J. Fluid Mech. (2013), vol. 736, pp. 4466.

c Cambridge University Press 2013 44

doi:10.1017/jfm.2013.525

Thermodynamic effects during growth and

collapse of a single cavitation bubble
Matev Dular1 , and Olivier Coutier-Delgosha2
1 Laboratory for Water and Turbine Machines, University of Ljubljana, Askerceva 6,
1000 Ljubljana, Slovenia
2 LML Laboratory, Arts et Metiers ParisTech, 8 Boulevard Louis XIV, 59046 Lille, France

(Received 4 December 2012; revised 26 August 2013; accepted 1 October 2013;

first published online 1 November 2013)
The thermodynamic effects associated with the growth and collapse of a single
cavitation bubble are investigated in the present paper by an experimental approach.
The study focuses on the temperature variations in the liquid surrounding the bubble.
Experiments are conducted in a cylinder partially filled with water at an ambient
temperature and atmospheric pressure. The bubble growth results from the expansion
of an initial air bubble, due to the pressure wave generated by a so-called tube-
arrest method. Several locations of the bubble, at different distances from the bottom
wall of the cylinder, are considered. The bottom wall is made of sapphire, which is
transparent to both the visible and infrared light spectra which enables temperature
measurements by a high-speed thermovision camera at a wavelength of 35 m. Water
is opaque to the infrared light spectrum, hence only temperatures in the boundary
layer and on the liquid vapour interface could be determined. A temperature decrease
of 3 K was recorded during the bubble growth while an increase up to 4 K was
detected during the collapse. Experimental results are compared to the predictions
of the thermal delay model based on the assumption that the bubble growth and
collapse are due to phase changes only. In this approach, the temperature variations
are related to the latent heat exchanges during the vapourization and condensation
processes. On the basis of these results, the respective effects of phase change and air
dilatation/compression in the bubble dynamics are discussed.
Key words: cavitation, drops and bubbles, phase change

1. Introduction
Cavitation is a phenomenon characterized by vapour generation and condensation
in high-speed liquid flows. It frequently occurs in industrial configurations such as
rotating machinery, injectors, and other hydraulic devices. Most of the time it is
accompanied by effects like vibrations, increase of hydrodynamic drag, changes in
the flow hydrodynamics, noise, erosion, light effects such as sonoluminescence, and
also thermal effects. The thermal effects are usually neglected, since the influence of
the temperature variations on the integral flow properties is small in common liquids
such as cold water (Hord, Anderson & Hall 1972). However, on the local scale each
individual bubble undergoes a considerable rise of temperature during the collapse

Email address for correspondence: [email protected]

 Thermodynamic effects during growth and collapse of a cavitation bubble 45
(Hauke, Fuster & Dopazo 2007), while significant local cooling of the liquid is also
expected within the cavitation areas (Franc, Rebattet & Coukon 2004).
Cavitation is usually initiated by a local pressure drop in the vicinity of a cavitation
nucleus. As the bubble grows, latent heat is supplied from the surrounding liquid to
the interface, creating a thermal boundary layer. The consequence is a small local
decrease of the liquid temperature, which results in a slight drop of the vapour
pressure. This delays the further development of the bubble, because a greater pressure
drop is needed to maintain the process. This phenomenon is known as thermal delay,
as it plays a moderating role in the development of cavitation (Brennen 1995).
Thermodynamic effects can usually be neglected in fluids for which the critical-
point temperature is much higher than the working temperature. On the other hand,
the effects become significant when the critical-point temperature is close to the
temperature of the fluid, as in the case of cryogenic fluids (Stahl & Stepanoff
1956). Therefore, the understanding and the prediction of the thermodynamic effect
is crucial in many applications; for example the turbopumps for liquid hydrogen LH2
and oxygen LOx in space launcher engines need to have an inducer rotor installed
upstream from the main impellers, in order to achieve high suction performance. The
inducer is designed to operate in moderate cavitating conditions, hence a minimum
pressure level in the tanks must be ensured, in order to avoid the occurrence of large
sheet cavities on the blades, which are often associated with large-scale instabilities.
Particularity well known is the failure of the Japanese H-II rocket due to rotating
cavitation in the LH2 turbopump (Sekita et al. 2001). To avoid such risk the pressure
in the tanks, and thus their structure and weight, cannot be reduced, which is one of
the fundamental limitations on the increase of the useful mass carried by the launcher.
The thermodynamic effect is favourable when pumping cryogenic fluids as it delays
the cavitation development in the cavitation area. Hence, the rocket engine turbopump
inducer, at given suction pressure, performs better than predicted without consideration
of thermal effects. However the degree of improvement depends on the strength of the
thermodynamic effect, which can at present only be roughly estimated, for example by
the thermodynamic parameter proposed by Brennen (1995).
Past studies have mostly concentrated on the consequences of the thermodynamic
effect rather than on the investigation of the mechanism itself. For example,
its amplitude has been estimated mainly by comparison between the cavitation
development: (i) in a liquid characterized by a large thermodynamic effect; and (ii) in
a liquid with negligible thermodynamic effect.
The first study on the thermodynamic effect was conducted by Stahl & Stepanoff
(1956), who investigated its consequence on pump performance. In the same period,
Sarosdy & Acosta (1961) reported differences in the appearance of cavitation
in water and Freon-113. Quantitative estimations of the thermodynamic effects
were first proposed by Ruggeri & Moore (1969) who measured the variations of
pump performance for various temperatures and fluids. Probably the most thorough
experimental set of data on cryogenic cavitation in Venturi sections, hydrofoil
sections and ogives was published by Hord et al. (1972) and Hord (1973a,b); it is
still considered as a benchmark for validating models of thermodynamic effects in
cavitation. More recent studies have focused on the influence of the thermodynamic
effect of cavitation on performance and cavitation instabilities in rotating machinery
such as turbopump inducers used for rocket propulsion. Franc et al. (2004) conducted
experiments both with water and refrigerant R114 to analyse the modifications of the
cavitation instabilities, and a similar study was also conducted by Cervone, Testa &
dAgostino (2005) with hot and cold water at 343 and 293 K, respectively.
46 M. Dular and O. Coutier-Delgosha
In these experimental investigations, temperature variations themselves are rarely
considered: one example is Fruman, Reboud & Stutz (1999) who measured the local
wall temperature under the cavity with five micro-thermocouples. More recently, on-
board measurements of temperature depression within the sheet cavity at the leading
edge of inducer blades were also recorded (Franc et al. 2010). Time-averaged
temperature variations and characteristic fluctuation frequencies related to cavitation
instabilities have been analysed.
Probably owing to the complexity of the experimental investigations, the
thermodynamic effects are nowadays usually estimated by rudimentary models, most
of them proposed between 1960 and 1990. The most commonly used parameters
in these approaches are the non-dimensional B-factor, which is a dimensionless
temperature depression (Stepanoff 1961), and dimensional parameters like (Brennen
1973), (Kato 1984) or (Watanabe et al. 2007).
To improve the prediction of the thermal delay, modelling of the heat transfers
associated with phase change needs to be improved. This requires experimental
investigations of the temperature variations in the liquid, due to vapourization and
condensation. Although a vast number of experimental studies of thermodynamic
effects in cavitation flow exist, almost none of them deals with the effect directly.
It is therefore essential to develop a technique that would enable direct measurements
of temperatures in cavitating flow.
The present study is devoted to the first direct measurements of the thermodynamic
effect. The analysis is focused on a single vapour bubble evolution. To initiate the
bubble growth, a so-called tube-arrest method is used (Chesterman 1952). It is
particularly suited for the present measurements since it does not introduce thermal
energy from an outside source to create a bubble; for example measurements where
the bubble was created by laser light or by electric discharge would be meaningless
as thermal energy would be transferred to the liquid and thus modify significantly
the local temperature. A non-invasive high-speed thermographic method was applied
to measure the temperature field in the liquid close to the bubble interface during its
successive growth and collapse. In addition, two conventional high-speed cameras were
used to simultaneously observe the bubble shape evolution. Experimental results were,
finally, compared to calculations based on the theories of bubble dynamics and the
thermal delay.

