ARCH1EF Lab. y.

Scheepsbouwkunde
Technische Hogeschool L

Deift

THE INERTIA OF THE WATER SURROUNDING A VIBRATING SElF

B F1&NK M. LEWIS, MEMBER OF CouNcIL*
[Read at the thirty-seventh general meeting of The Society of Naval Architects and Marine Engineers, held in New
York, November 14 and ES, 1929.1

1. The important elements which determine the natural frequency of a vibrating

ship are its mass and the distribution of this mass, and the rigidity of the hull structure.
In very general terms,
The natural frequency varies as rigidity
mass

It has long been recognized that the water surrounding a vibrating ship produces an
effect equivalent to a very considerable increase in the mass of the ship. In the present
paper we discuss this water inertia effect. The discussion will be limited to vertical flexural
vibration of two and three nodes.
At the present time the natural frequency of a ship can be calculated, if there is
any attempt at accuracy, only by more or less empirical methods; that is, by comparison
with similar ships of known frequency. The unknown elements which have so far pre-
vented a purely rational calculation are the water inertia and the rigidity of the hull in
flexure and shear.
We believe that the first item can be evaluated with fair accuracy by the methods
explained in this paper. Accurate knowledge regarding the second is still lacking. It is
recognized that the fiexural rigidity of a hull is somewhat less than that based on the nomi-
nal I of its cross-section, probably due to the tendency of compression members to shirk
their load. This point can be cleared up only by careful experiment. The classic experi-
ments on the Wolf give practically the only results at present available, l These were
made some twenty-five years ago on a hull of very light scantlings. Until further experi-
ments on hull rigidity have been made, the frequency calculations must rest upon an
empirical basis.
While this matter of the calculation of hull frequency is of great interest, we take
this opportunity to point out again that the road to the ideal vibrationless ship does not
lie in that direction. Such a ship can be produced only by the elimination of all periodic
exciting forces exerted on the hull from engines or propellers. So long as such periodic
forces exist, some vibration will exist in the hull at all speeds, although its amplitude may
be very small. If no periodic forces act upon the hull, there will be no vibration in t at
any speed, regardless of its natural frequency. Therefore, every effort should be made to
avoid such periodic forces, and if they can be avoided the natural frequency of the hull is
unimportant.
The evaluation of the water inertia by experimental model methods offers very
*Professor of Engineering, Webb Institute of Naval Architecture, New York.
tJ. H. Biles, "The Strength of Ships with Special fleference to Experiments and Calculations made upon H. f.
S. Wolf. Trans. Inst. of N. A., 1905.
 2 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
considerable difficulties. Experiments have been made by Nicholls, which we discuss later,
using a series of wooden blocks fastened to a steel strip, but the results are not applicable
to ship forms.
We attempt to evaluate the water inertia by purely mathematical methods. No
exact solution is possible, but with certain approximations and assumptions a result can
be obtained which we believe to be sufficiently accurate for any practical purpose. We
consider these approximations in order.
In Fig. 1, Plate 1, the full lines represent the underbody of a ship and the dotted
lines its "image," obtained by inverting the underbody about the water plane. It is
assumed (A) that for vertical vibration the inertia of the water surrounding the actual
ship with a free water surface is one-half of that of the ship and its image together, vibrating
totally immersed in a fluid extending to infinity in all directions. With such an assumption
the median plane is no longer a constant pressure surface, but the error involved is of the
order of the square of the velocity and, because of the small amplitude of vibration of a
ship, will be of very small magnitude. For transverse vibration this assumption will not
hold, and we make no attempt in this paper to evaluate the water inertia for this case.
It is assumed (B) that the frequency of a ship, and consequently the water inertia,
is the same whether it is moving or stationary. Experiments have been made on this
point, the frequency being determined with the ship under way and when at rest with the
propeller disconnected, and no change in the frequency could be detected.
It is assumed (C) that no eddies or other discontinuities in the fluid motion, due
to the vibration, are present. This is justified by the extremely small magnitude of the C,

fluid velocities, so that perfect stream-line motion should obtain even with abrupt changes
of shape.
Since the length of a ship is several times its beam, it is evident that the motion
of the water will, for the greater part, be parallel to transverse planes.
Consider a ship vibrating vertically. The kinetic energy of the surrounding water at
the instant of maximum velocity has a certain value. Imagine the water to be confined
by a series of thin transverse partitions extending to infinity so that it could move only in
transverse planes. Then the kinetic energy of the fluid would be greater than if it were not
so confined. We designate the ratio of these two kinetic energies as the longitudinal
inertia coefficient J.
Actual K. E. of surrounding fluid
-' - K. E. of surrounding fluid if motion is confined (1)
to transverse planes.

