-NOV. 1982
ARCH1F
SYMPOSIUM ON
"HYDRODYNAMICS OF SHIP AND OFFSHORE PROPULSION SYSTEMS"
HØVI K OUTSIDE OSLO, MARCH 20. - 25., 1977
"MEASUREMENTS OF THE RADIATED NOISE FROM VARIOUS FORMS OF LIFTING SURFACE CAVITATION"
By Steven J. Barker
University of California, Los Angeles
Ref.: PAPER 2/3 - SESSION 4
Lab. v.
ScheepsbouwkundeTechnische Hogeschool
'.jIiotheek
van deMeasurements of the Radiated Noi eAfcheepsb
'-vrthijnde
ICe
Various Forms of Lifting Surface C
V TEDATUM!
I
ABSTRACT
This paper describes the results of a regent series
of noise
measurements in the CaltechFiigh-Speed
Water Tunnel.
Model
hydro-foils of 15 cm span were mounted in the rectangular
test section of
the tunnel and run atvelocities from 600. to
1,500cm/sec.
The
static pressure was varied independently of the velocity to
produce
a wide range of cavitation parameter.
.Radiate.d noise was measured.
by a hydrophone mounted flush with the tunnel wall.
Cavitation noise
spectra were measured in the frequency range
from 200 Hz to 160 kHz.
Three distinct types of cavitation were studied: partial or
"blade surface" cavitation, full cavitation, and tip.vortex
cavita-tion.
For partial cavitation, vibrations of the hydrofoil mo4el
exéited by the cavity collapsearefound tobe an important source
of noiSe.
In full cavitatiOn this is not the case, although the
broad .band.nOise intensity is apparently greater.
Both partial and
full cavitation produce a strong spectral peak at a Strouhal
num-bér based upon cavity length of 0.5.
Noise from trailing vortex
cavities is found to be of. lower intensity than that of the
other
forms.
This is due to the fact that the vortex cavitation as i.t
occurs in the water tunnel is gaseous rather
than vaporous. .
Thevortex cavities do not show the same behavior with cavitation
parameter as the other forms.
Submitted by: Steven J. Barker, Assistant Professor
University of California
CONTENTS
Page
Introduction i
Experimental Geometry 5
Blade Surface Cavitation 8
Wake Cavitation 15
Trailing Vortex Cavitation 17
Summary and.Conclusions 19
REFERENCES 22
MEASUREMENTS OF RADIATED NOISE FROM VARIOUS FORMS OF LIFTING SURFACE CAVITATION
1.. Introduction
Cavitation on lifting surfaces in the ocean is often unavoidable,
particuiar].yin.high speed hydrofoils and screw propellors. The
noise generated by cavitation has thus been a subject of interest
for manyyears. However,, efforts to predict this noise fom. theory
have met with little success, and laboratory measurements of the
noise have been.incoinplete and unsatisfactory. Lifting surface
'cavitation occurs in several forms,' and in this study noise from
,eaóh of these fors. was measured independently. The measurements
of the study were made in a water tunnel that is unique in its low.
background noise level. . .
Cavitation produced noise is almost always of much greater
intensity than fully wetted turbulent flow noise at the same flow
velocity. The reason for the greater intensity is as follows.
Lighthill [1] showed that the momentum equation in ,a single-phase
flow, can. be written as . . .
4 -.
aVp
=
_
_ T .
,:
where the tensor T1 'is defined by
= PU1U - pj - (2)
In these equations, u is the i-component of the instantaneous
fluid velocity, p is the fluid density, is the stress tensor,
a0 is the local iseitropic speed of sound, and
i.j is the kroneker
momentum equation as a wave equation for the density with the
double divergence of the tensor T1 appearing as a forcing
fimc-tion on the right hand side. called the "fluctuating
Reynolds stress tensor," is quadrupole in nature since its
second derivative appears in the wave equation. The acoustic
quadrupole is. a relatively inefficient means .of converting
kinetic energy to sound. Most of the energy goes into near field
terms, representing localized reciprocating motions wh.ich do. not
propagate to infinity. To use Lighthili's example1 if a sphere
vibrates in a quadrupole fashion such that the acoustic
wave-length is twice the circumference, only 1/1000 of the energy of
the vibration is radiated as sound. For a corresponding dipole
motion of the sphere, 1/13 of the energy is radiated as sound.
The application of Lighthill's
analogy
to boundary layer flowshasbeen investigated by Curle [2] an4 Phillip.s [3].
For the case of cavitating flow, the: continuity equation.
can be written as . ..
pu. zh(x,t) . (3)
where th(x1t) is a distribution of mass sources representing the
rate of change in volume of cavitation bubbles. This would add
another .term to the right hand side of equation (1), namely
/at.