2. Theoretical background
A theoretical approach to the process of bubble expansion and implosion, which
is investigated here, can be found for example in Brennen (1995) or Franc &
Michel (2004). When the pressure drops, the bubble begins to grow due to both
water evaporation and gas expansion, i.e. expansion of the air contained inside the
initial nucleus. During the collapse, the process is reversed: the main mechanisms
are now condensation of vapour and compression of gases. In both phenomena,
the magnitude of temperature variations in the thermal layer around the bubble
depends strongly on the predominant process that drives the bubble size evolution:
gas expansion/compression and the phase changes (Toegel et al. 2000; Matula et al.
2002).

2.1. Isothermal or adiabatic process

To estimate which of the processes is dominant (expansion/compression or
vapourization/condensation), the energy balance of the gas inside the bubble is written
 Thermodynamic effects during growth and collapse of a cavitation bubble 47
as follows, where it is assumed that the initial bubble undergoes a pressure drop in
time 1t during which the radius grows from R up to R + 1R and the gas temperature
changes by 1T:
( 43 R3 )g cvg 1T = 1Q p(4R2 1R), (2.1)
where g and cvg are the density and the specific heat at constant volume of the gas,
1Q is the heat received by the bubble from the liquid, and p is the pressure at the
bubble interface. So, the left-hand side term stands for the variation of internal energy
of the gas during the time 1t, while the second term on the right-hand side is the
work of the pressure force at the bubble surface (as in many past studies, such as
Yasui 1997; Storey & Szeri 2001; Szeri et al. 2003, the terms for kinetic energy, the
energy of bubble deformation etc. are neglected in (2.1)).
Because of the temperature variation within the gas, a thermal boundary layer
develops in the liquid around the bubble. Its thickness is of the order of l 1t (Franc
et al. 2004), with l the thermal diffusivity in the liquid. The temperature gradient in
this boundary layer induces heat transfer by conduction, which drives the heat 1Q to
the bubble.
Fouriers law in the thermal layer can be written as follows:
1T
1Q l 4R2 1t = l l cpl 1t4R2 1T, (2.2)
p
l 1t
where l = l /l cpl with l the thermal conductivity in the liquid, l the liquid density,
and cpl the heat capacity at constant pressure in the liquid.
The energy balance equation (2.1) allows an estimation of the temperature variation
of the gas in the bubble:
1Tad.
1T , (2.3)
1t
r
1+
1tr
where the particular adiabatic case (1Q = 0) leads to 1Tad. = 3(p1R/g cvg R). If
we assume that the temperature variation inside the bubble is entirely due to the heat
exchanges (p(4R2 1R) = 0 which results in (4R3 /3)g cvg 1T = 1Q for (2.1)) and
combine this expression with (2.2) we get 1tr which is the characteristic time of the
heat transfer process at the bubble interface:
(g cvg R)2
1tr = . (2.4)
9l l cpl
The process can be considered adiabatic if the bubble lifetime is much shorter
that the characteristic time of heat transfer 1tr ; the bubble evolution is then mainly
driven by expansion and compression of gases. In the opposite case, if there is enough
time for the heat transfer to proceed until thermal equilibrium is reached (1tr  1t),
the process is closer to isothermal conditions ((2.3) gives 1T = 0), which means
that evaporation and condensation of water and water vapour are the main driving
mechanisms.
In the present experiments, a typical bubble in water at 20 C grew and collapsed
in 16 ms and its maximum radius was close to 9 mm (see for example figure 8).
Considering (2.4) with l = 999 kg m3 , g = 0.0173 kg m3 , cpl = 4182 J kg1 K1 ,
cvg = 717 J kg1 K1 and l = 0.653 W m1 K1 , then 1tr 0.5 ns is obtained.
The characteristic time of heat transfer (0.5 ns) is much smaller than the bubble
48 M. Dular and O. Coutier-Delgosha
lifetime (16 ms), which means that mass transfer (evaporation and condensation)
takes place and can be considered as the main reason for the measured temperature
variations. Expansion and compression may only occur at the very beginning of bubble
growth and just before it collapses to its minimum size, which also complies with the
findings of Akhatov et al. (2001) who predicted these two processes to be dominant
only during the first/last few per cent of bubble life. Therefore, the thermal delay
theory can be applied to interpret the results of the present measurements.

