Bubble Behavior in Hydrodynamic Cavitation:

Effect of Turbulence
Vijayanand S. Moholkar and Aniruddha B. Pandit

number of physical and chemical changes has been pursued

enthusiastically by sonochemists using ultrasound equipment
of various sizes, shapes and forms. The effect of cavitation as
observed in ultrasonic equipment has been attributed to the
zy
zyxw
Dept. of Chemical Technology, University of Bombay, Matunga, Bombay 400 019, India

Harnessing the energy associated with cavitation for a This analysis includes the computer simulations of modi-
fied bubble dynamics equation (Plesset, 1949) for hydro-
dynamic cavitation. The effects of various parameters which
affect the turbulence intensity (in terms of its magnitude and
frequency) and in turn its effect on the bubble behavior have
transient form of cavitation. Hydrodynamically generated been studied.
cavities are believed to behave in a stable cavitation mode
and, hence, are not very useful for the desired sonochemical Mathematical Formulationand Algorithm of
effects. Simulation
This work illustrates with numerical simulations the role
played by the turbulence in altering a single cavity behavior The specific assumptions in the mathematical modeling are
downstream of a cavitating orifice in a liquid flow. Specifi- as follows:

zyxwvu
zyxwv
cally, it looks into the effect of turbulent pressure fluctuations (1) A single bubble or cavity has been considered in isola-
in transforming otherwise stable cavitation into transient cavi- tion.
tation. (2) The cavitation medium is assumed to be inviscid and
Among the practical objectives of this study is a better un- incompressible and simulations are terminated at a point
derstanding of the bubble behavior in hydrodynamic cavita- where the instantaneous bubble size to the initial bubble size
tion, in particular the effect of various parameters such as ratio (R/R,) is 0.5 during the collapse.
intensity and frequency of eddies responsible for the turbu- (3) The relative motion between gas and liquid phase (that
lence downstream of the orifice that actually control the bub- is, the slip velocity) and hence the friction loss has been ne-
ble dynamics. Previously, the bubble dynamics problem has glected.
been analyzed on the basis of a single bubble and a linear (4) Heat- and mass-transfer effects have been neglected

z
pressure recovery profile downstream of the orifice neglect- due to very fast events.
ing the turbulent pressure fluctuations (Yan et al., 1988). This
approach may be adequate when the intensity of turbulence Cavitation inception
is quite low and thus pressure recovery from the vena con- A dimensionless parameter which has been widely used in
tracts to a downstream pipe position can be approximated by the study of hydrodynamic cavitation is the cavitation number
a linear expression with respect to a distance downstream (Ci) which is defined as

zyxwvutsr
of the orifice. It certainly loses validity when the intensity
of turbulence rises and pressure recovery is no more linear
especially with changing geometry of the orifice. In this case

zyxw
significant alteration are likely to occur in the local pressures
encountered by the vapor/bubble cavity during the passage
in the flow. In this article an attempt has been made to ana-
lyze the fundamental phenomenon, numerically with some where P2 is the recovered downstream pressure, and P, is
interesting results. This could be a useful basis for the effec- the vapor pressure of the cavitating medium at the operating
tive scaleup and design of hydrodynamic cavitation reactors temperature, p is the density, and v, is the average fluid ve-
by describing the role of downstream turbulence on the bub- locity at the orifice.
ble/cavity behavior. In the present analysis, we have assumed the cavitation
number to be 1 and the simulations are on that basis only.
Though in practice the cavitation inception number has been
Correspondence concerning this article should be addressed to A. B. Pandit. found to be a function of orifice diameter (Yan et al., 1988)

AIChE Journal June 1997 Vol. 43, No. 6 1641

 zyx
zyxwvutsr
zyxwvutsr
and could be greater than one indicating the influence of tur- For the estimation of the length scale ( 1 ) Prandtl eddy
bulent pressure fluctuations. model has been used. Prandtl eddy is a medium size eddy
Turbulence Modeling. For a pipe flow, the turbulent pres- and is also best suited for the geometry under consideration

zyxwvutsrqp
sure fluctuations are due to velocity perturbations as a result (since its size is a strong and only function of the diameter of

zyxwvut
of the formation of eddies. In turbulent flow, the instanta-
neous velocity in the x direction is given as

u, = 6, + n;
where 6, is the time-averaged velocity at any point in the
(2)
the conduit through which the liquid and eddies flow) was
selected for the analysis in this work.
According to this model the length scale is given as

