International Journal of Heat and Mass Transfer 180 (2021) 121646

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/hmt

Characterization of bubbles dynamics in aperiodic formation

Mohamad Farhat a,b,∗, Maxime Chinaud b, Philippe Nerisson a, Olivier Vauquelin b
a
Institut de Radioprotection et de Sûreté Nucléaire (IRSN), 13115 St Paul Lez Durance, France
b
Aix-Marseille Université, IUSTI UMR 7343, 5 rue Enrico Fermi, 13453 Marseille, France

a r t i c l e i n f o a b s t r a c t

Article history: Bubble-bubble interactions are relevant to industrial applications, for which air is injected at different
Received 26 March 2021 flowrates from submerged orifice. These interactions provoke aperiodic bubble formation that affects their
Revised 16 June 2021
departure and size distribution, which is revealed through our experimental data. Air is injected and
Accepted 27 June 2021
regulated at flow rate between 10−4 m3 .s−1 and10−3 m3 .s−1 from a submerged orifice of inner diameter
12 mm, corresponding to a gas Weber number between 0.16 and 16. Image analysis is carried out to track
Keywords: the formation and departing of bubbles, as well as the formation of a globule, which is a large bubble
Bubble interactions formed after the bubbles coalescence. Upon increasing the flowrate, the different patterns of bubbling
Aperiodic formation observed have been classified according to the variation of the coalescence position. In a conventional
Frequency of bubbling
method, bubbles volumes are determined by inspecting the departure frequency and assuming single
Bubble size
bubbling spherical pattern. Using the same assumptions and method, we find a good agreement with
literature models. Owing to the phenomenological description of the flow, the frequency and volume
of globules are characterized. A contrast between the conventional and phenomenological approaches is
exposed, revealing a weak applicability of conventional models in estimating bubbles sizes at high flow
rates.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction mass transfer phenomena and thus trapping efficiency of contami-

nants [4,5].
Bubble dynamics are highly regarded in the design of modern Extensive researches have been conducted to characterize the
engineering systems and present within numerous industrial ap- behavior and dynamics of bubble formation [6-12], where most
plications, notably those based on mass transfer mechanisms (food of them focused on the dynamics of the single bubbling regime
industry, water and sewage treatment, oil extraction) [1]. Recently, [8, 13–18]. Single bubbling regime considers that single bubbles
the fields studying the bubble dynamics have been highly devel- are formed periodically, without any coalescence with a previ-
oped, and attempts of describing the experimental aspects of bub- ous or following bubble. Most of the models intended to estimate
bles behavior, under different flow conditions, are increasing. Bub- the bubble sizes, within the single bubbling regime, by establish-
bles motion and behavior, especially during their formation, are ing force balance approaches, neglecting bubbles interactions, and
key for the performance of the different processes. In nuclear in- making assumptions of spherical shape of the bubble [10, 13, 19].
dustry, pool scrubbing is considered as a filtration system, notably In other experimental works, flow regimes were classified and
in FCVS, that aims to reduce the release of fission products to the the transitions from one regime to another were characterized
environment [2,3]. It relies on set of bubble dynamics and mass by different dimensionless numbers such as Reynolds, Froude and
transfer into the liquid phase that will be dependent of different Weber numbers [20-24]. Zhao et al. [22] applied combined Kelvin–
flow conditions. Hence, increasing the knowledge of the latter dy- Helmholtz and Rayleigh–Taylor instability analysis in order to ob-
namics will certainly enhance either the performance of such in- tain the critical Weber number Wecritical for transition from bub-
dustrial processes and the nuclear safety assessment through more bling to jetting regime. This critical Weber number was given
accurate modeling/simulations. Moreover, high degree of precision byWecritical = 10.5 (ρl/ρ )0.5 , such that Wecritical = 326 for wa-
g
of bubbles size is required due to its strong influence in controlling ter/air, and Wecritical = 420 for water/steam. In addition to that,
Hoefele [25] have described the transition from bubbling to jet-
ting above a nozzle in terms of a Weber number and modified
Froude number Frm, where for air/water system that critical num-
∗
Corresponding author. ber was found to be 250. Gaddis et al. [10] refer to the jet regime
E-mail address: mohamad.farhat@irsn.fr (M. Farhat).

https://doi.org/10.1016/j.ijheatmasstransfer.2021.121646
0017-9310/© 2021 Elsevier Ltd. All rights reserved.
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

