J. FZuidMech. (1993),VOZ. 249, p p .

241-259 24 1
Copyright 0 1993 Cambridge University Press

Wave-drift damping of floating bodies

By J. N. N E W M A N
Department of Ocean Engineering, Massachusetts Institute of Technology, Cambridge,
MA 02139, USA

(Received 14 April 1992 and in revised form 26 October 1992)

Wave-drift damping results from low-frequency oscillatory motions of a floating

body, in the presence of an incident wave field. Previous works have analysed this
effect in a quasi-steady manner, based on the rate of change of the added resistance
in waves, with respect to a small steady forward velocity. I n this paper the wave-
drift damping coefficient is derived more directly, from a perturbation analysis where
the low-frequency body oscillations are superposed on the diffraction field. Unlike
the case of body oscillations in calm water, where the damping due to wave radiation
is asymptotically small for low frequencies, the superposition of oscillatory motions
on the diffraction field results in an order-one damping coefficient. All three degrees
of freedom are considered in the horizontal plane. The resulting matrix of damping
coefficients is derived from pressure integration on the body, and transformed in
special cases to a far-field control surface.

1. Introduction
The linear exciting force exerted by water waves on a floating body is proportional
to the wave amplitude A , and acts with the same frequency w . Quadratic nonlinear
interactions a t second order result in a steady mean drift force of order A 2 which
is independent of time. More generally, in a realistic spectrum of ocean waves, slowly
varying second-order forces occur a t the differences u = Iwi -w,I between the
frequencies ( w i ,w j ) of each pair of spectral components. (Second-order interactions
also cause high-frequency forces which are important for certain types of offshore
platforms, cf. Lee et al. 1991, but these are quite different in their character, and are
not considered in the present work.)
Vessels moored in deep water can experience resonant low-frequency motions in
the horizontal plane, due to excitation from the slowly varying second-order wave
forces. Important practical examples are moored ships, where the mooring system
has a relatively small linear restoring force, and tension-leg platforms which are
similar dynamically to an inverted pendulum with the buoyancy force directed
upwards. The resonant response of these vessels is limited only by the relevant
hydrodynamic damping mechanisms.
A t low frequencies the conventional linear damping of body motions due to wave
radiation is negligible. For example, the horizontal exciting force acting on a fixed
three-dimensional body in long wavelengths is proportional to the pressure gradient
of the incident waves, or O ( a 2 )for waves of unit amplitude and low frequency u.It
follows from the Haskind relations (cf. Newman 1977) that the horizontal damping
coefficients are O(a).On the other hand, the second-order wave force acting on the
body tends to a finite limit equal to the mean drift force, as the difference-frequency
tends to zero. Thus, in the absence of more significant damping effects, resonant
242 J.N . Newman
second-order motions would occur with velocity proportional to O(cr-') and
amplitude proportional to O(c-').
An obvious alternative source of damping is viscous drag. However, the quasi-
steady drag force is proportional to the square of the relative velocity between the
body and surrounding fluid. From an equivalent-damping synthesis (cf. Faltinsen
1991) the resultant damping coefficient is formally of order c.
The relevance of ' wave-drift damping ' was suggested in an experimental study by
Wichers & van Sluijs (1979), where the oscillatory surge motions were measured on
two ship models restrained by spring moorings. Comparisons of the extinction rate
in calm water and in waves of differing heights clearly indicated the presence of a
damping force proportional to the square of the wave height. (See also Faltinsen
1991, figure 5.20, and Chakrabarti & Cotter 1992, figure 11.)
These experimental observations have been explained in a quasi-steady manner,
by considering the added resistance in waves due to steady translation of the body
with small velocity U. This force is proportional to the square of the incident-wave
amplitude, tending to the zero-speed mean drift farce as U+O, with the leading-
order correction proportional to U. The derivative with respect to U , evaluated at
U = 0, represents a force proportional to the velocity, which is interpreted as a
damping coefficient.
This quasi-steady explanation has been used as the basis for several theoretical
and computational studies where the diffraction problem is solved for a body moving
with a steady forward velocity U , in the presence of incident waves. The derivative
of the resulting mean force with respect to U is derived analytically, using pertinent
asymptotic analysis for U 4 1 (Nossen, Grue & Palm 1991 ; Emmerhoff & Sclavounos
1992). Alternatively, in the approach followed by Zhao & Faltinsen (1989), the
damping coefficient is evaluated by numerical differentiation from computations
with small non-zero velocity.
I n the present paper the phenomenon of wave-drift damping is considered in a
more direct manner, without introducing a steady forward velocity. Our approach is
motivated by the conditions of the experimental observations. Whereas the damping
due to wave radiation is asymptotically small with respect to the frequency of body
oscillations in calm water, a more significant damping force occurs in the presence of
an incident wave field. This suggests the use of perturbation methods to analyse the
higher-order interaction between low-frequency body motions with frequency cr, and
the diffraction problem for the fixed body in incident waves of frequency o.The
analysis is simplified by assuming that c -4 o. I n this respect the formulation is
similar to that of Agnon & Mei (1985),who employ the method of multiple scales to
analyse the corresponding two-dimensional problem for a rectangular body in
shallow water.
One feature of the present approach is that, like the conventional linear analysis
of floating-body motions, it is possible to consider not only the force due to
longitudinal surge motions, but also the more general case of arbitrary motions with
multiple degrees of freedom. I n practice the most important modes are translations
in the horizontal plane (surge and sway), and rotation about the vertical axis (yaw).
Our principal objective is to evaluate the three-by-three matrix, of the components
of the wave-drift damping horizontal force and vertical moment, due to low-
frequency oscillatory motions in the corresponding modes.
The perturbation expansion for the velocity potential is postulated in $2, and
appropriate boundary conditions are derived. I n $ 3 the low-frequency approximation
is introduced to simplify the free-surface boundary conditions for the required
 Wave-dt-ift damping of Jloating bodies 243
higher-order potentials. In $4 the hydrodynamic force acting on the body is
expressed in an analogous perturbation expansion, and the wave-drift damping
coefficient is derived from pressure integration as the component of the force in phase
with the body's velocity which is proportional to the square of the incident wave
amplitude. Integral relations are used in $5to replace local integration over the body
and free surface by integrals over a control surface in the far field. Comparisons are
made with the results of the quasi-steady analyses by Emmerhoff & Sclavounos
(1992) and by Grue & Palm (1993). In $ 6 the present results are discussed from the
standpoints of physical interpretation and computational implementation. Various
integral relations used in the analysis are derived in the Appendix.

