Towards Real-Time Optical Measurement of Microbubble Content in Hydrodynamic Test Facilities

Full text

1056
Towards real-time optical measurement of microbubble content in hydrodynamic test
facilities
1Patrick S. Russell*; 1Dean R. Giosio; 1James A. Venning; 1Bryce W. Pearce; 1Paul A. Brandner
V. Aumelas2and G. Maj2
1Cavitation Research Laboratory, University of Tasmania, Launceston, Tasmania, Australia
2YLEC Consultants, Grenoble, France
Abstract
Real-time measurement of microbubble concentrations is desirable in order to inform experimental results,
particularly in studies of cavitation physics. To develop these capabilities a controlled experiment using a
micro-fluidic T-junction to produce mono-disperse microbubbles was devised with the size and frequency of
microbubbles measured using a line-scan camera capable of acquiring 45k images per second. Measurements
were able to be obtained and reported in under 3 seconds from the triggering time. Tests were carried out in
quiescent water and implementation in non-stationary environments would extend the operational range. The
principal operating mode produced microbubbles on the order of 80 to 130 µm in size at frequencies ranging
from 750 to 3200 bubbles per second across the range of air and water pressures tested.
Keywords: microbubbles; cavitation
Introduction
Even small concentrations of microbubbles present within a fluid can greatly impact upon its mechanical properties. In
particular the inception of cavitation inside facilities and on board vessels is often controlled by the largest microbub-
ble which will pass through a region of low pressure. Consequently, real-time measurement of these concentrations is
desirable in order to inform predicted performance and experimental results.
In hydrodynamic flows of practical interest cavitation nucleation is invariably heterogeneous where microbubbles en-
trained in the flow or ejected from surface hydrophobic crevices provide sites of weakness, or nuclei, for the initiation
of phase change from liquid to vapour [1, 2]. The equilibrium of a microbubble becomes unstable below a critical
pressure, depending on its diameter, after which it will grow explosively. Once activated bubbles fill with vapour
and interact with the surrounding stationary or flowing liquid developing into macroscopic cavitation phenomena.
Rigorous experimental modelling of the inception and dynamics of hydrodynamic cavitation in water tunnels thus
requires control and measurement of the microbubble population. The variable pressure water (cavitation) tunnel
within the Cavitation Research Laboratory (CRL) at the Australian Maritime College has been developed with ancil-
lary systems for continuous artificial seeding and removal of microbubbles to provide controlled nuclei populations
in the test flow [3, 4]. To date this capability has been developed using direct or dilute injection of poly-disperse mi-
crobubble populations generated through the rapid de-pressurisation and cavitation of supersaturated water [5]. Whilst
poly-disperse populations (typically 10 to 100 µm in diameter) are always required to model real flows the use of
mono-disperse nuclei provides several advantages for basic research and for comparative experimental and computa-
tional work.
Microfluidic or lab-on-chip devices have been developed for mono-disperse generation of micro, or nano-bubble pop-
ulations for sono-fluidic or sono-chemical processes such as contrast agents or drug delivery vectors in medical appli-
cations [6]. These devices typically generate smaller bubbles than those suitable for nuclei and may involve the use of
surfactants [7]. Commercial devices using common materials and simple experimental set-ups generating microbub-
bles of order 10 to 100 µm at rates of order 103to 104have been developed by YLEC Consultants, France. The present
work is a collaboration between the CRL and YLEC Consultants to investigate the operational range and use of these
devices for mono-disperse cavitation nuclei seeding.
*Corresponding Author, Patrick Russell: Patrick.Russell@utas.edu.au
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
To characterise such devices, or measure populations within the CRL hydrodynamic test facility long range microscopy-
shadowgraphy is typically used. With the use of PIV cameras and dual pulse lasers bubble velocities may also be
measured. However, the processing of these high resolution image pairs is demanding such that results can not, at
present, be obtained in real-time. To address this problem a line-scan camera with a single row of densely spaced
pixels was used. The reduced number of pixels allowed the acquisition speed of the camera to be high and the spatial
resolution to be increased from standard PIV cameras in one dimension at the sacrifice of the second spatial dimen-
sion. When stacked these line measurements produce a space-time plot from which size and frequency information is
readily extracted.