of the flowrate, corresponding to the onset of bubble-bubble in-

Nomenclature teractions. This is consistent with the approach of Weber number
in characterizing the regimes, where Weber number is the ratio
db Bubble diameter [m] of inertia, i.e. injection flowrate to the surface tension. This latter
D0 Orifice diameter [m] tends to reserve the spherical shape of bubbles. Beyond the single
fb Frequency of bubble departures [Hz] bubbling regime, the classification of different morphologies differs
fG Frequency of globule formation [Hz] between authors, depending on the description and the nomencla-
U
Fr Froude number: Fr = √ b [-] ture of the morphology. In this paper, we referred to these regimes
g db
by an aperiodic bubbling, where departures of bubbles is not con-
G Gravitational acceleration [m.s−2 ]
trolled solely by flow conditions.
Qinj Injected air volume flow rate [m3 .s−1 ]
Aperiodic bubbling is characterized by a nonlinear behavior of
Qcalc Calculated volume flow rate [m3 .s−1 ] bubbles due to the bubble-bubble interactions, as inertial effects
r Radius of torus revolved by a pixel [m] dominate the interfacial forces. Shen et al. [26] linked this bubbling
ρg ∗D0 ∗Vin j
ReO Orifice Reynolds number: ReO = μg [-] regime to the loss of the regular manner of releasing discrete bub-
t Time [s] bles. Recently, different experimental works investigated the aperi-
tb Time for bubble growth and departure [s] odic formation of bubbles [24, 26-32]. The bubble-bubble interac-
Ub Bubble Velocity [m.s−1 ] tions comprise the coalescence phenomena, wake effect, and inter-
Vb Bubble volume [m3 ] ference of bubbles. According to Cai et al. [31], with increasing the
VG Globule volume [m3 ] injection flowrate, successive formation of bubbles is induced, then
Vinj Gas injection velocity [m.s−2 ] bubbles undergo coalescence and interference. In addition to Kyri-
Vpx Unitary volume of pixel [m3 ] akides et al. [23], the dependence of position of the coalescence on
inj ρg ∗D0 ∗V2 the injection flowrate is also reported in [27-28], such that when
We Orifice Weber number: We = σ [-]
the flowrates increases, the start of coalescence location descends
Wes Weber number for single bubbling regime [-]
towards the orifice.
Wecritical Critical Weber number [-]
Moreover, dynamics of trailing bubbles are influenced by the
x horizontal coordinate of a pixel [-]
wake of the leading bubbles, where the latter affects the growth
y vertical coordinate of a pixel [-]
acceleration of the trailing bubbles. Zhang et al. [28] have reported
z Thickness of revolved torus [-]
the vertical elongation and rapid detachment of bubbles due to the
Greek symbols wake effect. The impact of the leading bubble’s wake and bubbles
ρg Gas density [kg.m−3 ] interaction is also examined by Roig et al. [33]. They have pre-
ρl Liquid density [kg.m−3 ] sented accurate measurements of the relative motion and defor-
σ Surface tension [N.m−1 ] mation of two large bubbles released consecutively in a quiescent
μg Air dynamic viscosity [Pa.s] liquid confined in a thin-gap cell.
μl Liquid dynamic viscosity [Pa.s] Consequently it is shown that bubble dynamics are strongly af-
fected by these diverse possible phenomena. However, the char-
acterization of these dynamics remains of key importance despite
when the force due to the gas momentum exceeds the surface the chaotic variation in terms of bubbles size and frequency [24,
tension force such that Wecritical is equal to 4. Thus, the col- 27, 29–31,34]. This is relevant to many industrial processes, where
lapse of the surface tension force induces jetting instead of bub- air is injected at high flowrates, and at conditions that induce the
bling. Through the investigation of Kyriakides et al. [23] of this chaotic behavior of bubbles. However, the characterization of bub-
approach, they considered that the orifice Reynolds number ReO ble dynamics in aperiodic motion is still poorly characterized and
(ratio between inertial forces and viscous forces, calculated with in a preliminary stage. In these regimes, models developed to pre-
orifice diameter D0 and injection velocity Vinj ) is the parameter dict the size of formed bubbles in single bubbling are not precise
which controls the transition between the different regimes. They enough. Zhang et al. [28] developed a non-linear theoretical model
stated that beyond a certain flowrate, an irregular large bubble that takes into account the wake effect and the various interac-
is formed after the coalescence of two bubbles. Then, with in- tions. However, its validity is still limited within relatively low gas
creasing the flowrate, the coalescence’s position descends towards flow rates without defining the range of validity, and with making
the orifice until it occurs at the tip of the orifice. Therefore, their the assumption of spherical shape.
classification of bubbling phenomena considered the altitude of In the present paper, two approaches to characterize the bub-
the coalescence, besides the number of bubbles constituting this ble dynamics in the injection zone are investigated, where a con-
coalescence. trast between them is shown. In the first approach, which is com-
Through literature, we have observed that this single bubbling monly used for conventional models, departure frequency is de-
regime could be characterized according to a single Weber num- termined by tracking the bubble formation process. Then bub-
ber Wes , around 0.1, as shown in Table 1. Moreover, the transi- bles size is estimated upon the assumption of spherical bubbles
tion from single bubble regime effectively relies on the increase departing singularly, without considering bubbles interactions. In
the second approach more refined, a phenomenological descrip-
tion is proposed, which is relevant to the flow morphology, in
Table 1 order to characterize the large bubbles ‘‘globules’’ constituted af-
Upper limit of the single bubbling regime ac- ter the bubbles coalescence in the injection zone. Upon track-
cording to orifice Weber number.
ing the formation of globules, frequency of globule is determined
Authors Single bubbling regime and then volumes of globules are characterized. The compari-
Kyriakides [23] Wes < 0.04 son between both approaches in characterizing the bubble vol-
Badam [24] Wes < 0.1 umes shows the weak applicability of literature models to pre-
Cieslinski [27] Wes < 0.1 dict bubbles size, at flowrates where bubble-bubble interactions
Zhang [28] Wes < 0.08
exist.
Mosdorf [29] Wes < 0.06

2
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 1. Schema of the experimental setup.

2. Material and methods Thereafter, the contour of air bubbles (hereafter referred as
structures) is detected using the edge-detection algorithm, which
Different techniques have been used in the literature for in- is followed by returning a binarized image. Holes that could ap-
vestigation of bubbling through submerged orifice: electrical ca- pear inside the bubbles due to light refraction or the alignment
pacitance tomography ECT [31], phase Doppler [35], high speed of the light source are filled in order to have homogeneous bina-
framing [24], particle image velocimetry PIV [36], Laser-photodiode rized structure. After image treatment and detection of binarized
[27]. At high flowrates, bubbles behavior limit the capability of structures (see Fig. 2), a series of horizontal lines is imposed in
techniques that provide local measurements, due to the bubble- the MATLAB code, which could be considered as sensors, and are
bubble interactions. For that, Cieslinski et al. [27] have reported situated at different altitudes above the orifice. Each of these lines,
that laser photodiode system cannot be applied at high flowrates. when crossing the bubbles, will record 1 and 0 respectively cor-
In our experiments, bubble hydrodynamics measurements are car- responding to the dispersed phase (air) and the continuous phase
ried out mainly with a high speed camera technique, and based on (water) (Fig. 3).
images processing, allowing bubbles evolution tracking. Applying this technique to the acquisition of large number of
images will reproduce the so-called XT diagram, shown in the
2.1. Experimental setup Fig. 4. XT diagram represents the evolution of the bubbles at a
certain altitude for the whole time of the acquisition. This XT di-
The experimental setup used to study the bubble formation agram is discussed in details later with emphasizing its advan-
process is shown in Fig. 1. Air is injected from single nozzle sub- tages in the temporal description of bubbling. Although XT diagram
merged into stagnant liquid in a cubic tank. It is made of Plexiglas shows spherical, hemispherical and elliptical structures, neverthe-
plates forming a 0.34 m3 cube (0.7 m × 0.7 m × 0.7 m) and is less it has no relevance with the real morphological description of
filled with tap water to a height of 50 cm. According to Kumar these bubbles because it is only depicting one spatial dimension
and Kuloor [13] and Hayes et al. [37], this height of the liquid col- along the time. Moreover, after the binarization of the images in
umn does not affect the volume of bubbles forming for a constant the optical field, the binarized structures can be analyzed in de-
air flow rate. The inner-diameter of the nozzle is 12 mm, and the tail, to extract information about bubble size and shape, besides
air is supplied to the tank under constant flow conditions, by a tracking their motion, departure from the orifice, coalescences, and
compressor through a manometer at ambient temperature. The air breakup.
flowrate Qinj is regulated from 10−4 m3 .s−1 to 10−3 m3 .s−1 with a
volumetric flowmeter. 2.3. Observation and classification of flow patterns

