of freedom as defined in Fig. 1. The
three components of translation are surge parallel to the longitudinal axis,
heave in the vertical direction, and sway in the lateral direction orthogonal
to surge and heave. Rotational motions about the same axes are roll, yaw,
and pitch, respectively.
If the unsteady motions of the ship and the waves are of small amplitude,
systematic perturbation procedures can be justified, with the leading-order
solutions linear in these small amplitudes. Furthermore, the ambient seaway

FIG.1. The coordinate system and six modes of ship motion.


Theory of Ship Motions 223

can be decomposed into individual components which are unidirectional


and sinusoidal. In the jargon of this field, spectral analysis makes the study
of ship motions in regular waves applicable ultimately to an irregular
seaway. This synthesis is well known in other fields of applied mechanics,
particularly in random vibrations and acoustics where the analogies are
obvious.
The first application of spectral analysis to ship motions was made by St.
Denis and Pierson (1953). More recent developments are described by Price
and Bishop (1974), and in the proceedings of two symposia: Society of Naval
Architects and Marine Engineers (1974), and Bishop and Price (1975). The
present article is restricted to ship motions in regular waves.
The dynamics of ship motions are governed by equations of motion which
balance the external forces and moments acting upon the ship, with the
internal force and moment due to gravity and inertia. (Hereafter the term
force will be used in a generalized sense to include the moment.)
Assuming the ship to be in stable equilibrium in calm water, its weight is
balanced by the force of hydrostatic pressure. Similarly, the steady drag and
propulsive force are balanced. These steady forces may be neglected, and our
attention is focused on the unsteady perturbations.
The principal unsteady force acting upon an unrestrained vessel is due to
the hydrostatic and hydrodynamic components of the normal pressure
acting on the submerged surface. Additional force components which gen-
erally are neglected include the force on the ships propeller, viscous forces
acting on the submerged surface of the hull, and aerodynamic forces acting
upon the ship above the free surface.
With the assumption of small unsteady motions, of the ship and surround-
ing fluid, linear superposition can be applied. Thus we consider separately
the radiation problem, where the ship undergoes prescribed oscillatory mo-
tions in otherwise calm water, and the diflraction problem, where incident
waves act upon the ship in its equilibrium position. Interactions between
these two first-order problems are of second order in the oscillatory ampli-
tudes, and may be neglected in the linear theory.
The radiation problem may be decomposed further, by considering
separately the six degrees of freedom defined above. In each of these modes,
outgoing radiated waves will exist on the free surface.
The existence of radiated waves implies a complicated time dependence of
the fluid motion, and hence the pressure force. Waves generated by the body
at time t will persist, in principle for an infinite time thereafter, and the
resultant pressure force on the body will act similarly.
This situation can be described mathematically by a convolution integral,
with the fluid motion and pressure force at a given time dependent upon the
previous history of the ships motion. In this respect the irrotational flow due
224 J . N . Newman

to unsteady motions of a floating body is analogous to unsteady lifting-


surface theory, where in the latter problem the source of memory is the
shed vorticity in the wake. This aspect of ship motions is emphasized by
Cummins (1962) and Ogilvie (1964). Similar effects due to viscous separa-
tion are analyzed by Brard (1973). Experimental techniques which account
for the convolution effect are discussed by Bishop et a!. (1973) and by
Wehausen (1978).
For steady-state oscillatory motion a simpler description can be utilized:
since the linear pressure field is oscillatory in time, and proportional to the
amplitude of forced motion in each mode, the resulting force on the ship hull
must be of the form

1
6

Fi(t)= Re eio tjtij/, (i = 1, 2, ..., 6). (1.1)


j= 1

Here t j is the complex amplitude of the ships oscillatory displacement in


each mode of motion, o is the frequency, and t i j is a complex transfer
function which depends on the geometry of the ship hull, the frequency w,
and the forward velocity U. Re denotes the real part, which is implied
hereafter when the time dependence eiof is displayed.
The complex transfer function t i j can be expressed in terms of its real and
imaginary parts. From the physical standpoint a more useful decomposition
is
t I.J . = w2a.. - job..- c.. (1.2)
IJ IJ IJ

The coefficientsaij, bij,cijare real and correspond, respectively, to the force


components due to acceleration, velocity, and static displacement of the
ship. It is important to emphasize that the coefficients in (1.2) are not
constant, but depend on the same parameters as do t i j . In particular, the
coefficientsaij, bij, and cij will depend in general on the frequency w, and the
representation (1.2) is not meant to imply that t i j is a second-order polyno-
mial in o.
The damping codjcients bij in (1.2)are specified uniquely by the imaginary
part of t i j . The diagonal elements bii and suitable combinations of off-
diagonal elements, can be associated with the work done to oscillate the
ship. These can be related to the energy flux in the radiated waves, under the
usual assumption of an inviscid fluid.
The separation of the added-mass coefficients aijand restoring coefficients
cij is somewhat arbitrary. A physically appropriate subdivision can be af-
fected by defining
cij = - lim t i j .
0-0
Theory of Ship Motions 225

With this definition, the restoring coefficients cij can be associated with the
hydrostatic pressure gradient and, to a lesser extent, with the steady-state
dynamic pressure field.
In the diffraction problem the ship moves in its steady-state orientation, in
the presence of incident plane progressive waves of prescribed amplitude,
wavelength, and direction. The resulting oscillatory exciting force on the
ship hull is proportional to the wave amplitude A, and can be expressed in
the form
Fi(t)= AXieimf. (1.4)
Here Xi is a complex force coefficient which depends on the wavelength and
direction of the incident waves and on the ship geometry and forward speed.
In a fixed frame of reference the incident-wave frequency oois related to
the wavelength A and wavenumber K O = 2n/A by the dispersion relation.
For water of constant depth h,
wo = (gKo tanh K0h), (1.5)
and in the deep-water limit (h -+ 0 0 )
o0= (gK0).
In the moving reference frame of the ship, the incident waves arrive with
the frequency of encounter
o = 1010 - U K o cos /3I. (1.7)
Here /3 is the angle of incidence, between the phase velocity of the waves and
the forward velocity of the ship. In effect, (1.7) introduces a Doppler shift
between the wave frequency and the frequency of encounter. The frequency
of encounter is reduced infollowing seus (/3 = 0),whereas o is a maximum in
head seus (/3 = n).
The total oscillatory pressure force acting on the ship hull is the sum of
(1.1) and (1.4). The equations for unrestrained motion of the ship in a
prescribed incident-wave system follow from Newtons equations. Since
these equations of motion are linear and algebraic in ti,the only nontrivial
task is to predict the coefficients in (1.2) and (1.4).
The six modes of ship motions can be categorized in terms of the magni-
tudes of the corresponding restoring coefficients cii. These determine the
scale of the natural frequency in each mode and the resulting response
characteristics.
For surface ships (as opposed to submarines), small vertical motions are
opposed by a hydrostatic restoring force proportional to the waterplane
area S. The resonant frequency can be estimated by neglecting hydrodynam-
ic forces and equating the restoring force to the product of the ships mass
226 J . N . Newman

and acceleration. Since the mass is proportional to the displaced volume V,


the heave natural frequency is of order
(gS/V)2 = 0(g/T)/2, (1.8)
where T is the draft. The principal hydrodynamic effect is to increase the
effective mass of the ship and to reduce the natural frequency, but the order
of magnitude is not changed. The same estimate applies to pitch, since the
relevant moments of inertia of the waterplane area and ships mass are both
proportional to L?.In this respect, heave and pitch are dynamically similar.
The existence of a substantial static restoring force ensures that the ampli-
tude of the motions will be relatively small, except near the resonant
frequencies. The response at resonance depends on the magnitude of
the damping forces and the degree of tuning with respect to the exciting
forces.
The damping of pitch and heave motions is due principally to the radi-
ation of wave energy. In these modes the damping is generally subcritical but
sufficient to prevent highly tuned resonant response.
The exciting forces in pitch and heave are significant only if the wave-
length is comparable to or greater than the ship length. From the dispersion
relation it follows that excitation will occur principally from waves with a
characteristic frequency ooIO(g/L). For conventional ships L >> T, and
the estimate (1.8) indicates that resonance will occur in combination with
significant wave excitation only if the frequency of encounter is substantially
greater than the wave frequency. When the Doppler shift (1.7)is considered,
it follows that heave and pitch are most severe in head seas. Under these
circumstances the acceleration and structural loading on the ship hull are
most severe.
In most cases, heave and pitch are the modes of greatest practical impor-
tance. Moreover, the small amplitudes and large inertial effects of these
modes can be described by a linearized theory which assumes that the
unsteady motions are of small amplitude and that the fluid is inviscid. Thus,
theoretical approaches based on the methodology of applied mechanics
have been most successful in predicting these modes.
The static restoring moment in roll is small for conventional ships. This is
due in part to the narrow beam. In addition, the vertical position of the
center of gravity may be positioned to reduce this restoring moment, length-
en the natural period, and hence to reduce the angular acceleration. The
radiation damping is weak in roll, especially at the low frequencies near
resonance. Thus resonant motions occur with a large amplitude and with
significant nonlinear and viscous effects. No satisfactory method exists for
predicting the rolling motion of ships with engineering accuracy.
The remaining motions are in the horizontal plane, unopposed by
Theory of Ship Motions 227

hydrostatic restoring forces. The response is nonresonant, but motions of


large amplitude may occur at low frequencies in following waves. Partic-
ularly serious in this context is broaching due to the combined effects of
dynamic instability and prolonged unidirectional excitation. As in the case
of roll, the modes of surge, sway, and yaw may be influenced significantly by
nonlinear and viscous effects. A comprehensive discussion of ship motions in
following waves is given by Oakley et al. (1974).
Following a brief historical outline in Section 11, the remainder of this
article is devoted to a linearized analysis of ship motions in incident waves.
Slender-body approximations are applied to the ships hull form, without
restricting the scale of the wavelength or frequency. All six modes of ship
motion are considered.
In our analysis the fluid is assumed ideal, with irrotational motion, and
unbounded except for the submerged portion of the ship hull and the free
surface. Surface tension is neglected. The ship hull is assumed symmetrical,
in the port-and-starboard sense, about a vertical centerplane which con-
tains the longitudinal axis. The unsteady motions are assumed sinusoidal in
time, and they are of small amplitude. These assumptions, and the discussion
above, imply that the results may be of the greatest practical value for
predicting heave and pitch motions.
The exact and linearized boundary-value problems are formulated in Sec-
tion 111. Fundamental solutions are summarized in Section IV, including the
two- and three-dimensional source potentials or Green functions which
satisfy the linear free-surface boundary condition. Two-dimensional solu-
tions of the boundary-value problem adjacent to each section of the ship hull
are outlined in Section V.
The method of matched asymptotic expansions is used with these two-
and three-dimensional solutions, to derive a slender-body approximation for
the three-dimensional oscillatory flow field. This analysis is carried out for
the radiation problems in Section VI, and for the diffraction problem in
Section VII. In Section VIII these solutions are used to determine the
hydrodynamic pressure force acting on the ship.

11. History

Early sailing vessels favored trade-wind routes with following seas and
were unable to move with great speed to windward. For this reason, heave
and pitch were not of great importance whereas rolling motions in waves
were reduced by the stabilizing influence of the sails.
The first steamships could not attain high speed in head seas. However,
228 J. N. Newman

the absence of aerodynamic damping increased the importance of roll. This


may explain the initial study of rolling motions by William Froude (1861).
Subsequently, the importance of pitch and heave motions increased with the
power and speed of ships, attracting attention and study by the Russian
naval officer Krylov (Kriloff, 1896).
Froude and Krylov derived differential equations of motion for the iner-
tial and restoring forces of the ship. No attempt was made to analyze the
hydrodynamic disturbance associated with the presence of the ship hull.
Only the pressure field of the undisturbed incident waves was considered,
and the resultant force on the ship has become known as the Froude-Krylov
exciting force.
The first significant step to account for the hydrodynamic disturbance due
to a ship hull of realistic form was the steady-state wave-resistance theory of
Michell (1898). Michell assumed the ship to be thin, with small beams (B)
compared to the ship length, draft, and wavelength. In this respect, his
approach is related to the solution of the thickness problem in thin-wing
theory, which followed subsequently. Michell recognized the possibility of
extending his theory to include unsteady motions, but a promised sequel on
heave and pitch was never published.
Another major advance in accounting for the ships hydrodynamic distur-
bance followed in a study by Lewis (1929), of the added mass associated with
hull vibrations in structural modes. In this problem the characteristic
frequency is sufficiently large such that inertial effects are dominant, and
gravitational forces can be neglected. Thus wave effects are ignored, greatly
simplifying the analysis.
Lewis assumed the ship hull to be slender, and used a striptheory
approach to integrate the hydrodynamic force longitudinally in terms of the
two-dimensional characteristics of each transverse section. Not content with
this simplified approach, Lewis derived three-dimensional correction factors
by reference to the exact solutions for a prolate spheroid. This appears to be
the first development of a strip theory in ship hydrodynamics.
In an inviscid fluid, the vertical motion of a thin ship is equivalent math-
ematically to the wave-maker problem of a prescribed normal velocity on
a vertical plane. The latter problem was solved by Havelock (1929),during a
prolific career devoted primarily to the theory of wave resistance. It was not
until his last published paper, however, that Havelock (1958) explicitly
studied the oscillatory motions of a thin ship.
A comprehensive analysis of pitch and heave motions was made in two
papers by Haskind (1946a,b). Greens theorem was used to construct the
velocity potential due to the presence of the ship hull, and the necessary
Greens function or source potential was derived. The thin-ship approxima-
tion was invoked to solve the resulting integral equation. Haskind treated
Theory of Ship Motions 229

the initial-value problem for arbitrary time dependence, and the special case
of sinusoidal motion. A notable feature of Haskinds work is the decomposi-
tion of the velocity potential into a canonical form which includes separately
the solution of the diffraction problem and solutions of the radiation prob-
lem for each mode of oscillatory ship motion.
The thin-ship approximation was reexamined in a critical fashion by
Peters and Stoker (1957). A systematic perturbation procedure was adopted
with the ships beam and the unsteady motions assumed to be of the same
small order of magnitude. On this basis the Froude-Krylov exciting force is
the only first-order hydrodynamic force. This rather trivial first-order theory
essentially confirmed the approach of Froude and Krylov, but cast doubt
upon the value of Haskinds more extensive work.
The thin-ship approximation was refined by Newman (1961), with a more
accurate statement of the boundary condition on the oscillatory ship hull. A
systematic expansion in multiple small parameters was used to avoid the
results of Peters and Stoker (1957). Computations, however, of the damping
coefficients presented by Gerritsma et al. (1962) did not correlate well with
experiments.
It is obvious that the (inviscid) hydrodynamic disturbance due to vertical
motions of a thin ship is small in proportion to the beam. To avoid this
situation, Peters and Stoker (1957) advocated a complementary flat-ship
approximation with the draft small compared to the beam and length. This
leads to an integral equation similar in form to that of lifting-surface theory,
but with a more complicated kernel. Moreover, the intersection of the free
surface and the ship is a singular region which must be examined with great
care. Steady-state solutions have been derived for planing boats, as
described by Ogilvie (1977), but unsteady solutions are restricted to the case
of zero forward speed.
Typical ship hulls are elongated, with the beam and draft of the same
small order of magnitude compared to the length. Thus it is logical to
develop a three-dimensional approximation analogous to the slender-body
theory of aerodynamics. This was done initially for the steady-state wave-
resistance problem by Cummins (1956); subsequent references are given by
Ogilvie (1977). An important restriction resulted from the assumption, not
always explicitly stated, that the ship is slender relative to the characteristic
wavelength. Thus, the beam and draft were assumed small compared to the
wavelength scale U 2 / g as well as the ship length L. Equivalently, the Froude
number U(gL)-* was assumed to be of order one.
Unsteady solutions based on similar slender-body assumptions and
applicable to the prediction of ship motions in waves were derived by Ursell
(1962), Joosen (1964), Newman (1964), Newman and Tuck (1964), and
Maruo (1967). Here the long-wavelength assumption rZ = O(L)seemed phys-
230 J . N . Newman

ically appropriate and mathematically convenient. In this case, however, as


in the thin-ship approach, most nontrivial hydrodynamic effects are of
higher order by comparison to the hydrostatic restoring force and the
Froude-Krylov exciting force. Moreover, the inertial force due to the body
mass is of higher order, with the result that the leading-order equations of
motion are nonresonant.
Nevertheless, for shiplike vessels without forward speed, this simple long-
wavelength theory gives reasonable predictions of the pitch and heave mo-
tions, as illustrated in Fig. 2. An explanation of this fortunate situation is the
disparity between the natural frequencies and the wave frequencies where
the exciting force and moment are significant.
A different situation results if the ship moves in head seas with forward
velocity. Here the domains of resonance and significant excitation overlap,