2. Expansion of the velocity potential and boundary conditions

Consider the diffraction problem, resulting from the interaction of monochromatic
incident waves with frequency w and amplitude IAI, and also the radiation problem
resulting from oscillatory body motions < ( t ) in the horizontal plane with frequency
v. Three separate modes of motion are included : surge (parallel to the x-axis), sway
(parallel to the y-axis), and yaw (rotation about the vertical z-axis). An indicia1
notation ( j= I , 2,6) is used to denote each of these three modes of motion,
respectively, with the corresponding oscillatory displacement Cj sin ( d )and velocity
(rCicos(at). Since the wave-drift damping force and moment are linear in these
displacements, it is sufficient to consider a single degree of motion without regard for
nonlinear interactions between different modes. The phase of the incident wave is not
restricted, hence there is no loss of generality in defining the body motions to be in
phase with sin (d) ; similarly, when a complex representation is adopted for the
oscillatory time dependence, it will be assumed that Ej is a real coefficient. Later it
will be assumed that the frequency of the body motions is much less than that of the
incident waves, i.e. (T 4 w .
The fluid is considered to be infinitely deep, and the flow is assumed to be
irrotational. For the above inputs the appropriate perturbation expansion for the
velocity potential can be expressed in the form

Here the potentials $, and $mi depend on the space coordinates x. The first subscript
refers to the order of magnitude in A , and the second subscript refers to the mode of
motion. Thus q5m = O(Am)are the components of the diffraction solution, and q5mj are
potentials of the same order in A , due to the body motions. Superscripts are used
when necessary to denote harmonic time dependence in the respective frequencies.
The symbol Re denotes the real part of the complex expression. Without loss of
generality the potential q5io),the component of the second-order diffraction solution
whioh is independent of time, is assumed to be real. The remaining potentials
displayed on the right-hand side of (2.1)are complex. Terms which are conjugate to
those in (2.1) can be neglected, hence it is permissible to include only the complex
exponentials which have a positive imaginary argument when w > 0.
The functions q5j and ~,,c$, in (2.1) are governed by Laplace's equation in the fluid
domain, with appropriate boundary conditions specified on the body and free
244 J . N . Newrnan
surface. The boundary conditions are completed by requiring each potential to
vanish at large depths below the free surface. Except for the incident-wave potential
#I = ( g ~ / o eKz-iK(xcosp+vsina)
) 3 (2-2)
which is a specified component of the first-order diffraction solution $,, each
potential in (2.1) must satisfy the radiation condition of outgoing waves in the far
field. Tn ( 2 . 2 )g is the gravitational acceleration, K = w 2 / g is the wavenumber, and /3
denotes the angle of incidence relative to the positive x-axis. Since the phase is
unrestricted, the amplitude A is complex.
The first-order diffraction potential dl is subject to the boundary conditions
$198 =o on#,, (2.3)
and g$,, - =0 on x = 0. (2.4)
Here S , is the submerged portion of the body surface, in its mean position. The
subscript n denotes the normal derivative, with the unit normal vector n defined in
the positive sense to point out of the fluid domain, and hence into the interior of the
body. Subscripts (z, y, z , t ) denote partial differentiation with respect to the
corresponding variables.
The first-order radiation potential #oj satisfies the boundary condition
$Ojn = fq on x,, (2.5)
where the three components of the vector (nj>are defined by
n, = n,, n2 = nu, n6 = xny- yn,. (2.6)
The appropriate free-surface boundary condition is
g#oj2-cr2$oj = 0 on z = 0. (2-7)
The higher-order potentials in (2.1)satisfy inhomogeneous boundary conditions on
the body and/or the free surface. In the analysis to follow it will be necessary to
consider various products of the time-dependent potential (2.1)and its derivatives.
These products can be expressed in a similar form, with appropriate coefficients.
Thus, if two functions A(t) and B(t) are represented as in (2.1) with corresponding
coefficients amaand b,, the product C = AB can be represented in the same form.
The relevant low-frequency components of C are as follows :
cp) = t Re {a, b f } , (2.8)
c$) = &(a1b, +
bl),aoj (2.9)
c(-) 2(a1b,*,+ a; b,)
1j = 1 9 (2.10)
c(0) + + + +
2i = a p o j aojb p + ( a p l l * a: 6s a y * b, + a, b$T*). (2.11)
The coefficients of triple products can be derived by repeated application of the same
relations.
On the exact oscillatory position g,, of the body surface the kinematic boundary
condition is
yP,(x, t ) = &) n. - (2.12)
Boundary conditions for the potentials in (2.1)on the mean body surface are derived
by Taylor series expansion of the left-hand side of (2.12)to the mean position S,,, and
accounting for the rotation of the body-fixed normal vector n. The appropriate
modification of (2.12)follows from the analysis outlined by Newman (1978, equation
 Wave-drift damping of floating bodies 245
3.28). Neglecting terms of order 6
' yields the following boundary conditions for the
three separate modes of motion:
@1n = [At) n, - @ l m (2.13)
@2n = [ A t ) n,-62(4 @2yn, (2.14)
@6n = i6(t) (xng-ynz)-h(t) (X@6yn-y@6zn)+'55(t) (nx@6g-ng@6z) On #?J. (2.15)
The last pair of terms in (2.15) accounts for the rotation of the coordinate system.
The other terms in (2.13)-(2.15) are the results of Taylor expansion between the
oscillatory and mean body surfaces.
The boundary conditions (2.13)-(2.15) can be expressed on the unified form
@jn = &) nj-tj(t) qn(@j) on Sb? (2.16)
where %($) = $z) %(#I = $g> %($I = x$g-Y$x* (2.17)
Since V2$ = 0, each of the three functions defined by (2.17) is harmonic. Normal
derivatives of the same functions are denoted by
%%($) = $%n) % n ( $ ) = $ V n > 966,($) = X $ l / n - Y $ z n + n z $ y - n y $ x . (2.18)
When applied to the potential for steady-state translation of the body, the normal
derivatives (2.18) are equivalent to the so-called 'm-terms' which appear in the
quasi-steady analyses (Nossen et al. 1991 ; Emmerhoff & Sclavounos 1992). In the
present work the operators (2.17)are applied in a different manner, to the diffraction
solution with the body fixed.
Collecting the terms of the same order in (2.16)and using (2.8)-(2.11))the following
boundary conditions are derived on the mean position of the body:
$g = 0, (2.19)
$$: = &iiqn($l), (2.20)
$.$; = Bjn(#F). (2.21)
Next we consider the free-surface condition, which is expressed in exact form as
a
Qtt+g@, = - - V 2 - g V . V ( V 2 ) on z = 5, (2.22)
at