2.2. Data treatment Generally, flow regimes are associated with different operating
and boundary conditions, whether it is horizontal or vertical flow,
The measurement system for bubble hydrodynamics is com- such as the type of injection (single or multiple orifices, size of ori-
posed of a high-speed camera, a LED light source with a diffuser, fice), volumetric flowrates, material (hydrophobic and hydrophilic),
and associated to the CAM-WARE software The position of the and the dimensions of the flow device (micro, wall effect). As dis-
high-speed video camera is fixed at 40 cm from the cell, and it cussed previously, air is injected into a squared tank (0.34 m3 )
records the bubble formation process with a frame grabbing of 500 filled by tap water, from a plastic orifice (DO = 12 mm) which is
frames/s for an exposure time of 0.1 ms. Each acquisition of images submerged vertically. The multiphase flows that occurred in the
(1280 × 1024 pixels) lasts for 10 s, corresponding to 50 0 0 images. experiments could be characterized as bubbly regimes, however
The contrast of the raw images are adjusted and blurred to reduce comprising different morphologies depending on the flow rates
background noise. This is followed by removing noise from images that are regulated by the flowmeter (10−4 m3 .s−1 < Qinj < 10−3
through filtering the grey-scale images using a pixel-wise adapting m3 .s−1 ) corresponding to (0.16 < We < 16). The different bubbles’
low-pass wiener filter. morphologies are classified based on the position of the possible

3
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 2. Image processing; (a): Sobel filter, (b): Binarization, (c): Detection and implementing of contours.

Fig. 3. Data treatment; (a): Binarized image with an imposed digital sensor, (b):
Pixel value along the sensor.

Fig. 4. XT diagram.

coalescence of the bubbles. With the further increase of the flow

rate, the position of coalescence becomes closer to the orifice, as
shown by Kyriakides et al. [23], Cieslinski et al. [27], and Zhang at Fig. 5. Flow patterns obtained experimentally for different bubbling phenomena;
(a): Single bubbling and pairing of bubbles, (b): Formation of leading and trailing
al. [28]. The different flow patterns observed with the experimen-
bubbles followed by the coalescence above the orifice, (c): Coalescence at the edge
tal setup are illustrated on the image captures shown in Fig. 5. of the orifice, (d): Chaining and vertical elongation of bubbles.

2.3.1. . Single and pairing bubbling (double bubble) 0.15 < We < 0.63
Through the conducted experiments, even at the lowest flow leading bubble and accelerated, it elongates vertically and distorts
rate (Qinj = 10−4 m3 .s−1 ), a whole stage characterized by a sin- the shape of the leading bubble. As we increase the flow rate
gle formation of bubbles is not observed, however, it occurred par- within this regime, the frequency of bubble departure is observed
tially especially at Qinj = 10−4 m3 .s−1 . Single bubbles are formed to be increasing as well as the size of bubbles.
and rise up singularly. The stage where formation of two successive
bubbles that mostly coalesce far above the orifice, referred as pair- 2.3.2. . Coalescence above the orifice 0.63 < We < 2.55
ing regime, is dominant in this range of Weber number as shown As the flowrate increases, the position of the coalescence takes
in Fig. 5-(a). place closer to the orifice. This regime is characterized by the in-
The leading bubble which is not affected by wake of other bub- creased rate of the coalescence where a globule formed by the co-
bles takes foremost a spherical shape. However and after the for- alescence of two or three bubbles can be observed, as shown in
mation of the trailing bubble, which is entrained by the wake of Fig. 5-(b). In case of triple bubble coalescence, the first two bubbles

4
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

tracked in this display will be shifted to the third display. Through-

out this tracking, it is possible to investigate the formation of bub-
bling frequency and globules formation, mainly by analyzing the
evolution of bubbles in the first display.

3. Results and discussion

Through the different flow patterns presented, the approach of

globule formation could be introduced by two considerations. At
low flowrates where there is no bubble-bubble coalescence, each
bubble formed singularly is considered as a globule. Whereas, at
high flowrates (i.e. We > 0.63), the globule is the biggest structure
which can be formed from the orifice after the coalescence of 2, 3,
or more detached bubbles, before fragmentation. Characterization
of bubble dynamics is achieved by the coupling of temporal and
spatial description of bubbling for the different flowrates.

