- - JUN I 1976
ARCHEF
ABSTRACT
The linearized problem of potential flow around a twin-hull ship is solved by two sheet distributions of sources and
normal doublets. The thickness problem
is solved by the thin-ship approximation, while the lifting problem is solved by a
slender-ship approximation. General
formulas for the steady hydrodynamic force and moment acting on a hull are obtained as an extended form of Lagally's theorem for two floating lifting bodies. This theory is applied to the prediction of wave resistance of
Small-Waterplane-Area Twin-Hull (SWATH) ships. Computed
and experimental results are compared and presented.
I. INTRODUCTION
Recently, there has been a consider-able amount of interest in developing a particular type of twin-hull configura-tion, namely, the Small-Waterplane-Area Twin-Hull (SWATH) ship, as a new design
concept for a number of specific Navy
applications. As part of the initial
effort to investigate the hydrodynamic performance of SWATH ships, an analytical tool for predicting wave resistance was
developed. This paper presents the
results of subsequent efforts to extend the scope of analysis to include an investigation of the steady hydrodynamic force and moment on a twin-hull ship of
THE FORCE AND MOMENT ON A TWIN-HULL SHIP IN A STEADY POTENTIAL FLOW
by
WEN-CHIN LIN
Naval Ship Research and Development Center Bethesda, Maryland
a more general shape.
The wave resistance problem of catamarans has been investigated by
Lunde (1951) and Eggers (1955). In both
papers, a catamaran was treated as two
separate but identical thin ships, and the velocity potential of the problem was obtained from two sheet distributions
of simple sources (or, more precisely,
Havelock sources). However, since each
hull is in proximity to the oth2r, and since each will generally experience an asymmetrical flow field around it, dis-tributions of simple sources alone are insufficient for solving the problem. To account for the asymmetric flow field around each hull, distributions of
doublets normal to the sheets should be
simultaneously considered. Fortunately,
by virtue of linearization, the effects of doublet distributions may always be superimposed to refine the results initially obtained from source distri-butions without jeopardizing the initial results.
In this paper, the twin hulls are represented by two distributions of sources and normal doublets on their planforms (which, in the case of a symmetric ship, correspond to the ship
centerplane). The normal-doublet
distri-bution is also extended into the infinite downstream in the wake regions which
Lab.
v. Scheepsbouyikunde
Technische Hogeschool
trail each4u11. Hence, the so-called "lifting" effect is also included in
the present analysis. The mathematical
problem is formulated within the context of the linearized thin-ship theory. Therefore, the "thickness problem", i.e., the determination of the source density, may be solved immediately upon
applica-tion of the boundary condiapplica-tions. For
the "lifting problem", i.e., the deter-mination of the doublet density, an integral equation is initially derived with the full expression of the Havelock doublet as its kernel.
As a first approximation, the
kernel function of the original integral equation is replaced by the so-called "zero-Froude-number" Green's function.
mean camber surface of a demihull half thickness of a demihull
2b
hull-separation distance= total velocity potential = distribution potential
total fluid velocity relative to the ship
disturbance velocity planform of Hull I
wake sheet trailing Hull 1 disturbance potential due to source distribution
disturbance potential due to normal doublet distribution = local density of the source
distribution
= local density of the normal-doublet distribution
H
= ship draftL = ship length
B
= ship beamThis equation is further reduced to a singular integral equation of one varia-ble by introducing low draft-to-length ratio and small hull separation
distance-to-length ratio approximations. For this
reduced problem, both the source density and doublet density of the distributions
have been obtained. General formulas
for force and moment expressed in terms of these singularity-distribution
den-sities have been derived. The result,
as might be expected, is an extended form of Lagally's theorem for thin lifting bodies making a uniform motion in the
free surface. In addition to the
expressions obtained by Cummins (1957) for the steady case, terms involving integrations along the waterline and the trailing edge of a hull now appear in
NOMENCLATURE
(Additional nomenclature are defined as they appear in the text)
A673 =
scalar product of two vectors = Cauchy principal value integralA/03 =
vector product of two vectors2:
= bounding surface of a controlvolume excluding ship wetted surface
control volume bounded by and the wetted surface of Hull 1, S1
tt =
I'Ve plus the displacement ofHull 1
= (0,1,0) = the unit normal to $(:',)
- tangential component of
q
withrespect to
= normal component of F with
respect to
= = the difference
in at both sides of Sri,'
averaged value of the fluid velocities at both sides of the singularity distribution
F
= hydrodynamic force acting on Hull 1= hydrodynamic moment acting on Hull 1 with reference to the origin of the Oxyz-frame
7Rw = total wave resistance
_I C
t(-)
= = =these formulas. The significance of these additional terms for making better predictions of force and moment on a floating lifting body is being assessed by actual computations and by comparison
of the computed results with experi-mental data.
As an example of the application of the theory presented in this paper, the wave resistance of two SWATH ships has
been investigated. The results of
com-putations together with their comparisons with experimental results are presented. Unfortunately, the entire scope of the
investigation of force and moment on a twin-hull ship is not covered in this paper since this work is still in progress and further computations are being made. We shall continue to report the results of such investigations as they become available.
II. FORMULATION OF THE PROBLEM
Consider a twin-hull ship moving with constant forward speed U into
other-wise calm water of infinite depth. The
two hulls will be referred to as Hull 1
and Hull 2, respectively. Let Oxyz be a
right-hand Cartesian coordinate system moving together with the ship, with Oz
directed upward (against gravity), Ox in the direction of motion, Oy to port, and Oxy coinciding with the undisturbed free surface.
As usual, we shall assume that each hull is "thin" and has small camber, and that the two hulls are mirror images of
each other. We shall further assume that
Hull 1 is situated approximately in the Oxz plane, and Hull 2 is situated
approx-imately in the plane y = 2b. Thus, the
hull separation distance is 2b, and the plane y = b is a plane of symmetry for the whole flow field,
(2.3)
ship geometry may be described as follows: velocity relative to the ship as follows:
Using this coordinate system, the so that its gradient gives the total fluid
Hull I:
[V,
PC.
a)(2.1a)
Hu112: (2.1b)
cLs = 71'46, a-) 4-.26
where the subscript S and P indicate the starboard and port sides of the hull, respectively.
For convenience, let us define the mean camber surface and the half-thick-ness function, respectively, as follows:
6,4)1
(2.2a)
(2.2b)
Then, 1 -16,,,z-)= C6,-, )+ I 6,,,z) , ,1-6,,z)= C 6,,j) t6,4)Finally, we shall assume that the draft of the ship is H and the ship length is L, extending from -L/2 to L/2.
We shall now develop a potential flow of an inviscid fluid for a twin-hull ship. This development will be based upon the linearized theory of ship waves using the thin-ship approximation.
Let the total velocity potential be given by
?0
?x.
which represents the disturbance velocity
due to the ship. We shall use the
nota-tion et , i = 1,2,3 to represent the
three base vectors of the Oxyz coordinate
system. Then, by a systematic
perturba-tion expansion based on the thin-ship assumption, we may obtain the following boundary-value problem appropriate for
the first-order approximation of the
dis-turbance potential
0
.0 ( x>o,
o(1), ior x<o,
In addition,
0
is required to satisfythe following approximate kinematic con-ditions at the ship:
a.sX1411--pcs, to) 24(X, ufac (..o.t) ?), JJ, and 441,4) u ui 4re,
_ ?.1
a ,for
co),(o)
where Sib and 326 represent projections
of the wetted surfaces of Hull 1 and Hull 2 onto the vertical planes y = 0 and y = 2b, respectively, when the ship is at
rest. In wing theory terminology,
(2.4) (o)
and 526 are the planforms of Hulls 1 and
(2.5) 2, respectively.