1.25
X
X 0 F = 0.00
X F = 0.14
1.00

0.75

lSl
-
A
0.50 0

0
i
0.25

0.00 I I I

%&-
I I I I I I
a 5.0 2.0 1.0 0.5 0.3
c- x
L

FIG.2. Amplitude of heave motion, per unit wave height, predicted from the leading-order
low-frequency slender-body theory, and compared with experimental measurements for zero
forward velocity and a Froude number F = 0.14. This figure is reproduced from Newman
(1977), where a similar plot of the pitch motion is included.
Theory of Ship Motions 23 1

due to the Doppler shift toward higher frequencies of encounter. The effect
of this shift is indicated by the experimental data with forward velocity in
Fig. 2.
Naval architects have not awaited a three-dimensional theory of ship
motions which is both rigorous and practical. Instead the numerical solu-
tions of a simpler class of two-dimensional problems have been utilized,
where the body floats on the free surface and performs small oscillatory
motions without forward speed. The wavelength and body dimensions are
not restricted, but the solution is most useful, and nontrivial, when both
length scales are of the same order of magnitude. This type of problem was
first solved rigorously by Ursell (1949) for the heaving motions of a half-
immersed circular cylinder. Various extensions and generalizations have
been carried out subsequently, as described in the survey of Wehausen
,(1971). A compendium of two-dimensional results is given by Vugts (1968),
with more recent numerical techniques described by Chapman (1977) and
Mei (1977). Typical results for the heave added mass and damping of a
rectangular cylinder are shown in Figs. 3 and 4.
The first utilization of two-dimensional results in a three-dimensional
striptheory approximation for ship motions was made by Korvin-
Kroukovsky (1955). This work was self-contained in the sense that all of the

0.0I I I I
0.0 0.5 1.0 1.5

u4-
FIG. 3. Added-mass coefficients for a family of two-dimensional rectangular cylinders,

-.
based on the computations of Vugts (1968). With the normalization shown, the added-mass
coefficient is logarithmically infinite for w 0, and also for B/T -,0.
232 J . N . Newman
1

I.o

t
- b33
pe'w
0.5

0.a

FIG.4. Damping coefficients for a family of two-dimensional rectangular cylinders, based


on the computationsof Vugts (1968).The scale on the right gives the exciting-forcecoefficient,
in accordance with the Haskind relations (5.30).

relevant forces were considered in a linearized analysis of pitch and heave


motions in head seas. Use was made of concepts from the slender-body
theory in aerodynamics, supplemented by shrewd physical insight, to
account for the effects of forward speed. Some refinements and extensive
experimental comparisons were provided in a sequel by Korvin-
Kroukovsky and Jacobs (1957).
Theoretical workers were slow to accept the striptheory approach of
Korvin-Kroukovsky, due to the lack of a systematic and rational derivation.
This defect was of less concern to practical naval architects, who recognized
the computational simplicity of the strip theory and the generally satisfac-
tory agreement with experiments. In both respects, this nonrigorous
approach compared favorably with the thin-ship theory. Typical calcula-
tions for heave and pitch are compared with experimental data in Fig. 5.
More comprehensive evaluations of the strip theory are given by Salvesen et
al. (1970) and Gerritsma (1976).
Grim (1960) provided some of the first systematic calculations of the
two-dimensional added-mass and damping coefficients for shiplike forms,
suitable for use in the strip theory. Not content with this approximation,
Grim also proposed a heuristic correction based on the distribution of three-
dimensional sources on the surface of the ship, with the source strength
Theory of Ship Motions 233

FIG. 5. Heave and pitch motions predicted from strip theory and compared with experi-
ments for a Froude number F = 0.3. (From Newman, 1977.)

determined from the two-dimensional solution at each section. Further cor-


rections were introduced to account for the induced normal velocity of these
sources at other longitudinal positions along the ship hull.
By comparison to the slender-body results illustrated in Fig. 2, the superi-
ority of the strip theory for predicting pitch and heave motions in head
waves can be explained by the practical importance of the resonant
frequency regime. From the estimate (1.8) this is precisely the short-
wavelength regime where o z T / g = 0(1) and thus, for a slender ship,
o24g 9 1.
A rational foundation for the strip theory was suggested by Vossers (1962)
and Joosen (1964), based on the slender-body assumptions B/L = E << 1,
BIT = O(1). These authors showed that in the short-wavelength regime the
solution of the three-dimensional radiation problem in the near field adja-
cent to the ship hull is identical to the strip-theory solution.
In the absence of forward speed, this conclusion is fairly obvious: if a
slender ship is radiating short waves with slowly varying phase along the
length, these waves will be locally two-dimensional, and three-dimensional
interference effects are absent from the inner region. The principal effect of
forward speed is to introduce the convective derivative Ud/ax, but this is
dominated by the time derivative in the high-frequency regime.
A systematic analysis of the short-wavelength slender-body problem for
heave and pitch was carried out by Ogilvie and Tuck (1969). In addition to
234 J . N . Newman

the leading-order zero-speed striptheory results, Ogilvie and Tuck con-


sistently retained the higher-order terms of relative magnitude E . These
higher-order corrections are linearly proportional to U and thus provide a
rational approximation for the effects of the ships forward speed.
In spite of the work of Ogilvie and Tuck (1969), certain aspects of the strip
theory remain unsatisfactory from the rational viewpoint. The principal
questions concern the validity of the solution at lower frequencies, the emer-
gence of forward-speed effects only as higher-order corrections, and the
intractable nature of the diffraction problem in short incident waves.
From the underlying assumptions, it is clear that the strip theory is invalid
at low frequencies of encounter. An example is that the (two-dimensional)
added-mass coefficient for vertical motions is logarithmically infinite as the
frequency tends to zero, as shown in Fig. 3. This is of little importance for
predictions of ship motions in head seas since it is the product waij which
occurs in the equations of motion. Since the strip theory correctly predicts
the hydrostatic effects which are dominant as w -+ 0, the resulting equations
of motion are also correct for w -+ 0. (However,-this is a consequence of the
fact that the Froude-Krylov force is left in its original form; if it is expanded
consistently for large frequencies, the result is divergent for w -,0.)
In the strip theory, forward speed affects the hydrodynamic force simply
by introducing terms proportional to (U/w) and (Ulw). By assumption,
U / w = O(E),and thus the effects of forward speed are higher-order correc-
tions to the zero-speed leading-order theory. Ogilvie and Tuck (1969) retain
terms of relative order E/ while neglecting higher-order terms O(E).As such,
their theory is mathematically consistent, but at variance with most of the
intuitive versions of the theory which retain some additional corrections
proportional to (U/w).
The diffraction problem is complicated in the strip-theory synthesis by the
assumption of high-frequency, short-wavelength incident waves. These are
rapidly oscillatory along the ship length, and the scattering potential must
be sought as the product of a longitudinal oscillatory function times a slowly
varying solution of the Helmholtz equation. For head seas the latter prob-
lem is singular and a special analysis is required. Various schemes have been
devised to circumvent this difficulty, with Greens theorem used to replace
the diffraction problem by a simpler radiation problem. An alternative
approach outlined by Newman (1977) restricts the incident wavelength to be
intermediate between the ships length and beam; the resulting formula for
the exciting force is used in Fig. 5.
The conventional strip theory is deficient not only for low frequencies, but
also for high speeds. A complementary approach initiated by Chapman
(1975) uses a high-Froude-number approximation suggested by Ogilvie
(1967). The flow at each section along the ship is analyzed in a quasi-two-
Theory of Ship Motions 235

dimensional manner, with interactions propagated downstream by the free-


surface condition. This technique is discussed further by Ogilvie (1977).
Chapmans computations for the sway and yaw response of a thin ship are
supported by impressive agreement with experiments.
Ship motions are of primary interest offshore in deep water. However, the
operation in coastal waters of supertankers, and other ships of deep draft,
has increased the importance of the shaliow-water regime. In the slender-
body theory, the effects of finite depth are confined to the outer field for
h/L = 0(1),but for depths comparable to the beam and draft the near field is
affected and the order of magnitude of the hydrodynamic forces is increased.
Beck and Tuck (1972) have studied the latter regime, using shallow-water
approximations to simplify the results.
The necessity of analytic approximations may be questioned in the pres-
ent era of numerical fluid mechanics. Only Chang (1977) has reported
success with the direct numerical solution of the linearized three-
dimensional ship-motion problem. Calculations of the added-mass and
damping coefficients are presented for a realistic ship hull in all modes
except surge. The results show reasonable agreement with experiments,
except in the roll mode, and confirm the limitations of the striptheory
approximations.

111. Ihe Boundary-Value Problem

It is helpful to define three Cartesian coordinate systems, with xo = (xo,


yo, zo) fixed in space, x = (x, y, z) fixed with respect to the ship, and x = (x,
y, z) moving in steady translation with the mean forward velocity of the ship.
The space-fixed system xo is the simplest in which to express the free-surface
boundary condition, whereas the shipfixed system x is the best in which to
derive the boundary condition on the ships wetted surface. The steady-
moving coordinate system x is an inertial reference frame in which the
motions are periodic.
We take z, = 0 as the plane of the undisturbed free surface, the xo-axis
positive in the direction of the ships forward velocity, and the zo-axis pos-
itive upward. The steady-moving coordinate system is defined by the
transformation
x = (xo - Ut, yo, zo), (3.1)
with U the mean forward velocity of the ship. The shipfixed coordinate
system is defined such that x = x in steady-state equilibrium.
236 J . N . Newman

A. EXACTFORMULATION

With the assumptions noted in Section I, the fluid velocity vector V(x,, t )
is equal to VCP, with the velocity potential CP(xo, t ) governed by Laplaces
equation VCP = 0 throughout the fluid domain. The fluid pressure p(x,, t ) is
given by Bernoullis equation
p = -p(CPt + +vz + gz,) + pa. (3.2)
Here p is the fluid density, g is the gravitational acceleration, and pa is the
atmospheric pressure which is assumed constant. In (3.2) and hereafter,
when the independent variables ( x , t ) appear as subscripts partial differen-
tiation is indicated.
On the submerged portion of the ships surface S, the normal velocity is
equal to that of the adjacent fluid. The appropriate boundary condition is
(Vs - V ) n = 0 - on S, (3.3)
where V, is the local velocity of the ships wetted surface. The unit normal
vector n is defined to point out of the fluid domain.
The free surface is defined by its elevation zo = c(xo, y o , t). On this surface,
the kinematic boundary condition is expressed by means of the substantial
+
derivative DfDt 5 afat V . V, in the form
(D/Dt)(C - zo) = 0 on z, = c. (3.4)
Since the position of the free surface is unknown, an additional dynamic
boundary condition is imposed, that the pressure on the free surface is
atmospheric. From Bernoullis equation (3.2) it follows that
+ &ifz+ gz, =0 on z , = c. (3.5)
This boundary condition can be used to determine the free-surface elevation
from the implicit equation
c = - (1/9)(@t+ w2)zo=[. (3-6)
Since (3.5) holds on the free surface for all time, its substantial derivative can
be set equal to zero. This gives an alternative boundary condition for the
velocity potential,
CPtt + 2VCP * V a t + $VCP * V(V@ . V@) + gCPz, = 0 on zo = [.
(3.7)
Equations (3.3) and (3.7) are the principal boundary conditions of the
problem, valid on the ship hull and on the free surface, respectively. If the
fluid domain contains no other boundary surfaces, the additional require-
ments are imposed such that V -+ 0 as z, + - 00, and such that the energy
Theory of Ship Motions 23 7

flux of waves associated with the disturbance of the ship is directed away
from the ship at infinity. The latter is the radiation condition.
The problem stated above is exact within the limitations of an ideal in-
compressible fluid. However, the nonlinear free-surface condition precludes
solutions without further simplification. Moreover, there are unresolved
questions regarding the explicit form of the radiation condition, and the
singularities at the intersection of the ship hull and free surface. Further
progress requires the fluid motion to be small in some sense.

PROBLEM
B. THELINEARIZED

In the theory of water waves it is customary to assume the amplitude of


the oscillatory wave motion to be small by comparison to the wavelength.
Equivalently, the slope of the free surface is assumed small compared to
unity. With the neglect of second-order terms quadratic in derivatives of @,
Eq. (3.7) can be replaced by the linearized free-surface boundary condition
+
Ort gOZ,= 0 on zo = 0. (3.8)
Note that this condition is imposed on the mean position of the free surface,
since the difference between the value of @ or its derivatives on zo = [ and
zo = 0 is a second-order quantity.
There are extensive solutions of Laplace's equation and the linearized
free-surface condition (3.8), as described especially by Wehausen and Lai-
tone (1960). Of particular importance is the plane-progressivewave of con-
stant amplitude A and sinusoidal profile, for which the velocity potential in
deep water is given by
@ = ( i g A / o , ) exp[K,(z, - ix, cos fl - iy, sin fl) + io,t]. (3.9)
Here K O is the wavenumber, with I = 27r/K0 the wavelength, oo is the
radian frequency in the space-fixed reference frame, and fl denotes the angle
of wave propagation relative to the x,-axis. Substituting (3.9) into (3.8) gives
the dispersion relation (1.6).
In the steady-moving reference frame the velocity potential can be
redefined in the form
'('07 YO7 '07 = @(.
l) + U t , y, z, l)
+(x3 y7 z7 l). (3.10)
Thus, in accordance with the Lorentz transformation,
a - u - +(x, y, z, t).
a,= -
at
4(xo - U t , y o , z,, t) = -
(:t
(3.1 1)
238 J . N . Newman

Transforming the linear free-surface condition in this manner, it follows that

4,, - 2U4,, + U2&, + g& = 0 on z = 0. (3.12)


In the moving reference frame the plane-wave potential (3.9) takes the
form
4 = (igA/oo) exp[Ko(z - ix cos /3 - iy sin /3) + iot], (3.13)

where the frequency of encounter is defined by (1.7).


