TECHNISCHE UNIVERSITHT
taboratorium voor
Selteepehydromectaike
ArchletMekelweg 2, 282B co Delft
Tel.: 015 - 788873 -Fox 015 - 781838No. 175
August 1975
ON THE INTERACTION BETWEEN
A STRUT AND THE FREE SURFACE
T. Francis Ogilvie
This research was carried out under the Naval Sea Systems Command General Hydromechanics Research Program, Subproject
SR 023 01 01, administered by the David W. Taylor Naval
Ship R & D Center Contract No. N00014-67-A-0181-0053
Reproduction in whole or in part permitted for any purpose
of the United States Government.
Approved for public release; distribution unlimited,
THE
DEPARTMENT
OF
NAVA
410ot-clot-AND
MARINE E:G'INEERING
4061-THE UNIVERSITY OF MICHIGAN
COLLEGE OF ENGINEERING
-Bibliotheek van de
Afderag Sc4cep,s',:euw-cn Schecpynartunde
Technisce Ho,-,esch7.0!, Dept
DOCUME;STATiE :
9_ 4-15
DATUM'
ON THE INTERACTION BETWEEN A
STRUT
AND THE FREE SURFACE
T. Francis Ogilvie
This research was carried out under the Naval Sea Systems Command
General Hydromechanics Research Program, Subproject SR 023 01 01, administered by the
David W. Taylor Naval Ship R&D Center Contract No. N00014-67-A-0181-0053 Reproduction in whole or in part permitted for any purpose of the United States Government. Approved for public release; distribution unlimited.
1,,,V1Y0,,,. Department of Naval Architecture
44,
11V li
a. and Marine Engineering
.... ci
s
College of Engineering.o
.0 op The University of Michigan
7111 Ann Arbor, Michigan
48104 No. 175 August 1975
SECURITY CLASSIFICATION OF THIS PAGE (When Data Entered)
DD
1 JAN 73 1473 EDITIO CI NC,V ,1; IS OBSOLETEUNCLASSIFIED
REPORT DOCUMENTATION PAGE
BEFORE COMPLETING FORMREAD INSTRUCTIONS I. DEPORT NUMBER175 .
2. GOVT ACCESSION NO. 3 RECIPIENT'S CATALOG NUMBER
4. TITLE (and Subtitle)
ON THE INTERACTION BETWEEN A STRUT
AND THE
FREE SURFACE
5 TYPE OF REPORT & PERIOD COVERED
Final Report
1 Oct 72 - 31 July 75
6 PERFORMING ORG. REPORT NUMBER
7. AU THOR(s)
T. Francis Ogilvie
EL CONTRACT OR GRANT NUMBER(s)
N00014-67-A-0181-0053
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Naval Architecture and Marine
Engineering
The University of Michigan, Ann Arbor, MI
10 PROGRAM ELEMENT, PROJECT, TASK
AREA & WORK UNIT NUMBERS
61153N R023 01 SR 023 01 01
11. CONTROLLING OFFICE NAME AND ADDRESS
David W. Taylor Naval Ship Research and Development
Center Code 1505 Bethesda, MD. 20084 12. REPORT DATE
August 1975
13. NUMBER OF PAGES28 +
iii
14. MONITORING AGENCY NAME a ADDRESS(// different from Controlling Offic,)
Office of Naval Research
Arlington, VA 22217
15. SECURITY CLASS. (of this report)
Unclassified
15e. DECLASSIFICATION/DOWNGRADING
SCHEDULE
16 DISTRIBUTION STATEMENT (of this Report)
Approved for public release; distribution unlimited
17 DISTRIBUTION STATEMENT (of the abstract entered in Block 20, II different from Report)
18 SUPPLEMENTARY NOTES
Sponsored by the Naval Sea Systems Command, General Hydromechanics
Research Program administered by the David W. Taylor Naval Ship Research
and Development Center, Code 1505, Bethesda, MD. 20084
19 KEY WORDS (Continue on reverse si,je if neces,ary and identify by block number
STRUT
SURFACE-PIERCING STRUT
SLENDER-BODY THEORY
BOW WAVE
GHR Program
20. ABSTRACT (Contint, on reverse side If necessary nnd identify by block number)
A procedure is proposed for solving problems involving the
flow around surface-piercing struts.
There are three steps:
(1) An infinite-fluid problem must be solved, in which the flow
is antisymmetrical with respect to the plane of the undisturbed
free surface.
(2) From that solution, there is effectively a
distribution of normal velocity imposed on the plane of
anti-symmetry; the resulting wavelike disturbance can be found by a
UNCLASSIFIED
ORM
.
. '
UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGE(Wheri Dot. Entored)
UNCLASSIFIED
SECURITY CLASSIFICATION OP' THIS PAGE(When Pate Entered)
procedure which is effectively a slender-body analysis.
(3)The wavelike disturbance produces a disturbance
on the body
(strut), which must be canceled by yet another wavelike motion,
which can also be found locally from a slender-body analysis.
In order to justify the second and third steps, some
difficul-ties and advantages associated with slender-ship theory
are
examined.
It is shown how the usual axial source distribution
of slender-ship theory can be modified systematically
so that
the slender-ship wave resistance becomes identical to the
classical result of Michell for a thin ship.
The slender-ship
procedure has the further advantage that it is not limited to
thin bodies; in particular, the body boundary condition should
logically be satisfied on the exact location of the body surface,
ABSTRACT
A procedure is proposed for solving problems involving the flow around
surface-piercing struts. There are three steps: (1) An infinite-fluid problem
must be solved, in which the flow is antisymmetrical with respect to the plane
of the undisturbed free surface. (2) From that solution, there is effectively
a distribution of normal velocity imposed on the plane of antisymmetry; the
resulting wavelike disturbance can be found by a procedure which is effectively
a slender-body analysis. (3) The wavelike disturbance produces a disturbance
on the body (strut), which must be canceled by yet another wavelike motion,
which can also be found locally from a slender-body analysis. In order to
justify the second and third steps, some difficulties and advantages associated
with slender-ship theory are examined. It is shown how the usual axial source
distribution of slender-ship theory can be modified systematically so that the slender-ship wave resistance becomes identical to the classical result of Michell
for a thin ship. The slender-ship procedure has the further advantage that it
is not limited to thin bodies; in particular, the body boundary condition should logically be satisfied on the exact location of the body surface, not on the centerplane.
CONTENTS
INTRODUCTION
FAILURE OF SLENDER-BODY THEORY AT THE BOW
ADVANTAGES OF USING SLENDER-SHIP THEORY 13 ,
METHOD FOR SOLVING STRUT _PROBLEMS 18
REMAINING DIFFICULTIES 22
REFERENCES 25
APPENDIX /2 EQUIVALENCE BETWEEN TRANSLATING 31) SOURCE, AND
IMPULSIVE 2-D SOURCE 26
1
INTRODUCTION
This study was started because of the observation that the wave created at the bow of a fine wedge can be predicted fairly well (Ogilvie (1972)) on
the basis of a slender-body theory. In that analysis, it was assumed that an
incident stream experiences absolutely no disturbance upstream of the fine entrance of a slender body; this is not true, of course, but the assumption
nevertheless leads to predictions that are fairly accurate. The analysis was
not based on the conventional assumptions of slender-body theory, but it was
nevertheless a slender-body theory and as such it may be expected to be less
valid if applied to a strut-like body, although the example worked out indicated that this difficulty might not be so serious as intuition suggests.