1=0.08d

where d is the diameter of the conduit through which fluid

(8)

flowing fluid, and u; is the instantaneous fluctuating velocity.

and associated cavity flows. In the region near orifice where

zyxwvutsrq
It is convenient to express the amplitude of the fluctuating
fluid stream narrows down and flow area becomes equal to
velocities in the x direction as (6;)', which is the mean of the
the area of orifice the eddy size is 0.08 d o ( d o is the diameter
squares of the fluctuation velocities, this being necessarily
of the orifice). Thereafter, as the flow stream expands and
positive. Similar treatment can be given for velocity fluctua-
the flow area equals the area of pipe after full pressure re-
tions in other directions. The turbulent kinetic energy per
covery, the scale is 0.08 d , ( d , is the pipe diameter). To find
unit mass is thus given by
the frequency of turbulence (Eq. 71, the average length scales

zyxwvu
at these two extremes has been considered. Similarly, to esti-
K.E.= - (3) mate U' (Eq. 6) the same I has been considered with P, as
estimated earlier.
Bubble Dynamics. With the inclusion of surface tension
Hence, the rate of dissipation of turbulent kinetic energy is and viscosity effects the equation of bubble dynamics is given
given as by

zyxwv
I d
PM = --
2 dt
[
- (8;)' + q)'+
(8321 (4)

duces to

M-
zyxwvuts
which for isotropic turbulence (where 8; = 8; = VL = u ) re-

p ----(E')
3 d
2 dt
2

For isotropic turbulence, the fluctuation velocity 8' and the

length scale of eddy (1) can be defined in terms of the power
input per unit mass of the system P,,,. It is equal to the rate
This equation known as the Rayleigh-Plesset equation
forms the basis of this analysis (Plesset, 1949).
To confirm the validity of the numerical scheme, simula-
tions of bubble behavior reported by Miksis and Ting (1984)
were carried out with the current numerical scheme and these
were found to be identical.
For hydrodynamic cavitation in orifice flow, the mean pres-
sure recovery profile is assumed to be linear over which
turbulent pressure fluctuations can be superimposed. The
of dissipation of energy per the unit mass of turbulent fluid pressure P,, that is, the pressure at a large distance from the
and the relations are (Davies, 1972)

zyxw
bubble decides the bubble behavior which will change due to
the flow conditions assumed. Here the local pressure without
turbulence can be estimated on the basis of linear pressure
recovery as

The frequency of the velocity perturbations within the tur- (10)

bulent eddies is simply given as

8' where 7 is the pressure recovery time and P,,,,,,is the axial
fr= 7 (7)
pressure downstream of the orifice to be substituted for P, in
Eq. 9 for simulation.
In this analysis PM has been estimated by considering the The time for pressure recovery has been estimated by
permanent pressure head loss which is a function of the ratio Newton's equations. For an assumed recovery pressure (P,)
of the orifice to pipe diameter. The product of the head loss downstream of the orifice, the velocity near the orifice can be
and the volumetric flow rate will give the energy dissipation estimated from Eq. 1 for Ci= 1.0. This velocity reduces with
rate due to turbulence. The rate of energy dissipation due to an increasing distance downstream of the orifice and reaches
the pipe wall friction was found to be negligible ( 2 to 4% of the mean pipe velocity when full pressure recovery takes
turbulent energy dissipation rate) and hence neglected. This place. Pipe velocity has been estimated by equating volumet-
turbulent energy dissipation rate divided by the mass of the ric flow rates. Thus,
water in the region of pressure recovery (typically 8 pipe di-
ameters downstream orifice) gives p,.

