Vous êtes sur la page 1sur 11

2071

Observation of cavitation inception in separating


water ows through constricted channels
SWashio
1
, S Fujiyoshi
2
, and S Takahashi
3
1
Chugoku Polytechnic College, Employment and Human Resources Development Organization of Japan, 1242-1 Nagao,
Tamashima, Kurashiki City, Okayama, Japan
2
Izumo Murata Manufacturing Co., Ltd, 2308 Kaminaoe, Hikawa-cho, Hikawa-gun, Shimane, Japan
3
Graduate School of Natural Science and Technology, Okayama University, 3-1-1 Tsushimanaka, Okayama, Japan
The manuscript was received on 10 November 2008 and was accepted after revision for publication on 16 March 2009.
DOI: 10.1243/09544062JMES1438
Abstract: The research group including two of the present authors previously discovered a sur-
prising phenomenon in oil ows through constricted channels; at a certain owrate when there
is no cavitation occurring, an innitesimal cavity suddenly emerges on the wall where the ow
separates. The newborn cavity grows and splits releasing many minute bubbles to downstream,
as the owrate increases. To see if this phenomenonoccurs inwater too, water ows throughrect-
angular and cylindrical constrictions have been meticulously observed with very high resolution
in terms of both time and space. As a result it has been conrmed that the same phenomenon
does occur in separating water ows as well, although the developing process of a newborn cav-
ity in water ows is different from that in oil ows probably depending on a large difference in
Reynolds numbers between both ows. When the owrate is substantially increased from that
for the cavity inception, a large attached cavity is formed on the wall just downstream of the
separation point, as it occurred in corresponding oil ows. The pressure inside the cavity has
proved to be almost equal to the saturation vapour pressure of water.
Keywords: cavitation inception, point of separation, water, bubble, liquidsolid interface,
saturation vapour pressure, dissolved gas, tensile stress
1 INTRODUCTION
More than twodecades agoWashioet al. [1] succeeded
in creating a transient tensile wave in hydraulic oil
lines, recognizing that it propagates in oil at the same
velocity with the same attenuation as a compressive
one. Since their nding conicted with the prereq-
uisite of the traditional cavitation nuclei theory that
microscopic bubbles always exist everywhere in liq-
uid, a fundamental doubt arose about the theory,
leading them to re-examine the mechanism of cavita-
tioninception. After years of observations of cavitation
inseparating oil ows since then, they have discovered

