CROSS-SPECTRAL ANALYSIS OF SHIP MODEL MOTIONS:
SOME EXPERIMENTAL AND COMPUTATIONAL PROBLEMS
by John F. Daizell
Report 853 November 1961 DL Project GE 2245 SPONSORED BY BUREAU OF SHIPSFUNDAMENTAL HYDROMECHANICS RESEARCH PROGRAM TECHNICALtY ADMINISTERED BY
DAVID TAYLOR MODEL BASIN CONTRACT Nonr 263(09) PROJECT S-R009 01 01
DAVIDSON LABORATORY
Stevens Institute of Technology
Castle Point Station
Hoboken, New Jersey
Approved by
P. Breslin
TABLE OF CONTENTS iii Wave System 10 Computational Techniques 11 Summary 1k Acknowledgements 15 References 16 Abstract Page V I. Introduction 1 A. General 1
B. Recording and Data Reduction 2
C. Towing Arrangements 3
II. Effects of Towing Techniques 5
A. Test Setup 6
B. Data Analyses 7
ABSTRACT
This report presents the results of investigations of the low coherencies found in the first cross-spectral
analysis of ship model-tests. The major cause of these low coherencies was the computational technique used. The first
analyses failed to provide enough resolution in the rapidly varying cross spectra inherent in the experiment. Although
an optimum solution is not available, practical methods of improving the computational technique were developed.
I. INTRODUCTION
A. GENERAL
This report presents the results of a series of
experiments that were conducted to provide further
informa-tion on the causes of the low coherency that resulted when cross-spectral analyses were first applied to ship-model
tests, and forms a part of the attempt to substantiate the.
validity of spectral and cross-spectral analyses applied to ship-model motions (references 1 through
5).
Cross-spectral analysis was first used (in
1958)
toanalyze ship-model motions in random seas (reference k) by comparing model transfer-functions derived from experiments conducted in irregular head-seas with. those derived from experi-.
ments conducted in reguiax waves. The. attempt was partially
successful; derived amplitude and phase responses (transfer
functions) were in reasonable agreement. However, coherencies
(a measure of the linearity of the systeiri') between wave data
and heave or pitch data were obtained that were significantly
less than 1.0, though, the coherencies between heave and pitch
were about
0.95.
These anomo].ous results cast serious doubts on the validity of .analyzjng:shjp...modeis as linear systems.Before the crOss-speotral analysis technique was
used to reduce large amounts of data It was studied in greater
detail to determine whether or. not the low computed cohCrencies
were a consequence of the experimental and computational
methods. . . .. .
*A coherency of 1.0 implies perfect linear dependence of
out-put on inout-put. A coherency of 0.0 implies, two incoherent .pro cesses--that is', different and essentially independent.
B. RECORDING AND DATA REDUCTION
The first consideration was to determine whether
the electromechanical system, which converted the motions of
the model and the wave elevations to digitized time-series
suitable for cross-spectral analysis, contributed any non-linearities or had any odd frequency response characteristics.
A concurrent general program of equipment
improve-ment at Daidson Laboratory showed that the electromechanical
transducers used previously (reference k), coupled with
ampli-fiers anda digital data processing system (reference 6),had flat amplitude-responses over the range of frequencies
encountered in the model tests. Quantitatively, these
trans-ducer responses were flat well within the resolution of the
digital system, which was one part in 127 for full scale. (This resolution is about the same as can be expected from
manual readings of oscillograph tapes.) The phaseresponse of all transducers was as good; no measurable relative phase
error between different transducers was found. The standard test technique is to take no data unless static calibrations are linear to a tolerance not exceeding the resolution of the
digital system. Therefore, the only distortion of results that were attributable to the sensing and recording systems were reading errors associated with the resolUtion of the
digitizer.
Reading errors manifest themselves in spectral
analysis as white noise. The following equation is an apprOx-imate formula for the coherency of a linear system with both input and output contaminated by incoherent white noise
(reference
7).
Coherency = I
1.
s(o)11
11+
Illwhere s(a) and
are the spectral densities of whitenoise, and S(a) and s() are the spectral densities of the
input and output signals0
For the narrow-band processes involved in model work, an estimate of white-noise energy may be obtained by
noting that the scalar spectrum fails to go to zero at fre-quencies far removed from the significant band of energy.