where V = V@ is the fluid velocity vector. This boundary condition is transferred to

the mean free surface x = 0 using the following expansion for 5:
1 1
5 = --(~j,+gv2),=,
=--(~r,+~lr2+g~,,),_,+o(~3)
g 9

(2.23)

Using (2.23) in the Taylor-series expansion of (2.22) about z = 0,

1
@tt+g@, = -@t(@ttz+g@,,)-22-
9
K

2
+-@,(V,. K+ V . K,)-&V.V(V2)+O(CD4) on z = 0. (2.24)
g
246 J.N . Newman
For the same potentials associated with (2.19)-(2.21) the corresponding free-
surface boundary conditions can be written in the forms
g$F2 =.fi0), (2.25)
gqq&)-(w*a)z$$;) =fit', (2.26)
g$$!-CTz$$) =fi!) on z = 0. (2.27)
The inhomogeneous function on the right-hand side of (2.25) is

fi0) = Re { 8a ( - w2$T +g$Tz) I = Re {io#,$&,I. (2.28)

Here (2.4) has been used to eliminate the imaginary contribution from the first
derivative. The right-hand side of (2.26) can be evaluated in the form
io
f$) = -$I(
29
- ~ 2 $ o j z+S$Oj2,)* f-
ia
2g
($Oj)* ( - W2$lZ +g$,,,) -i(o * a)v(b,*v($oj)'.
(2.29)
Here the superscript ( f ) following a function in parentheses denotes the function or
its complex conjugate, respectively. The corresponding result for the right-hand side
of (2.27) will be derived in $3, under the approximation of small CT.
The boundary condition (2.25) implies that $!jo) is non-wavelike. I n deep water, the
right-hand side of (2.28) is o(l/R), for large horizontal radius R,since the terms in
parentheses vanish for plane waves. A more careful analysis in the Appendix shows
that $!j") = O(R-').

3. Low-frequency analysis
It is appropriate to consider the asymptotic forms of the potentials for c/ < o and
8 1 < g, where 1 is the characteristic lengthscale of the body and w21/g = O(1) is
implied. In the limit a = 0, (2.7) reduces to the rigid-free-surface condition. I n view
of the boundary condition (2.5) the potential is re-scaled in the form
$oi Vj? (3.1)
where the canonical potentials qjare real and satisfy the boundary conditions

