Professional Documents
Culture Documents
REVIEWS Further
Quick links to online content
STRUCTURES
Roger E. A. Arndt
by University of York on 08/08/13. For personal use only.
INTRODUCTION
Figure I Cavitation damage in the spillway tunnel of the Yellowtail Dam in Montana.
Courtesy of US Bureau of Reclamation.
review of the entire topic here. Instead, we review our present state of
knowledge of the physical phenomena that are common to all types of
applications, particularly the factors leading to the occurrence of cavita
tion, subsequent performance breakdown, and other deleterious side
effects such as vibration, noise, and damage. We attempt to bridge the
gap between the description of fundamental mechanisms and practical
experience. Cavitation is generally considered here as, an unwanted side
effect of fluid flow, althQugh, as already mentioned, there are useful
applications of the phenomenon.
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
Occurrence of Cavitation
Cavitation is normally defned as the formation o.f the vapor phase in a
liquid. The term cavitation (originally coined by R. E. Froude) can
imply anything from the initial formation of bubbles (inceptio.n) to.
large-scale, attached cavities (supercavitation). The formation ,of indi
vidual bubbles and subsequent development of attached cavities, bubble
clouds, etc, is directly related to reductions in pressure to some critical
value, which in turn is asso.ciated with dynamical effects, either in a
flowing liquid or in an acoustical field. Cavitatio.n is distinguished from
276 ARNDT
Figure 3 Cavitation in the wake of a sharp-edged disk. Courtesy of the Applied Research
Laboratory.
Figure 4 Cavitation in the tip vortices of a propeller. Courtesy of the Applied Research
Laboratory.
278 ARNDT
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
by University of York on 08/08/13. For personal use only.
Figure 5 Vortex and sheet cavitation on a marine propeller. Courtesy of the Netherlands
Ship Model Basin.
Cavitation Scaling
The fundamental parameter in the description of cavitation IS the
cavitation index
Po-Pv
0=--
u,2 (l)
iP
1
0
f
1.1
EFFICIENCY 100 jUL lMITED INO
STAGES CAVITATION
U
-
I
.
0=0,
j
RATIO, PERFORMANCE
I
NC 0.90 LOSS)
L
a>FULLY DEVELOPED
0.80 PA RTIA
Q)_. U
INOISE
I I I I I
& VIBRATION)
f
, I , I , 0<0,
ACOUSTIC ""
LEVEL - _
IdB)
.-
I I t I I o Oc
o 0.1 0.2 0.3 (LARGE
CAVITATION INDEX, 0
PERFORMANCE
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
LOSS)
(2)
where Hsv is the net positive suction head (Wislicenus 1965) and H is the
total head under which a given machine is operating. In a given
situation, a and aT are proportional. Another scaling parameter used
280 ARNDT
wVQ
(3)
J3/4
s=
[ gHsv
with w being rotational speed, Q the volumetric rate of flow, and g the
acceleration due to gravity. S, as defined in Equation (3), is a nondi
mensional number. In the United States, the quantity g is normally set
equal to unity, w is replaced by n in rpm, Q in gpm, and Hsv in feet. The
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
gHsv
-- =const, (4)
nD2
e
where De is the eye or throat diameter. These relations hold when the
kinematic condition for similarity of inlet flow is satisfied, i.e.
JL =const. (5)
nDe3
aT=
( ns )4/3 . (6)
S
Hence when the operating conditions are held fixed (ns=const), there is
a direct relationship between aT and S. The suction specific speed is
used extensively for pumps but has not received the same use in the
turbine field.
(7)
where Cpm is the mInImum pressure coefficient. In the case of a
hydrofoil or a strut, Cpm is a function of angle of attack, a, as well as the
shape of the body. The equivalent parameter for a turbomachine would
be the normalized flow rate, Q/nD3 In the case of a propeller, the
effective parameter is the advance (;oefficient, J= U/nD. Hence, for
CAVITATION 281
fixed geometry
-hydrofoils, struts
(J
c
=f{/nDnD3
U
-pumps and turbines
-propellers
(8)
shown with angle of attack as the independent parameter (in Figure 7).
Typical cavitation data for a propeller using J as the independent
parameter are shown in Figure 8. It is no accident that the two sets of
by University of York on 08/08/13. For personal use only.
60
5.0 v- 25 FPS
D-40FPS
0- 60FPS
ij'40
v
'"
w \ DAlLY(
en WATER TUNNEL TESTS
30
v
z
<t1t.
"Va'"
z
o ' !l
i= 2.0 v
;:! II
u
1.0
o ,_ I
20 15 10 5 0 -5 -10
ATTACK ANGlE IN DEGREES, a
0.4
ONSET OF
TIP VORTEX
CAVITATION
02
g
0.1 ONSET Of SACK CAVITATION
5
;) 0
0.3 0.4 0.5 06 0.7 0.8 0.9 10
J=LVnD
data are qualitatively similar. In fal;:t, hydrofoil data of the type shown
in Figure 7 can be used to predict the cavitating performance of a
propeller (Brockett 1966).
As already pointed out, resistance to cavitation in a turbomachine is
normally defined by the value of 0T where an observable reduction in
performance is noted as 0T is lowered below this critical value. The
same definition can be used for hydrofoil data, as shown in Figure 9.
Again the similarity of these data with those presented in Figure 6 is no
accident. In some situations, performance breakdown can also be
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
calculated (Acosta & Hollander 1959, Stripling & Acosta 1962, Stripling
1962, Wade 1967). However, the bulk of our information concerning the
resistance of a given machine to cavitation has been developed empiri
cally. An example is shown in Figure 10, where the minimum allowable
by University of York on 08/08/13. For personal use only.
-l 0.8
u '
k
..
8
'i
..
0.6 6
0
U
4
0-
0.4
--
0.2 -
- -
-
-
- .....