J obviously depends on the proportions and shape of the underwater body, and will.
approach unity as the ratio of length to beam. becomes greater and greater.
Among the limited number of three dimensional problems in fluid motion for
which an exact solution has been obtained is that of the ellipsoid of revolution. In Appen-
dix I we give the solution for the flow around an ellipsoid of revolution vibrating perpen-
dicularly to its axis in two and three nodes. We assume (D) that the longitudinal inertia
coefficient J for a vibrating ship is the same as for such a vibrating ellipsoid of revolution
having the same ratio of length to beam.
A closer approximation would probably be given if we took an ellipsoid of three unequal 4

axes, making these the length, beam, and twice the draft, but we have been unable to
obtain a mathematical solution for this case.
 INERTIA 0F WATER SURROUNDING A VIBRATING SHIP 3

A graph of J for two and three noded vibration is given in Plate 4. With assumption
(D) the problem is reduced to one of two dimensional flow.
11. While the problem of finding the two dimensional stream-line flow past a cylinder
of specified shape has never been solved, the powerful methods of conformal representation
yield solutions for a great variety of shapes, some of which may approximate more or less
closely to actual ship sections. Among these shapes around which the flow can be deter-
mined mathematically, we find that those given by the parametric equations
X [(1 + a)cos0+ bcos3û] (1 + a+ b)
Y = [(1 - a) sin G - b sin 30] (1 + a + b)
give an approximation to actual ship sections. The approximation is especially good for
the sections with flat bottoms and rounded bilges amidships where the water inertia effect
is most important.
In Plates 2 and 3 are drawn the sections having the foregoing parametric equations
for various values of Ralf Beam/Draft and constant b. The quarter section of the com-
plete cylinder (ship and image) is drawn in each case.
12. For a cylinder of elliptical section moving in a fluid the kinetic energy of the sur-
rounding fluid per unit length is given by*

2T
g

where
T = kinetic energy.
B = half breadth perpendicular to the motion.
w = weight per unit volume of fluid.
g = gravity constant.
y = velocity.
In this analysis we also take
L = half length or semi-major axis of ellipsoid.
B = half beam or semi-minor axis of ellipsoid.
D = draft:
For any other shape of section of the same axial breadth we write the energy as
CirE2 WV2
2T
g

and call C the inertia coefficient for that section.

The surrounding fluid has the same effect as the addition to the cylinder of a mass per
unit length
M = CirB2w
For our ship sections we will write the additional inertia mass to be added per unit
length as
M=3/2CJirJ32w (2)

*Lamb Hydrodynamics, Par. 70 (11) or see Formula (9), Par. 29 of this paper with b = 0.

I
 4 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
where J is the appropriate longitudinal inertia coefficient for the length beam ratio of the
ship and the 2 is introduced since we are interested in the ship alone and not in ship and
image together, as explained in Par. 6.
In Plates 2 and 3 the number noted on each section is the inertia coefficient C for
that section. These sections are drawn for ratios of Half Beam/Draft = of .2, .4, .6, .8,
1.0, 1.2, 1.5, 2.0, as marked on each. The sections marked 1.0 are ellipses.
There is also shown on each section the inertia coefficient for the rectangle and the
inertia coefficient for the rhombus of the same axial dimensions. These are obtained as
explained in Appendixes III and IV. Plate 4 gives C for the rectangle and rhombus for a
range of values of B/D. We assume (E) that the inertia coefficient for actual ship sections
can be determined by comparison with the figures of Plates 2 and 3.
The calculation for a ship is carried out as follows:
The body plan is drawn for the usual ten divisions. It is necessary to estimate C for
each of the stations. This may be done by visual comparison of the body plan with the
sections of Plates 2 and 3, or more accurately as follows: For each station the value of
Half Beam/Draft = H is determined. Using a pair of proportional dividers, each sec-
tion is transferred to tracing paper, its dimension being expanded or contracted so that
it will have the same half beam and draft as the section in Plates 2 or 3 having the nearest
larger H value. Another section is drawn for the nearest smaller H value.
Laying the sections on tracing paper over the appropriate figures in Plates 2 and 3,
the C value of each may be estimated. The C value of the actual ship section is obtained
by interpolation between the two values of C for adjacent H values.
For each of the stations the water inertia mass per foot of length will be given by (2),
where B is the half breadth on the water line at that station.
The curve of water inertia over the length of the ship is thus constructed and added
to the ordinary weight curve. The combined weight curve is used to calculate the natural
frequency by any of the methods which have been devised.
As an illustration we give the calculation for the ship whose weight curve was
given by Nicholls. * The characteristics of the ship are as given on Plate 5. While the
lines there shown are not the actual ones belonging to the weight curve given, they have
the same overall dimensions and coefficients and will serve for purposes of ilustration
The values of C at each station were determined as per Par. 14 and are noted on each
station. The J value from Plate 4 is .86, corresponding to L/B 10.2. Calculating the
water inertia mass at each station by (2), the curve of water mass is drawn to the same
scale as the weight curve. The total water inertia mass is 1,615 tons. The actual weight
of the ship is 1,378 tons. The distribution of this mass is equally important with its mag-
nitude, however. Calculating by the tabular method given by the writer,t we find the
curves for the neutral axis of the bending, when the water inertia is and is not taken int
account, to be as shown on Plate 5. With the water inertia allowed for, the nodes are
nearer the center and the ratio, amplitude at ends over amplitude at center, is considerably
greater, than when no such allowance is made. In the example cited this ratio has been
raised from 2.6 to 3.6. The frequency has been lowered in the ratio i 1.6S, or 23
per cent. Thus, while the mass of the ship has been increased 117 per cent, this would be
equivalent in lowering the frequency to an increase of only 68 per cent, provided the 68
per cent increase had the same distribution as the ordinary weight curve. The difference.
*The Vibration of Ships," H. W. Nicholls. Trans. Inst. of Naval Architects, 1924.
f'Vibration and Engine Balance in Diesel Ships." Soc. Naval Arch. and Marine Engineers, 1927.
 INERTIA OF WATER SURROUNDING A VIBRATING SHIP a