Corresponding author: Employment and Human Resources,


Development Organisationof Japan, ChugokuPolytechnic College,
1242-1 Nagao, Tamashima, Kurashiki City, Okayama 710-0251,
Japan.
email: washio_s@mx1.tiki.ne.jp; s1.washio@ehdo.go.jp
the following new fact that is presumably the real
inception of cavitation [24].
That is, as the owrate increases, an innitesimal
cavity suddenly emerges on the wall around where
the ow separates. The newborn cavity grows while
being attached on the wall, and then splits releasing
numerous minute bubbles to downstream. The inci-
pient cavity bornonthe wall is so small and the course
of its birth, growth, and splitting proceeds so fast that
the phenomenon can never be recognized with the
naked eyes. On the other hand, bubbles released from
the cavity by splitting grow big in turbulence devel-
oping downstream of the separation point and often
come out visible as the so-called cavitation bubbles.
All these ndings indicate that cavitationnuclei donot
inherently exist in liquid but are a posteriori produced
on the wall near the point of separation.
Regrettably though, there seem to be many cavi-
tation researchers who are still sceptical about this
discovery. Considering the liquidtraditionally usedfor
the cavitation research has overwhelmingly been not
JMES1438 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science
2072 SWashio, S Fujiyoshi, and S Takahashi
oil but water, their scepticism might be based on the
preconception that cavitation in water is fundamen-
tally different from that in oil for which the discovery
has been done. Fromthe hydrodynamic point of view,
however, there is no reason to doubt that a phe-
nomenon occurring in oil ows can occur in water
ones too, which Washio et al. [5] claim in the recently
published paper dealing with the mechanismof cavity
generation at the point of separation.
After all, in order for the discovery to be regarded
solid enough by many, the phenomenon should be
conrmedinwater ows too. Under thecircumstances
that researchers of water cavitation are indifferent to
the subject, the present authors have tackled it by
changing the test liquid from oil to water. The present
article provides the results obtained.
2 EXPERIMENTAL APPARATUS
2.1 Constricted channels for observation
The primary purpose of the present article is to make
clear whether the cavitation inception phenomenon
discovered in separating oil ows occurs or not in sep-
arating water ows. Therefore constricted channels
having similar geometries to the ones previously used
inoil ows were made of acrylic againfor observations
of water ows. That is, a block of either a rectangular
or a cylindrical shape was inserted into a test chan-
nel having a rectangular cross-section (10 mm high,
5 mm wide, and 60 mm long) to form either a rect-
angular [2] or a cylindrical constriction [4], which is
illustrated in Figs 1(a) and (b), respectively. The walls
of both blocks where ows were supposed to separate
were polished up using a bufng compound to make
them as smooth as possible. To determine the aver-
age velocity through the minimum section precisely,
the clearance of the constriction was measured using
a length-measuring machine witha microscope, every
time after the observation.
2.2 Water circulatory unit
Withtapwater usedas the experimental liquid, a water
hydraulic circuit illustrated in Fig. 2 is prepared for
the test. Water discharged from a vane pump with
a rated owrate of 14 litres/min and a rated pres-
sure of 1.1 MPa ows into the test constriction after
passing through a turbine ow meter and returns to
a reservoir tank. The owrate through the constric-
tion is regulated with a variable restriction set in the
bypass line branched at the out port of the pump
and is measured with the turbine ow meter. The
temperature of water is detected with a thermistor
thermometer mounted upstream of the constriction,
whereas the pressures upstream and downstream of
the constriction (Pu and Pd) are measured with strain
Fig. 1 Constrictedchannels usedfor test: (a) rectangular
constriction and (b) cylindrical constriction
Fig. 2 Water supply circuit
gage type pressure transducers of 5 and 1 MPa ratings,
respectively.
2.3 Means of observation
To observe the incipient cavity expected to emerge
suddenly on the wall where a water ow separates
with high resolution in terms of both time and space,
a super high speed video camera (maximum shoot-
ing rate: 1 000 000 fps) is used in combination with a
long distance microscope (maximum magnication:
900). To illuminate the observation area by trans-
mitted light, a 150W metal halide lump is set on the
opposite side of the camera across the test channel.
With the above-mentioned observation means, one
pixel on the CCD image sensor (312 260 pixels) cor-
responds to 3.2 m if an object is photographed by
the maximum magnication. Besides, to identify the
moment a cavity is created, the same method as was
previously adopted for oil ows is used again [24].
That is, a HeNe laser beam is transmitted close to
the point of separation after being focused down to
0.1 mm in diameter, and a sudden change of its trans-
mission brought about by the emergence of a cavity is
used as a triggering signal for the recording with the
super high speed video camera.
Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1438 IMechE 2009
Observation of cavitation inception in separating water ows through constricted channels 2073
3 RESULTS OF OBSERVATION