The density of the scalar spectrum at these remote. frequencies
is the magnitude of the white-noise spectrum, which by
defi-nition is constant for all frequencies.
The estimates of the noise-energy to signal-energy ratio for wave- or model-motion spectra, obtained at Davidson Laboratory, seldom have exceeded 0.02 at frequencies near the peaks of the spectra and increased as frequency increased or
decreased.
Thus, the coherency expected in the previous,
experi-ments was about 0.96 over the range of significant input and
output energy. At frequencies outside those associated with significant energy content, coherency is expected to go to
zero. This behavior was exhibited by the heave-pitch
coher-ency (reference )-i-); coherencies of about 0.95 were obtained
in the significant range of spectral energy. However, in the
first application of cross-spectral analysis to ship model
motions, both the wave-pitch and wave-heave coherencies showed
a kO% reduction near the spectral peaks instead of a k
reduc-tion (reference k). Therefore, the effects of white noise could not be considered to be one of the major causes of the
anomolous coherency results reported in reference k.
C. TOWING ARRANGE TS
In the case of pitch and heave, model motions were directly measured with reference to a single vertical heave
directly abeam of the point in the model where heave was measured; therefore, the wave probe was mounted ahead of the
model. Because of this, the experimental system was set up
so that the input was the wave ahead of the model,
and
theoutput was the heave (pitch) at the model (Figure 1).
The usual ideal system has the undistorted wave profile at amidships on the modelas the input. Therefore, the problem was to determine whether the experimental setup caused any coherencies lower than those of the ideal system.
Figure 1 shows the experimental setupused for the experiments reported in reference k. The model was permitted to drift relative to a constant-speed carriage and to oscillate
in surge under the influence of a constant tOwing-thrust. To
do this, (i) the model was towed by a sub-carriage that was permitted to move relative to a main carriage, and (2) towing weights were adjusted so that the mean speed over the run
equalled the carriage speed (Vc). A wave probe was mounted
on the constant-spèèd carriage.
The model moved relative to the main carriage afli
the distance (D) between the wave wire and the model was a
function of time--that is,
D = Do D1(t)
Do = Initial distance (constant)
D1(t) = oscillatory surge plus low-frequency drift, which
resulted from small changes in model speed during
the run. . ..
These experiments were conducted under moderate-wae.
condi-tions. During the experiments, no oscillatory surge was
apparent, however, low-frequency drift (speed change) was observed, which obviously was not linear with wave elevation In the sense of the assumptions used in linear systems analysis.
II.
EFFECTS OF TOWING TECH1'IQUESWhen waves
areassumed to be linear, the wave
eleva-tion at the model Oan be derived from the wave elevaeleva-tion
ahead
of the model. The amplitudes of all components do not change
and the phase is D2D/g (reference 1+) Thus, the theoretical transfer function of the wave at the model with respect to the wave ahead of the model is
Where, a is the wave frequency and D is a constant.
The transfer function of the response of the model
with respect to the wave at the model can be expressed as:
ió (w)
Z(We)e
z ewhere, We = w
for head seas
Z(CUe) = (response amplitude operator)V2
óz(We) = phase lag of response after the wave.
The theoretical transfer function of the model response with respect to the wave ahead of the model is
iE2D/g +
'Z(We)e z
When D = + D1(t) is substituted, the phase relations between wave elevations ahead of the model and the model response can not be uniquely defined with the present methods of analysis,
except when D1(t) = 0--that Is, with normal towing techniques,
the model response, with respect to the wave ahead of the model, is nonlinear.
An experiment was devised to check whether this was the cause of the low coherencies reported in reference .
A. TEST SETUP
A destroyer model was towed in long-crested waves of moderate height (average height 1/60 L) Three
dif-ferent towing techniques were useth
Model permitted to drift relative to a constant-speed carriage, and to oscillate
in surge while under constant towing-thrust; wave wire fixed on carriage--that is,
D1(t)
0, V/ V, D
= 6.0 feet.Model towed at constant speed, surge and drift restrained; wave wire fixed on carriage--that is, D1(t) = 0
V = V,
= 6.0
feet.Model permitted to drift relative to a constant-speed carriage, and to oscillate
in surge while under constant towing-thrust;
wave wire fixed on a small sub-carriage and moved longitudinally with the model--that
is D1(t) = 0 V jt V, D
= 6.0
feet.The tests were set up so that the effective mass
i'n longitudinal-motion of the model and sub-carriage system
was the same in technique 3 as in technique 1.