between these two percentages, the actual increase of mass and the virtual increase, shows
that the distribution ,f this mass is a vital factor and must be taken into account.
It can be stated that this large increase in mass is, in general, consistent with the il
elastic properties of this hull and its known frequency, but due to lack of accurate knowl-
edge regarding these elastic properties no precise correlation of the various factors which
fix the frequency can be made at the present time.
Certain general conclusions can be drawn.
The flat bottom amidships is of the greatest importance. Due to this the greater
part of the added mass is amidships, with the result that the ratio of amplitude at ends
over amplitude amidships is much greater than would otherwise obtain.
The water inertia is nearly independent of draft so that the natural frequency will
vary only slowly with the displacement, and it will not vary as the square root of the dis-
placement for different displacements of the same ship.
It is of interest to compare the observed frequency in the experiments of Nicholls*
with those calculated by the foregoing methods. Nicholls' experiments were made with
steel bars, supported at their free-free nodes, to which were fastened wooden blocks. The
frequency was determined in air and at various depths of immersion. The calculated fre-
quencies in water have been obtained from the observed air frequency on the assumption
that the frequencies vary inversely as the square root of the actual mass plus the water
inertia mass.
Case 1.-Bar of steel 30" X 2" X .312" carrying wood of thickness 13". Weight
of steel and wood in air 102.7 ounces.

In air In water

Depth of immersion O 1" 13"

Observed frequency 89.4 80.26 78.77 77.96 77.59
Ca1cuated frequency 81.0 78.3 77.3 76.8

Case 3.-Bar of steel 30" X 2" X .119" carrying wood of thickness 13'". Weight
of steel and wood in air 48.54 ounces.

In air In water

Depth of immersion O 1" 13C'

Observed frequency 49.6 40.57 39.13 38.61 38.50
Catcuated frequency 40.8 38.6 37.9 37.4

In the foregoing the LIB ratio is 15, the corresponding R value is .908, and for the
nodal position of a free-free bar J would be .93. Due to the restriction of the transverse
flow by the rectangular section, it is evident that J will be somewhat smaller, and we have
arbitrarily assumed it at .85. The C values were taken from Plate 4.
Case 2.-Wood triangular in section, weight of bar and wood in air 94.18 ounces. The
observed frequency in air was 89.48 V. P. S. and in water at 1 inches depth of immersion
p
T. 1. N. A., 1924.
 6 INERTIA OF WATER SURROUNDING A VIBPATING SHIP
82.83. The calculated frequency is 82.25. The longitudinal inertia coefficient was taken
as .85 and the section coefficient .730 from the curve of Plate 4 for the rhombus.
Since a ship form resembles the ellipsoid of revolution, for which the calculation is
theoretically exact, much closer than these rectangular bars, it is to be expected that the
water inertia effect will be estimated much closer for such ship forms than with these
experimental bars. It is possible, also, that surface tension has a slight modifying effect
in these small scale experiments.

APPENDIX I
The Inertia Coefficient of an Ellipsoid of Revolution, Vibrating in Two or Three
Nodes.
A system of orthoganal coordinates consisting of the confocal ellipsoids of revolution
= constant, the confocal hyperboloids of revolution p = constant, and the planes w =
constant, cutting the axis of the ellipsoid, is adopted. Then a soluti3n of Laplace's
equation V2 = O can be expressed in terms of the Associated Legendre Functions of
these coordinates. By choosing and combining the proper functions, the flow correspond-
ing to a transverse motion of the ellipsoid, or to a bending motion of two or more nodes,
can be obtained It will be seen by equation (22) that this "bending" motion is in reality
a shear, transverse section of the ellipsoid staying in the same plane. The distinction is
unimportant for our purposes, however.
It is impossible in a paper of this length to give any adequate explanation of these
mathematical methods, and for a preliminary study the reader should consult Chapter V,
Arts. 103 to 107, of Lamb's Treatise.* The notation here used is that adopted by Lamb
with the exception that r is used in place of and denotes
The result of this analysis is stated beginning at paragraph 25.
Properties of the ellipsoidal coordinates.
Referring to Fig. 2, Plate 1:
The distance between the foci is 2 k.
The semi-major axis of any of the ellipsoids of revolution is
(1)
The semi-minor axis B = k 2
i (2)
The volume of any ellipsoid k(2 - 1) (3)
If &, p, w are the ellipsoidal coordinates and x, , w polar coordinates, then the
relations between these coordinates are given by the equations
x= (4)
r = k(1 2) (2 1) (5)
w w
(6)
The element of arc of the ellipsoid is
k (.2
ds
(i - (7)

*Hydrodynamics, H. Lamb. Cambridge University Press.

