VISUALIZATION OF MERCURY CAVITATION
BUBBLE COLLAPSE
M. FUTAKAWA1), T.NAOE1), M. KAWAI2), N. TANAKA3)
1) Japan Atomic Energy Agency
Tokai-mura, Naka-gun, Ibaraki-ken, Japan, 319-1195
2) KEK
Tsukuba-shi, Ibaraki-ken, Japan, 305-0801
3) Ibaraki University
Hitachi-shi, Ibaraki-ken, Japan, 316-8511
ABSTRACT
Innovative researches will be performed at Materials & Life Science Experimental Facility in J-PARC, in which a mercury
target system will be installed as MW-class pulse spallation neutron sources. Proton beams will be injected into mercury
target to induce the spallation reaction. At the moment the intense proton beam hits the target, pressure waves are
generated in the mercury because of the abrupt heat deposition. The pressure waves interact with the target vessel
leading to negative pressure that may cause cavitation along the vessel wall. Localized impacts by micro-jets and/or shock
waves which are caused by cavitation bubble collapse impose pitting damage on the vessel wall. The pitting damage
which degrades the structural integrity of target vessels is a crucial issue for high power mercury targets. The visualization
of mercury cavitation-bubble collapse behaviour was carried out by using a high-speed video camera. In the experiment,
the pressure wave was generated in mercury by using electric magnetic force instead of proton beam injection because of
avoiding radiation hazard. The micro-jet and shock waves were clearly observed at the mercury micro-bubble collapse.
Localized impact was quantitatively estimated through comparison between experiment and numerical simulation in which
the micro-jet was simulated by droplet collision against solid wall. It was deduced from the numerical simulation that the
impact velocity of the micro-jet with the spreading velocity of ca 240 m/s was in the range from 100 m/s to 200 m/s.
INTRODUCTION
Mercury target system for spallation neutron sources, JSNS (Japan Spallation Neutron Source), will be installed at MLF
(Material Life Science Facility) in J-PARC (Japan Proton Accelerator Research Complex)[1,2]. At the moment proton
beams bombard mercury targets, pressure waves are caused by rapidly thermal heat deposition in mercury and
propagate into the target vessel wall[3]. The target wall is excited by the pressure waves to induce negative pressure
along the vessel wall. The negative pressure produces cavitation erosion on the vessel wall [4-7]. At MLF the proton beam
hits the mercury target at 25 Hz. The cavitation erosion, therefore, becomes a crucial issue for structural integrity and
lifetime estimation in the target [8]. MIMTM (electro-Magnetic IMpact Tesing Machine); a pulse generator driven by electric
magnetic force, was developed to systematically examine the cavitation erosion, so called pitting damage [9]. In general,
the cavitation erosion is known to be caused by the micro-jet and shock waves generating at cavitation bubble collapse,
i.e. the cavitation bubble collapse is associated with the damage on the solid interface.
In general, the visualization of cavitation bubble collapse was made in the case of water. A mechanism of
cavitation process was observed by high-speed photography. Researchers have discussed on the mechanism of
cavitation damage so far based on the detailed visualization of a single bubble collapse in water, in which the cavitation
bubble was formed by thermal shock using laser beam [10,11]. In the mercury, the laser beam is hardly absorbed not to
generate thermal expansion and it is opaque liquid. Therefore, we are focusing on the cavitation bubble formation along
the interface between liquid mercury and solid surface from the viewpoint of cavitation damage on the interface. The
mercury cavitation was generated by rapidly moving the solid interface which is driven by magnetic force. The
visualization of collapse behavior of mercury cavitation bubble was carried out by using two-type of high-speed video
cameras to understand relationship among cavitation (damage) intensity, localized impact, bubble size, acoustic emission,
etc. Localized impact was quantitatively estimated through comparison between the experimental results and the
numerical simulation in which the micro-jet was simulated by droplet collision against solid wall.
EXPERIMENTAL
Figure 1 shows a mercury chamber and a specimen in the MIMTM. The inner diameter and height are 100 mm and 15
mm, respectively. The impulsive pressure is imposed to the mercury through the disk plate specimen driven with the
striker controlled by the electro-magnetic force associated with input electric power. The magnitude of pressure is varied
by the imposed power into the MIMTM[9]. The pressure in mercury was measured with the pressure gage (ENTRAN
EPXH-011-25KP), which is installed into the mercury through the lid of the mercury chamber, as illustrated in Fig. 1. The
pitting damage is formed on the interface between mercury and inner-walls of chamber, i.e. the plate specimen, the lid of
chamber, etc. The morphology of pitting damage observed at the power of 560 W in the MIMTM is sufficiently equivalent
to that in the on-beam tests using MW-class proton beams [9]. The repeated frequency of pulses is 25 Hz which is the
same as that in JSNS. The morphology and depth profile of pits are observed using a laser microscope and an SEM.