Carriage speed was a constant 2.53 ft/sec and analogue and digital records of wave, pitch, and heave were recorded for each towing technique. Because of the
repro-ducibility of the irregular-wave program the detailed wave-time histories obtained for the three towing techniques
showed good correspondence. Records of the model drift relative to the constant-speed carriage for techniques 1
and 3 were similar in character and generally agreed within
10% in magnitude.
B. DATA PNALYSIS
Because of the limited length of the tank, several runs were necessary to cover the full irregular-wave program. Each run contained about 20 wave encounters; during five runs, about as many enáounters occur as are usually observed at a fixed point during a full irregular wave program. The recOrds were sampled at an interval of 0.2 sec (Thkay rule of thumb),.
and 30 lags were used for the analyses which consisted of spectra and cross-spectra for wave and pitch, and wave and
heave. Tukey degrees-of-freedom for the analysis of each
run were about
5.5; 27.5
for five runs.The energy spectra of wave, pitch, and heave,
obtained when different towing techniques were used. for tests
conducted under otherwise similar conditions, were almost
identical. The energy spectra and cross-spectra obtained from the five runs obtained with each test technique were averaged; amplitude and phase responses and coherency were calculated from these averages using the methods cited in
reference 4 (Figures 2 through 4).
Figure 2 shows the amplitude response of heave and pitch to wave obtained by taking the square root of the ratio
of the respective spectra. The data obtained with each of the test techniques are shown, as is regular-wave data.
Figure 3 shows phase lags of maximum upheave or
bow-up pitch after the time that the component .wave crest is at the LCG of the model. Results derived from the irregular-wave tests for the three test techniques and approximate
regular-wave results are shown.
Figure 4 shows coherencies. between wave-pitch and
C. DISCUSSION
Figure 2 shows that no significant, differences in
amplitude response occurred with the different test techniques. Data obtained during the irregular-wave tests agree
reason-ably well with regular-wave data. No different conclusions could be drawn from comparisons of the amplitude responses computed from the cross-spectra for the three test techniques. This result is in agreement with that from a similar investi-gation on the effect of towing technique on pitch and heave
in regular waves.
Figure 3 indicates that techniques 2 and 3 yield
nearly the same phase results. One technique restricts surge and drift of the model completely; the other permits both. The phase results of technique 1 do not agree with the others. The results for technique 1 were obtained by using the nominal
distance D
= 6.0
feet as the dIstance between the wave wireand the model. Actually, the recorded mean distance through-out the runs probably averaged slightly more than
6.0
feet.If D had been assumed somewhat greater when the results for technique 1 were computed, there would have been better
agree-ment with techniques 2 arid 3
Figures 2 and k indicate that the low coherencies were not óaused by model drift or surge nor were they caused by the varying distances between the wave wire and the model;
though this might be important under sea conditions more severe than those simulated in these experiments.
Under moderate wave conditions (similar to those
cited in reference )+):
Derived amplitude responses are not
sensi-tive to minor variations of speed during
the run or to the test technique.
Derived phase angles are not sensitive to minor speed changes during the run or
restrictions in surge; however, they are quite sensitive to variations in the
dis-tance between the wave wire and the model.
3.
Coherencies between wave and pitch or wave and heave displayed the same tendencies and magnitudes as those reported in reference 5 regardless of test technique, restriction in surge, or varying distance between wavewire and model.
As a result of these experiments it was tentatively recommended that future experiments be conducted with the
wave wire at a fixed distance from the model at all times.
Because neither the experimental techniques nor the data-gathering techniques appeared to be the cause of the low coherencies, the physical process was more closely examined.
III. WAVE SYSTEMS
The methods of analysis used in reference k depend heavily on a knowledge of the behavior of the waves between
the point of measurement and the model. Therefore, the
hypothesis drawn stated that the waves causing the response may not have been predictable from the measured waves.