(2.6)
(2.7)
Before proceeding further, let
us
observe the flow situation around the
ship. Because each hull is now cambered
and is situated in proximity to the other, each will experience a uniform flow coming
from an angle. Hence, in addition to the
usual thickness effect of a symmetric monohull ship, the so-called cross-flow
effect must also be considered. This is
true even if the camber is reduced to zero since, for two identical monohull ships sailing abreast, the flow around each ship will still be asymmetric with respect to its centerplane.
Since both the differential equation and the boundary conditions (2.6) and (2,7) are linear, we may consider the effect of
thickness and the effect of the position of the mean camber line (which now includes
the incidence) separately. This can be
done by assuming C(x,y) = 0 and t(x,z) = 0 in (2.7), respectively, and superimposing
the results. Thus, for example, the flow
around Hull I may be decomposed into two
problems as follows: (1) the flow around
a symmetric hull (with respect to the
plane y = 0) without the presence of Hull 2, and (2) the flow around a very thin hull (zero thickness), with or with-out camber, in the presence of Hull 2, Subsequently, these two problems will be referred to as the "thickness" and the
"lifting" problems, respectively. The flow around Hull 2 may be similarly
decomposed. This suggests that the
dis-turbance potential
0
may have thefollowing decomposition: such that -F I,/..__
,t.
(2.8)
=
where = 61,4/, 4v-) v , -2b,Z) = =I
=to, a) ?
?A,
Tx- , ?t,(x,.26,±0, a) 1.)?i" 7) 4.- 2x ,or
f oc,0.i)(2.9)
tm, tjaC 0.4 6, 24 *)a-(2.10)
;Or fx, zb, a)t S2bIn view of these boundary conditions and the corresponding flow problem each compo-nent potential represents, we may expect
the potentials Om and
0
to besymme-trical about the planes y = 0 and y = 2b,
respectively, while Olp.
and 4. should
be antisymmetrical about these two planes. Hence, and ;,t) t,.(-,
g, )=
-;,= t,(x,
(At,. 1-) =Our next task is to find solutions to the boundary-value problems given by
(2.6), (2.9), and (2.10). We shall try
to find the respective solutions by the method of Green's functions.
Solution to the Thickness Problem Let us first find the solution to
Ow
which is required to satisfy therequirements in (2.6) and the first
equation in (2.9). Since Oicr is expected
to be symmetrical with respect to the plane y = 0 (see 2.11), it suggests that
Oicr may be obtained by a distribution
of sources and sinks over the planform of
(0)
Hull 1,
S. Thus
a.) g
a,..)./x.de,,,(2.13)
sr
where(r(x
0 z0) is the local surface density of the distribution, and G (x,y,z,
x_OP,y0 ,z0) is known as a
Green's function. This function
satis-fies all the requirements in (2.6),
except at (x,y,z) = (x01y01z0), and has
the form
In this expression
A..
and H is harmonic everywhere in the lower
half space. The explicit expression of
the function G is well known (see, for example, p. 149 of Wehausen (1973)), and may be written in the following form:
= 71--, (2.11)
4 f%
,
X e...s[io,n)e-req. ( 0] (2.15)(2.12)
XA
R.
where TL,R 0)t(1-1,,f)2 I
= 3/U2,and the integral with respect to k is a Cauchy principal-value integral.
The function G for this specific case is called a "Havelock source" and,
in addition to having the properties of a source potential, satisfies the
linear-ized free-surface condition in (2,6), Thus, it is obvious from (2,13) and (2,14)
that 010- has the properties of a
single-layer potential, and, in particular, has the following jump property at the
sur-face of distribution, S16 '
(2.16)
6 (x.1,e: e.)-k
H
1-x.,a (2.14) -44.4,-
(x. = -tr(x.,From (2.16) and the first equation of (2.9), the local surface density,cr(xz,), of the distribution is found to be
related to the half-thickness function of a hull as follows:
_II
0(;
I) =
.4
In a similar manner, we may conclude that
tr(''1.4) = (2.18)
(')
where is the planform of Hull 2.
Since we assume that Hull 2 is a mirror
image of Hull 1, the local surface
density of the source distribution for
Hull
2 in (2,18) will still be given by(2.17).
Thus, both
OiT
and020-
are nowcompletely determined. Our next task
is to determine and 942fr,, .
Solution to the Lifting Problem Suppose that Hull 1 is of zero thickness, but cambered, and is situated in close proximity to Hull 2 which has
finite thickness. As was previously
discussed, describes this portion of
the flow field around
Hull 1,
By (2.11),
4s,
is antisymmetrical with respect tothe plane y = O. This suggests that Oto.
can be represented by a distribution of doublets of surface densityiu in the
x,z-plane. To satisfy the
antisymmetri-cal property of (2,11), the orientation of the doublet distribution should be
normal to the x,z-plane. Analogous to
the situation of an incident flow past a
finite thin wing, we shall assume that there is now a steady vortex wake trailing
the stern of
Hull
1 and remaining in thex,z-plane. We shall also assume a
similar physical picture for
Hull
2.Furthermore, since the wake is now a sheet of discontinuity, the doublet
dis-, (0)
tribution extends over the wake
S.
aswell as over the planform 5 of
Hull 1,
Thus,
S.
(c.)where the suffix 3, + J
indicatesthat the integral is taken over the
plan-form of
Hull 1
and the wake trailingbehind it.
Similarly, for
Hull
2, we have.1) = 5?- a6,,,.e;x,,,2, .),,&../z, (2.20)
+
where the distribution is now extending
over the planform of
Hull
2 and its wake.The minus sign in (2.20) is due to the
assumption that Hull 2 is a mirror image of
Hull 1;
hence, the doublet density for Hull 1 is equal in magnitude and oppositein orientation from that of Hull 2. From
(2,14), (2,19), and (2.20), it is clear
that both
0,,
and 02/, have theproper-ties of a double-layer potential. In
particular, we have:
f.) --
96 (x,-ai.)-= Cx,0, z-,)z S74),-and tm< .26 -a, R-.)--=+47r/-4-(x., t,), .S:rSince, by (2.11) 451).., is an odd function
of y, (2.21) gives (2.17)
(2.19)
9bip..06t
0, i.) (2.21) (2.22) +o,(x,,.
A few more observations can be made
for the doublet density . The
expres-sion for the pressure appropriate for the
first-order approximation may be obtained
from Bernoulli's law as
= f1j-
= f U.
I O.'s+(2.25)
a Oa,/
Since, in (2,25), A is the only
function which suffers a jump
discontin-uity across the sheet
5
, (a)
,the pressure difference on both sides of this sheet is
401
u
(2.26)
=
-47rf UThe assumption that the pressure is con-tinuous across the wake implies that
or
ptcx.,-E0)=-}41-1.,?-0) (2.27)
x
(G)
in the wake . That is, ),A is
con-stant along lines which are parallel to the direction of flow in the wake. Similarly, we may also set "6/"'" 21x. =
along the keel if we assume continuity of
pressure. At the keel we shall further
assume that the function is
continuous. This means that
it (s, ) 0,
(2.28)
everywhere.
Since JA
(X00
is no longer relateddirectly to the local geometry of the ship, it is more difficult to determine 951/... .
,However, the second equation of (2.9) may
now be used. If we substitute expressions
in (2.15), (2.18), and (2.20) into (2.9), we may write ,) /6(,' ; t,) e SZ+SI:
(2.29)
-
?1This is an integral equation to be solved for the unknown doublet density if.". , The
right-hand side of (2.29) involves only known quantities, since the source density
q-(xi0)
is already related to the localslope by (2.17). However, this equation
becomes extremely complex if the complete expression of the kernel function G is to
be substituted from (2,15). A closed-form
solution of this complete integral equation appears to be beyond our resources.
Although we may try to solve this equation numerically, it is not advisable to take a
brute-force approach to evaluate the function G in the form given by (2.15)
without further reduction. Since, in this
expression, G contains a double-integral term with a highly oscillating integrand, a direct numerical evaluation of this term
is not only time-consuming but prone to inaccuracy.