-
2
=
'
0
."
0 0.4 0.8 1.2. 1.6 2.0 2.4 2.8
Cavitation Number, IT
n pl/2
N= 5/4 (rpm,Hp,ft)
H
2.0.0.
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
1.0.0.
0..80.
0..60.
0..50.
by University of York on 08/08/13. For personal use only.
0.040.
I-
tl 0.30.
0
E
C1 0..20.
(j)
0..10.
0..08
0.0. I '----'-__...L-_--'-__.l.-_
0..4 0..8 1.6 3.2 604
wQI/2
Ns = gH)3/4 (non-dimensional)
(
LA
.28
.c
b-
a;
E
8 u-(l
"
::l.24 26 'Q
8
ur20
c
.2
4
<..l
__
........... ,.
't.
,
,/
/-
I'.,
"
.... a ........
OC
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
,,
E ,
Q) 0
'B..
''0.
;:;2
. 16
R= UL= 2xl05
.
u
by University of York on 08/08/13. For personal use only.
0 I.
0 15 30 60 90
e (de\lrees)
v 15 degrees
1.4 30 degrees
060 degrees
v! 1.2
1.0
0.8
0.1
~
0.2 0.3 0.4 D.!) 0.6 0.7 0.8 0.9
..
1.0
(rlUC
b. Covitating Flow
0.25
8
7. 0.20
Vl
0:
0.15
::l
2
<i 0.10
:r
a
e: 0.05
Vl
o
0.20 0.25 0.30 0.35 0.40 0.45 0.50
bO
0.75
,.-'
_-+--Iii---4-,ER lI
CPmin 1l5C ALIB
I I I
I T , ,
J
l
.1:0.60 f -t'
O,5CAlIBER 001'05.
' A e 58-Q7F.
... 2-IUPPM.
CPmin(MEASUREOI
TUNNEL 48" ORl.
o
i=-
0.4
CPmin(1.5CAliBERI
U
t- 4"
030
2:
iii
' . I"
1.5 CALIBER OOIVES
852-99"1'
w 01. 2-7.9 PPM
o
_____
CPmin(MEASUREDI
TUNNEL' 48" ORL
1 1 _1......J_I . .1UI. 1 I ' "
0 .15 _
1xl05 10xl05
REYNOLDS NUMBER.
Figure 13 Cavitation on streamlined bodies (Parkin & Holl 1954).
286 ARNDT
number. In both cases the data fall below the value predicted by the
basic scaling law [Equation (7)]. Reasonable correlation with Reynolds
number is achieved, but a definite size effect is still apparent. Holl &
Robertson (1956) provided some convincing evidence that Weber num
ber was also a significant parameter. This is in keeping with the unique
measurements of Kermeen et al (1956) who found that the flow was in
tension in the minimum pressure region of a streamlined body when
cavitation was incipient.
Another aspect of the scale-effects. problem is illustrated in Figure 14
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
(9)
o 0
e
fJ-
__
2. The critical pressure at inception is not the vapor pressure and can
be either higher or lower than Pv depending on the situation.
Ordinary liquids can sustain tension and more than one type of
cavitation process can occur in the same flow field. Bubble growth
by University of York on 08/08/13. For personal use only.
'
DATA FROM CHIVERS TEST SERIES I
1.10
1.05
1.00
.95
90
.85
nrr-c
o It f
I
A
A 10.4 C
1
.80
Ho .75
.70
.65
(J "
o
,: I
o! I
X
)( !
)(
+
"
Q61.4
985
52.4
61.3
1 1 0.6
l I
.60
I x
,I ,
.55
, , .I I , ,
.50
, , , , f f
,45
0 .020 .040 .060 080 .100 .120 .140 .160
O'T
Cavitation Nuclei
It is generally accepted that cavitation inception occurs as a conse
quence of the rapid or explosive growth of small bubbles or nuclei that
have become unstable due to a change in ambient pressure. These
nuclei can be either imbedded in th: flow or find their origins in small
cracks or crevices at the boundary surface of the flow (Acosta &
Hamaguchi 1967, Holl & Treaster 1966, Gupta 1969, Peterson 1968, van
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
der Meulen 1972). The details of how these nuclei exist have been
considered by many investigators. A coherent summary of this work is
offered by Holl (1969, 1970).
by University of York on 08/08/13. For personal use only.
[ R]
Prosperetti 1977):
(12)
(1962) discuss the source of nuclei and point out that in the turbulent
flow of a water tunnel, gas bubbles can be a source of nuclei. Presum
ably the bubbles are stabilized by rectified diffusion due to turbulent
pressure fluctuations. Arndt & Keller (1976) measured the nuclei in a
by University of York on 08/08/13. For personal use only.
R*
== [ NkO ] . (15)
8'17 S
290 ARNDT
3"S/R.
4
[
O'c = - Cpm +
"2PU
I 2
Measured nuclei size distributions art shown in Figure 17. The data are
(16)
for two water tunnels whose nuclei are mostly micro-bubbles and for
two other facilities (a water tunnel and a depressurized towing tank)
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
whose nuclei consist mostly of solid particles (Arndt & Keller 1976,
Gates & Acosta 1978). The data indicate that nuclei population that are
dominated by; micro.bubbfes are sens:itive to the relative saturation level
of dissolyed gas.
by University of York on 08/08/13. For personal use only.
(P.-Pla., mBar
100 10
,;.-
;:::: Propeller Tunnel (NSMB)
Vacu-Tank (NSMB)
HSWT (CIT)
YE
'z
.;f Idl
.8
E
:::J
2
lef
Ie! I
10 100 1000
Radius, f''''
Figure 17 Typical nuclei size distribution in test facilities and in the ocean (Arndt 1978,
Gates & Acosta 1978, Medwin 1977).