Visualization of cavitation bubble formation was made by using two-type of CCD video cameras: one is a high4
speed video camera (NAC MEMRECAM RX-6) whose frame rate was 2x10 f/s at its maximum, the other a super high6
speed video camera (SHIMADZU HPV-1) with 1x10 f/s at its maximum. The observed bubbles were formed on the
interface between mercury and a glass window which was installed in the lid of the mercury chamber. A trigger signal was
precisely controlled and input from the MIMTM to the camera at the onset of striker driving to investigate the relationship
between the time responses of imposed pressure in mercury and the acceleration measured at the striker and the optical
images of bubbles. The acoustic vibration was measured by using an accelerometer (RION PV-90B) fixed on the disk
plate specimen of mercury chamber.
(a)
(b)
(c)
(d)
Figure 1 Experimental apparatus for mercury cavitation: (a) Mercury chamber in which the impulsive pressure is imposed
through the disk plate specimen driven by magnetic force. (b) Set-up for visualization of mercury cavitation bubbles. (c)
MIMTM, electric Magnetic IMpact Testing Machine. (d) Schematic drawing for visualization of cavitation bubble formed
along the interface between the liquid mercury and the solid glass window.
RESULTS
Figure 2 shows the relationship between observed bubbles and time-responses of pressure and acceleration. In the figure,
the electric-current imposing time is defined as “0”, i.e. triggered time for photographs. The pressure decreased to ca. -
0.12 MPa and saturated around -0.12 MPa in the period from 0.4 ms to 1.5 ms, and increased rapidly again. The time
response of acceleration resulted from the macroscopic vibration of the disk plate in mercury chamber up to ca. 1.4 ms,
and then the acoustic emission with high frequency components higher than 15 kH was superimposed on the macroscopic
vibration. As for the observed bubbles, the growth of cavitation bubbles was clearly recognized up to 0.5 ms, and then the
increasing rate of bubble size got to be low after 0.5 ms. Around 1.5 ms, the bubbles disappeared rapidly. As a result, it
was readily understood that cavitation bubbles were formed under negative pressure and the bubbles are collapsed when
the pressure rises up to positive level and that the acoustic emission is induced by the bubble collapse. Let us look into in
more detail in the following.
Figure 2 Relationship between bubble formation and time-responses of pressure and acceleration at 560 W input
power. Acoustic vibration was emitted at the same time when the cavitation bubbles were collapsed.
Time responses
Figure 3 shows the time responses of acceleration measured at various input powers for the MIMTM. The amplitude of
acoustic emission increased with the input power in the range more than 295 W. The acoustic emission was hardly
observed in 150 W. The time up to the onset of acoustic emission increased with the input power. The time integrated
value of acoustic emission is associated with the eroded area by pitting damage [12]. The time when the acoustic
2
emission with more than 500 m/s was superimposed on the microscopic vibration was defined as the delay time. The
delay time was clearly dependent on the input power. Figure 4 shows the relationship between the delay time and the
input power. The delay time increases with the input power. It means that the magnitude of power might be deduced by
the delay time. Figure 5 shows the pressure time-responses at various powers. The imposing period of negative pressure
increased with the input power. The period while the pressure was saturated to be negative was defined as the saturated
time. Figure 6 shows the relationship between the saturated time, the pressure and the input power. The pressure
reached to less than -0.1 MPa in the range more than 185 W. More than 200 W the negative pressure gets to be between
-1.1 and -1.2 MPa, which is so-called critical negative pressure for the onset of cavity in mercury. The saturated time
increased with the input power.
Visualization of mercury cavitation bubbles
Figure 7 shows the visualized bubble at 560 W, which was formed along the interface between the liquid mercury and the
solid glass window. The frame rate is 10000 f/s and the time in each frame indicates the passing time from the onset of
the bubble growth. The bubble was recognized at 0.2 ms and grew rapidly up to 0.8 ms and the growth rate became very
low in the range from 0.8 ms to 1.4 ms and then it suddenly collapsed around 1.5 ms. Figure 8 shows the growth and
collapse behavior of bubble formed at 440 W. The bubble grew rapidly up to 0.6 ms and then the bubble size became
almost steady from 0.8 ms to 1.0 ms. The bubble started to shrink around 1.1 ms and collapsed perfectly around 1.2 ms.