Direct experiments were made to determine whether the theoretical transfer function relating two wave measure-ments a fixed distane apart would be verified by
ôross-spectral analysis. Two experiments wer:e made: the first with the wave probes five. feet apart, the second with the wave
probes 145 feet apart. The resulting scalar spectra were
essentially the. same for all wave probes, apparently verifying
that the amplitude part of the transfer function.was valid and
Indicating that the wave process was statistically stationary..
However, the coheréncies computed from the experiment with the probes spacedfive feet apart werelow. The value of coherency
was about
0.65
at the pa.k of the spectrum; the tendency was for coherencies to decrease with increasing frequency. Thisresult paralleled the coherency results between wave and model
reported In reference 14. The coherencles computed from the
experiment where the probes were spaced at 145 feet were so low
that the process could have been termed incoherent.
No immediate conclusions were drawn but it was thought that the mOdel could be temporarily eliminated from. the low-coherency problem and attention concentrated on the waves for two reasons;
l. The dual wave probe experiments showed the computed coherency to be strongly dependent
on the separation of probes.
2. Coherencies between model responses were. good (reference 14, pitch and heave).
IV. COMPUTATIONAL TECHNIQ,UES
The problem in the preceding section was brought to the attention of Professor W. J. Pierson of New York
University as a matter of interest arid he suggested a jOint
effort on the question of theadequacy of the cross-spectral computation technique prior to concluding that the waves generated in the tank did not approximate a linear system
This joint effort resulted in the major part of the solution to the low coherency problem (reference
5)
. The conclusion in reference 5 states: teThe loss of coherencywhen computed by standard techniques from samples of vector Gaussian processes, where, theoretically,the coherency ought to be one, is largely explained by the lack of resolution in the cross spectra and the effect of the convolving filter made necessary by the nature of the:flnite sample. High
coherencies can be regained by modifying the experimental design so as to obtain less rapidly varying, cross spectra
and by increasing resolution."
Before proceeding 'with model tests, another check
on the validity of the methods in reference 5 was undertaken.
In this case, instead of dealing with wave measurements at fixed points in the tank, a model experiment was simulated by replacing the model by a wave probe and moving both probes
at a constant speed of 2.92 ft/sec. Wave probe separation was 6.0 feet, and the translation velocity of both wires was
opposite to that of the Wave celerity. A moderate irregular wave program was used.
Three runs..at this speed 'were analyzed using the
data shifting method outlined an reference
5.
This method amounts tore-aligning the records so that the output which,occurred T seconds later than the input is analyzed :aB if it
had been measured simultaneously with the input. The
theo-retical phase relationships involved are shown in Figure
5.
The range of interest is that range where significant energyexists in the wave spectrum. The line labeled w2D/g
repre-sents the theoretical phase lag of the wave component measured at the wave probe furthest from the source of the waves after
the component at the first wire. The theoretical phases,
introduced into the data by shifting, are also plotted. Where w = wave frequency and We= encounter frequency, w2D/g
- eT
is the net theoretical phase after shifting.Thus, if the wave process is linear and stationary, the modulus of the derived transfer function would be 1.0, and
the arguement would be w2D/g - WeTs
Selected results of the analysis are shown in
Figures 6 through
9.
Figures 6 and 7 illustrate the consider-able change in the form of the crossspectrum brought about by the shifting process and the resulting improvement of coherency for one of the three dual wave wire runs.Figure 8 shows the scalar spectra resulting from the average of the three dual wave wire runs. The results
indicate that the spectrum encountered by the model will be virtually the same as that measured ahead of the model.
Figure 9 shows coherency between wave records
obtained at two moving points and the results of the derivation
of the, transfer function. The.coherency is 90$ or better
over the range where, the spectral energy is greater than 20$ of maximum. The deviations of the derived phase angles
from the theoretical are shown; an approximate 95% confidence
bound on the phase is indicated. (reference 8). In view of the width 'of the confidence bound, phase agreement with
theoretical is shown to be excellent where high coherencies
and significant energies prevail. Figure 9 also shows the derived amplitude response from scalar spectra and cross
spectra as well as 95 confidence bounds on the response derived from the cross spectra. Again, agreement with
theoretical assumptions is excellent. The average deviation from theory is about k, a very respectable accuracy for
this type of analysis.