3.1. Temporal description

Fig. 6. Tracking the bubble departure and formation of globules. Taking in consideration the non-periodic behavior of the bub-
bling at Weber number such as We > 0.1, it is important to investi-
gate the dependence of bubbling within the time on the flow con-
are coalesced and form an irregular bubble shape, after that, an- ditions, which could be correlated afterwards with the variation of
other bubble is entrained and accelerated by the wake of the dou- bubbles size. This can be seen in the given formula Vb = Qinj × tb
ble bubbles thus penetrating them. The entrainment of the bubbles holding the relation between bubble volume and the time cor-
leads to the penetration of trailing bubble into the leading one, responding to the growth of the corresponding bubble and then
thus inducing the irregularity of shape of the large single bubble. its departure tb . Whereas, variation of the time required for bub-
ble formation upon varying the flowrates, in other words the fre-
2.3.3. . Coalescence at the edge of orifice and chaining 2.55 < We < quency of bubbling due to the bubble-bubble entrainment, affects
16 the volume of bubble released from the submerged orifice.
The position of the coalescence continues to descend until
it reaches the edge of the orifice as we increase the injection 3.3.1. XT diagram
flowrate. Successive formation of bubbles with coalescence that XT diagram, in other words space-time diagram introduced on
forms one large globule can be identified. Two or three successive Fig. 4, represents the evolution of the bubbles at a certain altitude
bubbles are coalesced at the tip of orifice as shown in Fig. 5-(c), above the orifice for the axial position. For that, it consolidates vi-
and after that it is followed by another coalescence with a fourth sually the bubbling frequency. At low flow rates, the flow exhibits
bubble. This rate of successive coalescence is consolidated by the pairing bubbles at higher altitudes, which are departed singularly,
connectivity of the bubbles to the orifice. Coalesced bubbles are each one being considered as a globule, as shown in Fig. 7-(a), (d).
vertically elongated after several formation of bubbles until they For higher flowrates and at low altitudes, the coalescence phenom-
are finally detached from the orifice [23]. We refer to this phenom- ena of the trailing and leading bubbles is detected and exposed as
ena by chaining and it is more evident as we increase the flowrate shown in Fig. 7-(b), (c). At higher altitudes, XT diagrams show the
within the considered range of Weber number, as shown in Fig. 5- conception of globule formation beyond the coalescences of trail-
(d). The latter phenomena is also reported in Hur et al. [30] and ing and leading bubbles as shown in Fig. 7-(d), (e), and (f). Hur
Kyriakides et al. [23]. et al. [30] have shown the leading and trailing bubbles in the form
of two peaks in their detachment period distributions, while these
peaks could be visualized clearly through our XT diagrams. It is re-
2.4. Tracking of structures called that the latter diagram has no relevance with the real mor-
phological description of bubbles.
In order to characterize the bubble hydrodynamics within this
non-periodic behavior of bubbles formation, it is of prime impor- 3.3.2. . Frequency of bubbling
tance to track the formation, departure and the coalescence of As bubbles are affected by the wake and entrainment of each
bubbles. Thanks to image processing and binarization of images, other, a non-periodic formation of bubbles is observed. Cieslinski
structures are distinguished from the continuous phase. Beyond et al. [27] have reported the limitation in detection of the depar-
this treatment, these structures are labelled and their different ge- ture frequency due to the bubbles coalescence. Throughout the ex-
ometrical properties are extracted. In Fig. 6, the procedure of track- periments, the approach of bubbling frequency is considered by
ing is depicted. It consists of splitting the record of bubbles into tracking the departure of the bubbles and the formation of one
3 different displays recording at same instant t. The first display large globule. Hence, two means of bubbling frequency are investi-
tracks the structures as long as they are attached and connected to gated, the frequency of departure of bubbles fb and the frequency
the orifice. A departure event is recorded whenever there is a de- of globule fG .
tachment from the orifice, and since then, the second display con- An example of the difference between the two events is given
siders the departed structure. In case where an orifice connected in the Fig. 8. A bubble is formed and its neck is collapsing prior
bubble interferes with a bubble in the second display, the merged to its detachment (Fig. 8-a), so that after detachment, we consider
structure is re-shown in the first display. A coalescence is consid- that one departure event has been achieved (Fig. 8-b). Here, count-
ered whenever there is a merging of two or more structures result- ing an event for globule formation is not yet ensured, and this
ing in one labelled structure. Whereas, in case there is more than depends on the following stages subjected to the bubble. Conse-
one structure that is detached from the orifice, the second display quently, after the detachment, the entrainment of the latter bub-
tracks the new structure and therefore, the preceding structures ble induces the accelerated formation of a trailing bubble, which is

5
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 7. XT diagrams at different altitudes and for different regimes; at H = 1 cm, (a): Qinj = 10−4 m3 .s−1 , (b): Qinj =4 × 10−4 m3 .s−1 , (c): Qinj = 9 × 10−4 m3 .s−1 and at
H = 5 cm, (d): Qinj = 10−4 m3 .s−1 , (e): Qinj = 4 × 10−4 m3 .s−1 , (f): Qinj = 9 × 10−4 m3 .s−1 .

Fig. 8. Departure of bubbles resulting in globule formation.

vertically elongated and is still connected to the orifice (Fig. 8-c). method described in Section 2, the frequency of distribution is de-
After the detachment of the resulting globule, consisting of the co- picted in Fig. 9. At low flowrates ‘’pairing regime’’, doublet bub-
alesced bubbles, another departure event is considered as well as ble is formed after the acceleration induced by the leading bubble
an event of globule formation (Fig. 8-d). on the trailing one, due to the wake effect. However, this doublet
will not affect the growth of the following bubble. This phenom-
3.2. Conventional approach for predicting bubble volume Vb ena affects departure frequency, and it is shown in the distribution
of Fig. 9-(a). The position of coalescence descends as we increase
Models for determining the size of bubbles at single submerged the flowrate, thus the wake effect of bubbles intervenes contin-
orifice classified the different stages describing the bubble, and uously during the flow. The variation of departure frequency for
then established force balance approaches. Most of these models different flowrates is depicted in Fig. 10. There is a considerable
made the assumption of spherical shape of the bubble within the increase from low to intermediate flowrates until reaching a max-
two stage description [8, 10, 13, 20]. In the first stage, the bubble imum value at We = 1.43 (Qinj = 3 × 10−4 m3 .s−1 ). Afterthat,
expands while its base remains attached to the tip of the orifice with a further increase of the flowrate, the frequency decreases
(Fig. 8-b). Then in the second stage, when lifting forces exceed the until it fluctuates around a constant value. This is consolidated
restraining forces, the bubble base rises above the tip and stays in by the fact at regimes where inertia dominates over surface ten-
contact with the orifice through a neck (Fig. 8-c). Within this single sion, bubbling frequency tends to stabilize at constant frequency,
bubbling regime, bubbles interactions are so neglected. Upon these whereas bubble volume is proportional to gas flowrate [24]. Hur
considerations, characteristic bubble size by mean of determining et al. [30] reported that frequency of departure increases as the
bubble formation time tb = Vb /Qinj is such that tb = 1/fb . orifice diameter decreases, as they used 3 different sizes (0.6 mm,
1.0 mm, 1.8 mm) of orifices. Confirming their statement, our ori-
3.2.1. . Departure frequency fb fice (12 mm) is much larger, and we obtained lower frequency of
For the departure frequency, an event is recorded at the instant bubbles departure (about 3–4 times lower for a given We). Nev-
where there is a structure detachment from the orifice. In case this ertheless, our profile of frequency evolution, upon the variation of
bubble is re-attached to the orifice by mean of coalescence, the the flowrate, is consistent with Hur et al. data [30]. In both pro-
following departure of merged structure is considered as a second files, there is a maximum of detachment frequency at intermediate
event, as shown in Fig. 8. After recording bubbles departures with

6
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 9. Histograms of departure frequency fb ;(a): Qinj = 10−4 m3 .s−1 (We = 0.16), (b): Qinj = 6 × 10−4 m3 .s−1 (We = 5.7).