Hu (1961) has shown that for a simpler case, i.e., for y = 0 and yo = 0, the
k-integral of the double-integral term in (2.15) may be expressed in terms of the
exponential-integral function. Since the
exponential-integral function is well behaved and is also extensively tabulated, the degree of numerical difficulty appears to be considerably reduced.
Z.) =
(4
1A4(xe, # )
(2.23)
.2 7r"
Similarly, for Hull 2, we have from (2.22)
j"-(x,g0
=r
(x.,26+0 .-E0. .27(2.24)
,,0
=-To fill the need of the problem here, Flu's approach has been extended to
a general case of G. It is shown in
Appendix A that (2.15) may be written in the following form,
G x., IL + 12.
-n- ?ct.
IdA e
tuo
va- ,e
LvPc-,(4-,8)]a.= (1-)'
the rest of the variables are already
defined in (2.15). The special function
El is the "exponential-integral function" as defined by equation (A.4) of Appendix
A.
In (2.30), the double-integral term has been reduced to a single integral. Since the function E1 is extensively tabulated and its method of computation well studied, an efficient numerical
scheme for evaluating G may be developed from (2,30).
As an alternate approach, we may also
try to solve the integral equation (2.29)
by an iterative perturbation technique. In fact, for a yawed-ship problem, Hu
(1961) assumed perturbation expansions for both G and the doublet densityik and solved the integral equation of the problem by an iterative procedure.
In terms of (2.30), the expansion for G used in Hu may be written as follows,
(2.31)
(2.30)
The remaining terms of (2.31),
G(n)
, n 1,2 3 are obtained by
making use of the asymptotic expansion
of the E1 function. Their explicit
expressions are given by equations (A,18)
and (A,19) of Appendix A. Since the
asymptotic expansion of the El function is valid only for large arguments, and since the argument of the El function in (2.30) is now v (a.,. L) which is related to the inverse of the Froude number squared, the expansion (2.31) is valid only for low Froude numbers.
G(C)
i
In fact, n (2.31) corresponds
to the limiting case of setting Froude number equal to zero, and is referred to as the "zero-Froude-number" approximation
of G, Hu further reasoned that Cic° and
represent the lowest- and highest-order terms in the expansion, respectively,
and pursued the iterative approximations accordingly,
In Flu's report, the computational results so obtained compared favorably with corresponding experimental data. Therefore, it may be inferred that a similar technique may be applied here to
solve the integral equation (2.29). On
the other hand, some investigators have experienced difficulty in using the
iterative technique based on a perturba-tion expansion which has the zero-Froude-number approximation as its first term
in the series, and have warned that such a technique may not be valid, In any case, the information available in this regard
seems inconclusive in proving or
dis-proving the validity of such a technique. Continuing efforts are being made to
solve the integral equation (2.29) both by a direct numerical scheme and by the
iterative perturbation technique. It is
felt that, short of rigorous mathematical proof, one can draw a conclusion about the validity of the iterative perturbation
technique only after a careful examination
(2.32)
2' a de
4s09-73)7. (2.33) (0,4
a_d (11)= 4
a;E,
, 11 ,,,,,-of the computed results obtained by
different methods, such as those mentioned above, and by comparison between the
computed results and experimental data.
Zero-Order Solution
In the following we shall attempt to derive an approximate solution to the integral equation (2.29) by replacing the complete expression of G by Ci("°. Such a solution is equivalent to the zero-order approximation in the asymptotic
expansion (2.31). The problem will be
further simplified by adding three more assumptions concerning ship geometry: (1) that each hull has a rectangular
plan-form, (2) that the draft-to-length ratio
is small, and (3) that the ratio between the hull separation distance and ship length is also small and is of a com-parable order of magnitude as the
draft-to-length ratio.
Thus, (2.29) shall now be written as
U3T-C,.4.
/7
Cr (X. Y'r, ; N.).1x,if we assume
,c;.°11
(2.34)
In the above equation, \--T is to be
substituted from (2.32). After such a
substitution, however, each term in (2.34) must be further manipulated into a form suitable for taking the limit as indicated. Thus, the first term on the left side of
(2.34) becomes (x, 1, z) I ; 4.a N.)
?4t-St.
2 I' z. ( -gat ,16,-,f--,1'4(i."-Nt/ , #= I
.1,1d Itc.64.g.)r
f -(2.35)
/It
g. ) - i.)(2.36)
Sincesatisfies Laplace's equation, (2,35) may be replaced by
#
(--L) + 2L. (-L )1
747,(*, 1. ) = ' Dx: /Z 4: /1
(2.37)
Next, we adapt the usual assumptions made for a rectangular wing (for example, Robinson and Laurmann, p. 232) that:
(1) i-t(,(,z.) = 0 along the finite
boun-S
(o)daries of 1 b I v./ ; (2)
along the bow profile and its reflection,
x = L/2; and, (3)
0
along thekeel and its reflection, zo = +H . Then,
by making use of these assumed boundary conditions, of integration by parts, and
of the fact that = 0 in the
wake, the following results may be established: ti 1.4 fro g.)
a)
-and(2.38)
?/4)
N(xr
4) -% _iy?x E. Lg+
(E-Z02:11Z-(2.39)
The algebraic details for deriving these results may be found in Robinson and
Laurmann (1956, pp. 232-237). After
sub-stitution from (2.38) and (2,39) into (2,37), we obtain 4/, /
)
/
1 4 `4. Cx- ) - f.)t (xr, =a-'
I,
=
:,:/;4e, x.) aN e-J'at
(11.
)
The integrals in (2.40) are singular
for y = 0. It can be shown that in this
case the correct limiting value is obtained by taking the principal values of the
integrals. Thus, upon taking the limit,
% H
= idqlf°'"*)/211L)
g
?2t1L)
z6 £/t_ ?2/1 --m -e.) .1 (Y-.26)%,. N- N, (7-.26)'..(1-141where we have assumed
N.) cr 6r. , - i-,)
(2.46)
4..24 dy, S4' -N 44t-i. ti.,4)7) cr(x., a 6"
2i
(2.45)
-N [fr-xdN4 44.4(2.42)
where we have again made use of (2.36) and
the fact that 1/A.. satisfies Laplace's
equation.
(2.43)
Since there is no longer any singularity
at y = 0 in (2.43), (A,c,-E) may be
obtained immediately from (2,43) as
= 0.111:4 ?x?,'4.
(7f((:2ii';(4(°-4.:*():417
(2.44)
1- N. 4 (4- i.).` 4 L
The remaining term to be similarly manipulated is the last term on the right
side of the equal sign in (2.34). This
term represents the componert velocity
normal to SV+ 51(ci) contributed by the
source distribution over the planform of
Hull 2, Hence, let us write
i/LL-.N)
(2.40)
-4N
?i.
(-te)r-Following a similar manipulation as
in the case of Art), , we obtain
1,7 (,
t.4
(x-..)(N-e.) -,1d1.71 D'Ac Ji(--N)'4-(1-,z)71 N 11+
?A(- -"N`')(2.41)
?Z.N-=
A/4)c
.)
)Z-Note that upon taking the limit y 0, the
second double integral in (2.40) tends to
zero. Hence, only the principal values
of the first double integral and the last single integral of (2.40) remain
Next, let us derive a similar result for the second term on the left of (2,34), Clearly, this term represents the component
(0)
(0)velocity normal to + due to
the doublet distribution over the planform
and its wake of Hull 2, Hence, we shall
write )6,4)1') y, =
?yi
sT,s,Z
e=
?'
1 iv-1 ),16,-,)'+ (N-4) -A (N+)0=if./CC (x,
i.)
