Research Article 2321

Feishi Xu1
Characterization of Bubble Shapes in Non-
Noel Midoux2
Huai-Zhi Li2 Newtonian Fluids by Parametric Equations
Gilles Hébrard1
Nicolas Dietrich1,* Based on experiments with single air bubbles rising in stagnant non-Newtonian
fluids, an innovative model containing the aspect ratio (E) and two parameters
(a, b) was proposed and proved to be capable of characterizing the bubble shape
from spherical/ellipsoidal to prolate/oblate-tear with good accuracy. Several
impacts on bubble deformation were investigated, involving the rheological prop-
erties of the fluids and different forces exerted on the bubble, which were quanti-
fied by multiple dimensionless numbers (e.g., Reynolds, Eötvös, and Deborah
number). Within a wide range, the empirical correlations were obtained for
parameter b, and between a and b. Together with the shape model, a complete
system was set up for bubble shape characterization and prediction that will pro-
vide new ideas for future studies on bubble hydrodynamics.

Keywords: Bubble shape, Bubble dynamics, Gas-liquid system, Non-Newtonian fluid

Received: November 28, 2018; revised: July 01, 2019; accepted: July 16, 2019
DOI: 10.1002/ceat.201800690

1 Introduction begun to realize that, for fluids with high viscosity, no single
dimensionless number (Eo, We, or Ta number) is suitable to
Bubble-liquid systems are widespread in industrial areas such predict the bubble deformation. Therefore, more dimensionless
as oil extraction, wastewater treatment, and bioreactors. For numbers have been used and many empirical models have
these applications, the knowledge of bubble shape is of impor- been built to fit the results of the new experiments or numerical
tance since it is directly related to hydrodynamic behavior (e.g., simulations [25–28]. For instance, Liu et al. [29] suggested that,
terminal velocity, drag, surface area) and can influence the mix- in highly viscous fluids, the bubble shape is dramatically
ing level and the efficiency of energy, mass, and momentum affected by the viscosity and that the effects of We should be
transfer. considered along with those of Re. However, the aspect ratio
The observed shapes of individual bubbles in free motion in suffers from the shortcoming of being unable to characterize
Newtonian fluids have been well investigated and can be divid- the details of shape or irregular forms (e.g., a cap bubble and
ed mainly into three categories: spherical, ellipsoidal, and an ellipsoidal bubble would have the same aspect ratio). A few
spherical-cap or ellipsoidal-cap [1, 2]. Grace et al. [3] have pro- researchers have tried to use other parameters to characterize
posed a generalized graphical correlation or the so-called the bubble deformation. Myint et al. [30] introduced a distor-
‘‘Grace diagram’’, which delineates the shapes of bubbles in tion factor to quantify the asymmetry between the upper and
Newtonian fluids, providing a guideline for further studies on lower parts of the bubble divided by the major axis. Another
the bubble shape [4–6]. On the other hand, although numerous study, by Yan and Zhao [31], used the bubble circularity,
studies deal with bubbles rising in non-Newtonian media, only defined as the ratio of the circumference of an equivalent circle
a few of them have discussed and analyzed the bubble shape to that of the real bubble projection. Although some valuable
[7–12]. It is generally agreed that non-Newtonian fluids have conclusions have been drawn from the previous work, most of
complex rheological properties (e.g., shear thinning, viscoelas- the expressions mentioned above could only handle quasi-
ticity, etc.) leading to a variety of bubble shapes, such as tear- ellipsoidal bubbles. For a bubble with a more complicated con-
drop or elongated bubbles [13–18].
To quantitatively characterize the bubble shape, many
–
researchers used the aspect ratio or eccentricity, i.e., the ratio 1
between the lengths of the major and minor axes of the bubbles. Dr. Feishi Xu, Prof. Gilles Hébrard, Dr. Nicolas Dietrich
[email protected]
Pioneering work by Moore [19] and Tadaki and Maeda [20]
Université de Toulouse, CNRS, INRA, INSA, Laboratoire d’Ingénierie
theoretically correlated the aspect ratio of the bubble with the des Systèmes Biologiques et des Procédés (LISBP), 135 Avenue de
Weber number and the original Tadaki number (Ta = Re.Mo0.23), Rangueil, 31077 Toulouse, France.
respectively. Their correlation was modified and improved for 2
Prof. Noel Midoux, Prof. Huai-Zhi Li
small inertial effects [21] and became applicable to clean and Université de Lorraine, CNRS, Laboratoire Réactions et Génie des
contaminated bubbles [22–24] for a large range of Morton Procédés (LRGP), 1 rue Grandville, BP 20451, 54001 Nancy Cedex,
numbers (Mo) [25]. More recently, many researchers have France.