If the ship is stable, and if the amplitude A of the incident wave system is
small, the oscillatory motions of the ship and surrounding fluid will be
proportional to A. Linearization of the unsteady problem can be justified on
this basis. However, suitable geometric restrictions must be placed on the
ship hull to ensure that the steady-state disturbance is small. Before restrict-
ing the geometry for this purpose, we first consider the general case where
only the oscillatory flow is linearized.
An overbar will be used to denote the velocity potential due to the steady
forward motion of the ship,
q x 0 , t) = UI$(x). (3.14)
The velocity vector of the steady flow relative to the moving reference frame
is
w = UV(8 - x). (3.15)
The boundary condition on the hull surface in its steady-state position S
takes the form
W.n=O onS. (3.16)
In the moving reference frame the nonlinear free-surface condition is
+ w .v(w) + gI$), = o on z = c, (3.17)
with the steady free-surface elevation given by the implicit formula
e = -(1/2g)(W2 - v),,,. (3.18)
The total potential can be written in the form
q x , , t ) = 4(xy t ) = UI$(X) + cp(x, t), (3.19)
where the unsteady component cp is assumed small. Neglecting second-order
terms in cp, the free-surface boundary condition is
4w * V(W2) + d),+ cp,, + 2 w . vcp, + w V ( W - Vcp)
*

+ # V q . v(W) + gqz = 0 on z = [. (3.20)


Theory of Ship Motions 239

The corresponding expression for the free-surface elevation is

Here the error is O(cpz), and the last form of (3.21) follows from (3.18) and a
Taylor-series expansion. Using this formula to solve for the difference (C - r)
gives
c = e - 4% + w . Vcp)/(g + w - W,)l,=i. (3.22)
The contribution from the steady terms in (3.20) can be evaluated by
expanding from c to c and by using (3.17). Thus, the unsteady velocity
potential is governed by the first-order free-surface condition

+ cp,,+ 2 w . vcp, + w . V(W . Vcp)


+ +Vcp . V ( W ) + gcp, = 0 on z = c. (3.23)
If the perturbation of the steady flow due to the ship is neglected, W = - Ui
and (3.23) reduces to (3.12).
The linearized boundary condition on the hull requires a similar analysis.
We begin by decomposing the velocity V, in the form
V, = Ui + a, (3.24)
-
where a = x x is the local oscillatory displacement of the ships surface
and the overdot denotes time differentiation in the reference frame of the
ship. Since a is a small oscillatory quantity, this vector displacement can be
expressed as
a =6 +R x XI. (3.25)
Here 6 and R denote the unsteady translation and rotation of the ship,
relative to the origin x = 0. Substitution of (3.15), (3.19), and (3.24) in (3.3)
gives the unsteady boundary condition
cp,=a*n-W*n onS. (3.26)
The last term must be retained, in spite of (3.16), since W = O(1) and the
-
difference in the boundary values of W n on S and S is O(a).
The two first-order contributions to W n on S are from the rotation of
the shipfixed coordinate system, and from the gradient of the steady flow
240 J. N . Newman

field. After accounting for both effects,


(W * n), 2 ([W - R x W + (a * -
V)W] n)s. (3.27)
Invoking the steady boundary condition (3.16), and substituting (3.27) in
(3.26), it follows that
(p,, = [a + R x W - (a V)W] * n
3 on S , S. (3.28)
Since each member of (3.28) is O(a),this boundary condition can be applied
either on S or S with the difference O(a2).
An alternative to (3.28) can be derived using (3.25), in the form
q,, = [a + (W . V)a - ( a . V)W] - n on S , S. (3.29)
The first two terms in brackets give the rate of change of a in a frame of
reference moving with the steady flow. Finally, since a and W have zero
divergence, a vector identity gives the more compact expression
(p,, = [a + V x (a x W)] - n on S , S. (3.30)
The boundary condition (3.30) was derived by Timman and Newman
(1962) to account in a consistent manner for the interaction between the
steady and oscillatory flow fields. In most prior ship-motion analyses an
incomplete form of (3.30) was used, which led to an erroneous asymmetry of
the coupling coefficients between heave and pitch.
If the perturbation of the steady flow field due to the ship is neglected in
(3.29), W = -Ui, and thus
qn= [(; - Ug)a]*n

= [a - U(R x i)] * n. (3.31)


The term proportional to U can be interpreted as the product of the ship's
forward velocity and the angle of attack due to pitch and yaw.

c. LINEARDECOMPOSITION
OF THE UNSTEADY POTENTIAL

Since the unsteady motions are assumed small, the potential (p in (3.19)
can be decomposed linearly into separate components due to the incident
wave, each of the six rigid-body motions, and the scattered disturbance of
the incident wave. With the restriction that the unsteady motions are sinus-
oidal in time with the frequency of encounter w, the motions of the ship are
denoted by
5 = (51, 5 2 , 53)eiOt, (3.32)
R = (al,R2, 03)ei"'
= (54, 5 5 , 56)ei0'. (3.33)
Theory of Ship Motions 24 1

With this notation, the unsteady component of the velocity potential can
be expressed as

(3.34)

Here cpo is the incident-wave potential of unit amplitude,


cpo = (ig/o,) exp[Ko(z - ix cos fl - iy sin p)], (3.35)
and cp, is the scattered potential such that
(d/dn)(cpo + cp7) = 0 on S. (3.36)
The components cpj ( j = 1, 2, . . . , 6) in (3.34) are the radiation potentials
due to motions of the ship with unit amplitude in each of six degrees of
freedom. From (3.25) and (3.29), these are governed by the hull boundary
conditions
cpj, = ionj + Umj on S. (3.37)
Here the components n j are defined as
(n1, n2, n 3 ) = n, (3.38)
(n4, n 5 , H 6 ) = (x x n), (3.39)

and, following Ogilvie and Tuck (1969),


(ml, m2, m 3 )= rn = - (n . V)W, (3.40)
(m4, m,, m6) = -(n * V)(x x W). (3.41)
On the free surface, the radiation potentials satisfy the boundary condi-
tion (3.23). The same condition applies to the sum (cpo + q7),but since the
incident wave satisfies (3.12), the scattered potential cp7 satisfies an inho-
mogeneous form of (3.23) with derivatives of cpo on the right-hand side.

D. SPECIAL
CASES

The steady flow field W is a major complexity in the free-surface and hull
boundary conditions. Moreover, while it is implicit that W is known in the
boundary conditions for cp, the solution of the nonlinear steady-flow prob-
lem is beyond the present state of this field. Thus, regardless of whether the
steady-flow problem is of direct interest, it must be simplified in order to
solve for the unsteady flow.
The simplest case of a moving ship is obtained when the hull shape is
restricted to be a small perturbation from a plane which contains the x-axis.
242 J . N . Newman

As in the analogous case of thin-wing theory, perturbations of the steady-


streaming flow - Ui are small, in proportion to the thickness of the hull, and
to leading order the boundary conditions (3.12) and (3.31) apply. Examples
are the thin ship and the flat ship, mentioned in Section 11.
While it is more appropriate to the study of submarines than it is to
surface ships, the assumption that the ship hull is deeply submerged
should be noted as an alternative to restrictions on the shape of the hull
surface. The leading-order steady flow near the body is that for motion in an
infinite fluid, and on the free surface the steady disturbance from the body
can be linearized. Thus the unsteady problem can be treated with relative
ease. This assumption was utilized extensively in the collected works of
Havelock (1963), although studies there of the unsteady problem generally
were carried out with the incomplete hull boundary condition (3.31) instead
of (3.28-3.30).
The steady-flow problem also can be simplified if one adopts the slow-
ship assumption that the forward velocity is small in some sense. This is a
topic of contemporary importance in wave-resistance theory (Baba and
Hara, 1977; Keller, 1978). The implications for unsteady motions in waves
have not been explored. From the boundary conditions it is clear that as
W + 0, the zero-forward-speed case will result, except possibly for a regime
where the frequency of encounter is small. Singularities in the steady-state
solution may be expected to occur at the bow and stern stagnation points.

E. SLENDER
SHIPS

Finally we consider the consequence of restricting the geometry of the


ship hull such that its beam (B) and draft (T) are small compared to the
length (L) by factors of order E G 1. It is convenient to choose the unit of
length such that L = O(l), whereas (B, T) = O(E).To leading order the com-
ponents of the factors in (3.38-3.39) reduce to
n = (n1, n 2 , 4 ) , (3.42)
(x x n) = (yn, - zn,, - x n 3 , xn2). (3.43)
The first components of these two vectors are O(E),whereas the remaining
contributions are O(1). We shall retain the first components in order to
derive nontrivial results for ship motions in surge and roll.
In the inner region where (y, z) = O(E),a coordinate stretching argument
can be used to show that gradients in the y-z plane are of order 1 / ~ Except
.
for the diffraction potential rpo + rp,, gradients in the longitudinal direction
are 0(1),and the velocity potential is governed to leading order by the
Theory of Ship Motions 243

two-dimensional Laplace equation


4yy + 422 = 0. (3.44)
Equation (3.44) applies in particular to the steady potential 0, and from
the hull boundary condition (3.16) 6 = O(E2).tThus, to leading order in E,
the components in (3.40-3.41) reduce to
m = - ( n 2 a/ay + n3 a/az)v$, (3.45)
-
m4 = - n 2 4 * + n36y + ym3 - zm2, (3.46a)
m, = -xm3 + n3, (3.46b)
m6 = xm2 - n 2 . (3.464
The factors rn, and m4 are O(E),whereas the remaining components of
(3.45-3.46) are 0(1), as in (3.42-3.43).
The free-surface boundary condition must be dealt with separately in the
inner and outer regions. In the outer region (y, z) = 0(1), gradients of the
potential are 0(1) in all three directions and, since the steady potential is
O(E), the leading-order free-surface condition is (3.12). In the inner region,
the steady free-surfacecondition (3.17) is dominated by the second term, and
the elevation (3.18) is O(E).The leading-order free-surface condition for the
steady problem is the rigid-wall boundary condition
$,=O on Z = O . (3.47)
The resulting inner solution for 6 and the factors mj in (3.45-3.46) are
independent of the forward velocity U.
The free-surface condition for the unsteady potential rp can be derived in
the inner region from (3.23). Using the above results for I$, it follows that
(Pa - 2urpxt + u2(Px* + 26,rp,t + 6yyrpt - U 6 y y r p x
- 2u6x,(P, - 2U6YrpXY + 4 y 2 ( P y y + 3 6 y 6 y y ( P , + 9rpz
=0 + O(~rp,~ V r p ) on z = 0. (3.48)
Since d/az = O ( E - ) ,the last term on the left side of (3.48) is O(V/E).The
remaining terms are O(rp), if the time derivatives and gravity are assumed to
be O(1). Thus, with the assumption that the frequency of encounter is 0(1),
the rigid-wall boundary condition (3.47) applies also to the unsteady poten-
tial. I n this case there are no wave effects in the inner region, to leading order in
the slenderness parameter E.

t Strictly, 6 = O(E log E). Logarithmic error factors will be deleted unless their display is
essential.
244 J . N . Newman

Wave effects can be introduced in the inner problem to leading order if


or if o = O(&-(l/)). In this circumstance the leading-order
a2/atz = O ( E -l),
terms of (3.48)are
+
qtt g q z = 0 on z = 0, (3.49)
corresponding to the zero-forward-speed condition (3.8). Justification for
assuming that o = O(E-()),or that oB/g = 0(1),can be based on the fact
that this is the resonant frequency domain (1.8) for heave and pitch. As
noted in the Introduction, this high-frequency regime is of the greatest prac-
tical importance for ship motions in head seas. The fact that it leads to a
relatively simple free-surface condition and, ultimately, to a correspondingly
simple strip theory, is an additional reason for its study, but not the only
one.
The error in (3.49) is a factor 1 + O ( E ~ /If~ the
) . terms of that order are
retained a more accurate free-surface condition results, in the form
+
qff g q Z - 2Uq,, + 2$,,qyf+ $Yyqt= 0 on z = 0. (3.50)
The free-surface boundary conditions (3.47) and (3.49-3.50) lead to two
separate theories for ship motions, applicable respectively for long and short
wavelengths or low and high frequencies. These are discussed in the Intro-
duction, and in greater detail by Ogilvie (1977). In Sections VI-VII we shall
develop a more general approach, which seeks to unify the two separate
theories. The velocity potentials q j in (3.34)will be derived from an asymp-
totic analysis which is valid to leading order in the body slenderness, for all
values of o I O(E(/)).The free-surface conditions to be satisfied in the
outer and inner regions are (3.12) and (3.49),respectively. With the sinusoi-
dal time dependence indicated in (3.34), it follows that
- 0 2 q j - 2ioUqjx + Uq,, + g q j z = 0 on z = 0 (3.51)
in the outer region, and
-w2qj + gqjz = 0 on z = o (3.52)
in the inner region. The first term of (3.52) is of higher order for o = 0(1),
but not for the extended frequency regime o < O(&-(/)).

IV. Fundamental Solutions

The interaction of a body with an exterior flow field can be represented by


suitable distributions of sources, dipoles and higher order multipoles. In the
simplest example, a point dipole may be used to represent the uniform flow
past a circle or sphere. More complete multipole expansions are applicable
for other body shapes, with the singularities situated at a point or on a line
Theory of Ship Motions 245

within the body. In the most general case, sources and normal dipoles may
be distributed in a continuous manner on the body surface, with Greens
theorem used to derive integral equations for the unknown source strength
or dipole moment.
The source potential is known also as Greensfunction and will be denoted
here by the symbol G. This is the fundamental singularity, since dipoles and
higher-order multipoles can be derived from the source by differentiation.
The elementary three-dimensional source potential is G = - (47cr)- with
r = 16 - x I the distance between the source and field points. An analogous
result holds in two dimensions, with G = (27c-l log r. In both cases the
normalizing factor is such that the source generates a unit rate of flux.
It is possible to use the elementary source potential for free-surface prob-
lems, as shown by Yeung (see Bai and Yeung, 1974).In linearized problems,
however, it is common to use a modified source potential satisfying the
free-surface condition, radiation condition, and (for infinite depth) the con-
dition of vanishing at z = - co.When expressed in terms of this singularity,
the velocity potential will satisfy all of the boundary conditions except that
on the body. For a particular body geometry and normal velocity one then
seeks appropriate distributions of surface sources and/or dipoles, or interior
multipole expansions, so as to satisfy the body boundary condition. Source
and multipole solutions which satisfy the linearized free-surface condition
are described systematically by Wehausen and Laitone (1960, Section 13).
In slender-body theory the source and dipole potentials are particularly
useful for the outer solution, where the body boundary condition is absent.
Typically, the outer solution consists of sources and transverse dipoles, dis-
tributed on the longitudinal axis of the body. For a slender ship these
singularities are on the free surface, and the net flux associated with vertical
motions implies the need for sources, whereas lateral motions of the ship
hull in relation to the surrounding flow are represented by transverse
dipoles.
Before considering the three-dimensional source potential we first discuss
the simpler result for two dimensions, which will be used subsequently for
the inner solution.