The purpose of this study is to apply a similar concept to problems of surface-piercing struts, to determine the limitations of such an approach, and to develop modified procedures as far as possible to overcome those limitations. A major side-result has appeared too, and it may be much more significant than
the original goal: It is found that the problem can be broken into two parts,
one of which is a conventional infinite-fluid problem, the other a problem in wave propagation which can be treated in much the same manner as problems in which the governing equation is a typical hyperbolic wave equation.
In the light of classical ship hydrodynamics, it might be wondered why anyone would even suggest applying slender-body theory to a strut problem;
thin-ship theory might be a more reasonable approach. Aside from the
demon-strated partial success of the earlier analysis, the following reasons may be cited for at least considering the slender-body approach:
1) Thin-ship theory is based on the assumption that the field variables
change slowly in the lateral direction near a ship. This is actually true over
most of the surface of a ship, but it is wrong near a ship bow. A strut may be
visualized for some purposes as a fine ship bow (without the rest of the ship!),
and so the thin-body approach may be expected to fail in a strut problem. As
will be seen, slender-ship theory does not necessarily suffer from a correspond-ing handicap.
1 2 1
-Directly behind a surface-piercing strut moving at an angle of attack,
there is a discontinuity in free-surface elevation; behind a symmetrical strut,
there is a discontinuity of slope of the free surface. Such discontinuities
can be analyzed on the basis of a slender-body approach, but no valid trea-ment
based on a thin-ship model has been found possible. This problem is potentially
important in determining the effect of a forward strut on after struts. Within the framework of thin-ship theory, the solution of problems
involving lateral assymetry is terribly difficult. The computational problems
can readily be seen in the work of Daoud (1973), who solved the case of a flat
plate of zero thickness moving ahead with a steady yaw angle; the
two-dimen-sional singular integral equation of linear lifting-surface theory becomes
almost intractable because of the inclusion of free-surface effects. Such problems can be much less complicated when treated within the framework of
slender-body theory.
Of course, there are some serious objections to applying slender-body
theory to strut problems. Ordinary slender-ship theory, as developed for
example by Tuck (1963), fails catastrophically when applied to ships with
wedgelike bows, no matter how fine they are. As one manifestation of this
failure, the theory predicts infinite wave resistance for such a ship form;
this ridiculous result is the result of errors in the theory
near the bow,
and so one might expect a similar failure in the treatment of strut problems.
In addition, of course, one must face up squarely to the objection that
a strut is not a slender body, and any attempt to analyze strut problems by
slender-body theory requires considerable justification.
In this report, the failure of the slender-ship theory is examined for
the case of a ship with a fine wedgelike bow. It is shown how slender-body
theory can be modified so that the far-field expansion (to one term) is identi-cal to that of thin-ship theory (which is essentially a far-field theory). This equivalence shows that the bow anomaly in slender-ship theory can be avoided.
In developing the above equivalence, we find that the finite draft of the
ship, particularly near the bow, causes the generation of waves which seem to
arise from points upstream of the bow. The significance of this result is not
3
-entirely understood. It may possibly explain the observation reported from
time to time that ship waves seem to occupy a Kelvin angle with apex forward
of the bow. However, some reservations may be in order with respect to this
interpretation.
In any case, it seems evident that a strut-like body must cause more upstream disturbance than a slender body of shallow draft, and such upstream effects will cause havoc in a slender-body analysis predicated on the
non-existence of upstream disturbances. A method for treating such effects is
proposed. This method does not appear to offer the possibility yet of being
applied in a routine manner, but, if some special cases can be computed, they may indicate the extent of validity of simpler procedures.
It may be noted that the proposed method shows explicitly that the origin of ship-generated waves ought to appear partly to be ahead of the ship.
The general plan of attack here is to resolve first those special diffi-culties that might prevent the application of slender-body theory to strut problems and then to develop that application so as to take advantage of the special advantages of slender-body theory.
FAILURE OF SLENDER -BODY THEORY AT THE BOW
It has long been recognized that slender-body theory gives poor predictions of wave resistance, and this shortcoming has been blamed on the fact that, in
the far-field representation, the slender-body source distribution lies entirely
on the level of the free surface. Thus, the strong decay with depth in
wave-making effectiveness of submerged sources does not appear in the slender-ship representation.
This interpretation is misleading, but it is nevertheless instructive to
0
is the ship centerplane,
0
b(x,z) is the hull offset at (x,z) = g/U2 .
-4-recall how it comes about. One form of Michell's integral for wave resistance
is the following: R 40a 2
"
=I
7/2
36 [p2(6) 4. Q2(6)] de sec (1) TrU2 where 0 P(0) +i
oe)
=ff
bx(x z) eKzsec2eeiKxsece dx dz (2)In the Michell-Havelock theory, bx(x,z) is proportional to the local source
density, a(x,z) , in fact, a(x,z) = 2Ubx(x,z) . If all of the sources are
concentrated at z = 0 , and if we define
0
E(x) = 2U dz bx(x,z) = U A' (x) , (4)
-H(x)
where H(x) is the keel depth at station x and A(x) is the area of the
submerged cross-section, then
1iKx sec
eE(x)
P + iQ = d
2U x e = 2U
E* (K sec e )
(5)where E* is the Fourier transform of E(x) and L is the domain of the ship
length. Michell's integral then becomes:
K3)
1
5
which is a standard form for the wave resistance of a slender ship, as given,
for example, by Tuck (1963); Yo is a Bessel function in the notation of
Abramowitz and Stegun (1964). The source density, E(x) , is defined to be
identically equal to zero outside of the ship.
This correspondance between thin-ship theory and slender-ship theory has been noted by numerous writers, most of whom have then assumed that the poor wave-resistance predictions of slender-body theory are caused by setting
z = 0 in the integrand of the expression for P+iQ, that is, in (2).
However, for a wedgelike body, the failure of the slender-body formula
When this is substituted into (7), the answer is indeterminate (infinite). The
same is true if we compute E*(k) for substitution into (6): In the sense of
generalized functions, E*(k) contains one term, -2UHai/k , which leads to a
divergent integral in (6).
Michell's integral gives a reasonable prediction (even if not very accurate)
is catastrophic. Let the
being at x = 0 . If the
ship be located between x = 0
half-angle of the entrance is
o x < 0 and x = L , a , then the bow A (x) = (8a)
in the neighborhood of the
2Hotx + 0 (x2) , 0 <
bow. It follows that
x , E (x) = UA' (x) =
(0
x < , (8b) E'(x) = 2UHa 2UHa + 0 (x) d(x) + ,0< x
(8c ) pK R = 2 7r/2 2de sec 3e IE*(K sec
HI
0
k2
dk1E*(k)12
(6) 7 P f -Ik2 -K2 KThe last form can be changed into the following:
R =
-LId
dx E'( ) E' (x) Y0(Klx-) (7)2
0
6
-for the wave resistance of a body with a wedge bow, and so the error in
slender-ship theory involves much more than just the inaccuracy of overlooking
the depth attenuation of waves generated by submerged sources. The
disconti-nuity at the bow end of the line of sources in the slender-body far-field repre-sentation is completely unacceptable for certain purposes -one of which is the
computation of wave resistance. Such a discontinuity is intolerable because
it leaves behind in its track an entire line of singularities. The waves so
generated carry out energy at an infinite rate. If the line of sources were
submerged the slightest amount, this would not happen. Thus, allowing for
depth of source submergence (at least near the body ends) changes the entire character of the free-surface flow.