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2322

tour (e.g., cap, teardrop), it is clear that

neither the aspect ratio nor the circularity
can properly characterize the bubble shape.
Therefore, in this paper, a new model
using the parametric equations is proposed
to directly depict the complete profile of
the bubble. Experiments will be imple-
mented to study individual air bubbles of
different sizes (equivalent diameters:
0.7–7 mm) rising in stagnant non-Newto-
nian fluids (aqueous solution of polyacryl-
amide or xanthan gum) by means of high-
speed photography and image post-pro-
cessing technologies. The objective mainly
is to investigate the overall impact (includ-
ing the different dimensionless numbers
Figure 1. Experimental setup for bubble shape investigation.
and the rheological properties of the fluid)
on the bubble shape.
is found that the xanthan solutions have a much higher viscosi-
ty than the PAAm solutions, especially at low shear rates
2 Materials and Methods (< 10 s–1). Meanwhile, both of the two solutions present a simi-
lar flow behavior. Under low and high shear rates, the viscosity
2.1 Experimental Setup approaches a Newtonian plateau where the value of viscosity
tends to be constant. Between these two plateaus, the viscosity
The experimental setup for studying the bubble shapes is decreases with higher shear rate, which is known as the shear-
shown in Fig. 1. A column (volume: 100 · 100 · 300 mm3) thinning property. This evolution can be characterized by the
was filled with 2 L of the liquid under study. A single bubble classic Carreau model [33, 34]:
was generated by injecting air into the column through a
horizontal nozzle located 20 mm above the column bottom, m  m¥  n1
¼ 1 þ ðl_gÞ2 2 (1)
by a syringe pump (Harvard Apparatus, PHD 22/2000) with m0  m¥
a flow rate of 10 mL h–1. Stainless-steel nozzles of different
calibers ( » 0.5–1 mm) were used to make bubbles of differ- where m is the variable viscosity (Pa s), m0 is the viscosity at the
ent sizes (equivalent diameters: 0.7–7 mm). A high-speed zero shear rate (Pa s), m¥ is the viscosity at the infinite shear
camera (Photon SA3; resolution: 1024 · 1024 pixels) was rate (Pa s), g_ is the variable shear rate (s–1), l is the relaxation
placed next to one side of the column and focused on the time (s), and n is the power index.1)
rising bubble with a vertical image area ( » 40 ·40 mm2). The For each solution, under different concentrations, multiple
recording rate of the bubble images was 2000 frames per sec- samples were tested and fitted with the Carreau model. The fit-
ond. The image area was illuminated by a backlight board ting results are in agreement with the experimental ones with
placed at the opposite side of the column. The images ac- the determination coefficient r2 > 97 %, proving that the Car-
quired by the camera were transmitted to the computer and reau model is suitable for describing the viscosity variation of
processed with the professional software Photron FASTCAM the investigated non-Newtonian fluids in our cases. The aver-
Viewer (PFV). All the experiments were performed at aged results of the regression of the viscosity are shown in
293.15 K and under atmospheric pressure. Tab. 1 as well as other general physical properties. In this study,
the shear rate g_ near the bubble was estimated by g_ ¼ U=Deq
with the experimental data: bubble rising velocity U and equiv-
2.2 Materials and the Rheological Properties alent diameter Deq. This is the characteristic shear rate near the
equator of a spherical bubble. The effective shear rate may be
Two typical non-Newtonian solutions that are widely used in slightly different for large, non-spherical bubbles, but for sim-
the industry [32] were chosen for the experiments: poly(acryl- plicity we use the same definition in all cases [35]. The meth-
amide-co-acrylic acid) partial sodium salt (PAAm) (Sigma- ods for calculating U and Deq are presented in Sects. 2.3 and
Aldrich, CAS: 62649-23-4) and xanthan gum (Sigma-Aldrich, 2.4.1.
CAS: 11138-66-2). The solutions were prepared by dissolving
powder of substances in deionized water, under five concentra-
tions for PAAm (0.1, 0.25, 0.5, 1, 1.25 wt %) and three for xan-
than gum (0.1, 0.25, 0.5 wt %).
The rheological properties of both PAAm and xanthan solu-
tions were measured by a rheometer (HAKKE MARS III,
Germany) at 293.15 K, with the shear rates ranging from 0.001 –
to 10 000 s–1. The measured flow curves are plotted in Fig. 2. It 1) List of symbols at the end of the paper.