(
1-.26)4) z..)4 -N (z-. (4ae#0 -,?14X-Because of the relationship between the source-distribution density and the local
ship-hull slope as given by (2,17), the assumption (2.46) means that the ship geometry is also an even function of Z, i.e., symmetric with respect to the
Oxy-plane, so that
.6(10,,,L.,)/x
The integral equation for the doublet-distribution density appropriate for the flow around a ship with two thin hulls may now be obtained by substituting the results obtained in (2.41), (2.44), and
(2.45) into (2.34). Hence,
-N
N
{(1(7-1:."24.L
)
(.-x.)a+ ei4 (I-4).
.44'4
2C(.,2) r% 2
,,,
1,Jab.",1"-,4e,
r.)
This is an integral equation of the
first kind for . Although
we may try to solve this equation numeri-cally, it appears that, for the problem
considered here, further simplifying
assumptions may be introduced. To do
this, we observe that the draft of a ship is generally much smaller than its length. Furthermore, for the steady forward motion of a twin-hull ship, the cross-flow effect becomes a problem of concern only when
the two hulls are sufficiently close to
each other. Hence, the two assumptions
introduced earlier concerning the draft-to-length ratio and separation
distance-to-length ratio will now be used. They
may be stated as
(2.48)
Thus the radicals Lc%-.0'.(7_ zOT2 and
(z-1.)'rz
may both bereplaced by IX-X- in (2.47). For
example, two of the terms in (2,47) will be approximated as follows;
(2.49)
lz-i.) (xx,) zo)
After making substitutions from (2.49) and (2,50) and carrying out the integration with respect to xo, the left side of the equal sign in (2.47) becomes:
r*
I
di,
2/4("'?') f(2.47) "
E 4
(2-2,)')
(2.51)To treat the last double-integral term in (2.47) similarly, we integrate by parts with respect to xo first before
replacing the radical
E(A-x)2.46,2
with X - . Thus, this term becomes:
= ° -1/
-jdX.
,x
/(x-x.)L + °=
I' I41'7-4;4::::)):
-#That is, H/L and 2b/L are both small and
their orders of magnitude are comparable, Substituting (2.51) and (2.52) into
so that the usual approximations made for (2.47), we obtain an approximate integral
low-aspect ratio wings may be applied here, equation for 1/"111. as follows:
(2.52) (2.50)
-.'W:(xuTg..)/ax.
44 0-4)1) (x'x.)(4-4,)t
(x-x.)/x-x."
' d 0-Eat- /1°)
(
j
46'4. (a-i-.)z-#
DC /f"
0,-(x,4)dz.
_ + 4 b?),
-HThis equation, aside from being simpler than the original one in (2,47), has an interesting physical significance
which shall be discussed here. It can be
shown, by integrating by parts and by making use of the fact that log
(/-ri
satisfies Laplace's equation in the plane perpendicular to the x-axis (i.e., in y- and z-variables), that(2.53) is precisely the same equation as
27 _N/ (x, 10)
i(g-g°f+ /1"t)%...0
_
(?- io)L 1 di,=
0-8(x, (./111( Z.,' ? -N (2.53)In the cross-sectional plane, x = constant, the left side of the equal sign in (2.53a) represents the net y-component velocity contributed by the two line-distributions of horizontal doublets, located at yo = 0 and yo.= 2b and extending from zo = -H to
zo = H. On the other hand, the right side
of this equation represents the cross-flow velocity due to both the camber of Hull 1
and the source distribution representing
Hull 2. Thus (2.53a) states that the
doublet density may be determined by
equating these two velocities when eval-uated at y = 0 (which corresponds to the
planform of Hull 1). Hence, ).k is now
to be determined two-dimensionally at
each cross-section. This is in agreement
with one of the familiar conclusions obtained from the slender-body approxima-tion.
To find the doublet density , we
note first that (2.53) is a complete
singular integral equation. Except for
some particular types (see, for example, Gakhov, §51) for which closed-form
solutions may be obtained, the solution of such an equation is generally carried out by a reduction to a Fredholm integral
equation. A number of techniques are
available. Here, we shall use the method
of Carleman-Vekua. Namely, we shall
eliminate the singular integral by solving the corresponding dominant equation (for example, Gakhov, pp, 186-194),
Morgan (1962) derived an integral equation of this type for a ducted pro-peller problem, and used the same method
to solve the equation. A similar approach
will be applied here to solve (2.53). First, let us rewrite (2.53) in the following form: 2.)
=
0e, z ) (2.54) --i whereF_
I De
x-rr ,/
4(/ 1 )
X (1*..Z ),1-1 -N Z., 4 h."-+Note that in (2.55) we have made a
sub-stitution for cr from (2.17).
By making use of a known inversion formula, (2.54) may be solved in a closed form if Fo is regarded temporarily as a
known function, Thus, according to this
formula, (2,54) may yield the following result:
e),zft
F Fc,-14
?,t (x,.)
d (2.56) -H E0 But '))0, 2.0 /1-1(x, iv) --/C(.. -NJ =o , (2.57),
Z. (2.55) at y = O. (2.53a) ?)140c.i-ie
tCy-y'..)1+
/A. .I ?Z -17-2110-Zt.,/ 1 /-17-
F
a.)di'
z where rm k,,( a'=
(4-Z)Z 40,
-H N)t, a) = _
_I?t(x.
-z ?), -rr-?x
(2.60)H
462+(z-Note that (2,58) is a Fredholm
equation of the second kind for
l',P'hz
.However, it may not be solved in this form since there are singularities at z . +H. Instead, let us define
where
(f0.2'),
-7' (ki)[.4W.e,Z!)1.1 ,#= -÷
F,(x'Z.)
dz °The kernel function K(z,z') and the known function F2(x'z) are both defined
by Cauchy principal-value integrals. In
general, for a given ship geometry, (2.62)
may be solved for
*Ala
by a numericalThen, in terms of this new unknown function, (2.58) may be written as
*(x,i)
I
k
(2.62)
(2.61)
III. HYDRODYNAMIC FORCE AND MOMENT
Thus far, both the thickness and lifting problems previously formulated
have been solved. That is, both the
corresponding source and doublet densi-ties may now be determined explicitly for
a given ship geometry. The logical next
step is to find the hydrodynamic force and moment acting on the ship in terms of
these known singularity densities.
Formu-las derived for such purposes are general-ly referred to as Lagalgeneral-ly's theorem.
Recall that we have obtained the flow field of our problem by two sheet-distributions of sources and normal (or
transverse) doublets. Across such
sur-face distributions of singularities, two important properties of the fluid velo-city are of particular interest to our
investigation, namely: (1) that part of
the velocity generated by a source dis-tribution which has a jump discontinuity in its component normal to the surface, and (2) that part generated by a normal doublet distribution which has a jump discontinuity in its component tangential
to the surface. These two properties
shall now be used to derive general expressions for force and moment.
In the subsequent development, we shall derive formulas for force and
moment acting on Hull 1 alone. Let
St
and
S2
be the wetted surface of Hull 1and Hull 2, respectively. Let $,, and
S,w
represent the plus and minus sidesof the wake sheet of Hull 1, respectively. Next, consider a volume of fluid,
which is bounded by the wetted surface
SI
and a control surfaceZ
enclosingHull 1 and its wake sheet. Thus, part
since we have assumed that tA is even in scheme without much difficulty. Finally,
z and also vanishes along the keel, Hence, the doublet density /4.,,.(,,,$) may be
com-upon substitution from (2,55) for Fo, pletely determined from the solution for
(2.56) becomes:
11-kxh,i
.(2.58)
(2.59)
(2.63)
(2.64)
1of consists of the free surface and
both sides of the wake sheet. Hence,
the combined surface
S
and /
boundsonly fluid. As before, we shall define
the positive direction of the unit
normal to this bounding surface
Si
to be pointing away from the fluid
volume -V-c . We shall assume that
contains no singularities, and therefore
Hull 2 lies outside of
Va
. For asteady-state problem we shall assume that
the control surface
E
moves with theship.