The wavemaking properties of a submerged source distribution can be represented by a longitudinal distribution of sources along the free surface
above the location of the actual sources. When this has been demonstrated, it
will be apparent how to fix up slender-ship theory for the region near the bow.
One form of the Kelvin-Havelock unit-source potential is the following:
1(1
1 ]cPs(x,Y,z)
74,7i r'v(
412 + k2eim(z+c) 472 1I
dk eikx dm m e-l-m 1 I iK2 -k2 dk k cos kx e K e-kly1/1 - k2/k2 71. 0 -1 1 dk ki/k2K2
e- LK2i z+ COS kx sin (Id y I Vk2/K 2 -1)
k21,7_,
1K/k2 K2
r 1
1 dk k
e- K 1 - 'I sin k_x cos (kyVk2/K2 - 1) .
7
-This is very similar to the form preferred by Michell himself; it can be derived by using separation of variables on the Laplace equation, then superposing all solutions obtained in that way so that the proper singularity is represented and
the free-surface condition is satisfied. In (9), the source is located at
(9) im2 k2 m + ik2/K =
co
-K .7
-(0,0,C) , with c < 0 , the stream of speed U flowing in the positive x
direction; also, r = [x2 +y2 + ) 2] 1/2 r = [x2 +y2 + (z+) 2] 1/2
The last term in (9) is odd with respect to x , and the other terms are
all even. The odd and even terms separately represent wavelike motion both
upstream and downstream of the source. The upstream waves must cancel, and so
there is a doubling downstream. Accordingly, we can represent all wavelike
motion downstream of the source by taking twice the odd term in (9), i.e.,
2 dk k k2,
e K sin kx cos (Jsyik2/K2- 1) for x > 0 .
(1)S(x'Y'z) K2
(10)
We can go yet one step further to obtain a more useful expression: Rewrite
the integrand using the relationship: 2 sin kx cos (kp/k2/K2 -1)
k 42/K2 -1)
sin (kx + sin (kx +kly1 )k2/K2 -1) .
Downstream of the source, only the first of these terms gives a significant wave motion, since the second has no point of stationary phase. Therefore we can write approximately:
k21
1 rdk k
- --iz+Cq)s(x,y,z)
=
-2K'
e < sin (kx - /k2/K2 - 1) . 7i/k-It is no longer necessary to add the restriction that x > 0 , since the above
integrand has no points of stationary phase for x < 0 .
The potential for the wave motion generated by a thin ship can now be expressed approximately as follows:
q)Ts(x,Y,z) = - -7 dE
OC a(E,)
1 IL f°
fdk k
e-tlz
+c!
s
.in[k(x-E) - kI y116c2/K2-1]
0 -H (E) K (12)
where a(x,z) = 2Ubx(x,z) on the centerplane, as before. Extend this definition
by requiring that a(x,z) E 0 outside the centerplane. Now define:
0
/K p*(k) = dx e-ikx dz a(x,z) ek2z
-H(x) We then have:
(13)
-(PTS(X,Y,Z)
1 K r-dk kJ] ]
A2
-K2 ek2z/Keik(x-Iyil/k2-K2/K)p*(k) . (14) co KOn the other hand, the wave motion generated by a line of sources of density E(x) along the axis is given approximately by the potential:
K
1
LI
dk k ek2z/Keik(x - lylik2-K2/K)z.(k)4)55(x,Y'z) - 27i ik2 -K2
These are the same if we identify E* and p* , that is, if
E*(k)=
j dx e-ikxE(x)0
= dx e-ikx dz a(x,z) ek2z/K (16)
-H(x)
The significance is shown in some special cases. It should be noted, in
particular, that (14) and (15) represent the same wave motion if (16) is satis-fied. A consequence is that the two lead to identical values of wave resistance.
Point Source. Suppose that a(x,z) = a06(x)5(z-z0) . Then:
E*(k) = a0ek2z0/K
which implies that
.,/K
a
E(x) . -21
f
dk eikx ek2z u27r -. / K KX = Cj01 47IZ0 e 0 ( Zo < 0 ) .
The total strength of E , that is,
f
E(x)dx , is the same as the strength,a , of the original source, but it is spread out over the longitudinal axis.
0
In principle, it is spread out over an infinite length, but actually it is
fairly well concentrated, especially if the source is shallow.
In no sense does the line distribution produce local effects comparable
to those of the original point source. In fact, the potential corresponding
to the line distribution is analytic in the lower half-space, even at the
loca-tion of the "equivalent" point source. What is important for our purposes is
(15)
(17)
9
-that a free-surface line distribution of sources can produce the same waves as the submerged point source, but the line distribution extends fore and aft of
the point source. Furthermore, the distribution is analytic in x .
Vertical Line of Sources. Suppose that
a06(x) , -H < z < 0 0 , z < -H . Then: Elc(k) = L.S)"_ (1 e-k2H/K k2 and: 1 I
Kx2/411
E (x) = 00{- 'XI - erf /Kx2/4H} +
ViTrr
KI- e (18')A longitudinal distribution of sources given by (18') produces the same wave motion as the vertical line of sources of constant density a per unit height.
0
It may be noted that E(x) is continuous, but its first derivative is not.
It is convenient to define a function which is proportional to the above E(x)
-KX
K!xlr 2/4H
M(x) =
2H (1 erf VKx214H) + e
It may be noted that FriT<M(x)
is
dimensionless and depends on the singlevariable, X = /<x2/4H , that is,
1 -X2
VH/K M(x) = - X [1 - erf X] + e
)77
This function is displayed in Figure 1. It is a kind of spreading function.
Strut-Like Body. Let a symmetrical strut be defined in terms of its
offsets as follows:
y = ± b(x) , 0 < x < L , -H < z < 0 .
The corresponding thin-ship source distribution is given by:
a (x,z) = 2 U b' (x) ,
0< x < L
, -H < z < 0(19')
,
(18)
M* (k) =
10
-I x-I
FIGURE 1.
THE FUNCTION
A
M(x) EXPRESSED IN .TERMS OF THE SINGLE VARIABLE Kx2Ix!.
4H
From the previous result (for a vertical line of sources), we can immediately write down the corresponding longitudinal source distribution:
E(x) = 2UH dE b' (E)
M(x-E)
.(20)
0Ordinary slender-body theory gives the same result if M(x)
is
replaced by(5(x) , that is E(x) = 2UHb'(x) . The representation in (20) shows that the
ordinary slender-body-theory source distribution is smeared out.
Since (20) has the form of a convolution integral, we can readily rewrite it in terms of Fourier transforms:
1
E(x) = dk eikx M*(k)
E(k)
27 where
Kr
0_ e-k2H/K) J k'H (21) (21')the transform of (19), and E*0(k) is the transform of the source density
func-tion in ordinary slender-body theory, that is, in the strut case,
0
to
1.21.4 1.6 1.8 2.0
.2
-T (X) =
specifying the keel line). We can then find E(x) explicitly, as follows,
aUK f dk eikx[l
-e-k2H/1{1
- 2eE(x) = -iLk/2 + e-iLk] T1i k3 -CO = 2aUH {T (x) - 2T
(X -
L/2) + (X L) (22) where X -X2 1 sgn x {- X2 - e + + 2X2) erf X} I 1/7(22')
with
x =
1X1 = /Kx2/4H , as before. Figure 2 shows E(x)/2aUH , as wellas E0(x)/2aUH , for the case L/H = 9 , F = U = 0.2 .