Fig. 10. Variation of departure frequency as function of Weber number.

Table 2 experimental works regarding the gas injection into liquid use it.
Correlations developed to evaluate the volume of bubble.
Given uncertainties on Wecirtical value in literature (see above),
References Correlation Gaddis correlation has been extrapolated to the complete range
Qinj 2 3 of We studied experimentally. As mentioned before, these correla-
Krevelen [20] Vb Krevelen = 1.722( g
) /5 tions have made the assumption of spherical shape of the bubbles.
4/ 4/
81 μ Q 5 3 Hence, the equivalent diameter db of sphere (see Fig. 12) is com-
Gaddis [10] Vb Gaddis l l
135 Q 2
= π6 [( 6gDρo σ ) 3 + ( g ρ l πinj ) + ( 4 π 2injg ) ] /4 
Q 2 3/ puted such that db = 3 π6 Vb . As expected regarding to the Vb
Davidson [8] Vb Davidson = 1.138( inj g
) 5
results, db fits well the models considered.
On the basis of the morphological description of bubbles
throughout our experiments, the two stage description (expansion,
flowrates, and as injection velocity increases, the frequency tends detachment) as well as the spherical assumption are not consis-
to a constant value. tent as observed experimentally (see Section 2.3). It is considered
as a critical limitation, especially in aperiodic formation of bub-
3.2.2. . Bubbles volume Vb bles (non-single bubbling), amid bubble-bubble interactions that
Under the constant gas inflow Qinj [10, 24], we computed deforms considerably the shape of bubble (non-spherical shape).
Vb =Qinj × tb = Qinj /fb . This volume corresponds to the bub- This inaccurate characterization of bubbles can have a negative
ble formed without exhibiting any coalescence, supposing a single impact on the modelling of bubble dynamics in industrial appli-
bubbling regime. Some of the correlations commonly used for pre- cations, where air or non-condensable gas are mainly injected at
dicting the bubble volume Vb in literature, are shown in table 2. high flowrates. For characterizing the representative bubbles of the
Gaddis et al. [10] defining the validity range of their model such flow, we investigated in the following paragraphs the volume of
that 0 < We < 4. They justified that by referring to a critical large bubbles, called globules, resulting from the bubbles coales-
Weber number Wecritical = 4, for the transition from single bub- cence in the injection zone.
bling regime to jetting regime. It is recalled that these correlations
were developed, within the single bubbling regime, by establishing 3.3. Phenomenological approach and globule volume
a force balance approaches. Bubble-bubble interactions were ne-
glected, and assumption of spherical shape of the bubble has been In aperiodic behavior, bubble-bubble interactions affect the size
made. of the detaching bubble, thus broaden the size distribution [26,
In Fig. 11 the good agreement of the characteristic bubble vol- 30]. Wake effect of leading bubbles influences the size of trail-
ume for the different correlations is shown, especially that of ing bubbles [28], accelerates the departure of a certain bubble thus
Davidson et al. [8], which is very common in literature and most minimizing its size. Bubble-bubble coalescence at the edge of the

7
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 11. Comparison between bubble characteristic volume Vb obtained with models from literature.

Fig. 12. Comparison between equivalent diameter db with models from literature.

3.3.1. . Frequency of globules fG

For the frequency of globule, the detection of globule forma-
tion is recorded by tracking the growth and the rise of bubbles
that have detached from the orifice. Therefore, an event is recorded
when a complete formation of a globule is achieved, whether it
is a result of coalescence of two or more bubbles, when it is de-
tached completely and finally from the orifice. It is noted that
for characterizing the formation of globules, an additional post-
treatment of the bubble evolution presented in Section 2 is re-
quired. The numerical quantification of the globules formation is
Fig. 13. Globule formation at different altitudes.
done through analyzing a chronograph of bubbles evolution (see
Fig. 14), on which we are capable of differentiating between the
events of bubbles formation, detachment, and coalescences.
Binarization errors (t = 2.55 s, t = 3.12 s) and channeling ef-
fects (t = 3.34 s) are filtered automatically. This is done by set-
orifice increases the size of the departed bubble. In relevance for ting a threshold of minimum period time in order to consider
what is given above, we intend as a first step to avoid the con- a new event. Then the events of bubble formation and depar-
troversial determination of the size of each bubble departure from ture are manipulated manually (t = 2.83 s), upon the observa-
the orifice (We > 0.63), though we opted to characterize the size of tion of the chronograph. In Fig. 15, the distributions of globule
globules, i.e. as mentioned before the biggest structure that can be frequencies is shown. In comparison with Fig. 9, we can observe
formed in this zone, before fragmentation. Given that, the position that the frequency of departure (Fig. 9-a) is the same as the fre-
of globule formation varies according to the position of the coales- quency of the globule (Fig. 15-a) for low Qinj = 10−4 m3 .s−1 . This
cence and also to the number of bubbles forming this globule [23, could be justified by the fact that there is no coalescence phenom-
27-28]. Besides, the temporal and size characterizations depend on ena at low flowrates [24, 31]. However, this occurs after consid-
the tracking of globule formation, which could occur at different ering each departure bubble as a globule, since this pairing oc-
altitudes as shown in Fig. 13.

8
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 14. Chronograph of bubbling events; Bubble birth (solid lines), Breakup followed by coalescence and re-attachment to the orifice (dashed lines), and coalescence (dotted
lines); (a): Raw chronograph, (b): Filtered chronograph.

Fig. 15. Histograms of globule frequency fG ; (a): Qinj = 10−4 m3 .s−1 (We = 0.16), (b): Qinj = 6 × 10−4 m3 .s−1 (We = 5.7).

curs at high altitudes beyond the tracking. As the flowrate in- frequency but at lower frequencies and with maximum frequency
creases, bubble-bubble interactions are generated and coalescence corresponding to Qinj = 2 × 10−4 m3 .s−1 (We = 0.63).
of bubbles occur. A globule event occurs after several bubbles co-
alescence. This results in a difference between the events count- 3.3.2. . Globules volume VG
ing for bubbles departure (Fig. 9-b) and that of globule formations Beyond the tracking process, the globule volume is computed
(Fig. 15-b): fG ≤ fb . For the globule frequency evolution as shown numerically, once its definitive size is achieved (detected as men-
in Fig. 16, it takes nearly the same profile of evolution of departure tioned in the methodology above). For that, a cylindrical revolution
of the pixels is assumed around an equidistant line figure (17- a),

9
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 16. Variation of globules formation’s frequency as function of Weber number.