A. THETWO-DIMENSIONAL
GREENFUNCTION

For two-dimensional flow in the y-z plane, the free-surface condition


(3.52) is applicable. The corresponding source potential is given by We-
hausen and Laitone (1960,Eq. 13.31). In the present notation, with the source
point at the origin in the free surface,
2cos(ky)
IOm
1
GZD(y,z) = -- lim (44
2R p + o + dk k - (w - ip)2/g *
246 J. N. Newman
The parameter p can be interpreted as a Rayleigh viscosity coefficient,
representing a fictitious dissipation which suppresses incoming waves at
infinity. Alternatively, this parameter can be associated with a complex
frequency such that the factor ei(a-ip)fgrows slowly from a state of rest at
t = - GO.In either case, the limit p -,O + determines the contour of integra-
tion in the complex k'-plane, such that the radiation condition is satisfied.
Since the imaginary part of the pole k' = (a- ip)2/g is negative, p may be
set to zero in (4.1)if the contour of integration is deformed to pass above the
pole at k' = w2/g = K .
An alternative form for (4.1)is in terms of the exponential integral

defined such that the complex parameter u is exterior to a branch cut along
the negative real axis. After a reduction it follows that
1 1
G,,(y, Z) = -- Re{eK('+'Y)El(Kz+ iKy)} + - ieK(z-ilyl). (4.31
2n 2
The residue term in (4.3)results from deforming the contour in (4.1)to avoid
the branch cut.
The asymptotic properties of the two-dimensional wave source can be
obtained from the corresponding approximations of the exponential integral
(Abramowitz and Stegun, 1964,Eqs. 5.1.11 and 5.1.51). For small values of
Kr the sourcelike logarithmic singularity is displayed in the approximation

Here y = 0.577. . . is Euler's constant, and (r, 8) are polar coordinates such
that y = r sin 8, z = - r cos 8.The error in (4.4)is a factor 1 + O(K2r2).For
large values of K Iy I the asymptotic approximation of the exponential inte-
gral confirms the outgoing two-dimensional plane waves in the form
G z D &eK('-'IYI) for K l y l 9 1. (4.51

B. THETHREE-DIMENSIONAL
GREENFUNCTION

The three-dimensional source potential which satisfies the outer free-


surface condition (3.51) corresponds physically to a source of oscillatory
strength, moving with constant velocity U.This source potential was derived
initially by Haskind (1946a),and subsequently by Brard (1948).The solution
is described by Wehausen and Laitone (1960,Eq. 13.52)and by Lighthill
Theory of Ship Motions 247

(1967). With the source point at the origin on the free surface, this source
potential is given by
exp[kz + ik(x cos 8 + y sin 8)]
1
G(x, y, z ) = -2lim
8ff p + o +
j0
m
kdk
0
2n
d8
k - (o- ip + Uk cos 8)/g
(4.6)
Ultimately we shall distribute these sources along the longitudinal axis,
and an inner approximation of this distribution will be required. For this
purpose Fourier transforms are particularly helpful. Thus we shall analyze
the Fourier transform of (4.6), in the form
m
G*(y, z ; k) = [
-m
dxeikXG(x,y, 2). (4.7)

After substituting the solution (4.6),and using generalized harmonic analysis


to perform the integral in (4.7), it follows that
1 . exp[z(k2 + u) + iyu]
G*(y, z; k, lim du
K) = - -
4n p + o + j-, (k + uz)1/2- (o- ip - Uk)/g

where u = k sin 8. Here, for future convenience, we define


K = (a - Uk)/g. (4.9)
The parameter p may be set to zero in (4.8),if the contour of integration is
deformed to avoid the poles at f(K - k)/. For k < K the poles are sym-
metrically situated on the real axis, and the sign of the imaginary part as
p + O + is such that the contour should pass above or below the pole
according as u(w - U k ) is positive or negative, respectively. For k > K the
poles are imaginary, and no deformation of the contour is necessary in (4.8).
The value of G* for k = 0 is the longitudinal integral of the three-
dimensional source potential, which reduces t o the two-dimensional source
potential (4.1). To confirm this we note that

1 exp[zlul + iyu]
G*(y, z ; 0,K ) = --
411
j-,
du
(uI - ( o - i p ) / g
= GzLdY, 4. (4.10)
Approximations similar to (4.4) and (4.5)can be derived for the transform
G*(y, z ; k, K). For this purpose we shall assume that k = O(1).
An asymptotic expansion of (4.8) for K r -g 1 is derived by Ursell (1962,
Eq. 2-19), in the special case U = 0. In the present notation, Ursells result
248 J . N . Newman

can be expressed in the form

G*(y, z ; k, K ) E (1/2n)(1 + Kz)[log(* Ik I I ) + y + ( I 1 - k Z / K ZI ) - ( ' I 2 )


COS-' ( K / l k l )- II
)cosh-' ( K / l k l )+ ni
- K Z + KyO ,

where the upper or lower expression in braces is applicable according as


1 (4.11)

K/ I k I 5 1, respectively. The error in (4.11) is a factor 1 + O ( K 2 r Zk2r2).


,
This approximation is analogous to (4.4),except for an additional homo-
geneous solution of the form (1 + K z ) F ( K / l k ( ) .
Generalization of (4.11) for U # 0 simply requires that the parameter K is
replaced by K and also requires that the conjugate contour of integration is
used if w - Uk < 0. After a straightforward reduction it follows that
G*(y, z ; k , K)= G*(y, z; 0, K ) - (1/2n)(1 + K z ) f * ( k , K , K )
+ O ( ( K - K ) r , K2r2,k z r z ) , (4.12)

where G*(y, z ; 0, K) is defined by (4.10),and

f * ( k , K , K) = log(2K/ I k I ) + ai - ( I 1 - k Z / K Z I ) - ( I i z )
.\ cos-'(K/lkl) - a
\cosh-'(K/ I k I ) + ni sgn(w - U k ) ' 1 (4.13)

The asymptotic approximations (4.11) and (4.12) are valid for K r < 1,
irrespective of the magnitude of K. This can be confirmed by writing (4.8)in
terms of the nondimensional variables Ky and Kz before deriving (4.11).By
the same argument, the requirement that k = O(1) can be replaced by the
less restrictive assumption that k/K = O(1).
The functionf* defined by (4.13)tends to zero for K % 1, with the limiting
behavior
f* = log(K/K) + O ( K - ' )
= O(K-('/')). (4.14)
A complementary approach is required for the short-wavelength regime
K B 1. First we deform the contour of integration in the expression (4.8)for
G*, into the upper or lower half of the complex plane u + iu, according as
y 2 0. The contour can be deformed ultimately to a large semicircle
I u + iu I = 00, except for a branch cut I u I > I k I along the imaginary axis.
There is no contribution to the integral from the large semicircle, but one
must include the residue from the pole situated in the appropriate half-plane
Theory of Ship Motions 249

for p > 0. For short wavelengths such that o - Uk > 0 and K > 1 k 1, the
final result of this procedure is the expression
G*(y, Z ; k, K) = )i(l - k/K) exp[Kz - i Iy I (K - k)]
exp[-u lyl + iz(u - (4.15)
(u - k2) + iK k2)21.

For large K z K, the last term can be approximated by

where K l is the modified Bessel function, and the error is O(Ky)-. After
expanding the modified Bessel function for small kr and substituting the
result in (4.16), we obtain the approximation
1 cos 9
G* = - i(1 - k/K)-( exp[icz
2
-i 1 y I (K - k)] +-
211Kr
+ O(KY)-.
(4.17)
Comparison with the limiting value of (4.17) for k = 0 gives the result
G* z GZD, (4.18)
with the error a factor 1 + O(ky/K, K1y, ( K y ) - ) .

V. TweDirnensional Bodies

In the inner region close to a slender ship hull, its boundary surface is
approximated by a long horizontal cylinder having a two-dimensional
profile defined by the local cross section of the ship. We shall refer to such a
cylinder as a two-dimensional body.
In the radiation problem of forced oscillatory motions, in the y-z plane,
the inner flow is governed by the two-dimensional Laplace equation (3.44).
The resulting velocity potential is denoted by 4 ( y , z ) to distinguish this from
the outer three-dimensional potential q ( x , y , z). In both cases the complex
timedependent factor eio is implied.
For oblique waves incident upon a two-dimensional body, the three-
dimensional diffraction potential can be expressed as the product of a sinus-
oidal function of x, and a two-dimensional function @(y, z). The latter is
governed by the Helmholtz equation, reducing to Laplaces equation in the
special case of beam seas.
There is an extensive literature on wave radiation and diffraction by two-
250 J . N . Newman

dimensional bodies. Numerical results have been obtained for a variety of


body profiles, using several different methods of solution. These are
described in the surveys of Wehausen (1971) and Mei (1977).Our discussion
in this section is limited to the derivation of analytic properties which are
needed subsequently in the three-dimensional slender-body analysis.

A. RADIATIONPROBLEMS

Forced motions of the three-dimensional ship hull in sway (j= 2), heave
(j= 3), and roll (j= 4) can be related directly to the same motions of the
two-dimensional body in the y-z plane. Pitch and yaw motions will be
related ultimately to appropriate translations of the two-dimensional body
in heave and sway, respectively. The remaining surge mode (j= 1) corre-
sponds to a dilation of the two-dimensional body, with normal velocity
proportional to the longitudinal component of the unit normal vector on the
ship hull.
Thus we must consider the four radiation problems of surge, sway, heave,
and roll. In each case the two-dimensional Laplace equation (3.44) and
free-surfacecondition (3.52) apply. With the normalization (3.34), the corre-
sponding potentials satisfy the boundary condition
4Jn. = iconj (j= 1, 2, 3, 4) (5.1)
on the body profile. Restricting the body to be symmetrical about y = 0, the
surge and heave potentials are even functions of y, whereas the sway and roll
potentials are odd. Each boundary-value problem is completed by imposing
a radiation condition of outgoing plane waves at y = k 00 and by requiring
that the motion vanish for z + - 00.
Following an approach introduced by Ursell(1949), we shall express the
solutions for surge and heave in the form

c
m
cos 2mO K cos(2m - l)e
+j=ajGzD+ m=l a j m ( 7 + (2m- 1) rZm-l (j= 1, 3).

(54
In this expansion G,, is the two-dimensional source potential (4.1), and the
higher order multipoles have been combined to form the wave-free poten-
tials in braces. The source strength aj and coefficients aimare unknowns
which must be determined from the boundary condition (5.1) on the body
surface. In practice this leads to an infinite system of simultaneous equa-
tions, which can be truncated and solved by numerical methods. Ursell
(1949) proves that this process is convergent for a circular body profile. For
more general body profiles the expansion (5.2) is valid for symmetric mo-
Theory of Ship Motions 25 1

tions exterior to a circle of constant radius which encloses the body, as


shown by Ursell (1968a).
Since the wave-free potentials in (5.2) vanish at infinity, the radiated waves
are associated only with the source term. Using (4.5) with (5.2) it follows that
4J. N- 2 l j a . eJ K ' z - ' l Y l ) ( i g / o ) A eK(Z-'IyI) as lyl +GO. (5.3)
Here
Aj = t(o/g)aj (j= 193) (5.4)
is the complex amplitude of the radiated wave. Since the potential & j is
normalized for a unit amplitude of motion, (5.4) is nondimensional.
Similar results hold for sway and roll. The appropriate singularities can be
obtained by differentiation of (5.2). After redefining the unknown coefficients
it follows that

(5.5)
Here the unknown coefficients p j , ah are determined from the boundary
condition (5.1). Since the radiated waves are associated only with the wave
dipole,
c$j z (ig/o)A eK(' ' iY) as y - , +GO, (5.6)
where
A = -+ i ( o K / g ) p j ( j= 2,4). (5.7)
Hereafter we shall assume that these two-dimensional radiation potentials
are known. The hydrodynamic pressure then may be determined from the
linearized form of Bernoulli's equation (3.2),
p = - iop4e'"'. (5.8)
Ignoring the restoring force cij in (1.2), due to the hydrostatic pressure, the
remaining components of the pressure force are associated with the added-
mass and damping coefficients. These can be computed by integration of
(5.8) over the submerged portion of the body profile,

o Z A i j- iwBij = - i o p jp n i 4 j dl. (5.9)

Here A i j and Bij denote the two-dimensional values of the added-mass and
damping coefficients, and P denotes the submerged portion of the body
profile.
252 J . N . Newman

Other properties of the added-mass and damping coefficients are derived


by Wehausen (1971). Included in that survey are the computed results for a
family of rectangular cylinders, including the coefficients for heave results
reproduced here in Figs. 3 and 4, as well as similar results for sway and roll.
These calculations and additional results for other body profiles are due to
Vugts (1968).

B. THEDIFFRACTION
PROBLEM

If a progressive wave system of the form (3.35) is incident upon a two-


dimensional body, the fluid motion will be periodic along the body axis. The
effect of forward speed along this axis can be ignored, since the body is
two-dimensional. Therefore the analysis can be performed in a frame of
reference fixed with respect to the fluid, and there is no need to distinguish
the wave frequency oofrom the frequency of encounter o.
With the incident-wave potential defined by
cpo = ( i g / o ) exp[K(z - ix cos /I - iy sin /I)], (5.10)
the scattered potential cp7 is subject to the condition that the total potential
cpo + ip7 has zero normal velocity on the body. Since the flow is periodic in
the x-direction, we can write
cpj(x, Y , z ) = Oj(y,z)e-"", j = 0, 7, (5.11)
where
1 = K cos /I (5.12)
is the longitudinal component of the wavenumber.
With these definitions, the boundary condition (3.36) takes the form
= io(n, - in, sin j?)exp[K(z - iy sin /I)],
07,, (5.13)
on the body profile. After substituting (5.11) in the three-dimensional
Laplace equation, the two-dimensional functions Oj satisfy the Helmholtz
equation
+
Ojyy Ojzz- PO,= 0. (5.14)
Both Oo and O7 vanish for z -,- 00 and satisfy the free-surface condition
KOj - Ojz= 0 on z = 0. (5.15)
The scattering function O7 satisfies a radiation condition of outgoing waves
for y + +a.
Expansions similar to (5.2) and (5.5) can be applied to the symmetric and
antisymmetric portions of 07,except for the singular case sin /I = 0, where
Theory of Ship Motions 253

I
the two-dimensional solution is unbounded for I y --* 00. The proof is given
by Ursell (1968a).
The appropriate wave-free singularities which satisfy the Helmholtz equa-
tion involve the modified Bessel functions K,,,(lr);these are exponentially
small at infinity. The corresponding source function which satisfies (5.14)
can be derived from the Fourier-transformed three-dimensional Green's
function G*, by setting U = 0 and k = 1 in (4.8). With these substitutions,
and ignoring the wave-free functions, it follows that

(5.16)

where Z, denotes the source strength and M , denotes the dipole moment.
The radiated waves in the far field can be derived by substituting (4.17) in
(5.16), with the result
I
CD, 2 *(iC, csc p I f K M , ) exp[K(z - i I y sin /3 I )I, (5.17)
for y -, f00. Since the incident wave amplitude is unity, the radiated wave
amplitude at y sin /3= - 00 is equal to the rejection coejjcient
R = ( 0 / 2 g ) ( C ,csc /3+ i K M , ) sgn(/3). (5.18)
Similarly, the total wave amplitude at y sin /3 = + 00 is equal to the trans-
mission coeficient
T=1 + ( 0 / 2 g ) ( C , csc /3 - X M , ) sgn(/3). (5.19)
In (5.18) and (5.19), --R < /3 < A. If the body profile is symmetric about
y = 0, the dipole moment M , is an odd function of b, whereas the source
strength C,, and hence R and T, are even functions of /3.

C. APPLICATIONSOF GREEN'S
THEOREM

Green's theorem may be applied to the pair of functions (Yl, Y 2 )in the
plane x = constant, in the form

f (YlY2n- Y2Yln)dl =
C S
- Y2V2Yl)dS.
(Y1V2Y2 (5.20)

The closed contour C is the boundary of the simply connected domain S .