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2323

situate the major axis of the bubble

along the vertical direction. The
bubble profile (Fig. 3d) was finally
extracted by recognizing the edge
of the bubble spot (Fig. 3c).
To calculate the velocity of the
bubble, the centroid (xi, yi) of the
bubble spot in each image frame
(Fig. 3b) was recognized. Then the
distance between the centroids in
two successive frames was divided
by the time interval Dt (1/2000 s),
Figure 2. Flow curves (viscosity versus shear rate) for (a) PAAm solutions and (b) xanthan gum thus giving the rising velocity of
solutions. the bubble:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Table 1. Materials and their physical properties.a) ðxiþ1  xi Þ2 þ ðyiþ1  yi Þ2
U¼
Dt
–1 –3
Composition Concentration [wt %] s [mN m ] r [kg m ] m0 [Pa s] m¥[Pa s] l [s] n (2)
PAAM 0.1 69 998 0.109 0.003 5.91 0.38
For a given experimental condi-
0.25 67 999 0.115 0.004 6.77 0.40 tion, the averaged values of U were
calculated by considering 20 pairs
0.5 66 1000 0.120 0.005 8.19 0.42
of images, and the standard devia-
1 65 1003 0.146 0.007 9.18 0.42 tion is smaller than 1.25 %. Herein,
1.25 65 1004 0.159 0.008 11.86 0.43
it can be assumed that the velocity
on the image area has attained the
Xanthan 0.1 69 998 0.36 0.001 15.12 0.47 terminal velocity.
0.25 67 999 3.6 0.002 26.11 0.33 After the image processing, the
bubble profile is put into the Carte-
0.5 66 1001 69 0.004 59.26 0.20 sian coordinates, making the bot-
–1 tom of the bubble coincide with the
a) T = 293.15 K; the viscosity is valid for the shear rate ranging between 0.001 and 1000 s .
origin and the major axis of the
bubble along the ordinate (see
2.3 Image Processing Fig. 4a). The coordinates of each point of the bubble profile
were then detected to fit the parametric equations, which will
The raw images recorded from the experiment cannot be used be discussed in next section.
directly for the bubble shape optimization. An image process-
ing procedure was implemented in MATLAB (R2017b) to
extract the bubble profile from the raw images. The specific 2.4 Characterization of Bubble Shapes
steps are displayed in Fig. 3. Firstly, a reference image concern-
ing the background, which was taken before the passage of the 2.4.1 Equivalent Diameter of the Bubble
bubble, was subtracted from the raw image (Fig. 3a). The image
thus obtained was then transformed into a binary image As shown in Fig. 4a, the images obtained from the experiment
(Fig. 3b) on which the bubble was depicted by a black spot after were two-dimensional bubble profiles. To calculate the equiva-
the background noise had been removed. Since the bubble may lent diameter of the bubble, a reconstruction of the three-
slant while rising in the liquid, a rotation was implemented to dimensional (3D) bubble was implemented, considering that
the bubble shape was axisymmetric
with respect to the major axis of
the bubble profile. Then the 3D
bubble was divided into a set of
multiple small circular truncated
cones (Fig. 4b). The volume of each
cone equals [36]:
ph  2 
Vi ¼ R þ r2 þ Rr (3)
3

Figure 3. Schematic views of the image processing steps (example of a bubble rising in a where R and r are the radius of the
0.5 wt % xanthan solution). lower and upper cross-sections,

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2324

respectively, and h is the height of the cone. These three vari- right-hand side of the table (as shown below), i.e., the shapes
ables could be directly obtained from the bubble profiles with a cusp at the bottom. In order to extend these equations
(depicted in Fig. 4a). The total volume of the bubble is the sum to be used for more types of shapes like spherical, ellipsoidal,
of all the small cones, and the equivalent diameter could be cal- and cap bubbles, a modification was implemented on the
culated as below: expression of the parametric equation for the folium shape by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi adding extra variables to the equation. The appropriate expres-
P
3 6 N i¼1 Vi
sion of the parametric equation was determined through trial
Deq ¼ (4) and error. A new parametric model is proposed as below:
p
8
> 2xC t
< x ðt Þ ¼
1 þ ty2 t ˛ ½¥; þ¥ (5)
>
: y ðt Þ ¼ 
A