2.2. Thermal delay

The notion of thermal delay can be most simply presented in the case of a spherical
bubble in infinite liquid. As the pressure at infinity drops, an initial spherical nucleus
begins to grow and becomes a much larger bubble. The evaporation process involved
in this growth requires the latent heat L to be supplied by the liquid at the bubble
interface. In the case of an isolated bubble, an estimation of the heat flux transferred
from the liquid to the gas can be obtained from
Fouriers law in (2.2), using the order
of magnitude of the thermal boundary layer l 1t:
1Q Tb T
q= l , (2.5)
4R 1t
2 l 1t
where q is the heat flux per unit surface, Tb is the liquid temperature at the bubble
interface, and T is the ambient liquid temperature. Assuming that the bubble size
evolution is entirely due to vapourization or condensation, the energy balance equation
can be written as:

q = l LR, (2.6)
where R is the growth rate of the bubble, which gives an estimation for the
temperature difference:

R t v L
Tb T = . (2.7)
l l cpl
If we further simplify the analysis by neglecting the initial size of the nucleus and
assuming that the growth velocity of the bubble is of the order of R/1t (see Brennen
1995) we get:
R v L
Tb T = . (2.8)
l 1t l cpl
Equation (2.8) provides an order of magnitude of the temperature difference
expected in the liquid close to the bubble interface. For the present study, water
at 20 C was used, and the bubble grew to a radius of 9 mm in approximately
10 ms. With l = 999 kg m3 , v = 0.0173 kg m3 , cpl = 4182 J kg1 K1 , l =
0.653 W m1 K1 , L = 2454.3 kJ kg1 and l = l /l cpl = 1.56 107 m2 s1 , a
cooling of 2.3 K is obtained.
Heat transfer from the liquid to the bubble is possible only if the temperature
Tb inside the bubble is smaller than T . Hence, the vapour pressure inside the
bubble pv (Tb ) is also smaller than its value pv (T ) in the liquid bulk. Consequently,
the pressure imbalance between the bubble and the reference point at infinity
pv (T ) pv (Tb ) increases, so that the growth rate of the bubble is reduced. To estimate
this variation of vapour pressure, the following equation is based on the Clapeyron
 Thermodynamic effects during growth and collapse of a cavitation bubble 49
relation where the vapour density is neglected compared to the liquid one:
1 1 dpv T dpv
 
L=T . (2.9)
v l dT v dT

The thermodynamic effect in terms of vapour pressure difference can thus be

expressed as:
dpv v L1T
1pv = pv (T ) pv (Tb ) 1T . (2.10)
dT T
Combination of (2.8) and (2.10) finally leads to:
R (v L)2
1pv = , (2.11)
l 1t l cpl T
which gives the decrease of the vapour pressure in the vicinity of the bubble due
to the latent heat flow. The last equation, (2.11), will be used in 4 together with
Brennens parameter (Brennen 1995) to model the spherical bubble dynamics and
the temperature at the interface.

3. Experiments
One of the simplest ways of creating a single cavitation bubble is the so-called tube-
arrest method, which was first used by Chesterman (1952). The technique is based
on a vertically mounted cylindrical tube containing liquid which is pulled downwards
against tensioned supports and released. After the tubes arrest the liquid continues its
upward motion generating a tension pulse, ab initio.
The experimental arrangement is shown in figure 1 and is for the most part the same
as the one used by Chen & Wang (2004) in their experiments. A vertical acrylic glass
tube (1) with inner diameter of 30 mm is open at the upper end. It is put into slides
(2) so it can move freely in the vertical direction. The tube is 1 m long and filled with
water up to the 700 mm level. Distilled water was used. In addition it was boiled for
10 min and cooled down to ambient temperature prior to the experiments to remove as
many cavitation nuclei as possible.
The tube is fastened indirectly to a spring (3) pinned at one end. For the generation
of a cavitation bubble, the tube is pulled downward for some specified length, the
spring thus being compressed. The former is then released, and is quickly pushed
upward by the spring, and then suddenly arrested by a barrier (4) after shooting up the
short distance.
A thin rod (5) with a diameter of 2 mm was inserted vertically into the water. In
the bottom part it had a small dent so that a small air bubble with a radius of 0.5 mm
could be positioned within it (the bubble served as an artificial cavitation nucleus). The
rod was suspended along the centreline of the tube.
During the sudden arrest, a tension wave is produced in the water column, starting
from the tube bottom and travelling upward. The bottom of the tube is closed by
an observation window (6) made out of sapphire glass which is transparent in both
the visible and infrared light spectra. Two high-speed conventional black-and-white
cameras recorded from the side and bottom views, while a high-speed IR camera was
used to measure temperatures from the bottom (through the sapphire glass window).
The velocity Vs at which the tube was suddenly stopped was varied from 0.5 to
2 m s1 . For that purpose, the tube displacement was set to 10 mm. These conditions
lead to spherical growth of the bubble, with a maximal diameter of 20 mm reached
50 M. Dular and O. Coutier-Delgosha

2 5

BW1
BW

IR
2

F IGURE 1. Experimental setup for creating a single cavitation bubble by a tube arrest method.
For a description of the parts see the text.

in a time 1t close to 10 ms. Owing to the vicinity of the bottom wall there was a
tendency of the bubble to form a micro-jet and to collapse asymmetrically (Philipp
& Lauterborn 1998). The extent of asymmetry depended on the initial distance of
the nucleus from the wall. The bubble position is commonly defined by the non-
dimensional distance between the wall and the bubble:
h
= , (3.1)
Rmax
where Rmax is the maximal bubble radius and h is the distance of the bubble centre
from the wall: for > 1, the bubble does not touch the wall, for = 1, the bubble
just touches the wall, and for < 1, the bubble is in contact with the wall during
some period of time.
Observation of the bubble was performed with two high-speed cameras: Fastec
Imaging HiSpec4 2G mono and Motion Blitz EoSens mini 1, which captured images
from side and bottom views, respectively. The cameras could capture images at
523 f.p.s. at 3 Mpixel and 506 f.p.s. at 1 Mpixel resolution, respectively. For the
present experiments, the cameras were synchronized and recorded at 6000 f.p.s. at
a reduced resolution. Figure 2 shows visualizations of bubble growth and collapse in
three different configurations. In the first case, (a), = 0 (the bubble is in contact
with the wall during its whole lifetime); in the second case, (b), = 0.8 (the bubble
 Thermodynamic effects during growth and collapse of a cavitation bubble 51
(a) 10 mm (b) 10 mm (c) 10 mm

F IGURE 2. Visualization of bubble growth and collapse for non-dimensional standoff

distances(a) = 0, (b) = 0.8 and (c) = 1.3. Side view on the left and bottom view
on the right.

is away from the wall at the beginning, but in contact later on); in the last case, (c),
= 1.3 (the bubble is never in contact with the wall).
It can be observed that the collapse of all three bubbles is strongly driven by
the micro-jet (Plesset & Chapman 1971): due to the vicinity of the bottom the
bubble becomes asymmetrical and a liquid jet penetrates it (particularly well seen
at t = 12 ms for = 0.8 and at t = 16 ms for = 1.3). Also in all three cases, the
rebound or splashing can be seen after the collapse (a torus of micro-bubbles nicely
captured by the bottom view camera at t = 18 ms for = 0.8 and at t = 22 ms
52 M. Dular and O. Coutier-Delgosha

(a) (b) (c)

IR IR IR

F IGURE 3. Localization of temperature measurements (black circle) for (a) < 1, (b) = 1
and (c) > 1.