The bubble size at 440 W is smaller than that at 560 W and the lifetime of bubble forming from the onset to the collapse is
shorter at 440 W than that at 560 W. The relationships between the bubble size, the lifetime and the input power were
investigated systematically as shown in Fig. 9. The measurements were carried out on bubbles observed in the area of 4 x
2
8 mm around the center of the glass window. The bubble size and the lifetime are increased with the input power. In fact,
the light to get the image of bubble was reflected on the inside wall of bubbles, as illustrated in Fig.1. Note that the size of
bubble measured on the image does not give the real bubble size.
2
1.8
295W
Delay time, ms
Response acceleration, m/s
2
150W
340W
440W
1000
560W
0
1.4
1.2
Delay time
-1000
0
1.6
0.5
1
1.5
2
2.5
3
3.5
1
100
4
200
300
Figure 3 Time-responses of acceleration at various powers.
Acoustic emission was superimposed in more than 295 W.
500
0
Pressure, MPa
0.05
0
-0.05
1.5
1
-0.05
0.5
-0.1
0
-0.12MPa
Saturate time, ms
150W
215W
295W
440W
560W
0.1
600
Figure 4 Relationship between the delay time and
the input power. The delay time increased with
the input power.
0.15
Pressure, MPa
400
Power, W
Time, ms
-0.1
-0.15
-0.15
0
0.5
1
1.5
2
Time, ms
Figure 5 Time-responses of pressure imposed in mercury.
Critical negative pressure for cavity generation in mercury
was recognized to be between -1.1 and -1.2 MPa.
0
100
200
300
400
500
-0.5
600
Power, W
Figure 6 Relationship between the pressure, the
saturate time and the power.
The solid interface was damaged by the localized impact due to the shock and/or micro-jet generating at the
6
moment of collapse of cavitation bubbles. A super high-speed camera with maximum frame rate of 10 f/s was used to
catch the image of continuous collapse behavior. Figure 10 shows the bubble collapse behavior at 560 W, i.e. the
5
phenomena between 1.4 ms and 1.5 ms shown in Fig. 7, which was taken with 2.5x10 f/s. The time in each picture
indicates the inversely passing time from the last frame. The bubble was shrinking gradually from -124 μs to -80 μs. At -72
μs, the micro-jet was recognized to hit against the glass window, and spread out along the glass window from -72 μs to 20 μs. The bubble was collapsed perfectly around -20 μs. The something like mist arose at -4 μs and 0 μs and expanded
out circumferentially, which might be associated with the propagation of shock waves driven by the bubble collapse.
However, it was hardly observed in Fig. 10 that the shock waves propagated along the interface before or after the mist
6
spreading. As increasing the frame rate to 10 f/s, the incident induced by shock wave propagation was recognized more
clearly, as shown in Fig. 11. The size of the bubble decreased up to -20 μs and it was not almost recognized at -15 μs and
-11 μs. It was observed in the images at -10 μs and -9 μs that the annulated mist band was brought about and radiated
circumferentially. The outside edge of the mist band went out from the frames after -8 μs, on the other hand the inside
edge moved out relatively slowly from -8 μs to 0 μs.
Figure 8 Growth behavior of cavitation bubble at 440 W
0.5
2
0.4
1.5
0.3
0.2
1
Lifetime, ms
Maximum bubble radius, mm
Figure 7 Growth behavior of cavitation bubble at 560W
0.1
0
100
200
300
400
500
0.5
600
Power, W
Figure 9 Relationship between the maximum bubble radius, the lifetime and the input power.
Figure 10 Collapse behavior of mercury cavitation bubble with micro-jet injection. It was observed that the micro-jet
was injected to the glass wall from -72 μs to -40 μs during bubble collapsing. The mist was brought about
at -4 μs after completely bubble collapse and radiated circumferentially at 0 μs.
Figure 11 Visualization of mercury cavitation-bubble collapse with the incident induced by shock wave propagation.
The size of the bubble was decreased up to -20 μs and it was scarcely recognized at -15 μs and -11 μs. It was
observed in the images at -10 μs and -9 μs that the annulated mist band was brought about and radiated
circumferentially. The outside edge of the mist band went out from the frames after -8 μs, on the other hand the inside
edge moved out relatively slowly from -8 μs to 0 μs.
