MODELING A CAVITATING
VORTEX
Prof.dr.ir. G. Kuiper, TUDeift I MARIN
Report 1119-P
June 1997
ASME Fluids Engineering Division Meeting,
FEDSM'97. June 22
- 2,, 1997
TU Deift
Faculty of Mechanical Engineering and Marine TechnologyShip Hydrornechanics Laboratory
NOMENCLATURE
Martijn X. van Ri Jsbergen Propulsor Group
Maritime Research institute Netherlands 2, Haagsteeg, P.O. Box 28 6700 AA Wageningen, The Netherlands Phone: +31 31,7 493441, Fax: +31 3i7 493245
eMail: mvrijsberg@marin.nl
ABSTRACT
The purpose of this paper is to develOp a description of a cavitating vortex at various pressures. No consistent model for
such a situation is available in literature. To develop such a
description, systematic tests in
the MARIN high speed
cavitation tunnel have been carried ot with a cavitating hubvortex.
The thickness of the cavitating core and the tangential
and axial veloctity distribution are measured at several tunnel
pressures at a Reynolds number of 6.5 . iO5.
The radial momentum equation is used to integrate the
tangential velocity distribution to give a relation between the
cavitation number and the thickness of the cavitating core.
Taking conservation of kinetic energy and angular
momuentuni into account the relation between the tangential velocity distribution and the cavitating core diameter has been
investigated.
a,, cavitating core radius Cg gas concentration in water Cp,,, minimum pressure coefflcient D diameter of the test section E kinetic energy
L angular momentum
Po pressureatr = R
vapor pressure
r radius
R radius of the test sect ion
U axial velocity in the test section tangential velocity in the testsection
[m] [ppm] [-i [m] [J] [kgm2s I [Pa] [Pa] [m] [ml [mis] [mis]
1997 ASME Fluids Engineering Division Summer Meeting FEDSM'97 June 22- 26, 1997
FEDSM97-3266
MODELING A CAVITATING VORTEX
Gert Kuiper ITUDeift
R&D Department
Maritime Research Institute Netherlands
2, Haagsteeg, P: Box 28
6700 AA Wageningen, The Netherlands Phone: ±31 317 493273, Fax: ±31 317 493245 eMail: g.kuiper@marin.nl V, et 13 V p a a1
tangential velocity in the test section angle of attack
Henry's constant
wavelength
dynamic viscosity of water density of water
cavitation nümber
PoPu
cavitation number at inception
pU2
INTRODUCTION
Cavitation On a ship propeller begins nearly always with cavitation in the tip vortex. The ship speed at which cavitation
begins 'is called the inception speed. For Navy ships this
inception speed is very important because cavitation increases
the radiated noise significantly It is therefore important to
contrOl cavitation inceptionof the tip vortex.
Tip vortexcavitation isa most complex form of cavitation on a propeller A drastic simplification is the two dimensional cylindrical vortex. Even after this simplification, inception of cavitation remains rather 'intractable because it depends on the Reynolds number and on the nuclei content of'the flow. Since neither parameter can be properly scaled, strong scale effects occur when model tests are carried out, see ITFC (1993); In
practice the nuclei content of a tunnel is kept as constant as
possible and a simple emperical relation between the incepient cavitation number and Reynolds number is used (McCormick,
1962).
The condition at which the cavitating vortex disappears
[mis] [deg] [Pa/ppm] Em] [rn/S] [kg/rn3] E-] [-] Copyright© 1997 by ASME
Figure 1. Adaption of test section for LDV
(desinence) can be different from the inception condition
(hysteresis). However, there are indications that the radius of a cavitating vortex is uniquely related to the cavitation number and that this relation is independent of the Reynolds number and of the nuclei content (Kuiper,198l). Measurements of the
cavitating core are also given by Gowing et al. (1995). They
also added polymers and this did change the mentioned
relation, which may indicate a Reynolds dependence. BothKuiper and Gowing observed that the radius of the cavitating
core was rather independent of the position behind the tip,
although Gowing points out that at high air contents, diffusion occurs and the pressure in the cavitating core becomes higher
than the vapor pressure.