These are the velocity potentials for translation or rotation of the body, with unit
velocity, in the presence of the 'rigid' free surface. For small values of a21/g, (3.1)
applies throughout an inner domain which is large compared to 1 and the wavelength
2xg/02 of the diffraction problem, but small compared to the wavelength 27cg/a2 of
the low-frequency oscillations. Hereafter our attention will be restricted to this inner
region. From the free-surface condition (2.7) it follows that the imaginary part of $oi
is of order 8.
Next we consider (2.26) and the associated functions (2.29), which define the free-
surface boundary conditions for the potentials $if).
I n the low-frequency analysis it
is convenient to define the auxiliary potentials
$y- qq;) = ps, (3.4)
$$) + $$;) = aQj. (3-5)
 Wave-drift damping of Jloating bodies 247
From (2.20),the corresponding boundary conditions on the mean position 8, of the
body surface are
4 n = iqn(A)r (3.6)
Q j n = 0. (3.7)
Similarly, using (2.26) and (2.29) on x = 0, and neglecting terms of order ( r 2 ,
94,-oaq = 0, (3.8)
gQi, - 02Qi = 2 0 4 -k iqb1vizz - 2iwVq4 Vq,. - (3.9)
Since the boundary conditions (2.20) and (3.6) do not involve cr, it follows that the
functions Q, and 4 defined by (3.4) and (3.5) also are independent of (r. The error in
neglecting (r is a factor 1 +O(a2).
A potential which satisfies the boundary conditions (3.6) and (3.8) is easily
constructed in the form = igj(&), but this violates the radiation condition since
the incident wave is part of qbl. For j = 1 , 2 this problem can be overcome simply by
adding an extra term proportional to with the results
Pl = ic$ls-KcosP$l, Pz= i&-KsinPqb,. (3.10)
For j = 6 the appropriate extra term involves the derivative $la with respect to the
wave heading angle p:
p6 = i96($1)+i#1,8. (3.11)
To confirm these solutions, note that each sum vanishes for the incident-wave
potential (2.2), hence (3.10) and (3.11) satisfy the far-field radiation condition. In
view of the boundary condition (2.3) there is no contribution from the second terms
to the boundary condition (3.6),and thus the validity of (3.10)-(3.11) is established.
In the analysis of Qiattention is first given to the term 2wP, on the right-hand side
of the free-surface boundary condition (3.9).The general solution can be expressed
in the form
Qj = ( 2 w / g ) q K + qj, (3.12)
where the subscript K denotes differentiation with respect to the wavenumber. The
potential qj is subject to the boundary conditions
qjn =-(2w/g)4Kn On sb, (3.13)
and gqiz -02qi = ioqbl vjZz -
-2io Vg51 Vpli on z = 0. (3.14)
Since 4 is a solution of the homogeneous free-surface condition, its effect on the
right-hand side of (3.9) is secular. For large values of the horizontal radius R the
radiation condition implies that
-
, and thus the solution (3.12) is non-
R-i emiKR
uniform in the far field, with the asymptotic behaviour Qj Rie-iKR.This does not
result in practical difficulties in the analysis to follow, provided the domain
considered is suitably restricted.
Since vj is asymptotic to a Rankine dipole, the right-hand side of (3.14)is of order
l/R3 in the far field. The solution of this boundary condition is non-trivial, but
uniform at infinity with the same far-field form as a first-order radiating wave.
Thus the conventional far-field radiation condition is applicable for the regular
component q, .
Next we consider the functionf$) defined by (2.27).The possible contributions are
indicated from the complete third-order free-surface condition (2.24),and involve the
following combinations of lower-order potentials and their drivatives :
(h4lj)3 ($23 4 0 , h (@>$oj).
248 J . N . Newman
Only the first combination is independent of $oj, and hence of order one as g + O .
After substituting the potentials and $if)
with appropriate time-dependent
factors in the terms on the first line of the right side of (2.24), and collecting all
components with the time dependence eiut,the limit for f $ ) as g + 0 is obtained in the
form
A; - &4mw;i -4:L) @(WZ $3 - -

(m72* $!$? 1+ 4;)*

- $1 -9 )I @$12 -
-
122

~ i 0 [ - 9 ~ ~ e Z + ~ ~ T , , - $ 1 p j * Z e + p , *$ l * Z I . (3.15)
In the far field the second derivatives in (3.15) can be replaced b y K 2 , and it follows
that f$) = o(l/R). Thus in the limit as a+O the boundary condition (2.27) implies
that $1); is non-wavelike, vanishing algebraically in the far field in the same manner
as
The solutions (3.10) can be substituted in (3.15) for j = 1 , 2 with the result

m - WWT $122- $1 $3= iq(j+4)). (3.16)

Thus, in the limit c-t 0, the potentials
$2) = iq($?)) (3.17)
are solutions of the boundary conditions (2.21)and (2.27)for ( j = 1,2). Similarly, for
j = 6,
j+$) - ig6(f2(0))
+if$), (3.18)
and $1): = ig8(@jO)) (3.19)

4. The hydrodynamic pressure force and damping coefficients

Perturbation expansions similar to (2.1) can be assumed for the pressure and the
resulting force (and moment) acting on the body. The appropriate terms to consider
for the component of the force or moment in the same direction as the mode ti,due
to the diffraction field, the oscillatory motion in the mode ti,and their interaction,
are
F,(t) = Re {Flieiwt+F$) +Fii) e2iwt+ ...
+ tj[F,,,eid +FiG)ei(w+dt +F(-) e i ( w - d t
1Zi
+ ~ i ; eiut
j + ~ ( 2z i+j ) e i ( z w + d t +F~;C&) e U Z w - d t + . . .I}. (4.1)
Here F16is the first-order exciting force and B$) is the second-order mean drift force
for the fixed body. Foijis the first-order force due to a unit motion &, which can be
expressed in the usual form

