See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222194928

A look at three measurement techniques for bubble size determination

Article in Experimental Thermal and Fluid Science (EXP THERM FLUID SCI) · October 2005
DOI: 10.1016/j.expthermflusci.2005.03.018

CITATIONS READS

43 10,726

5 authors, including:

A. Vazquez Elizabeth Salinas-Rodríguez

Universidad Nacional Autónoma de México Metropolitan Autonomous University
8 PUBLICATIONS 135 CITATIONS 37 PUBLICATIONS 222 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Factors involved in the bubble entrainment when a drop collides with a liquid surface View project

All content following this page was uploaded by A. Vazquez on 20 October 2017.

The user has requested enhancement of the downloaded file.

 Experimental Thermal and Fluid Science 30 (2005) 49–57
www.elsevier.com/locate/etfs

A look at three measurement techniques for bubble

size determination
A. Vazquez a, R.M. Sanchez a, E. Salinas-Rodrı́guez a, A. Soria a,*
, R. Manasseh b

a
Group of Dispersed Multiphase Systems, Universidad Autónoma Metropolitana—Iztapalapa, Dep. I.P.H., Av. San Rafael Atlixco
No. 186, Col. Vicentina C.P. 09340, Mexico, DF
b
Energy and ThermoFluids Engineering, CSIRO Manufacturing and Infrastructure Technology, P.O. Box 56, Highett, VIC 3190, Melbourne, Australia

Received 1 October 2004; accepted 13 March 2005

Abstract

Three methods for measuring the volume and the equivalent radius of air bubbles at detachment from the tip of different capillary
tubes in quiescent water were tested. The aim was to provide an accurate cross-calibration of two of the methods against a standard
laboratory method. The inverted funnel method is the laboratory standard and performed within a 0.5% repeatability error for 50
bubble sets. The passive acoustic method performed within an accuracy between 97% and 99% with respect to the inverted funnel
method. The photographic method gave an accuracy between 88% and 96%. After improvement of the photographic method by an
empirical cutting edge criterion, its accuracy was raised between 95% and 99%. The bubble shape at detachment was found to be
fitted by a CassiniÕs oval or by a BernoulliÕs lemniscate according to the formation time. Qualitative observations on the bubble
formation stage indicated an early vertical nose growth for the larger diameter capillary tubes at short formation times. The
major-axis vertical and horizontal lengths were measured as functions of time at the formation stage.

2005 Elsevier Inc. All rights reserved.

Keywords: Bubble generators; Bubble size measurement; Passive acoustic; Inverted funnel; Visualization techniques

1. Introduction is widely known that the bubble diameter and its motion
affect the design of mass transfer equipment [20,21]. The
Gas–liquid bubble flow is essential in a great number gas injector design is meaningful in the determination of
of industrial processes such as biochemical processes: bubbling regimes and their evolution, the bubble diame-
glucose–gluconic acid biotransformation [1], fermenta- ter and motion can be studied experimentally with high
tion in bioreactors [2], biological wastewater treatment quality results where there is only one bubble. However,
[3]; and other chemical processes [4–6]. Gas–liquid flows the bubbles are generated by porous ceramic stones and
also occur in the natural environment, such as in geo- pipes or plates with multiple orifices (spargers), flexible
physics [7–11] and aerosol dynamics [12]. In spite of membranes, agitated tanks or capillary tubes, and of
the simple design of industrial equipment, the behavior course, complex bubbly flows can result with multiple
of bubbles is complicated and difficult to predict. scales of motion and bubble size.
Understanding the behavior of a single bubble can The need to understand complex bubbly flows has
support a better knowledge of the overall behavior and inspired the development of numerous techniques over
has been the subject of many studies [13–19]. Besides, it the last decade that can also specify the shape and vol-
ume of bubbles with improved accuracy and high-speed.
*
Corresponding author. Fax: +52 55 5804 4900. Such methods are useful in practical complex flows, or
E-mail address: [email protected] (A. Soria). at least have the potential to be useful. However, there