2 b
1 þ at

2xC
E¼ (6)
yA

where a, b are two parameters describing the bubble shape, E

is the aspect ratio, 2xC is the length of the minor axis of the
bubble profile depicted in Fig. 4a (m), yA is the length of the
major axis of the bubble profile depicted in Fig. 4a (m), and t is
a numerical variable ranging between negative and positive in-
finity.
By adjusting a and b, the modified parametric equation can
be used to describe the profile of different bubble shapes that
are encountered in non-Newtonian fluids. The validation of
Figure 4. Schematic diagram of the (a) bubble profile and the model is discussed in Sect. 3.1.
(b) the processing method for the equivalent diameter of the In a measurement, the lengths of the major axis yA and the
bubble. minor axis 2xC can be obtained directly from the bubble images
(Fig. 3c). On the other hand, an optimization should be per-
formed to determine the values of a and b. Supposing that the
2.4.2 Bubble Shape Equation edge of the bubble profile can be divided into N points, the
coordinates of point i (xi,exp, yi,exp) on the experimental image
As in the example shown in Fig. 4, one of the typical shapes in are definite. For a pair of arbitrary values (a, b) from the func-
non-Newtonian elastic fluid is the teardrop-like shape. It would tion of the ordinate in Eq. (5), it has:
be interesting if this kind of shape could be described by simple sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
parametric equations. On the geometry, there exist several 1   1
mathematical curves that are similar to the teardrop-like bub- ti ¼ yi;exp =yA b  1 (7)
a
ble shape, such as teardrop, simple folium, and piriform. Their
parametric equations are given in Tab. 2, with the correspond- By substituting ti back into the function of the abscissa in
ing profiles. Eq. (5), the calculated abscissa xi,cal can be obtained as:

Table 2. Approximate models of the profiles of a bubble or

2xc ti
xi;cal ¼ (8)
droplet: parametric equation and example. 1 þ ti2
(

Teardrop t n a=1 The values of the parameters a and b are then optimized by
x ¼ a sin t sin
2 n=3 minimizing the sum of the differences at all points between the
y ¼ a cos t
t∈[-¥,+¥] abscissa calculated with Eq. (8) and the one from the experi-
8 ment:
Simple fo- > at a=1
<x ¼
lium 1 þ t2 2 t∈[-¥,+¥] X
N
> a xi;cal  xi;exp
:y ¼ (9)
1 þ t22 i¼1
8
Piriform < a2 a=1
x¼ cos 3 t sin t b=1
The difference is then compared to the equivalent diameter,
: b
y ¼ a cos 2 t t∈[-¥,+¥] and the relative deviation of the fitting result is defined as fol-
low:
PN
xi;cal  xi;exp
i¼1
It should be noticed that the equations in the table can only err ¼ (10)
be used to characterize the specific shapes displayed on the NDeq