The relation between the radius of the cavitating vortex
and the cavitation index may be used to define an inception
condition from the cavitating tests. Very often the radius of the
cavitating core is taken at the location where the vapor
pressure occurs, e.g. Arndt and Keller (1992). In that case,using a Rankine vortex, angular momentum is conserved but kinetic energy is not conserved, however.
Therefore this relation between cavitating core radius and
pressure (cavitation number) has been investigated in more
detail.
EXPERIMENTS
In the present research the behaviour of a cavitating hub
vortex is studied, instead of the (usual) tip vortex of an
elliptical foil, studied e.g. by Gowing (1995). The high speed cavitation tunnel of MARIN is particularly suitable to generate this kind of vortex. Because of it's small test section, both the tangential velocity distribution and the cavitating core radius are relatively easy to measure.
The tunnel, build of 5" stainless steel pipes, is driven by a centrifugal pump which induces velocities in the test section up to 60 ni/s. A heat exchanger can be used at high speeds to keep the tunnel water temperature constant. The test section is
a ventury with a contraction ratio 1:10 and has a constant circular cylindrical diameter of 40 mm over a length of 60
mm. The walls of the test section are adapted to make laser
Figure 2. LDV setup with: 1=Iaser, 2=mirror box, 3=beam spacer, 4=rotatlng device, 5= focussing lens
velocimetry possible, see Fig. 1.
The laser beams enter the test section through a flat
perspex plate, followed by a box, filled with water. Next, they pass through a film with a thickness of 0.2 mm to finally enter the flow. Thus the refraction of the beams in the test section is minimal.
Generation of a hub vortex
A vortex is created in the test section by a stator
positioned in the stilling section, upstream of the test section. The stator consists of four stator blades mounted on a central
hub of radius 0.2 R. The stator blades have a NACA 0020
section at a fixed angle of attack of 15°.
During the first tests of the experiments it was observed that the cavitating core wandered too much. This wandering was reduced by giving the trailing edge of the hub of the stator
a flat back (instead of a sharp point), thus creating a sharp
edge where separation is well defined. A further reduction of the wandering was attained by placing a receiving stator in the difusor with a sharply pointed hub in the upstream direction. In this way the vortex core was guided to the centre of the test
section.
Measuring tanqentlal and axial velocities
In Fig. 2 a diagram of the total LDV setup is given. The laser used in the experiments is a HeNe laser with a wavelength ? = 632.8 nm, a beam diameter of 1.0 mm and a measuring control volume of 0.lxl.0 mm.
To measure the axial as well as the tangential
velocitycomponent with a 1D laser, the rotating device is used. This device can rotate the two parallel beams over 90°. In this way
the x- and z-component of the velocity distribution can be measured.
2
7 6 2 1-o OI
1
- U.
,
o , o
Vt calculated standard deviation VI 8 7 U)>3
2 o o U 0.005 0.01 rEm] Vt calculated o standard deviation VtU.
I
00, 0 0.015 0.02Figure 5. Tangential velocity distribution with correct centre position
Data reduction
In
a non cavitating
vortex, the tangential velocitydistribution can be directly measured by measuring the vertical velocity on the horizontal axis through the core of the vortex,
see Fig. 2. But when the core of the vortex is cavitating, it
disturbs the forward scattered light. Hence the vertical velocity
component is measured off-axis and the tangential velocity
must be calculated. To do this, the vertical velocity component, the position of the measuring volume and the position of the centre of the vortex must be known. Also the assumptions must
be made that the real
tangential velocity distribution isaxisymmetric and that there are no radial velocities.
The mean position of the centre of the vortex is difficult to determine visually, due to the wandering of the vortex core. Therefore vertical velocity measurements were made on
trajectories parallel to the y-axis, at about one maximum core diameter above and below the vortex core.