0894-1777/$ - see front matter
2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermflusci.2005.03.018
50 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57

is no calibration standard for them. The most used 2. Experimental set-up and procedures
methods are: high-speed photography [22–27]; electrical
impedance tomography [28], capillary suction probes This chapter for experimental set-up is divided into
[20], optical waveguided sensors [29,30]; endoscopic the experimental tank and bubble generators, the
optical probes [31,32]; capacitance probes [23] and pas- measuring techniques and the procedure.
sive acoustic detection [14,15,33]. There are some dis-
agreements among them, since some offer greater 2.1. Experimental tank and bubble generators
rapidity of detection but are constrained by the sensing
element sensitivity or by data interpretation issues (pas- The experiments were done in a rectangular glass
sive acoustic, capacitance and electric impedance meth- tank 0.80 · 0.80 · 0.60 m and 0.006 m in wall thickness,
ods), while others depend on the purity of the liquid large enough to neglect the wall effects [37]. The bubbles
medium and lighting effects (photographic and optical were generated at 0.506 ± 0.006 m below the liquid level.
methods). In fact, the lighting may produce a play of Fig. 1 shows the schematic experimental apparatus. The
several reflections and refractions. These should be con- temperature in the tank for the experimental runs was
sidered in order to determine a threshold level for the T = 24 ± 0.5 C; the water density q = 997 kg/m3 (bal-
bubble size estimation, since some reflections on the ance hydrometer), the viscosity l = 0.900 · 10
3 kg/
frontal part of the bubble may reduce the apparent (m/s) (Ostwald viscometer) and the surface tension
bubble size [22,34]. r = 72.80 · 10
3 N/m (Fisher tensiometer). Single bub-
Although many studies on bubble size have been made bles were generated at the bottom center line of the tank,
using these various methods, there are only a few system- using a specially-designed L-shape adaptor, 0.0012 m
atic studies comparing optical and acoustic methods, inner diameter and 1.12 m length (see Fig. 1), connecting
such as Longuet-Higginns et al. [35], Manasseh [36], the electronic valve to microcap capillary tubes (Drum-
Manasseh et al. [15] and Zhu et al. [32]. Moreover, no mond Scientific Company) with flat tops. Four capillary
published studies so far have determined the relative tubes with internal diameters ranging from 0.39 · 10
3
accuracy of the methods against a calibration standard to 0.99 · 10
3 m ±1% were used. Air was supplied with
for precisely-formed, single bubbles. In order for acoustic a regulated pressure pump, filtered and saturated with
and optical bubble-sizing methods to be applied in com- water in a globe flask in order to avoid the bubble vol-
plex bubbly flows, an accurate cross-calibration of these ume changes produced by water absorption [38]. An
techniques against a laboratory standard is required. usually closed micro-electronic valve (PSV-1, Aalborg
In this paper we obtain the size of single air bubbles Instruments and Controls Inc.), was opened at a given
in a viscous Newtonian fluid (water) in an effectively time and an air pulse was sent through the pipe, con-
unbounded domain by three methods: the photographic, necting the globe flask to the air circuit, up to the needle
the acoustic and the inverted funnel. We took as a lab- where the bubble was formed and detached. The pres-
oratory-standard reference the results obtained by the sure up-stream the valve was measured by a manometer
inverted funnel method, also known as the capillary suc- (Metron Instruments, 0–43.09 · 102 N/m2), before a
tion probe. This method is the most accurate since the pulse was produced. Bubbles were generated in the stag-
bubble size is computed from the direct volume quanti- nant liquid every 3 min in order to allow the liquid
fication. For the acoustic technique, a commercial piezo- motion induced by the previous bubble to be damped,
electric element with a lower sensitivity than that of a and avoiding hydrodynamic interactions other than
conventional hydrophone (e.g. the Bruel and Kjaer type those due to the isolated bubble motion.
8103) was used. For the photographic method, a high-
speed video camera (Motion Scope System 8000S, 2.2. Measuring techniques
MASD-Red Lake Inc.) was used. In order to find the
bubble size, the bubble shape was fitted by a BernoulliÕs 2.2.1. Video imaging
lemniscate or by a CassiniÕs oval function. We also Bubble generation was recorded with a high-speed
observed the development of a nose at the bubble top video camera (Motion System 8000S, MASD-Red Lake
at an early stage of the bubble formation. The vertical Inc.) at 4000 frames per second. The camera was oper-
(a) and horizontal (b) major-axis lengths in this genera- ated in the full frame mode which had a resolution of
tion process were measured as they evolved in time. 100 · 98 pixels. An optical array made of a Precise
With these purposes, the paper is organized as follows: Eye lens (1-61453) with less than 0.14% distortion and
In Section 2 we describe the experimental set-up, we dis- a Biconvex KB7 lens with 100 mm focal length were
cuss the main experimental methods from the literature used, giving one pixel per 9.6 · 10
3 mm. The images
and how the experiments took place. In Section 3, we were digitized directly from the cameras. Illumination
discuss our results and finally, we conclude, pointing was provided by a 500 W halogen lamp which illumi-
out some advantages, limitations and possible general- nated the rear container wall. Radiation from the lamp
izations of the various methods. was tested to be negligible by temperature measurements
 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57 51