This term is a simple acoustic. SOurce, which is much more efficient
than either the dipole or quadrupole. Because of its greater
acous-tic efficiency, weexpec.t.a cavitating flow to radiate sound at
a higher. intensity than a fully. wetted flow of the same energy
Each of these different acoustic source. types will produce a
different dependence of. the noise intensity upon flow veloäity.
Lighthiil showed that the acoustic quadrupoles of his equation
should yield a dependence of noise intensity upOn U8. Similarly,
acoustic dipoles will result in a U6 dependence and simple sources
a U dependence. The assumptions made. in reaching these
conclu-sions are that the amplitude of the fOrcing function scales with
pU2 and that the frequencies present in the turbulent flow scale
with U. Previous experimenters have compared the behavior of
various frequency bands in a measured noise spectrum with these laws and made inferences about dipole, quadrupole, and even
octu-pole radiation. However, the velocity dependence predict-ions for
the different sources are for the total noise intensity and not
for the spectrum level at a particular frequency. Since these
predictions were made under the assumption that frequencies
pre-sent in the flow scale with velocity, the velocity dependence of
the spectrum level cannot be interpreted, in this way..
Cavitation on bodies moving through water. appears in Several
forms, and these may generate noise by different mechanisms. For
the purposes of the present experiment we shall discusS blade
sur-face cavitation, wake cavitation, and tip vortex cavitation.
Blade surface cavitation, also known. as "partial. cavitation,"
consists of a vapor cavity which begins near the leading edge of the suction side of a hydrofoil and collapses on the surface
of the foil somewhere further aft. Blade surface cavitation may
travelling bubble cavitation, the cavity is a mixture of liquid
and small vapor bubbles which change rapidly in size. Such a
cavity has a cloudy or diffuse appearance. On the other hand, a
stationary cavity has a well defined interface with a glassy
ap-pearance, within which the pressure is essentially vapor pressure.
Wake cavitation, also referred to as "full" cavitation,
occurs when the cavity collapses downstream of the trailing edge
of the mode]... The cavities generally have the glassy appearance
of a stationary cavity near their upstream end.,, changing to the
frothy look of travelling bubble cavitation further aft. Between
partial. and full cavitation there. is a region of 11unsteady"
cavi-tation,.'in which thecavity collapse occurs very near the trailing
edge of' the model.. In this case. the cavity fluctuates rapidly
in length, causing the forces on the foil to vary by more than
100% Any type of measurement is difficult to make in this
re-gime, and none were made in these tests.
The trailing vortex wake of a lifting three 'dimensional
hydro-fOil has .a low pressure core which can cavitate. Trailing vortex
cavities occur not only at the erds of hydrofoils, but more
conmion-ly at the tips of ,screw propellers. These cavities usually have a
glassy appearance, and are found to behave very differently from
the other forms mentioned above. They will be discussed below in
Section 5. . '
It might be. asked why further measurements of cavitation
,rioise are important, since there have been several such
proper-ties of the CaItech High-Speed Water Tunnel, which will be
brief-ly described below. In addition, past measurements have not
characterized the noise from the different forms of cavitation
described above. SOme of the previous experiments have employed
rotating blades as' the Source of cavitation [4], [5].. Such
mea-surernents are of limited use since the cavitation parameter sigma
is not constant along the length of the blade. '(Sigma is defined
by a=(p_P)/½Pu2, where pis free stream pressure,
PC is cavity
pressure, and U is free stream velocity.) There have also been
other water
tunnel
measurements [6], [7], but these differgreat-ly from the present study.
2. Experimental Geometry
2.1 High-Speed Water Tunnel.
These experiments were conducted in the Caltech 'High-Speed
Water Tunnel, which is deScribed in detail in Reference [8]. This
tunnel has two features that are very Important to the measurement
of cavitation noise. One is its extremely low background noise
level, which makes it possible to measure radiated noise from
even fully wetted boundary layers under. some conditions. The
higher intensity of cavitation noise is far above the background.
The other relevant feature of th tunnel is the. "resorber" section.
The resorber' is a vertical Oylinder 12 ft . in diameter and 58 ft
high whose top is located about 20 ft below the test section.
The volume of the resorber is approximately 45,000 gallons. Water
leaving the test section passes first through the drive pump and
then through the .resorber. The time required for the. water to pass
at its bottom causes complete re-absorption of entrained air
bubbles generated by cavitating flow in the test Section.
The test section of the tunnel is rectangular; 6 inches x 30
inches in cross section and 48 inches in length. The maximum
velocity is a.bout 80 ft/sec and the static pressure can be varied
independently of velocity from vapor pressure to about three
at-mospherés. .. . .