3.1 Rectangular constriction
In the rst place, with the rectangular constriction
(Fig. 1(a)) employed, the wall in the neighbourhood
of the ow separation point, that is, the square edge
at the entrance of the constriction, was observed from
the top-view angle with the laser beam transmitted
along the entrance edge for triggering as shown in
Fig. 3, whereas the owrate was kept constant. Con-
sequently, it has certainly been recognized that a
microscopic cavity does suddenly emerge on the wall
just downstream of the entrance edge where there
are no observable cavities and bubbles before the
emergence. A typical result obtained by the video
photography at 1 000 000 fps is displayed in Fig. 4;
in Fig. 4(a), some images arbitrarily picked out of
the original video data are arranged chronologically
to demonstrate the phenomenon of cavity inception,
whereas Fig. 4(b) shows thebirthandgrowthof acavity
at its very beginning stage by enlarging some consec-
utive images taken at 1 s intervals. In these images,
the entrance edge appears as a line vertically dividing
each picture into halves. Moreover, the ow direction
is fromleft to right as shown by the arrow, which is the
same for all the video images provided in the present
article.
In Fig. 4, it has clearly been recorded that a cavity
as small as 10 m in diameter suddenly emerges and
subsequently grows on the wall near the edge, verify-
ing that the same phenomenon of cavity inception as
was discovered in separating oil ows occurs in sepa-
rating water ows too. Although the newly born cavity
is attached on the wall and grows in the direction par-
allel to the edge just for a short period (Fig. 4(b)), it
soon moves downstream and splits (Fig. 4(a)). Judg-
ing from Fig. 4(a), the rate of this cavity movement is
much smaller than the average velocity u in the con-
striction, which presumably indicates that the cavity
is detached from and is carried close to the wall by
the main ow. Such behaviour of a developing cavity
is obviously different fromthat in a separating oil ow,
where an incipient cavity elongates along the edge
while keeping attached to the wall, as is recognized by
Fig. 3 Laser beamtransmission along edge of rectangu-
lar constriction for triggering of video recording
Fig. 14 in reference [2]. This difference of cavity devel-
opment in oil and water ows is probably based on
the difference of owstructure between the two ows.
That is, the Reynolds number of an oil ow through
the constriction is quite small and thereby the sepa-
ration ow there is just locally turbulent but basically
laminar. On the other hand, the water owin the con-
striction is entirely turbulent with a Reynolds number
about 100 times higher than that of the corresponding
oil ow.
Once a cavity emerges on the edge, it meteori-
cally develops and splits, supplying the separation
region downstream of the edge with many minute
bubbles, some of which expand explosively in the
shear layer of the separation ow and become visi-
ble even to the naked eye. In Fig. 5, the behaviours of
such expanding bubbles are demonstrated with their
top-view images taken at 500 000 fps. Here one can
see that bubbles grow big and split in vigorous tur-
bulence just upstream of where the separation ow
reattaches to the wall. In particular, the bubbles indi-
catedby arrows showtypical examples that explosively
grow in microseconds. Although all those bubbles are
considered to originate from an incipient cavity gen-
erated on the edge, it is hard to recognize it in these
images. However, it was possible with oil ows. Actu-
ally in a separating oil ow through a rectangular
constriction, it was observable that an incipient cav-
ity having grown to a string shape attached on the
entrance edge intermittently emitted bubbles, which
explosively developed upstream of the reattachment
point (see Figs 11 and 12 in reference [2]).
Moreover, Fig. 6 shows a similar cavitating owpho-
tographed from the side at the same frame rate as
Fig. 5. Here, the bubble pointed by an arrow in the
second frame (4 s) seems to emerge suddenly in the
midst of ow on the separation streamline and grows
rapidly afterwards. Probably Arakeri [6] and Katz [7]
recognized similar phenomena in their observations
of separation water ows, which must have lead them
to the understanding that an explosive growth of
invisible bubble nuclei existing inherently in liquid is
a start of cavitation [6, 7]. As is known fromthe discus-
sion above, however, it seems to be the fact that those
nuclei are not inherent but newly produced from an
incipient cavity started on the wall.
That assertion might be reinforced by another fea-
ture of Fig. 6 that the separation region looks dimmer
than the surrounding main ow. Presumably a dim
image of the separation region indicates that there
are numerous micro-bubbles within the region but
not outside it, and it is nothing but a cavity newly
born on the edge that can supply those bubbles exclu-
sively to such restricted area as downstream of the
edge.
On the basis of all the discussion above, the present
article would assert that the real start of cavitation in
separation water ows is the sudden emergence of a
JMES1438 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science
2074 SWashio, S Fujiyoshi, and S Takahashi
Fig. 4 Sudden emergence and subsequent ssion of incipient cavity on entrance edge of rect-
angular constriction (top view) (u = 23.9 m/s, Re = 25 000, = 0.71, = 22.7