1642 June 1997 Vol. 43, No. 6 AIChE Journal

 zyxwvutsrqpo
zyxwvu
zyxwvutsrqpo
where d , and d p are the diameters of the orifice and the
pipe through which the stream flows. With the knowledge of
u, and up the time of pressure recovery has been estimated
using Newton’s first and third laws of motion.
Using the mean pressure as given by Eq. 10 and the use of
increases. The life of the bubble/cavity also increases. The
increased recovery pressure increases both the intensity and
frequency of turbulence affecting the cavity in a manner simi-
lar to the changing acoustic field in the ultrasonic cavitation
case.
Bernoulli’s equation (between that particular point and a The drastic change in the cavity behavior can be seen by
point where full pressure recovery takes place) an estimation comparing the simulations with or without turbulence effects.
has been made of the local velocity at any particular point With turbulence, the oscillatory stable cavitation behavior
downstream of the orifice. The use of Bernoulli’s equation transforms into a transient cavitation, similar to the acoustic
for this purpose has been only at steady-state conditions as cavitation.
the turbulent pressure variations have not been considered
(they are likely to cancel out as uniform turbulence intensity
has been considered over the entire volume, which is consid- Effect of pipe-size downstream of or@ce
ered for the pressure loss or energy dissipation). The turbu- The simulations conditions have been listed in the caption
lent velocity fluctuations have been superimposed on it by

zyxwvut
zyxwvutsrqpon
zyxwv
of Figure 2. Figure 2 indicates that with an increase in the

zyxwvutsrqpon
assuming a sinusoidal velocity variation in the instantaneous pipe size the maximum cavity radius before collapse in-
local velocity with a frequency estimated by Eq. 7. Thus, the creases, resulting into a higher-pressure pulse. This alter-
new instantaneous local velocity is given as ation in the cavity behavior is due to the increased scale of
turbulence (assumed to be a function of pipe and orifice di-
u,, = u, + sin (2afTt) (12) ameters) which reduces the frequency of turbulence as the P,
or P, value is only marginally altered (the effect of skin fric-
where ur is the local mean velocity and u,, is a function of tion is small). The reduced frequency allows the cavity to grow
time with t being the numerical integration step. Use of this further resulting in a larger radius before the collapse. Again,
instantaneous velocity has been made to estimate the instan- a transformation from stable to transient cavitation is evident
taneous local static pressure using Bernoulli’s equation of the from this figure as a result of consideration of turbulence in
following form liquid.

Effect of /3
For a fixed pipe size two different values of /3 have been
This instantaneous pressure which involves the effect of considered for the simulation. Figure 3 indicates that with an
increase in p the life of the cavity before its violent collapse

zyxwv
turbulent pressure fluctuations is then used instead of P,,,,,”
given by Eq. 10 for the solution of Eq. 9 as a variation in P,. increases although the maximum radius of the cavity bubble
The Runge-Kutta fourth-order method was adapted for the is not altered significantly. The explanation for this effect can
numerical solution of Eq. 9. be given as follows: A change in the orifice to pipe diameter
With the initial conditions as R = R,, dR/dt = s = 0, at t ratio alters the permanent pressure head loss. Specifically,
= 0, the solution has been presented as radius-time or pres-
the permanent pressure head loss is 73% of the orifice pres-
sure-time history of a single cavity starting from its inception sure differential for a orifice to pipe diameter ratio of 0.5,
at vena-contracta with P, varying from an average pressure while for a p of 0.75 it is around 60% of the orifice pressure
of P, at vena contracta until the complete recovery of the differential. Therefore, intensity of the turbulence is in-
pressure to the designated value of P2. versely proportional to p (i.e., PM decreases with an increas-
ing p). Therefore, as discussed in the previous subsection
the increase in p decreases turbulent intensity and hence in-
Results and Discussion creases the bubble life. However, the maximum bubble size
Results are provided of numerical solutions described ear- reached during the transient collapse doesn’t change much
lier for the variety of operating conditions. For brevity, only (specifically a change in p from 0.5 to 0.75 alters the maxi-
representative solutions have been depicted in the subse- mum bubble size reached during bubble oscillation by less
quent figures. The parameters which have been found to af- than 10%).
fect the cavity/bubble behavior are
(1) Recovery pressure (P,) Effect of initial bubble radius
(2) The pipe size downstream of the orifice (d,)
(3) Orifice to pipe diameter ratio ( p ) Simulations were performed for two different bubble sizes
(4) Initial cavity or bubble size ( R , ) (viz., 10, 100 microns) under otherwise similar conditions. The
The result of the simulations indicate the following effect parameters of simulations are listed in the caption of Figure
on the bubble/cavity behavior. 4. The sizes of the cavities were chosen to cover the widest
possible size distribution expected in hydrodynamic flow situ-
ation (Yan et a]., 1990).
Effect of recovery pressure It is evident from the results that the ratio R / R , before
The conditions of simulations have been listed in the cap- collapse is higher for smaller bubbles. Therefore, it is obvious
tion of Figure 1. It indicates that with an increase in the that the contribution to the cavitational effect due to the
recovery pressure the maximum cavity size before collapse pressure pulse is higher due to smaller bubbles, since the