2.2 Transducer Geometry
The chief difficulty in measuring radiated noise in ,a steady
state flow is to distinguish radiated sound from the wall
pres-sure fluctuations (also called "self noise" or "pseudo-Sound")
created by the turbulent boundary layer over the face of the
transducer itself.. This was done in the past in this tunnel by
placing a focusing acoustic reflector outside of the test
sec-tiôn and locating the. transducer at its focus [9). However the
test section wall through which the sound must pass before
reach-ing this reflector is a very poor acoustic window. . The
config-uration . used in the present study employs a thin diaphragm
mounted.
flush with the tunnel wall, with a flooded cavity behind it (See
Figure 1).. A piezoelectric transducer is located at the back of
this cavity. The diaphragm transmits nearly all sound into the
cavity, but has .the effect of averaging the boundary layer wall
pressure fluctuations over its full area. (The diaphragm is 1½
inches in diameter, which is much larger than the tunnel wall
boundary layer thickness.) Thus the wall pressure fluctuations
detailed
explanation
of thismechanism, see Reference (10].) The lowest resonant frequency of the flooded transducercavity is 25 kHz, but prominent peaks in the noise spectrum are
not observed at this frequency or any. of its harmonics. A more
likely source of reverberation is the 6 inch spacing between the
two largest parallel walls, of the test section. The fundamental
frequency of this
standing
wave is between 4 and 5 kHz, where 'amild peak is observed in the clear tunnel spectrum. Due to 'the
reverberant nature of the test section, the absolute measured
acoustic lee1s may not be' significant. However, the goal of
these tests is to establish trends of, the noise spectrum,
in-cluding its variation with velocity, cavitation number, and model
configuration. ' .
The transducer output is coupled to a .high input impedance
preamplifier, and then to a pre-whitening amplifier. The
out-put of this is fed into a Bruel and Kjaer Audio Spectrometer,
which reads out 1/3 octave band levels for bands whose center
frequencies range from 100 Hz to 160 kHz. The band levels are
converted to spectrum levels' in decibels referred to I dyne per
square centimeter, and the resulting spectra are plotted by
computer. To establish the background noiSe level of the
tun-nel, power spectra were measured in the clear tunnel
configura-tion and are shown in Figure 2. Spectra are plotted for test
section velocities of 20, 25, 30, 35, 40, 45, and 50 ft/sec.
The clear tunnel Spectra are independent of the static pressure
.3. Blade Surface Cavitation
Noise measurements on blade surface cavitation were made
over a wide range of sigma and angle of attack. Test section
velocity was varied from 20 to 50 ft/sec. One of the hydrofoil
models used was an NACA 66-210 section with a six inch chord.
The model spanned the six inch width of the test section, so that
the mean flow was nearly. two dimensional. The model was made of
aluminum, and its surface was painted and polished to a mirror
finish. A xnicroflash photo of blade surface cavitation on this
model is
shown.in
Figure 3.Let us consider first the effect Of varying sigma while
holding velocity, and angle of attack constant. (The cavity
pres-Sure is assumed to equal vapor pressure in all measured values
of siqta.) Figure. 4 shows four spectra measured at eight degrees
angle of attack and 20 ft/sec tunnel velocity. In the lowest
spectrum, labeled sigma co the pressure is high enough that no
Oavitation occurs; this establishes, a background level. In the
spectrum for sigma = 6.6 there is no visible cavitation, but the
spectrum level for frequencies above 10 kHz has increased by.
between. 2 and 7 db.. This increase at high frequencies is not
detectable to the ear. There is still no visible cavitation at
sigma = 4.9, but the spectrum level at the highest frequencies
has been increased by about 40 'db. The low, frequency part of
the spectrum remains unchanged, and the change in sound is. barel.y
audible. At sigma = 4.2 the cavity is clearly visible and. about
48 db, but the changes at the low frequencies are still relatively
small.
Figure 5 shows the fully wetted spectrum again plus two spectra
for still, lower values of sigma. At sigma = 3.5 the cavity length
is 2. mm and the significant increases in spectrum level now extend
to frequencies belOw 1 kHz. For the lowest sigma of 2.2, the
cavity is 30 mm long and the spectrum level increases extend down
to 300 Hz. Beginning with sigma = 3.5, the spectrum level at the
highest frequencies decreases as the pressure is lowered. The 30
mm long cavity actually produces less noise at 100 kHz than the
invisible incipient cavitation at sigma = 4.9. All of these
vapor cavities have, the cloudy. or' diffuse appearance of travelling
bubble cavitation, as seen in Figure 3.
For the first visible.cavitatjon (sigma =.4.2) we can form 'a
frequency based upon the flow velocity, and the observed cavity
length' of 1 mm. This frequency is approximately 6 .kHz. At. sigma =
4.2 there is very little increase in, the radiated sound (relative
to sigma = ) for frequencies below 6 kHz, and it is near this
frequency that. the spectrum begins to deviate from that of the
fully wetted flow. This suggests that, at. least for incipient
cavitation, the sound may be produced by acoustic sources created
by the growth and collapse of very small bubbles passing through
the region of cavitation. For larger cavities we shall Se below
that another mechaniam becomes important:. the forces exerted by
Note that the spectra in Figures 4 and 5 are of dissimilar
shapes. Each spectrum has prominent peaks occurring at different
frequencies. The general trend is that as sigma decreases, energy
content shifts to lower frequencies and the broad band nOise level
increases.