F,,, = - (irr2AoZi rrBoij), + (4.2)

where the real coefficientsA,, and B,, are the added-mass and damping coefficients.
The higher-order force component F$; can be expressed in the analogous form
= - (ia2A2ij + CT&,). (4.3)
This force is of second order in the wave amplitude, and first order in the motion
amplitude.
 Wave-drift d a m p i n g of floating bodies 249
For asymptotically small values of the frequency c,A,, = O( l ) , whereas B,, =
O(a7)as noted in the Introduction. By comparison, the highest-order added-mass
coefficient A zii is of secondary importance, but the corresponding damping coefficient
Bzij is significant since B,, is asymptotically small.
The most direct approach to evaluate Fi:] is from pressure integration on the body.
The analysis is carried out for a floatin3 body with a time-varying wetted surface fl,,,
mean surface S,, waterline contour C,,, and mean waterline contour C,, with the
restriction that the body surface is smooth and vertical at the waterline. Integration
around the closed contour C b is defined in the positive sense with respect to the
enclosed boundary surface X,, i.e. in the counter-clockwise direction when viewed
from above the origin. For a submerged body the integrals over the waterline can be
neglected. Ultimately a fixed control surface S, also will be used, which surrounds the
body in the far field. The portion of the mean free surface between S, and S, will be
denoted S,. The intersection of S, and S, is the contour C,.
The horizontal components of the pressure force and the vertical component of the
moment are evaluated using Bernoullis equation, in the form

As in the case of the body boundary condition (2.13),the pressure is transferred from
flb to its mean position S,. I n addition, the contribution from the time-varying
intersection of fib with the free surface z = 5 is expanded as a line integral on cb.
The contribution from the transfer of the pressure involves the expansion
PI& = PIS, +E$) q ( P )+ O ( E 3
The linear correction does not affect the damping coefficients in (4.2)or (4.3)since it
is out of phase with the body velocity, and the hydrostatic pressure does not
contribute to (4.4). Thus the only contribution from integration over the mean
surface is

-P /JSb (A+tv4 * V4)n, ds. (4.5)

The contribution from the contour C,, includes the vertically integrated hydrostatic
pressure contribution -ipgc, and the Taylor expansion of the dynamic pressure.
The resulting line integral is

The last term in the first integral accounts for the integrated hydrostatic pressure ;
in the second integral (2.23) is used. Terms of higher order than (4.3)are neglected.
The last term in (4.6), which contributes terms of order a& = O(a2) to (4.3), is
neglected hereafter.
Our objective is to evaluate Bzij,the wave-drift damping coefficient in the direction
i due t,o a velocity in modej. It is convenient to simplify the notation by defining the
new coefficients
gij= B,,Jp. (4.7)

9 FLM 249
250 J.N . Newman
Considering only the real part of the force coefficient F$),and substituting the
appropriate components of the velocity potential (2.1) in (4.5)-(4.6),

A$ = Rell,,, ( i $ ~ ~ ) + V ~ ~ o ) . V p l j + g V $ T . V g i ) , , d s

Hereafter the symbol Re is deleted, with the understanding that the real part is
implied in all of the following equations.
Equation (4.8) provides an explicit relation for the wave-drift damping coefficients.
The principal difficulty is in evaluating the various higher-order potentials including
P,,Qj,$!jo), and $$). The dependence on the last two can be removed by further
analysis, using Greens and Stokes theorems together with appropriate boundary
conditions on the body and free surface. After substituting the boundary condition
(3.2) in the first term of the surface integral, and Stokes theorem in the form (A 2)
for the second and third terms,

%= ,I ( i ~ D ) ~ * n + ~ ~ ~ , ( ~ ~ O ) ) + ~ ~ ~ ~ n ( ~ ~ ) ) *

+s
1
kb [qjfi(o)-$qbr(q- iV$l*Vq+.)]
To evaluate the contribution from the first term in the surface integral of (4.9),
n, dl. (4.9)

Greens theorem is applied using the boundary conditions (2.21), (2.27), and (3.3),
with the result

SSJSb(i~D~~,+~~$.(BP))dS

where the integral which remains over S,, is defined as

(4.11)

Various alternative formulae for evaluating (4.11) are given in the Appendix. The
only non-zero elements are I,,, 12,,I,,, and
The contribution from the last integral of (4.10) is

where (A 4) and (2.28) have been used.

 Wave-drift damping of Jloating bodies 251
Substituting these results in (4.9),

The integrand in the last integral vanishes for j = 1,2, as a result of (3.16). For
j = 6, it follows from (3.18) that

where (2.28) and (A 3) have been used.

Comparing (4.13) and (4.14) with the original expression (4.8),integrals over the
free surface are introduced but the higher-order potentials $$) have been removed
and the second-order diffraction potential $io) only contributes via the dipole
moments pi in the evaluation of I, from (A 16). The latter contribution is present
only for the coefficients q6, %6, Be1, and &?682, The special role of $lo) has been
emphasized by Grue & Palm (1992, 1993) with respect to the coefficients B16, BZ6.

5. Far-field analysis
Momentum relations have been used in the quasi-steady analyses by Grue & Palm
(1992, 1993) and Emmerhoff & Sclavounos (1992),to relate the wave-drift damping
coefficients to integrals over a control surface 8, in the far field. Similarly, in the
present analysis of unsteady body motions, energy conservation could be used to
relate the work done by the damping coefficients to the rate of energy flux in the far
field. However, higher-order potentials and higher-order terms in the low-frequency
approximation must be considered in both the momentum and energy approaches.
Alternatively, integral theorems can be applied to the results of 94 with the objective
of replacing integrals in the near field, over the body and free surface, by integrals
over S,. This analysis is carried out below.
From the boundary condition (3.7) and Greens theorem, the integral over Sbin
(4.13) can be expressed as

Invoking the inhomogeneous condition (3.9) for Q,, the contribution from S, is

9-2
252 J.N . Newman
The last term in (5.2) can be transformed using (A 3). With the substitutions $ =
$19i($:) and y? = qj,and the boundary condition qjz= 0 imposed, (5.2)is equal to