According to slender-body theory, the difference between E(x) and E0(x)
is of higher order than E0(x) . However, slender-body theory is not valid
near the body ends, and so the approach presented above is not necessarily
inconsistent in the sense of asymptotic analysis. In the computation of wave
resistance, the difference is absolutely critical, for, as we have noted, the
wave resistance corresponding to E0(x) is infinite, whereas the wave
resist-ance of the distribution E(x) is identical to the value given by Michell's
integral in the thin-ship idealization, and the latter is not grossly wrong.
ax , 0 < x < L/2 b(x) = a (L -x) , L/2 < x < L , ) -H < z < 0 , we have 2aUH , 0 < x < L/2 Z0 tx) = - 2aUH , L/2 < x < L 0 otherwise = 2aUH {H(x) - 2H(x- L/2) + H(x-L)}
where H(x) is the Heaviside step function (not to be confused with H(x)
Cr
E(k)
=0
dx e-ikx[2UHb1(x)] . (21")
Cr
For a strut with the shape of a double wedge of half-angle a , that is,
.
-F,
= 0.2
L/H
9
E(x)
2aUH
ORDINARY SLENDER-BODY THEORY
<MODIFIED THEORY
FIGURE 2.
SLENDER BODY SOURCE DISTRIBUTIONS FOR A SYMMETRICAL DOUBLE-BONDED WEDGE
ADVANTAGES OF USING SLENDER-SHIP THEORY
if appropriate assumptions are made with respect to Froude number, there are two aspects of slender-ship theory that make it potentially superior to
thin-ship theory, namely,
the interference between the bow wave and the bow form is included; the interference between the transverse waves and the entire body is included.
The second is of little or no interest in the strut problem. However, it
should be noted and recognized because it means that one of the primary
short-comings of thin-ship theory is thereby avoided, the so-called "sheltering
effect." This problem has been treated by Reed (1975).
On the other hand, the first may be of great importance in strut problems.
What it means is this: At each cross-section near the bow, waves are created
by the widening local beam. These waves appear locally as the diverging waves
of the usual Kelvin pattern. In thin-ship theory, these waves are all simply
added up. However, in fact, each such component suffers interference from the
other sections of the bow. Thus, the waves which we consider to be generated
at x = x1 have velocity components which violate the body boundary condition
for all x x/ . With the incident stream flowing in the positive - x
direc-tion, this phenomenon is not critical for x < x1 , but it may be very important
for x > x1 These interference effects are formally of higher order in
thin-ship theory, but they need not be in slender-thin-ship theory depending on the
assumptions made about rates of change of field variables.
A procedure for handling this phenomenon was proposed by Ogilvie (1972), although the only problem worked out in detail by him was based on the extra
assumption that this phenomenon could indeed be overlooked. Such an assumption
was not logically necessary in his analysis; it was only convenient in solving
a specific problem. This phenomenon was included in an analysis by Tuck (1973)
for the problem of a nearly flat body, and it led to significant consequences. The flat-body problem was also partially solved by Maruo (1967). A general
pro 13 pro
14
-cedure for handling these problems has been developed in detail by Daoud (not
yet published). All of these works were effectively anticipated by Cummins
(1956), although he did not solve any specific problems.
The basis of these analyses is all the same, although the rationalization
has taken many forms. Succinctly stated, the problem can be formulated as
follows: On the free surface, the boundary condition is taken as
(1)xx K(1)z =
°
z = 0 , (23)where, as before,
K =
g/U2 . On the body, there is a kinematic condition,4
= - Un1
where n is (strictly) a unit vector normal to the body, and nl is its x
component; for computing the left-hand side of (24), n may be considered as
a unit vector lying in a transverse plane, oriented normally to the contour of
the cross-section in that plane. The Laplace equation is assumed to be
satis-fied approximately in the two dimensions of the transverse planes:
(1,1,17 q,zz = 0
(25)
Finally, it is assumed that there is no disturbance forward of the bow, that is,
E 0 for x < 0 . A formal derivation of this problem was provided by Ogilvie
(1972), and it was embedded into a hierarchy of problems by Ogilvie (1974). The
following discussion largely follows the latter.
A fundamental solution of the problem set by (23) and (25) is:
CO 6 (x) r K r' (x;y,z) - log - H 2 (x) I ek ( z+ c ) cos ky sin (x) x) (26) 7r 7 0
where 6(x) is the Dirac function, r = [(y_T-)2+(,_)2]1/2 ro
[(y_n)2+(z+)2]1/2 , and H(x) is again the Heaviside function. The potential
given by (26) represents the flow due to an "impulsive source" at
(C)
< 0 , if we think of x as a time-like variable. In fact, if we replace
x/U
by t (time), the problem set by (23) - (25) corresponds precisely to a 2-D
problem in which the body shape changes with time (instead of with x ).
A solution of the complete problem given by (23) - (25) can now be given (24)
-Tr
0 0
C(E)
15
-by superposing fundamental solutions, as in (26), over the contour of the body at all cross-section, as follows:
1/2 , 1
(Y-11)2 + (z-C)
1
14)(x;y,z)
= -:r7 di a (x;r1,C) log (17-T1)2+ (z+)2)
C(x) (27) kz x K I dk ef
cl sin
/a-(x-C)
f
di a( ;71(),C(C))eIg(E)cos k(y-n( ))where C(x) denotes the contour of the body cross-section at a particular x
and u(x;y,z) is a source distribution over C(x) . A solution is not in hand
of course until a is known, and it must be found as the solution of an integral
equation derived from (24). The integral equation can be expressed in a
rela-tively simply way if we define C(x) as the image above z = 0 of the contour
C(x) and furthermore require that u(x;y,z) = -u(x;y,-z) . Then substitution
of (27) into (24) yields:
1 , 1
_ akxiy,z)
f
di u(x;n,c) n2(y-) + n3(z-C)2 27 C (Y.'1-1)2 C(x)+C(x)
(Z-02
= - Un1 + f(x;y,z)where
(nn2'n3) = n
, the outward unit vector on the body, andkz
x K dk ke sin f(x;y,z) -7 147.1.)4. 17(x-)
0 0 .1di ) ekC{-n2sin k(y-n) +n3 cos k (y-ri)1 .
C(C)
The integral in (28) is given a principal-value sense. In (28), the first term
on the right-hand side obviously comes from the right-hand side of (24). The
second term, f(x;y,z) , represents the normal velocity component on the
contour C(x) due to the upstream disturbances. Equation (28) has a simple
interpretation in which the basic character (and difficulty) of the free-surface
problem is removed: We defined a as an odd function in z , for convenience
in writing (28). But this means that a represents a source distribution that
could be used to satisfy a condition cl) = 0 on the plane z = 0 . The compli-(28)
(28')
-,
16
-cations of the free surface have been displaced entirely to the right-hand
side, and (28) is a relatively simple Fredholm equation over C(x) and
E(x)
.The right-hand side of (28) does contain a inside the triple integral,
and so it may seem that this term should be included with the initially unknown
terms on the left-hand side of the equation. If this were true, there would be
little chance of solving (28). Fortunately, f(x;y,z) depends on a( ;n,,C)
only for E < x . This occurs because of the factor sini47(x-) that appears
in (28'). If we solve (28) step by step, starting at x = 0 , the quantity in
(28') will be completely known at any x . Thus f(x;y,z) properly belongs on
the right-hand side, adding to the nonhomogeneous part of the problem. Of
course, (28) is also an integral equation over the x domain, and, from that point of view, it is a Volterra equation.