We now consider noise spectra measured for a model
configura-tion run at different speeds for a fixed value of, sigma. Speed
is. varied over the range from 20 to 40 ft/sec in these runs, and
sigma is held constant by raising the static pressure as speed
is increased. The size and appearance of the cavity are observed
to be constant over this speed range. Two very interesting
char-acteristics can be seen in these spectra. One of theseis shown
in Figure 6, which is a plot of spectra for 20, 25, 30, 35., and
40 ft/Sec at sigma 2.46. Here the cavity forms near the leading
edge of the mOdel and is about 5 mm long. The spectra for the five
different speeds are remarkably similar in shape, with peaks occur-i
ring at the same frequencies over the entire speed range. Figure
7. shows another series of spectra for sigma =. 2.15; the cavity
length is. .8 mm. Again the spectra have the. same shape at all
speeds, with the major peaks occurring at the same frequencies.
However, the shapes of the spectra in Figure 7 are quite
differ-ent from those of. Figure 6.
The similarity in the Spectral "signature" at different speeds
is not
consistent
with the concept of the noise being produceddirectly by the
expansion
and collapse of cavity bubbles. For ascale with the velocity. The fact that. they do not in the above
cases suggests the following. A well developed (as opposed to
incipient) blade surface cavity will exert large fluctuating
forces on the surface of the hydrofoil in the region of collapse.
These fOrces can be expected to excite the normal modes of
oscil-lation of the foil itself. The resulting vibrations of the
hydro-foil would then be a very efficient noise generating mechanism.
The frequencies of these vibrational modes are iüdependent of
the flow velocity, hence the similarity in spectral signatures
at all speeds. Increasin.g the speed only increaSes the energy
being 'fed into these vibrations, so that the whole spectrum moves
upward rather uniformly.
bn
the other hand, a change in sigma willchange the Spectral, signature because for a different cavity length
the fluctuating forces are being applied on a different part of the
foil. This would tend to excite different modes,: or tO excite the
same modes in different propgrtions.
Further evidence ui support of this
explanation
of thecon-stant spectral. Signature can be obtained. Measurements
on two
models, of identical shape but very different vibrational
charac-teristics Should yield different results for the same flow
condi-tions.' An effective way to change the vibrational characteristics
of a model is.to vary the
mounting
configuration. Hydrofoilmodels in the water tunnel are normally cantilevered from the
test section wall 'opposite the waIl in which the transducer is
located. A plug was constructed to f it into the latter wall Such
this way the same model could be studied when cantilever mounted
or when fixed at both ends. Spectra for the two different
con-figurations are different in shape, Supporting the hypothesiS
that cavitation-induced foil vibration is a mechanism of noise
production.
The variouS sets of spectra discuSSed. above were chosen
be-cause they exhibited the characteristic of constant spectral
sig-nature In many other spectra we observe: peaks which Shift to a
higher frequency as velocity is increased for
constant
sigma. Thisis the behavior we would expect for the mechanism of direct
radia-tion from the expansion an collapse of cavitation bubbles. The
spectra of Figure 8 are for a piano-convex hydrofoil with a four
inch chord at a. five degree angle of attack. Sigma is 1.74. and
the velocities shown are 20, 25, 30, 351 and 40 ft/sec. The
most prominent peak
in
these spectra occurs at 500 Hz for 20 ft/secand 1,000 Hz for 40 ft/sec. We can form a Strouhal number based
on the observed cavity length (L) of 6 nun, the free stream
velo-city. U, and the frequency
n of
the spectral peak. The resultingvalues of nL/U for the four different velocities are 0.52, 0.53,
0.54, .0.48, and 0.52. These values are constant to within the
resolution of. the spectrum analyzer, and they are all very
close to 0.5. This indicates that the time required for a
cavita-tion bubble to traverse the length of the cavity is one half the
period of the fundamental noise frequency.
Similar data for a lower sigma value of 1.54 are shown in
Figure 9. In this case the cavity length is 10 mm and the
0.50, 0.47, 0.45, and 0.49. Thus the Strouhal number is again
about 0.5, in this case for a much larger cavity. Data for a
very small cavity, L = 3 xtn, are shown in Figue 10. Here the
primary peak in the spectrum occurs between 1 and 2 kHz.. The
Strouhal numbers are 0.50, 0.50, 0.53, 0.46, and 0.50. In these
spectra we also see two prominent peaks whose freqi.iencies do not
vary with Speed, at 5 kHz and. 31.5 kHz. The appearance of both
constant Strouhal number and cOnstant frequency peaks is seen
in much of the data. The former generally occur at lower
fre-quencies than the latter., The constant Strouhal number. peaks do
not exhibit strong harmonics, while the conStant frequency peaks
often. do. .