The exact position of the centre of the vortex is found by
visual check of the appearance of the tangential velocity distribution, calculated with a chosen centre position. The correct centre position is defined as the position where the
chosen coordinates of the centre position give a smooth curve for every tangential velocity distribution, compare Figs. 4 and 5.
3 Copyright © 1997 by ASME
7 Figure 4. TangentIal velocity distribution with incorrect centre position
4
eFigure 3. Typical cavitatlng core surfaces
From the position of the measuring control volume, the
position of the centre of the vortex and from the measuredvertical velocity component, the tangential velocity distribution can be derived.
To examine the velocity distribution accurately, the whole LDV setup can be traversed in z and y direction with steps of 0.1 mm.
The forward scattered light
is focussed on a photo
detector, the signal of the detector is processed by a tracker and
the analog output of the tracker is used as the input for a
measurement pc. Most measurements are done at a sample rate
of 16 kHz with a number of samples of 8000. After visual check of the signal, the mean value of the velocity and it's
standard deviation are calculated.
Measurinq the cavitating core diameter
Observation of the cavitating core with the naked eye
using stroboscopic light showed substantial disturbances on the
surface of the cavitating core, see Fig.
3. Therefore themeasurements of the cavitating core diameter where done by
making stroboscopic video recordings with 50 frames per
second to have some redundacy in
the number of the
recordings.
As a reference, recordings of a ruler placed in the test
section were made to determine the real diameter of the
cavitating core.
To observe the dynamics of the disturbances, high speed video recordings were also made. Figure 3 shows eight stages which are placed in order by high speed video observation at 27 kHz. These observations show that even though a diameter can be determined
at many times (frames
i to 3), significantdisturbances occurred, which were moving with the flow.
I
60.015 0.02
0.005 0.01
0.102 £ 0.114
o
.0.645 * 0.653
Figure 6. Measured tangential velocity distributions at several cavitation numbers
Experimental results
In total 6 tangential velocity distributions were measured
at 6 different cavitation numbers, the Reynoldsnumber was
kept constant at 6:5. iO5. The gas concentration was between 3.9 and 5.3' ppm. The cavitation number is defined as:
PoPv
wherein P0 is the static pressure at the test section wall, Pv is the vapor pressure and U is the mean axial velocity measured
by the LDV.
The Reynoldsnumber is defined as
UD
ReV
wherein D is the diameter of the contraction and y is the
kinematic viscosity. Figure 6 shows that the ambient pressure
(and thus the thIckness of the cavitating core) has
little influence on the tangential velocity distribution. Figure 7 shows the region close to the core in detail. Arrows indicatethe radius of the cavitating core a
at several cavitationnumbers and some indicative lines are given.
For O. I <a <0.4 'the gradient of the tangential velocity 'is already very high at r '2a. Due to the wandering of the vortex core in this range of cavitation numbers, no tangential velocity
measurements could be made for r < 2a,
(2)
Figure 7. Comparison of measured tangential velocity distributions
Figure 8. Measured axial velocity distributions
For a = 0.645 the tangential velocity distribution seems
to remain constant and data are found even inside the
cavitating core. This is due to the high intermittancy of the
cavitating core, see frame 5 of Fig. 3.
The axial velocity distributiöns show the opposite; At cavitation numbers of 0.1 to 0.2 the axial velocities seem to
remain constant. But at cavitation numbers of 0.4 to 0.65 the axial velocities near the core are up to 13% higher, see Fig. 8.
The diameter of the cavitating core is measured as a
function of the cavitation number, mainly at a constant
Reynolds number of 6.510, with different gas concentrations
of the water (Cg = 3.9, 4.1, 5.3, 6.5, 8.5 and 10.8 ppm), see
Fig. 9. The diameter measurements were done in a a interval
which was limited on one side (at small cavitation numbers
and large core diameters) by vortex breakdown. On the other side (at high cavitation numbers and small core diameters) the
a interval was limited by absence of cavitation. The largest
value of a at which core diameter measurements were carried out was at the condition where a minimum of 10 cylindrical
181 17 16 19 15 V VF -VA V - V
..ii'
7i.