for = 1.3). This is caused by the radial flow of the micro-jet which induces
secondary evaporation (Tong et al. 1999).
In the case = 0 (figure 2a), the shape of the bubble is half-spherical during most
of its lifetime; only at the very end of the collapse, at t = 15 ms, does the bubble
shape become severely distorted. In the case = 0.8 (figure 2b), the bubble is initially
spherical but the shape becomes asymmetrical when it contacts the wall at t = 6 ms,
and a very distinctive micro-jet forms during the collapse (at t = 12 ms). In the final
case = 1.3 (figure 2c), the bubble grows and collapses far from the wall. The
growth is therefore spherical, apart from the high mode instability, small distortions
of the interface (Chen & Wang 2005). The collapse however is still significantly
influenced by the vicinity of the wall and a micro-jet clearly occurs even in this case
(at t = 16 ms).
In the three configurations, the evolution of the temperature field is investigated
with a high-speed thermovision camera CMT384SM Thermosensorik. Its sensitive
wavelength range lies between 3 and 5 m. It should be recalled that water has very
specific properties regarding the propagation of infrared light even a very thin layer
of liquid (10 m or even less) is absolutely opaque to the whole range of the infrared
spectrum (Hale & Querry 1973). This implies that the thermal camera, in configuration
> 1, systematically records the temperature at the boundary between the sapphire
glass and water. In case the = 1, it is expected that this temperature is almost that
of the liquid/vapour interface, at the bottom and/or upper point of the bubble, since
vapour is transparent to infrared light. In the case < 1, when the bubble is attached
to the glass, the recorded temperature is that of the bubble interface at the upper part
of the bubble, for the same reason (figure 3).
The opacity of water makes it appropriate for investigation of the thermodynamic
effect in the present configuration. Indeed, if a transparent or semitransparent liquid
were used, interpretation of the images would be more complex.
Two different acquisition frequencies were applied during the experiments. At a
low frequency of 600 f.p.s. (i.e. 1/10th of the frame rate of the two conventional
cameras), the main features of temperature variations were obtained with a sufficient
 Thermodynamic effects during growth and collapse of a cavitation bubble 53
time resolution of the images to analyse the dynamics of the temperature field. The
frame size at this frequency was 128 128 pixels and the integration time was 500 s
(one pixel corresponded roughly to 0.2 mm 0.2 mm area). At a higher frequency
(3840 f.p.s.) the frame size had to be reduced down to 32 32 pixels with the
integration time of only 40 s. This setting enabled better temporal resolution of the
measurements, which enabled a more accurate comparison with the thermal delay
theory.
The camera was calibrated in-situ. Uncertainty of the mean operating temperature of
water was checked by comparison with measurements by an A-class Pt100 sensor and
a discrepancy of 0.2 K was found. However, our main goal was to quantify relative
differences in the non-uniform time-dependent temperature field. For a single element
on the temperature sensor of the thermocamera, the noise equivalent temperature
difference (NETD) is less than 30 mK.
To test the repeatability of the installation (generation of the bubble and its
temperature response) we generated four bubbles at the same conditions at a non-
dimensional standoff distance = 0.8. The results are shown in figures 8 and 9
in 5. The differences in the size of the bubble are insignificant (figure 8b at
= 0.8), while somewhat larger discrepancies between the test runs are observed
in the temperature evolution (figure 9b at = 0.8). Nevertheless one can still claim
that the measurements are repeatable.

4. Modelling
For further analysis of the experimental data, and discussion of the mechanisms that
drive the temperature variations, comparison of the experimental time evolutions of the
temperature with results of modelling is required. In the theoretical model we assume
the bubble to remain spherical, which means that the influence of the bottom wall is
neglected. The RayleighPlesset equation (Plesset 1949) including thermal effects is
considered:
 3
3 2 R0 2S R
 

l RR + R = pv (Tb ) p + pg0 4 . (4.1)
2 R R R
Equation (4.1) can be rewritten as follows:
 3
3 2 R0 2S R
 

l RR + R + 1pv = pv (T ) p + pg0 4 . (4.2)
2 R R R
Considering (2.11) and introducing the parameter (Brennen 1995) defined by
= (v L)2 /l2 cpl T l , the term 1pv can be written as 1pv = R tl which finally
leads to:
 3
3 R0 2S R
 
l RR + R 2 + R tl = pv (T ) p + pg0 4 , (4.3)
2 R R R
which gives the bubble radius evolution according to time t. It should be noted that
other approaches for calculation of bubble size evolution in tubes exist (for example
Chen & Wang 2005). However, although their approach works well for bubbles with
size comparable to the tube size, the bubble evolution in the initial and final stages
is not correctly predicted. As the velocity of the interface in these periods is of
primary importance for the calculation of the thermodynamic effect, such a model is
not suitable in the present study.
54 M. Dular and O. Coutier-Delgosha
In 5, (4.3) is solved numerically to obtain the bubble radius evolution as a function
of time. Then (2.8) is applied to calculate the time evolution of the temperature Tb .
Results are compared to the results of experimental measurements.

5. Results
The present section focuses on the analysis of the temperature measurements during
the growth and collapse of the bubble. First, the data recorded with a low frame rate
(600 f.p.s.) are shown and discussed. Then, measurements with the higher frame rate
(3840 f.p.s.) are compared to the results of the RayleighPlesset and thermal delay
model to clarify the understanding of the experimental visualizations.