 INERTIA OF WATER SURROUNDING A VIBRATING SHIP 7

and the length of an elementary normal

dn (2_1)
The cosine of the angle between the tangent and the z axis is given by
-
cosO
=(-
A solution of Laplace's equation is given by
475 = ß1P1'(p.)Q'() ± f33P3'(i2)Q3'()J cos w (10)
Where P' and Q' are the associated Legendre functions of /1 and and 8' and 8 are
arbitrary constants.
The above solution represents a motion of the ellipsoid in which the axis is deformed
into the arc of a parabola. The relative values of ß and 8s determine the position of the
axis with respect to the parabola and consequently the position of the nodes.
In like manner the solution
= [j32P2'(ji)Qz'() + ß4P4'(z)Q4'()Jcosw (11)
wiLl represent a 3-noded motion, in which the axis is deformed into a cubical parabola.
The kinetic energy of the motion is given by
*
2T = - dS (12)

or

2T = - pffç, - rdwds

and substituting the values of r and ¿s , previously given,this reduces to

2T = - iepjfc (2 - 1) dj1dw
o is the particular value of associated with the surface of the ellipsoid, over
which the integration extends.
Substituting the value of for the 2-noded solution:

2T = - kp - 1)
r+1 çl
.' _°/ [ß1P1'(,i)Q'() + ß3P3'(j4Q3'()]
(15)
X [ßPi'&i)Qj'() + ß3'C)Q3'()J} CoS2wd/Ldw
where represents differentiation with respect to
We have

fcos2wdw = (16)

f± P1'()P3'()d = O Lamb Art. 87(4) (17)

*Lamb Art. (46) 14.

 8 INERTIA OF WATER SURROUNDING A VIBRATING SHIP

=
Lamb Art. 87(6)
[P3'()Jd =
f-1 J

So that equation (15) reduces to

27' = kpir(o2 - 1){ ßj2Qj(-)1/() +
In like manner for 3-noded vibration the energy equation is
27' = kpir(r02 - 1){ ß21Q2'(r)Q2'(r) ±
22. The velocity of the surface at any point normal to itself will be
(V1 - V31z2) cos O cos w 2-noded
(V21 - V4M3) cos O cos w 3-noded
and this must equal the velocity of the fluid normal to the surface which will be
a-
a ar an
combining (22) (23) and (8) (9) there is obtained

= - (V1 - VaM) -(1 - ,

k cos w
2
/
= {ßlPl'(M)j'() + ß3P3'(M)l'(-)J cos w (24)
for = . In all the following equations it will be understood that s =
= (1 - p2)J(M) = (1 - 2)
1

(25)
= (1 M2)P3(M) (SM2 - 1)(1 M2)

Lamb Art (85)

Substituting the above in (24) and equating the coefficients of like functions of M there is
óbtained
2V3k
= 15 (2 - 1) Q'()
(26)
(V1 - V3/5)k- 4
ßl
- - 1) Q'(r) J

and substituting in the energy equation (20) there is obtained for 2-noded vibration
2T = - TPk3r2[ (Vi V3/5)2' +
Q'(r) 525

In like manner for 3-noded vibration

2T = - 15 (V- - 3 174/7)2 + 160

9.352
1742 Qiífl
Q'(r)J
 w-- eL*'r'n

INERTIA OF WATER SURROUNDING A VIBRATING SHIP

The Legendre Q functions are defined by the equations
Q'() = (2 - 1) )

= log
±1

f,
= +(3 1)log
(29)

Q() = 14
(5 -3
- 3) log .- i
= (35 30 + 3) log -+ 1 - ±4
Performing the necessary differentiations upon the equation (29), substituting i
equations (27) and (28) and simplifying there is finally obtained for 2-noded vibration
2T M [(V1 - 1/3/5)2 R1 + T1c V32 R3] (30)
and for 3-noded
M[2
2T =
_ V4)
R2± th V42R4] (31)

where M is the mass of an ellipsoid of density p and R1 R2 R3 R4 are functions of s given

by the equations.
32
4log1
t-_1
4log1
-_i
3-2.._6
-2_
42log+l 3--2
R. =

(5
1 -i
- ) log + 1
6-7
15 -4 13 2
(32)

R3 =
4 (15 - li ) log +1 45 4
- 63 -2 + 16

(35 4
- 15 2) log 105 15 + 16
-
Çt
R4
4 (140 - 135 -2 ± 15) log s' ± 1 420 - 685 + 259

We wish to obtain the ratio of the actual kinetic energy to that which would
exist if the motion of the fluid were confined to transverse planes. The results for a cir-
cular cylinde' show that the energy in the latter case would be equal to the energy of the
ellipsoid itself if of density p.
The area of any transverse section of the ellipsoid will be irr2 J2 (1 2) (.2 1) and
the velocity at any section for 2-noded vibration
y = (V1 - V32)
 VI

lo INERTIA OF WATER SLTRROUNDING A VIBRATING SHIP

then
2T' = pk (2 l)f' (1 - p2) (V1 - V3p2)2 dp

2T' = M[(V1 - V3/5)2 + 1V32] (33)

and likewise for 3-noded vibration

2T' M[2 -5 + Tb V42] (34)

25. Summary of results and numerical evaluation.

The coefficient J, defined as the ratio of the kinetic energy of the flow around the
vibrating ellipsoid, to that which would obtain if the flow were confined to transverse
planes, is given by the equations
(V3 - V3/5)2 R1 + T V3 R3
J2 node - (V1 - V3/5)2 + h V32
- (V2 - 4 V4)2 R2 + b V4 R4
J3 node
(V2 -
4 V4) + 7b V42
The R's are functions of s as given by equation (32). determines the proportions
of the ellipsoid, the ratio semi-major axis over semi-minor axis being given by L/B =
-
We have evaluated the R functions for certain values of and they are given in
the table below with corresponding values of L/B. Graphs of the R functions over L/B as
a base are also given in Plate 4.
TABLE i
L/B R1 R3 R3 R4