The surface integral vanishes if, in this domain, both functions satisfy the
same governing equation (3.44) or (5.14).
We shall apply (5.20) to the fluid domain in the y-z plane, between the
body profile P and a pair of vertical contours at y = k 00. These are con-
nected by horizontal lines at z = 0 and z = - 00 to form a closed contour.
254 J . N . Newman

There is no contribution from the horizontal lines to the contour integral in


(5.20),provided that the functions Y i satisfy the linearized free-surfacecon-
dition (5.15) and vanish for large depths. Thus (5.20) gives the result

- Y2Yln)dl =
(Y1Yzn -I 0

-m
dz[YIYz, - Y z Y l y ~ Z ? m . (5.21)

In this form, Greens theorem can be used to derive various relations for the
forces acting on the body and for the characteristics of the radiated waves at
infinity.
In the simplest example, we apply (5.21) to two solutions of the radiation
problem +i and dj. Since these are subject to the radiation condition at
infinity, the right side of (5.21)vanishes. After using the boundary conditions
(5.1) and comparing the left side with the pressure force (5.9), it follows that
the added-mass and damping coefficients are symmetric, i.e., A, = Aji and
BII. . = BI..t
If (5.21) is applied to the radiation potential +i and its conjugate $i,the
left side of (5.21) is proportional to the damping coefficient Bii. The contri-
bution from the right side is nonzero. After a reduction using (5.3)or (5.6),it
follows that
Bii I
= (pg2/co(3) Ai 12 (i = 1, 2, 3, 4). (5.22)
Alternatively, this relation can be derived from energy conservation.
If (5.21) is applied to the diffraction solution Oo + (P, and its conjugate,
there is no contribution from the left side due to the body boundary condi-
tion. The contribution from the integration at infinity gives the familiar
result
IRIZ+ ITl2=1. (5.23)
This relation also can be derived from energy conservation in an obvious
manner.
Alternatively, if (5.21) is applied to the diffraction potential with an angle
of incidence 8, and the conjugate of the diffraction potential with angle of
incidence R + 8, it follows that
RT + RT = 0. (5.24)
Here the symmetries noted after (5.19) are used, and (5.24) holds only for a
body profile symmetrical about y = 0.
Further relations can be obtained by applying Greens theorem to suitable
combinations of the radiation and scattering problems. TO preserve the
Helmholtz equation for both functions, we define a class of radiation prob-
lems where the normal velocity on the two-dimensional body surface is
periodic in the same manner as (5.11). This corresponds physically to a
Theory of Ship Motions 255

forced sinuous motion which propagates along the cylinder with phase veloc-
ity w/l.To be more specific, generalized radiation functions are defined to
satisfy the same conditions as 0,, except on the body profile where
ajn= i o n j ( j = 1, 2, 3, 4). (5.25)
These functions can be expanded in the same manner as the scattering
potential, with source strength C j for the symmetric modes ( j= 1, 3) and
dipole moment M j for the antisymmetric modes (j= 2, 4). The radiated
waves in the far field may be defined in a similar form to (5.17). For
y - , +a,
Q j z (ig/w)Ajexp[K(z - iy I sin B I )I, (5.26)
where
Aj = 9(0/g)cjlcsc BI ( j = 1, 3), (5.27)
Aj = -*i(oK/g)Mj ( j = 2, 4). (5.28)
The same results hold for y + - 00, provided the sign of (5.26)is reversed for
j = 2,4. When cos = 0, the wave amplitudes (5.27-5.28) reduce to the
corresponding values defined by (5.4) and (5.7).
If Greens theorem (5.23) is applied to the diffraction function a0 +

iw
and the generalized radiation function Oj, one obtains the result

jp nj(Oo+ 0,)dl = -i(g/o)Jj ( -sin.


sin /3( ) c
for j =-l , 3
- 2, 4
The left side of (5.29) is proportional to the linearized pressure force on the
). (5.29)

body profile, due to the interaction of the fixed body with the incident waves.
With the notation (1.4) it follows that the two-dimensional exciting force X j
can be related to the wave amplitude of the generalized radiation function, in
the jth mode, by means of the formula

(5.30)

This is the two-dimensional form of a more general result known in ship


hydrodynamics as the Haskind relations. Equation (5.22)can be combined
with (5.30) to relate the damping coefficient and the magnitude of the excit-
ing force, as shown in Fig. 4.
A different result follows if the generalized radiation function is replaced
by its conjugate Gj. In this case

- io jp nj(mo + (D7)dl = i ( g / o ) * J j ( R_+ T )( -Isinsin. BIj?) for 0 1.j=l 3



= 2, 4
(5.3 1)
256 J . N . Newman

Adding (5.29) to (5.31) gives the relation

(5.32)

These linear equations can be solved for the reJection and transmission
coefficients R and T in terms of the ratios Aj/Aj.A corollary is that the
phase of the radiated waves (5.27-5.28), and hence the phase of the exciting
force (5.30), is equal to the argument of -h(R f T) for symmetric or anti-
symmetric modes, respectively. Since R and T are independent of the par-
ticular modes, the two phase angles are likewise invariant with respect to
the distribution of normal velocity on the body profile.
Finally, we combine (5.32) with (5.18-5.19),and obtain relations for the
source strength X, and dipole moment M , ,

C, = -(g/u)~sin~ l ( + 1 Aj/jj) ( j = 1,3), (5.33)


M , = -i(g2/03) s g n ( ~ ) ( l -Zj/Zj) ( j = 2,4). (5.34)

The relations derived in this section from Greens theorem are special
cases of more general results which are summarized by Newman (1976).

D. LONGWAVELENGTH
APPROXIMATIONS

When the wavelength is long compared to the dimensions of the body


profile, the free-surface condition can be replaced to leading order by the
rigid-wall boundary condition (3.47). In this limit the source strength for
surge and heave can be determined from a continuity argument,

I,ah dl = -4Zj ( j = 1, 3). (5.35)

Substituting the boundary condition (5.l), it follows from geometrical con-


siderations that
XI = -2ios(x), (5.36)
x 3 = -2ioB(x). (5.37)
Here S(x) is the submerged area of the body profile; S(x) = dS/dx; and B(x)
is the beam, or width of the body profile at the waterplane. Derivations of
these results are given by Newman and Tuck (1964, Appendix 2).
The dipole moment for sway can be determined by considering the body
profile plus its image above the free surface, in an infinite fluid. For this
Theory of Ship Motions 257

double-body the dipole moment is given by the formula


M , = -2io(S + m,,/p), (5.38)
where 2m,, is the added mass of the double body. Equation (5.38) results
from a theorem due to G. I. Taylor, which is rederived and extended by
Landweber and Yih (1956).The dipole moment due to rolling motion can be
derived in a similar manner, noting that on the double body the normal
velocity for roll is an even function of z. The corresponding dipole moment is
given by
M4 = 2iw(szB + B3/i2 - m,4/p). (5.39)
Here zB < 0 is the vertical coordinate of the center of buoyancy, or the
centroid of the submerged profile, and 2mZ4 is the added-mass coupling
coefficient for the double body, between roll and sway.
The same estimates apply for the source strength and dipole moment of
the generalized functions Oj.The boundary condition (5.13) for the scatter-
ing potential implies the long-wavelength approximation
O, z -03 + i sin POz (5.40)
and, hence, from (5.36-5.37),
x, z 2ioB(x), (5.41)
M , z 2 4 s + m z z / p )sin /3. (5.42)
The amplitudes of the radiated waves can be found using (5.27-5.28).
These results are summarized as follows:
A, z -iKSIcsc 81, (5.43)
A, z - K 2 ( S + m,,/p), (5.44)
A, z -iKBIcsc 81, (5.45)
,T4 z K(SZ, + ~ ~ / -1 m,4/p).
2 (5.46)
More accurate approximations for the phase angles of the radiated waves
follow from (5.33-5.34) and (5.41-5.42) in the form

arg(Aj) z
IL
--
2
+ K B I csc fl I ( j = 1, 3), (5.47)

arg(d,) z IL - K 2 ( S + m z , / p ) I sin /3 I ( j = 2,4). (5.48)


These can be used with the Haskind relations (5.30) to approximate the
exciting forces, and with (5.32) to obtain approximations for the reflection
and transmission coefficients.
258 J . N. Newman

VI. Slender-Body Radiation

Hereafter the slender-body assumption is invoked for the ship hull with
the slenderness parameter E defined as the ratio of the beam (B) or draft (T),
divided by the length (L). It is convenient to presume a length scale such that
L = O(l), and thus (B, T) = O(E).Following the method of matched asymp-
totic expansions, approximate solutions for E 4 1 are derived separately in
the outer region (y, z) = 0 ( 1 ) and in the inner region (y, z) = O(E).These
two separate solutions then are required to match in a suitable overlap
domain E 6 (y, z) 4 1.
In this section we analyze the radiation problem for each mode of rigid-
body motion. The corresponding velocity potential 'piis defined? by (3.34).
The three-dimensional Laplace equation and linearized free-surface condi-
tion (3.51) apply in the outer region, together with a radiation condition and
the requirement that the solution vanishes for z + - 00. The inner solution
is governed by the two-dimensional Laplace equation (3.44), the linearized
free-surface condition (3.52), and the boundary condition (3.37) on the ship
hull.
The method of matched asymptotic expansions has been applied to the
radiation problem for a slender ship by Newman and Tuck (1964), for the
case where the characteristic wavelength I = O(1), and by Ogilvie and Tuck
(1969) for I = O(E).Here we seek a more general "unified" approach, which
is valid for all wavelengths 1 IO(E).
The objective of a unified theory requires a careful analysis of the match-
ing error. For this reason the errors in the inner and outer solutions will
be estimated, and the overlap region will be chosen to minimize the largest
of these. The accuracy of each solution will be indicated by an error factor 8,
with logarithmic factors neglected in these estimates.

PROBLEM
A. THEOUTER

Since the ship hull is collapsed onto the longitudinal x-axis as E + 0, the
outer solution can be constructed from a suitable distribution of singular-
ities on this axis. For the symmetric modes (j = 1,3,5) the three-dimensional
source potential (4.6) is the appropriate singularity, whereas the antisymme-
tric modes ( j = 2, 4, 6) require axial distributions of transverse dipoles.
(These statements are somewhat intuitive, but they will be confirmed ulti-
mately by matching the results with the inner solution.)

t Throughout this section the integer j takes all values from one to six, unless otherwise
noted.
Theory of Ship Motions 259

The strength of the source and dipole distributions will be denoted by


4 j ( x ) and d j ( x ) , respectively. The potential for a source at the point x = l is
obtained by shifting the longitudinal coordinate in (4.6), and the transverse
dipole potential can be derived by differentiating with respect to y . Thus the
outer solution is expressed in the form

with the convention that 4 j = 0 if j is even, and d j = 0 if j is odd.


The Fourier transform of (6.1) is derived from the convolution theorem in
the form
pi*
(
+d.* -
= qj*
aY
)
a G*(Y, 2 ; k, K), (6.2)

where G* is defined by (4.74.8). Assuming the source and dipole distribu-


tions vary slowly along the hull, the Fourier transforms 4j* and dj* will tend
to zero if k 9 1. Thus it is reasonable to assume that k = O(1), in seeking
asymptotic approximations of (6.2).
Inner approximations of (6.2) for small values of the coordinates ( y , z) can
be constructed from the results of Section IV. For small values of Kr, the
approximation (4.12) is used to give

Here G z D andf* are defined by (4.1) and (4.13), and the error in (6.3) is the
factor
+
d = 1 O ( ( K - K ) r , K 2 r Z ,k2r2). (6.4)
Alternatively, for K 1 y ) 9 1, (4.18) gives the approximation

with the error factor


d = 1 + O ( k Z y / K ,Ky, ( K y ) - ) . (6.6)

B. THEINNER PROBLEM

In constructing the inner solution we shall ignore temporarily the match-


ing requirement, replacing this by the two-dimensional radiation condi-
tion. The result can be identified with the strip-theory synthesis, and a
superscript (s) is used to distinguish this solution.
260 J . N . Newman

In view of the boundary condition (3.37) on the body profile, the strip-
theory solution can be expressed in the form
cpp = + u$j
$Ij (6.7)
where, on the body profile,
+j,, = iconj, (6.8)
$.Jn = m i . (6.9)
The factors n j and mj are defined by (3.42-3.43) and (3.45-3.46), respec-
tively. The two-dimensional potentials 4j and $ j are governed by the
Laplace equation (3.44),satisfy the free-surface condition (3.52), and vanish
as 2 - -aoO. These boundary-value problems and their solutions do not
involve the forward velocity U.
The two-dimensional potentials $ j ( j = 1,2,3,4) are analyzed in Section
V,A. Analogous results can be derived for d j if the factors mj are known. In
practice a numerical solution is required not only for the two-dimensional
potentials, but also for the factors mj.Hereafter, we shall assume that these
are known. The remaining potentials for pitch ( j = 5 ) and yaw ( j = 6) follow
from the definitions of n j and mjr
cpp = - xcpy + (U/iw)&, (6.10)
cpt)= xcp$ - (U/iw)&. (6.11)
In general the matching requirement with the outer solution will differ
from the condition of outgoing radiated waves satisfied by the strip-theory
potentials. Moreover, the potentials (6.7)are unique (assuming unique solu-
tions of the two-dimensional radiation problems), and depend only on the
local cross-sectional geometry of the hull. Since the outer solution includes a
longitudinal function of x which depends on the three-dimensional shape of
the ship hull, a more general solution is required in the inner region.
In the classical slender-body theory of aerodynamics, and also in the
long-wavelength case where the rigid free-surface condition (3.47) holds in
the inner region, the inner solution is generalized simply by adding an
arbitrary constant C, which may depend on x. In the present case, the
free-surface condition (3.52) requires a nontrivial homogeneous solution.
Since the homogeneous solution satisfies a boundary condition of zero
normal velocity on the hull, it can be identified physically with the scattering
of incident waves by the fixed body. A solution which is symmetric about
y = 0 can be derived by combining two waves of equal phase incident from
opposite sides of the body. An antisymmetric solution can be derived sim-
ilarly, with incident waves of opposite phase. In either case standing waves
will exist in the far field of the two-dimensional problem.
An alternative derivation of the homogeneous solutions follows by
Theory of Ship Motions 261

observing that the boundary conditions (6.8) are purely imaginary, hence
is a homogeneous solution. Similarly, from (6.9), Im($j) is a homo-
geneous solution which will differ from Re(4j) by a multiplicative constant.
With an arbitrary multiplicative factor, only one homogeneous solution is
required in each mode, and we shall take this to be (4j+ Bj). With this
choice, the general form of the inner solution is given by
'pj = Cpy + Cj(X)(4j+ Bj). (6.12)
Here C j ( x ) is a function to be determined by matching with the outer
solution.
The outer approximation of the inner solution (6.12) can be derived from
the expansions (5.2) and (5.5). Since the radial derivatives of these expan-
sions are O(1) in the inner region, as r + O(E),
ajm= O(E'"'O~,
&'"pj). (6.13)
Thus, the wave-free potentials can be neglected for r B E , and the only con-
tribution to the outer approximation of the inner solution is from the source
or dipole.
For the potential 4 j ( j = 1, 2, 3, 4) the source strength 01,3or dipole
moment pz,4is defined by the two-dimensional solution. These definitions
are readily extended to the potentials for yaw and pitch where, in accordance
with (6.10-6.11), os = - x o 3 and p6 = x p z . With the same convention as in
(6.1), the outer approximation of the potential 4j is

(6.14)
A similar representation can be applied to the potentials $j,

G
-
DjGzD. (6.15)
For convenience in subsequent expressions we also define the differential
operators
S j = Dj + UDj, (6.16)
R j = Dj + Dj. (6.17)
With these definitions, the outer approximation of (6.12) is
q j z SjGzD + Cj(DjGzD + D j G z D )
= (Sj + C j R j ) G z D- iCjDjeKzcos K y . (6.18)
262 J . N . Newman

The last result follows from (4.3). The Fourier transform of (6.18)is given by
'pj* E [S? + (C,Rj)*]G2D- i(CjD,l)*eKzcos K y . (6.19)
The errors in the inner solution may be summarized in the following
manner for r & E. The two-dimensional Laplace equation and body bound-
ary condition involve second-order errors in the ratio of longitudinal to
transverse gradients, or a factor of 1 + O(kr)'. The two-dimensional free-
surface condition (3.52) contains an error factor 1 + O(K'I2kr). Neglect of
the wave-free potentials involves the error factor 1 + O(EZ/r2). Thus the
cumulative error in (6.19) is the factor
d = 1 + O(k2r2, K1I2kr,E2/r2). (6.20)

C. MATCHING

The inner and outer solutions are matched in a suitable overlap domain
E 6 r 4 1 to determine the unknown source strength and dipole moment of
the outer solution and the coefficients C , in the inner solution. Initially it will
be assumed that the overlap region is close to the ship, in terms of the
wavelength, and thus Kr 4 1. This assumption will become invalid for short
wavelengths, at which point a separate approach will be adopted.
Matching of the inner and outer solutions is carried out in the Fourier
domain. Thus we equate (6.3) to (6.19), and obtain the relation

- (S,+ CjRj)*G2D- i(CjBj)*eKzcos K y .