It was concluded that by using the methods reported
in reference
5,
the difficulties involved in deriving, thecoheröncy and transfer function of the Ideal wave to model system from the experimental system would be largely overcome.
It must be noted that neither the work cited in reference 5 nor that cited In this report has outlined an optimum procedure for crossspectral analysis. It is certain
that rapid oscillation of the cross spectrum is to be avoided, whatever the experiment, If the lengthof sample Is. limited
(reference 5)
An approach to optimizing the analysis would be to operate on the data so that the phase response of output to
input is a constant multiple of 7r/k. This would have the
effect that co and quadrature spectra.would have the same shape as the scalar spectra and that resolution optimum for scalarspectral analysis would be optimum for crossspectral
analysis. Because the exact phase between input and output is almost always unknown, this is Impossible in principle
and could only be. approximated by an iterative process.
V. SU1YINAPZ
The investigations into the effect of experimental techniques on wave-model coherencies disclosed no large
direct effects. However, the experimental setup previously used indirectly affected the coherency because rapidly
oscil-lating cross spectra were produced. This condition resulted in insufficient resolution of cross spectra when standard
computational techniques were used. Insufficient resolution in the analysis of the data, in turn, accounted for the low
coherencles of reference k. A practical method of overcoming
VI. ACKNOWLEDGEMENTS
The author wishes to acknowledge the contributions of Professor W. J. Pierson, Jr. of New York University who indicated the way to successful practical solutions of the
coherency problem. The author also wishes to acknowledge the assistance of those members of Davidson Laboratory who assisted
in these investigations; particularly, Mr. Wilbur Marks arid Mr. P. G. Spens.
VII. REFERENCES
Lewis, E. V. and Numata, E.: "Ship Model Tests in Regular
and Irregular Seas," September
1956,
E.T.T. Report567.
Lewis, E. V. and Daizell, J.: "Motions, Bending Momentand Shear Measurements on a Destroyer Model in Waves,"
E.T.T. Report
656,
April1958.
Daizell, J: "Ship Model Tests in Irregular Waves with a Broad Spectrum," April 1958, E.T.T. Note
k71.
11.. Daizell, J. F., and Yamanouchi, Y.: "Analysis of Model
Tests Results In Irregular Head Seas to Determine Motion Amplitudes and Phase Relationships to Waves," Ship Behavior at Sea, Second Summer Seminar, June
1958.
PIerson, W. J., Jr. and Daizell, J. F.: "The Apparent Loss of Coherency in Vector Gaussian Processes due to
Computa-tional Procedures with Application to Ship Motions and Random Seas," Joint Report, Davidson Laboratory,NewYork University, Department of Meteorology and Oceanography,
September
1960.
Spens, P.: "A Digital Recording System for Model Tests in Irregular Waves," DL Note
550,
Presented at the 12th Meeting of ATTC, Berkeley, CalifornIa, August1959.
Coleman, T. L., Press, H. and Meadows, M. T.: "An
Evalua-tion of Effects of Flexibility on Wing Strains in Rough Air for a large Swept Wing Airplane by means of Experiment-ally Determined Frequency Response inctions with an
Assessment of Random Process Techniques Involved," NACA
TN k291, July 1958.
Goodman, N. B.: "On the Joint Estimation of the Spectra, Cospectruin, and Quadrature Spectrum of a Two-Dimensional
Stationary Process," March 1957, Scientific Paper No. 10, Engineering Statistics Laboratory, New York University.
MODEL
I
SUBCARRIAGE.
CONSTANT-SPEED CARRIAGE
FIGURE
ITEST SETUP
II-vcRAIL
TOWINGWEIGHT
I-I
Ui=
ZO
LL cr.% LLJU).Q-w
WW
ctx
I-Ui
c
-j--0I
C-) I-0 540
30
20
(POINTS
FROM REGULAR
WAVETESTS SHOWN As: 0)
FIGURE 2
AMPLITUDE RESPONSES DERIVED FROM POWER
SPECTRUM ANALYSES (SURGE INVESTIGATION).
PITCH
0/
TECHNIQUE TECHNIQUE L.q.\.__TECHNIQUE
'\ \ I 2o
I
v, 3r(
_./,.\
4 I-e'
RADIANS/SECOND I I'O I 12.
14-
D-.