5.1. Temperature measurements

Figures 46 show (for the three experimental test cases: non-dimensional standoff
distance = 0.8, = 1 and = 1.3) five successive characteristic situations observed
during the bubble lifetime. For each situation, the bottom and side views of the bubble
are displayed together with the corresponding temperature field. In the three cases,
time t = 0 corresponds to the bubble inception.
According to (2.8), cooling of the liquid due to the transfer of latent heat during
the evaporation process should occur in the initial stage of the experiment. The case
= 0 (figure 4) leads to the simplest analysis, as the bubble is attached to the
glass surface during the whole duration of the experiment. This condition enables
continuous measurements of the temperature at the bubble surface (see figure 3a).
The cooling is observed slightly after the beginning of the bubble growth. The
maximum temperature drop 1T during the evaporation stage is 2 K. As the growth
of the bubble decelerates the temperatures starts to grow. The rate of temperature
increase is slow at first, but it is significantly accelerated as the cavity collapses at
t = 16 ms in the form of a micro-jet. The acquired temperature field at the instant
of bubble collapse should represent the temperature elevation of the liquid micro-jet,
which is of the order of 3 K. This temperature increase may be due to the transfer
of latent heat during the vapour condensation, and/or to the heat transfer between
the vapour, whose temperature increases drastically during the gas compression, and
the liquid micro-jet that flows through the bubble. Indeed, temperature elevations of
several thousands Kelvin may occur in the vapour during a very short period of time at
the end of the bubble collapse (Brennen 1995). However, such high values apply only
to spherical collapse of bubbles, which is not the case in the present study.
After the bubble collapse, the heat is convected and diffused over the whole
observation window, so the liquid temperature increases slightly (the change between
the initial and final temperatures is 0.3 K) and no temperature deviation is recorded
during the secondary evaporation observed by conventional cameras at t = 18 ms.
At a bigger standoff distance = 0.8 (figure 5), the evolution is similar, but the
temperature decrease is detected much later in the bubble lifetime, 6 ms after the
start of the evaporation, which corresponds to the moment when the bubble contacts
the wall. A maximum temperature drop of 1.2 K is obtained. The smaller temperature
change in this configuration is a result of delayed acquisition of the temperature
field as the opacity of water in the IR spectrum does not enable detection of the
temperature variations associated with the first stage of bubble growth, before it
reaches the wall. At t = 14 ms the bubble collapses in the form of a micro-jet. The
acquired temperature field at this time (t = 14 ms) represents the temperature of the
liquid micro-jet in the centre, which has been slightly increased by flowing through the
vapour bubble that surrounds the jet. The temperature of the jet is 1.2 K above the
 Thermodynamic effects during growth and collapse of a cavitation bubble 55

F IGURE 4. Evolutions of the bubble shape and temperature field for the case = 0.

initial one. After the collapse the heat is again convected and diffused over the whole
observation window. However, the change in liquid temperature is 1T = 0.7 K, which
is larger than in the previous case at = 0.
In the third case displayed in figure 6 ( = 1.3), the bubble never touches the
sapphire window, so no cooling is detected (because of the water opacity in the IR
spectrum). However, a hot spot (0.8 K of temperature increase) appears 18 ms after
the bubble started to grow. It can be presumed that this local temperature elevation is
associated with the micro-jet that hits the glass surface. Like in previous cases the final
temperature is slightly higher than the initial one (1T = 1 K).
It seems that the change between the initial and final temperature is dependent
on the bubble standoff distance. This may be due to variations of the maximum
temperature reached by the vapour during the final stage of the collapse (a bubble
which is further from the wall collapses in more spherical shape which initiates higher
gas temperatures) and also the difference in the jet velocity as it hits the window
(according Plesset & Chapman 1971 it increases with the standoff distance).
56 M. Dular and O. Coutier-Delgosha

1.2

0.8

0.4

0.4

0.8

1.2

F IGURE 5. Evolutions of the bubble shape and temperature field for the case = 0.8.

For a more detailed representation of the thermal effect, vertical cross-sections

through the centreline of the images of the temperature fields (like the ones in
figures 46) were made and put together in a time sequence. This way, one dimension
(horizontal one) is sacrificed (one can assume the temperature field to be almost
axisymmetric) but a much clearer representation in time is obtained; the diagrams in
figure 7 display the temperature variations as a function of time (horizontal axis) and
bubble radius (vertical axis) for the three cases.
For = 0, the bubble is attached to the observation window right from the start of
its growth. It can be observed that the cooler region grows very quickly (in 3 ms)
to its final extent of 17 mm. The region then slowly shrinks which implies that the
collapse is slow. The reason for this peculiar behaviour is the fact that the bubble
collapses asymmetrically: the height of the bubble decreases more rapidly than its
 Thermodynamic effects during growth and collapse of a cavitation bubble 57

1.2

0.8

0.4

0.4

0.8

1.2

F IGURE 6. Evolutions of the bubble shape and temperature field for the case = 1.3.

radius which consequently leads to the formation of a micro-jet. As the jet hits the
surface of the window, a rapid and very brief increase of temperature is detected.
The case = 0.8 is similar but now the growth phase seems to be longer and the
collapse phase shorter. Indeed, due to opacity of the water, the temperature at the
bubble interface can be detected only through the part of the bubble that is attached
to the wall. As the bubble slowly attaches to the wall, a bigger and bigger part of the
bubble interface becomes visible, which results in the detection of a slowly increasing
cooler region. The collapse in this configuration has almost the same duration as in the
case = 0: a very quick change to higher temperatures is recorded at the moment of
the micro-jet impact.
In the third case = 1.3, the bubble remains unattached to the wall. Therefore,
the cooling due to evaporation does not reach the liquid located at the wall, and no
temperature variation is detected during this phase. Only the impact of the micro-jet
58 M. Dular and O. Coutier-Delgosha

(a) 3
10

2
5

R (mm)
1
0
0
5
1
10
2
4 1 6 11 16 22

(b)
10 1.2

0.8
5
0.4
R (mm)

0 0

0.4
5
0.8
10
1.2
5 10 15 20 25 30

(c)
10 1.2

0.8
5
0.4
R (mm)

0 0

0.4
5
0.8
10
1.2
17 22 27 32 37 42
t (ms)

F IGURE 7. Temperature variations over time through vertical cross-sections of the

bubble for non-dimensional standoff distances (a) = 0 (for t = 4 to 22 ms),
(b) = 0.8 (for t = 5 to 30 ms) and (c) = 1.3(for t = 17 to 42 ms). Bubble growth begins at
t = 0.

leads, as previously, to a slight increase of the wall temperature. However, it is not

as clear as in the previous cases, probably because of the heat diffusion during the
distance it travels prior to impact with the wall. Conversely, as already mentioned, the
general increase of liquid temperature after the bubble collapse can be clearly observed
 Thermodynamic effects during growth and collapse of a cavitation bubble 59
in that case: the temperature increase is close to 1 K, compared with the one before
the experiment.
In order to more precisely record the temperature evolution and the maximum
amplitude of temperature variations in the experiments, the frequency of the camera
was increased up to 3840 f.p.s. This setting enabled better temporal resolution of the
measurements, but for this we needed to decrease the spatial resolution: the image
was reduced to 32 32 pixels which corresponds to a region of 6 6 mm only and
consequently the recorded temperature fields were practically homogeneous and were
treated as such. The results of these measurements are shown and compared to the
predictions of the RayleighPlesset and thermal delay theory in the following section.