Noise measurements made for incipient blade surface
cavita-tion show some very different properties. Incipient cavitation
refer. to flow at a value of sigma such that cavitation is just
beginning, and the vapor cavity or bubbles may not be visible: at
all. The :incipient cavitation nimiber a1 refers to the value of
sigma at which cavitation begins as sigma is lowered. The
"desi-nent" cavitation number ad is the value at which cavitation ceases
as sigma .is increased. This. is generally slightly greater than.
the incipient value. Both a1 and ad are found to be weak functions
of velocity, increasing with the velocity in most. cases. Measured
values for a particular model configuration depend on the facility
in which the measurements are made, and the Caltech water tunnel
yields lower values of a than most other tunnels [11]. This is
described above. There is some ambiguity in the definition of
inq.pient cavitation number, since there is' no universally
ac-cepted criterIon for what defines the onset of cavitation.
As was Shown in Figure 4, incipient cavitation is first
detected acoustically, at very high frequencies.. High intensity
noise will be radiated at frequencies above the audible band
before there is any visual indication of cavitatIon. Figure 11
shows spectra at four different speeds for a cavitation number of
305, which is slightly below the incipient value. These
àpec-tra'have two interesting characterist-ics: (1) at frequencIes
above 1 kHz, the spectra are relatively flat, that is, the noise
is nearly "white." and. (2) at frequencies abov 6 kHz a change
in flow velocity has very little effect on the spectrum. The
flatness of the high frequency part of the spectrum indicates
that mOdel vibrations are not important in this case. The lack
of dependence of spectrum level upon velocity is difficult to
understand. Even for simple source radiation, the total energy
radiated should vary as U4.
The explanation for this anomalous behavior of the spectrum
level may lie in the fact that all velocity dependence predictions
from Lighthill's equation are for the total sound intensity and
not for the spectrum level at a particular frequency. As stated
above., an assumption in these predictions is that the frequencies
present in the flow scale with the flow velocity. The effect
that. this scaling has upon a particular part of the spectrum
by Figure 12.
This figure. shows spectra for incipient
cavita-tion on a different. model, and has the same characteristic region
of velocity independence above 10 kHz.
However., at 100kHz the
spectra begin to diverge rapidly and have developed a 13 db spread
at the highest frequency of 160 kHz.
This behavior again points
out the errors that can be made in drawing concluSions from the
velocity dependence of the spectrum level at a particular frequency.
4.
Wake Cavitation
If the cavitation niznber is low enough in flow over the models'
described above, the cavity length will exceed the ôhord length and
the collapse of,the cavity will occur downstream of the trailing
edge.
.The cavitation then changes from partial or.blade surface
cavitation to full cavitation.
.In the case of certain
bluff-based shapes, the collapse of the cavity occurs downstream of the
trailing edge for all values of sigma.
We now consider the noise
radiated from a model of this type.
The model used for these stud-.
ies was a 7.5° wedge with a six inch span and six inch chord,
run
at zero angle. of 'attack with the sharp edge pointing upstream.
The base or trailing edge of. this wedge was 0.78 inches thick.
Vapor cavities formed with this model have a different
appear-ance from the blade surface cavities discissd above.
Here the
cavity has a more vagtie outline; its boundaries are not well
defined, as seen in Figure 13.
.The reasoh for this appearance
can be seen in microf'lash photographs, such as Figure 14..
The
cavitation is of the travelling bubble type, and the bubbles
occupy a relatively low volume fraction of the cavity region.
Adis-tribution suggests the shedding of von Karxnan vortices from the
base. Globular masSes of bubbles can be seen in the cores of
these vortices (Figure 14). The Reynolds ntirnber based upon the
thickness of the model is about 200,000.
Figure 15 shows a series of spectra measured with the wedge
for a sigma of 2.27. Several features. of these spectra. should be
pointed out. (1) The broad band noise level for this form of
cavitation is 5 'to 10 db higher than for a blade 'surface .cavity
of 'the same
dimensions
and at the same velocity. (2) Much higherlevels are seen in the 100. to 500 Hz band than for blade surface
cavitation. (3) The spectra are "whiter" at the higher frequencies
and there are no constant frequency peaks, indicating that model
vibration is not a factor. The average downward slope of the
spectrum is about 3 dbpr octave, which is a (frequency)
be-havior of spectrum level. (4) The prominent peak occurring
be-twèen 100 and 200 Hz has constant Strouhal number. The value of
nL/U based on the best estimate of the cavity length L is 0.5, th
same value found for blade surface cavitation.