The total momentum of the fluid in
the volume tf, is
where io is the mass density of the
fluid, and i is the total fluid
velo-city relative to the ship as defined by
(2.4). Then, by making use of the
momentum theorem (see, for example, Wehausen (1973), p. 102), we may obtain an expression for the force acting on Hull 1 as
F
(3.2)
(Sn
(3.1)
in which 1, is the pressure, and
n
is the unit normal to . Next,
recall that the generating singularities representing Hull 1 are distributed only
on its planform
S,
Let (S)and
-represent the plus and minus sides (0)
of
Si6
, respectively. To simplify ourderivation, we shall assume that the planform of Hull 1 lies entirely within
Si
, the wetted surface of Hull 1.Then, the volume bounded by the combined
(o)
surface (
,S,V)
andkS,, )
alsocon-tains no singularities. Let this volume
be denoted by Atc . Note that the
difference between -V-, and -V-, in this
case is the displacement of Hull 1.
Let us now rewrite (3.2) as follows.
"L'i6L7
es`:) .-tan...
.
(3.3)
Applying Gauss'Gauss' theorem to the first sur-face integral in (3.3) results in
[V
f(V)9 ]
(3.4)where we have used a variant of Gauss' theorem,
5t
(g n Lis
III[(B.v)AA (i7,9 )
and the fact that
V.
= 0 for anincompressible fluid. Note that Euler's
equation of motion may be written with reference to the Oxyz frame as follows:
- -
T
If we neglect the external body force
IXI , then for steady-flow problems we
have
-74-f(iG9
= o
(3.5)
From (3.5) and (3.4) it is clear that the first surface-integral term in (3.3)
vanishes identically. Hence, the
expres-,
-sion for the force, (3.3),-becomes simply-:' In view of the decomposition (2:8),
' Let us write
in Which
a
substitution has been madefrom Bernoulli's equation for the
pres-sure
1,
Let us now define the unit normal
vector to the planform S,(r7) to be 1,
such that it is pointing in the positive
'-direction. Then, according to our
convention, the nOrmal vector m in (3.6)
is related to 1r; AS folloWs:
1
(ST14. n, =I -I! ,I,.. (s,71-.
,(0),
Recognizing the planform Jib to be a
sheet
pf
discontinuity.,, letus
alsodefine
c /1(c, F(i..no,.4) (SZ1
Furthermore, We shall decompose the
fluid velocity
I.
on either sideof
s to)
into the normal and tangentialcomponents as follows.
9,
Substituting (3.9), (3.8), and (3.7) into (3,6) and working out the necessary algebraic manipulations, we
may obtain an expression of
F
as
F
-fiRcricr)
÷i(c.c(f...-,LC)s-sT
-
(4.4 n.(4-1-1,..-2-y]ds.
( SZ1)+ ,
as,
in which we have defined
C41 ,-,
qt =
(t
(977'4- <7)
On the other hand, the discontinuity in the tangential component will bewritten
oo
(,-)
1-t -
9'4
r:(3.13)
In terms of (3.12) and (3.13), the three terms In the integrand of (3.10) beoome;
CZ" V)(Ct
4' 2( gv") g7',)
it 4 4 ircr
6"4" -;z1- ,( C4 0 (fL."-- 417r
e-(3.9)
)*(
g'-')/. = (lt
0.10) With substitUtions of the results
obtained in (i), (.40, and the
Thus, 9.e represents the uniform stream,
and the 4sturbance velocity due to
Hull 2, On the other hand, y.,
and
VIC
represent the disturbancevelo-cities due to the normal doublet and
(o),
source distributions on Sib
respec-tively; If we now decompose %fir and
yir
into the normal and tangentialcomponents according to (3.9)', then fran (2.16) and (3.11) we have, 6.;
Cil
- 4 rz--
(3.12)
F
f(l.f1)115 f. (3.,6) here (3. ,7.1)iI- f
4 f(pni #.-11W2.14,,
t
V UX OL,4 959,' )(c).4an-
e, 4 + VI,. t V (3171(3.8)
= = ,expression for the force in (3.10) becomes,
F
=-4f0- (-) cis
IL
,
j)-v]is
47r //a.
is
,(3.15)
s`ftwhere 'I- = q-t+ qv- represents the
average value of the velocities on both
(0)
sides of 316 . The first term of
(3.15) is exactly the familiar form of Lagally's theorem for the force acting
on a source distribution. However, (3.15)
states that for a continuous sheet dis-tribution, across which the velocity suffers a jump discontinuity, the average value of the velocities on both sides of
the sheet should be used when applying Lagally's theorem to compute the force.
Since (3.15) represents the force due to a simultaneous distribution of sources and normal doublets, the second integral must represent the force on
the doublet distribution. Nevertheless,
it is not in the familiar form of Lagally's theorem since it is now expressed in terms of vorticity rather than explicitly related to the density of the doublet distribution.
It will be shown next that the second integral in (3.15) can indeed be brought into the familiar form of
Lagally's theorem for doublets. However,
because the singularity distribution now represents a lifting body, additional terms must be added to the usual form of
Lagally's theorem for doublets. To
derive this result, we shall make use of the properties of the potential of
normal-doublet distribution. Since the
subsequent analysis involves integration by parts, in order to simplify our deri-vation we shall assume that Hull 1 has a
rectangular planform. Hence, the
boun-dary conditions imposed, following (2.37), on the doublet density and its deriva-tives for a rectangular planform will
still be used here. Extension to a
plan-form of a more ship-like shape is
straightforward but adds additional com-plications to the final result.
To begin with, let us write the second integral of (3.15) as
Recall that in (3.13) we defined as
the jump discontinuity in that component
of <1. which is tangential to
Thus from (3.13), (3.11), and (2.21) we have
Let us now examine each component
of in (3.18). After integration
by parts, the x-component gives
o
fi I;q-s
<ix!. 4 7rfJ7 ?
4,r,111 A 2.de. But if we define (3.16)-4-n-C?,"),
0 ?i4 ) (3.17)Substituting (3.17) into (3.16) and
noting that
-r
= (0,1,0) and ithe expression for Fr, becomes
F..=
4irark-
)
(3.18)ET
I
(x,/.1)/
/
{_L (2956,,,,t) (3.19)
ax
that is, the limit y 0 is to be
taken after the partial differentiation with respect to y, then
2),
With substitution from (3.20), the last
expression for becomes
Fi1=:-476r/a&de
S:
-47rd
1,g41
IE(3.21) (3.20)
The first term of (3.21) is the standard form of Lagally's theorem for a distribution of doublets with its
axis in the positive y-direction. The
second term of (3.21), on the other hand, can be identified with the induced
drag of a wing. As is indicated in
(3.21), the doublet density and the
(0) "averaged" velocity normal to
S
ib are both to be evaluated at x = -L/2, which is the trailing edge (i.e., the stern)of the planform of Hull 1. Note also
that (3.21) appears to provide an inter-esting insight into the difference
between a lifting and a non-lifting body.
For simplicity, let us consider the case of a thin asymmetric body making a uniform motion in an infinite fluid (which is incompressible and inviscid). So long as the flow field is to be generated by a distribution of
singu-larities on the planform of the body, both sources and normal doublets will
be required. Since in the wake the
doublet density remains constant along the direction of the uniform stream,
i.e., /...()(,z) AA (- L,2) , if we
assume /./. (x.,
0=
0
for any xo< -L/2,then there will be no wake trailing the
thin asymmetric body. In this case the
induced-drag term in (3.21) vanishes. Thus the flow problem is solved as a non-lifting potential problem which has its velocity field continuous
every-where outside the body. On the other
hand, if
/A
is non-zero at thetrail-ing edge, then there will be both a wake trailing behind the body and induced
drag due to the wake. Clearly, in this
case, the flow problem is solved as a lifting problem.