CAVITATION 291
than 50 mb). However, there is evidence that the flow field around a
given body produces a screening effect such that larger bubbles tend to
move out of the critical-pressure region leaving only smaller bubbles
with more negative critical pressures to be active in the cavitation
process (Johnson & Hsieh 1966). Very little has been done to investigate
the question of the influence of nuclei size distribution on cavitation.
What information we do have does indicate that this can be an
important factor in laboratory experiments (Keller 1972, 1974, Gates &
Acosta 1978). Most of the theoretical work on the influence of nuclei on
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
O.B . . , . ,-, , . l
-- -- - -- - -- -CPmin
cli:boo t
bU 0 0
C' 0
x 0
v&
'-!--
0.6 00 0 0 0 0 00 ''-HEMISPHERICA
w
0 NOSE
z 0 0 000 a 0
0
0.4
; ERMAN - SCHIEBE <Y'0
;}
u
BOOV o
.
,
.
r
0.2 '----'-----'_-'--_1..._ r
2.2 x
"
xx
x "
" o
,u 2.0
:I:
" "x
COl
1.8 Ii'
2
....
g 1. 6
AI'9.WIttr
u -0- Temp . e
1.4 lin.! (em) ( (OC)
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
z
iii x I 2.54 n 25
...
o " 2 5.08 n 25
l. 121.92 em (48 In.) G) 4 10.16 n 25
ARL WAlER TUNNEL
Data from Robertson. GlnllY. HoII 119571
1 2.54 107 41,7
) 6 1- I l 12
I.C 4 10 I
114 l Ii -'-
by University of York on 08/08/13. For personal use only.
I I I
REYNOLDS NUMBER,
Figure 19 Desinent cavitation as a function of Reynolds number for 1/8 caliber ogives
(Arakeri et al 1978).
1.0
0.5
0
- G>
'- 'J
1.0 INVISCID FLOW
-0.5
0.5
-1.0
4Im. TRANSITION TURBULENCE
REGION (PRESSURE FLUCTUATION)
8- 0
J
..
-1.5
'0.5
-2.0 MODEL
-1.0
-2.5 1.0
0 0.5 1.0
X/C
a. Laminar Seoaration b. Transition Effects
Figure 20 Viscous effects on cavitation on streamlined bodies (Huang & Peterson 1976).
294 ARNDT
- (19)
or if it occurs at the trailing edge of the separation bubble,
(Jc = Cp s - Cpts (20)
by University of York on 08/08/13. For personal use only.
about Cptt and Cpts. Suggested values are - 0.1 and - 0.35, respec
tively.
The effect of viscosity is to change the magnitude of (Jc in model and
prototype as well as the position of cavitation. Viscous effects can
dominate at model scale and are relatively unimportant in the proto
type. Typically, the flow can be in a state of tension in the model (if
(Jc< Cpm), whereas the critical pressure in the prototype is typically Pv
-
CAVITATION 295
FLOW
DIMEN- DATA
SYMBOL IRREGULARI T Y SIONS SOURCE 0 b C
t:J. TRIANGlES TWO 1-lCU1960 0361 0.196 0.152
h[j";
;;
-.I
hh.o.1751
0
1--1---/ ,
CIRCULAR TWO HClLL.1960 0.344 0.267 0.041
ARCS
-i
'Jl1-"j
& HEMISPHERfS THREE &NSal 0.439 0298 00108
h oO.5d
1966
POO328 n::q
hod
d ):."r
0.632 0.451
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
": , . ,
1966
POOO+
by University of York on 08/08/13. For personal use only.
---I
1 -
CTCfp
Ql0
* BOliN -1972
NACA 16-012
HYDROfOIL TESTS
hO U. b
CTcfp = C(-Hlt.)
& "
0.01 --1 I
1.0 10 100 1000 10.000
and the skin friction coefficient, Cr, has been determined (Arndt &
Ippen 1968):
(1cfp = 16Cf { distributed roughness } . (30)
Thus, the best available method for estimating the effect of surface
finish is through the use of Equation (26). (1cfp can be determined from
experimental data [Equation (29)] or from Equation (30), whichever is
appropriate. Holl (1965) presents a semi-empirical method for estimat
ing (1efp' An application of this method for predicting the inception
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
7.0
NACA 16 - 212 HYDROfOIL ( a 'd') U"'C
h
.. -
U", -
l C J.
r--
3.0
2.0
ClX10'3
1.0 cre -Cpm 0.409
-t'CONSTANT .0.004
I I I I
0.2 04 06 08 1.0
.!.
e
5.0
NACA 16 12 HYDROFOIL WITH
DISTRIBUTED ROU(;HNESS
4.0
30
ereR
ere
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
2.0
1.0
by University of York on 08/08/13. For personal use only.
Vortex Cavitation
Vortical motion of all types is common in fluid machinery and hy
draulic structures, in tip vortices on propellers, and in the clearance
passages of pumps and turbines, hub vortices trailing from propellers
and Kaplan turbines, secondary flows in corners, etc. Vortex cavitation
is a major consideration for erosion in ducted propellers, surging in
draft tubes, propeller-induced hull vibrations, etc. There are two signifi
cant aspects to this problem from a cavitation point of view: the
minimum pressure in the cores of vortices can be very Reynolds-number
dependent and cavitation nuclei can be entrained in the low-pressure
core for relatively lengthy periods of time.
The minimum pressure in a vortex can generally be predicted from a
Rankine model
wherein r is the circulation in the flow and 'c is the core radius. The
radius of the core is generally unknown and related to several parame
ters. In addition, it is often difficult to estimate r, especially for
CAVITATION 299
(32)
by University of York on 08/08/13. For personal use only.
- 10.0 1 I T ' 1 1 1 1 1 T
t
TESTS
-IN
-5.0
o
_O
....
& __ ) 0
a = 5
WATER TUNNEL TESTS
c..