C, and
1 000 000 fps). (a) Images of cavity emergence and (b) enlarged consecutive images of
newborn cavity at incipient stage
Fig. 5 Explosive growth of bubbles downstream of separation point in rectangular constriction
(top view) (u = 23.7 m/s, Re = 23 000, = 0.70, = 19.7

C, and 500 000 fps)


Fig. 6 Sudden emergence and subsequent explosive growth of a bubble in separation shear layer
(side view) (u = 26.8 m/s, Re = 22 000, = 0.71, = 15.2

C, and 500 000 fps)


Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1438 IMechE 2009
Observation of cavitation inception in separating water ows through constricted channels 2075
Fig. 7 Emergence and development of incipient cavity in accelerated water ow through
rectangular constriction (u = 0 34.9 m/s, = 29.8

C, and 125 000 fps)


cavity at thepoint of separation. Accordingly, anexplo-
sive growth of bubble nuclei in a separating shear ow
is not the start of cavitation but a consequence of it,
contrary to what Arakeri [6] and Katz [7] considered.
The same phenomenon of inception and subse-
quent development of a cavity at the point of sepa-
ration was observed also in a transient ow through
the rectangular constriction, when the owrate was
rapidly augmented from zero. Some of the obtained
video images are chronologically displayed in Fig. 7,
where the process in which a tiny cavity suddenly
emerges on the edge, explosively develops and splits
is clearly recognized again.
3.2 Cylindrical constriction
To nd out if the same phenomenon of cavity gener-
ation occurs even on a smooth wall where the ow
separates, the rectangular constriction was replaced
by the cylindrical one (Fig. 1(b)). A laser beam was
used again for triggering of the photography with the
super high-speed video camera. Since the ow sepa-
rates the cylindrical surface a little downstream of the
narrowest section, a laser beamwas transmittedabout
2 mm downstream of the section at the middle of the
channel width, as is illustrated in Fig. 8.
When the owrate was kept constant at a cer-
tain level, the transmission of a laser beam suddenly
decreased. Withthis dropof laser transmissionusedas
a triggering signal for the video camera, the ow over
the cylindrical wall around the separation point was
Fig. 8 Laser beam transmitted at point of separation
on cylindrical surface for triggering of video
photography
photographed fromthe top viewangle at 1 000 000 fps.
The obtained images are shown in Fig. 9; in Fig. 9(a),
some images appropriately chosen from the original
video data are chronologically arranged to demon-
strate the cavity inception phenomenon on a smooth
surface, whereas enlarged images of the cavity just
after its birth are consecutively displayed at 1 s
intervals in Fig. 9(b).
Figure 9 reveals that a cavity as small as 10 m sud-
denly emerges on a smooth wall near the separation
point around where there were no recognizable bub-
bles or cavities before. Although the newly generated
cavity grows keeping an elliptical shape for just a short
period (Fig. 9(b)), it quickly loses its initial shape, tak-
ing on a disorderly form by splitting and expanding
quite irregularly (Fig. 9(a)). Onthe other hand, as Fig. 8
in reference [4] or Fig. 2 in reference [5] shows, a cavity
born on a cylindrical wall in an oil ow kept a trim
elliptic shape until having grown much larger than
the initial one and split not irregularly but neatly into
two parts rst. Consequently this irregularity in the
development of an incipient cavity is considered to be
characteristic of separating water ows.
The video images obtained by photographing the
cavity inception phenomenon at 500 000 fps from the
side viewangle are displayedinFig. 10 inthe same way
as inFig. 9. Here, it has beenclearly recordedagainthat
a tiny cavity suddenly emerges where there was noth-
ing like bubbles andcavities, grows a little inanorderly
way (Fig. 10(b)) but soon splits irregularly (Fig. 10(a)).
By this side-viewobservation, it has beenconrmed
again that a cavity is really generated in a water ow
separating from a smooth surface as well. Here, it
should be pointed out that there is one particular dif-
ference between Fig. 10 and its counterpart previously
obtainedfor anoil ow, that is, Fig. 3inreference[5]. In
the latter, not only an incipient cavity but also its mir-
ror image reected on the acrylic surface was clearly
recorded. Since the upstream tips of those real and
mirror images coincided at one point in those pic-
tures, it was concluded that the upstream tip of the
cavity was always attached to the acrylic surface from
its inception. Moreover, onthebasis of this conclusion,
it was presumed that the cavity was created not inside
oil but on the interface between oil and the acrylic
wall. In Fig. 10 for a water ow, however, there are no
JMES1438 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science
2076 SWashio, S Fujiyoshi, and S Takahashi
Fig. 9 Top view images of a cavity suddenly emerging on cylindrical wall near where ow sep-
arates (u = 26.2 m/s, Re = 28 000, = 0.50, = 19.1