Fig. 1. Sketch of the experimental set-up.

inside the tank. The opposite wall was covered with a dif- In this work, a piezoelectric element (Radio Shack
fusion screen to avoid undesired reflections and refrac- Co., #273-0073) was adapted to be used as a hydro-
tions as much as possible. The experiment took place phone and placed inside an aluminum holding tube with
when the electronic valve was activated, triggering both a diameter of 0.0205 ± 0.005 m and 0.52 ± 0.005 m
capture systems: the video camera and the hydrophone, length. This element was connected to a wireless micro-
in a computer-controlled environment. phone in order to amplify and convert the electric signal
into an acoustic one. The use of a wireless transducer
2.2.2. Acoustic sensor also offers a practical advantage, since industrial mea-
The acoustic method was introduced by Minnaert surements in some inaccessible geometries could be
[39], who related the bubble detachment from a nozzle made remotely. The latter was captured with an audio
to its acoustic signal. In this analysis, the bubble volu- card (Sound Blaster 64 PCI) and analyzed with Cool
metric vibration at small amplitudes was approximated Edit Pro software 2.0 (Syntrillium Corp.). Following
by the classical simple harmonic oscillator. Neglecting Manasseh et al. [15], the hydrophone was located
surface tension as well as dissipation effects and assum- 0.05 m away from the bubble generator in order to min-
ing adiabatic conditions, the Minnaert resonance imize the perturbation on the bubble dynamics and to
frequency is given by maximize the quality of the acoustic signal as much as
sffiffiffiffiffiffiffiffi possible. Testing acoustic signals showed the tank was
1 3cP sufficiently large to neglect reverberation. The hydro-
m¼ ð1Þ
A
2prequ q phone detected the bubble sound at detachment. The
bubble size was obtained by analyzing the acoustic sig-
where m is the frequency in Hz, c is the ratio of specific nal for the first five cycles with a fast Fourier transform
heats for the gas, q is the liquid density, P is the absolute (Fig. 2). We obtained the dominant frequencies in the
liquid pressure, and rA equ is the bubble radius. It can be range of 1–2 kHz, as expected from Eq. (1).
shown that surface tension corresponds to a second- Although commercial hydrophones (e.g. Bruel and
order effect [35]. This method was fully reviewed by Kjaer type 8103), have linear frequency responses in
Leighton [40] for spherical bubbles, and it has been the 1–10 kHz band, the piezoelectric element frequency
applied successfully in engineering equipment such as response used in our experiments exhibited a non-linear
aeration systems [15,41]. It has also been used for mon- behavior, as can be seen in Fig. 3 where a variable sound
itoring the equipment status and the physical and chem- pressure dependent upon the frequency is apparent. In
ical changes in different processes at real time [42]. Its addition, the novel combination of a piezoelectric ele-
great advantages are low cost and reliability. Moreover, ment and wireless microphone, while offering many
since the devices used as transducers (hydrophones) are practical advantages, appeared to saturate in ampli-
made of piezoelectric materials, this technique is inde- tude rather than displaying the classical exponentially-
pendent of fluid properties. decaying acoustic pulse observed by many authors
52 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57

The bubble volume was then estimated from the air

bridge inside the capillary tube. Considering the air as
an ideal gas, the bubble size was corrected for pressure
and surface tension effects, giving a 6% of bubble
volume change.

2.3. Procedure

A triggering circuit (TI-01 HIQUEL timer) was

adapted for synchronizing the camera, the PS-1 valve
and the hydrophone. In this way each individual bubble
was recorded by both the video system and the hydro-
phone and then collected by the inverted funnel. The
volumes and the bubble growth curves were evaluated
by video image analysis, using IMAQ Vision Builder
6.1 Software (National Instruments Corp.).
Fig. 2. Acoustic procedure: (a) acoustic signal, (b) sequence of five
cycles and (c) dominant frequency.

3. Results and discussion

3.1. Sizes

We studied the bubble shape and size at detachment

for four inner diameter (/) capillary tubes, as shown in
Fig. 4. The origin of the time axis in Fig. 4 was selected
as the time of the bubble detachment. The bubble
detachment time (t = 0) was taken as half the interval
between the last frame where the bubble is in contact
with the capillary tube and the first frame where the
bubble is separated from the capillary tip (called the
detachment frame). In this way, the actual detachment
time is coincident with our definition, with an un-
certainty bound equal to ±0.125 ms.
Two evolution patterns for bubbles detaching from
the 0.99 mm inner diameter capillary tip are shown in
Fig. 4. The bubble sizes and shapes at detachment for
Fig. 3. Typical receiving frequency characteristics of piezoelectric
element.