This formulation of the problem is actually well-rooted in work early in
the century by Kelvin, Lamb, Havelock, and others. They frequently treated
time-dependent flows in terms of impulsive pressure distributions, which, in
the presence of a free surface, create ongoing fluid motions that can be
deter-mined by methods which are generalizations of the Cauchy-Poisson solution. The
corresponding treatment of steady-motion 3-D ship problems depends essentially
on the transformation of such problems to sequences of 2-D problems, through
application of the slenderness concept. But, after that has been done, the
solution is formally equivalent to the solution of the corresponding time-depen-dent problem, as described above.
Equation (28) permits a direct comparison with the usual thin-ship theory.
First of all, there is no integral equation at all in thin-ship theory, since
the source distribution lies on a plane; therefore the integral term on the
left-hand side of (28) vanishes. Secondly, the right-hand side has two terms.
The first is exactly the same as in thin-ship theory. The second, as already
pointed out, represents the normal velocity component at any section due to the
upstream sources; this term is missing in thin-ship theory too, since the body
boundary condition is satisfied on the centerplane. In determining a so that
both terms on the right-hand side of (28) are accounted for, we have incorpor-ated the diffraction of waves by the ship itself.
17
-A procedure such as the above would not be valid in problems of low-speed
aerodynamics. We have taken advantage of the wavelike nature of the free-surface
problem to treat it almost as if the governing equation were hyperbolic. There is no such basis for overlooking the elliptic nature of the corresponding aero-dynamic problems.
The above procedure is more nearly valid in the free-surface problem as the
disturbance is limited to a region near the free surface. This statement depends
on a result proven by Ursell (1960). The above solution in two dimensions is
equivalent to the true 3-D solution to the extent that we can simplify the latter
to just a system of diverging waves. Ursell showed that near the track of a
translating pressure point the diverging-wave part of the solution (as obtained
by applying the method of stationary phase) dominates the solution. Furthermore,
this dominant part can be evaluated properly by the method of stationary phase
even at points close to the pressure point. Obviously, this is true only because
the solution of the pressure-point problem is so highly singular along the track. The solution corresponding to a submerged singularity is well-behaved on the free surface along the track, but Ursell's conclusion is probably valid to some extent if the submergence is very small.
The equivalence between the 2-D solution and the diverging waves of the
3-D solution is discussed in Appendix A. One point is especially remarkable
and should be noted here: The fluid motion close behind a point singularity
can be treated approximately as a slender-body flow, even though such a
singu-larity appears to violate all of the basic ideas of slender-body theory. Of
course, no such conclusion is possible in infinite-fluid (aerodynamics) prob-lems, and so one may speculate that slender-body theory ought to be more accurate in free-surface problems than in aerodynamics.
METHOD FOR SOLVING STRUT PROBLEMS
In the last section, we showed how a slender-body free-surface problem
can be solved approximately as a sequence of 2-D problems. The procedure has
the advantage that it includes effects of interference between the body and
its own waves. Also, there is no great handicap in treating unsymmetrical
bodies, including yawed bodies (such as struts), which lead to formidable troubles in thin-ship theory.
However, the method has two distinct shortcomings in applications to
struts and a third shortcoming in applications to bodies of extended length
in the longitudinal direction:
It is a slender-body theory, and as such it can hardly be used on
strut-like bodies.
It is based on an assumption of no disturbance ahead of the body.
This is apparently not serious for a truly slender body of shallow submergence,
but it may be expected to be important in strut problems.
Transverse waves are ignored.
We show here how to overcome the first and second objections. The third
will only be discussed briefly in the following section.
Let the potential be written as the sum of two parts:
(1)(x,y,z)
=epo(x,y,z)
+ ybi(x;y,z) , (29)where 1)0 satisfies the conditions:
(150 = 0 on z = 0 ; (30)
2(11)r°1 = - Unl , on body; (3L)
ci) (1) 0 in fluid region.
OxxOyy
ozzThe sum of the two terms in (29) must satisfy (23), and so cl), satisfies:
18
-(32) +
-+K
01z lxx =00z
gC + Ucl)lx = 0 UC - 00z-
Olz
-
19 -on z = 0 . (33)Thus, 01 is the velocity potential for a fluid motion which could be caused by the presence of an applied pressure distribution on the free surface. If we let the equivalent pressure distribution be represented by
po(x,y)
, its form is given byPox(x,Y) = pUK00x(x,y,0) . (34)
This interpretation is useful in solving for 01 , but it must be used
care-fully. A precise interpretation of (33) shows that there is an imposed vertical
velocity distribution on the plane z = 0 , not an imposed pressure field. The
two separate boundary conditions which combine to give (33) are as follows:
on z = 0
where (x,y) is the free-surface elevation; from the first of these, we have:
(x,y) =
-01x(x,y,0) (35)
It is all right to solve for 01 as if there were an applied pressure field,
given by (34), but we must use (35) to find
0x,y)
.We also
require1
to satisfy the kinematic body boundary condition:9n
0 on body. (36)
Finally, we revert to the discussion of the last section to justify an
assumption
that1
is approximately the solution of a 2-D problem:01yy Cblzz 0 in fluid region. (37)
It should be noted that our conditions are near to ideal for making this
assumption. In particular, the disturbance that leads to the definition of
occurs on the plane z = 0 , and the equivalence between 3-D and 2-D
problems is most nearly valid under such circumstances.
In principle,
pox(x,y)
is defined over the entire plane z = 0 outsideof the body. However, its magnitude drops off very rapidly with distance from
the body. From far away, the fluid motion generated by any translating body
in an infinite fluid appears as if it could have been generated by a
longitudi-nally oriented dipole. When that body moves below a surface on which = 0
the motion appears far away as if it had been generated by a quadrupole. Thus
we expect 4)
0
to have the following form far away from the body: Cxz
= (x2 +y2 +,2)5/2
where C is a constant. On z = 0 , we have
Cx
(POz (x2 +
y2)
5/2 rand this is proportional to
pox
, by (34). Thus, the disturbance whichgenerates 4)1 drops off ahead of the body as x The procedure proposed
for determining 4)1 is not valid at great distances behind the disturbance,
and we have a disturbance, pox , which extends upstream to infinity. In
prin-ciple, the procedure violates the conditions of its own validity. However, this
is not likely to cause difficulty, since the upstream disturbance is very small
except possibly just ahead of the body. Furthermore, whenever a smoothly
dis-tributed impulsive pressure is applied to the free surface in a 2-D problem,
the wave motion quickly disperses from that region. Only the possible
occur-rence of transverse waves, not included in this analysis, will vitiate the
solution. This possibility can be safely dismissed too, either on the basis of
actual observation or on the basis of the usual Kelvin wave analysis.
Ahead of the body, the 4), problem involves only conditions (33) and(37). Formally, the solution can be written:
(i)1(xiY,z) = - e dn cos k(y-n)
ITPU VI
pox(,n) siniTt(x-)
, (38)ck
1 dk kz
I
0 x < 0 .
The surface elevation is given by:
(x,Y)
=-
20 -dkI
dn cos k(y-n) 1I
r1)I
dEPox(E,T1)
cos(x-)
(39) 0x< 0.
Aft of the bow or leading edge of the body, the solution for cl), comprises
two terms, one like (38), a second like (27). In finding the function o to
be used in the latter, it is probably most convenient to solve first for 4)1
-as in (38), the body boundary condition being ignored, then to compute
41/3n
on the body from that solution for substitution into the right-hand side of(28) in place of Uni after which a second component of qb, may be found by
solving (28) and substituting the result into (27).