Figure 16 shows results from the same model at a lower value
of sigma, 1.79. The appearance of tIe cavity is.much more dense
than in the higher sigma case, and photographs show .a higher
den-sity of bubbles distributed over the sameregion of space. The
visible length of the. cavity is about the same. As shown in
Figure 16, there is only a slight change in the radiated noise
spectrum. The broad band level is about
I
db higher for thelower sigma, and there is a very slight shift of'energy toward
It is a
well known
fact that "ventilating" a base cavity by bleeding atmospheric air into it has some effect upon thenoise.. Figure 17 shows this effect quantitatively for the wedge
model at a sigma of 1.79. We see that the effect of vent-ilation
at ow frequencies, below 300 HZ, is not very significant. At
the high frequency end of the spectrum, noise reductions on the
order of 24 db result from ventilation. The downward slopE of
the spectrum rezfla-ins uniform with ventilation, but is increased
.frorn about 3 db per. octave, to 5 db per octave. The reduction .
in the broad band noise, level is from 4 to, 7 db, depending on
the Speed. .
5. -Trailing Vortex Cavitation
- Trailing vortex cavitation is observed to occur both on
hydro-foils and submerged screw propellers.. A trailing vortex wake is.
formed behind any three dimensional lifting surface, and the
dy-namics of these vortices have, been studied extensively, both in
theory. [12W) and experiment [13]. The Static pressure distribution
across a trailing vortex has a minimum at the center of the
vor-tex core, and this presSure minimum can be low enough to produce
cavitation.
Trailing vortex cavitation was produced in the Caltech water
tunnel in the following way. An MACA 4412 section hydrofoil with
a three. inch span and three inch chord was mounted on the test
section wall. The moel planform extended halfway across the six
inch width of the test eotion. This model developed sufficient
cay-ity at Static pressures high enough that no blade surface
cavita-tion occurred, as can be seen in Figure 18. Thus the.noise
pro-duced by the trailing vortex cavity could. be measured in a flOw
with no other significant noise sources.
The. characteristics Of the vortex cavities produced in this
study run completely counter to those of the other forms of
cavi-taton.
The most significant difference is the failure oftrail-ing vortex cavities to scale with sigma. In the previous cases,
the size, and appearance of the vapor öavity depended only on sigma,
and were nearly independent of velocity or time. For, trailing
vortex cavities the size and shape do not depend solely on sigma,
nor are they constant in time.
When the trailing vortex cavity is first formed, it is
cylin-drical in shape and about ¼ inch in dianieter. The cavity may
be-gin right at the tip.ofthemodel, or at'higher pressures it may
begin a foot or more downstream of the model. If the cavity
be-gins downstream of the model, its upstream end. will oscillate back and forth in 'the streantwis.ê. direction over a distance of
several inches. The' cavities have no visible downstream end,
ex-tending far into the diffuser section of the water tunnel. The
surface of' a vortex cavity is glassy in appearance except at its
upstream end. If constant. flow conditions are maintained for
sev-eral minuteS, t'e' cavity gradually thickens and. develops a twisted
ribbon shape.. The width of this ribbon is about 3/4 inch and its
thickness about ¼ inch. The twisting wavelength of the ribbon is on
'the order of a foot and is inverSely proportional to the angle of
These properties of the trailing vortex cavities suggest that
they are largely gaseous rather than vaporous. Instead of being
composed entirely of water vapor, these cavities are mostly gases
which have been taken out of solution. The measured dissolved air
content
of the water tunnel is about. 10 parts per million, andvaries very little.
Noise measurements of the trailing vortex cavities support
the conclusion that they are gaseous. Figure 19 shows spectra
for vortex cavitation at a sigma of 1.70 and an angle of attack
of seven degrees. The signal to noise ratio is greater than one
only .at frequencies above 600 Hz. The trailing vortex cavities
produce much less noise at all frequencies than the other forms.
of cavitation. The downward slope of the spectrum is greater for
vortex cavitation; in the example of figure 19 it is 7 db per
oc-tave. The low noise levels are further evidence of gaseous
cavi-tation because gas bubbles do not undergo the violent collapse of
vaporous cavitation bubbles. If both blade durface and vortex
cavitation take place simultaneously in the water tunnel, the
vortex cavitation will contribute little to the noise spectrum.
The same will be true of vortex cavitation on full scale hydrofoils
and propellers if the cavities in those cases are also gaseous.