DISCUSSION
The visualization of bubble collapse behavior is important to quantitatively evaluate the impact force to impose the pitting
damage on the interface as cavitation erosion. In general, the shock wave and micro-jet have a significant effect on
forming the pitting damage. In the mercury cavitation erosion, the micro-jet impact phenomenon is considered to play a
more important role for the pitting damage than the shock wave because of a high density of mercury. Futakawa, et al.
estimated the impact velocity of the micro-jet against the solid wall through the comparison on the depth profile of pits
between numerical simulation on mercury droplet collision and experimentally measured depth [13]. For the estimation
they were focusing on the ratio of the open diameter to the depth of pits. The locus curve of the ratio, which is independent
of the size of droplet, was obtained. Finally the impact velocity was estimated to be from 200 m/s to 300m/s at 560 W.
Here, the impact velocity of micro-jet is estimated from the observed continuous images of bubble collapse
behavior with the micro-jet injection. Numerical simulation on mercury droplet spreading along the glass window after
collision against it was carried out to evaluate the relationship between the impact velocity and the spreading velocity. The
numerical model and procedure are described elsewhere [13]. Figure 12 shows the micro-jet injection observed in Fig. 10.
The image analysis was carried out to get more prominent profiles of the micro-jet spreading, as shown in Fig.12. As a
result, the spreading velocity of the micro-jet was evaluated to be 241 m/s. The numerical analysis on the droplet collision
represents appropriately the spreading behavior of mercury droplet after colliding against the solid glass wall, as illustrated
in Fig.13. It was deduced from the numerical simulation that the impact velocity of the micro-jet with the spreading velocity
of ca 240 m/s was in the range from 100 m/s to 200 m/s. The impact velocity estimated from the photographs of micro-jet
is lower than that from the pit depth profile. The pit formation could be affected by the shock wave that imposes enough
pressure to locally deform the surface of 316ss with the yield stress of 200 MPa. Ikohagi, et al. carried out the numerical
simulation to investigate the effect of micro-jet and shock wave on the pit formation, which is dependent on the distance
between the bubble and the interface [14]. They concluded that the pit profile formed by water cavitation is affected by
both the pressure wave and micro-jet. As mentioned above, the effect of micro-jet might be dominant on the pit formation
because of more than 10 time higher density as compared with water. Nevertheless, the effect of the pressure wave might
be not ignored because of high surface tension of mercury that could become a driving force for pressure waves. We have
been developing, therefore, the numerical simulation code to analyze the bubble collapse behavior in mercury including
the micro-jet and pressure wave behaviors [15]. The effects of shock wave and/or micro-jet on pit formation in mercury
cavitation will be discussed in the forthcoming studies based on the numerical simulation.
The velocity of the annulate mist band was evaluated from the piled-up images shown in Fig. 14. After the bubble
collapsing completely, the mist was spread symmetrically from the center position of bubble. The front and back edges of
the mist recognized by the difference of contrast were plotted at each measuring time in Fig. 15. It was seen that the front
edge of annulate mist band moved at 1487 m/s that is nearly equal to the sound velocity of mercury and the back edge at
243 m/s, approximately. The mist might be consisted of micro-bubbles which were generated by the excitation of nuclei
( i.e. microscopic impurities, dissolved gas, etc.) at the passage of shock waves [16]. As well, the effects of interface on
the mist generation and disappearance should be considered; the interaction between liquid and solid, stress wave
propagation on the surface of solid, etc.
Figure 12 Micro-jet spreading profile after collision
against the glass wall window.
Analytical model
Deformation
Contour of velocity
Figure 13 Numerical simulation on mercury droplet
collision against solid wall.
Distance from bubble center, mm
Figure 14 Sequential images of bubble collapse showing the expanding an annulated mist band. After the bubble
collapses completely, the annulated band of mist radiates from the center position of the bubble. In the figure, the
vertical axis represents the distance from the center of the bubble collapse, averaged over a narrow horizontal band
through the center in Fig. 11. The horizontal axis in the figure is time.
10
8
1487 m/s
6
243 m/s
4
2
0
0
2
4
6
8
10
Time, μs
Figure 15 Mist expanding behavior described by distance from bubble center position and time. The moving velocity of
outside edge of the annulate mist band is equivalent to the sound velocity in mercury.
CONCLUSION
The mercury target system will be installed as MW-class pulse spallation neutron sources. At the moment the intense
proton beam is injected into the mercury target, pressure waves are generated in the mercury because of the abrupt heat
deposition. The pressure waves interact with the target vessel leading to negative pressure that may cause cavitation
along the vessel wall. Localized impacts by micro-jets and/or shock waves which are caused by cavitation bubble collapse
impose pitting damage on the vessel wall. The visualization of mercury cavitation-bubble collapse behaviour was carried
out by using high-speed video cameras and electric magnetic force instead of proton beam. It was confirmed through the
measured time-responses of pressure and acceleration and visualised images of bubble growth and collapse behaviour
that the acoustic emission was induced by the bubble collapsing, which brings about the micro-jets and the shock waves.