Applying a similar analysis to the
two remaining components of F,M. in
(3.18), we may obtain the following results, ----
477-f
#11'
chci,
o-& 47r!
(/
A( 9
I4t -M I+471-f j (0,
, )1 (3.22)2..
in which the second term is to be inte-grated along the trailing edge, and the third term is to be integrated along the waterline.
In summary, let us write the force acting on Hull 1 as
where
-477-1 IIa dx.1,-4
(3.24)
and F
has already been given by (3.22).(3.23)
an
Moment
To obtain formulas for the hydro-dynamic moment acting on Hull 1, we may use an analysis similar tO that uSed for'
deriving the force expression. However,
moment of momentum will be used rather
than momentum. We shall use the origin
of the Oxyz frame as the point of reference
for the moment expression. Thus, in place
of (3.1), the momentum, we shall consider the moment of momentum with respect to the origin 0 = .(0,0,0) in the control
volume as
follows:,
ih which
X =
(x,y,z) is the positionvector of a point P
(x,y,z)
in thecontrol volume.
Then, by a Similar Analysis applied
to YC.
,
we may obtain the expressionfor the moment acting on Hull 1 with reference to the origin of the Oxyz frame, as follows
1 z,
s,
crx
),z;
s:4
f
f,
= ivocr
Noi,
(3.26)where
Pio,-
and A4A- represent thefirst and second integrals in. (3,26), respectively.
ty a similar analysis following (3.16) we may obtain the final expression
for AV
,
as follows (3.25) e IA4 f
I 14-2.( g
sX--4rf
/Pt
Wx4
sC,, -1-47:17,142c, (-1
g,o)ji
'ft+47rfj
fit2..(n (0,4' E)11
. (3.27) lmoAgain, in this expression, the first two integrals represent the well known
result of Lagally's theorem. The last
two integrals in l3.27) are required only if the flow problem is being solved as a
lifting-surface problem. In particular,
the last integrals in (3.22) and (3.27) appear only for surface ship problems since they are to be evaluated along the
undisturbed waterline, z = O. These
integrals are obtained as the result of
integration by parts. Their physical
Meaning and numerical importance to a better prediction of the force and moment
are not yet Well understood- Hehte,
further investigations are required to assess their contributions,
IV. APPLICATION TO 'SWATH SHIPS
With the source density 0-6c,zey
and doublet density )1U6cZ0 obtained
in Chapter II, and the general formulas for force and moment derived in Chapter. III, we may now proceed to compute all three components of force and moment. However, since this work is still in progress and further computations are. being made, we shall not be able to covet' the entire scope of this investigation here, but shall continue to report the results as they become available... In this paper, we shall present the results ,of Our specific application of the theory,
namely: the prediction of wave
resis-tance of the Small-Waterplane-Area Twin, Hull (SWATH) Ships_
=
-N
A SWATH demihull is characterized by having a combination of an elongated, slender, round body which is totally sub-merged, and a thin strut which joins the submerged main body and extends above
the free surface. There has been a
con-siderable amount of interest in develop-ing such configurations as a new design
concept. As a part of the effort to
investigate the hydrodynamic performance of SWATH ships, an analytical tool for predicting the wave resistance was
developed. Such a tool may be obtained
by applying the theory presented in this paper.
To begin with, observe that the main body of a SWATH demihull has a round shape, and its transverse dimen-sion is wider than that of the strut. Such a configuration tends to prevent the flow from going around the keel. Hence, a SWATH demihull is less likely to behave as a lifting surface as
com-pared to a conventional catamaran. Thus,
in applying the theory to SWATH ships, it was assumed that wake may be neglected
in the initial investigation. Such an
approach does not represent a serious compromise in accuracy since, as a con-sequence of linearization, the effect of the wake may be superimposed on the results of the initial investigation as a later refinement, if necessary.
To obtain the wave resistance formu-la, we only have to be concerned with the x-component of equation (3.23). Since we have assumed that Hull 2 is the mirror image of Hull 1, the resultant wave resistance of the twin-hull ship is
twice that experienced by Hull 1. Thus,
with the effect of wake neglected, we may obtain from (3.23):
=
F
(
Fr+
F,)
= z (R
(4.1)
where
R.. = 16frpk:
fa.
sic'ett.c...(2bk. sees *me)°
(4.2)
R. R.,. = /de5ec'esin(2bk.s.c'es,ne)
X[R. , (4.3)
Re?. = 16
frfk: f
de see8[1- cos(2 bk. sec2E3 sonX [P;
(4.4)
k $ee0
cr(.x.z)e cosl (kx sece),
(a' S' sm
= k. secle fidx,i)....(v.s.)
kZsec'e{c5c7s-1(k/Sete)
o.
k= 3/u2
, andS
is theprojec-tion of et! and
Sft
onto the Oxzplane.
The quantity represents the
wave resistance experienced by one hull contributed by the source distribution
alone.
Similarly,-Rpiu
is that due tothe transverse dipole distribution alone,
and Pcp.4-Ritvo- is due to the
inter-action between the sources and the trans-verse doublets.
No-Cross-Flow Assumption and Its Consequences
Here we shall discuss a special
con-dition on camber. Under such a condition,
each demihull will experience no
cross-flow effect. Recall the integral equation
(2.34) which is to be solved for the
unknown doublet density )4 . Suppose
now that a demihull is so designed that its camber satisfies the following relationship:
.?D .24
(21-.7x.)
(4.7)
S'
Then, since the right side of equation (2.34) vanishes, we may conclude that
}-1(xe,?..)=0,
(4.81 (4.5) (4.6) ,X e ! ) 32if the integral equation has a unique
solution. Thus, if the condition (4.7)
is satisfied, then the force and moment
associated with Ak vanish. In
particu-lar (4.1), the wave resistance formula, simplifies to merely
2 (4.9)
Thus, the relationship (4.7) represents
an interesting design application. If
we apply the low draft-to-length ratio and the small separation-distance-to-length ratio approximations used in obtaining (2.52), the relationship (4.7) may be further simplified as follows:
N
Z) = - dzo
7r 46'4 (4.10)
-N
with the boundary condition
i)=0
and C(_1-/z,z)=0 . According to
(4.10), the no-cross-flow camber varies
with z as well as with x. In practice,
however, it would be more convenient and sufficiently accurate to use an averaged value of C(x,z) so that it would be
con-stant along the draft and would vary
only along the length. If we let -C. be
such an averaged value, then from (4.10) we have:
Ftw-i(xZ-) d
-(4.11)
1
/ 46")[4'210-1-:T )/d.
Z,H7r
Thus, the averaged no-cross-flow camber may now be determined for each station by using (4.11).
Results of Computations and Comparison with Experimental Data
In making wave resistance computa-tions for SWATH ships, we have assumed that the ship either has camber which
satisfies the no-cross-flow requirement of (4.11), or a low
beam-to-separation-distance ratio. Under such an
assump-tion, the simplified formula (4.9) may be considered an acceptable
approxima-tion. Thus, with this simplification,
only the source distributions are now
involved. Nevertheless, as can be seen
from (4.2) and (4.5), the actual compu-tation is still quite involved for an arbitrarily given ship geometry.
To take advantage of SWATH geometry, the source distribution is further simpli-fied to the sum of a sheet distribution representing the thin strut, and a line
distribution representing the body. With
this simplification, only the thickness of the strut and the cross-sectional area of the body appear explicitly in the
wave-resistance formula. However,
carry-ing out the computation still requires the strut thickness and body sectional
area to be described analytically. This
may present problems since hull geometry is usually given in terms of offsets.
To solve this problem, a special curve-fitting technique has been developed. This technique employs the Chebyshev
series, utilizes the offsets of strut thickness and area curve of the body, and provides analytical expressions for the strut-thickness function and the
body area curve. Analytical development
of this technique is presented in Appendix B.