Fig. 17. Tracking and numerical computation of globule volume VG; (a): Tracked globule, (b): Schema of unitary volume of a pixel.

which bisects the structure radially into two equal parts at each volume evolution of bubbles, while they’re attached to the orifice,
vertical position of the structure. Actually, one part is rotated at corresponding to Qinj = 8 × 10−4 m3 .s−1 . Low amplitude peaks
360° around the equidistant line with axi-symmetric assumption corresponding to one bubble or globule are observed. Moreover,
(the two volumes induced by each part are equal in this case). The high amplitude peaks corresponding to structures attached to the
unitary volume of a pixel corresponds to torus of revolution such orifice are observed, which are formed of several bubbles, due to
that its thickness z = 1 pixel. Hence, this volume is equivalent to coalescences, vertical elongation, and chaining. The post-treatment
of these signal aims to reconstruct the filtered signals representing
Vpx = π r 1 2 z − π r 2 2 z (1) the formation of globules through two successive processes.
The first process considers the rectification and filling of the
Where r1 and r2 are the radii of the internal and external torus cracked crests appearing at t = 0.4 s and t = 1.15 s in Fig. 19-(a).
revolved by the pixel. Knowing that the coordinates of pixel (x) As structures are tracked as long as they are attached to the ori-
corresponds to its center, so we have: fice, the bubble volume is computed until its departure (Fig. 20-a),
1 then the computation begins for another bubble that is undergoing
r1 = x + (2) formation that appears as a slip in the signal like at t = 0.4 s and
2
t = 1.15 s in Fig. 19-(a). Consequently, as the trailing bubble pen-
1 etrates into the previous bubble, a jump in the signal can be seen
r2 = x − (3)
2 since the new structure, formed by these two bubble, is attached
to the orifice and so it is tracked (Fig. 20-b), therefore, the globule
Thus, the volume of revolution (1-pixel torus) is computed for
volume is computed until its final departure (Fig. 20-c). Hence, in
each pixel (rotation, see Fig. 17-b), such that
order to avoid overestimation in the quantification of peaks rep-
Vpx = 2π x z (4) resenting the globules formation, the filling of the cracked crest
is done as shown in Fig. 19-(b). The second process relies on the
Eventually, the volume of the structure is the sum of all torus
splitting and differentiation of crests, corresponding to chaining
volumes calculated along the vertical equidistant line from the bot-
formed by more than 2 bubbles, into globules. This could been
tom pixel yb until the top pixel yt , such that:
seen in Fig. 19 at t = 0.55 s and 0.6 s, where there was two peaks

yt
existing in one crest. However in the filtered signal they are spitted
VG = 2π x y z (5) into two crests with each has its own peak, representing a globule
i=yb formation. It is justified by the fact the chaining of bubbles repre-
Therefore, signals recording the evolution of bubbles volume, sent several coalescence of globules, which are vertically elongated.
which are tracked until their final detachment from the orifice, are Therefore the post-treated signals are constructed where the
constituted for each flowrate. Fig. 18 shows a raw signal for the maxima of their peaks represent the final volume of globules VG

10
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 18. Evolution of bubbles volume as function of time for Qinj = 8 × 10−4 m3 .s−1 (We = 10.2).

Fig. 19. Post-treatment of volume signal; (a): Raw signal, (b): Treated signal.

Fig. 20. Tracking of globules; (a): Departure of bubble at t = 0.4 s, (b): Coalescence and reattachment of the structure to the orifice at t = 0.44 s, (c): Departure and
formation of the globule at t = 0.5 s.

at the instant of detachment. A histogram is constructed to inspect Fig. 22 shows both the variation of bubble volume Vb and aver-
the distribution of globule volumes, for each flowrate, as shown in age globules volume VG as function of Weber number, upon vary-
Fig. 21. Moreover, Gaussian curves fitting the histograms provide ing the flowrate. At low flowrates, globule average volume is com-
an average globule volume for each flow rate. These results have parable to the characteristic bubble volume, thus fitting the mod-
shown that increasing the flowrate influences the variation of bub- els. This is due to the fact that at the corresponding regime, the
bles volume while conserving static bubbling frequency [24]. conception of globule formation is not attained yet, although the
At low flow rates, bubbles tend to exhibit spherical and ellipti- acceleration of bubbles formation is induced. In other words, the
cal shapes, however, the increase of flowrate induces more bubble- volume of bubbles in this regime corresponds to single bubbles
bubble interactions, so that the shapes become irregular depending formed without prior coalescence. However, with the increasing
on the position and number of coalescences. of flowrate and the onset of globules formation, the divergence of

11
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

Fig. 21. Histograms of VG distribution; (a): Qinj = 3 × 10−4 m3 .s−1 (We = 1.4), (b): Qinj = 7 × 10−4 m3 .s−1 (We = 7.8), (c): Qinj = 10−3 m3 .s−1 (We = 16).

Fig. 22. Difference between classical and phenomenological determination of bubble volume.

shows the computed flowrate experimentally (Qcalc = VG x fG ) as

function of the real injected flowrate Qinj , adjusted during exper-
iments. The satisfactory agreement consolidates the phenomeno-
logical approach of globule characterization, as well as the axisym-
metric assumption for VG computation.
This non-negligible difference between Vb and VG confirms
the impact of bubble-bubble interaction. The comparison between
both approaches shows the weak applicability of literature models
to predict bubbles size at high flowrates, and suggests the phe-
nomenological approach in characterization of bubbles dynamics.