In the contour integral (3.2)has been used, and there is no contribution from C, since
cpj is o(l/R) in the far field. After substituting (5.1) and (5.3)in (4.13),

Here in the first integral over S , (A 3) has been used, together with the boundary
conditions (2.3) and (3.3).
After using Stokes' theorem in the form (A4) to transform the two contour
integrals over G,,

(5.5)

The first integral over S, vanishes f o r j = 1,2, as noted following (4.13).Similarly,

after using (3.10) to evaluate Pj,the second integral over S, vanishes for i = 1 , 2 . For
i = 6 the contribution from the second integral cancels I,, except for the last two
terms in (A 16). For j = 6 the first integral over S, must be evaluated using (4.14),
and (3.11) may be used to evaluate p3. The final result for all cases except i = j = 6
can be expressed in the form
 Wave-drift damping of j-loating bodies 253
Here Sij is the Kroenecker delta function, equal to one if i = jand zero otherwise, and
p l ,p2 are the horizontal components of the effective dipole moment associated with
the potential &f') as defined by (A 13). Except for this dipole moment, (5.6) depends
only on the first-order diffraction potential and on the interactions (3.4)-(3.5)
between $1 and qj.The same dependence was noted a t the end of 54,with respect to
the near-field analysis.
Finally, in the case i = j = 6,

Note that in (5.6) the only integrals which remain are in the far field, but in (5.7
an integral remains over the free surface.
If i = j=I=6, (5.6) reduces to the form

After applying Stokes' theorem over S,, in a form analogous to (A 2),

where (2.4) is used. Except for differences of notation this formula is identical to
equation (72) of Emmerhoff & Sclavounos (1992).
Another variant of the far-field representation is derived by using the function pi
in (5.1). For i = j this leads to the relatively simple result
in place of gi(q5f)
rr
(5.10)

which is valid for all three values of i . A feature of (5.10)is that the contribution from
the secular component of (3.12)can be evaluated by differentiation of a non-singular
integral :

I n this form explicit dependence on the derivatives I<K and PiKn is removed, but the
latter derivative still must be evaluated on the body surface in the boundary
condition (3.13).Since the functions which remain in the integrand of (5.11)satisfy
the radiation condition, far-field asymptotic approximations can be substituted for
254 J . N . Newman
each potential in the same manner as in the evaluation of the mean drift force (cf.
Newman 1967).
Collectively, (5.6)and (5.10)can be used to evaluate all of the damping coefficients
.@$* from far-field integrations. However the apparent computational advantage of

these relations over (4.13)is offset by the fact that, regardless of which approach is
followed, the higher-order potentials qj must be evaluated on the body surface.

6. Discussion
The approach followed in the present work is to consider the low-frequencymotion
of a floating body as a perturbation of the incident-wave diffraction problem where
the body is fixed in position. Two timescales are involved, one associated with the
body motions at the frequency (T and the other corresponding to the incident wave
frequency o. The analysis is based on the assumptions that the incident-wave
amplitude A and body motions E are both small, and that (T < o.Unlike other works
which use a quasi-steady analysis, the relevant damping force acting on the body is
derived without introducing a vanishingly small forward velocity of the body and
considering the derivative of the force with respect to this velocity.
The results confirm that, whereas low-frequency oscillations in calm water result
in wave radiation and damping asymptotically small with respect to the frequency
(r, the same quantities are of order one in IT if the oscillations are superposed upon

an incident-wave field. More specifically, for horizontal low-frequency oscillations of