C, and 1 000 000 fps). (a) Emergence


and subsequent growth of cavity and (b) very beginning stage of cavity inception
Fig. 10 Side view images of a cavity suddenly emerging on cylindrical wall near where ow sep-
arates (u = 24.3 m/s, Re = 27 000, = 0.56, = 19.8

C, and 500 000 fps). (a) Emergence


and subsequent growth of cavity and (b) very beginning stage of cavity inception
mirror images recorded, which means it would be dif-
cult to make a discussion about where the cavity is
created. The reason of no mirror images for a water
ow through the acrylic test channel is not yet clear,
but might be explained as follows: oil has a refrac-
tive index similar to that of acrylic but water does
not, which probably causes the cylindrical surface to
turn out black in its video images as is seen in Fig. 10
and thereby prevents mirror images of a cavity from
appearing in those pictures.
As is pointed out above with regard to Fig. 9, the
metamorphosis of a newborn cavity in a separating
water ow is quite irregular and accordingly takes
a different process each time the cavity develops,
which was not the case with an oil ow though. To
demonstrate the lack of reproducibility in the cavity
development, two more different types of metamor-
phoses of an incipient cavity are shown in Fig. 11.
When the owrate is further increased from that for
the cavity inception, cavitation in the cylindrical con-
striction becomes more vigorous with largely devel-
oped cavities attached on the cylindrical wall. The
picture in Fig. 12 shows typical ones, which have been
quite familiar to researchers who observed cavitation
in water ows separating from smooth walls [7, 8].
According to the assertion of the present article, it is
Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1438 IMechE 2009
Observation of cavitation inception in separating water ows through constricted channels 2077
Fig. 11 Different developing processes of incipient cavity on cylindrical surface (side viewimages,
500 000 fps). (a) u = 25.1 m/s, Re = 29 000, = 0.54, and = 21.5

C. (b) u = 25.1 m/s,


Re = 29 000, = 0.54, and = 21.5

C
Fig. 12 Attached cavitation on cylindrical surface for
a larger owrate (u = 45.1 m/s, Re = 64 000,
= 0.19, and = 30.8