AIChE Journal June 1997 Vol. 43,No. 6 1643

 m
0

zyxwvut
3 1.10
3
I2
4
m
zyxwvutsrqpo
zyxwvuts
zyxwvutsrqpon
1.00
P;

3
d
0

31.10

a 1.00
;

zyxwvutsrqp
d
P; d
0.90 0.90
0.000 0.002 0.004 0.001 0.001 0.010 0.000 0.002 0.004 0.006 0.001 0.010
Dimensionless Time Dimensionless Time

&
a'
0.04 4 a"
\
0.05

a 0.04
0 0.09
3 ."
+16 0.09
= 0.02 m
0
:0.02
0.01 2 0.01
0)
h
a 0.00 a
0.00

-0.01
0.
Dimensionless Time Dimensionless Time
(b)

Dimensionless Time

zyx
Figure 1. Radius history and pressure pulses of bubble oscillations under turbulent and nonturbulent conditions:

zyxwvutsrqp
zyxwvutsrqponmlk
Effect of recovery pressure.
Parameters for simulations:

P
p2
atm 2 do
in.
Ro
Pm
fT
kHz
'/T
kHz
Figure No. without
Turbulence
Figure No. with
Turbulence
0.5 3 1.5 0.75 10 1.492 0.05185 la lc
0.5 5 1.5 0.75 10 1.926 0.06713 lb Id

1644 June 1997 Vol. 43, No. 6 AIChE Journal

 2
3
p.
m
3
zyxwvu
zyxwvutsr
zyxwvutsrqpo
zyxwvutsrqpo
2 1.10

1.00
2
21.10
f
a 1.00
4
d
0.000
0.00 zyxwvutsrqpon
I 0.002 0.004 0.006
Dimensionless Time
0.006 0.010
B

Dimensionless Time

0.06 '
\
a
3
6
0.05

0.04

0.03
'
2

2a
0.05

0.04

0.03
a
*
2 0.02 :0.02
g 0.01
e8 0.01
h h
a a 0.00
0.00

-0.01 -0.01
0. 0 0.
Dimensionless Time Dimensionless Time
(b)
275
250- 250 -
225 -
200 -

Dimensionless Rme Dimensionless Time

a
0
a .
\ :

izo0;
J.

.
gsoo-
\
&SO-

2200-
a
zy
zyxwvutsr
2 150-
e8 100-
a I2 50-

0
50 - L
Dimensionless Time

zyxwv
Figure 2. Radius history and pressure pulses of bubble oscillations under turbulent and nonturbulent conditions:
Effect of pipe size downstream of the orifice.
Parameters for simulations:

p2 do Ro fT 1/T Figure No. without Figure No. with

P atm in. Pm kHz kHz Turbulence Turbulence
0.5 3 1 0.5 10 2.239 0.07778 2a 2c
0.5 3 1.5 0.75 10 1.492 0.05185 2b 2d

AIChE Journal June 1997 Vol. 43, No. 6 1645

 0
a
\
a
1.10
3
a
ID

zyxwvuts
zyxwvutsrqponml
$1.00
d
a

0
.I
Y

ad
:0.02
2g
I.
R
zyxwvutsrq
zyxwvutsrqpon
zyxwvutsrqponmlkji
0.04

0.03

0.01

0.00
4 zyxwvuts
Dimensionless Time

I \

0
3
ad
:0.02
28
h
0.000 0.002 0.004
0.00

0.06

0.05

0.04

0.03

0.01

0.00
Q.006 0.008
Dimensionless Time
0.010

-0.01 -0.01
0. 0. 1
-0
Dimensionless Time Dimensionless Time

450 -

400 -

350 -

I
\I \ I
2300-
\
a250-

d3150$
- 125 / \I .&lo-
d
a 150-
100-

50 -
25
0
0. 0 0.61 0.62 0.63 O.& 0.65 0.66
Dimensionless Time Dimensionless Time