5.2. Comparison with simulations by the RayleighPlesset equation combined with the
thermal delay theory
To investigate whether the present experiment corresponds with the theoretical
approach presented in 2 and 4, the measured and calculated evolutions of bubble
size are first compared. The tube walls and the sapphire observation window certainly
influence the bubble dynamics, which is a notable difference with the assumptions
involved in the theoretical approach where a bubble in an infinite liquid is assumed.
The determination of the radius evolution in the experiments is based on the
estimation of the vapour volume V at each time by image processing. Then, a
radius r of a sphere that would accommodate this volume is derived simply from
the equation: rV = 3V/4)1/3 . Figure 8 shows the evolution of the measured bubble
radius rV compared with that calculated by the RayleighPlesset equation (4.3).
The initial condition for each case can be derived from the pressure drop estimated
according to the water hammer equation:
1p = l cl Vs , (5.1)
where 1p is the pressure drop and Vs is the tube velocity just before it is suddenly
stopped. Values of velocity in the experiments for cases = 0, = 0.8 and = 1.3
are 0.84, 0.76 and 1.4 m s1 , respectively, as measured optically from the high-speed
camera recordings. According to (5.1) these result in 1p = 1.21, 1.08 and 2.02 MPa,
for = 0, = 0.8 and = 1.3, respectively.
Measurements (by a hydrophone Reson TC4013) of the pressure evolution at
the bottom of the tube show that the pressure first drops to approximately the
negative value of the water hammer pressure. The measured pressure recuperation
is, unexpectedly, much slower and a positive pressure pulse is seen only after 10 ms.
We suspect that the slow pressure recovery is related to a very big cavitation bubble
which significantly influences both the dynamics of the pressure wave and also the
measurements. Since we have no means of determining the real pressure evolution
we used a combination of theoretical and measured pressure evolutions for solving
the RayleighPlesset equation; there an almost instant pressure drop to the minimum
pressure (water hammer pressure) and an almost instant increase to vapour pressure
were assumed. The vapour pressure then persisted until the end of the simulation.
The initial bubble radius was the same as observed in the experiment R0 = 0.5 mm.
These values are used to obtain numerically the bubble radius evolution in figure 8. It
is interesting to observe that even during the bubble collapse, where the bubble was
not spherical, the measured and calculated bubble radii fit almost perfectly.
The diagram at a non-dimensional standoff distance = 0.8 shows the evolution of
the bubble radius for four different bubbles; since the differences are insignificant one
can consider the tests to be repeatable.
60 M. Dular and O. Coutier-Delgosha

(a) 10
9
8
7

R (mm)
6
5
4
3 Experiment
2 Simulation
1

0 2 4 6 8 10 12 14 16 18 20

(b) 10
9
8
7
R (mm)

6
5
4
3 Experiment
2 Simulation
1

0 2 4 6 8 10 12 14 16 18 20

(c) 10
9
8
7
R (mm)

6
5
4
Experiment
3
2 Simulation
1

0 2 4 6 8 10 12 14 16 18 20
t (ms)
F IGURE 8. Measured and calculated (for a spherical bubble) bubble radius as a function
of time for non-dimensional standoff distances (a) = 0, (b) = 0.8 (The grey symbols
show results from several tests it can be seen that the experiment is highly repeatable.) and
(c) = 1.3.

It can be observed that the calculation also predicted a rebound of the bubble; this
was also seen in the experiments, but the vapour volume could not be estimated with
acceptable uncertainty. Moreover, the temperature variations could not be recorded
after the collapse of the initial bubble, because of the opacity of the water in the IR
spectrum, which makes the study of the rebound of little interest in the context of the
present work.
Diagrams displayed in figure 9 show the comparison between the experimentally
measured and calculated temperature evolution. A fair agreement between the model
 Thermodynamic effects during growth and collapse of a cavitation bubble 61

(a) 5

3 Experiment
Simulation
5
2 2 6 10 14 18 22 26 30

(b) 5

3 Experiment
Simulation
5
2 2 6 10 14 18 22 26 30

(c) 5

3 Experiment
Simulation
5
2 2 6 10 14 18 22 26 30
t (ms)

F IGURE 9. Measured and calculated (for a spherical bubble) temperature as a function of

time for non-dimensional standoff distances (a) = 0, (b) = 0.8 (The grey symbols at
show results from several tests it can be seen that the experiment is highly repeatable.) and
(c) = 1.3.

and the experiments is obtained in the configuration = 0. The maximum amplitude

of the cooling (2 K), at the beginning of the process, is especially well predicted by
the RayleighPlesset approach. However, the models predicts an almost instantaneous
temperature decrease at t = 0, while only a relatively sharp gradient is obtained in the
experiment. The temperature evolution after the initial cooling, until the peak due to
the bubble collapse at 15 ms, is nicely reproduced by the model, but the amplitude of
the peak is not correctly predicted: 1T = 12 K in the model, while only 1T = 3.6 K
was measured. This may be partially related to the assumption of a spherical bubble,
62 M. Dular and O. Coutier-Delgosha
and also it has been assumed that both bubble evaporation and collapse are triggered
by phase change, which is probably valid for the bubble growth, which is slow, but
much more questionable for the collapse. Indeed, only the final stage of collapse
leads to the very high temperature increase in the vapour phase, and thus to the brief
temperature elevation measured at the interface. This is corroborated by the difficulty
in detecting experimentally the maximum value of the temperature increase, which
has required accelerating the camera speed as much as possible. Consequently, air
compression may play a significant role in the bubble size variation, during this final
stage of collapse.
For = 0.8 the situation is similar, but as the bubble is further from the wall
the cooling is not detected immediately in the experiment. It is obtained at t 5 ms,
i.e. when the bubble contacts the wall (see figure 5). Again, the predicted amplitude
of temperature decrease is very close to the measured one. It can also be seen that
the temperature increase associated with the final stage of the collapse is predicted
to occur at the same time as it was measured, however a much higher amplitude is
given by the model. As in figure 8 this diagram also shows the recorded temperature
evolutions for the four parallel tests at the same conditions. Again one can conclude
that the discrepancies between the test runs are insignificant and that measurements are
repeatable.
In configuration at = 1.3 only heating can be detected experimentally. It has been
suggested in the previous discussion of the results that this temperature increase
is obtained when the micro-jet hits the surface of the observation window. The
temperature deviation is first detected 8 ms after the micro-jet inception jet (see
figures 2, 6 and 9). This corresponds well to the time it takes the jet to travel the
distance of l = 2 Rmax 24 mm from the top point of the bubble (jet inception point)
to the wall (the jet velocity was estimated from several consecutive images and is
roughly 3 m s1 ).