Experimental overview
The L10 device from Ylec Consultants was mounted within a quiescent tank at atmospheric pressure and supplied
by independently pressurised air and water lines. The size and frequency of bubbles produced was dependent on the
balance of these pressures. The device was deemed to be in its principal operating mode when the standard deviation
of the bubble size was 10% of the measured mean, with a single production frequency.
The T-junction device was tested in a 0.05 m square acrylic tank filled with distilled water. The junction was fitted
through the base of the tank on a mount to which was connected the 4 mm pressurised air and water supply lines.
Air pressure was regulated then conditioned through a 1.0 μm filter before it reached the device. Water was supplied
from a reservoir of distilled water pressurised by air connected to a second regulator. Water and air pressures ranging
between 1.5 and 7 bar absolute in 0.5 bar increments were tested, measured using two Siemens Sitrans P DS III,
Range: 0-500 kPa, Model number: 7MF4333-1GA02-2AB1 absolute pressure transducers. These were connected to
a National Instruments 6366 USB-DAQ which sampled the pressures for 1 second during image acquisition at a rate
of 1000 Hz. The mean and standard deviation of these measurements were recorded. The standard deviation of all
pressure measurements was below 0.01 bar and the mean within 0.05 bar of the nominal pressure.
Bubbles were illuminated using a Constellation 120 W white 5600K LED light positioned directly behind the acrylic
tank. Images were captured using a Teledyne DALSA Linea Mono linescan camera with 8192×1 pixel resolution.
The single line of pixels was sampled 10000 times at a rate of 45 kHz. These rows of pixels when stacked produced
a space-time plot which was recorded on board the camera and then transmitted as a single frame to MATLAB for
processing. The camera was coupled to a Questar QM100 long-range microscope using a 1.5×Barlow lens then
bellows giving a field of view of approx 900 μm with a spatial resolution of 0.121 μm/pixel. A schematic of the
overall experimental set-up is shown in figure 1. For each condition 4 ensemble images were collected and processed
separately. All data is presented as the mean of these measurements with error bars of two standard deviations applied
where appropriate.
Figure 1: Schematic of experimental setup for the measurement of microbubble size and production rate using shadowgraphy
10th International Symposium on Cavitation - CAV2018
Baltimore, Mar
y
land, USA, Ma
y
14 – 16, 2018 CAV18-05211
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1057
Towards real-time optical measurement of microbubble content in hydrodynamic test
facilities
1Patrick S. Russell*; 1Dean R. Giosio; 1James A. Venning; 1Bryce W. Pearce; 1Paul A. Brandner
V. Aumelas2and G. Maj2
1Cavitation Research Laboratory, University of Tasmania, Launceston, Tasmania, Australia
2YLEC Consultants, Grenoble, France
Abstract
Real-time measurement of microbubble concentrations is desirable in order to inform experimental results,
particularly in studies of cavitation physics. To develop these capabilities a controlled experiment using a
micro-fluidic T-junction to produce mono-disperse microbubbles was devised with the size and frequency of
microbubbles measured using a line-scan camera capable of acquiring 45k images per second. Measurements
were able to be obtained and reported in under 3 seconds from the triggering time. Tests were carried out in
quiescent water and implementation in non-stationary environments would extend the operational range. The
principal operating mode produced microbubbles on the order of 80 to 130 µm in size at frequencies ranging
from 750 to 3200 bubbles per second across the range of air and water pressures tested.
Keywords: microbubbles; cavitation
Introduction
Even small concentrations of microbubbles present within a fluid can greatly impact upon its mechanical properties. In
particular the inception of cavitation inside facilities and on board vessels is often controlled by the largest microbub-
ble which will pass through a region of low pressure. Consequently, real-time measurement of these concentrations is
desirable in order to inform predicted performance and experimental results.
In hydrodynamic flows of practical interest cavitation nucleation is invariably heterogeneous where microbubbles en-
trained in the flow or ejected from surface hydrophobic crevices provide sites of weakness, or nuclei, for the initiation
of phase change from liquid to vapour [1, 2]. The equilibrium of a microbubble becomes unstable below a critical
pressure, depending on its diameter, after which it will grow explosively. Once activated bubbles fill with vapour
and interact with the surrounding stationary or flowing liquid developing into macroscopic cavitation phenomena.