4. Conclusion

Bubble-bubble interaction is considered to be key phenomena

of aperiodic formation of bubbles, especially upon operating with
high flowrates. Experimental work is carried out to reveal the im-
pact of these interactions on the formation of bubbles at a sub-
merged orifice. Based on the different bubble’s morphologies that
Fig. 23. Validation of the experimental flowrate with the real injected flowrate.
have been observed, we first investigated the bubble-bubble in-
teractions and then classified different bubbling regimes depend-
ing on the position of coalescence. The biggest bubble formed
average bubble volume is apparent. Moreover, this difference in- above the orifice, after the coalescence of bubbles, is referred by
creases with the increase of flowrates, which means that VG is not a globule. On the basis of the morphological description of bub-
fitted in models anymore. bles throughout our experiments, two approaches in characteriz-
We postulate that the formation of globule is the relevant vol- ing the bubbles sizes are investigated. The first conventional ap-
ume of formed bubble along the different flow morphologies in proach is based on the same assumptions than most of literature
the injection zone. In order to show the consistence of VG , Fig. 23 models, where the single bubbling regime and spherical shape of

12
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

bubbles are assumed. The comparison of characteristic bubble size [5] B. Aboulhasanzadeh, G. Tryggvason, Effect of bubble interactions on mass
(Vb , db ), which is computed after determining the bubble depar- transfer in bubbly flow, Int. J. Heat Mass Transf. 79 (2014) 390–396, doi:10.
1016/j.ijheatmasstransfer.2014.08.027.
ture frequency fb , shows a good agreement with literature mod- [6] Z.L. Yang, Truc-Nam Dinh, R.R. Nourgaliev, B.R. Sehgal, Numerical investiga-
els. However, this approach is not consistent with flow morphology tion of bubble growth and detachment by the lattice-Boltzmann method, Int.
experimentally observed. Therefore and through a proposed phe- J. Heat Mass Transf. 44 (2001) 195–206, doi:10.1016/S0017-9310(0 0)0 0101-0.
[7] L. Davidson, E.H. Amick Jr, Formation of gas bubbles at horizontal orifices,
nomenological approach, we aimed to characterize experimentally AIChE J. 2 (3) (1956) 337–342, doi:10.10 02/aic.690 020309.
the globules size. First, the globule frequency fG by tracking the [8] J. Davidson, B. Schuler, Bubble formation at an orifice in a viscous liquid, Trans.
globules formation through image processing is quantified. Then Inst. Chem. Eng. 38 (1960) 144–154.
[9] T. Miyahara, M. Iwata, T. Takahashi, Bubble formation pattern with weeping at
its volume VG is computed by reconstructing the shape of bubbles
a submerged orifice, J. Chem. Eng. Jpn. 17 (6) (1984) 592–597, doi:10.1252/JCEJ.
based on axisymmetric assumption. The globule sizes are found to 17.592.
be generally larger than sizes determined by the classical approach, [10] E. Gaddis, A. Vogelpohl, Bubble formation in quiescent liquids under con-
stant flow conditions, Chem. Eng. Sci. 41 (1) (1986) 97–105, doi:10.1016/
showing a weak applicability of literature models to predict bub-
0 0 09- 2509(86)85202- 2.
bles size at high Weber number in the injection zone of such flows. [11] J. Best, A. Kucera, A numerical investigation of non-spherical rebounding bub-
This phenomenological approach is validated by comparing Qcalc bles, J. Fluid Mech. 245 (1992) 137–154, doi:10.1017/S0 0221120920 0 0387.
(VG , fG ) to the real injected flowrate Qinj , which are consistent. [12] G. Yang, B. Du, and L. Fan, “Bubble formation and dynamics in gas–liquid–solid
fluidization—A review” Chemical engineering science, vol. 62, no. 1–2, pp. 2–
This disparity between phenomenological investigation and con- 27, 2007. https://doi.org/10.1016/j.ces.2006.08.021
ventional approach is relevant for industrial applications, where [13] R. Kumar, N. Kuloor, The formation of bubbles and drops, Adv. Chem. Eng. 8
air is often injected at high flowrates. In this paper, awareness is (1970) 255–368, doi:10.1016/S0065-2377(08)60186-6.
[14] A. Marmur, E. Rubin, A theoretical model for bubble formation at an orifice
raised in modelling of such applications regarding the precision submerged in an inviscid liquid, Chem. Eng. Sci. 31 (6) (1976) 453–463, doi:10.
of dynamics quantification, notably mass transfer mechanisms. In- 1016/0 0 09-2509(76)80 030-9.
deed, at high Weber number, the contrast of bubble volumes deter- [15] K. Terasaka, H. Tsuge, Bubble formation at a single orifice in highly viscous
liquids, J. Chem. Eng. Jpn. 23 (2) (1990) 160–165, doi:10.1252/jcej.23.160.
mined by the two approaches would necessarily lead to more sig- [16] A. Dani, P. Guiraud, A. Cockx, Local measurement of oxygen transfer around
nificant differences between the relative exchange surfaces, closely a single bubble by planar laser-induced fluorescence, Chem. Eng. Sci. 62 (24)
linked to mass transfer phenomena. For example, if we suppose a (2007) 7245–7252, doi:10.1016/j.ces.2007.08.047.
[17] L. Liu, H. Yan, G. Zhao, Experimental studies on the shape and motion of air
spherical shape of bubble at We = 16, and then compute the sur-
bubbles in viscous liquids, Exp. Therm. Fluid Sci. 62 (2015) 109–121, doi:10.
face areas determined by the different approaches, we would have 1016/j.expthermflusci.2014.11.018.
an underestimation of 23% for each bubble. Besides, this margin of [18] S. Aoyama, K. Hayashi, S. Hosokawa, A. Tomiyama, Shapes of ellipsoidal bub-
bles in infinite stagnant liquids, Int. J. Multiphase. Flow 79 (2016) 23–30,
error will probably increase with the non-spherical shape shown in
doi:10.1016/j.ijmultiphaseflow.2015.10.003.