D.
IS..
0.102 ,o 0.205 'A 0.409 y 0.645Standard DeviatiOn (Va)
4
05. o.
f pU2
(i)
o 0.002 0.004 0.006 0.008 r Em]5 4.5 E 3.5 8
o,3
C (0 2.5 w w 1.5 E w 0.5 o OD a * A A..
$.
s IO s'
a Cg =3.9 ppm Cg=4.1 ppm a Cg=53 ppm a Cg=85 ppm e Cg=8.5 ppm a Cg =10.8 ppm Stamiard dedation«)) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 11 o oFigure 9. Diameter of the cavitating core as a function of the cavitation number
bubbles in
20 sec
of video
recording were observed(desinence).
With a high gas concentration in the water, inception
occurred at larger values of the cavitation number than with a
low gas concentration in the water. This influence on the
cavitation inception number by the concentration of dissolved gas in the water can bedescribedby
cY.=Cpmin+
¡3Gg(3)
The difference in a is calculated (using Henry's constant
6700 Pa/ppm and U = 164 ms) to be 0.34. This agrees well with the measured difference in a1 between the maximum and minimum gas concentration, see Fig. 9. Since in this testsetup the vortex existed for some time, diffusion effects can indeed
be present.
MODELING A CAVITATING VORTEX
Kûiper (198'!) tised aspirai vortex approach with three regions to describe a cavitating vortax. Two regions are distinguished in the present analysis:
i. A vapor core, treated as a solid core, rotating with an
outer 'velocity equal to the fluid velocity. The pressure at
this cavitating core radius a may be assumed to be the
vapor pressure P.
2. A rolled-up vorticity region, in which the tangential'
velocity distribution is' assumed to have the form
where,R is the radius of the test section.
The parameters a and b in this velocity distribution and' the cavitating core radius a are assumed to be a function of the
pressure P0 at r = R.,But since no forces, act on the vortex when
changing the pressure, there must be conservation of angular momentum and kinetic energy.
The pressure distribution in the vortex can be found by integrating the momentum equation in radial direction
¿IP
v2(r)
(5)
'r
Integrateting this equation using the tangential velocity
distributiOn of Eq. 4 over the region between r = a and R, gives the relation between the cavitation number and the
radius of the cavitating core:'
bU2 a a2
(1 (aC\2b
'R'
(6)The angular momentum per
unit length containedbetween the radii a. and' R is
L =
2,rpjv(r)r2dr
(7)The kinetic energy of the tangential velocity distribution per unit length contained between the radii a and R is
E =
lrpjv2(r)rdr
(8)Out of the measurements, five conditions are chosen at which the tangential' velocity distribution (Eq. 5) is determined with it's core radius. These' conditions are listediñ Tab. i.
This leads to three equations (Eqs. 5,7 and 8) with three unknowns'(a, b, as).
By regression' analysis, the parameters a and b are
determined using Eq. 6 to fit into the, measured core radii as a function of the cavitation ñumber, see Fig. 9. This resúlts in a
=084 m'' Isec andb = 0.61.
Copyright © 1997 by ASME a a s O .a
r
a < r'<'R (4') 0.8 E E 0.6 wI
0 0.4 w 0.2a iø5o 0645 0A09 0.205 0i'02 a [m] 2.5 lOE4 3.751O 5.75iO 8.5-1 4 1.251'OE3
Table 1. Conditions used ¡n modeling of the cavitating; vortex 14 12 10 E
4->6
4 2 -°-- 0.645 fr 0.409 -4-0205 -.0-0.102 is-Figure 10. Comparison of tangential veloóity distributions, calculated by exercise 1
Exercise I
In the first exercise these values of a and b are used to
describe the velocity distribution (Eq. 4) of the smallest
cavitating core radius ac. In this condition the energy equals 1.3 J (Eq. 8)and the angular momentum is 0.019 kgin2s' (Eq.7).
Maintaining this calculated energy and momentum in
combination with another cavitation number o, the new core
radius and the velocity distribution (described by the new
parameters aandb) are now calculated again.