= (6.21)

The dominant terms in this matching relation are the antisymmetric contri-
butions associated with the dipole potential, of order l/r, and symmetric
contributions from the source potential, of order log r.
First we consider the antisymmetric terms in (6.21), corresponding to the
modes j = 2, 4, 6. Equating the factors of the dipole terms gives

dj* = [pj + Ufij + C j ( p j+ pi)]* ( j = 2,4, 6). (6.22)


The long-wavelength approximation (5.48) can be used with (5.7) to show
that the dipole moment p, is imaginary with the error a factor 1 + O(K2c2).
Thus the interaction term proportional to C , may be neglected in (6.22).
After inverting the Fourier transforms,
d, = p j + Ufi, ( j = 2,4, 6), (6.23)
Theory of Ship Motions 263

and the dipole moments are identical in the inner and outer solutions. The
remaining antisymmetric part of (6.21) is a higher order contribution from
the last term, proportional to sin Ky.
To leading order the antisymmetric solution is strictly two-dimensional in
the inner region and given correctly by the striptheory approach. This is a
familiar situation in slender-body theory, where lateral body motions with-
out a net source strength contain no longitudinal interactions. Ogilvie (1977)
refers to this as a primitive strip theory.
Next we consider the symmetric modes, which are dominated in (6.21)by
the source potential. Equating the factors of G z Dgives a relation for the
source strength
qj* = ( S j + CjRj)* ( j = 1, 3, 5). (6.24)

Equating the remaining terms in (6.21) of order one gives

qj*f* = 2ni(CjDj)* ( j = 1, 3, 5). (6.25)

A comparison of (6.4) and (6.20) indicates the cumulative error in (6.24)


and (6.25) to be the factor
d=1 + O(k2r2,K112r,K 2 r 2 ,c2/r2). (6.26)

The optimum location of the overlap region may be selected to minimize


this error.
For sufficiently small values of K, the dominant contributions to (6.26)are
O(K*r, E2/r2).The optimum value of r is such that these two errors are
equal and thus
,.= 0(&2/3~-(1/6) ). (6.27)

With this choice, the maximum error in (6.26) is O ( E ~ K ~provided ),


K &-(li2).
For K > E - ( / ~ ) , the dominant contributions to (6.26) are O ( K Z r 2E2/r2).
,
Equating these defines the optimum overlap region
r =O(E/K)~. (6.28)
The corresponding error in (6.26) is O(EK).
As K + O(E-l ) a separate matching must be constructed from the short-
wavelength inner approximation of the outer solution. Proceeding on this
basis with (4.18), (6.21) is replaced by

G z D2 ( S j + C j R j ) * G z D- i(CjDj)*eKzcos Ky. (6.29)


264 J. N . Newman

The solution of (6.29) is the strip-theory result

(6.30)

cj*= 0, (6.31)
in accordance with the short-wavelength analysis of Ogilvie and Tuck
(1969). The error factor in (6.29-6.31) is

8 = 1 + O(kzrz,K*r, c2/r2, ( K y ) - ) . (6.32)

The maximum error in (6.32)is O(K-(13)if the overlap region is defined by


r = O(K-(5/6)). Since the alternative error in the long-wavelength matching
is O(EK),the short-wavelength matching should be adopted if K 2 E - ( ~ / ~ ) .
At this stage we observe that the long-wavelength results (6.23-6.25) are
consistent with the striptheory results (6.29-6.31) for K 9 1, sincef* van-
ishes in accordance with (4.14). For this reason (6.23-6.25) are valid in
general, for all wavelengths such that K IO(E-), and will be used exclu-
sively hereafter.
The inverse transforms of (6.24-6.25) can be expressed in the form
qj = Sj + CjRj, (6.33)

2nicjDj = JL qj(<)f(x - <) d<, (6.34)

f o r j = 1,3,5. The kernelf(x) is the inverse transform of (4.13).Elimination


of C, gives an integral equation for the outer source strength:

= g,(x) + Ukj(X) ( j = 1, 3, 5). (6.35)


Here the two-dimensional source strengths have been substituted for the
differential operators using (6.14-6.17). The relations (5.4) and (5.32) can be
+ 1) by (1 - R - T),where R and Tare the
used to replace the factor ( 0 , / 8 ~
two-dimensional beam-sea reflection and transmission coefficients.
The kernel in (6.35) is identical to that which accounts for longitudinal
interactions in the long-wavelength slender-body theory (Newman and
Tuck, 1964). On the other hand, the source strengths 0, and 6, are derived
from the striptheory solution. Thus the integral equation (6.35) provides a
blend between the two limiting theories, such that the result is valid for all
intermediate wavelengths.
Theory of Ship Motions 265

The term containing the integral in (6.35)tends to zero for the two limit-
ing regimes K = 0(1) and K = O ( E - ) . For long wavelengths the factor
+
( a j / S j 1 ) = O ( K E )from (5.4) and (5.43).For short wavelengths the kernel
vanishes, in accordance with (4.14).To leading order it follows that
qj E bj + u&j. (6.36)
This approximation can be refined by iteration.

D. THEINNERSOLUTION

At this stage the inner solution (6.12) is determined. The two-dimensional


potentials 4j and t$j ( j = 1, 2, 3, 4) must be obtained numerically at each
section along the hull. The coefficients C j in (6.12) vanish for ( j = 2, 4, 6),
and are otherwise determined from (6.34) in terms of the outer source
strength qj. The latter is determined from the integral equation (6.35) or by
approximation from (6.36). The resulting inner solution can be expressed in
the form

The unified inner solution (6.37) is the principal result of our analysis.
This velocity potential is a linear superposition of the strip-theory solution
(6.7),and the homogeneous solution (4j + $ j ) . The homogeneous solution is
multiplied by the same longitudinal integral of the outer source strength
which appears in the long-wavelength slender-body theory.
In the long-wavelength regime K = 0(1),the inner free-surface condition
is the rigid-wall boundary condition (3.47). This governs the two-
dimensional potentials 4j and $, and thus qy).The homogeneous inner
solution is a constant and the last term in (6.37) is a function only of the
longitudinal coordinate x. The outer source strength qj is given explicitly in
terms of the net flux at each section, in accordance with (5.36) and (5.37).
Longitudinal interference is accounted for by the kernel f (x - t),but trans-
verse interactions are neglected. This is the ordinary slender-body theory
described by Newman and Tuck (1964) and by Ogilvie (1977).
In the short-wavelength regime K = O ( E - ~ the) , kernel is of higher order
and (6.37) is dominated by the potential cpy). Transverse interference is
accounted for in this two-dimensional potential, but longitudinal interac-
tions are negligible. This is the striptheory solution.
The unified potential (6.37) is valid for all wavenumbers K I O(E- ) . In
the special case of zero forward velocity this solution reduces to that derived
266 J . N . Newman

by Mays (1978) and outlined by Newman (1978). The latter reference also
treats an analogous problem in acoustic radiation.
The unified result may be compared with an interpolation solution
derived by Maruo (1970) for the case U = 0. Maruos approach is rather
different, but the only change in the final result is that the homogeneous
solution in (6.37) is replaced by (1 + K z ) , and the amplitude of the two-
dimensional striptheory potential is modified accordingly to satisfy the
boundary condition on the body. Maruos source strength is governed by an
integral equation similar to (6.35), with the kernel simplified by the restric-
tion to zero forward velocity.

VII. Slender-Body Diffraction

In this section we consider the diffraction problem for a slender ship, in


the presence of incident waves. With an incident wave of unit amplitude
defined by ( 3 . 3 9 the scattered potential q7 must be determined such that
the total potential po + q 7 satisfies the boundary condition of zero normal
velocity on the ship hull.
Our approach here is similar to that used for the radiation problems in
Section VI. The conditions in the opening paragraph of that section apply.
The principal difference is with respect to the boundary condition on the
ship hull, which for the scattered potential is given by
pTn = -iw,(n, - in, cos p - in, sin
p)
exp[K,(z - ix cos p - iy sin p)] on S. (74
Here oo is the incident wave frequency, in a fixed reference frame, and
K O = o o 2 / g is the corresponding wavenumber.
The diffraction problem is simplified by the absence of forward-speed
effects in (7.1), by comparison to the factors mjand the corresponding solu-
tions $ j in the radiation problem. On the other hand, the normal velocity
(7.1) is oscillatory along the ships hull, at a rate which is proportional to the
longitudinal wavenumber component K O cos p.
In the beam-sea case (cos b = 0),the right side of (7.1) is slowly varying
for all values of K O I O(E-). Thus the beam-sea diffraction problem can be
treated in an identical manner to the radiation problems, except that the
striptheory solution in the inner region corresponds to the two-dimensional
scattering problem.
The boundary condition (7.1) is simplified also for head or following seas
(sin = 0),but this simplification is deceptive. Head seas propagating along
Theory of Ship Motions 267

the two-dimensional body in the inner problem are diffracted over a trans-
verse width that increases without limit. Thus it is not possible to derive a
conventional striptheory solution in the inner region.
The singular nature of the head-sea diffraction problem was established
by Ursell (1968a,b). Ursells proof states that head seas cannot propagate
along an infinitely long cylinder in a periodic manner, unless the diffraction
potential is unbounded at large distances from the cylinder axis.
A detailed analysis of the head-sea diffraction problem has been carried
out by Faltinsen (1971) for the case where the incident wavelength is O(E).
To leading order the incident wave is canceled in the near field by an equal
and opposite longitudinal wave. Ursells unbounded solution is utilized in a
higher order inner solution, and matched with the outer solution in a con-
sistent manner. A singularity is encountered at the ships bow, of the sort
which generally occurs in short-wavelength scattering problems. Maruo and
Sasaki (1974) present a modified approach intended to remove this singular-
ity. Both solutions are discussed further by Ogilvie (1977, 1978).
For a ship moving in head seas with U = O(l), the frequency of encounter
o is increased by the Doppler shift (1.7). The regime of resonant pitch and
heave motions (1.8) coincides with incident wavelengths of order cl/, inter-
mediate in scale between the ships length and transverse dimensions. In this
regime the head-sea problem can be analyzed in a relatively simple manner
from the long-wavelength slender-body theory.
A solution of the diffraction problem will be derived from the unified
slender-body approach which was developed for the radiation problems in
Section VI. This theory is intended to apply for incident wavenumbers
K O IO ( E -), but we shall concentrate on the regime where K O I O(&-()).
Beam seas will emerge as a relatively simple limit, but for head seas the
unified solution is singular for all wavenumbers. The alternative long-
wavelength assumption will be used to provide a simple remedy for this
defect. A more fundamental extension of the unified theory is warranted to
include head seas, but this task is left for future research.

A. THEOUTERPROBLEM

Since the scattering potential (p7 differs from the radiation potentials of
Section VI only with respect to the body boundary condition, the boundary-
value problems in the outer region are identical. The outer solution (6.1)is
applicable directly to the potential (p7, with unknown source strength q , and
dipole moment d 7 . The Fourier transform of the outer scattering solution is
given by (6.2), with j = 7.
268 J . N . Newman

B. THEINNER PROBLEM

The oscillatory longitudinal factor of the boundary condition (7.1) sug-


gests expressing the inner solution for the diffraction potential in the form
(p7 = Q, exp( - iKox cos fl) = Q exp( - dox). (74
After substituting in (7.1), and neglecting the longitudinal component n,, the
function Q, satisfies the boundary condition
On= -iwo(n, - in, sin fl) exp[Ko(z - iy sin fl)] on S. (7.3)
Since (7.3) is slowly varying in the x-direction, the same behavior is expected
of Q,. Substituting (7.2) in the three-dimensional Laplace equation and ne-
glecting longitudinal derivatives of Q then gives

a,, + - lo% = 0. (7.4)


Thus Q, is governed by the two-dimensional Helmholtz equation. The
leading-order free-surface equation can be derived from (3.49) or (3.52),

KoQ,-Q,=O onz=O. (7.5)


Here (1.7) has been used to replace the frequency of encounter by the
incident-wave frequency oo, and K O = wo2/g.
The problem defined by (7.3-7.5) is identical to the two-dimensional dif-
fraction potential Q7 derived in Section V,B, except for the appearance of the
frequency oo in place of o.Thus the striptheory solution of the inner
problem is the solution of the two-dimensional diffraction problem, which
will be denoted here by Q7. The general solution of the inner problem is
given in a form analogous to (6.12):

@ = @7 + (74
Since the real and imaginary parts of the boundary condition (7.3) are
antisymmetric and symmetric, respectively, the general homogeneous solu-
tion is given by
@(h) = C,(@, + as)+ C,(@, - a,). (7.7)
Here

and (Cs,
C,) are arbitrary constants.
Theory of Ship Motions 269

The outer approximation of this inner solution can be obtained from


(5.16) in the form

G*(y, Z, lo, KO) + CsE7 - C , R 7 (C*- G*). (7.9)

After using (4.15) to evaluate the last factor, and taking the Fourier trans-
form, it follows that

C s z 7- C a M , -)aYa * eKoz cos(Koy sin B). (7.10)

This outer approximation of the inner scattering potential is analogous to


(6.19) for the radiation problem. The error in (7.10) is the factor (6.20), with
K O substituted for K , and with the Fourier parameter k replaced by lo if
lo > k. Since (D is slowly varying with respect to x, the significant domain for
(7.10) is k = O(1).

C. MATCHING

The Fourier transform (7.10) of the inner solution is to be matched with


the transform (6.2) of the outer solution. Recalling the oscillatory factor in
(7.2), it is necessary to shift the transform parameter from k to k - lo in
(7.10), or alternatively to shift from k to k + lo in the outer solution. Adopt-
ing the latter approach, (6.2) is rewritten in the form
(P7*(Y,z ; Q = [q7*(jt)+ d,*(Q(a/ay)lG*(y, 2; IT, 3, (7.11)
where
K=k+lO (7.12)
and
K = (0 - UE)/q = (00 - Uk)/g. (7.13)
Matching of the inner and outer solutions is performed by equating (7.10)
and (7.11) in a suitable overlap region. Since the transforms G* in these two
270 J . N . Newman

solutions contain different arguments, it is not possible to match their fac-


tors directly. Instead, the long-wavelength inner approximation (4.12) of G*
is used in the solutions, with (4.4), to derive the relation
1
- q,*(E)(l + Kz)[log(Kr) + y + ni -f*(& II-, K) - Kz + KyO]
2n
1
+- d,*(E)(l + Kz)[(sin O)/r + KO]
271
1
z - [C, + Cs(C7+ X,)]*(l + K,Z)
2n
[log(Kor) + y + ni - f * ( l o , K O ,K O )- Koz + KoyO]

+-1
2n
[ M , + C,(M7 - &f7)]*(1 + Koz)[(sin O)/r + KoO] - ilcsc flleKoZ

. [(Cs&)* cos(Koy sin fi) + (C,l\si,)*(Ko sin fi) sin(Koy sin /I)].
(7.14)
Proceeding as in Section VI,C, the antisymmetric terms in (7.14) are
matched by neglecting the higher-order interactions, proportional to C,,
with the result
d7*(E) = M7*(k). (7.15)
After inverting this Foilrier transform the outer dipole moment is given by
d 7 ( x ) = M 7 ( x )exp( - ilox). (7.16)
Once again there is no interaction in the antisymmetric solution, and the
only difference between the inner and outer dipole moments is the oscilla-
tory factor in the definition (7.2) of the inner solution.
The dominant symmetric terms in (7.14) are proportional to log r. Equat-
ing these,
q7*(Q = [C7 + CS(C7 + %)I* (7.17)
Using this result, the remaining symmetric terms in (7.14) can be equated to
give the relation
274CsC7)* = 47*(l;)f7*(k9 lo, 4, (7.18)
where from (4.13),

[
f,*(k, lo, k) = 1 sin fl I log( I l o / E l ) - ( 1 1 - R / i 2I ) - ( ' I 2 )

- II
cos-'(i?/~E~)
(cosh-l(i/ I + xi sgn(oo- U k )
+ [cosh-'(sec /3) + nil. (7.19)
Theory of Ship Motions 27 1

Equations (7.17-7.18) are analogous to (6.24-6.25) in the radiation


solutions.
The error factor in (7.14-7.18) can be deduced in a similar manner to
(6.26). With k = O(l), the error is a factor
I = 1 + O(KO2r2,&'/r2). (7.20)
The optimum overlap region is defined by r = O ( E / K ~ )with ~ ' ~this
; choice
the error factor is 1 + O(&KO).A separate analysis must be performed for
K O-+ O ( E - ' ) ,using (4.17) to approximate G* in (7.10) and (7.11). We shall
not repeat that analysis here, since the results are similar to the radiation
solutions, and serve only to confirm the validity of (7.15-7.18) for the ex-
tended regime KO < O ( E - ' ) .
The Fourier transforms (7.17) and (7.18) can be inverted, after noting the
shift in the parameter (7.12). The result is a pair of simultaneous equations,

q7 = [C, + C,(C7 + C,)] exp(-il,.x), (7.21)

(7.22)

wheref7(x) is defined by its Fourier transform (7.19). From (7.21), the outer
source strength can be expressed in a similar form to (7.16),
q7(x) = Q7(x) ex~(-ilox), (7.23)

with Q7 a slowly varying unknown function of x.