morphological description. Hence, mass transfer estimation would [19] M. Jamialahmadi, M. Zehtaban, H. Müller-Steinhagen, A. Sarrafi, J. Smith, Study
be more precise by considering the proposed phenomenological of bubble formation under constant flow conditions, Chem. Eng. Res. Des. 79
approach, within bubbling flows exhibiting bubble-bubble interac- (5) (2001) 523–532, doi:10.1205/02638760152424299.
[20] D.v. Krevelen, P. Hoftijzer, Studies of gas-bubble formation, Chem. Eng. Prog.
tions. 46 (1) (1950) 29–35.
[21] McCann, R. Prince, Regimes of bubbling at a submerged orifice, Chem. Eng. Sci.
26 (10) (1971) 1505–1512, doi:10.1016/0 0 09- 2509(71)86042- 6.
Declaration of Competing Interest
[22] Y.-.F. Zhao, G.A. Irons, The breakup of bubbles into jets during submerged gas
injection, Metall. Trans. B 21 (6) (1990) 997–1003, doi:10.1007/BF02670270.
The authors declare that they have no known competing finan- [23] N. Kyriakides, E. Kastrinakis, S. Nychas, A. Goulas, Bubbling from nozzles sub-
cial interests or personal relationships that could have appeared to merged in water: transitions between bubbling regimes, Can. J. Chem. Eng. 75
(4) (1997) 684–691, doi:10.1002/cjce.5450750405.
influence the work reported in this paper. [24] V. Badam, V. Buwa, F. Durst, Experimental investigations of regimes of bubble
formation on submerged orifices under constant flow condition, Can. J. Chem.
Eng. 85 (3) (2007) 257–267, doi:10.1002/cjce.5450850301.
CRediT authorship contribution statement
[25] E.O. Hoefele, J.K. Brimacombe, Flow regimes in submerged gas injection, Met-
all. Trans. 10 (1979) 631–648, doi:10.1007/BF02662566.
Mohamad Farhat: Resources, Methodology, Software, Investiga- [26] Y. Shen, L. Hu, W. Chen, X. Fu, Periodic and aperiodic bubbling in submerged
tion, Writing – original draft. Maxime Chinaud: Resources, Con- gas-liquid jets through a micro-channel, Phys. Fluids 29 (4) (2017) 047104,
doi:10.1063/1.4979483.
ceptualization, Software, Investigation, Writing – review & editing. [27] J. Cieslinski, R. Mosdorf, Gas bubble dynamics—Experiment and frac-
Philippe Nerisson: Supervision, Writing – review & editing. Olivier tal analysis, Int. J. Heat Mass Transf. 48 (9) (2005) 1808–1818
Vauquelin: Supervision, Writing – review & editing. https://doi.org/10.101-6/j.ijheatmasstransfer.20 04.12.0 02.
[28] L. Zhang, M. Shoji, Aperiodic bubble formation from a submerged ori-
fice, Chem. Eng. Sci. 56 (18) (2001) 5371–5381, doi:10.1016/S0 0 09-2509(01)
Acknowledgement 00241-X.
[29] R. Mosdorf, T. Wyszkowski, Experimental investigations of deterministic chaos
appearance in bubbling flow, Int. J. Heat Mass Transf. 54 (23–24) (2011) 5060–
This work was performed in the French Institut de Radiopro- 5069, doi:10.1016/j.ijheatmasstransfer.2011.07.023.
tection et de Sûreté Nucléaire (IRSN), with the financial support of [30] Y.G. Hur, J.H. Yang, H. Jung, S.B. Park, Origin of regime transition to turbulent
French Région Sud - Provence-Alpes-Côte d’Azur and Electricité de flow in bubble column: orifice-and column-induced transitions, Int. J. Multiph.
Flow 50 (2013) 89–97, doi:10.1016/j.ijmultiphaseflow.2012.11.001.
France (EDF). [31] Q. Cai, X. Shen, C. Shen, G. Dai, A simple method for identifying bub-
bling/jetting regimes transition from large submerged orifices using electri-
References cal capacitance tomography (ect, Can. J. Chem. Eng. 88 (3) (2010) 340–349,
doi:10.1002/cjce.20295.
[1] R. Clift, J. Grace, M. Weber, Bubbles, Drops and Particles, Academic Press, N.Y, [32] H. Li, Y. Mouline, L. Choplin, N. Midoux, Chaotic bubble coalescence in non-
1978. Newtonian fluids, Int. J. Multiph. Flow 23 (4) (1997) 713–723, doi:10.1016/
[2] D. Jacquemain, S. Guentay, S. Basu, M. Sonnenkalb, L. Lebel, J. Ball, H. Allelein, S0301-9322(97)0 0 0 04-9.
M. Liebana, B. Eckardt, N. Losch, et al., OECD/NEA/CSNI Status Report on Fil- [33] S.G. Huisman, P. Ern, V. Roig, Interaction and coalescence of large bubbles ris-
tered Containment Venting, Organization for Economic Co-Operation and De- ing in a thin gap, Phys. Rev. E 85 (2) (2012) 027302, doi:10.1103/PhysRevE.-85.
velopment, 2014 tech. rep.. 027302.
[3] J. Wen, H. Gu, Z. Sun, Y. Zhou, A theoretical model and experiment valida- [34] X. Shen, J.P. Schlegel, S. Chen, S. Rassame, M.J. Griffiths, T. Hibiki, M. Ishii, Flow
tion on filtration characteristics of methyl iodide in bubble column, Int. J. characteristics and void fraction prediction in large diameter pipes, in: Fron-
Heat Mass Transf. 114 (2017) 1263–1273, doi:10.1016/j.ijheatmasstransfer.2017. tiers and Progress in Multiphase Flow I, Springer, 2014, pp. 55–103, doi:10.
07.023. 1007/978- 3- 319- 04358- 6_2.
[4] Y. Abe, K. Fujiwarab, S. Saitob, T. Yuasab, A. Kaneko, ‘Bubble dynamics with [35] F.O. Gréhan, T. Girasole, G. Gouesbet, F. Durst, C. Tropea, Measurement of bub-
aerosol during pool scrubbing, Nucl. Eng. Des. 337 (2018) 96–107, doi:10.1016/ bles by phase doppler technique and trajectory ambiguity, in: Developments
j.nucengdes.2018.06.017.

13
M. Farhat, M. Chinaud, P. Nerisson et al. International Journal of Heat and Mass Transfer 180 (2021) 121646

in Laser Techniques and Applications to Fluid Mechanics, 1996, pp. 290–302, 27–June 1, 2001, New Orleans, LA, USA), Tulane Univ., New Orleans, 2021
doi:10.1007/978- 3- 642- 79965- 5_19. icmf530[CD-ROM].
[36] A. Fujiwara, et al., Flow structure around rising bubble measured by PIV/LIF [37] W. Hayes III, B.W. Hardy, C.D. Holland, Formation of gas bubbles at submerged
(effect of shear rate and bubble size), in: E.E. Michaelides (Ed.), Proceed- orifices, AIChE J. 5 (3) (1959) 319–324, doi:10.10 02/aic.690 050314.
ings of the ICMF-2001 (4th International Conference on Multiphase Flow, May

14