6. Discussion
The results displayed in figure 9 reveal that the liquid heating observed during
the bubble collapse is of higher amplitude than the cooling measured during the
bubble growth. This point may be related to the abrupt character of the final phase
of the bubble collapse, which generates high levels of temperature and pressure in
the gas (Yasui 1997; Storey & Szeri 2000). Such a high temperature, of the order of
thousands Kelvin, induces heat exchange by conduction with the surrounding liquid,
which results in a significant temperature rise at the interface, in addition to that
due to vapour condensation. Results obtained in the present experiments suggest that
conduction at the bubble interface counterbalances the fact that part of the vapour
contained in the bubble does not condense during the collapse, because of the very
high speed of bubble compression at the final stage of the process. This phenomenon,
which has been investigated by several authors (for example Storey & Szeri 2000;
Akhatov et al. 2001) is due to the fact that the vapour located at the centre of the
bubble does not have a sufficient time to diffuse towards the bubble surface, so some
vapour is trapped inside the bubble during the most violent stage of the collapse.
Results also suggest that the bubble growth is not caused by vapourization only.
Indeed, the case = 0 in figure 9 shows a period of temperature stagnation at the
beginning of the bubble expansion (in the first image taken by the IR camera no
temperature difference was measured approximately between t = 0 and t = 0.25 ms),
before the liquid cooling is observed. This implies that at the initial stage of
bubble growth, the high velocity at the interface (see figure 8) does not allow the
 Thermodynamic effects during growth and collapse of a cavitation bubble 63
vapourization to occur. To investigate this point, the characteristic time scale of bubble
growth bubble can be compared to the time scale related to phase change PC , as
proposed by Storey & Szeri (2000):
R
bubble = (6.1)
R
and
s
R 2 Mvapour
PC = , (6.2)
9Rgas Ti
where Mvapour is the molecular mass of the vapour, Rgas is the constant of perfect
gases, and Ti is the temperature at the bubble interface (here we assume the
temperature field within the bubble to be homogeneous, hence Ti = Tb ). The parameter
is the so-called accommodation coefficient which is included in the expression
for the evaporation/condensation flux across the bubble interface, according to the
HertzKnudsenLangmuir formula which is derived from the kinetic theory of gases.
Detailed expressions for these fluxes can be found for example in Fuster, Hauke &
Dopazo (2010). The accommodation coefficient represents a resistance to evaporation
or condensation at the interface during the bubble evolution. Although there is no
general consensus regarding this parameter, = 0.4 is generally used in most of the
recent studies (Yasui 1997; Storey & Szeri 2000, 2001)).
If we consider the initial stage of the bubble growth, i.e. the first 250 s mentioned
previously, the order of magnitude of the bubble radius and the bubble growth velocity
can be estimated from figure 8 as R = 1.7 mm and R 7 m s1 , respectively. This
leads to bubble = 2.5 104 s, while PC = 3.1 104 s is obtained with = 0.4,
Tb = 293 K, and Rgas = 8.31 J mol1 K1 . These orders of magnitude of the time
scales confirm that vapourization may have insufficient time to occur during the initial
stage of bubble growth, which means that the bubble velocity is mainly due to air
expansion. It explains why no water cooling is detected by the thermal camera during
the first 50 s of the process.
Later in the sequence, for example at t = 2 ms, the bubble radius reaches
R 6 mm and the bubble velocity decreases to R 1 m s1 , which leads to
characteristic time scales bubble = 6 103 s and PC = 103 s. At this point, bubble
has become significantly higher than PC , so the bubble growth is essentially driven by
vapourization, which is consistent with the cooling 1T = 3 K detected by the camera
(see figure 9).
A similar phenomenon is expected at the end of the bubble lifetime, during the
last abrupt phase of the bubble implosion. Similar velocities as in the initial stages of
bubble growth can be observed, hence similar time scales bubble and PC are obtained.
However, no evidence of the time delay can be seen from the thermal measurements,
because, as mentioned previously, high levels of temperature within the bubble during
this final step of the process also contribute to the heating of the surrounding liquid,
so the temperature increase at the wall is observed even if vapour condensation is of
minor importance at the end of the process.
One can see that in cases = 0 and = 0.8 (figures 4, 5 and 9) the liquid
temperature starts to increase above the initial level at about t 14 ms, which is
significantly later than the beginning of the bubble collapse (t 10 ms). In both cases
the bubble touches the surface of the glass at this stage, so the thermal boundary layer
in the liquid due to heat exchanges with the gas should be immediately visible with
the thermal camera. This suggests that the temperature increase detected by the camera
64 M. Dular and O. Coutier-Delgosha
is an effect of the large heating of the air inside the bubble, due to its compression
during the second part of the collapse; at the very end of the collapse the temperature
difference in the boundary layer is mainly due to the conductive heat transferred from
the hot inner part of the bubble to the liquid, while the condensation process only
slightly affects the liquid temperature.
This is again confirmed by the case = 1.3 (figures 6 and 9). In this configuration,
the bubble is far from the wall during its implosion, so the increase of the liquid
temperature is convected by the re-entrant jet towards the wall (figures 2 and 7).
Figure 7 shows that the temperature increase at the wall is spread over a large
area, and it is continuously detected between t = 18 ms and t = 24 ms, i.e. during
Theat 6 ms. As already mentioned, the velocity of the jet was about Vjet 3 m s1 .
The duration of the high temperature detected at the wall implies that the origin of the
re-entrant jet temperature increase has a vertical size that can be roughly estimated as
H = Vjet Theat = 18 mm, which is of the order of magnitude of the bubble diameter
and much larger than the thickness of the thermal boundary layer on the bubble wall.
This shows that the heat convected by the re-entrant jet is not due to the convection of
the thermal boundary layer around the bubble but due to the heat conduction of the jet
flow as it penetrates the bubble.

7. Conclusions
In the present study, the first direct measurements of the thermodynamic effect on
a single cavitation bubble have been obtained. A non-invasive thermographic method
was used to measure temperature fields in the liquid during the growth and collapse of
a single cavitation bubble, which was initiated mechanically without the introduction
of heat sources.
Results show that during the bubble growth the temperature of the liquidvapour
boundary layer decreases. At the collapse a reversed process was observed as the
temperature increased. The contribution of four processes (evaporation, expansion,
condensation and compression) was discussed. Based on a comparison of characteristic
times of conductive and latent heat flows it was concluded that the two main
mechanisms that drive the heat fluxes are evaporation and condensation, while the
other two, expansion and compression, contribute significantly only at the very
beginning and very end of the bubble lifetime. Consequently, the theory of the
so-called thermal delay (Brennen 1973), which is based on this assumption, was
applied, and it was found that the predicted temperature variations in the liquid are
in relatively good agreement with the present measurements. This suggests that the
thermal delay theory is valid in the present case for most of the bubble evolution,
which is further indirect evidence of the predominance of mass transfers in the bubble
evolution. Definite confirmation of the thermal delay theory is beyond the scope of the
present work and could possibly be obtained after additional experiments in various
other liquids; however the success of such experiments is questionable due to the
unique property of water that it is opaque to the infrared spectrum.
Further work will include studies of direct temperature measurements in developed
hydrodynamic cavitation with the purpose of improving numerical methods for
prediction of cavitation in thermosensible fluids.