--.' tHEAVE
TECHNIQUE TECHNIQUE TECHNIQUE I. 2RADIANS/SECOND
T
TECHNIQUE 2 TECHNIQUE 3
-340
-320
3.0
-280
-260
2-220
4
--- 360
-300
-280
4PITCH
REGULAR WAVE DATA
RADIANS/SECOND
I I I'O I 12
- 340
REGULAR WAVE
DATA--320
HEAVE
TECHNIQUE TECHNIQUEA
TECHNIQWe, RADIANS/SECOND
I :1 10 1 l'2FIGURE 3
TECHNIQUE IPHASE ANGLES
DERIVED
FROM
CROSS
.2
FIGURE 4
COHERENCES (SURGE INVESTIGATION)
.
WAVEPITCH
I A/
/
- ...
I
.\
I ..'. : I . 1.09'.
.8 IStFIb,.
4&
.7-
C.)..:
___
/
A:
.. ... 0 .5 . TECHNIQUE TECHNIQUE TECHNIQUE I 2 33-L:J
.2-
.1-4
W,
6.1
8 IRADIANS/SECOND
10112
14w2V
D WIRE SPACING
6.0 FT.
VSPEED
2.92 FT./SEC.
TRECORD SHIFT
1.1 SEC.9
(ABSOLUTE VALUE)
ESTI MATED
RANGE OF INTEREST
,ppr
We, RAD.ISEC.
8
12 1620
I 1FIGURE 5
(I)
w
I-U)z
w -J4
I-C) w 0 U)0
C-) U) w I-U)z
Lii4
a
-+15
-+10
CO SPECTRA ORIGINALDATA\\
24
I I I I I I I I I I I I I I 1 I I I I\i
DATA PHASE SHIFTE I I I 10 12 I I1'
I
I I I/
' I ' 'AHEAD SEA" SPEED 2.92
-Ft/SEC. WIRE AHEAD
CONSIDERED AS "INPUT -WIRE FOLLOWING AS
"OUTPUT' WIRE SPACING
6FT., LtO.IO SEC., 40
-LAGS.I
/
We , RADIANS/SECOND 16 1820 22 24 26
I I I I I QUAD SPECTRADATA PHASE SHIFTED
,'-ORIGINAL DATA
I I I iI!'
/.
\ /
We, RADIANS/SECOND 68
tO(2 '/14
. 16 1820. 22
24 26
H ii.
.1 i i.. I,,I
FIGURE6
..I
SHIFTED DATA
WAVE ENERGY>20% OF MAXIMUM
Ik
WAVE ENERGY>I0% OF MAXIMUM
-I-.
0e RADIANS/SECOND
UNSHIFTED \ I DATA I F'jFIGURE 7
COHERENC.IES FROM
ANALYSES OF SHIFTED
AND
UNSHIFTED
DATA.
RUN 0106
HEAD SEA SPEED 2.92 FT./SEC., WAVE WIRE SPACING
6 FEET. WAVE WIRE AHEAD CONSIDERED
THE INPUT,.
THE WIRE FOLLOWING, THE OUTPUT. M = 0.10 SEC.
4o LAGS.
-1.0
-.9
-.8
-.7
C.)z
LU-.6
LU=
0
C)-.5
-.4
-.2
8 12 1620
24
I I I I30
-25
-20
-'5
6I
I AIt
FOLLOWING WIRE LEADING WIREAVERAGE OF RUNS 0106-0108
WAVE WIRES 6 FT. APART
"HEAD SEA" SPEED:
2.92 FT/SEC.
t= 0.1 SEC., 40 LAGS
DATA SHIFTED 1.1 SEC.
TUKEY "DEGREES OF
FREEDOM." ABOUT 24
We, RADIANS/SECOND
8 10 12 14 16 1820 22
I I I I I I24 26
FIGURE 8
WAVE SPECTRA: DUAL WAVE
WIRE
EXPERIMENT
COHERENCY
0.9---
0.8-0.7
+40 DEVIATIONS OF COMPUTED ARGUMENT FROM
THEORY-+30
95% CONFIDENCE BOUNDS+20
-+10
1.00.9--RANGE OF SIGNIFICANT
WAVE ENERGY We, RADIANS/SECOND I 12 I'620
1 S.'.--,A
MODULUS/
/
FROM' RATIO OF SPECTRA-CROSS SPECTRA THEORY
'I'
\__j
\..,. 95% CONFIDENCE BOUNDS (GOODMAN) .0.8--I I I I I4
8 12 1620
FIGURE 9
COHERENCY AND DERIVED TRANSFER FUNCTIONS
DUAL WAVE WIRE EXPERIMENT:
AVERAGE OF 3 RUNS, "HEAD SEA" SPEED 2.92 FT/SEC., M:
0.1 SEC., 40 LAGS, DATA SHIFT, 1.1 SEC., RUNS 0106-08, WAVE
WIRES 6 FT. APART.