1.000 oo 1.000 1.0 1.0 1.0 cylinder

1.001 22.4 .9888 .9728 .9539 .9319
1.002 15.82 .9806 .9530 .9220 .8902
1.003 12.95 .9731 .9368 .8972 .8586
1.005 10.02 .9604 .9103 .8588 .8084
1.007 8.48 .9495 .8890 .8291 .7704
1.010 7.11 .9348 .8608 .7894 .7248
1.015 5.83 .9141 .8234 .7414 .6702
1.020 5.07 .8963 .7932 .7037 .6286
1.030 4.174 .8672 .7468 .6476 .5694
1.050 3.275 .8226 .6806 .5746 .4944
co 1.0 .5000 .3333 .2500 .2000 sphere

26. The shape of the axis is given by the equations t

V3 - y3 = y 2-noded
V2 p - V4 3 = y 3-noded
 INERTIA OF WATER SURRO'JNDING A VIBRATING SHIP 11

where i varies from O at the origin to + or - i at the ends of the major axis, so that by
taking V1/V3 and V2/V4 of the proper value the nodes of the motion can be placed at any
desired point.
If we take V3 = 5v1 for 2-noded vibration, the nodes will be at M = 1.2 = .447,
or they are at .447 the length of the semi-major axis from the center. This is very nearly
at the point found in a vibrating ship. Likewise, with V = 7/3V2 for 3-noded vibration
the nodes will be at M = O, .65, a position corresponding fairly close to the nodes of actual
ships. With these ratios the equations (35) (36) reduce to
J2 node = R3
(37)
J3 node = R4
For slight variations in the nodal position the variation in J would not be very great,
and considering the approximate nature of the step from the ellipsoid to the ship it is
considered that the equation (37) give the result with sufficient accuracy. It may be
noted also that these positions of the nodes are the ones corresponding to the minimum
kinetic energy of the flow.

APPENDIX II
THE INERTIA COEFFICIENT FOR CYLINDERS OF APPROXIMATE SHIP SECTION

27. We show how the shapes of Plates 2 and 3 and the energy of the flow around each
are obtained.
Let z, y be the coordinates of any point of a two dimensional flow and , the values
of the velocity potential and stream-line functions at the same point. Then the equation
z= (x+iy) =f(+i)
where f designates any arbitrary function represents a two dimensional stream-line flow. *
The flow in the x, y plane can be further transformed to any other plane, say X, Y,
by the relation
Z = X ± iY = f (x + iy)
since (X + 1Y) is still a function of ± iv'.
We take the flow past the circle
x2±y2=l or r=l
and transform it by means of the relation

Z=X±iY = z+ +- =z±az'±bz3 (1)

but
z = z ± iy = re1 =
so that
X + IY = e8 + ae° + be318 (2)
and since
=cos O + i sin O
Lamb, Chapter IV.
12 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
we have, by equating the real and imaginary parts
X= (1+a)cosû+bcos3O
}(3)
Y = (1 - a) sin O - b sin 3 0
These are the parametric equations of the transformed circle.
The semi-axis parallel to the flow will be of length
X1 = (1 + a + b)
and perpendicular to the flow
Yj=(1a+b)
so that the ratio Half Beam over Draft will be
1+a+b
H BID
/ 1a+b (4)

For the flow past the circular cylinder and ' are given by
= U(r+--)cos0 = U(r_--) sinO (5)

and for the boundary r = i

=2Ucos0
This represents the flow past a stationary cylinder. We wish to obtain the energy
of the flow of a moving cylinder in a fluid stationary at infinity; we must accordingly add
to the values of ç and in f) the terms
'4=UY (6)
giving
= U[(1 - a) cosO - bcos3û]
= - U[(1 - a) sinO hsin3ü] } (7)
for the moving cylinder.
The kinetic energy per unit length of cylinder i given by
=27
2T = p / = irU2p [(1 - a)2 + 3 b2]
Jo (8)

on substituting the above values of and and integrating.

If the semi-axis of the transformed circle is to be B, instead of (1 - a ± b), the above
energy must be multiplied by the ratio
B2/(i - a ± b)2
giving
r(1 - a)2 ± 3 b1
2T=U2PB2L(l+b) J
(9)

The kinetic energy of a circular cylinder of :iadius B is given by

2T = 7rU2pB2

*Lamb Art. (68).

 INERTIA OF WATER SURROUNDING A VIBRATING SHIP 13

Therefore the inertia coefficient of the transformed circle is

c- (1a)2+3b2
(1ab)2
30. Solving equations (4) and (10) simultaneously for a and b leads to the equations
b -- H[ (2CHH+ fl + (H+ 1) i4C 3]
2(CH2_H2_H_1) r

a= (1 -(1+
H)(1 + b'
H) (12)

and the parametric equations of the transformed circles when of draft D will be
E (1+a)cosG+bcos3O
X=DL (1+a+b)
(1 - a) sin O - b sin 3 0
Y=D L (1 + a + b)
X being the dimension in the direction of motion and the line O = O coinciding with the
X axis.
Each shape of Plates 2 and 3 corresponds to definite values of C and H marked
upon it. For each of these shapes "b" was calculated by (11) and 'a" by (12), then the
parametric equations (13) were evaluated and a simple graphical construction used to
draw the shapes. The inner curve in each case represents the limiting cusped form. The
outermost curve has been taken in each case so as to obtain the best approximation to a
rectangle with rounded corners. The stream-lines could be constructed for each of these
shapes, but this is not essential to the present investigation.