the body, the matrix of calm-water damping coefficients B,, = O((T'), whereas the
analogous wave-drift coefficient B,, = O(A2).Thus the relative importance of these
two damping coefficients is in proportion to the ratio d / A 2 . The small value of the
calm-water damping coefficient can be associated with the fact that the far-field
energy flux is associated with long waves, the amplitude of which is negligible due to
their asymptotically large lengthscale relative to the body. The situation is different
in the presence of the diffraction field, where the basic wavelength is comparable to
the body scale and the interaction of these waves with the body motions leads to a
modification of the energy flux associated with the scattered field.
It is interesting to compare the present analysis with that of the quasi-steady
approach, e.g. the work of Emmerhoff & Sclavounos (1992). The linear potentials
and q50j are the same, but the interaction potential & is somewhat different. In the
quasi-steady analysis the difference component (3.4) does not appear, and the
interaction potential is formally equivalent to (3.5).Although the potential is
absent, an equivalent term is included in the inhomogeneous free-surface condition
(3.9) to account for the effect of steady forward velocity on the frequency of
encounter. As shown in $5 the final results for the wave-drift damping coefficientsgI1
and .@ ,,equivalent in these different approaches.
are
The quasi-steady analysis of Nossen et al. (1991) is somewhat different. Instead of
the first inhomogeneous term on the right-hand side of (3.9),the secular component
of the interaction potential appears as a consequence of differentiating the zero-speed
Green function to obtain a linearized correction for forward velocity. This
construction was first suggested by Huijsmans & Hermans (1985). It has been
extended by Grue & Palm (1992, 1993) to include the coefficients BZl and gel, with
similarities to the present results for these coefficients.
A significant feature of the present method is the ability to include angular (yaw)
oscillations of the body, about the vertical axis. This mode of motion cannot be
accounted for in the quasi-steady approach, except possibly by considering a slow
 Wave-drift damping of Jloating bodies 255
steady rotation of the body. Unlike the conventional added-mass and damping
coefficients in (4.2), the matrix of wave-drift damping coefficients 9jgiderived here
appears to be asymmetric, and there is no obvious way of relating the forces due to
yaw motions to the corresponding moments due to surge or sway.
Our results for yaw ((4.14)and (5.6)-(5.7))involve differentiation of the first-order
diffraction solution with respect to the angle of incidence p ; this can be interpreted
in the quasi-steady sense as the correction of the incidence angle due to the bodys
yaw oscillations. Equation (5.7),for the yaw damping moment, includes an integral
over the free surface, but the same coefficient is expressed completely in terms of a
far-field integration in the forms (5.10)-(5.11).The latter formulae do not involve
explicit differentiation with respect to the heading angle, but this is implicit via the
functional PBdefined by (3.11).
In the present work it is assumed that the unsteady motions <(t)are sufficiently
small to justify perturbation expansions about the stationary mean position of the
body. This assumption is not made explicitly in the quasi-steady analyses, which
assume only that the corresponding velocities &t) are small. At first glance this
distinction appears to be significant, since low-frequency horizontal excursions of
offshore platforms generally occur with substantial amplitudes. However, in the
context of deriving only the wave-drift damping coefficients, i.e. the component of
the total hydrodynamic force which is linear in ( ( t ) , the magnitude of <(t) is
irrelevant. Thus, despite the different initial assumptions concerning the order of <(t),
our results (5.8)-(5.9)are identical to those of Emmerhoff & Sclavounos (1992).
Far-field integration is generally considered to be simpler or more accurate than
direct pressure integration on the body or intermediate results such as (4.13).
Asymptotic relations can be used to evaluate the components of the velocity
potential, and integrals over the control surface 8, can be reduced to azimuthal
integrals in terms of the far-field scattering amplitude. A more specific advantage of
the far-field evaluations here is that the secular component of the higher order
potential Qi can be evaluated as the derivative of an integral with respect to the
wavenumber, as in (5.11). On the other hand, the most difficult task envisaged in
numerical implementation is the evaluation of the potential qi,as the solution of the
boundary conditions (3.13)-(3.14). Even in the far-field analysis this solution is
required locally on the body, in order to evaluate the far-field scattering amplitude.
Thus there is no obvious advantage in evaluating the wave-drift damping coefficients
in the far field, and direct use of (4.13) may in fact be simpler. Numerical
implementation of the present analysis is required to confirm this conjecture, and to
demonstrate the practical value of our results.
Several restrictions should be recognized in the present analysis. These include the
assumptions of infinite fluid depth, no first-order body motions, and the consideration
of low-frequency motions only in the horizontal plane (modesj = 1 ,2 ,6 ).The effects
of finite depth are relatively simple to account for, including the second-order
component of the incident-wave potential required in the diffraction solution. First-
order body motions can be accommodated by including the corresponding linear
radiation potentials in but the boundary condition (2.3) must be modified and
this will affect much of the subsequent analysis. Low-frequency vertical motions of
the body may be important in certain applications; the principal difficulty
anticipated in this extension of the analysis is that the corresponding components of
the operator (2.17), including vertical derivatives, will complicate the reduction of
the integrals over the free surface. Each of these possible extensions will be useful in
practical applications.
256 J . N. Newman
This work was conducted under the Joint Industry Project Wave effects on
offshore structures. The author is grateful to the sponsors, including the Chevron Oil
Field Research Company, Conoco Norway, the National Research Council of
Canada, Norsk Hydro, PetrobrAs, Saga Petroleum, the Shell Development Company,
Statoil, and Veritas Research. Several errors in the original manuscript were noted
by Dr P. J. Clark, and one of the referees corrected a fundamental error in the
Appendix.

Appendix
A variant of Stokes theorem (Milne Thomson 1955, $2.51) can be applied in the
form

Here S is an open surface, C the boundary contour, and the integration around C is
in the positive (counterclockwise) direction with respect to the normal vector n.
For application to the second and third terms in the surface integral of (4.8)the
substitution q = q5V$ is made, with (A 1 ) applied on the body surface 8, and the
contour C, in the plane x = 0. The x,y-components of (A 1) are then given
respectively by setting i = 1,2 in the equation

JJSb(Vq5.ogrnid8 = ~ ~ s b [ % ( d ) r , + d ~ , ( $b )d$*n,dl.
l~+~ (A 2)

Here it is assumed that V2$ = 0. To verify that (A 2) also holds in the case i = 6, the
products y# and xq5 are substituted for q5 with i = 1 , 2 , respectively; the difference
between these two results is equivalent to setting i = 6 in (A 2).
Alternatively, consider the vertical component of (A 1 ) on the free surface,

sS,,(Vd.V$)ds = {~sf(e,r,+d$zz)d8+f cb+cc 611.ndl. (A 3)

In the contour integral the normal derivative is directed out of 8, in the same plane.
This integral is to be evaluated over both contours Cband C, in the positive direction
(counterclockwise when viewed from above).
In the special cases where $ = xi,(A 3) is reduced to the more familiar form of
Stokes theorem,

This formula also can be extended to include the case i = 6, using the same procedure
described following (A 2).
Next we consider the integrals Itj as defined by (4.11). From the boundary
conditions ( 3 . 2 ) ,

I, = JT,,(nj%(@)) -ni %($go))) ds. (A 5)