Rigorous experimental modelling of the inception and dynamics of hydrodynamic cavitation in water tunnels thus
requires control and measurement of the microbubble population. The variable pressure water (cavitation) tunnel
within the Cavitation Research Laboratory (CRL) at the Australian Maritime College has been developed with ancil-
lary systems for continuous artificial seeding and removal of microbubbles to provide controlled nuclei populations
in the test flow [3, 4]. To date this capability has been developed using direct or dilute injection of poly-disperse mi-
crobubble populations generated through the rapid de-pressurisation and cavitation of supersaturated water [5]. Whilst
poly-disperse populations (typically 10 to 100 µm in diameter) are always required to model real flows the use of
mono-disperse nuclei provides several advantages for basic research and for comparative experimental and computa-
tional work.
Microfluidic or lab-on-chip devices have been developed for mono-disperse generation of micro, or nano-bubble pop-
ulations for sono-fluidic or sono-chemical processes such as contrast agents or drug delivery vectors in medical appli-
cations [6]. These devices typically generate smaller bubbles than those suitable for nuclei and may involve the use of
surfactants [7]. Commercial devices using common materials and simple experimental set-ups generating microbub-
bles of order 10 to 100 µm at rates of order 103to 104have been developed by YLEC Consultants, France. The present
work is a collaboration between the CRL and YLEC Consultants to investigate the operational range and use of these
devices for mono-disperse cavitation nuclei seeding.
*Corresponding Author, Patrick Russell: Patrick.Russell@utas.edu.au
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
To characterise such devices, or measure populations within the CRL hydrodynamic test facility long range microscopy-
shadowgraphy is typically used. With the use of PIV cameras and dual pulse lasers bubble velocities may also be
measured. However, the processing of these high resolution image pairs is demanding such that results can not, at
present, be obtained in real-time. To address this problem a line-scan camera with a single row of densely spaced
pixels was used. The reduced number of pixels allowed the acquisition speed of the camera to be high and the spatial
resolution to be increased from standard PIV cameras in one dimension at the sacrifice of the second spatial dimen-
sion. When stacked these line measurements produce a space-time plot from which size and frequency information is
readily extracted.
Experimental overview
The L10 device from Ylec Consultants was mounted within a quiescent tank at atmospheric pressure and supplied
by independently pressurised air and water lines. The size and frequency of bubbles produced was dependent on the
balance of these pressures. The device was deemed to be in its principal operating mode when the standard deviation
of the bubble size was 10% of the measured mean, with a single production frequency.
The T-junction device was tested in a 0.05 m square acrylic tank filled with distilled water. The junction was fitted
through the base of the tank on a mount to which was connected the 4 mm pressurised air and water supply lines.
Air pressure was regulated then conditioned through a 1.0 μm filter before it reached the device. Water was supplied
from a reservoir of distilled water pressurised by air connected to a second regulator. Water and air pressures ranging
between 1.5 and 7 bar absolute in 0.5 bar increments were tested, measured using two Siemens Sitrans P DS III,
Range: 0-500 kPa, Model number: 7MF4333-1GA02-2AB1 absolute pressure transducers. These were connected to
a National Instruments 6366 USB-DAQ which sampled the pressures for 1 second during image acquisition at a rate
of 1000 Hz. The mean and standard deviation of these measurements were recorded. The standard deviation of all
pressure measurements was below 0.01 bar and the mean within 0.05 bar of the nominal pressure.
Bubbles were illuminated using a Constellation 120 W white 5600K LED light positioned directly behind the acrylic
tank. Images were captured using a Teledyne DALSA Linea Mono linescan camera with 8192×1 pixel resolution.
The single line of pixels was sampled 10000 times at a rate of 45 kHz. These rows of pixels when stacked produced
a space-time plot which was recorded on board the camera and then transmitted as a single frame to MATLAB for
processing. The camera was coupled to a Questar QM100 long-range microscope using a 1.5×Barlow lens then
bellows giving a field of view of approx 900 μm with a spatial resolution of 0.121 μm/pixel. A schematic of the
overall experimental set-up is shown in figure 1. For each condition 4 ensemble images were collected and processed
separately. All data is presented as the mean of these measurements with error bars of two standard deviations applied
where appropriate.