APPENDIX III
THE INERTIA COEFFICIENT OF A RECTANGLE

In estimating the inertia coefficient for sections with sharp bilges a knowledge of
its value for the rectangle is useful. While we would consider it probable that the equa-
tions for the flow past a rectangle have been given before, we have been unable to locate
them.
A solution may be effected by thb transformation of Schwarz.
Let
z=x+iy and W=ç+iL'
Then a rectilinear flow may be transformed into a flow with a polygonal boundary by
the transformation
= (W - &) (W - 2) 1
(W - j.

where i, etc., are the particular values of c at the corners of the polygon and a1, a2,
,

etc., are the internal angles at corresponding corners.*

* For further explanation see Electricity and Magnetism, J. H. Jeans. Cambridge University Press, Art. (322),
or Lamb Art. (73).
 14 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
33. Applying this to the rectangle, let k be the values of at the corners of the
rectangle and i be the values where the central stream-line meets the rectangle, as
shown in Fig. 3, Plate 1. Then
a2a-7r
and the transformation becomes
d
j-= (W+ 1)-(W- 1)3 (W+ k)(W- k)3
or

dW (1)
J (W'-1)
We are interested in evaluating this around the boundaries of the rectangle only.
For this ' = O and we have
rc - k2)
Z=X+iYJ(l1)d (2)

Along the D side of the rectangle varies from O to k, and we have

x_-f(k2
12d (3)

Along the B side, after multiplying (2) by i

-k
(4)

For large values of ', x = so that the velocity of the undisturbed stream is -1.
In the equation for the D side (3), let
p-
= k sin C, 8, = sin-' /k
Then
x= aÇ8, k cos2 8,
_de,
J i - k2 sin2 Ci (5)

This is an elliptic integral of known form, having the solution*

= [E (ei, k) - (k')1 F (or, k)] (6)
where F and E are the elliptic integrals of the first and second kind, respectively, and k
is the complementary modulus defined by
(k')1 ± k2 = i
If the half width parallel to the motion (D side) be D,, then
D, = [E (-i-e k) - (k1)2 F k)] (7)
(-i-,
More conveniently, if
k=sina k'=cosa
= [E (-i-, sin a) - cos2 a p' (i-, sin a)] (8)

The Integral Calculus, Joseph Edwards, Vol. 1, Art. (391).

 INERTIA OF WATER SURRGUNDING A VIBRATING SHIP 15

Taking the equation for the B side (4) and making the substitution
sin02= J1_2
Ii_k2 Il_k2
we have
C0S2 0
y = (1 - k2) de2
- J o V 1 - (fr')2 sin2 02
the solution of which is
y = E (UI, k') fr2 F (02, k')
and if the half width perpendicular to the direction of motion is B1
B1 = E cos a) - sintaF , cos a)
The kinetic energy is given by
2T = - «V dS (12)

where «V refers to the flow around a moving rectangle in a fluid stationary at infinity
(c refers to the flow past a stationary rectangle).
Along the B side of the rectangle
dS=dy
and (12) becomes
2T=4p1 (çDj)dy (13)

Along the D side (12) is zero.

From (4)
dy (j2)
(«pl - k2)
dçb

so that
«p («p2 -
2T - 4p :
(1 «p2)3 d_ BIDI]
integrating
2T = 4p cos2a - B1D1]

Dividing by irp B12 the section inertia coefficient C is found

cos2a 4D1
C PECT. =
B12 ir B1

The procedure is then to calculate B1 and D1 for various assumed value of a by

(8) and (11). Then C is given by (16) and BI/D1 = H gives the proportion of the rectangle.
The values of the complete elliptic integrals E and F or K can be taken from any table of
the elliptic functions. *
*Pierces Table of Integrals, or Smithsonian Tables of Elliptic Functions.
 16 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
Below are tabulated the results for certain values of , and these results are also
plotted on Plate 4.
TABLE 2

O 15 20 25 30 35 40 43
H 16.8 8.85 5.23 3.31 2.18 1.47 1.165
C 1.0 1.095 1.152 1.211 1.276 1.348 1.427 1.476

Flat
Plate

a 45 47 30 55 60 65 70 75

H 1 .858 .681 .439 .302 .191 .113 .0595

C 1.5131 1.551 1.611 1.720 1.846 1.995 2.177 2.412

Square

APPENDIX IV
THE INERTIA COEFFICIENT FOR A RHOMBUS MOVING AXIALLY

Referring to Fig. 4, Plate 1, it will be seen that a = d3 yir az ir(3 2y)

and therefore the appropriate Schwarzian transformation is
(I,,
=(W2_1)y-1 W
2(1-')')

or
Z = re
= f J4,T 2(i-y)
(W - 1)'
along the sides = O. Let r denote the distance along these sides from the point marked
dW (1)