[35,40,43]. However, the only relevant parameter to

MinnaertÕs expression is the frequency and therefore,
the sound pressure changes are irrelevant to the bubble
size measurements provided there is only one bubble at a
time [15,44].

2.2.3. Inverted funnel

Another way to measure the bubble diameter is the
inverted funnel method, described in detail by Leighton
and Walton [43]. The bubble is trapped in an inverted
cone and induced into a capillary tube where the length
of the air column is measured. In this paper, bubbles
were caught by an inverted pyrex-glass funnel (KIM-
BLE/KONTES) with 58 cone angle. The funnel was
connected to an internal radius 0.42 ± 0.01 mm capillary
tube provided with a graduated scale. A syringe at the Fig. 4. Bubble formation and detachment at the tip of capillary tubes
tube top sucked the air bubble into the capillary tube. for four inner diameters.
 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57 53

both patterns are different and seem to be dependent

upon the elapsed time taken for the bubble to form at
the capillary tip. Small variations in the initial upstream
pressure at the capillary tip may be found responsible
for the different detachment shapes. These variations
could be appreciated by the size of a liquid plug inside
the capillary tip at the beginning of the experiment.
The length of this liquid plug changed between 0.002
and 0.012 m. In this discussion two data sets for bubbles
detached at the 0.99 mm inner diameter capillary tip are
shown, each one corresponding to a 50-bubbles sample
set with a specific bubble shape and size. Fig. 5. Bubble edge perimeter fitting (white line embodying the
The calculation of bubble volume from images of bubble) to a BernoulliÕs lemniscate and to a CassiniÕs oval, respectively.
arbitrarily distorted bubbles has been studied by Lunde
and Perkins [22] and Zhu et al. [32] (and references
therein). We performed three procedures on the photo- An equivalent radius, rP2
equ , was found from the volume,
graphic method for the evaluation of the bubble size V, obtained by the best fit integration of the solid of
at detachment. As a first procedure, (Photographic revolution around the major axis, according to the
Method I) the values of the vertical (a) and horizontal expression
(b) major-axis lengths were obtained for each bubble,  13
with the so-called Advanced Edge Detection algorithm 3V
rP2
equ ¼ ð4Þ
(from the IMAQ Vision Builder). In this algorithm the 4p
bubble edge is defined as the locus where the maximum The third bubble size estimation procedure for the
difference in grey tone values between adjacent pixels is photographic method (Photographic Method III) begins
obtained. Mean values for the 50 bubble sets were then with the superposition of the 50 images in a set, in order
evaluated
pffiffiffiffiffiand the mean equivalent radius, defined as to get a smoothed image with a better edge definition.
rP1
equ ¼ ab was also computed. A second procedure The bubble edge is defined with the so called, Clustering
(Photographic Method II), for bubble size estimation Threshold algorithm from the IMAQ Vision Builder.
at detachment, was done by fitting a BernoulliÕs lemnis- After a multi-clustering process, two mean grey values
cate or a CassiniÕs oval, according to the observed shape, are obtained, identifying the bubble and its surround-
as could be seen in the detachment frame (t = 0.125 ms) ings. The bubble edge is defined as the locus of grey level
in Fig. 4. Of course, the lemniscate and the oval func- equal to the average of the mean values. After superpo-
tions presume the bubble is axisymmetric, which is sition of the images, the a and b axes are obtained in the
clearly the case at formation in quiescent liquid. How- resulting picture, the bubble volume is evaluated consid-
ever, analogous contour-measurement methods, e.g. ering the bubble shape and finally the equivalent bubble
that of Lunde and Perkins [22] may be applicable in radius is evaluated from Eq. (4).
more general bubbly flows. The present analysis is Fifty bubbles from each capillary tube were analyzed
intended to quantify the error inherent in the optical by the three methods to generate statistically meaningful
method, presuming the best available contour measure- equivalent bubble radii, requ, as well as their correspond-
ment is used. We selected the best edge fitting to one of ing standard deviations. Furthermore, to quantify the
the mentioned curves (Fig. 5), by substituting the mea- differences between the acoustic and the photographic
sured horizontal (b) and vertical (a) axes in the analyti- methods on one hand and the inverted-funnel method
cal expressions, given by on the other, an accuracy definition was introduced as
 