Figure 1: Schematic of experimental setup for the measurement of microbubble size and production rate using shadowgraphy
10th International Symposium on Cavitation - CAV2018
Baltimore, Mar
y
land, USA, Ma
y
14 – 16, 2018 CAV18-05211
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1058
Processing method
A small portion of an image frame is shown in figure 2. Images were acquired using the MATLAB image acquisition
toolbox. Once in MATLAB each row of pixels was summed together to create a time series of integral intensity.
Peaks extracted from the series correspond to the passage of a bubble through the measurement area. The single row
of pixels corresponding to the maximum of each peak was extracted and the diameter of the bubble calculated via
thresholding of the smoothed pixel intensity. Two methods were used to collect bubble production frequency. The first
was extracted from the fast Fourier transform (FFT) of the intensity time series. In periodic production regimes this
was sufficient to assess production frequency. As the device reached the limits of its operating envelope the size and
frequency of the bubble produced fluctuated. The second method simply estimated the integral production frequency
by dividing the number of bubbles observed in the intensity time series by the length of the sample. Comparison of
this frequency estimate with the peak frequency observed in the FFT of the series allowed immediate determination
of when the device was operating in its intended manner. Table 1 shows a summary of the differences between these
frequencies. Large numbers signalled that the device was no longer producing mono-disperse bubbles.
Frequency ∆f(hz) pair (bar)
Difference 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
pwater (bar)
1.5 --- ----- - - - -
2.0 --- ----- - - - -
2.5 --- ----- - - - -
3.0 --- -1 2 3 1 3 776 - -
3.5 --77 1 1 0 1 1 1083 68 - -
4.0 --116 2 2 1 3 2 3 378 - -
4.5 - 26 3 13 2 1 1 1 2 2 2-
5.0 -- - 1 0 1 3 1 373 17 594 -
5.5 ---633 1 1 0 2 37 1723 1334 -
6.0 --- ---1 5 15 1 21 897
Table 1: For each condition the table shows the absolute difference (in Hz) between two measures of the production frequency. The first measure
uses the dominant bubble production frequency from largest peak in the FFT of pixel row intensity. The second measure of frequency is gathered
by dividing the total number of bubbles observed over the acquisition time. Large values in this table indicates that the device was operating outside
the intended principal operating mode. Conditions whose frequency measures are in agreement, ∆f<50 Hz, have been highlighted in blue. A
difference in frequency greater than 50 Hz is marked in red. A dash indicates that for this pressure combination the device either did not produce
bubbles or was not tested.
Figure 2: A sample image (left) - pair =5 bar, pwater =4.5 bar - showing 700 lines from the line-scan camera. Here the vertical axis is time, with
spatial coordinates on the horizontal axis to produce a space time plot. A portion of the regular stream of similarly sized bubbles observed in the
image has been enlarged and stretched (centre). Each row of pixels in this centre image has been summed and non-dimensionalised by the minimum
and maximum value of the overall series to produce the plotted row intensity time series (right). This time series is used to detect bubbles and find
the centre location of each bubble for sizing.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Results
The range of the principal operating mode can be established from examination of Table 1. For convenience the
pressure ratio is defined with the air pressure as the numerator. It was observed that a pressure ratio near unity is
required to remain inside the principal operating mode. The device remained stable over the largest range when an air
pressure of 3-5 bar(abs) was used. At low pressure ratios there was a rapid cessation of production with the device
simply ejecting water upon exceeding the critical ratio. In contrast, the breakdown of consistent production at high
ratios was more subtle. The device would continue to operate but sporadically shift in and out of periodic production
into a chaotic production mode. Qualitatively it was observed that this transition is often due to the disturbances
created by coalescence further up in the bubble train, however this is not revealed in the data presented here.
With the conditions of interest established, the relevant data was plotted in terms of the size and frequency in figure
3. Naturally, lower air pressures produced smaller bubbles. Lower air pressures also produced bubbles at lower
frequencies. As water pressure increased the bubble size reduced and the production frequency increased. This
indicates bubble pinch-off occurred quicker inside the junction at these pressures. The largest bubbles of consistent
size were 133 µm in diameter, produced at an air pressure of 5 bar and water pressure of 4 bar. While the smallest
bubble size was approximately 70 µm, the two conditions where this size was observed were very close to device
breakdown (pAir =3 bar, pwat er =5.5 bar & pair =5 bar, pwat er =6 bar). The smallest stable condition then was
83 µm at an air pressure of 3 bar and water pressure of 5 bar.