With two Chebyshev-series represen-tations, one for the strut thickness function and one for the body area curve, the expression for wave resistance may be put in the following final form:
Rw = 2 (Rs+ R. Rs.) (4.12)
where R' RB' and RSB represent the
S
wave resistance of one strut, one body, w =
i(--vt,
and the interaction between strut and
body, respectively. Furthermore, each
of the three components is now expressed as a finite sum of the products of
Chebyshev coefficients and auxiliary
functions. For example, Rs is given as
follows;
Rs = (41-es-ra Lsros)
x
{A,ATsmn+B.,BW,,,,n1
, (4.13)nI
where T and L are the maximum thickness
and length of the strut, respectively,
and
4
is a dimensionless number relatedto ship speed. The equations for Rs and
RSB have a similar structure and are
given by equations (B.9) through (B.17) in Appendix B.
A computational program has been developed on the basis of those final
sets of equations. Since the three
principal equations for Rs, Rss, and Rs have a similar structure, we shall use
the one for R as an example for
discuss-ing some of the computational aspects of
the program. In the above equation for
R the Chebyshev coefficients' Asm and
S'
Bsm are determined solely by
strut-thickness offsets, while the auxiliary functions, Tsmn and Wsmn' are determined by the draft-to-length ratio and the speed (or, to be precise, the Froude number); that is, the auxiliary functions do not vary with changes in strut shape. Thus, in case there are several alter-nate designs to be investigated, the
auxiliary functions need to be computed only once provided those designs have
the same draft-to-length ratio.
In making the wave resistance com-putations, the major portion of computer time is taken up by the auxiliary func-tions, while the Chebyshev coefficients require only a minor amount of effort.
Hence, the present development is especially suitable for investigating
the effect of hull-form variations on
wave resistance. In particular, it would
be a straightforward extension to formu-late an optimization problem to investi-gate the theoretical hull form of
"minimum" resistance. The procedure developed in Lin, Webster, and Wehausen (1963) may be easily adapted to the
present formulation. The computer
pro-gram developed on the basis of the present analytical work has been applied
to most of the SWATH models investigated
at NSRDC. Only the results of two
SWATH models will be reported in this
paper as typical examples. Since no
experiment was conducted to measure wave resistance directly, comparison of theo-retical predictions can be made only
against residuary resistance. Despite
the fact that residuary resistance con-sists of the so-called "form drag" as well as wave resistance, good
qualita-tive agreement between residuary resis-tance and theoretical wave resisresis-tance was observed.
Schematic diagrams and geometric characteristics of the two SWATH models are given in Figures 1 and 2 and Table I,
respectively. The demihulls of both
models have similar cross-sectional shapes, the major difference being that SWATH III has two "straight" (i.e., zero camber) demihulls, while SWATH IV has a slight camber on each demihull intended to minimize the cross-flow effect.
Another distinction is in the waterplane shapes, as can be seen from the diagrams.
A number of test cases has been devised to investigate the merits of this theoretical tool in making
resis-tance predictions. First, to test its
ability to reflect the effect of a small change in hull geometry, the demi-hull of SWATH III with different strut sm
AWNIII .12.1AL t
-
d.)
I I 11551SL IN F. 20 19 19 17 lb 15 14 13 IN ill ç IS 6 5 4 3 2 1 0 STATFIGURE 1 - REDUCED' SCALE AWINGOF WTH ill FORESENTED BY WI 5275
CMOIER
11/10402I L100
20 19 II 17 16 15 14 la 12 11 10 9 11 7 6 5 4 1 2 I 0
,}47 lows
FIGURE 2 - RIM], SCALE DRAWlit OF NAN IV KRUM) BY MIL 576?
shapes was used. Results of theoretical
predictions and the corresponding residuary resistance coefficients are
presented in Figure 3. In this figure,
the same main body was used for both
cases, Although the only difference
is in the strut shape, experimental data show marked differences in the residuary-resistance characteristics
at lower speeds. This trend is very
well predicted by the theory. However,
at the higher speed range, the compari-son between theory and experiment is not as favorable for this specific
example. Although both theory and
experiment indicate that differences in the resistance characteristics will be small at higher speeds, theory predicts
a different trend. One thing that may
help explain this reversed trend is the fact that the "original" strut possessing a "Coke-bottle" shape may have a higher-than-usual percentage of "form drag" and therefore a higher residuary resistance,
TABLE I
Ship Dimensions and Coefficientz For SWATH Demihulls
Note: For a Twin Hull Configuration the
Separation Distance Between Demihull
Center-line is 3.68 feet, model scale. This
corresponds to a beam/separation distance ratio of 0.106.
The theory, on the other hand, is intended to predict only wave resistance and does not account for the viscous form drag of the Coke-bottle effect.
Figure 4 shows a comparison between
single- and twin-hull ships. Since the
demihulls of SWATH III are "straight", each is exactly one half of a twin-hull
ship. In this figure, the solid- and
dotted-line curves show the predicted wave resistance characteristics of the twin-hull (i.e., SWATH III) form and its
demihull, respectively. For most speeds,
the twin-hull interference effect amounts Wetted Surface (sq ft) (S) 17350 17540 LCB/LOA 0.479 Length/Diameter of Body 16.55 16.02 Cp (Body) 0.758 0.758 C (Strut) 0.709 0.740 T/Ls(Strut) depth to E body 0.035 1.12 0.035 1.06 diameter of body
Design Speed in Knots 32.0 32.0
Scale Ratio 20.4 20.4
SWATH III SWATH IV
NSRDC Model Number 5276 5287
Length of Body (ft)(Ls) 267 288
Length of Strut (ft)(L,) 226 227
Diameter of Body (ft)(D) 17.3 18.o
Beam of Strut (ft) (t) 8.0 8.0
Depth of Submergence
to g Body (ft) 19.4 19.0
Total Draft of Demihull
(ft) 28.0 28.0 Displacement (tons, s.w.) 3760. 3960. -,
08
0.2 025 03
SWATH III DEMIHULL
SWATH (CAMBERED DEMIHUL LS)
28 FT DRAFT 8 75 FT SPACING FOR PROTOTYPE RESIDUARY RESISTANCE COEFFICIENTS FROM EXPERIMENT
, MODEL FREE TO TRIM 8 HEAVE
.^, MODEL CAPTIVE
10 12 14 16 18 20 22 24 26 28 30 3; 34 36 PROTOTYPE SPEED IN KNOTS (226 FT WL LENGTH)
STRUT SHAPE
SPEED-LENGTH RATIO V/VLSTRUT
FIG 3 -EFFECT OF STRUT VARIATIONS-COMPARISON BETWEEN
THEORY AND EXPERIMENT
---2 0 7
114 16 113 20 2.2 24
SPEED-LENGTH RATIO V/VLSTRUT
FIG 4 -EFFECT OF TWIN - HULL INTERFERENCE ON RESISTANCE CHARACTERISTICS
THEORETICAL PREDICTION OF WAVE RESISTANCE COEFFICIENT
1 10 1.2 4 16 18 20 22 24 SPEED-LENGTH RATIO V/VL T 04 0.5 0.6 FROUDE NUMBER v/V91 STRUT
FIG 5 -COMPARISON BETWEEN THEORETICAL
PREDICTION AND DATA FROM TWO EXPERIMENTAL TECHNIQUES
to an increase in resistance. However,
at certain speed ranges (for example, the speed-length ratio between 1.3 and 1.5), the interference effect becomes favorable. In this case, the twin-hull ship would have less resistance than would two
demi-hulls travelling individually. The
experimental data appear to substantiate this predicted trend very well.