C)
beyond question that this type of cavitation also starts
froma microscopic cavity born on a wall near the sep-
aration point, although that point of view might not
yet be shared with other cavitation researchers.
4 MAXIMUMDEVELOPEDSTATIONARY CAVITY
INRECTANGULAR CONSTRICTION
4.1 Process of development
In an oil ow through a rectangular constriction, a
cavity generated at one point on the entrance
edge elongated parallel to the edge as the owrate
increased. When the elongating cavity extended over
the whole edge, it suddenly expanded in the down-
stream direction and formed one large stable cavity,
which covered a substantial area from the edge to
downstream [2]. Moreover, this cavity had such a
unique property that its inside was maintained near
vacuum.
With a separating water ow through the constric-
tionincontrast, anincipient cavitydoes not stayonthe
edge and easily splits supplying the separation region
withbubbles, as was observedabove. Notwithstanding
the difference of the developing process for an incip-
ient cavity, however, a large stable cavity is eventually
formed in the water ow too, if the owrate is further
increased.
The process of such cavity formation is demon-
strated in Fig. 13 with top-view images of cavitation
at three different stages of development. Fig. 13(a)
shows a vigorous cavitation with numerous bubbles
in the separation region. When the owrate is more
increased, cavitation becomes more vigorous and a
stable cavity is partially formed in the constriction as
Fig. 13(b) shows. In an oil ow, though, the expan-
sion of a string-shaped cavity towards downstream
occurred so fast that an intermediate cavity like this
couldnot be formedstably. As the owrate is increased
even more, a large cavity shown in Fig. 13(c) eventu-
ally emerges, spreading over the whole width of the
constricted channel. Here circular objects within the
cavity are not bubbles but water droplets remaining
on the channel wall.
4.2 Changes of wall pressures in constriction
during cavity development
Previously in an oil ow through a rectangular con-
striction, the pressures were measured with six pres-
sure holes aligned on the constriction walls in the ow
direction during the expanding process of an incipi-
ent cavity [2]. With ordinary pressure transducers of
MPa rating used for the measurement, the pressure
inside the maximum developed cavity in an oil ow
JMES1438 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science
2078 SWashio, S Fujiyoshi, and S Takahashi
Fig. 13 Transition fromvigorous cavitation to maximumdeveloped attached cavity in rectangular
constriction: (a) vigorous cavitation (u = 26.3 m/s and Re = 17 000), (b) partially formed
attached cavity (u = 33.5 m/s and Re = 22 000), and (c) maximum developed attached
cavity (u = 39.9 m/s and Re = 27 000)
turned out to be too lowto distinguish froma vacuum.
This time the similar measurement was conducted in
a water ow, as the owrate was gradually increased
until a stable cavity developed in such a way as shown
in Fig. 13.
A rectangular constriction having a 0.7 mm clear-
ance and four pressure holes of 0.7 mmin diameter on
the walls was newly made of acrylic for the purpose
(Fig. 14). To measure the pressures upstreamof, inside
and downstream of the constriction, strain-gage pres-
sure transducers of 5 MPa (for upstream) and 1 MPa
(for inside and downstream) ratings were employed.
The obtained data are plotted in Fig. 15 with the
averagevelocityuintheconstrictionandthepressures
in absolute scale taken on the abscissa and ordinate,
respectively. Among the data, the pressure P
2
provides
a typical example of how the wall pressures behave
during the cavity development. At rst, P
2
is larger
than 0.1 MPa abs and increases with the owrate. As
the owrate is further increased, it suddenly drops
down to several kPa abs around at u = 15 m/s and
keeps the same value for larger u. This sharp drop
indicates that a growing cavity has covered the detect-
ing hole for P
2
. In the reverse course of reducing the
owrate, the pressure suddenly bounces back from
several kPa abs to above 0.1 MPa abs almost at the
same u as it dropped. This sudden recovery of P
2
means that the cavity has shrunk and the P
2
hole has
Fig. 14 Rectangular constriction with four pressure
holes on constriction walls
P