Bernoulli’s lemniscate r ¼ ½cosð2hÞ  104 9=50 ð2Þ  IF 
requ
riequ 
A1
ð5Þ
Cassini’s oval r ¼ a rIF
equ
v2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
u 4 Here the superscript IF is for the inverted funnel
u
u4 b method, and the superscript i indicates the acoustic or
 t cosð2hÞ 
sin2 ð2hÞ5
a the photographic method. In Table 1, the resulting
equivalent radius and its accuracy for the three methods
ð3Þ are shown.
where In Table 1 it can be observed that the Photographic
  Method I performs with an accuracy from 86% to 99%
b with respect to the inverted funnel. The Photographic
h ¼ arctan
a Method I performs with the best accuracy (around
54 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57

Table 1
Bubble mean radius and its accuracy for the three methods
Capillary Bubble mean radius (mm)
diameter,
Inverted funnel Photographic Photographic Photographic Passive acoustic
/ (mm)
method ±0.5% Method I ±5% Method II ±1% Method III ±1% method ±0.3%
0
rIF
equ rPequ A rPequ A rPequ A rA
equ A
0.99 2.45 2.44 0.99 2.30 0.94 2.21 0.90 2.36 0.97
0.99 1.94 1.92 0.99 1.69 0.88 1.98 0.98 1.92 0.99
0.79 1.97 2.24 0.86 1.81 0.92 1.85 0.94 1.99 0.98
0.63 1.82 1.86 0.98 1.73 0.96 1.74 0.96 1.80 0.99
0.39 1.39 1.41 0.99 1.31 0.95 1.28 0.92 1.39 0.99
Each statistics calculated for 50 bubbles.

0.99); however, it should be stressed that the worst accu- Recent investigations by Leifer et al. [45] indicate that
racy also occurred under Photographic Method I, for off-axis external reflections cause a bright ring which
the 0.79 mm inner diameter capillary tube. Moreover, obscures the true bubble edge. Neglecting this effect,
the repeatability of this procedure was within 5%. The particularly if an inappropriate threshold is used, may
Photographic Methods II and III performed better than cause bubble size underprediction as large as 10–15%.
Photographic Method I in their repeatability, which was They also mention that the appropriate threshold is
around 1%. Nevertheless, as noted, Methods II and III slightly below the local background intensity. It was fol-
had lower accuracies than Method I, varying from lowing these observations and those of Nomura and
0.88 to 0.96 for Method II and from 0.90 to 0.98 for Naruse [46], who considered stationary-scene multiple
Method III. images to reduce random obscuring or drop-out noise,
The passive acoustic method, on the other hand, per- that we developed the Photographic Method III. In
formed well in both, accuracy and repeatability, since its Fig. 7 the bubble size evaluated by the three photo-
repeatability was 0.3% and its accuracy range between graphic methods are shown, together with the inverted
0.97 and 0.99. This suggests that the piezoelectric funnel results for comparison.
element performs well for these bubble sizes.
In Fig. 6 the bubble size measured by all the three 3.2. Growth
methods is shown, illustrating the bubble radius under-
estimation by the photographic method. This result Although the focus of the present paper is the com-
should be understood as due to the introduction of parison of experimental methods, owing to the compre-
reflections at the bubble surface, which cause an obscu- hensive measurements made, we are also able to report
ration of the actual bubble edge and thus a bias in the very detailed observations of the bubble growth at the
measured size. capillary. We measured the vertical (a) and the horizon-

2.6
2.6

2.4
2.4
2.2
2.2
requ (mm)

2.0
requ (mm)

2.0
1.8

1.8
1.6

1.6
1.4

1.4 1.2
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 φ (mm)
φ (mm)
Fig. 7. Bubble size measured by photographic methods and the (j)
Fig. 6. Bubble size measured with (j) inverted funnel, (s) Photo- inverted funnel as a reference. (s) Photographic Method I, (})
graphic I and (n) acoustic methods. Photographic Method II and ( ) Photographic Method III.
 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57 55