Bubble production rates were of order 103to 104bubbles per second. In contrast to bubble diameter production
frequency increased with water pressure but also with air pressure. The lowest production frequency was 770 Hz
(pair =3 bar, pwater =3.5 bar), while the highest consistent frequency was 3200 Hz (pair =5 bar, pwater =5.5 bar).
Higher, stable, mono-disperse production frequencies were achieved, but they were very close to device breakdown
and could not be repeated when the combination was later re-tested. Further extreme cases were found that produced
bubbles outside of these size and frequency bounds but they were either inconsistent or likely to disappear during
testing.
A plane was fitted through the size data as a function of water and air pressure.
Size(pair,pwat er)=110 +13.4pair −13.83 pwater (µm)(1)
In equation form it is clear that the mean bubble size produced was 110 µm and that variation of water and air have
similar influence but opposite effects. The size reduced or increased by approximately 13.5 µm as the difference
between these two pressures increases by 1 bar.
The adjusted R-squared residual of this fit was 0.76. Consequently approximately 24% variation in the size was
left unexplained. This result is not as strong as would be desired. To improve this result, due to the speed with
which measurements may be collected, electronically controlled pressure regulators could be used to conduct a more
detailed sweep with increased repetition of tests. However, before testing it was assumed that the device was operating
independent of its supply pressure history. Observations during testing indicate that this may not be true across
short time periods at all conditions. Acquisition of samples were slowed to give time for these effects to decay -
approximately 2 minutes for each pressure combination - but this effect may have contributed to errors observed
here. The cause for this effect may lie in the pressure supply system. Detailed tracking of the evolving production
characteristics following large pressure changes is to be conducted and compared to the supply pressure measurements
in time.
Conclusion
Bubble sizes were measured using a linescan camera to collect size and frequency measurements in near real time.
Tests were carried out in quiescent water and implementation in non-stationary environments would extend the op-
erational range. However, in this environment the principal operating mode produced microbubbles on the order of
80–130 µm in size at frequencies ranging from 750–3200 bubbles per second were produced across the range of air
and water pressures tested. Bubbles near these sizes could be used as seeded cavitation nuclei within hydrodynamic
test facilities through dilute or direct injection.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1059
Processing method
A small portion of an image frame is shown in figure 2. Images were acquired using the MATLAB image acquisition
toolbox. Once in MATLAB each row of pixels was summed together to create a time series of integral intensity.
Peaks extracted from the series correspond to the passage of a bubble through the measurement area. The single row
of pixels corresponding to the maximum of each peak was extracted and the diameter of the bubble calculated via
thresholding of the smoothed pixel intensity. Two methods were used to collect bubble production frequency. The first
was extracted from the fast Fourier transform (FFT) of the intensity time series. In periodic production regimes this
was sufficient to assess production frequency. As the device reached the limits of its operating envelope the size and
frequency of the bubble produced fluctuated. The second method simply estimated the integral production frequency
by dividing the number of bubbles observed in the intensity time series by the length of the sample. Comparison of
this frequency estimate with the peak frequency observed in the FFT of the series allowed immediate determination
of when the device was operating in its intended manner. Table 1 shows a summary of the differences between these
frequencies. Large numbers signalled that the device was no longer producing mono-disperse bubbles.
Frequency ∆f(hz) pair (bar)
Difference 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
pwater (bar)
1.5 --- ----- - - - -
2.0 --- ----- - - - -
2.5 --- ----- - - - -
3.0 --- -1 2 3 1 3 776 - -
3.5 --77 1 1 0 1 1 1083 68 - -
4.0 --116 2 2 1 3 2 3 378 - -
4.5 - 26 3 13 2 1 1 1 2 2 2-
5.0 -- - 1 0 1 3 1 373 17 594 -
5.5 ---633 1 1 0 2 37 1723 1334 -
6.0 --- ---1 5 15 1 21 897
Table 1: For each condition the table shows the absolute difference (in Hz) between two measures of the production frequency. The first measure
uses the dominant bubble production frequency from largest peak in the FFT of pixel row intensity. The second measure of frequency is gathered
by dividing the total number of bubbles observed over the acquisition time. Large values in this table indicates that the device was operating outside
the intended principal operating mode. Conditions whose frequency measures are in agreement, ∆f<50 Hz, have been highlighted in blue. A
difference in frequency greater than 50 Hz is marked in red. A dash indicates that for this pressure combination the device either did not produce
bubbles or was not tested.