M
P
a
.
a
b
s
0
0.5
1
P
u
0
0.05
0.1
P
1 inc.
dec.
0
0.05
0.1
P
2
0
0.05
0.1
P
3
0
0.05
0.1
P
4
10 20
0
0.1
0.2
u m/s
P
d
Fig. 15 Variation of pressures in water ow through
rectangular constriction during formation of
the maximum developed attached cavity
been exposed to the owing water again. In an oil ow
through a rectangular constriction there was an evi-
dent hysteresis between these increase and decrease
processes [2], but there isnt for a water ow. To dis-
cern the two processes in Fig. 15, the arrows denoted
by inc and dec are added near their proles.
The sharp drops and rises of pressure imply that
a cavity grows and shrinks quite rapidly even with a
little change of the owrate around a critical value.
Among all the pressures, the behaviour of the pres-
sure P
1
, whose detecting hole exactly faces that of P
2
on the opposite wall just 0.7 mm apart and always
contacts the water owing outside the cavity, seems
interesting. That is, P
1
also drops and bounces back
in synchronization with P
2
, but fairly increases with
Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1438 IMechE 2009
Observation of cavitation inception in separating water ows through constricted channels 2079
10 15 20 25
0
1
2
3
4
measured
from literature
P
2

k
P
a
.
a
b
s
= 20.7 m/s
Fig. 16 Pressures inside the maximumdeveloped cavity
the owrate after the drop, which is denitely differ-
ent from the behaviour of P
2
. The similarity of the
descending prole between P
1
and P
2
shows that an
incipient cavity emerging in a ow reduces the pres-
sure around it to the same level as its inside. On the
other hand, the difference between the two after the
drop implies that a considerable pressure gradient
occurs in a very thin ow layer between the cavity
surface and the upper wall.
4.3 Pressure inside the cavity
It has turned out in Fig. 15 that the pressure inside
the incipient cavity is very low but denitely has a
certainlevel above a vacuum. Thenthe following ques-
tions naturally arise: how low is it and how is its level
determined? To answer these questions, the pressure
P
2
was precisely measured as the owrate was kept
constant. To measure pressures as small as several kPa
in this measurement, a pressure transducer capable of
detecting pressures between a vacuum and the atmo-
spheric with a resolution of 10 Pa was introduced in
place of the above-used conventional ones.
Moreover, an experimental technique was intro-
duced to obtain accurate data. When a cavity covers
the pressure hole, water within the hole is sucked out
into the cavity but some droplets inevitably remain. To
remove completely those droplets that might hinder
accurate measurement, the pressure transducer was
detached when the maximum cavity was generated.
After air was sufciently sucked into the cavity to get
ridof all the droplets, the hole was pluggedupwiththe
transducer again.
If water is circulated without cooling, its temper-
ature is gradually elevated by uid energy loss. The
pressures were measured in that natural heating pro-
cess this time. The obtained data are plotted in Fig. 16
with the water temperature and the pressure in the
absolute scale taken on the abscissa and the ordinate,
respectively. Here the black dots show the measured
data, whereas the solid line indicates the saturation
vapour pressure of water citedfromthe table [9]. There
is a close agreement between the two, which indi-
cates that the inside of the cavity is almost lled with
saturated vapour of water.
The above conclusion may look self-evident, but it
is not necessarily so considering it was not the case in
oil ows. Actually the pressure inside a similar stable
cavity generated in an oil ow was very low but much
higher thanthe saturationvapour pressure of oil. Gen-
erally speaking, it is not only vapour but also dissolved
gases that come out intoa cavity surroundedby liquid.
Since the dissolvedgases shouldkeepcoming out until
their partial pressures balance with their solubilities,
the total pressure inside the cavity will eventually be
pushed up to the level fairly higher than the saturation
vapour pressure, which actually occurred for a cavity
in oil. However Fig. 16 shows that it does not happen
with water. Probably, this discrepancy between oil and
water can be attributed to a large difference of vapour