The solution as constructed above has some important virtues. First of
all, for a strut-like body, the first approximation satisfies the Laplace
equa-tion in three dimensions. Thus, the solution well below the free surface is an
accurate solution of the actual problem. Also, it can be obtained by existing
numerical methods. Secondly, this first approximation is invalid only near the
free surface, and a slender-body approach is valid there for finding the
neces-sary correction. The fact that the body itself is not slender is irrelevant.
Thirdly, the fact that the free surface elevation or its slope may be discon-tinuous transversely at the trailing edge of the body is not a source of
difficulty at all. The 2-D problems that have to be solved are similar to the
classical Cauchy-Poisson problem, which starts with the imposition of a delta-function elevation or impulse, and the actual surface discontinuities are much
less severe than these. In fact, no special procedure need be introduced for
handling these effects, since distributions of fundamental solutions like that in (26) are quite capable of representing these discontinuous "initial conditions"
and their effects automatically.
-REMAINING DIFFICULTIES
The procedure described above makes no provision for representing
trans-verse waves or their effects. This is probably of little or no importance in
strut problems. However, our procedure could be useful in solving problems,
say, of yawed ships if this deficiency were removed.
This difficulty can probably be worked out with use of the method of
matched asymptotic expansions. A far-field solution would be expressed in
terms of a line of singularities on the longitudinal axis (possibly augmented
by a sheet of singularities'on the centerplane). The strength of the
singu-larities would have to be obtained by matching to the near-field solutions.
This would be straightforward except near the bow, where we have not been able
to devise practical matching procedures. If we obtain a near-field 2-D solution,
as by Ogilvie (1972), we can match it with a far-field solution like that
gener-ated by the source distribution given in (22). The matching is similar to that
described in Appendix A. There is no "overlap region" in the usual sense. Rather,
the outgoing 2-D waves created by a distribution of sources on the centerplane can
be matched to the diverging wave system of the corresponding sources in three
dimensions. If the 2-D near-field solution is not expressed in terms of soarces
on the oenterplane, then the singularity distribution in the far field is not
readily determined. Undoubtedly it will be possible to develop an integral
equa-tion, the solution of which will give the far-field singularity distribution. But
we have not done this, and, until it is done, the transverse waves cannot be
con-sidered along the body in the near field.
The numerical analysis necessary in working out the new procedure will not
be easy, although Daoud has done many of the essential steps and will publish
his material soon. One difficulty may possibly be serious. The solution of the
clpo problem set in (30)-(32) is weakly singular along the line of intersection
of the body and the plane z = 0 . We can see this easily by magnifying a small
neighborhood of this line of intersection, as in Figure 3. For z < 0 , the
normal velocity component is approximately a constant in this small region. For
z > 0 , the normal velocity component has the same magnitude but opposite sign.
If this were a 2-D problem, with
= y+iz ,
the complex velocity potential23
-in the cross-section plane would be approximately
f = C iC [log - 1]
IT
with C equal to the value of co/ri on z < 0 . The 2-D velocity components
in general are given by:
cpy - ici)z = f'
(c)
= ((log )/y2 +z2 + i tan-1 (z/y)) .IT
Thus cP has the prescribed values on y = 0 , and
2C
= -
-7-T- log/y2 + z2 .This very weak singularity appears on the right-hand side of (33), and so the
"effective pressure distribution," pc) , has a singular behavior. In principle,
this should not give trouble, since it is so weakly singular. But it may be
very awkward numerically.
FIGURE 3. SINGULARITY IN (1)0 AT
INTERSECTION OF BODY AND PLANE Z=0
24
-We may note that this singularity could have been avoided had we defined
cp differently. In place of (30), suppose that we had required that
4
/3zD 0
= 0 on z = 0 . Then (33) would have become:ci) lz = and the
lxx Oxx '
right-hand side of this condition is not singular in the same way. We did not
choose to proceed this way for the following reasons: For a body of shallow
submergence, there is experimental evidence that the assumption of = 0 on
the free surface ahead of the body is a very good assumption. See, for example,
Ogilvie (1972). For such shallow bodies, our procedure should not require
large corrections to the lowest approximation. This requirement is satisfied
only when we start as indicated in the cpo problem of (30) - (32).
In problems of yawed struts, the most important phenomena may well be
cavitation and ventilation, about which we have said nothing at all. It is
just possible that the first of these, cavitation, can be analyzed by the
pro-cedures discussed in this report. In particular, the appearance of a cavity on
the strut is probably not very sensitive to the actual free-surface disturbance;
if the free surface is replaced by a (I) =osurface, a prediction of the
occur-rence and extent of a cavity would give a good first approximation for practical
purposes. Photographs taken during experiments with struts at the Naval Ship
Research & Development Center strongly suggest that this will be the case.
The prediction of ventilation is another matter. Its inception is probably
the result of flow instability around the leading edge, which certainly cannot
be predicted reliably by any methods such as those discussed here. It is con-ceivable that the flow in a ventilated condition could be predicted as well as
the flow around a cavity, if one is willing to assume a priori that ventilation
has indeed occurred. However, the occurrence of ventilation is so overwhelming
in its effects that the prediction of inception is the most important aspect of
its study. Cavitation, on the other hand, seems to build up gradually, starting
with very small cavities at locations of concentrated low pressure and gradually
enveloping a large part of the strut. From the point of view of predicting the
overall flow around a strut, the details of the cavity flow are not terribly significant unless possibly they provide the trigger for inducing ventilation.
In any case, the theory for predicting cavity flow even in the absence of a free surface needs more development before the free surface is considered more.
REFERENCES
Abramowitz, M., & Stegun, I. A., Handbook of Mathematical Functions, Mathe-matics Series, 55, National Bureau of Standards, 1960, Washington.
Cummins, W. E., The Wave Resistance of a Floating Slender Body, Ph. D. thesis, The American University (1956).
Daoud, N., Force and Moments on Asymmetric and Yawed Bodies on a Free Surface, Report NA 73-2, College of Engineering, University of California,
Berkeley (1973).
Maruo, H., "High- and Low-Aspect-Ratio Approximations of Planing Surfaces," Schiffstechnik, 14 (1967) 57-64.
Ogilvie, T. F., "The Wave Generated by a Fine Ship Bow," Proc. Ninth' Symposium on Naval Hydrodynamics, 1972, ACR-203, 1483-1525, Office of Naval Research, Washington, D. C.
Ogilvie, T. F., "Workshop on Slender-Body Theory, Part I: Free Surface Effects," Report 163, Dept. of Naval Architecture & Marine Engineering, The University of Michigan (1974).
Reed, A. M., Wave Making: A Low-Speed Slender-Body Theory, Ph. D. thesis, The University of Michigan (1975).
Tuck, E. O., The Steady Motion of Slender Ships, Ph. D. thesis, Cambridge
Uni-versity (1963). [See also: "A Systematic Expansion Procedure for Slender
Ships," J. Ship Research, 8:1 (1964) 15-23)
Tuck, E. O., "Low-Aspect-Ratio Flat-Ship Theory," Report 144, Dept. of Naval Architecture & Marine Engineering, The University of Michigan (1973).
Ursell, F., "On Kelvin's Ship-Wave Pattern," J. Fluid Mechanics, 8 (1960) 418-431.
-APPENDIX A: EQUIVALENCE BETWEEN TRANSLATING 3-D SOURCE AND IMPULSIVE 2-D SOURCE
The potential for the fluid motion caused by a translating point source of unit strength in three dimensions was given by (9), and simplified forms
con-taining the essential wave motion were given in (10) and (11). Let us start with
the form given in (11). Under the usual conditions for applying the method of
stationary phase, we find that the integrand has two points of stationary phase
if x > 0 and
fl <
2/-2-x , namely, for k given byk =
X2 X2
-8y2/x2 .