We, RADIANS/SECOND
Copies
1 Admiralty Experiment Works
Haslar, England
Attn: H. F. Lafft, Esq.
1 Engineering Index, Inc.
29 West 39th Street New York, New York
Attn: Editor
8 Bureau of Ships
Department of the Navy
Washington 25, D.C. Chief
Attn: Technical Info. Branoh (Code 335) 3
Code 106
Preliminary Design (Code 420) 2
Hull Design (Code 440)
Scientific and Research(Code 442) 1
1 Bethlehem Steel Company
Shipbuilding Division
Central Technical Division Quincy 69, Massachusetts
Attn: Mr. Hollinshead DeLuce
1 Bureau of Naval Weapons
Department of the Navy
Washington 25, D.C.
Attn: Chief
2 Canal De Esperiencias Hldrodinamicas El Pardo (Madrid), Spain
Attn: Sr. M. Acevedo, Director
Applied Mechanics Reviews
Southwest Research Institute 8500 Culebra Road
San Antonia 6, Texas
Attn: Editor
50 David Taylor Model Basin
Washington 25, D.C.
Commanding Officer and Director
Attn: Code 513 5 Department of the Navy
Washington 25, D.C.
Chief of Naval Research
Attn: Code 438
Code 466
2 Gibbs and Cox, Inc. 21 West Street
New York 6, New York
Attn: Mr. W. F. Gibbs Mr. W. Bachman
1 Hamburgische Schiffbau Versuchsanstalt Bramfelder Strasse 164
Hamburg 33, West Germany
Attn: Dr. 0. Grim
2 Hamburgiache Schiffbau Versuchsanstalt Bramfelder Strasse 164
Hamburg 33, Germany
Attn: Dr. H. W. Lerbs, Director
1 Hydrodynamics Laboratory
California Institute of Technology Pasadena 4, CalifornIa
Attn: Director
1 Naval Warfare Analysis Group Operations Evaluation Group The Pentagon
Washington 25, D.C.
Attn: Dr. Manley St. Denis, Chairman
1 Institute of Engineering Research University of California
Berkeley 4, California
Attn: Prof. H. A. Schade, Director
DISTRIBUTION LIST
Copies
1 Institute of Engineering Research
Department of Engineering University of California Berkeley 4, California
Attn: Dr. J. V. Wehausen
1 Institute Najionale per Studied Esperienze
de Architettura Nava].e
Via della Vasca Navale 89
Roma-Sede, Italy
1. Kryloff Shipbuilding Research InBtitute
Leningrad, U.S.S.R.
Attn: Dr. A. I. Voznessensky
1 Maritime Administration
Engineering Specification Branch Main Propulsion Section
Washington 25, D.C.
Attn: Mr. Ceasar Tangerini, Head
3 Massachusetts Institute of Technology
Department of Naval Architecture and
Marine Engineering
Cambridge 39, Massachusetts
Attn: Prof. M. A. Abkowitz Prof. G. C. Manning Prof. P. Mandel
2 National Physical Laboratory Teddington, Middlesex
England
Attn: Dr. A. Silverleaf
Dr. F. H. Todd, Director
1 Ned. Scheepsbouwkundig Proefatation
Wageningen, Netherlands
Attn: Dr. Ir. 0. Vossers
1 Ned. Scheepsbbuwkundig Proefstation
Haagsteeg 2
Wageningen, The Netherlands
Attn: Dr. Ir. J. D. Van Manen
1 Newport News Shipbuilding and Dry Dock Co.
Engineering Technical Department Newport News, Virginia
Attn: Mr. John Kane
1 New York University
College of Engineering
University Heights New York 53, New York
Attn: Dr. W. J. Pierson, Jr. Oceanics, Inc.
ilk East 40th Street New York 16, N.Y.