Figure 2: A sample image (left) - pair =5 bar, pwater =4.5 bar - showing 700 lines from the line-scan camera. Here the vertical axis is time, with
spatial coordinates on the horizontal axis to produce a space time plot. A portion of the regular stream of similarly sized bubbles observed in the
image has been enlarged and stretched (centre). Each row of pixels in this centre image has been summed and non-dimensionalised by the minimum
and maximum value of the overall series to produce the plotted row intensity time series (right). This time series is used to detect bubbles and find
the centre location of each bubble for sizing.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Results
The range of the principal operating mode can be established from examination of Table 1. For convenience the
pressure ratio is defined with the air pressure as the numerator. It was observed that a pressure ratio near unity is
required to remain inside the principal operating mode. The device remained stable over the largest range when an air
pressure of 3-5 bar(abs) was used. At low pressure ratios there was a rapid cessation of production with the device
simply ejecting water upon exceeding the critical ratio. In contrast, the breakdown of consistent production at high
ratios was more subtle. The device would continue to operate but sporadically shift in and out of periodic production
into a chaotic production mode. Qualitatively it was observed that this transition is often due to the disturbances
created by coalescence further up in the bubble train, however this is not revealed in the data presented here.
With the conditions of interest established, the relevant data was plotted in terms of the size and frequency in figure
3. Naturally, lower air pressures produced smaller bubbles. Lower air pressures also produced bubbles at lower
frequencies. As water pressure increased the bubble size reduced and the production frequency increased. This
indicates bubble pinch-off occurred quicker inside the junction at these pressures. The largest bubbles of consistent
size were 133 µm in diameter, produced at an air pressure of 5 bar and water pressure of 4 bar. While the smallest
bubble size was approximately 70 µm, the two conditions where this size was observed were very close to device
breakdown (pAir =3 bar, pwat er =5.5 bar & pair =5 bar, pwat er =6 bar). The smallest stable condition then was
83 µm at an air pressure of 3 bar and water pressure of 5 bar.
Bubble production rates were of order 103to 104bubbles per second. In contrast to bubble diameter production
frequency increased with water pressure but also with air pressure. The lowest production frequency was 770 Hz
(pair =3 bar, pwater =3.5 bar), while the highest consistent frequency was 3200 Hz (pair =5 bar, pwater =5.5 bar).
Higher, stable, mono-disperse production frequencies were achieved, but they were very close to device breakdown
and could not be repeated when the combination was later re-tested. Further extreme cases were found that produced
bubbles outside of these size and frequency bounds but they were either inconsistent or likely to disappear during
testing.
A plane was fitted through the size data as a function of water and air pressure.
Size(pair,pwat er)=110 +13.4pair −13.83 pwater (µm)(1)
In equation form it is clear that the mean bubble size produced was 110 µm and that variation of water and air have
similar influence but opposite effects. The size reduced or increased by approximately 13.5 µm as the difference
between these two pressures increases by 1 bar.
The adjusted R-squared residual of this fit was 0.76. Consequently approximately 24% variation in the size was
left unexplained. This result is not as strong as would be desired. To improve this result, due to the speed with
which measurements may be collected, electronically controlled pressure regulators could be used to conduct a more
detailed sweep with increased repetition of tests. However, before testing it was assumed that the device was operating
independent of its supply pressure history. Observations during testing indicate that this may not be true across
short time periods at all conditions. Acquisition of samples were slowed to give time for these effects to decay -
approximately 2 minutes for each pressure combination - but this effect may have contributed to errors observed
here. The cause for this effect may lie in the pressure supply system. Detailed tracking of the evolving production
characteristics following large pressure changes is to be conducted and compared to the supply pressure measurements
in time.
Conclusion
Bubble sizes were measured using a linescan camera to collect size and frequency measurements in near real time.
Tests were carried out in quiescent water and implementation in non-stationary environments would extend the op-
erational range. However, in this environment the principal operating mode produced microbubbles on the order of
80–130 µm in size at frequencies ranging from 750–3200 bubbles per second were produced across the range of air
and water pressures tested. Bubbles near these sizes could be used as seeded cavitation nuclei within hydrodynamic
test facilities through dilute or direct injection.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1060
Figure 3: The diameter of bubbles produced (top) decreases as water pressure increases for a constant air supply pressure. Diameters increase with
air pressure. Error bars denote two standard deviations. (bottom) Bubble production frequency increases both with air and water pressure.