The integralsI,,, I Z 2andI,,

, are obviously equal to zero. For I,, the integrand in (A 5 )
is equal to the vertical component of n x V#, and the integral vanishes by another
 Wavedrift damping of floating bodies 257
variant of Stokes theorem (Milne Thomson 1955, $2.51, equation 2). Thus I,, =
-Il2= 0. The only non-zero cases are

Here Stokes and Greens theorems have been used, together with the boundary
conditions (3.2) and (3.3).After reversing the signs for the lower elements of (A 6)
and using (A 3) with the boundary conditions (2.3), (2.25) and (2.28),

An alternative representation can be derived by applying Greens theorem to 4):

and (rpi-xi), for i = 1,2. Since both potentials satisfy homogeneous Neumann
conditions on the body,

Jsf $i:)(vi-xi) + JISc ($PA(Ti- xi)-#io)(vtn - ni))a= 0. (A 8)

The asymptotic form of $?) is required to ascertain the contribution from the last
integral in (A 8). For this purpose it is convenient to assume that S , is a circular
cylinder of large radius R about the vertical axis. If the divergence theorem is used
with the boundary conditions (2.19) and (2.25),

where (A 3) is used to derive the last contour integral. To estimate this contour
integral Greens theorem is applied to and its conjugate. Since the boundary
conditions on the body and free surface are homogeneous,

In the far field the asymptotic form of is such that, for large R,
$,(I?, 8, x ) = 8,O)eKZ+O(R-). (A 11)
Here the estimate of the error follows from the spherical-harmonic wave-free
potentials (Havelock 1955, equation 8), and from the far-field asymptotic expansion
of the corresponding Green function (Newman 1985, equation 6). With (A 11)
substituted in (A lo), the vertical integration over S, can be performed, with the
resulting estimate
Reifcc$l$~ndZ= O ( E 2 ) . (A 12)
258 J . N . Newman
Thus (A 9) is of order (F), implying that $$') behaves like a Rankine dipole in the
far field, with the asymptotic approximation

+
p -
p*V(l/r).
Here p is the dipole and r = (R2 2);. The only non-vanishing contribution to the
(A 13)

integral in (A 8) over S,, as the radius of the control surface is increased to infinity,
is
rr

J, (&%,-$5:: q )dx = - 2npi

Evaluating the last integral in (A 8 ) and using (A 3),

Substituting (A 15) in (A 7) gives the alternative expressions

REFERENCES
AQNON,Y. & MEI, C. C. 1985 Slow-drift motion of a two-dimensional block in beam seas. J . Fluid
Mech. 151, 279-294.
WBARTI, S.K. & COTTER,D.C. 1992 Hydrodynamic coefficients of a moored semi-
submersible in waves. J . Oflshore Mech. Arctic Engng 114, 9-15.
EMMERHOFF, 0. J. k SCLAVOUNOS, P. D. 1992 The slow-drift motion of arrays of vertical
cylinders. J . Fluid Mech. 242, 31-60.
FALTINSEN, 0. M. 1991 Sea Loads on Ships and Oflshore Structures. Cambridge University Press.
GRUE,J. & PALM, E. 1992 The effect of second order velocities on drift forces and drift moments.
In 7th Intl Workshop on Water Waves and Floating Bodies, Val de Reuil, Prance (ed. R. Cointe).
GRUE,J. & PALM, E. 1993 The mean drift force and yaw moment on marine structures in waves
and current. J . Fluid Mech. 250, 121-142.
HAVELOCK, T. H. 1955 Waves due to a floating sphere making periodic heaving oscillations. Proc.
R. XOC.LO&. A 231, 1-7.
HUIJSMANS, R. H. M. & HERMANS, A. J. 1989 A fast algorithm for computation of 3-D ship
motions at moderate forward speed. I n Proc. 4th Intl Conj. on Numerical Ship Hydrodynamics,
Washington (ed. J. H. McCarthy), pp. 24-31. National Academy of Science Press.
LEE,C.-H., NEWMAN, J. N., KIM,M.-H. & YUE,D. K. P. 1991 The computation of second-order
wave loads. I n Proc. 10th Intl Conf. on Offshore Mechanics and Arctic Engineering, Xtavanger,
Norway, Vol. 1 (A) (ed. S. K. Chakrabarti, H. Maeda, A. N. Williams, D. G. Morrison 6
M. Hendriksen), pp. 113-123. ASME.
MILNE-THOMSON, L. M. 1955 Theoretical Hydrodynamics. Macmillan.
NEWMAN, J. N. 1967 The drift force and moment on ships in waves. J . Ship Res. 11, 1, 51-60.
NEWMAN, tJ. N. 1977 Marine Hydrodynamics. MIT Press.
NEWMAN, J. N. 1978 The theory of ship motions. Adv. AppZ. Me&. 18, 221-283.
NEWMAN, J. N. 1985 Algorithms for the free-surface Green function. J . Engng Math8 19, 57-67.
 Wave-drift damping of floating bodies 259
NOSSEN,J., GRUE, J. & PALM,E. 1991 Wave forces on three-dimensional floating bodies with
small forward speed. J . Fluid Mech. 227, 13k160.
WICHERS,J. E. W. & SLUIJS,M. F. VAN 1979 The influence of waves on the low-frequency
hydrodynamic coefficients of moored vessels. Proc. Offshore Technology Conf., Houston, OTC
3625.
ZHAO,R . & FALTINSEN, 0. M. 1989 Interaction between current, waves and marine structures. In
Proc. 5th Intl Conf. on Numerical Ship Hydrodynamics, Hiroshima, Japan (ed. K. Mori), pp.
51S52.5. National Academy of Science Press.