Elimination of C , from (7.21) and (7.22) gives an integral equation for Q 7 ,

Q ~ ( x-) ( 2 4 - ' ( & / x 7 + 1) '[L Q 7 ( 5 ) f 7 ( ~ - 4 ) d4 = C7(x). (7.24)

This integral equation is similar in form to (6.35). Once again the integral
term in (7.24) vanishes in the limits K OE 4 1 and K O B 1 so that to leading
order (7.24) is approximated by
Q7b) = C7(x). (7.25)

D. THEINNERSOLUTION

The solution in the inner region follows from (7.6-7.8), with the coefficient
C, determined from (7.22) and the (antisymmetric) coefficient C , = 0. The
result is of the form
272 J . N . Newman

Here @, is the solution of the two-dimensional diffraction problem outlined


in Section V,B, X7 is the corresponding source strength, and QS is the sym-
metric part of @., In all cases the incident-wave frequency wo and wavenum-
ber K O apply to the two-dimensional solution. In view of the definitions
(7.12-7.13) and (7.19), this inner solution is not dependent on the frequency
of encounter o,or the corresponding wavenumber K , but depends instead
on the frequency wo and wavenumber KO.A physical interpretation for
this distinction is given by Newman (1977).
For short wavelengths the kernel in (7.26) vanishes, and the striptheory
result follows. This two-dimensional solution has been studied by Bolton
and Ursell (1973), Choo (1975), and Troesch (1976).

E. THELONGWAVELENGTH
SOLUTION

Since the solution @, of the two-dimensional diffraction problem, is sin-


gular in head or following seas (fl= 180" or 0", respectively) the unified
solution (7.26) is not well behaved in these limits.
In the long-wavelength regime this difficulty does not arise. The essential
change is to simplify the inner boundary-value problem by assuming that lo
and K O are much less than E - ' in (7.4) and (7.5). The two-dimensional
Laplace equation (3.44) and rigid free-surface condition (3.47) follow. The
source potential for tbe inner solution is (271)- log r, and the homogeneous
solution is a constant.
With these simplifications the outer approximation of the inner solution
(7.9) is replaced by
1
@ 2 - (Z,
271
+ M, $) +
log r C,(x), (7.27)

with the source strength C7 and dipole moment M , given by(5.41)and (5.42).
Matching follows by equating the Fourier transform of (7.27) to the left
side of (7.14). The relation for the outer dipole moment (7.15-7.16) is un-
changed, but (7.17-7.18) are replaced by
47*(lt) = Z 7 * ( k ) , (7.28)
2nC,*(k) = q,*(Q[log(K) + y + xi - f * ( & 2, K)]. (7.29)
After inverting these transforms and substituting the source strength from
(5.41), it follows that
q7(x)= - 2iw0 B ( x ) exp( - ilox), (7.30)

2nC7(x)exp( - ilox) = q7(x)[log K + y + nil - L


q7(5)f(x - t) d5. (7.31)
Theory of Ship Motions 273

Here the kernelfis the inverse transform of (4.13)' as in (6.34-6.35).


As intended, the source strength q , and constant C, are well behaved in
(7.30-7.31) as sin fi + 0, and these results can be utilized for head and
following seas. On the other hand, the factor log K in (7.31) suggests
difficulties for large wavenumber. This is a defect of the long-wavelength
slender-body theory which is absent from the unified results.

VIII. The Pressure Force

The six components of the pressure force (Fl, F,, F 3 )and moment (F4,
F s , F 6 ) can be expressed in the form

F, = ljpn, dS
S
(i = 1, 2, . . . , 6). (8.1)

Here S is the submerged portion of the ship hull, and the factors n, are
defined by (3.38-3.39). The pressure p is determined from Bernoulli's equa-
tion (3.2), and can be separated into hydrodynamic and hydrostatic
components.
The hydrostatic pressure force is analyzed by substituting p = -pgz in
(8.1). The nonvanishing components (i = 3 , 4 , 5) are analyzed by Wehausen
(1971) and Newman (1977). The results are simplified for linearized motions
of a ship hull which is symmetrical about y = 0. Further simplification
follows by adding the gravitational force and moment due to the ship's mass,
and assuming equilibrium when the unsteady motions vanish. The latter
assumption implies that the ship's weight is balanced by the buoyancy force
pgV and that the horizontal coordinates of the center of gravity and center of
buoyancy are coincident. With these restrictions, the force components due
to the sum of the hydrostatic pressure and the ship's weight are given by

F3 = -pg(t3SOO - tSSlO)e'"', (8.2)


F4 = -Pg[So, + ~ ( Z B- Z G ) ] C ~ ~ ' ~ ' , (8.3)
FS = pg{s10t3 - + v(zB -
[sZO zG)]<5}ei0'. (8.4)
Here zB and zG denote the vertical coordinates of the centers of buoyancy
and gravity, and

sij = ljX'yJ dx dy,


where the integral is over the plane z = 0, interior to the ship hull.
274 J . N . Newman

The steady component of the hydrodynamic pressure follows by substitut-


ing the velocity field W(x) in Bernoullis equation. From symmetry, the only
contributions to (8.1) are for (i = 1, 3, 5). The steady portion of Fl is the
(negative) wave resistance, which is balanced by the propulsive force. The
steady vertical force and pitch moment are balanced by static sinkage and
trim, which modify the equilibrium position of the ship hull but d o not
contribute directly to the unsteady force or motions.
Hereafter we consider the unsteady component of the hydrodynamic pres-
sure force, with the usual assumption that the oscillatory motions of the ship
and the fluid are small. Neglecting second-order terms in Bernoullis equa-
tion (3.2), the pressure is given by

An additional contribution to (8.1) results from the oscillatory position of


the ships surface S with respect to the mean surface S. From a Taylor-series
expansion the total oscillatory pressure in (8.1) is

p = -p(q, + w . vq + +x . vwz)s. (8.7)


The last term in (8.7) gives a force proportional to the unsteady displace-
ment of the ship, and hence an additional contribution to the restoring
coefficients cij in (1.2-1.3). This force is due to the unsteady motion of the
ship within the steady pressure field. A similar contribution from the oscilla-
tory change in the upper boundary of S is noted by Timman and Newman
(1962).
For a slender ship, the last term of (8.7) is approximated by

The resultant force from this change in pressure is O ( E )relative to the


restoring-force components (8.2-8.4). For horizontal translation the contri-
bution from (8.8) is zero and the only first-order effect from (8.8) is a sway
force, and moment, due to a static yaw angle. The latter are analogous to the
aerodynamic lift force and moment on a slender body, with important effects
on the low-frequency steering maneuvers of ships. For oscillatory motions in
waves these generally are neglected.
The remaining contributions to the linearized unsteady pressure force are
due to the added-mass and damping coefficients defined in (1.2) and the
exciting force defined by (1.4). These are discussed separately below, using
the solutions for the radiation and diffraction potentials derived in Sections
VI-VII.
Theory of Ship Motions 275

A. ADDED MASSAND DAMPiNG

With the notation of (l.l), our task is to evaluate the transfer function
t . .= - p
IJ jj(ioqj+ W . Vqj)nidS. (8.9)
Here Bernoullis equation has been used in the form (8.7), together with
(3.34) and (8.1). The real and imaginary parts of (8.9) give the added-mass
and damping coefficients defined by (1.2).
The term in (8.9) proportional to the steady velocity field W can be
transformed by means of a theorem due to Tuck (Ogilvie and Tuck, 1969,
Appendix A),
11
S
(W . V q j ) n ,dS = - U 1s q j m i d S
S
- U fqj&ni
C
dl. (8.10)

This result follows from Stokes theorem, and the fact that V W = 0. The -
last integral in (8.10) is over the boundary of S, i.e., the intersection of the
ship hull with the plane z = 0. Since the rigid free-surface condition (3.47)l
applies to &, the line integral is of higher order and can be ignored.
Substituting (8.10) in (8.9) and using (6.7) and (6.12) for the unsteady
potential q j , it follows that

t I.J. = - p jj(ion, - Umi)[c$j+ U J j + C j ( 4 j+ Bj)] dS. (8.11)


S

The two-dimensional potentials 4j and $ j are defined in Section VI,B, and


the interaction coefficient Cj( x)is determined from the source strength of the
outer solution by (6.34).
Equation (8.11) may be rewritten in the form

where
(8.13)

T!;)= - p U - mi4j) dl, (8.14)


P

TI; = pU2 mi$j dl, (8.15)


P

T $ ) = -pCj 1 (ioni
P
- Umi)(+j+ $ j ) dl. (8.16)
276 J . N . Newman

From Green's theorem (5.21), and the boundary conditions (6.8-6.9),


TIIO.') = T$)g2), (8.17)

(8.18)

TI;) = (- 2ipgAjCj/K)(Ai + UA,). (8.19)


Equation (5.32) can be used with (6.34) and (6.37) to show that the contribu-
tion from (8.19) to the integrated force (8.12) is symmetric with respect to the
indices i and j. Thus the total three-dimensional force (8.12) satisfies the
reverse-flow theorem of Timman and Newman (1962), i.e., ti;) = t);), where
the superscript denotes the direction of the forward velocity.
The local force (8.13) can be related to the two-dimensional added-mass
and damping coefficients (5.9), and the contribution to (8.12) is the zero-
speed striptheory result. The contribution from (8.15) is similar, but re-
quires a separate analysis of the potentials 4j.There is no contribution from
(8.16) to the fluxless modes (j = 2, 4, 6) where C j = 0. From (8.18) the
integrals defining TI;) vanish for i = j , and give nonzero contributions only
for the cross-coupling coefficients.
For coupling between heave ( j = 3) and pitch ( j= 5), we note from(3.43)
and (3.46b) that 4, = -x&, and m, = -xm3 + n3. From (8.18) it follows
that
7''')
35 -
- - " ( 15 )3 -
- (W4TS"J. (8.20)
By a similar argument, for sway and yaw,

T':Q = - Thy = - (U/io)T$"J. (8.21)

Thus these contributions to the cross-coupling coefficients can be expressed


in terms of the "zero-speed" potentials 4j.
In the short-wavelengthstriptheory regime, the interaction coefficientsC j
vanish, in accordance with (6.31). This leaves only the integrals (8.13-8.15)
to be considered. For heave and pitch there are two variations of the result-
ing formulas. In the intuitive approach, gradients of the steady-state velocity
field are neglected with the result that the only nonzero elements of mi are
m, = n3 and m6 = -nz. Thus the two-dimensional potentials 4, vanish
except for iw4, = and = -&. With these simplifications in
(8.12-8.15),
(8.22)

(8.23)
Theory of Ship Motions 277

It
b53 I- J;.
- - T&Jxdx & ( U / i o ) t 3 , , (8.24)

t55 = S, T&Jx dx + p(U/o)t33, (8.25)

t66 = jL T$iX2 dx -t p(U/W)2t22. (8.26)

These equations are essentially? identical to the striptheory results for


heave and pitch derived by Salvesen et al. (1970) and by Newman (1977).
Somewhat different results are obtained in the systematic approach of
Ogilvie and Tuck (1969),as noted in the Introduction. With o = O(&-()),
the moment proportional to (U/w) is discarded as a higher-order effect.
Additional cross-coupling terms are added to (8.24) from the free-surface
boundary condition (3.50). The heave force t 3 , is unchanged from (8.22).
The comparative merits of these different approaches to strip theory have
not been resolved, although Faltinsen (1974) has shown that the comparison
with experiments is improved by using the Ogilvie-Tuck cross-coupling
coefficients.
The results (8.12-8.16) are not limited by the strip-theory assumptions
and apply more generally for all frequencies and wavelengths. The principal
4,
complication is that the forward-speed potentials and the factors mjmust
be determined, as well as the interaction coefficients Cj. Such calculations
have not been performed in the general case U # 0.
For zero forward velocity (V = 0), calculations of the heave and pitch
coefficients by Mays (1978) show good agreement with exact three-
dimensional computations for slender spheroids with E < $. Computations
of the added mass and damping at zero velocity also have been performed by
Maruo and Tokura (1978), based on the interpolation solution of Maruo
(1970) which is described in the closing paragraph of Section VI. Maruo and
Tokura show good agreement of their results with experimental data. The
difference between Maruos theory and the unified solution is of little practi-
cal significance for U = 0, and both offer substantial improvement relative
to the ordinary slender-body theory and the strip theory.

B. THEEXCITING
FORCE

The exciting force (1.4) is the result of the pressure associated with the
diffraction potential. The coefficient X i in (1.4) can be interpreted as the

t In Salvesen et al. (1970), a transom-stem correction is introduced for ships where the
after end of the hull is not pointed; the validity of this correction is questionable. In Newman
(1977) a different coordinate system is used, and the expression for the pitch moment contains
an error in the sign of the term proportional to (V/o)*.
278 J . N. Newman

complex amplitude of the exciting force due to an incident-wave system of


unit amplitude. With the diffraction potential substituted in (8.9), this
coefficient is given by
xi= - P \jni(io + W V)(cpo +
* (p7) dS
s
= -P jjs (ioni - Umi)(cpo + ( ~ 7d
)s, (8.27)

where the last form follows from (8.10).


A direct evaluation of the exciting-force coefficient follows by substituting
the inner solution (7.26) for the diffraction potential:

Xi = -p j exp(il,x) dx
L P
(iconi - Urni)[@, + O7+ C,(x)(Os + a,)]dl.
(8.28)
The contributions proportional to the factors ni can be evaluated directly
from the two-dimensional zero-forward-speed exciting force (5.30). The con-
tributions from the factors mi depend on the steady-state solution, and the
interaction coefficient C , is dependent on the forward speed. For the fluxless
modes (i = 2, 4, 6), and also in the short-wavelength regime, C , does not
contribute to (8.28) and the exciting force coefficients are linear functions of
the forward velocity.
Since Stokes theorem has been used in the last form of (8.27), the inte-
grand in (8.28) cannot be interpreted as the local force. To emphasize this
distinction we note the special case of a long parallel middle body where
the ship hull is cylindrical and, in the inner region, W = - Ui. Neglecting the
interaction coefficient, the only effect of the gradient operator in (8.27) is on
the oscillatory factor exp(ilox ) , with the result that (iw + W - V) = io,,and
the local exciting force is independent of the forward speed. This is
confirmed physically by the fact that a long cylinder may be moved axially
(in an inviscid fluid) without affecting the local pressure field except near the
ends. In this connection we recall the discussion following (7.26).
Greens theorem can be applied to the three-dimensional potentials in
(8.27), or alternatively to the two-dimensional functions in (8.28). In both
cases the result is a form of Haskinds relations, with the solution of the
diffraction problem replaced by an appropriate radiation solution.
To derive Haskinds relations in the three-dimensional form, following
Newman (1965), the boundary conditions (3.37) are combined with (8.27) to
give
xi = - P jf
s
CPG)((P~ + ( ~ 7 d) s . (8.29)
Theory of Ship Motions 279

Here the superscript (- ) denotes the reverse-flow solution of the radiation


problem, with negative forward velocity. This radiation potential satisfies
the free-surface condition (3.51) with U replaced by - U . After an applica-
tion of Greens theorem to (8.29) with the boundary condition (3.36)
imposed,
Xi = - p Jj (cp{,cpo
- cp,cp~-) dS. (8.30)
s
Here a line integral similar to the last term in (8.10) is neglected.
In this form, the total exciting force on the ship is expressed in terms of the
solution of the radiation problem. The inner solution for cpj- may be sub-
stituted, with the integral performed over the mean surface of the ship hull.
Alternatively, this integral can be performed over a closed surface at large
distance from the ship, where the far-field asymptotic form of the radiated
waves may be used.
McCreight (1973) has extended the Haskind relation (8.30) to the strip-
theory regime of Ogilvie and Tuck (1969), including the higher-order terms
in the free-surface condition (3.50). Equation (8.30) is unchanged, but the
integral cannot be evaluated at infinity due to the inhomogeneous free-
surface condition in the higher-order solution. McCreight notes that the
leading-order contribution to (8.30) is the Froude-Krylov force. The next
term in a systematic perturbation expansion can be expressed in terms of
the zero-speed two-dimensional potential 4.
If Greens theorem (5.21) is applied to the two-dimensional functions in
(8.28), we find after a reduction using (5.33) that