This is done for the four remaining cavitation number&
The results are given in Fig. 10. This Figure shows that the
energy and the angular momentum remain near the core when
the pressure decreases. The calculated values of a
as a
functionof a are given in Fig. 11.
Exercise 2
In the second exercise a similar calculation of the energy
and angular momentum is made, but now based on the
conditiOn with the largest core radius. In this case the energy equals 1.19J and the angular momentum is0.019kgm2s'.
2.50E.03 2.00b03 5.00E-04 0.000.00
-j
s
noasured -e-- exercise 1 -û-- exerce2-
Fgression fitFigure 11. Comparison: of calculated values of a with
measured values
Figure 12. Comparison of tangential velocity distributions, calculated by exercise 2
It Should be kept in mind that this redúction in energy
(compared with exercise 1) is caused by the fact that the saine
velocity distribution (the same values a and b) is used at
different pressures..There is a rather large discrepancy between the measured radius and the radius from the regression line, (for a = OE102),
using these values of a and b (Fig.
11). This causes an overestimate of the cavitation number (a = 0 128) at this largest core diameter.14 Ici= ]_.o_o.128 12 i -Q-0205 -.6-0409 10 -0- 0.645 --1.06 I 4-0.000 0.005 0.010 0.015 0.020 r (m] 05 1.5 a o 0.000 0.005 0.010 r LrnJ 0.015 0.020
> 14.00 12.00 10.00 8.00 6.00 4.00 200 0.00 a I b 0=
I
II-O6S3(rm) . .tnd,(Vl) 0- O.ft, ,,g., -+-O.132(,,,,c 3) -fr-O2(.oro 3) I-°-O4O9(,*, 3) -a-0645(.,erc 3) -.- I OIO(,,,c 3) U O.1(n.) OE1I4(n.) O.2t6(n.) S O.4O9(n.) 84S(n.) 0.000 0.005 0.010 r (nil 0.015 0.020Figure 13. Comparison of tangential velocity distributions calculated by exercIse 3 withmeasured
tangential velocity distributions
The result of exercise 2 is given in Fig. i2 This figure
shows that the energy and the angulár momentum remain even closerto the core than with exercise 1. Thecalculatedvalues of a as a function of a are given in F:g. 11.
Exercise 3
Neither exercise i nor exercise 2 predict the measured
a,-a rela,-ation very well (Fig. 11). Therefore, a,-a third model ha,-as been tried,
to satisfy the measured values of a and the
measured a value.
From the measurements it is clear that only the velocity
distribution close to the core (inner region) changes as a
function of the pressure. Hence a fixed velocity distribution is
taken for 0.006 < r < 02 m (outer region). A concentric
tangential velocity distribution for both the inner and outerregion is now chosen according toEq. 9.
C
(9)
The parameters c and d for the
outer region are
determined by regression analysis.
14.00 12.00 10.00 ' 8.00 > 6.00 4.00 2.00 0.00
Figure 14. Comparison of tangential velocity distributions, calculated by exercise i with measured
velocity distributions
Table 2. Loss of kinetic energy and angular momentum
The parameters c 'and d for the inner region can now be determined from' the conditions that the velocity distribution at the inner/outer boundary is continuous and the pressure at' the core radius is equal to the vapor pressure. So the measured a-a rela-ation is ma-ainta-ained.
The values c and d in the inner region give the velocity
distribution forthe various pressures.
The results of this exercise are shown in Fig 13, denoted by
'(exerc 3)' together with the measured tangential velôcities,, denoted by (meas)'.
The subsequent loss of angular momentum and' kinetic
energy can now be cálculated and is shown in Tab 2. The percentages are normalized with respect
to the smallest
diameter (a =2.5lOE4 m).
The loss 'of kinetic energy, shown by Tab. 2, can also be seen in' Figure 13. The inner region velocity distribution of the
largest diameter, (a =0.102 (exerc. 3'))does not have velocities as high as the measured velocities (a =0.102 (meas.)).