<.J - L O
is
/:'" 'n'
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
AT THE HUB, 0
TUNNEL
WALL
p{
by University of York on 08/08/13. For personal use only.
PRESSURE TAP.
- 0. 1 5 I I I . 'II I
wherein K is an unknown factor (0.3 < K < 1), a is the dissolved gas
content, and f3 is Henry's constant.
Figure 25 contains cavitation data for the same configuration shown
in Figure 24 obtained at various velocities and with water containing
varying amounts of gas. In all cases the cavitation numbers are higher
than the absolute value of the pflssure coefficient measured at the cone
CAVITATION 301
-
I , r-
0::
THEORY
UI
III 6 + . K o.
l
1 2 P V;'
o; - C
l:
P
::l
Z WHERE KoO.4
. 7 iii '" 8.4 PPM (MocE !!IASIS)
OESINENT
U
CAVITATION 0 ", . 2.1 PPM
Z 6
UI
o
5
8
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
II! 4
IE 3
PRESSURES MEASURED UPSTREAM TIP OF CONE
by University of York on 08/08/13. For personal use only.
I I I I
10 15 20 25 30 35
VELOC ITY, Voo (tt/SI!IC)
Figure 25 The influence of dissolved gas on hub vortex cavitation (Holl et a1 1972).
tip and in most cases higher than the calculated values based on
upstream pressure measurements. The data are well represented by
Equation (33), using K = 0.4. The theoretical curves are based on calcu
lated values of Cp o It is interesting to note that during the entire test
program, cavitation was found to occur at critical pressures higher than
the vapor pressure.
(35)
wherein -
Cpm is usually negligible. The factor C1 takes into account
CAVITATION 303
both the statistics of the pressure field and any effects associated with
the influence of turbulence on the nuclei size distribution and their
response to the unsteady pressure. As already discussed, nuclei size
influences the critical pressure. Negative pressure peaks that induce
cavitation should have a statistical occurrence that is relatively high,
and a duration that is comparable to the natural period of the nuclei for
cavitation to occur at a local pressure approaching the vapor pressure.
Thus C1 is a factor involving both the probability density and frequency
content in the pressure field. If all the energy in the pressure field has a
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
frequency content much higher than the natural frequency of the nuclei,
there would be a lowering of (Jc associated with a lower value of Per'
Hence, one could conclude that if the turbulent intensity remains
constant, a reduction in the scale of a turbulent shear flow could reduce
by University of York on 08/08/13. For personal use only.
O'c ' Generally, this is unlikely (cf Arndt & Daily 1969), but as already
noted (Jc does vary with orifice size (Lienhard & Goss 1971). Measured
values of C1 range from about 7 in a turbulent boundary layer (Arndt &
Ippen 1968) to about 10 in a jet (Rouse 1 953).
The theoretical determination of a turbulent pressure field is ex
tremely complex. However, considerable progress has been made over
the last five years in the understanding of turbulent pressure fluctua
tions in free shear flows in an Eulerian frame of reference. Of particular
importance is the development of pressure-sensing techniques which
under certain circumstances can lead to reliable measurements of pres
sure fluctuations. The first theoretical arguments on the pressure fluctua
tions associated with turbulent flow appear to be due to Obukhov and
Heisenberg (Batchelor 1 953). Heisenberg argued that Kolmogorov scal
ing should be possible for small-scale pressure fluctuations. Batchelor
(195 1) was able to calculate the mean-square intensity of the pressure
fluctuations as well as the mean-square fluctuating pressure gradient in
a homogeneous, isotropic, turbulent flow. This work was extended by
Kraichnan (1956) to the physically impossible but conceptually useful
case of a shear flow having a constant mean velocity gradient and
homogeneous and isotropic turbulence. Apparently there were no at
tempts made to extend this theoretical work until the 1970s when
George (1974) and Beuther et al (1 977a, b, George & Beuther 1977)
applied the concepts developed by Batchelor and Kraichnan to the
calculation of the turbulent pressure spectrum in homogeneous, iso
tropic, turbulent flows with and without shear. When compared with
experimental evidence gathered in turbulent mixing layers, the theory is
found to be remarkably accurate.
All this work indicates that relatively higher-intensity pressure
fluctuations can be expected in a shear flow than in the idealized case of
304 ARNDT
1.- = C _u
'
where
{ C
1 .4 isotropic flow,
(36)
"iPU2 U2 C > 1 .4 shear flow.
-- -0.02
by University of York on 08/08/13. For personal use only.
THEORY
========---0007
p'
.04 =::::::-
K
0.0.00247
0. 1/2pUj2 -
Where
YIR -- _ __
K' L ( )
Y/R
L--------------4 I ,
2 3 5 6 7 8
;(/0
Figure 26 Rms pressure intensity in the :mixing zone of a submerged jet, theory and
experiment (Barefoot 1972).
- IX
- ..=>
.. (75)
Q. N
a. Q... -6.00
co
2
-7.00
turbulent pressure field for boundary layers than for free turbulent
shear flows. Not only is the theoretical problem made more difficult
(impossible to the present) by the presence of the wall, the experimental
problem is considerably complicated by the dynamical significance of
the small scales near the wall. Thus, in spite of over two decades of
concentrated attention, we cannot say with confidence even what the
rms wall pressure level is, although recent evidence points to a value of
(Willmarth 1975)
p' = Cp u . 2 , C = 2 to 3, (37)
where u. = V'I'/p . The basic problem is that the most interesting part
of a turbulent boundary layer appears to be the region near the wall
where intense dynamical activity apparently gives rise to the overall
boundary-layer activity. While the details of the process are debatable,
most investigators concur on the importance of the wall region in
overall boundary-layer development. Unfortunately, under most experi
mental conditions, the scales of primary activity are smaller than
standard wall pressure probes can resolve. Thus we have virtually no
information concerning the contribution of the small scales to the
pressure field. Arndt & George (1978) suggest that the small scales are
significant and can possibly be a dominant factor in very-large-scale
flows (e.g. boundary layers on ships, spillways, etc).