- 0521
0 0
&ooo{
a
q1000- ~1000-
a R
0
3 750-
d

zyxwvu
a
u
5 500-
5M)-
m
m
u
k 250- 6 250-

zyxwvut
- 0
0.00 0.02 0.04 0.08 0.08
Dimensionless Time Dimensionless Time

Figure 3. Radius history and pressure pulses of bubble oscillations under turbulent and nonturbulent conditions:
Effect of orifice to pipe diameter ratio.
Parameters for simulations:

p2 do Ro fT 1/T Figure No. without Figure No. with

P atrn in. cLrn kHz kHz Turbulence Turbulence
0.6 3 2 1.2 10 1.056 0.04391 3a 3c
0.75 3 2 1.5 10 0.961 0.05693 3b 3d

1646 June 1997 Vol. 43, No. 6 AIChE Journal

 B
21.10
3
p:
1.20

5 1.00
zyxwvutsrqponm
7 I 1.50 , zyxw 1

z
0.000
0.90 zyxwvutsrqpon
j
I 0.002 0.004 0.006
Dimensionless Time
0.008 0.010
I

Dimensionless Time

0.07

0 0.06
!? 0.05
pI

3d 0.04

-
0.09

2u 0.02

zyxwvutsr
m
0.01
a
0.00

-0.01 -0.01
0.0 0 0.002 0.004 0.008 0.006 0. 0.
Dimensionless Time Dimensionless Time

zyxwvutsrqponm
(b)

300 1 sOi

\
-_-
0 350-
a
asoo-
.$250

2
V
d
a 200 -
150-
-
zyxwvuts
zyxwvutsrqpon
zyxwvutsrqpo
10

5
Dimensionless Time

m
100-
a
50-

0
0. AM
Dimensionless Time

Figure 4. Radius history and pressure pulses of bubble oscillations under turbulent and nonturbulent conditions:
effect of initial bubble radius.
Parameters for simulations:

p2 do Ro fT 1/7 Figure No. without Figure No. with

P atm in. fim kHz kHz Turbulence Turbulence
0.5 3 1.5 0.75 10 1.492 0.05185 4a 4c
0.5 3 1.5 0.75 100 1.492 0.05185 4b 4d

AIChE Journal June 1997 Vol. 43,No. 6 1647

the maximum radius before the violent collapse.
zyxwvuts
zyxwvu
magnitude of the pressure pulse produced by these bubbles
during collapse is proportional to the ratio of R/R,, R being
The variables studied were found to affect mainly two as-
pects of the bubble behavior: (i) Maximum bubble radius
reached during the oscillations, and (ii) bubble life.
In hydrodynamic cavitation, bubbles formed flow along the
stream. The higher the bubble life, more of the liquid down-
Conclusion
Simulations have been performed to assess the influence of
various flow and geometrical parameters on the turbulence
intensity and thus on the bubble behavior in hydrodynamic
orifice flow cavitation. In spite of the assumptions made in
the algorithm, the simulations do indeed represent some in-
teresting physical phenomenon.
It was found that the bubble behavior changes drastically
zyxwv
stream of the orifice is affected by the cavitational effect. The
simulation studies provide an idea about the variables that
can be manipulated to control the above cavity/bubble be-
havior for maximum cavitational effects.
(1) A rise in the discharge pressure and hence the final
recovery pressure will result in an increase in the active cavi-
tational volume downstream of-the orifice with increased cav-
itational intensity.
( 2 ) The pipe size downstream of the orifice does provide
under the turbulent flow conditions. As compared to the os-
cillatory behavior under nonturbulent conditions; the bubble another means of control over cavitational intensity and ac-
behavior under turbulent conditions is transient and resem- tive volume.
bles the behavior of a cavity under acoustic cavitation. This is (3) The orifice to pipe diameter ratio provides a possible
an interesting result and opens up a number of possibilities means of control over cavitationally active volume without
for the design of cavitating reactors for the so-called sono- changing the cavitational intensity.
chemical effects. It means that acoustic cavitating conditions (4) Although there is a wide variation in the initial bubble
as generated in ultrasonic equipments can be more simply sizes that are naturally generated downstream of the orifice,
generated in hydrodynamic flow situations by manipulating the contribution of small bubbles to the cavitational effect
turbulence levels. This conclusion has been given credence observed in hydrodynamic cavitation reactor appears to be
if one compares the identical metal erosion rates obtained significant.
under acoustic and hydrodynamic cavitating conditions. To
obtain, identical metal erosion rates, the pressure pulses pro- Acknowledgment
duced by collapsing cavities need to be of similar magnitude ABP acknowledges the funding of the DST (Government of India)
which can only be obtained with transient cavitation (with for the project.
turbulence) and not by stable cavitation (without turbulence)
(Hansson et al., 1977).