tal (b) major-axis lengths during the formation stage. gated shape with a greater vertical axis, and continuing
Different growth patterns for lemnisca-shaped bubbles up to detachment.
and for oval bubbles are apparent (see Figs. 8 and 9). For lemnisca-shaped bubbles (Fig. 9), the initial
The beginning of the formation process was taken as growth of the vertical and the horizontal axes were
the first time when one or both axes increased its initial found similar, when no stagnant gas meniscus was pres-
slope. This definition was convenient since an incipient ent at the beginning. However, the bubble shape was not
stagnant gas meniscus was observed at the tip for some close to spherical, with a nose formation apparent at its
capillary tubes, before the growth process started. The top. The behavior of this nose could be observed for
initial horizontal length was found greater than the ver- the largest diameter capillary tube, where a subsequent
tical length when the stagnant gas meniscus was retardation in vertical growth was found, followed by
observed (Figs. 8 and 9). On the other hand, when no a constant growth rate. Meanwhile the horizontal axis
stagnant gas meniscus was observed, both initial lengths continued its growth, forming the typical lemniscate
were found to vanish (Fig. 9). While the oval bubblesÕ length (greater than the typical oval horizontal length).
formation time was found dependent on the gas flow A further crossover of both axes was noted close to
rate, it seemed to be independent and close to 23 ms detachment.
for lemnisca-shaped bubbles.
The initial vertical growth rate was found greater
than the horizontal one for oval bubbles (Fig. 8), tend- 4. Conclusions
ing to a nearly spherical shape, crossing over to an elon-
Two measurement methods for determining the vol-
ume and the equivalent radius of air bubbles were eval-
5
uated. This analysis was made by comparison with a
a0.99 third, the laboratory-standard inverted-funnel method,
4 b0.99 at bubble detachment in a quiescent fluid (water). The
a0.63 inverted funnel method performed within a 0.5% repeat-
b0.63 ability error for a 50-bubble set and was taken as a ref-
growth length (mm)

3
erence for comparison with the other methods due to its
reliability [40]. A first evaluation of the photographic
2
method, which included our use of analytic functions
to fit the bubble profile, gave accuracy between 86%
1 and 99% within a 5% repeatability error. Several possi-
ble error and bias sources were pointed out, one due
to the lighting reflections and refractions which could
0
confuse the bubble-edge image with the background,
causing a bias in the size estimation. We selected an
-60 -50 -40 -30 -20 -10 0
t (ms)
image superposition method in order to improve the
edge definition criterion, which together with an equiva-
Fig. 8. Growth length for oval bubbles. lent radius estimation based on the volume evaluation
considering the true bubble shape, gave improved bub-
ble size estimation with accuracy between 90% and
5
a0.99 98% within a 1% repeatability error. While the most re-
b0.99 cent improvements on the photographic method are
a0.80 simply based on a technique of superposing the whole
4
b0.80 set of data, further improvements should have as its the-
oretical foundation an analysis of light scattering. The
growth length (mm)

a0.40
3 b0.40 bubble shape at detachment was found to be fitted by
a CassiniÕs oval or by a BernoulliÕs lemniscate, depend-
2 ing on the formation time. The bubble growth at the
capillary tip was followed during the formation stage
and its evolution was qualitatively characterized. In
1
more complex bubbly flows, the bubbles would not nec-
essarily be axisymmetric and the photographic method
0 seems to be the only one that can measure non-spherical
-25 -20 -15 -10 -5 0
bubbles accurately, when the actual bubble shape is de-
t (ms)
sired. Moreover, if the bubble volume is the only impor-
Fig. 9. Growth length for lemnisca-shaped bubbles. tant size variable, the acoustic method should be
56 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57