Both experimental and theoretical information on the spectral char
acteristics of turbulent pressure fields are cast in an Eulerian frame of
reference. The problem of cavitation inception requires pressure infor
mation in a Lagrangian frame of reference. Arndt & George ( 1978)
point out that Kolomogorov scaling has been successful in an Eulerian
frame of reference and suggest that similar scaling will be valid for
Lagrangian spectra. Based on this assumption, they derived expressions
for the spectrum of a turbulent pressure field. However, there is no
experimental evidence available to SUDDort their findinl7!,:
306 ARNDT
.
1
The minimum pressure must fall below a critical level.
2. The minimum pressure must persist below the critical level for a
finite length of time.
u.2TB
-- < 1.
II
!
1.0 . '
, . . ' I
o Smooth, Daily a Johnson ('56)
II Sowteeth, Arndt
a Ippen ('68)
Boundary
Loyer Sand, Messenger ('68) ,
Sond, Huber 1'69)
Wire Screen, Bechlel ('70"
0.1
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
IOOO C,
boundary-layer flow
(39)
free-shear flows
In this figure, Cf is computed either from the measured wall shear in the
case of boundary-layer flows or from turbulence measurements made in
air at comparable Reynolds numbers for the case of a free jet and a
wake. The measured value of C is only significant for the case of the
p
disk wake and the pressure data were determined from the experimental
work of Carmody (1964). The available data seem to be well approxi
mated by the relation
(Jc +Cp = 16Cr (40)
which was originally proposed for boundary-layer flow by Arndt &
Ippen (1968). These- <!lata might seem to imply that a relatively simple
scaling law a:lready exists and then further imply that the previous
discussion in this paper on turbulence effects is superfluous. This is not
the case. Observations of the bubble growth in turbulent boundary
layers indicate that the growth rate stabilizes at a constant value during
most of the growth phase. Using the measured growth rate, the levels of
local tension are computed to be quite small, of the order of 20 to 100
millibar (Arndt & Ippen 1968). Observations of bubble growth in
308 ARNDT
turbulent boundary-layer flow over rough walls indicate the life time for
bubble growth is a fraction of the Lagrangian time scale, <5= 13/u',
actually of the order of h/u . Cavitation occurs roughly in the center of
the boundary layer with a tendency for the zone of maximum cavitation
to shift inward as (u. TB)lhs decreases from about 1.5 to approximately
0. 1 . In the cited boundary-layer experiments Cp is negligible. For this
case, uc 16Cf
The fact that a strong depend4nce on Reynolds number can be
observed even in free-shear flows has already been illustrated in Figure
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
14. These data were obtained in two water tunnels and a new depres
surized tow-tank facility located at the Netherlands Ship Model Basin.
The water-tunnel data are for cavitation desinence, whereas the tow-tank
by University of York on 08/08/13. For personal use only.
It was found that the tow-tank data agreed with this relationship at
relatively high Reynolds numbers. Equation (41) was developed from a
model that assumes laminar-boulildary-layer flow on the face of the
disk. It would be expected that this condition would be satisfied at
higher Reynolds number in a tow tank than in a highly turbulent water
tunnel. At high Reynolds number (and also high velocity, where gas
content effects are negligible), there is a continuous upward trend in the
data with increasing Reynolds number. This underscores the need for
further work as suggested in the introduction to this section.
To summarize, cavitation inception in turbulent shear flows is the
result of a complex interaction b:tween an unsteady pressure field and
distribution of free stream nuclei. There is a dearth of data relating
cavitation inception and the turbulent pressure field. What little infor
mation is available indicates that negative peaks in pressure having a
magnitude as high as ten times the root-mean-square pressure can
induce cavitation inception. This fact alone indicates that consideration
should be given to the details of the turbulent pressure field. It is
suggested that two basic factors related to the pressure field enter into
the scale effects. First, as the sl;:ale of the flow increases, cavitation
nuclei are relatively more responsive to a wider range of pressure
fluctuations. Second, the available evidence indicates that large devia-
CAVITATION 309
tions from the mean pressure are more probable with increasing
Reynolds number. This would explain some of the observed increases in
cavitation index with physical scale. In view of the almost total lack of
information on the statistics of turbulent pressure fields (aside from
some correlation and spectral data) and the potential importance of this
knowledge to understanding cavitation, it is strongly recommended that
further work be carried out.
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
DISSOLVED-GAS EFFECTS
shorter than typical times for bubble growth by gaseous diffusion. For
example, in the cited cavitation data, a typical residence time for a
nucleus within a large eddy is roughly I I 15 of a second. At a gas
content of 7 ppm and a jet velocity of approximately 10 mis, inception
occurs at a mean pressure equivaknt to a relative saturation level of
1 .25. Epstein & Plesset (1 950) show that for growth by gaseous diffu
sion alone, 567 seconds is required for a 10-3 cm nucleus to increase its
size by a factor of 10. However, the local pressure within an eddy is
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
much less than the mean pressure and highly supersaturated conditions
can occur locally. Arndt & Keller (1 976) also reported large gas-content
effects in their experiments with disks when the flow was supersaturated
(Figure 29). The magnitude of the effect also depends on the number of
by University of York on 08/08/13. For personal use only.
"
c: 5
I UnsaturatedI
c:
o
>4
o
l.)
20 J A -- 11, 12
Velocity , m /sec
Figure 29 Influence of dissolved gas on cavitation in the wake of a sged disk
(Arndt & Keller 1976).