Xi = i ( p g / K , ) I sin fl I exp(il,x)(& - U j i )
L

. [l + C,(AIi/zi - Ji/Ji)]dx (i = 1, 3, 5), (8.31)


and
I

X i = -i(pg/K,) sin B exp(il,x)(& - U A , ) dx ( i = 2, 4, 6). (8.32)


JL

Here the local exciting force at each section is expressed in terms of the wave
amplitude of the generalized radiation function &i and the corresponding
forward-speed function &. $or the antisymmetric modes the exciting force
follows directly from (8.32) but for the symmetric modes (i = 1, 3, 5) one
must determine the interaction coefficient C , of the diffraction problem, by
solution of (7.21-7.22).
If the inner solution of the three-dimensional radiation problem is sub-
stituted in (8.30), the result is an integral along the length involving two-
dimensional solutions of Laplaces equation. By comparison, the integrands
280 J. N. Newman

of (8.31) and (8.32) contain solutions of the two-dimensional Helmholtz


equation. It is not obvious that these alternative integrals for the total excit-
ing force will be equivalent, even in the asymptotic sense for small values of
the slenderness parameter. Computations of the heave and sway exciting
forces have been made by Troesch (1976) for the case U = 0, using these two
alternative expressions; the results are in satisfactory agreement with each
other and with experiments.
In the intuitive strip theory initiated by Korvin-Kroukovsky (1955), thZ
local exciting force at each section is derived from a relative-motion
assumption. This states that the exciting force can be expressed in the same
form as the pressure force of the radiation problem, but with the ships
velocity and acceleration replaced by the relative motions of the incident
wave, at a suitable mean depth. (In accordance with G. I. Taylors theorem,
there is an additional component of the exciting force, equal to the product
of the incident-wave acceleration and the mass of fluid displaced by the hull.)
In this relative-motion approach the pressure force of the radiation prob-
lem is expressed in terms of the added-mass and damping coefficients, eval-
uated at the frequency of encounter w. Similar expressions for the exciting
forces are derived by Salvesen et al. (1970), using the two-dimensional Has-
kind relations. In modified results derived by Newman (1977) the added-mass
and damping coefficients are evaluated at the incident-wave frequency wo.
An argument in favor of this modified approach is that, along a parallel
middle body, the exciting force is independent of the forward velocity of the
ship.

REFERENCES

ABRAMOWITZ, M., and STEGUN,I., eds. (1964) Handbook of Mathematical Functions. U.S.
Gov. Print. ON., Washington, D.C.
BABA,E., and HARA,M. (1977). Numerical evaluation of a wave-resistance theory for slow
ships. Proc. Int. Con$ Numer. Ship Hydrodyn., 2nd. pp. 17-29. Univ. California, Berkeley.
BAI,K. J., and YEUNG,R. W.(1974). Numerical solutions to free-surface flow problems. Proc.
Symp. Nav. Hydrodyn., 10th ACR-204. pp. 609-647. ON. Nav. Res., Washington, D.C.
BECK,R. F., and TUCK,E. 0. (1972). Computation of shallow water ship motions. Proc. Symp.
Nav. Hydrodyn., 9th ACR-203, pp. 1543-1587. ON. Nav. Res., Washington, D.C.
BISHOP,R. E. D., and -Ice, W. G., ads. (1975). Proc. lnt. Symp. Dyn. Mar. Vehicles Struct.
Waoes. Inst. Mech. Eng., London.
BISHOP,R. E. D., BURCHER, R. K., and PRICE,W. G. (1973). The uses of functional analysis in
ship dynamics. Proc. R.SOC.London, Ser. A 332,23-35.
BOLTON, W. E., and URSELL, F.(1973). The wave force on an infinitely long circular cylinder in
an oblique sea. J . Fluid Mech. 57, 241-256.
BRARD,R. (1948). Introduction a letude theorique du tangage en marche. Bull. Assoc. Tech.
Marit. Aeronaut. 47, 455479.
BRARD,R. (1973). A Mathematical Introduction to Ship Maneuverability, Rep. No. 4331.
Nav. Ship Res. Dev. Cent., Bethesda, Maryland.
Theory of Ship Motions 28 1

CHANG,M.4. (1977). Computations of three-dimensional ship motions with forward speed.


Proc. Int. Con/: Numer. Ship Hydrodyn., 2nd, pp. 124-135. Univ. California, Berkeley.
CHAPMAN, R. B. (1975). Numerical solution for hydrodynamic forces on a surface-piercing plate
oscillating in yaw and sway. Proc. Int. Con/: Numer. Ship Hydrodyn., l s t , pp. 333-350.
David W. Taylor Nav. Ship R & D Cent., Bethesda, Maryland.
CHAPMAN, R. B. (1977).Survey of numerical solutions for free-surface problems. Proc. Int. Cot$
Numer. Ship Hydrodyn., Znd, pp. 5-16. Univ. California, Berkeley.
CHOO,K.Y. (1975). Exciting forces and pressure distribution on a ship in oblique waves. Ph.D.
Thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts.
CUMMINS, W. E. (1956).The wave resistance of a floating slender body. Ph.D. Thesis, American
University, Washington, D.C.
CUMMINS, W. E. (1962). The impulse response function and ship motions. Schiffstechnik 9,
101- 109.
FALTINSEN, 0. (1971). Wave forces on a restrained ship in head-sea waves. Ph.D. Thesis,
University of Michigan, Ann Arbor.
FALTINSEN, 0. (1974). A numerical evaluation of the Ogilvie-Tuck formulas for added mass
and damping coefficients. J. Ship Res. 18, 73-85.
FROUDE,W. (1861). On the rolling of ships. Inst. N a n Archit., Trans. 2, 180-229.
GERRITSMA, J. (1976). A note on the application of ship motion theory. Schiffstechnik 23,
181-185.
GERRITSMA, J., KERWIN,J. E., and NEWMAN, J. N. (1962). Polynomial representation and
damping of Series 60 hull forms. Int. Shipbuilding Prog. 9, 295-304.
GRIM,0. (1960). A method for a more precise computation of heaving and pitching motions
both in calm water and in waves. Proc. Symp. Nav. Hydrodyn., 3rd ACR-65, pp. 483-524.
Off. Nav. Res., Washington, D.C.
HASKIND, M. D. (1946a). The hydrodynamic theory of ship oscillations in rolling and pitching.
Prikl. Mat. Mekh. 10, 33-66. (Engl. transl., Tech. Res. Bull. No. 1-12, pp. 3-43. SOC.Nav.
Archit. Mar. Eng., New York, 1953.)
HASKIND, M. D. (1946b). The oscillation of a ship in still water. Izv. Akad. Nauk SSSR, Otd.
Tekh. Nauk 1,23-34. (Engl. transl., Tech. Res. Bull. No. 1-12, pp. 45-60. SOC.Nav. Archit.
Mar. Eng., New York, 1953.)
HAVELOCK, T. H. (1929). Forced surface-waves on water. Philos. Mag. [7] 8, 569-576.
HAVELOCK, T. H. (1958). The effect of speed of advance upon the damping of heave and pitch.
Inst. Nav. Archit., Trans. 100, 131-135.
HAVELOCK, T. H. (1963). Collected Works, ACR-103. Off. Nav. Res., Washington, D.C.
JWEN, W. P. A. (1964). Oscillating Slender Ships at Forward Speed, Publ. No. 268. Neth.
Ship Model Basin, Wageningen.
KELLER, J. B. (1978). The ray theory of ship waves and the class of streamlined ships. J. Fluid
Mech. (in press).
KORVIN-KROUKOVSKY, B. V. (1955). Investigation of ship motions in regular waves. SOC.Nav.
Archit. Mar. Eng., Trans. 63, 386-435.
KORVIN-KROUKOVSKY, B. V., and JACOBS, W. R. (1957). Pitching and heaving motions of a ship
in regular waves. SOC. Nav. Archit. Mar. Eng., Trans. 65, 590-632.
KRILOFP, A. (1896). A new theory of the pitching motion of ships on waves, and of the stresses
produced by this motion. Inst. Nav. Archit., Trans. 37, 326-368.
LANDwEBeR, L., and YIH,C. S. (1956). Forces, moments and added masses for Rankine bodies.
J. Fluid Mech. 1, 319-336.
LEWIS,F. M. (1929). The inertia of water surrounding a vibrating ship. SOC.Nav. Archit. Mar.
Eng., Trans. 37, 1-20.
LIGHTHILL, M. J. (1967). On waves generated in dispersive systems by travelling forcing effects,
with applications to the dynamics of rotating fluids. J. Fluid Mech. 27, 725-752.
282 J . N.Newman

MCCREIGHT, W. R. (1973). Exciting forces on a moving ship in waves. Ph.D. Thesis, Massa-
chusetts Institute of Technology, Cambridge, Massachusetts.
MARUO,H. (1967). Application of the slender body theory to the longitudinal motion of ships
among waves. Bull. Fac. Eng., Yokohama Natl. Uniu. 16, 29-61.
MARUO,H. (1970). An improvement of the slender body theory for oscillating ships with zero
forward speed. Bull. Fac. Eng., Yokohama Natl. Uniu. 19, 45-56.
MARUO, H., and SASAKI, N. (1974). On the wave pressure acting on the surface of an elongated
body fixed in head seas. J . SOC. Nov. Archit. Jpn. 136,3442.
MARUO,H., and TOKURA, J. (1978). Prediction of hydrodynamic forces and moments acting on
ships in heaving and pitching oscillations by means of an improvement of the slender ship
theory. J . SOC.Nau. Archit. Jpn. 143, 111-120.
MAYS,J. H. (1978). Wave radiation and diffraction by a floating slender body. Ph.D. Thesis,
Massachusetts Institute of Technology, Cambridge, Massachusetts.
MEI,C. C. (1977). Numerical methods in water-wave diffraction and radiation. Annu. Rev. Fluid
Mech. 10, 393-416.
MICHELL, J. H. (1898). The wave resistance of a ship. Philos. Mag. [5] 45, 106-123.
NEWMAN, J. N. (1961). A linearized theory for the motion of a thin ship in regular waves. J . Ship
Res. 3:(1), 1-19.
NEWMAN, J. N. (1964). A slender-body theory for ship oscillations in waves. J . Fluid Mech. 18,
602-618.
NEWMAN, J. N. (1965). The exciting forces on a moving body in waves. J . Ship Res. 9, 190-199.
NEWMAN, J. N. (1976). The interaction of stationary vessels with regular waves. Proc. Symp.
N a n Hydrodyn., 11th pp. 491-501. Mech. Eng. Publ., London.
NEWMAN, J. N. (1977). Marine Hydrodynamics. MIT Press, Cambridge, Massachusetts.
NEWMAN, J. N. (1978). Wave radiation from slender bodies. Proc. Symp. Appl. Math. Dedicated
to the Late Prof: Dr. R . Timman pp. 101-115. Sijthoff & Nordhoff, Groningen.
NEWMAN, J. N., and TUCK,E. 0. (1964). Current progress in the slender-body theory of ship
motions. Proc. Symp. Nao. Hydrodyn., 5th ACR-I 12, pp. 129-167. Off. Nav. Res., Washing-
ton, D.C.
OAKLEY, 0. H., Jr., PAULLING, J. R., and WOOD,P. D. (1974). Ship motions and capsizing in
astern seas. Proc. Symp. Nau. Hydrodyn., 10th ACR-204, pp. 297-350. Off. Nav. Res.,
Washington, D.C.
OGILVIE,T. F. (1964). Recent progress toward the understanding and prediction of ship mo-
tions. Proc. Symp. Nau. Hydrodyn., 5th ACR-112, pp. 3-128. Off. Nav. Res., Washington,
D.C.
OGILVIE, T. F. (1967). Nonlinear high-Froude-number free-surface problems. J . Eng. Math. 1,
2 15-235.
T. F. (1977). Singular-perturbation problems in ship hydrodynamics. Adu. Appl. Mech.
OGILVIE,
17, 91-188.
OGILVIE, T. F. (1978). End effects in slender-ship theory. Proc. Symp. Appl. Math. Dedicated
to the Late Prof: Dr. R . Timman pp. 119-139. Sijthoff & Nordhoff, Groningen.
OGILvIE, T. F., and TUCK,E. 0. (1969). A Rational Strip Theory for Ship Motions, Part 1,
Rep. No. 013. Dep. Nav. Archit. Mar. Eng., University of Michigan, Ann Arbor.
PETERS, A. S., and STOKER, J. J. (1957). The motion of a ship, as a floating rigid body, in a
seaway. Commun. Pure Appl. Math. 10, 399490.
PRICE,W. G., and BISHOP, R. E. D. (1974). Probabilistic Theory of Ship Dynamics. Chapman
& Hall, London; Wiley (Halsted), New York.
ST. DENIS,M., and PIERSON, W. J. (1953). On the motion of ships in confused seas. SOC.Nau.
Archit. Mar. Eng., Trans. 61, 280-354.
SALVESEN, N., TUCK,E. O., and FALTINSEN, 0. (1970). Ship motions and sea loads. SOC. Nau.
Archit. M a r . Eng., Trans. 78, 250-287.
Theory of Ship Motions 283
SOCIETY OF NAVAL ARCHITECTSAND MARINE ENGINEERS (1974). Seakeeping 1953-1973 : Tech.
Res. Symp. S-3. SOC.Nav. Archit. Mar. En&, New York.
TIMMAN, R., and NEWMAN, J. N. (1962). The coupled damping coefficientsof symmetric ships. J.
Ship Res. 5(4), 34-55.
TROESCH, A. W. (1976). The diffraction potential for a slender ship moving through oblique
waves. Ph.D. Thesis, University of Michigan, Ann Arbor.
URSELL,F. (1949). On the heaving motion of a circular cylinder on the surface of a fluid. Q.J.
Mech. Appl. Math. 2, 218-231.
URSELL,F. (1962). Slender oscillating ships at zero forward speed. J . Fluid Mech. 19,496-516.
URSELL, F. (1968a). The expansion of water-wave potentials at great distances. Proc. Cambridge
Philos. SOC.64, 811-826.
URSELL, F. (1968b). On head seas travelling along a horizontal cylinder. J. Inst. Math. Its Appl.
4,414427.
VOSSERS,G . (1962). Some applications of the slender-body theory in ship hydrodynamics. Ph.D.
Thesis, Delft University of Technology, Delft.
VUGTS,J. H. (1968). The Hydrodynamic Coefficientsfor Swaying, Heaving and Rolling Cylin-
ders in a Free Surface, Rep. No. 194. Shipbuilding Lab., Delft University of Technology,
Delft.
WEHAUSEN, J. V. (1971). The motion of floating bodies. Annu. Rev. Fluid Mech. 3, 237-268.
WEHAUSEN, J. V. (1973). The wave resistance of ships. Adu. Appl. Mech. 13, 93-245.
WEHAUSEN, J. V. (1978). Some aspects of maneuverability theory. Proc. Symp. Appl. Math.
Dedicated t o the Late Prof Dr. R . Timman pp. 203-214. Sijthoff & Nordhoff, Groningen.
WEHAUSEN, J. V.,and LAITONE, E. V. (1960). Surface waves. In Handbuch der Physik (S.
Flugge, ed.),Vol. 9, pp. 446-778. Springer-Verlag, Berlin and New York.
WEINBLUM, G. P., and ST. DENIS,M. (1950). On the motions of ships at sea. SOC.Nau. Archit.
M a r . Eng., Trans. 58, 184-248.

Вам также может понравиться