6. Suzmnary and Conclusions
Cavitation-produced flow noise is inherently more intense than
fully wetted boundary layer noise because the kinetic energy of the
flow is more.efficiently converted to acoustic energy. The ratio
The water tunnel is thus an effective means of studying this type
of flow noiSe. The major limitation of a water 'tunnel study is
that. the closed test sectiOn geometry makes it difficult to
mea-sure absolute acoustic levels. However, in these tests we have
been interested in relative changes in noise resulting from ôhanges
in flow parameters or model configuration.
in blade surface or partial cavitation two different mechanisms
of noise production were. Observed. One of these is direct radiation
from, the rapid expansion and collapse of vapor bubbles in the
cav-ity. This is characterized by spectral peaks which occur at a
constant Strouha]. number as velocity is varied with sigma held
constant. The Strouhal number based on cavity length of the low-.
est of these peaks is consistently 0.5. The second noise
genera-tion mechanism is vibragenera-tion of the hydrofoil model excited by the
collapse of the vapor cavity. This type of noise is
character-izedby spectral peaks occurring at constant frequency for all
velocities. Changing the structural"properties of the hydrofoil
model has a measurable effect on these spectral peaks.
Wake cavitation produces noise with a whiter spectral
distri-bution than that of blade surface cavitation. The total intensity
is higher for wake cavitation, and there is a high energy content
at very high frequencies. Node]. vibration does not appear to be
a factor in this type of cavitation. The cavitation of trailing
vortex wakes, as it occurs 'in the water tunnel, is apparently
gaseous in nature. The violent collapse characteristic of vapor
cavities does, not occur, and as a result the radiated noise levels
noise is detectable only at frequencies above 5 kHz.
In comparing the relative importance of various mechanisms of producing flow noise, it can be said that cavitation noise
will predominate over boundary layer radiated. noise under almost
any conditions. Incipient cavitation produces high intensities
at very high frequencies, so that in some frequency bands small
cavities will radiate more noise than large ones. Under partial
cavitation, foil vibrations are important and the structural
pro-perties of the hydrofoil will affect the radiated noise. Full
cavitation produces high levels throughout the measured spectrum,
but these can be reduced a great deal by ventilation, particularly
at high frequencies. Tip vortex caviation may not be an
Curie, N., 1955 dynamic Sound," Phillips, O.M., Plane Turbulent p. 327'. REFERENCES
.1. Lighthili., N.J., 1952 "On ound Generated Aerodynamically,"
Proc. Roy.Soc. London,A, 211, p. 564.
"The Influence of Solid Boundaries upon,
Aero-PrOc. ROy. Soc. London A,231, p. 505.
1956 "On the Aerodynamic Surface Sound from a
Boundary Layer," Proc. Roy. Soc. London A, 234,
Meilen, R.H., 1954 "Ultrasonic Spectrum Of CavitationNoise
in Water1" J. Acous. Soc. Am. 26. '
Ross, D. and' McCormick, B, 1948 "Propeller Blade-Surface Cavitation Noise," Penn. State College report NOrd 7958-115.
Song, S.C. and Silberman, E., 1961 "Experimental Studies of
Cavitation, Noise in a Free-Jet Tuine1,":St. Anthony F411$
Hydraulic Lab., U. of Minn., Tech, paper No. 33, series B.
Ripken, J.F.-and Killen, J.M., 1959 "A Study of: the' Influence
of Gas Nuclei on Scale Effects and Acoustic Noise for mci-pient Cavitation in a Water Tunnel," St. Anthony Fails
Hydrau-ho Lab., U. of Minn., Tech, paper No. 27, series B.
'8'. Knapp, R.T. and Levy, J., 1948 "The Hydrodynamics Laboratàry
of the California Institute of Technology," Trans. ASNE 70,
p. 437. '
'
Knapp, R.T., 1945 "Cavitation NoiSe from Undeqater Projectiles,"
California Institute of Technology, report No. 6.l-sr207-19]O'.
Corcos,G.lvl., 1963 "Resolution of PreSsure in Turbulence," J.
Acóus. Soc'.Axn. 35, p. 192.
il. Arakeri, V.H. and Acosta, A.J., 1974 "Cavitation Inception
Observations on Two Bodies at Supercritical Reynolds Numbers,"
Calif. Institute of Technology, Div. of Eng. report No. E183.2.
Moore, D.W. and Saffman, P .G., 1973 "Axial Flow in Laminar
Trailing VorticeS," Proc. Roy. ,Soc. London A, p. 491.
Baker, G.R., Barker, S.J., Bofah, K.K., andSaffman, P.G.,'1974
"Laser Anemometer Measurements of Trailing Vortices inWater,"
ACKNOWLEDGMENT
The author would like to acknowledge the
-24-List of Figures
1.. Transducer Geometry.
2. Clear Tunnel Noise Spectra:
0
20 ft/sec.;0
25 ft/sec;30 ft/sec;
V
35 ft/sec; 40 ft/sec; 45 ft/sec;50 ft/sec.