preferred, since acoustic data may be processed easily and sensitivity study, Continental Shelf Res. 22 (16) (2002) 2409–
and faster than image processing. Also in dark or turbid 2428.
[9] H. Medwin, N.D. Breitz, Ambient and transient bubble spectral
media, the acoustic method is available, while the photo- densities in quiescent seas and under spilling breakers, J. Geophys.
graphic method is not. Res. 94C (1989) 12571–12759.
The second measurement method evaluated was pas- [10] S. Vergniolle, G. Brandeis, Strombolian explosions 1. A large
sive acoustics, which performed with accuracy between bubble breaking at the surface of a lava column as a source of
97% and 99% and a repeatability of 0.3%, suggesting sound, J. Geophys. Res. 101 (B9) (1996) 20433–20447.
[11] S. Vergniolle, G. Brandeis, J.C. Mareschal, Strombolian explo-
the piezoelectric element performed appropriately for sions 2. Eruption dynamics determined from acoustic measure-
bubbles from 1.39 mm to 2.45 mm equivalent bubble ments, J. Geophys. Res. 101 (B9) (1996) 20449–20466.
radius. [12] S.K. Friedlander, Smoke, Dust, and Haze, Oxford University
In conclusion, the improved optical method shows Press, New York, 2000.
promise for detailed laboratory studies of bubble forma- [13] P.C. Duineveld, The rise velocity and shape of bubbles in pure
water at high Reynolds number, J. Fluid Mech. 292 (1995) 325–332.
tion processes, yielding accurate data on bubble shapes [14] R. Manasseh, Y. Yoshida, M. Rudman, Bubble formation
as well as sizes. An optical method with good contour- processes and bubble acoustic signals, in: Third International
fitting could possibly also be used for laboratory studies Conference on Multiphase Flow, Lyon, France, June 1998, p. 426.
of turbulent bubbly flows, provided access for imaging is [15] R. Manasseh, R.F. La Fontaine, J. Davy, I. Sheperd, Y.G. Zhu,
possible. The acoustic method was demonstrated to be Passive acoustic bubble sizing in sparged systems, Exp. Fluids 30
(6) (2001) 672–682.
equally accurate, and moreover, feasible using a piezo- [16] T. Maxworthy, C. Gnann, M. Kürten, F. Durst, Experiments on
electric element instead of a conventional hydrophone. the rise of air bubbles in clean viscous liquids, J. Fluid Mech. 321
This offers a major cost-saving, potentially making the (1996) 421–441.
passive acoustic technique of widespread industrial rele- [17] A. Biesheuvel, L. van Wijngaarden, Two-phase of equations for a
vance with inexpensive equipment. It is applicable to a dilute dispersion of gas bubbles in liquid, J. Fluid Mech. 148
(1998) 301–318.
very wide variety of practical industrial and laboratory [18] H. Oǧuz, A. Prosperetti, Dynamics of bubble growth and
systems. detachment from a needle, J. Fluid Mech. 257 (1993) 11–145.
[19] A.W.G. de Vries, A. Biesheuvel, L. van Wijngaarden, Notes on
the path and wake of a gas bubble rising in pure water, Int. J.
Acknowledgments Multiphase Flow 28 (2002) 1823–1835.
[20] M. Barigou, M. Greaves, A capillary suction probe for bubble size
measurement, Meas. Sci. Technol. 2 (1991) 318–326.
The authors gratefully acknowledge Prof. Ira Leifer [21] F.J. Montes, M.A. Galan, R.L. Cerro, Comparison of theoretical
for useful advice and comments and the Universidad and experimental characteristics of oscillating bubbles, Ind. Eng.
Autónoma Metropolitana for financial support. Chem. Res. 41 (2002) 6235–6245.
[22] K. Lunde, R.J. Perkins, A method for the detailed study of bubble
motion and deformation, in: A. Serizawa, T. Fukano, J. Bataille
(Eds.), Proc. Int. Conf. Multiphase Flow, Multiphase Flow 1995,
References Elsevier Science B.V., Amsterdam, 1995, pp. 395–405.
[23] I. Leifer, G. De Leeuw, L.H. Cohen, Optical measurement of
[1] J. Klein, M. Rosenberg, J. Markoš, O. Dolgoš, M. Krošlák, L. bubbles: System, design and application, J. Atmos. Oceanic
Krištofı́ková, Biotransformation of glucose to gluconic acid by Technol. 17 (10) (2000) 1392–1402.
Aspergillus niger—study of mass transfer in an airlift bioreactor, [24] M. Wu, M. Gharib, Contamination is not the only way to slow
Biochem. Eng. J. 10 (2002) 197–205. down rising air bubbles, in: American Physical Society 53rd
[2] N.A. Mensour, A. Margaritis, C.L. Briens, H. Pikington, I. Annual Meeting of the Division of Fluid Dynamics, Washington
Russell, New developments in the brewing industry using DC, 2000.
immovilized yeast cell bioreactor systems, J. Inst. Brew. 103 [25] M. Wu, M. Gharib, Experimental studies on the shape and path
(1997) 363–370. of small air bubbles rising in clean water, Phys. Fluids 14 (2002)
[3] A. Prakash, A. Margaritis, R.C. Sauders, S. Vijayan, High L49–L52.
concentrations ammonia removal by the cyanobacterium plecto- [26] K. Ellingsen, F. Risso, On the rise of an ellipsoidal bubble in
nema boryanum in a photobioreactor system, Can. J. Chem. Eng. water: Oscillatory paths and liquid-induced velocity, J. Fluid
77 (1999) 99–107. Mech. 440 (2001) 235–268.
[4] J.R. Boles, J.F. Clark, I. Leifer, L. Washburn, Temporal [27] S.S. Alves, C.I. Maia, J.M.T. Vasconcelos, A.J. Serralheiro,
Variation in natural methane seep rate due to tides, coil oil point Bubble size in aerated stirred tanks, Chem. Eng. Sci. 89 (2002)
area, California, J. Geophys. Res. 106C11 (2000) 27077–27086. 109–117.
[5] J. Magnaudet, I. Eames, The motion of high-Reynolds number [28] M.C. Kim, S. Kim, H.J. Lee, Y.J. Lee, K.Y. Kim, An
bubbles in inhomogeneous flows, Annu. Rev. Fluid Mech. 32 experimental study of electrical impedance tomography for the
(2000) 659–668. two-phase flow visualization, Int. Commun. Heat Mass Transfer
[6] R. Clift, J.R. Grace, M.E. Weber, Bubbles, Drops and Particles, 29 (2002) 193–202.
Academic Press, New York, 1978. [29] B. Saberi, K. Shakourzadeh, J. Militzer, Application of fiber optic
[7] W.E. Asher, L.M. Karle, B.J. Higgins, P.J. Farley, I. Leifer, E.C. probe to the hydrodynamic study of an industrial fluidized bed
Manahan, The influence of bubble plumes on air–seawater gas furnace, Proc. Int. Conf. Fluid Bed Combust. 14 (1997) 595–601.
transfer velocities, J. Geophys. Res. 101C (1996) 1207–12041. [30] M.R. Rza˛sa, A. Pla˛skowski, Application of optical tomography
[8] I. Leifer, R. Patro R, The bubble mechanism for transport of for measurement of aereation parameters in large water tanks,
methane from the shallow sea bed to the surface: a review Meas. Sci. Technol. 14 (2003) 199–204.
 A. Vazquez et al. / Experimental Thermal and Fluid Science 30 (2005) 49–57 57