8y2 - 8y2
The upper sign gives the usual diverging waves, and the lower sign the
trans-verse waves. The dividing value is k =
KI/72-
(for which the inner square rootin (Al) is zero). We note the following in general:
K < k < transverse waves;
< k < diverging waves.
If y2/,2 << 1
, the expression in (Al) can be approximated:
Kx/21y1 (diverging waves)
k =
K(1 +y2/2x2) (transverse waves) .
Then, from (11), by applying the method of stationary phase, we obtain: r-- KX2
K
-
2-1-(1)s(X,y,Z)
=-/
e 4y21 isin(-Kx2-L)
7ry 4y 4
/I7
-K1Z.-4-C1-
sin (Kx + 7/4)Trx
The first term represents the system of diverging waves. If the source is
actually located on the free surface ( C = 0 ), the approximation for this term
is valid near the entire track of the singularity, even close to the singularity
itself. The same is not true of the second term, representing the system of transverse waves; the approximation for this term is only valid for large values of KX . However, the transverse waves are significant only at a considerable
26
-(Al)
(A2)
--
27-distance behind the source. (These statements represent a loose interpretation
of results proven by Ursell (1960) for the case of a moving pressure point.)
In the region of interest to us, i.e. y2/x2 << 1 but KX small, the
most important part of the wave motion comes from the part of the integral in
(11) for which k is very large, in fact, for
k/K >>
1. We use this fact asfollows: In (10), which is equivalent to (11) for x > 0 , expand the integrand
for large k by taking
K2
/k2 -K2 = k (1 + )
We then have
= 1
1- 7772
cos (kyl/k2/K2 -1) = cos k2Y
and so the entire integrand becomes k2y
e K sin kx cos
Now let k = , so that this integrand becomes:
k'l x cos k'y and dk = 1-= d k ' 2 V,71TT (10) now becomes: CO cips(x,y,z) _ ek'(V-C) 7 Kki
COS k'y
sin1/57x
.Since the value of the integral depends only on the higher range of k , it does
not change much if we replace the lower limit by zero. Then this expression
becomes identical to that given by (26) for x > 0 , and the latter is the
potential for the fluid motion in two dimensions caused by an impulsive unit
source at x = 0 .
If the reader is not convinced by the preceding loose demonstration, he
can estimate the integral in (26) by the method of stationary phase, and he will
z+0
28
-obtain the first term on the right-hand side of (A2).
The above results mean that the flow close behind a translating source can be analyzed approximately by treating it as a 2-D flow in the cross-planes. There are some restrictions on this conclusion however:
The equivalence breaks down well before one gets back to the "first" transverse wave.
The equivalence is most accurate for a source at the free surface; it rapidly becomes invalid for any significant submergence of the source.
UNIVERSITY OF MICHIGAN REPORT NO.
175October 1975
DISTRIBUTION LIST FOR REPORTS PREPARED UNDER THE GENERAL HYDROMECHANICS RESEARCH PROGRAM Commander
40
David W.
Taylor Naval ShipResearch and Development Center
Bethesda, Ed
20084 Attn: Code1505 (1)
Code 5614 (39)
1 Officer-in-Charge
Annapolis Laboratory Naval Ship Research and
Development
CenterAnnapolis, Maryland
21402
Attn: Code 5642(Library)
Commander
Naval Sea Systems Command Washington, D. C. 20360 Attn: SEA 09G32 (3 cys) SEA 03512 (Peirce)
SEA 037 ADF
SEA 0322 SEA 033
12*
Director
Defense Documentation Center 5010 Duke Street
Alexandria, Virginia 22314 1 Office of Naval Research
800 N. Quincy Street
Arlington, Virginia 22217
Attn:
Mr. R. D. Cooper (Code
438) 1 Office of Naval ResearchBranch Office
492 STIer Street
Boston,
Mass
022101 Society of Naval Architects and Marine Engineers
74 Trinity Place
New York, New York 10006 Attn: Technical Library
1 Sun Shipbuilding & Dry Dock Co.
Chester, Penna 19000
ABC
Attn: Chief Naval Architect
1 Sperry Systems Management Division Sperry Rand Corporation
Great Neck, N. Y. 11020 Attn: Technical Library
1 Office of Naval Research Branch Office (493) 536 S. Clark Street Chicago, Illinois 60605 1 Chief Scientist
Office of Naval Research Branch Office
1030 E. Green Street
Pasadena. CA 91106
1 Office of Naval Research Resident Representative 715 Broadway (5th Floor) New York, New York 10003 1 Office of Naval Research San Francisco Area Office 760 Market St., Rm 447
San Francisco, Calif 94102
2
Director
Naval Research Laboratory Washington, D. C. 20390
Attn: Code 2027
Code 2629(0NRL)
1 Commander
Naval Facilities Engineering Command (Code 032C)
Washington, DL. C. 20390
1 Library of CorLgress
Science & Technology Division Washington, D.- C. 20540
1 Stanford Research Institute Menlo Park, CA 94025
Attn: Library G-021
2 Southwest
ReseaLh
Institute P. O. Drawer 28510San Antonio, Texas 78284
Attn: Applied Mechanics Review Dr. H. Abramson
Attn: Library
3 Institute of Hydraulic Research
The University of Iowa
Iowa City, Iowa 52240 1 Commander ADEF
Attn: Library
Dr. L. Landweber BDE
Kennedy BDE
Naval Undersea Center San Diego, CA 92132
Attn:
Dr. A.
Fabula (6005)Dr. J.
1 Commander . BDEF
Naval Electronics Laboratory. 2 Officer-in-Charge
Center (Library) Naval Undersea Center
San Diego, CA 92152 Pasadena, CA 91107
Attn: Dr. 3. Hoyt (2501)
8 Commander Library (13111) ADEF
Naval Ship Engineering Center Center Building
Prince Georges Center
1 Newport News Shipbuilding &
Hyattsville, Maryland 20782 Dry Dock Company
Attn: SEC 6034B 4101 Washington Avenue
SEC 6110 Newport News, Virginia 23607
SEC 6114H Attn: Technical:Library
Dept-SEC 6120
SEC 6136 1
,
Library
SEC 6144G Naval Underwater Systems Center
SEC 6140B ADEF Newport, R. I. 02840
SEC 6148 ADEF
1 Research Center Library
1 Naval Ship Engineering Center Waterways Experiment Station
.Norfolk Division ABCE Corp ofEngineers BCDE
Small Craft Engr Dept P.O. Box 631
Norfolk, Virginia 23511 Vicksburg, Mississippi 39180
Attn: D. Blount (6660.03)
1 Eastern Research
Group--1 Library (Code 1640) P.O. Box 222
Naval Oceanographic Office Church Street Station
Washington, D. C. 20390 New York, New York 10008
1 Technical Library
DE'
1 Esso International
Nava/ Proving Ground Design Division, Tanker Dept.
Dehlgren, Virginia 22448 15 West 51st Street ABCD
New York, New York 10019
1 Comr,?nder (ADL)
Naval Air Development Center 1 AFFOL/FYS (J. Olsen) DE
Warminster, Penna 18974 Wright Patterson AFB
Dayton, Ohio 45433
1 Naval Underwater Weapons Research
& Engineering Station (Library) 1 Dept. of Transportation
Newport, R.I. 02840 Library TAD-491.I
400 - 7th Street S.W. .