CAVITATION 311
= 4 2D ! ( Coo - Cs ) [ ! +
'"R g
1
V7TDg! t
]. (43)
Inertial effects are neglected and the difference between the partial
pressure of gas and vapor and the external pressure is balanced by
surface tension [Equation (1 1)]. If one uses the perfect-gas law, the mass
of gas in the bubble at any instant is
3(Poo-Pv + 2SIR )
.
4
N= '3 7TR k() (44)
Differentiation of Equation (44) and equating to the right-hand side of
].
Equation (43) yield the bubble wall velocity
R = Dg1( Coo - Cs ) [ + 1
dt Poo-Pv 4 SIR R ocr;-: V7TUg lt
(45)
---;;e- + '3 M
Equation (45) can b e integrated after noting Cs is a function of bubble
l In this section C , e;" and Coo are expressed in terms of mass (k moles) per unit
g
volume. Normal water-tunnel practice is to express the concentration in ppm on a
mole/mole basis. To avoid confusion, a will symbolize concentration measured on a
mole/mole basis.
3 12 ARNDT
radius
2S
Pg Ii" +Poo -Pv
Cs = {i = {J (46)
(47)
In comparing the time for a bubble to grow to ten times its original
value (Rj Ro = 10), the ratio of growth times is
Ig (without convection) r
-=--------:-- =1 .93 ,v Pe (48)
tg (with convection)
CAVITATION 3 13
, _ Po -Pc (49)
(1 = -
U2
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
I
"2P 0
8
- _ .I r L
- ,
(I.
'
Physical SuCllion in Brennen s Experimenls
r
gos- water solution
u..... -r -------,L .. ,
----.I C
---I A
Boundary
__
_ - Turbulent
---+t
Layer
Boundary
Layer, 0= cons1CInt
\
><s __:____ ____
Xo gas only
Figure:30 lIiielilize'd entl'liinment model for heat and mass transfer effects (Brennen 1 969)
3 14 ARNDT
Brennen's calculations is
(50)
-
tude. The theory implies that when(ver a cavity is maintained in a flow
that is locally supersaturated (Coo Cs > 0) there will be diffusion of gas
into the cavity and Pc >PV ' This accounts for the observed effects of gas
content already cited. For example, Billet & Weir (1975) measured the
flow into a ventilated cavity on a zero-caliber ogive. They maintained
the cavity pressure at a value equal to saturation pressure for the
dissolved gas in the flow and determined the following empirical rela
tionship:
dN
= 0 .0042 P
g (L)
0 .69
RedO.16 Fr o.13U; d 2 (53 )
dt kO d
,
wherein Fr= uo / Vid Noting that Cs =pg/{3 and that Pc =Pg +Pv'
Equations (53) and (50) can be combined to yield
kO
K
'
0=0 + pCoo P ( 1 + <7,)1/2(<7')1/2 (54)
o 13
tP Uo2 Re .41 Fr .
where differences in geometry bttween the Brennen theory and the
Billet-Weir experiments are absorbed in the constant K. Equation (54)
implies a new scaling parameter for gas content (pa/ t p U02). Although
the experiments of Wade & Acosta. (1966) display the trend indicated by
Equation (54), no dependence on the parameter pa/ t pU02 is noted.
This question will have to be explored further.
In summary, cavitation inception can exceed the expected value by
an amount Kap/ t pul, 0.3 < K < 1 in vortical flows, separated flow,
CAVITATION 315
THERMODYNAMIC EFFECTS
A similar effect may occur for cavitation inception but the issue is
cloudier.
In both cases the key factor is the difference between Pv computed
from bulk temperature and the actual value of Pv within the bubble or
by University of York on 08/08/13. For personal use only.
cavity wall which is necessary for the supply of heat to vaporize liquid
as required during bubble growth or to make up vapor diffused into the
wake of a stable cavity. This whole question is of critical concern when
attempting to relate cold-water tests to cavitation inception in liquid
metals or to pump performance in hot water or in more exotic fluids
such as cryogenics. A question of economics is involved since cold-water
tests always give conservative results and a particular device may end
up being over-designed for the task at hand.
The B-factor theory was first offered as an explanation of the ob
served scale effect on performance breakdown (Stahl & Stepanoff 1956,
Stepanoff 1961, 1964). The essence of the theory is that there is a
ecessary temperature depression in the liquid to produce the vapor.
Thus the vapor exists in equilibrium at temperatures less than the bulk
temperature of the liquid. This results in a correction to (J (always
positive):
o
= Pv =
B
( )
Pv 2 "A?
(55)
IP U2 P O,cpp 2
B for
aa=O.5
Substance Temp. @3.5 m/s
(56)
, PO -Pc
(1 = - (57)
I
"2P U02
wherein Ie is cavity length and Pc is the cavity pressure. It is further
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
(60)
Nv+ N
Qr- - = --g
NT
(61)
Pc Pv +Pg
318 ARNDT
or
Ng
Pg = . ' - -. Pc ' (62)
Nv + Ng
Nv
Pv = . . Pc (63)
Nv +N
g
where Nv and Ng are the mass-flow rates of vapor and non-condensible
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
gas (in moles per second). In the ca.se of a ventilated cavity, there are
two sources of non-condensible gas: from the ventilation system and via
the process of diffusion across the cavity wall. Ideally, the cavity can be
ventilated at a partial pressure Pg su(:h that the saturation concentration
by University of York on 08/08/13. For personal use only.
at the cavity wall is identical with the concentration in the free stream
(Billet & Weir 1975):
cs = P/3g (64)
Pg
l, Ng l . (66)
Pv Nv
Thus the mass flow of vapor into a cavity under natural conditions can
be estimated by simulating the same cavity development (same a') by
ventilating and measuring the ventilation flow rate. The assumption is
then made that the mass flow of vapor in a natural cavity is equal to the
gas-flow rate measured in a ventilating cavity. The unknowns in a given
problem are CA , CQ , and Nu. The:se can be measured and correlated
with leiD, Reynolds number, Froude number, Weber number, and
Prandtl number. Substitution of the empirically derived coefficients in
Equation (58) yields the temperature drop, M. Comparison between
theory and experiment is presented in Figure 3 1 for two different ogives
at two different velocities. Good agreement between theory and experi
ment is achieved. Unfortunately, these are simple flow configurations
and the results cannot be directly e:xtrapolated to the case of a pump or
turbine. The relationship between fl(J and flO can be determined from
the Clausius-Clapeyron equation
(J 2pv flO
P U2 Boo .