Attn: Mr. Wilbur Marks
1 Office of Naval Research
Branch Office 1000 Geary Street
San Francisco 9, California
Attn: Commanding Officer
Office of Naval Research Branch Office
346 Broadway
New York 13, New York
Attn: Commanding Officer Office of Naval Research
Branch Office Navy 100, F.P.0.
New York, New York
Attn: Commanding Officer Office of Naval Research Branch Office
495 Sumner Street
Boston 10, Massachusetts
Copies
1 Office of Naval Research Branch Office
The John Crerar Library Building 10th Floor
86 East Randolph Street Chicago 1, Illinois
Attn: Commanding Officer
1 Office of Naval Research Branch Office
1030 East Green Street Pasadena 1, California
Attn: Commanding Officer
1 Office of Ship Construction Maritime Administration Washington 25, D.C.
Attn: Mr. V. L. Russo, Deputy Chief
2 Ordnance Research Laboratory Pennsylvania State University P.0 Box 30
University Park, Pennsylvania
Attn: Dr. G. Wislicenus, Director
Mr. McCormac
]. Prof. B V. Korvin-Kroukovsky
P.O. Box 2k7
East Randolph, Vermont
1 Reed Research, Inc. 10kB Potomac Street, N.W. Washington 7, D.C.
Attn: Mr. S. Reed
1 Ship Hydrodynamics Laboratory Ship Division
National Physical Laboratory Faggs Road
Feltham, Middlesex, England
Attn: G. J. Goodrich, Eaq.
1 Skipsmodelltaken
Norges Telmiske Hogakole
Trondheim, Norway
Attn: Dr. J. Lunde, Director
1 Society of Naval Architects and Marine Eng. 7k Trinity Place
New York, New York
St. Anthony Falls Laboratory University of Minnesota
Minneapolis, Minnesota
Attn: Dr. L. G. Straub, Director
1 National Council for Industrial Research,TNO
12 Koningekade P.O. Box 29k
The Hauge, Netherlands
1. Hydro-og Aerodynamisk Laboratorluin Hjortekaersvej 99
Lyngby, Copenhagen
Denmark
Attn: Prof. C. W. Prohaska
1 Ir. J Gerritsma
Supt. Shipbuilding Laboratory
Delft Technological University
Prof. Mekeiweg 2 Delft (Netherlands)
DISTRIBUTION LIST
Copies
1 Transportation Technical Research Institute
Ministry of Transportation
1-1057 Mejiro-Cho, Toshima-Ku
Tokyo, Japan
Attn: Dr. Shiro Kan
1 Universitaet Hamburg Lammeraleth 90 Hamburg 33, Germany
Attn: Prof. 0. P. Weinblum, Director
1 University of California
College of Engineering
Berkeley k, California
Attn: Mr. 0. J. Sibul
1 University of Michigan
Department of Naval Architecture and
Marine Engineering. Ann Arbor, Michigan
Attn: Prof. R. B. Couch, Chairman
1 University of Notre Dane
Department of Engineering Mechanics
Notre Dame, Indiana
Attn: Dr. A. 0. Strandhagen, Head
2 U.S. Naval Ordnance Laboratory
White Oak
Silver Spring, Maryland
Commander
Attn: Library
3 U.S. Naval Ordnance Test Station Pasadena Annex
3202 East Foothill Blvd.
Pasadena, California
Attn: Commander
Technical Library
Head, Thrust Producer Section
1 U.S. Naval Ordnance Test Station Propulsion Division (P8063) 125 S. Grand Avenue
Pasadena, California
Attn: Mr. Sam Elhai
1 U.S. Naval Research Laboratory (Code 2000) Washington 25, D.C.
Attn: Director
3 Webb Institute of Naval Architecture
Glen Cove, New York
Attn: Administrator
Post Graduate School for Officers Prof. E. V. Lewia
1 Division of Mechanical Engineering National Research Council
Ottawa, Canada
1 Director
Statens Skeppsproviningsanstalt Glbraltargatan 50
Goteborg 2kC, Sweden
1 Bassin d'Essais des Carenes 6 Boulevard Victor