Figure 4: Size is plotted against both air and water supply pressures. A plane of best fit using least square residuals is created though the data.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Acknowledgements:
This project was supported by the Defence Science and Technology Group (Mr. Brendon Anderson and Dr. David
Clarke), the University of Tasmania, and the US Office of Naval Research (Dr. Ki-Han Kim, Program Officer) and
ONR Global (Dr. Woei-Min Lin) through NICOP S&T Grant no. N62909-15-1-2019.
References
[1] Christopher E. Brennen. Cavitation and bubble dynamics. Cambridge University Press, 2013.
[2] Jean-Pierre Franc and Jean-Marie Michel. Fundamentals of Cavitation. Springer Science & Business Media,
2006.
[3] PA Brandner, Y Lecoffre, and GJ Walker. Development of an Australian National Facility for Cavitation Research.
In Sixth International Symposium on Cavitation, 2006.
[4] P. A. Brandner, Y. Lecoffre, and G. J. Walker. Design Considerations in the Development of a Modern Cavitation
Tunnel. In 16th Australasian Fluid Mechanics Conference, 2007.
[5] D Giosio, BW Pearce, and PA Brandner. Influence of pressure on microbubble production rate in a confined
turbulent jet. In 20th Australasian Fluid Mechanics Conference, 2016.
[6] Joshua Owen, Paul Rademeyer, Daniel Chung, Qian Cheng, David Holroyd, Constantin Coussios, Peter Friend,
Quentin A Pankhurst, and Eleanor Stride. Magnetic targeting of microbubbles against physiologically relevant
flow conditions. Interface focus, (5), 2015.
[7] Shelley Lynn Anna. Droplets and Bubbles in Microfluidic Devices. Annual Review of Fluid Mechanics, 48(1),
2016.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1061
Figure 3: The diameter of bubbles produced (top) decreases as water pressure increases for a constant air supply pressure. Diameters increase with
air pressure. Error bars denote two standard deviations. (bottom) Bubble production frequency increases both with air and water pressure.
Figure 4: Size is plotted against both air and water supply pressures. A plane of best fit using least square residuals is created though the data.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
Acknowledgements:
This project was supported by the Defence Science and Technology Group (Mr. Brendon Anderson and Dr. David
Clarke), the University of Tasmania, and the US Office of Naval Research (Dr. Ki-Han Kim, Program Officer) and
ONR Global (Dr. Woei-Min Lin) through NICOP S&T Grant no. N62909-15-1-2019.
References
[1] Christopher E. Brennen. Cavitation and bubble dynamics. Cambridge University Press, 2013.
[2] Jean-Pierre Franc and Jean-Marie Michel. Fundamentals of Cavitation. Springer Science & Business Media,
2006.
[3] PA Brandner, Y Lecoffre, and GJ Walker. Development of an Australian National Facility for Cavitation Research.
In Sixth International Symposium on Cavitation, 2006.
[4] P. A. Brandner, Y. Lecoffre, and G. J. Walker. Design Considerations in the Development of a Modern Cavitation
Tunnel. In 16th Australasian Fluid Mechanics Conference, 2007.
[5] D Giosio, BW Pearce, and PA Brandner. Influence of pressure on microbubble production rate in a confined
turbulent jet. In 20th Australasian Fluid Mechanics Conference, 2016.
[6] Joshua Owen, Paul Rademeyer, Daniel Chung, Qian Cheng, David Holroyd, Constantin Coussios, Peter Friend,
Quentin A Pankhurst, and Eleanor Stride. Magnetic targeting of microbubbles against physiologically relevant
flow conditions. Interface focus, (5), 2015.
[7] Shelley Lynn Anna. Droplets and Bubbles in Microfluidic Devices. Annual Review of Fluid Mechanics, 48(1),
2016.
10th International Symposium on Cavitation - CAV2018
Baltimore, Maryland, USA, May 14 – 16, 2018
CAV18-05211
© 2018 ASME
Downloaded From: http://ebooks.asmedigitalcollection.asme.org on 03/08/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use