Acknowledgement
M.D. would like to thank the European Space Agency (ESA) which enabled the
study within the scope of the project Cavitation in Thermosensible Fluids.
 Thermodynamic effects during growth and collapse of a cavitation bubble 65
REFERENCES
A KHATOV, I., L INDAU, O., T OPOLNIKOV, A., M ETTIN, R., VAKHITOVA, N. & L AUTERBORN, W.
2001 Collapse and rebound of a laser-induced cavitation bubble. Phys. Fluids 13, 28052819.
B RENNEN, C. E. 1973 The dynamic behavior and compliance of a stream of cavitating bubbles.
Trans. ASME: J. Fluids Engng 95 (4), 533541.
B RENNEN, C. E. 1995 Cavitation and Bubble Dynamics. Oxford University Press.
C ERVONE, A., T ESTA, R. & D AGOSTINO, L. 2005 Thermal effects on cavitation instabilities in
helical inducers. J. Propul. Power 21, 893899.
C HEN, Q.-D. & WANG, L. 2004 Production of large size single transient cavitation bubbles with
tube arrest method. Chin. Phys. 13 (4), 564570.
C HEN, Q.-D. & WANG, L. 2005 Luminescence from transient cavitation bubbles in water. Phys. Lett.
A 339, 110117.
C HESTERMAN, W. D. 1952 The dynamics of small transient cavities. Proc. Phys. Soc. B 65,
846858.
F RANC, J.-P., B OITEL, G., R IONDET, M., JANSON, E., R AMINA, P. & R EBATTET, C. 2010
Thermodynamic effect on a cavitating inducer-part II: on-board measurements of temperature
depression within leading edge cavities. Trans. ASME: J. Fluids Engng 132 (2), 021304.
F RANC, J.-P. & M ICHEL, J.-M. 2004 Fundamentals of Cavitation. Fluid Mechanics and Its
Applications, vol. 76, Springer.
F RANC, J.-P., R EBATTET, C. & C OUKON, A. 2004 An experimental investigation of thermal effects
in a cavitating inducer. Trans. ASME: J. Fluids Engng 126, 716723.
F RUMAN, D. H., R EBOUD, J. L. & S TUTZ, B. 1999 Estimation of thermal effects in cavitation of
thermosensible liquids. Intl J. Heat Mass Transfer 42, 31953204.
F USTER, D., H AUKE, G. & D OPAZO, C. 2010 Influence of the accommodation coefficient on
nonlinear bubble oscillations. J. Acoust. Soc. Am. 128 (1), 510.
H ALE, G. M. & Q UERRY, M. R. 1973 Optical constants of water in the 200 nm to 200 m
wavelength region. Appl. Opt. 12, 555563.
H AUKE, G., F USTER, D. & D OPAZO, C. 2007 Dynamics of a single cavitating and reacting bubble.
Phys. Rev. E 75, 066310.
H ORD, J. 1973a Cavitation in liquid cryogens II hydrofoil. NASA CR-2156.
H ORD, J. 1973b Cavitation in liquid cryogens III ogives. NASA CR-2242.
H ORD, J., A NDERSON, L. M. & H ALL, W. J. 1972 Cavitation in liquid cryogens I Venturi. NASA
CR-2054.
K ATO, H. 1984 Thermodynamic effect on incipient and development of sheet cavitation. In
Proceedings of International Symposium on Cavitation Inception, New Orleans, LA,
pp. 127136.
M ATULA, T. J., H ILMO, P. R., S TOREY, B. D. & S ZERI, A. J. 2002 Radial response of individual
bubbles subjected to shock wave lithotripsy pulses in vitro. Phys. Fluids 14, 913921.
P HILIPP, A. & L AUTERBORN, W. 1998 Cavitation erosion by single laser-produced bubbles. J. Fluid
Mech. 361, 75116.
P LESSET, M. S. 1949 The dynamics of cavitation bubbles. J. Appl. Mech. 16, 277282.
P LESSET, M. S. & C HAPMAN, R. B. 1971 Collapse of an initially spherical vapour cavity in the
neighborhood of a solid boundary. J. Fluid Mech. 47 (2), 283290.
RUGGERI, R. S. & M OORE, R. D. 1969 Method for prediction of pump cavitation performance for
various liquids, liquid temperature, and rotation speeds. NASA TN D-5292.
S AROSDY, L. R. & ACOSTA, A. J. 1961 Note on observations of cavitation in different fluids.
Trans. ASME: J. Basic Engng 83, 399400.
S EKITA, R., WATANABEL, A., H IRATA, K. & I MOTO, T. 2001 Lessons learned from H-2 failure and
enhancement of H-2a project. Acta Astron. 48 (512), 431438.
S TAHL, H. A. & S TEPANOFF, A. J. 1956 Thermodynamic aspects of cavitation in centrifugal pumps.
Trans. ASME: J. Basic Engng 78, 16911693.
S TEPANOFF, A. J. 1961 Cavitation in centrifugal pumps with liquids other than water. J. Engng
Power 83, 7990.
66 M. Dular and O. Coutier-Delgosha
S TOREY, B. D. & S ZERI, A. J. 2000 Water vapour, sonoluminescence and sonochemistry. Proc. R.
Soc. Lond. A 456, 16851709.
S TOREY, B. D. & S ZERI, A. J. 2001 A reduced model of cavitation physics for use in
sonochemistry. Proc. R. Soc. Lond. A. 457, 16851700.
S ZERI, A. J., S TOREY, B. D., P EARSON, A. & B LAKE, J. R. 2003 Heat and mass transfer during
the violent collapse of nonspherical bubbles. Phys. Fluids 15, 25762586.
T OEGEL, R., G OMPF, B., P ECHA, R. & L OHSE, D. 2000 Does water vapour prevent upscaling
sonoluminescence? Phys. Rev. Lett. 85, 31653168.
T ONG, R. P., S CHIFFERS, W. P., S HAW, J. S., B LAKE, J. R. & E MMONY, D. C. 1999 The role of
splashing in the collapse of the laser-generated cavity near a rigid boundary. J. Fluid Mech.
380, 339361.
WATANABE, S., H IDAKA, T., H ORIGUCHI, H., F URUKAWA, A. & T SUJIMOTO, Y. 2007 Steady
analysis of the thermodynamic effect of partial cavitation using the singularity method. Trans.
ASME: J. Fluids Engng 129 (2), 121127.
YASUI, K. 1997 Alternative model of single-bubble sonoluminescence. Phys. Rev. E 56, 67506760.