Figure 5 shows the theoretical pre-diction and its comparison with
experi-mental data for SWATH IV. Data from two
model-experiment techniques, one with the model captive and the other with the model free to trim and heave, are
pre-sented. In contrast with SWATH III,
SWATH IV has camber which makes the flow field more closely satisfy the no-cross-flow assumption of the present theory. Thus, the agreement between theory and experiment in this case is the most striking among all the cases
investi-gated. It is also evident from this
figure that the theory agrees better with the data from the captive-model experi-ments.
Comparison of the computed and experimental data shows that, for SWATH ships, we may expect generally good correlation between theoretical wave
resistance and residuary resistance. A
number of reasons which are rather obvious may explain such good agreement. First, the SWATH ships have thin and slender demihulls which well satisfy the
assumptions of the theory. Second, the
traditional method of estimating viscous drag (i.e., Froude's method) must be quite satisfactory, at least qualita-tively, for SWATH ships so that residuary resistance consists mostly of wavemaking
resistance. Third, the speeds of
inter-est for SWATH ships fall in a higher Froude number range for which the theory
is known to be more reliable. As was
previously mentioned, lifting effect seems to be less important to SWATH
con-THEORY (WAVE) EXPERIMENT (RESIDUARY) SWATH 10 L\ DEMIHULL OF SWATH LI
0
z 1.0 0 w 0 cr 0 6 08 iO 2 2.0 .0 06 0.8 "> 3.0 al4.0 z 3.0 -0 .0. I-0SHAPE THEORY IEXPERIPNT
(OWE) tRESIDUARy)
0
0 0 2 0 1.0 k2 1.4 .1.6 2.4 . STRUT 0 0 0 0,7figurations. For a conventional cata-maran, however, lifting effect may become
rather significant. In this case, the
dipole distributions as well as the source distributions must be considered simultaneously.
ACKNOWLEDGMENT
The author is indebted to Professors R. Timman and J. N. Newman for many help-ful ideas and discussions; to Mr. V. J. Monacella for continued encouragement and guidance through the course of this
investigation; and to Miss Claire E. Wright for her expert assistance in editing and preparing the manuscript.
This work was supported under the Naval Ship Research and Development Center's General Hydromechanics Research Program and was funded by the Naval Ship Systems Command under Task Area
SR 023 01 01.
REFERENCES
Cummins, W.E. The force and
moment on a body in a time-varying
potential flow. J. Ship Res., Vol. 1,
No. 1 (1957) PP. 7-18 & 54.
Eggers, K. 'Ober
Widerstands-verhalt nisse von Zweikorperschiffen. Jahrb. Schiffbautech. Ges. Band 49
(1955), pp. 516-537; ErOrt. 538-539. Gakhoy, F.D. "Boundary Value Problems" (1963) English translation by I.N. Sneddon, Pergamon Press.
Havelock, T.H. Wave resistance:
the mutual action of two bodies. Proc.
Roy. Soc., 2er. A, Vol. 55 (1936), pp. 460-472.
Hu, P.N. Forward speed effect on
lateral stability derivatives of a
ship. Davidson Lab. Report 6-829
(Aug 1961), Stevens Inst. of Tech.,
38 pp + 2 figs.
Kellog, O.D. "Foundations of Potential Theory," Dover Publications, N.Y. (1929) ix + 384 pp.
KOchin, N.E. On the wave-making
resistance and lift of bodies submerged
in water. (Russian Org., 1936)
Trans-lated in Soc. Nay. Archs, and Mar. Engrs. Tech. Res. Bull. No. 1-8 (1951), 126 pp.
Landweber, L.; Yih, C.S. Forces,
moments, and added masses for Rankine
bodies. J. Fluid Mech. Vol. 1 (Sep
1956), pp. 319-336.
Landweber, L. Lagally's Theorem
for multipoles. Schiffstechnik Bd. 14
(1967), pp. 19-21.
Lin, W.C.; Day, W.G., Jr. The
still-water resistance and propulsion characteristics of
Small-Waterplane-Area Twin-Hull (SWATH) Ships. AIAA/
SNAME Adv. Marine Vehicles Conf.(1974), San Diego, Calif., 14 pp.
Lin, Wen-Chin; Webster, W.C.; Wehausen, J.C. Ships of "minimum"
total resistance. Int. Sen. Theoret.
Wave Resistance, Ann Arbor (1963), pp. 907-948; disc. 949-953.
Lunde, J.K. On the linearized
theory of wave resistance for displace-ment ships in steady and accelerated
motion. Trans. Soc. Nay. Archs. and
Mar. Engrs. Vol. 59 (1951), pp. 25-76; disc. 76-85.
Magnus, W.; Oberhettinger, F.; Soni, R.P. "Formulas and Theorems for the Special Functions of Mathe-matical Physics." Springer-Verlag, New York, Inc. (1966), VIII + 508 pp.
Morgan, W.B. Theory of the
annular airfoil and ducted propeller. Sym. Nay. Hydrodyn., Washington, D.C.
(1962), pp. 151-197.
Muskhelishvili, N.I. "Singular Integral Equations." English Trans-lation by J.R.M. Radok, F. Noordhoff N.V.
Fien, P.C.; Lee, C.M. Motion and
resistance of low-waterplane-area
catamaran. Sym. Nay. Hydrodyn., Faris
(1972).
Robinson, A.; Laurmann, J.A. "Wing Theory." Cambridge University Press (1956), ix + 569 pp.
Wehausen, J.V. The wave resistance
of ships. "Advanced in Applied Mechnn-ics," Vol. 13, pp. 93-245, Academic Press, New York (1973).
Yokoo, K.; Tasaki, R. On the
twin-hull ships. Rept. Transportation
Tech. Res. Inst. of Japan, No. 1, 20 pp.
(1951); No. 2, 9 pp. (1953). Translated
In U. of Michigan Dept. of Nov. Arch. and Mar. Eng., Rept. No. 33 and 34
(1969).
Zucker, R.D. Lagally's Theorem
and the lifting body problem. J. of
Ship Res. Vol. 14, No. 2 (1970), pp. 135-140.
EXPONENTIAL-INTEGRAL FUNCTION AND HAVELOCK SOURCE POTENTIAL The Havelock source potential given
in (2.15) may be expressed in terms of the well tabulated exponential-integral
function. To this end, we first rewrite
(2.15) in the following form:
(-;6, 0,2, IT
-
db/ d
eAZI
-4-id,
(4.4..49) eGE,....4oci 0 )(Atz-f-i.6,-,)4,7 6,414. (w--.7.) a -4-5 9_
(A.1)I
IT"
e
04.-x .)t.jC.4
tvt;-4-where a ----
z + z< o,1) = ko sec20 >
0,
and = (x-x. ) cos() + (y-ye)
sine.
The third term in (A.1) involves a double integration and thus presents the
major computational difficulty. Hence,
our primary concern is to express the k-integral in terms of a well-known
func-tion for which computafunc-tions may be
per-formed efficiently.
Let us first introduce the follow-ing change of variable:
= - eiv)(a-÷-:ez)
(A.2)Then, by making use of the definitions of exponential-integral functions, the Cauchy integral theorem, and a suitable choice of the paths of integration in a complex plane, the following identity may be established:
APPENDIX A
vo,,z)
IF;(7.4,,,,,2-4±: 7r
I
,0;
(A.3)
The special functionsEt and Ei in
(A.3) are known as exponential-integral functions and are defined as follows:
,...
-t
E, 6414 =
,..,
(a)= -7-11fr >°
Further information concerning these functions may be found in Abramowitz and Stegun (1964, pp. 228-252).
Next, we substitute (A.3) into the
double-integral term of (A.1). However,
since the
e
integral ranges from -7r to+tr, and variable 2Z changes sign in this interval, care is required in making the
substitution. Let us introduce the plane
polar coordinates /9 and such that
x-xo .(ocosis
Then, the variable Z., may be written as
-.(A-,..),cs#9
- /e9cc-4e74)(A.5)
Thus, the double-integral term in (A.1) becomes