A = (67)
CAVITATION 319
c
9J
by University of York on 08/08/13. For personal use only.
B
o CAVITY
---
- tol
U.., -
8
.LJ
R - OI4
8 - - _
.., i:=l::J
6
4
VElOC lTV ' 36.6 m/set
I I I I I
o
40 50 60 70 80 CJO
FREE-STREAM TEMPERATURE, B ... Ideg CI
Pi =Pv( OR ) + Pg , (68)
[ 1
correction is (Plesset 1957)
R ma:x Pv dpv
IIpv =
(69)
dO
3 YDItm ('pI P I
320 ARNDT
y} PV;P
inertial effects dominate,
R= ; (70)
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
R = '" If O - OC R ) (7 1 )
V; APv v'D:i
by University of York on 08/08/13. For personal use only.
A dfv
B =
'
Pv
K
{ 1
1
inertial effects dominate,
thermodynamic effects dominate.
PI
N o concrete evidence i s available t o support this conjecture. Holl &
Kornhauser (1970) defined cavitation inception to occur when a given
nucleus grows to a macroscopic s ize in the time available for growth.
Their calculations of bubble growth including heat-transfer effects indi
cate that there should be a strong thermodynamic effect for inception
on hemispherical-nosed bodies in hot water and in Freon 1 1 3. However,
the experimental results were found to agree more closely with isother
mal calculations. Presumably other effects, such as noncondensible gas
and variations in viscosity with temperature, reverse the predicted
decrease in Dc with temperature (Holl et al 1 972, 1 976). Thus the subject
of thermodynamic effects on inception remains open to further investi
gation.
CONCLUDING REMARKS
ACKNOWLEDGMENTS
This article was prepared under grants from the US Air Force of
Scientific Research and the Seed Research Fund of the St. Anthony
Falls Hydraulic Laboratory. A special word of thanks should go to
Professor Dr. Ir. J. D. van Manen and Dr. Ir. M. W. C. Oosterveld, of
the Netherlands Ship Model Basin, for making possible a sabbatical
leave that allowed time for a portion of the literature review. Dr. Ir. J.
H. J. van der Meulen, also of NSMB, was my gracious host and
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
Literature Cited
Acosta, A. J. 1973. Hydrofoils and hydro Flows, ASME, New York, pp. 1 15-20
by University of York on 08/08/13. For personal use only.
foil craft, Ann. Rev. Fluid Mech. 5 : 161- Arndt, R. E. A. 1 974a. Cavitation incep
84 tion and how it scales: A review of the
Acosta, A. J. 1974. Cavitation and fluid problem with a summary of recent re
machinery. Cavitation, pp. 383-96. Conf. search, Proc. Symp. High Powered Pro
held at Heriot-Watt Univ., Edinburgh, pulsion of Large Ships, Wageningen, The
Scotland, Sept. 1974. London: Inst. Netherlands, Part 2, pp. XXI. 1 -65
Mech. Engrs. (available from the Netherlands Ship
Acosta, A. J., Hamaguchi, H. 1967. Cavita Model Basin).
tion inception on the I.T.T.C. standard Arndt, R. E. A. 1974b. Pressure fields and
head fonn. Calif. Inst. Technol. Hydro cavitation. Trans. 7th IAHR Symp.,
dyn. Lab. Rep. No. E-149 Vienna, Part I, pp. IX, 1-20
Acosta, A. J., Hollander, A. 1 959. Remarks Arndt, R. E. A. 1976a. Cavitation on model
on caviation in turbo machines. Calif. propellers with boundary layer trips.
Inst. Technol. Eng. Div. Report No. 79. 3. ASME Cavitation and Polyphase Flow
Acosta, A. J., Parkin, B. R. 1975. Cavita Forum, pp. 30-32
tion inception-a selective review, J. Arndt, R. E. A. 1976b. Semiempirical anal
Ship Res. 1 9 : 1 93-205 ysis of cavitation in the wake of a sharp
Arakeri, V. H. 1973. Viscous effects in in edged disk, J. Fluids Engrg. 98:560-62
ception and development of cavitation on Arndt, R. E. A. 1978. Investigation of the
axi-symmetric bodies. Ph.D. thesis Calif. effects of dissolved gas and free nuclei
Inst. Techno!., Pasadena, Calif. on cavitation and noise in the wake of a
Arakeri, V. H. 1975a. A note on the transi sharp edged disk. Proc. Joint
tion observations on an axisymmetric IAHR/ASME/ASCE Symp. Design and
body and some related fluctuating wall Operation of Fluid Machinery, Colo. State
pressure measurements. J. Fluids Engrg. Univ., Ft. Collins, Colo. Vol. II, pp.