Localized impact was quantitatively estimated through comparison between experiment and numerical simulation in which
the micro-jet was simulated by droplet collision against solid wall. It was deduced from the numerical simulation that the
impact velocity of the micro-jet with the spreading velocity of ca 240 m/s was in the range from 100 m/s to 200 m/s. The
annulated mist band expanding phenomenon caused by the shock wave propagation with nearly sound speed, ca. 1480
m/s, was visually recognized.
Acknowledgments
The authors are grateful to Emeritus Prof. N. Watanabe of JAEA for his encouragement and Dr. Y. Ikeda for his fruitful
advice. This research was partly supported by Japan Society for the Promotion of Science under Grant-in-Aid for Scientific
Research 17360085, 17560740 and 16206097.
References
[1] Nagamiya S., "Overview of the J-PARC Project", Neutron News, Vol. 16, No. 1 16(2005).
[2] Ikeda Y., "1-MW Pulse Spallation Neutron Source (JSNS) of J-PARC", ibid, 20(2005).
[3] Futakawa M., Kikuchi K., Conrad H., Stechemesser H., Pressure and stress waves in a spallation neutron source
mercury target generated by high-power proton pulses, Nucl. Inst. Meth. in Phys. Res. A, 439,1-7(2000).
[4] Futakawa M., Kogawa H. and Hino R. ,Measurement of dynamic response of liquid metal subjected to uniaxial strain
wave, J. Phys. IV France 10 Pr9-237-242(2000).
[5] Futakawa M., Kogawa K., Hino R., Date H. and Takeishi H., Erosion damage on solid boundaries in contact with liquid
metals by impulsive pressure injection, Int J. Imp. Eng. 28 123-135 (2003).
[6] Haines J. R., Farrell K., Hunn J. D., Lousteau D. C., Mansur L. K., McManamy T. J., Pawel S. J., and Riemer B. W.,
Summary of Mercury Target Pitting Issue, SNS-101060100-TR0004(2002).
[7] Futakawa M., Naoe T., Tsai C.C, Kogawa H., Ishikura S., Ikeda Y., Soyama H., Date H., Pitting damage by pressure
waves in a mecury target, Jour. Nucl. Mat. 343, 70-80, (2005).
[8] Futakawa M., Wakui T., Kogawa H., Ikeda Y., Pressure wave issues in high power mercury target, Nucl. Inst. Meth. in
Phy.Res. A 562 676-679 (2006).
[9] Futakawa M., Naoe T., Kogawa H., Tsai C., Ikeda Y., Pitting damage formation up to over 10 million cycles -off-line test
by MIMTM-, J. Nucl. Sci. Tech., 40, 895-904 (2003).
[10] Shima A., Tomita Y., Gibson D. C., Blake J. R., The growth and collapse of cavitation bubbles near composite
surfaces, J. Fluid Mech. Vol. 203 (1989), pp. 199-214.
[11] Phillipp A., Lauterborn W., Cavitation erosion by single laser-produced bubbles, J. Fliiud Mech. Vol. 361, (1998)
pp.75-116.
[12] Futakawa M., Naoe T., Kogawa H., Ikeda Y., Damage diagnostic of localized impact erosion by measuring acoustic
vibration, J. Nucl. Sci. Tech., 41, 1059-1064 (2004).
[13] Futakawa M., Naoe T., Kogawa H., Date H., Ikeda Y., Micro-impact damage caused by mercury bubble collapse,
JSME Int. Jour. A, 234-239 (2005).
[14] Ohira H., Shin B. R., Ikohagi T., Numerical study on behavior of material deformation by bubble collapse and droplet
impact, Turbomachinery,Vol.28,No.2(2000),pp.101-110,(in Japanese).
[15]Tanaka N., Suzuki M., Futakawa M., Numerical analysis of mercury cavitation in MW-scale spallation neutron source
system, Nucl. Inst. Meth. in Phys. Res. A 562 (2006) 680-683.
[16] Obreschkow D., Kobel P., Dorsaz N., Bosset A., Nicollier C., Farhat M., Cavitation bubble dynamics inside liquid
drops in microgravity, PRL 97 094502(2006).