O. Then by the elimination of O from (1) there is obtained

f(1 21-y

the length of a side being

/'i 2(i-y)
r1= / o (1
dç

The definite integral can be expressed in terms of the Gamma Function giving

Ti =
-2r( ) r ()
The kinetic energy will be given by
2T = - pfçb' d
where 4/ and st" refer to the flow past a moving cylinder in a stationary fluid
= -X
= ,ì'y -
 INERTIA OF WATER SURROUNDING A VIBRATING SHIP 17

and
2T=-pf(çS-x)dy
2T = 4sin-yir ; r' 3-2
2)1_7
areaof
- rhombus]
0(1

which will reduce to

2T = p[2ir(l - y) - arca]
The half width perpendicular to the direction of motion is
B, = r sin yir
and parallel to the motion
D, = - r, cos yir
The area
2B,D, = - 2r,' cos yir sin
Dividing (7) by ir p B,' and simplifying, there is obtained the inertia coefficient C:
2ir' (1 - + 2 cot ir
r2 q
- ) r2 () sin 2
yir
ç9)

the proportions of the rhombus being given by

= H = - tan 'yir (10)

The table below gives the evaluation of these equations for certain values of y.
Values of log r (X) can be taken from Pierce's Table of Integrals or from Brownlee's
Table of Log r (X).* A gràph of C is also shown on Plate 4. Note that for the square
moving comer on the coefficient is half that of the square moving flat on, a result which
can also be arrived at by considerations of superposed flows.

TABLE 3

-Y .5 .65 .7 .75 .9 .95

II co 1.962 1.375 1.0 .325 .191
C 1.0 .8430 .7986 .7566 .6389 .600

Flat Square
Plate

LIST OF ILLUSTRATIONS

Plate 1.Miscellaneous sketches.

Plates 2 and 3.Cylinders of approximate ship sections.
Plate 4.Inertia coefficients.
Plate 5.Data for example given in Article 15.
*TraCtS for Computers. Cambridge Univ. Press.
 18 INERTIA OP WATER SURROUNDING A VIBRATING SHIP

DISCUSSION
Ta PRESIDENT: We are ready for the discussion on this paper.

MR. JOHN L. BOGERT, Member: No one seems to want to discuss this mathematical
paper, but there is one aspect of it which is very interesting to all of the members. Professor
Lewis says in one part of his paper that, regardless of any theoretical treatment of this
subject, the one thing which appeals to every practical man is, that vibration of a hull is
something that won't occur, provided the cause of vibration in the interior of the hull is
largely eliminated. He therefore stressed in his paper the importance of doing away with
all those factors which go to produce vibration.
There isn't any question at all but what that is a highly desirable thing, and of course,
naturally, to our minds occurs the fact that the turbine-driven ship and the electrically
driven ship fall largely in that category, whereas the reciprocating driven ship is much more
subject to those actions.
There is one other way in which this thing can be reached, and that is by the avoidance
of synchronism of impulse; and, just as we recognize the fact that we are obliged to have a
rolling ship as long as the ocean rolls, so, of course, the solution naturally has been that of
destroying the synchronism of the rolling impulse by means of the Frahm anti-rolling
tank and things of that kind.
I think that point hardly has been brought out here to those who haven't read the
paper, and I think it is one of the strong points that Professor Lewis makes in his paper.

MR. THOMAS C. RATHBONE, Visitor: The accuracy of prediction of the natural fre-
quencies of complex systems, such as are represented by hulls and complicated machine
structures, depends on the, accuracy of information as to the mass moments and elastic
scales involved, and on the closeness with which the assumed schematic system approaches
the actual conditions encountered. As the author states, the more reliable predictions have
been arrived at through interpolations or extrapolations based on existing experimental
data. It may be pointed out that these experimental data include the water inertia effect.
The author has accomplished a valuable work in affording a quantitative evaluation of
the water inertia factor. It will be extremely interesting to apply this information to the
numerous cases wherein the calculated frequency did not agree with the actual frequency
found.
There has been little information available as to the damping effect of the water.
Experience on large steel land structures which are free to vibrate, and possess no damping
other than internal friction of the metal itself, indicates that enormous inhibitive forces
must be supplied by the water surrounding a ship, preventing the hull from responding
violently and in a great variety of modes to small periodic forces. In so far as amplitude
of vibration is concerned, the damping effect of the water skin is probably much more
potent than the inertia effect on the natural frequency of the hqli and its relation to the
disturbing frequency.

MR. J. Locwoon TAYLOR, D. Sc., Visitor: I have read with considerable interest
Professor Lewis's mathematical results, having recently been engaged in some similar
calculations. As Professor Lewis suspects, the case of the rectangular section has been
 INERTIA OF WATER SURROUNDING A VIBRATING SHIP 19

worked out (Riabouchinski, International Congress of Mathematicians, Strasbourg, 1920),

and it so happens that he anticipates the solution for a rhombus in a paper of my own which
is at prcscnt in the press. I also worked out the case of the ellipsoid with two nodes, but
arrived at somewhat different results from the author, who remarks that his solution really
represents a shearing motion rather than actual flexure. The distinction is, however,
of some importance, and if the actual boundary condition for flexure, allowing for the lon-
gitudinal movement of the surface, which is of the form
= (a - X ) - + 2.xy ax
2 2