zyxwvu
Literature Cited
Independent studies (Shirgaokar et al., 1997) carried out
on a high-pressure homogenizer having geometry similar to
that of a hydrodynamic cavitation reactor have also con-
firmed this change in cavity behavior from stable to transient.
The model reaction of the decomposition of the aqueous
KI solution used by Prasad Naidu et al. (1994) was attempted
on high-pressure homogenizers. It was discovered that by ma-
nipulating the discharge pressures and hence the fluid veloci-
(1972).

zyxw
Davies, J. T., Turbulence Phenomena, Academic Press, New York

zyx
Hansson, I., K. A. Morch, and C. M. Preece, “A Comparison of U1-
trasonically Generated Cavitation Erosion and Natural Flow Cavi-
tation Erosion,” Proc. of Ultrasonics Int. Conf., Brighton, U.K., p.
267 (1977).
Miksis, M. J., and L. J. Ting, “Non-Radial Oscillations of Bubbles
including Thermal Effects,” J . Acoust. SOC.Amer., 76,897 (1984).
Plesset, M. S., “The Dynamics of Cavitating Bubbles,”J. Appl. Mech.
ties through the valve gap it was possible to decompose KI in Trans. ASME, 16,277 (1949).
the aqueous solution releasing iodine which was measured Prasad Naidu, D. V., R. Rajan, K. S. Gandi, R. Kumar, S. Chan-
drasekharan, and V. H. Avakeri, “Modelling of a Batch Sono-
spectrophotometrically. It is generally agreed that for the chemical Reactor,” Chem. Eng. Sci., 4x61, 877 (1994).
decomposition of KI in aqueous solutions, transient cavity

zyxwvuts
Rayleigh, Lord, “On the Pressure Developed in a Liquid Due to
behavior is necessary (Prasad Naidu et al., 1994). Thus, Collapse of a Spherical Cavity,” Phil. Mag., 34, 94 (1949).
decomposition of KI in aqueous solutions as observed in Shirgaonkar, I. Z., and A. B. Pandit, “Studies in Kl Degradation
the high-pressure homogenizer confirms the existence of Using Hydrodynamic Cavitation,” in press (1997).
Yan, Y., and R. B. Thorpe, “Flow Regime Transition Due to Cavita-
transient cavity behavior. This behavior as predicted by the tion in the Flow Through an Orifice,” Int. J . Multiphase Flow, 16(6),

zyxw
zyxwv
earlier discussions is only possible under turbulent flow con- 1023 (1990).
ditions (homogenizer valve Reynolds number > 4,000) and Yan, Y., R. B. Thorpe, and A. B. Pandit, “Cavitation Noise and Its
thus can be treated as a preliminary evidence of the transfor- Suppression by Air in Orifice Flow,” Proc. Int. Symp. on Flow In-
mation of cavity behavior from oscillatory to transient (No. duced Vibration and Noise, Chicago, ASME, p. 25 (1988).
KI decomposition was observed at Re < 4,000). Manuscript received Apr. 11, 1996, and revision received Jan. 27, 1997

1648 June 1997 Vol. 43, No. 6 AIChE Journal