97:82-87 543-56
Arakeri, V. H. 1975b. Viscous effects on Arndt, R. E. A. 1979. The influence of
the position of cavitation separation from dissolved and free gas content on cavita
smooth bodies, J. Fluid Mech. 68: 779-99 tion. Boeing Co. Doc. No. D 324-51505-1
(3 plates) Arndt, R. E. A., Daily, J. W. 1969. Cavita
Arakeri, V. H., Acosta, A. J. 1973. Viscous tion in turbulent boundary layers. Cavi
effects in the inception of cavitation on tation State of Knowledge, pp. 64-86.
axisymmetric bodies, J. Fluids Engrg. ASME
95 : 5 1 9-27 Arndt, R. E. A., George, W. K. 1978. Pres
Arakeri, V. H., Acosta, A. J. 1976. Cavita sure fields and cavitation in turbulent
tion inception observations on axisym shear flows. 12th Symp. Naval Hydrotiyn.,
metric bodies at supercritical Reynolds Washington, DC, pp. 327-39
numbers, I. Ship /{es. 20:40-50 Arndt, R. E. A., Ippen, A. T. 1968. Rough
Arakeri, V. H., Carroll, J. A., Holl, J. W. surface effects on cavitation inception. J.
1978. A note on the effect of short and Basic Engrg. 90:249-6 1
long laminar separation bubbles on desi Arndt, R. E. A., Keller, A . P. 1976. Free
nent cavitation. Proc. Meas. in Polyphase gas content effects on cavitation incep-
324 ARNDT
tion and noise in a free shear flow. Proc. mental investigations on incipient cavita
IAHR Symp. Two Phase Flow and Cavi tion in different liquids. ASME Pap. No.
tation in Power Generation Systems, 72WA /FE31
Grenoble, France, pp. 3- 16 Bossel, H. H. K. 1967. Inviscid and viscous
Arndt, R. E. A., Holl, I. W., 8ohn, 1_ c., models of the vortex breakdown phe
Bechtel, W. T. 1979. The influence of nomenon. PhD thesis. Univ. Calif.,
surface irregularities on cavitation per Berkeley, Rep. No. AS-67-14. viii + 78 +
formance. J. Ship Res. 23: 1 57-70 6O pp.
Baker, C. B., Holl, I. W., Arndt, R. E. A. Brennen, C. 1969. The dynamic balances
1976. The influence of gas content and of dissolved air and heat in natural cav
polyethelene oxide additive upon con ity flows. J. Fluid Mech . 37: 1 15-27
fined jet cavitation in water. ASME (1 plate)
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
Cavitation and Polyphase Flow Forum, 'Brennen, C. 1970. Cavity surface wave pat
pp. 6-8 terns and general appearance. J. Fluid
Barefoot, G. L. 1972. Fluctuating pressure Mech . 44:33-49 (2 plates)
characteristics of perturbed and unper Brennen, C. 1973. The dynamic behavior
turbed round free jets. M. S. thesis. Penn. and compliance of a stream of cavitating
by University of York on 08/08/13. For personal use only.
Chan, Y. Y. 1977. Noise generated wave Advisory Group for Aerospace Res. De
like eddies in a turbulent jet. AIAA J. velop. Conf. Proc., No. 4, Part 2, pp.
1 5(7):992- 1001 813-54. London: Technical Editing and
Chivers, T. C. 1969. Cavitation in centri Reproduction, Ltd.
fugal pumps. Proc. Inst. Mech. Eng. 184, Gates, E. M., Acosta, A. J. 1978. Some
Pt. I, No. 2 effects of several free-stream factors on
Daily, J. W. 1949. Cavitation characteris cavitation inception on axisymmetric
tics and infinite-aspect ratio characteris bodies. 12th Symp. Naval Hydrodyn.,
tics of a hydrofoil section. Trans. A SME Washington, DC, pp. 86- 108
7 1 :269-84 Gelder, T. F., Ruggeri, R. S., Moore, R. D.
Daily, J. W., Johnson, V. E. 1 956. Turbu 1966. Cavitation similarity considera
lence and boundary layer effects on cavi tions based on measured pressure and
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org
tation inception from gas nuclei. Trans. temperature depressions in cavitated reo
ASME 78: 1695- 1706. gions of Freon 1 14. NASA TN D-3509
Daily, J. W., Lin. J. Q., Broughton, R. S. George, W. K. 1974. The equilibrium range
1959. The distributio of the mean static of turbulent pressure spectra. Bull. Am.
pressure in turbulent boundary layers in Phys. Soc. 19: 1 158
George, W. K., Beuther, P. D. 1977. Pres
by University of York on 08/08/13. For personal use only.
tion hysteresis. J. Basic Engrg. 88: 1 99- Univ., Edinburgh, Scotland, Sept. 1974.
212 London: Inst. Mech. Engrs.
Holl, J. W., Wislicenus, G. F. 1961. Scale Johnson, V. E., Hsieh, T. 1966. The in
effects on cavitation. J. Basic Engr. fluence of the trajectories of gas nuclei
83:385-98 on cavitation inception. 6th Symp. Naval
Hydrodyn., Washington, DC, pp. 163-82
by University of York on 08/08/13. For personal use only.
Shen, Y. T., Peterson, F. B. 1978. Unsteady additives. PhD thesis. Univ. Twente, The
cavitation on an oscillating hydrofoil. Netherlands
12th Symp. Naval Hydrodyn., Washing van der Meulen, J. H. J. 1978. A holo
ton, DC, pp. 362-84 graphic study of the influence of
Silberman, E., Schiebe, F. R. 1974. A boundary layer and surface characteris
method for determining the relative cavi tics on incipient and developed cavita
tation susceptibility of water. Cavitation, tion on axisymmetric bodies. 12th Symp.
pp. 101-8. Cont. held at Heriot-Watt Naval Hydrotlyn ., Washington DC, pp.
Univ., Edinburgh, Scotland, Sept. 1974. 433-51
wndon: Inst. Mech. Engrs. van Wijngaarden, L. 1967. On the growth
Stahl, H. A., Stepanoff, A. J. 1956. Ther of small cavitation bubbles by convec
modynamic aspects of cavitation on tive diffusion. Int. J. Heat Transfer
Annu. Rev. Fluid Mech. 1981.13:273-326. Downloaded from www.annualreviews.org