is fulfilled, instead of neglecting, as Professor Lewis virtually does, the second term, the
resulting coefficients are appreciably modified. I find for J2 the value
2(2G2)
i
1R 3xC1429G4)
in piace of R2, the reduction being about 8 per cent for G = 1.01, while the corresponding
reduction for the three-node coefficient is even greater. To take an extreme case, the
coefficient replacing R2, which corresponds to the case of "one-node" vibration (rotation),
will be zero for the sphere instead of one-third. it is capable of mathematical proof that
the inertia coefficient is independent of motion ahead, for the ellipsoid, and there is no
reason to doubt that this is a perfectly general result.
With regard to the author's general remarks, it is not easy to endorse the statement
that hull rigidity is a comparatively unknown factor, since the recognition of the impor-
tance of shear deflectionand in exceptional cases of buckling under compressionhas,
by bringing into line the calculated and observed deflections in practically all the cases
for which experimental data are available, removed the uncertainty in this direction. The
principal element of doubt in the frequency calculation has been the inertia of the water,
and, if the results of such mathematical investigations as Professor Lewis has presented
can be correlated with actual ship data, it should be possible for this doubt to be removed;
and the fact that I was, I believe, the first to suggest the application of "perfect fluid"
hydrodynamics to this problem* gives me especial pleasure in congratulating him on his
achievements in this direction.
PRoFEssoR Lrwis: I am in agreement with Mr. Rathbone that the damping effect of
the water in suppressing the numerous possible modes of vibration of a hull is undoubtedly
very great. This damping must be due either to the surface waves, produced by the
vibratory motion of the vessel, or by viscosity and eddies. Beyond this little can be stated
at the present time.
In preparing a paper containing original work of this character, the question of priority
is always a difficult one, and I wish to thank Mr. Taylor for calling attention to the previous
paper of Riabouchinski on the flow past a rectangle. In commencing this work I recog-
nized that the surface condition stated by Mr. Taylor would represent a bending, while
that I have used represents a shear. It seemed unlikely, however, that the difference
would be very great for the proportion of an ellipsoid which would apply to a ship and
would certainly be smaller than the error involved in passing from the ellipsoid to the ship.
Further, there is no way of determining, a priori, whether the bending or shear conditions
would give the better results for the ship. The conditions for shear, which are very con-
*Trans. N. E. C. Inst., 1927S.
20 INERTIA OF WATER SURROUNDING A VIBRATING SHIP
siderably simpler, were therefore used. I am surprised that Mr. Taylor should find the
difference to be as great as 8 per cent. I am unable to follow Mr. Taylor in his equation,
for I note that this would make J2 negative for values of G2 greater than 2, a result which is
physically impossible. I trust that these difficulties will be cleared up with the publication
of Mr. Taylor's own results.
THE PRESIDENT: Before passing on to the next paper I wish to assure Professor Lewis
of our appreciation of the work he has done in preparing this paper, and of our appreciation
also of the splendid work that he has done along this line in recent years. His researches
and practical work in regard to vibration and the causes of vibration and elimination of
those causes have been among the greatest things contributed in recent years to the develop-
ment of the science of our profession.
The next paper on our list is No. 2, on "Great Lakes Bulk-Freighters," by Mr. William
Bennett, Member.

Mr. Bennett presented the paper.

4
Transactions Society Naval Architects and Marine Engineers, Vol. 37, 1929. Plate I
To illusi rate paper on "The Inertia of i/ic Water Surrounding a Vibrating Ship,"
by Frank M. Lewis, Member.

ôo
L t' 2lcroN OF FLO\'v'

F%G.3
c'ci

q Ç3 -zr)
o
PSECTtON
4
OF FLOI

¿
 -. -

Transactions Society Naval Architects and ivíarine Engineers, Vol. 37, 1929.

To illustrate paper on "The Inertia of the Water Surrounding a Vibratin

by Frank M. Lewis, Member.

K .3

)\-- LZO
rine Engineers, Vol. 37, 1929. Plate 2

aper ou "The Inertia of lite Water Surrounding a Vibrating Ship,"

by Frank M. Lewis, Member.
7'ransactions Society Naval Architects and 1'Iarine Engineers. Vol. 37, 1929.
To illustrate paper on "The Inertia of the Water Surrounding a Vibrating Ship,"
by Frank M. Lew, Member.
:wal Architects and Marine Engineers. Vol. 37, 1929. Plate 3
liust rate paper on "The Inertia of I/ic Waler Surrounding a Vibrating Ship,"
by Frank M. Lewis, Member.
Transactions Society Naval A rciiitects nd Marine Engineers, Vol. 37, 1929. Plate 4
To illustrate paper on "The Inert-ia of the Water Surrounding a Vibrating Ship,"
by Frank M. Lewis, Member.

r
.5
F NO9 P45ÇVO5
C.

.4

L
NE.TR COFFcfNjTs.
5CLa CCURVES1 OVER
. ., . .O ZO 2JR Z'r Z' 2 ' 3.O

:3 -
LL fk
f9
cP\(ESUNEk1
IO ri t2. '3
Bt9r
;,
13 vr
Transactions Society Naval ArcJ:itects and Marine Enginèers, Vol. 37, 1929. Plaie 5
'To illustrate paper on "The Inertia of tue Water Surrounding a Vibrating Ship;'
bi Frank M. Lewis, Member.

>
D
u ILL

LLItLL
Li
H >>

zz
oC
J-2
LlJui