7 Copyright © 1997 by ASME A a u O.It4)im.) O.2(rau.) -O.4(,mX.) 0645) - O653(nX) I) I) -I) 645(,x,,c I) --o-o7rL)ex=r. UI.OIIo,c.I) L X lId ':
IIi
---1 -a AE[%] AL[%] 1.050 0 o 0.645 0.8005
0.409 3.0006
0.2059l
0.32 0.102i68
0.76 0.000 0.005 0.010 r (mj 0.015 0.020DISCUSSION
To develop an alternative criterion for inception it is necessary to know the a-a relation in the inception range.
From Eq. 6 this relatiön can be written as
'C
a26 (10)
From the regression of the data, the value of b was found
:to be 61. Kuiper (1981) found values clOse to 05 for a
propeller tip vortex. The power b depends oñ the tangentialvelocity distribution, hence the difference in tangential velocity distributions between the two situations: hub vortex cavitatiòn
and propeller tip vortex cavitation, apparently are not very large.
The results show that
with the given parametricdescription of the tangential velocity distribution (Eq. 4), the cavitating core radius can be analytically calculated at various
pressures (Eq. 6). The effects of pressure changes on the
tangential velocity distributions could be estimated using conservation of energy and momentum, within reasonable accuracy.
In exercises 1 and 2, one point on the experimentally determined a- relation was taken and 'the other points were predicted using conservation of eñergy and angular
momentum. The result is given in Fig. i 1, in which also the
experimental data of the cavitating core radius are given with their95% confidence levels. Exercise I, which starts at a large
cavitation number overpredicts the core radius at smaller cavitation numbers Exercise 2 which starts
at a small
cavitation number,. underpredicts the core radius at higher pressures. Within' a limited range of pressure variations thecurves are reasonably well predicted'.
From Figs 10 and 12 it can' be concluded that at decreasing pressures the velocity increase occurs at radii close to the
cavitating core.
The technique
in which the velocitydistribution of various pressures is kept the same for all
pressures (Gowing, 1995) is therefore only slightly wrong in
this case.
In Fig. 14 the calculated velocity distribution from
exercise i (Fig. 10)
is plotted in combination with the
measured points. The shape of the curve is reasonably wellpredicted, although the calculated velocities are systematically
lower than the measurements The results of exercise 1 are
slightly closer to the measurements than those of exercise 2.. When the starting point is taken at the measured velocity distribution
in combination with the condition
that thepressure at a equals the vapor pressure, the velocity
distribution is not well predicted (Fig. 13').
The measured tangential velocities close to the core are
larger than predicted in' exercise 1
(Figs 7 and 14). This
means that even more energy is present around the core than predicted using conservation of energy in tangential direction.
From Fig. 8 it can 'also be seen that axial energy disappears
with decreasing pressure. This could show up in the tangential velocity distribution, but has not been modelled in the present
exercises.
REFERENCES
Arndt, R. E. A., and Keller, A. P., 1992, "Water Quality Effects on Cavitation Inception in a Trailing Vortex", ASME Journal of Fluids Engineering, Vol. 114 ¡pp. 430-438.
Cavitation Committee, 1993, "Report of the Cavitation Committee", International Towing Tank Conference, San Francisco.
Gowing, S., and' Briançon-Marjollet, L., 'and Frechou, 'D.,
'and Godeffroy, V., 1995, "Dissolved Gas. and Nuclei Effects on Tip Vortex Cavitation Inception and Cavitating Core Size International Symposium on Cavitation, Cav95, Deauville,
France.
Kuiper, G., 1981', "Cavitation Inception on Ship Propeller Models Thesis Technical University DeIft
Kuiper, , 1979, "Some Experiments with Distinguished
Types of Cavitation on Ship Propellers", ASME Internatiönal
Symposium on Cavitation Inceptiòn, 'New York.
McCormick jr, B. W., 1962, "On cavitation Produced by a Vortex Trailing from a Lifting Surface",, ASME Journal