1 Commanding Officer (L31) Washington, D. C. 20590
Naval Civil Engineering Laboratory
Port Hueneme, CA 93043
1 Gibbs & Cox, Inc.
21 West Street
Lockheed Missiles & Space Co. New York, New York 10006
P.O. Box 504 Attn: Technical Info. Control
Sunnyvale, CA 94088
Attu: Mr. R. L. Wald, Dept 57-74 Hydronautics, Inc.
Bldg. 150, Facility 1 Pindell School Road
Howard County
Laurel, Maryland 20810
-1 Portsmouth Naval Shipyard
Technical Library
Portsmouth, N. H. 03801
1 Puget Sound Naval Shipyard
Engineering Library
Bremerton, Wash 98314
1 Long Beach Naval Shipyard
Technical Library (246L)
Long Beach, CA 90801
1 Hunters Point Naval Shipyard
Technical Library (Code 202.3)
San Francisco, CA 94135
Pearl Harbor Naval Shipyard Code 202.32
Box 400, FPO
San Francisco, CA 96610
1 Mare Island Naval Shipyard
Shipyard Technical Library Code 202.3
Vallejo, CA 94592
1 Assistant Chief Design. Engineer for Naval Architecture (Code 250) Mare Island Naval Shipyard
Vallejo, CA 94592
2 Florida Atlantic University
Ocean Engineering Department
Boca Raton, Fla 33432
Attn: Technical Library Dr. S. Dunne
2 College of Engineering
University of Notre Dame Notre Dame, Indiana' 46556 Attn: Engineering Library Dr. A. Strandhagen
3 U. S. Naval Academy
Annapolis, Maryland 21402
Attn: Technical Library Dr. Bruce Johnson Prof. P. Van Mater, Jr. Naval Postgraduate Sdhool
Monterey, CA 93940
Attn: Library, Code 2124
1 Capt. L. S.
McCready,
USMS Director, National MaritimeResearch Center'
U. S. Merchant Marine Academy
Kings Point, L.I., N.Y. 11204
1 U. S. Merchant Marine Academy
Kings Point, L.I., N.Y. 11204
Attn: Academy Library
1 Library
The Pennsylvania State University Ordnance Research Labbratory P. O. Box 30
State College, Penna 16801
1 Bolt, Beranek & Newman DEF
1501 Wilson Blvd
Arlington, Virginia 22209
Attn: Dr. F. Jackson
1 Bolt, Beranek & Newman
50 Moulton Street
Cambridge, Mass 02148
Attn: Library
1 Nielsen Engineering Research Inc.
510 Clude Avenue
ABDF
Mountain View, CA 4043
1 Bethlehem Steel Corporation
25 Broadway
New York, New York 10004
Attn: Library (Shipbuilding)
1 CALSPAN Corporation
P.O. Box 235
Buffalo, New York 14221 DE
Attn: Dr. A. Ritter
Aerodynamics Res. Dept.
1 University of Hawaii
Department of Ocean Engineering 2565 The Mall
Honolulu, Hawaii 96822
Attn: Dr. C. Bretschneider
ABCD
BDEF
3 California Institute of Technology
Pasadena, CA 91109
Attn: Aeronautics Library
Dr. T. Y. Wu
Dr. A. J. Acosta ABDE
Charleston Naval Shipyard Technical Library
Naval Base
Charleston, S. C. 29408
Norfolk Naval Shipyard Technical Library
Portsmouth, Virginia 23709
1 Philadelphia Naval Shipyard
Philadelphia, Penna. 19112 Attn: Code 240
University of Kansas
Chm Civil Engr Dept Library DE
Lawrence, Kansas 60644
1 Oceanics, Inc.
Technical Industrial Park Plainview, L.I., N.Y. 11803
Department of Ocean Engineering
Massachusetts Institute of Technology Cambridge, Mass 02139
Attn: Department Library
Prof. P. Leehey DEF
Prof. P. Mandel-Prof. M. Abkowitz Dr. J. Newman 1 University of Illinois Urbana, Illinois 61801 DE Attn: Dr. J. Robertson
St. Anthony Falls Hydraulic Laboratory University of Minnesota
Mississippi River at 3rd Avenue S.E. Minneapolis, Minnesota _55414
Attn: Prof. E. Silberman
Mr. J. Wetzel BDEF
Mr. F. Schiebe BDEF
Mr. J. Killen DEF
Dr. C. Song BCDE
Department of Naval Architecture and Marine Engineering
University of Michigan
Ann Arbor, Michigan 48104
Attn: Library
Dr. T. F. Ogilvie Prof. F. Hammitt
1 Catholic University of America
Washington, D. C. 20017 . ABDF
Attn: Dr. S. Heller, Dept of -Civil &.Mech.Engr 1 Colorado State'University
Foothills Campus
Fort Collins, Colorado 80521 Attn:
Reading Roan,Engr Res
1
-Mr. V. Boatwright,
Jr.
R
D ManagerElectric Boat Division
General Dynamics Corporation
Groton, Coon 06340
University of Bridgeport Bridgeport, Conn 06602 Attn: Dr. E. Uram
Cornell University
Graduate School of Aerospace Engr
'them
New York 14850 - BDAttn: Prof. W. R. Sears 4 University of California
Naval Architecture Department College of Engineering Berkeley, CA 94720 Attn: Library Prof. W. Webster Prof. J. Paulling Prof. J. Wehausen 3 Davidson Laboratory
Stevens Institute of Technology' 711 Hudson Street
Hoboken, New Jersey
07030
Attn: LibraryMr. J. Breslin Mr. S. Tsakonas
1 Department of Mathematics St. John's University Jamaica, New York 11432 Attn: Prof. J. Lurye
1 Tracor, Inc. 6500 Tracor Lane Austin, Texas
78721
College, of Engineering Utah State University
Logan, Utah 84321
Attn:
Dr. R. Jewson
1 Mr. Robert Taggart3933 Walnut Street
Fairfax, Virginia 22030 Ocean. Engr Department
Woods Hole Oceanographic Inst. Woods Hole, Mass 02543
Worcester Polytechnic Inst. Alden Research Laboratories
Worcester, Mass 01609 ADE
Attn: Technical Library
1 Applied Physics Laboratory
University of Washington ABCDF
1013 N. E. 40th Street Seattle, Washington 98105 Attn: Technical Library
BDEF Center ABCD ABDI AB AB AB ABC
-&
4Webb Institute of Naval
ArchitectureCrescent Beach Road
Glen Cover, L.I., N.Y.
11542Attn:
Library
Prof. E. V. Lewis
Prof. L. W. Ward
1
National Science Foundation
-
Engineering Division Library
DE1800 G Street N. W. Washington, D. C. 20550
University of Connecticut
Box U-37 -DE
Storrs, Conn 106268 Attn: Dr. 'V. ScottronHydraulic Research Lab
1 Applied Research Laboratory P.O. Box 30
State College, Pa 16801
Attn: Dr. B. Parkin, Director
Garfield. Thomas Water Tunnel
Dr. Michael E. McCormick
Naval Systems Engineering, Department
U. S. Naval Academy
Annapolis, Maryland
21402Dr. Douglas E. Humphreys (Code 712) Naval Coastal Systems Laboratory
Panama City, Florida 32401
ABODE_
Page. = 7'
pt
Stanford University'
Stanford, CA. 94305
Attn:
Engineering
Library
Dr. R. Street
2
3
1