3 Hydrofoil blade Surface Cavitation.
NACA 66 Model
06.6;
4.9; . NACA 66 Model o 3.5;A
2.2. NACA 66 Model o 25;L 30;V
NACA 66 Modelat 20 ft/sec. Sigma.
= 0
fully wetted;V'4.2.
-at 20 ft/sec. Sigma
.0
fully wetted;at Sigma =2.46. U = 020 ft/sec.;
3.5; 40.
at Sigma = 2.15. Velocity - same as Figure
6.
Piano-Convex Model at Sigma = l.74. Velocity -same as
Figure. 6.. .. . .
Piano-Convex Model
at
Sigma = 1 54. Velocity - same asH Figure 6.
NACA 66 Model at Sigma= 2.41. . Velocity - same as. Figure 6.
ii. InOipient Cavitation, sigma = 3.05..
30;
V
3.5.Incipient Cavitation, sigma 1.40.
V35; 40.
Hydrofoil Wake Cavitation.
Wake Cavitation - Micrôflash photograph.
Wake Cavitation Spectra,
Figure 6.
16. Wake Cavitation Spectra,
Figure. 6. .
Trailing Vortex Cavitation.
Trailing Vortex Cavitation Spectra,
0
30 ft/sec;0
U
= 0
20 ft/sec;0
25;U = Q25 .ft/s,ec; 30;
Sigma .= 2.27.. Velocity - saite as
Sigma 1.79. Velocity = same as.
Sigma = 1.70.
17. Effect of Ventilation, on Wake Cavitation Spectr
U=20
U=25
Unventilated
0
Ventilated
/,/i/, / /////////////
Metal Wail
1/32" Diaphragm
Figure 1.
Transducer Geometry.
Plexiglass Wall
Flooded Cavity
1/4" Transducer
6"
UFREQUENCY (KHZ)
Figure 2.
Clear Tunnel Noise
Spectra: El 20 ft/sec;
0
25 ft/sec;A
30 ft/sec; S735 ft/sec; $40 ft/sec; 045 ft/sec;so
ft/sec.-S
FREQUENCY (KHZ)
Figure 4.
NACA 66 Model at 20 ft/sec.
Sigma
= 0
fully wetted; Q6.6; Lt4.9; c74.2. 10 100FREQUENCY (KHZ)
Figure 5.
NACA 66 Model at20
ft/sec.
Sigma = Ufully wetted;
0
3.5;
40,-
20zD
Figure
111111
I I1111111
I I1111111
FREQUENCY (KHZ) ,.10. 100NACA 66 Model at Sigma
2.46..
U' = 1320 ft/sec;
.0 25
3,0; V 35;
.4.0. ..20
-Cl)-40 -
-60 I I40 20 w
>0
w -J 20 .60: L FREQUENCY (KHZ). Figure 7.NACA 66 Model at Sigma =2.15.
Velocity - same as Figure 6.
I.. 'I .I. I. I F I I i i I 10. 100
FREQUENCY (KHZ)
Figure 8.
Piano-Convex Model at Sigma = i.74.
Veiàcity - same as
40 -
V
0
-40 -
-60'
1.1
I11111
I I1111111
I I I111111
FREQUENCY (KHZ). Figure 9...Piano-Convex Model at Sigma = 1.54.
Velocity - same as
40 -
.
200
10
2O
60 I I Iii ii
I I I I i iii
I I I I i iii
110
100 FREQUENCY (KHZ) FigurelO.NACA66 Model at Sigma = 2.41.
40 -
V
0
20
-)
00
J
C.)UJi20 -
a- Cl) I1.111111
I I1111111
-40 - 60 I II
Iiii,
I 1 10 100 FREQUENCY (KHZ) Figure 11.Incipient Cavitation, sigma = 3.05.
U = 1] 20 ft/sec;
0 25;
40 20 -40 -60 1 10 100 FREQUENCY (KHZ)
Figure
12.Incipient Cavitation, sigma = 1.40.
U
.Q 25 ft/sec; L 30;
Figiire' 14.
Wake Cavitation
40 20
-j
LU>
LU -JD
I- C-) LU C,) -20 -40 -6011,1111
I I.11.11111
1 I1111111
10 100 FREQUENCY (KHZ) Figure 15.Wake Cavitation Spectra, Sigma = 2.27.
Velocity - same as
10
FREQUENCY '(KHZ)
Figure
16.
Wake Cavitation Spectra., Sigma
1.79.
Velocity - same as
Figure 6.
:
-40 -60 I I
111111
I I I111111
FREQUENCY (KHZ) Figure 17.Effect of Ventilation on Wake Cavitation Spectra.
U=20
U=25
Unventilated [10
Ventilated I I iui1
100 1 10Figure
18.
40 20 -40 -60
FREQUENCY (KHZ)
Figure
19.
Trailing Vortex Cavitation Spectra, Sigma
= 1.70. 0 30 ft/sec;