pressure between the two liquids. As the result of that
difference, the inside of a cavity developed on the wall
downstream of the separation point is mainly occu-
pied by vapour in a water ow but by gases separated
from the surrounding liquid in an oil ow.
5 CONCLUSIONS
The inception and subsequent development of a cav-
ity in separating water ows have been observed in
cylindrical and rectangular constrictions. The results
obtained are summarized as follows.
1. As was the case with separating oil ows, cavitation
in separating water ows also starts with a sud-
den emergence of a microscopic cavity on the wall
where the ow separates. With this result, it can be
concluded that a birth of a microscopic cavity at
the point of separation is the universal mechanism
of cavitation inception in separating ows of any
liquid.
2. Unlike the behaviour of a cavity in a separating oil
ow, an incipient cavity in a water ow separating
fromaprojectedwall does not attachfor longonthe
wall downstream of the separation point but starts
to split apart soon after its generation. This differ-
ence inoil andwater ows is presumably explained
by the fact that the former are basically laminar,
whereas the latter turbulent.
3. Similarly to what has previously been conrmed in
oil ows through a rectangular constriction, a large
stable attached cavity is formed on the wall down-
stream of the separation point in water ows too at
a large owrate, although the process of formation
is somewhat different from that in oil ows.
4. The pressure inside the thus formed cavity is kept
almost at the saturation vapour pressure of water
at its temperature. However that tendency has not
beenfoundinoil ows. Actually, the pressure inside
a cavity formed in an oil ow within a rectangular
constriction is very low but is considerably higher
than the vapour pressure of oil.
JMES1438 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science
2080 SWashio, S Fujiyoshi, and S Takahashi
ACKNOWLEDGEMENTS
The manufacturing of the experimental appara-
tus owes a lot to Mr Yoshihiko Tamura who is
a senior technician in the Faculty of Engineering
OkayamaUniversity. His assistanceis herebygratefully
acknowledged.
REFERENCES
1 Washio, S., Takahashi, S., Konishi, T., and Moriwake, H.
Creation and observation of tensile waves in oil column.
JSME Int. J., 1994, 37(2), 342348.
2 Washio, S., Takahashi, S., Uda, Y., and Sunahara,
T. Study on cavitation inception in hydraulic oil
ow through a long two-dimensional constriction.
Proc. IMechE, Part J: J. Engineering Tribology, 2001,
215(J4), 373386. DOI: 10.1243/1350650011543619.
3 Washio, S., Takahashi, S., Kawahara, S., and Kishitani,
M. Study on the cavitation mechanism in hydraulic oil
ow using a needle projection. Proc. IMechE, Part J:
J. Engineering Tribology, 2002, 216(J1), 2734. DOI:
10.1243/1350650021543861.
4 Washio, S., Takahashi, S., and Yoshimori, S. Study
on cavitation starting at the point of separation on a
smooth wall in hydraulic oil ow. Proc. IMechE, Part
C: J. Mechanical Engineering Science, 2003, 217(C6),
619630. DOI: 10.1243/095440603321919554.
5 Washio, S., Takahashi, S., Uemura, K., Iwamoto, T.,
and Ogata, T. Singular properties of ow separation
as a real cause of cavitation inception. Proc. IMechE,
Part C: J. Mechanical Engineering Science, 2008, 222(C4),
667678. DOI: 10.1243/09544062JMES835.
6 Arakeri, V. H. and Acosta, A. J. Viscous effects in the
inception of cavitation on axisymmetric bodies. Trans.
ASME, J. Fluids Eng., 1973, 95, 519527.
7 Katz, J. Cavitation phenomena within regions of ow
separation. J. Fluid Mech., 1984, 140, 397436.
8 Holl, J. W. and Carroll, J. A. Observations of the vari-
ous types of limited cavitation on axisymmetric bodies.
Trans. ASME, J. Fluids Eng., 1981, 103, 415424.
9 JSME. JSMESteamTables, vol. 29, 1999(TheJapanSociety
of Mechanical Engineers, Tokyo).
APPENDIX
Notation
Pu absolute pressure upstream of constriction
Pd absolute pressure downstream of constriction
Re Reynolds number
u average velocity at narrowest section of
constriction
temperature of water
cavitation number Pd/(PuPd)
Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1438 IMechE 2009
Reproducedwith permission of thecopyright owner. Further reproductionprohibited without permission.

Vous aimerez peut-être aussi