[31] J. Chabot, S.L.P. Lee, A. Soria, H.I. de Lasa, Interaction between [38] D. Manley, Change of size of air bubbles in water containing a
bubbles and fiber optic probes in a bubble column, Can. J. Chem. small dissolved air content, Br. J. Appl. Phys. 11 (1960) 38–42.
Eng. 70 (1991) 61–68. [39] M. Minnaert, On musical air-bubbles and sounds of running
[32] Y. Zhu, J. Wu, R. Manasseh, Rapid measurement of bubble water, Phil. Mag. 16 (1933) 235–248.
size in gas–liquid flows using a bubble detection technique, [40] T.G. Leighton, The Acoustic Bubble, Academic Press,
in: Proc. 14th Australasian Fluid Mechanics Conference, Cambridge, UK, 1994.
Adelaide University, Adelaide, Australia, 10–14 December [41] R. Manasseh, Passive acoustic analysis of complex bubbly flows,
2001. in: 16th International Congress of Chemical and Process
[33] H. Chanson, R. Manasseh, Air entrainment processes in a circular Engineering, Prague, Czech Republic, 22–16 August 2004.
plunging jet: Void-fraction and acoustic measurements, J. Fluids [42] J.W.R. Boyd, J. Varley, The uses of passive measurement of
Eng. Trans. ASME 125 (2003) 910–921. acoustic emissions from chemical engineering processes, Chem.
[34] C. Bongiovanni, A. Dominguez, J.P. Chevaillier, Understanding Eng. Sci. 56 (2001) 1749–1767.
images of bubbles, Eur. J. Phys. 21 (2000) 561–570. [43] T.G. Leighton, A.J. Walton, An experimental study of the sound
[35] M.S. Longuet-Higginns, B.R. Kerman, K. Lunde, The release of emitted from gas bubbles in a liquid, Eur. J. Phys. 8 (1987) 98–104.
air bubbles from an underwater nozzle, J. Fluid Mech. 230 (1991) [44] R. Manasseh, A. Nikolovska, A. Ooi, S. Yoshida, Anisotropy
365–390. in the sound field generated by a bubble chain, J. Sound Vib. 278
[36] R. Manasseh, Acoustic sizing of bubbles at moderate to high (4–5) (2004) 807–823.
bubbling rates, in: M. Giot, F. Mayinger, G.P. Celata (Eds.), [45] I. Leifer, G. De Leeuw, G. Kunz, L.H. Cohen, Calibrating optical
Experimental Heat Transfer, Fluid Mechanics and Thermo- bubble size by the displaced-mass method, Chem. Eng. Sci. 58
dynamics, Edizioni ETS, Pisa, 1997, pp. 943–947. (2003) 5211–5216.
[37] V. Filderis, R.L. Whitmore, Experimental determination of the [46] Y. Nomura, H. Naruse, Reduction of obscuration noise using
wall effect for spheres falling axially in cylindrical vessels, Br. J. multiple images, Pattern Anal. Machine Intell., IEEE Trans. 10
Appl. Phys. 12 (1961) 490–497. (1988) 267–270.

View publication stats