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2325

3 Results and Discussion

3.1 Validation of the Parametric Model

Firstly, before dealing with the result from the

experiment, the proposed parametric model is veri-
fied with respect to its capability of characterizing
the bubble shape. According to Eqs. (5) and (6),
the parametric equations for the bubble shape
involve three parameters: a, b, and the aspect ratio,
E. The evolution of the bubble shape as a function
of a and b is depicted in Fig. 5, in which the pro-
files of the bubbles are shown for aspect ratio E = 1
(Fig. 5a) and E = 0.5 (Fig. 5b). It could be observed Figure 5. Evolution of the shape versus the parameters a and b; (a) E = 1,
that the value E controls the global shape in that (b) E = 0.5.
the profile becomes more oblate with smaller value
E (e.g., spherical bubble to ellipsoidal bubble). On
the other hand, the parameters a and b control the
detail of the shape. The value of a can change the
position of the minor axis (horizontal) of the bub-
ble profiles. When the value of b is fixed (e.g. b = 1)
and a increases from 0.5 to 2.5, the minor axis is
moving from near the top of the bubble to the bot-
tom of the bubble. On the other hand, b is related
to the appearance of the cusp at the bottom of the
bubble. Herein, the bubble deformation can be
roughly summarized as follows:
– For b < 0.6, there is a clear cusp at the bottom of
the bubble. The typical inverted teardrop
shape appears at a high value of a (e.g., Fig. 5a:
(a, b) = (2.5, 0.4)).
– For 0.6 < b < 0.8, the bubble becomes progres-
sively rounded at the bottom and the cusp grad-
ually disappears as the value of b increases.
– For 0.8 < b < 1, there is no longer a cusp and the
profile becomes more and more rounded. The
quasi-ellipsoidal bubble appears at this interval. Figure 6. Shape regime map (Grace map) for bubbles in PAAm and xanthan
In particular, the circular profile (spherical bub- gum solutions (dark gray curves: log Mo; gray patterns in the background: typi-
ble) and the perfectly elliptical profile (ellipsoidal cal bubble shape near the corresponding experimental points).
bubble) appear when the values a and b are both
equal to 1.
– For b > 1, the bottom of the profile tends to be flat. The typi- and the typical bubble shapes (gray patterns in the back-
cal cap shape appears when the values of a and b are both ground) extracted from the raw images (Fig. 3c) are also
sufficiently large (e.g., Fig. 5b: a = 1.5, b = 1.5). depicted near the corresponding experimental points.
Therefore, by adjusting the values of parameters a, b, and E,
the majority of the bubble shapes appearing in non-Newtonian rgD2eq
fluids can be characterized precisely. Based on the parametric Eo ¼ (11)
s
equations, the investigation of the bubble shape can be carried
out by analyzing the evolution of these parameters. rUDeq
Re ¼ (12)
m

3.2 Bubble Shape Regimes gm4

Mo ¼ (13)
rs3
The resulting bubble shapes in different non-Newtonian solu-
tions (i.e., both PAAm and xanthan gum solutions) are shown In PAAm solutions, the range of the Mo number is relatively
as bubble shape regime map, the so-called Grace map (Fig. 6). small (–9 <logMo <–7) within the concentration range of the
This bubble shape regime map is plotted in terms of the study (0.1–1.25 wt %). The data covers the spherical bubble and
dimensionless Re and Eo numbers and log Mo (dark gray the ellipsoidal bubble regimes. When increasing Eo and Re, the
curves), which can be calculated as defined in Eqs. (11)–(13), bubble becomes bigger and the shape changes from spherical

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2326

to ellipsoidal. Regarding the concentration of the solution, it around the bubble due to the shear-thinning property. On the
can be seen that Re increases in a dilute solution, making the other hand, when a bubble is rising in a concentrated PAAm
bubble shape become more oblate and even causing wobbling solution (1–1.25 wt %), the surface oscillation no longer exists
due to the surface oscillation. due to the higher viscosity. The bubble thus encounters more
In xanthan solutions, the range of the Mo number is larger resistance and its velocity is much slower. Moreover, there is
(–9 <log Mo <3) under the investigated concentration no evident velocity discontinuity under our experimental con-
(0.1–0.5 wt %). The data covers not only the spherical and ellip- ditions, indicating that the bubble terminal velocity could be
soidal bubbles but also the prolate and teardrop ones. A similar less influenced by the bubble shape deformation (from ellipsoi-
tendency is found for the bubble deformation as in the PAAm dal to inverted teardrop). This result is consistent with the
solutions. In particular, the cusp appears for high Eo numbers literature [38], in which the same behavior of rising bubbles
(Eo > 1) and becomes more significant with the increase of the was found in carboxymethyl cellulose (CMC) solutions and
solution concentration or for lower Re numbers. Under this xanthan gum solutions.
condition, the bubble is also elongated vertically, which is
uncommon in Newtonian fluids.
Compared with the Grace map given by Clift et al. [37] for 3.4 Bubble Shape Parameters
bubble shapes in Newtonian fluids, the regimes of the spherical
bubble and the ellipsoidal bubble in the present study are con- To better analyze the bubble shape, the bubble profile was fitted
sistent when Re > 1. However, the bubble shapes are invariably by the parametric models introduced in Sect. 2.4.2. It should be
spherical when Re < 1, in contrast to those in non-Newtonian noticed that the bubble shape is assumed to be stable without
fluids. In fact, in non-Newtonian fluids, as the bubble velocity surface oscillation. The data for a big bubble (Deq > 6 mm) ris-
could be extremely low due to the complex rheological proper- ing in a dilute PAAm solution (0.1–0.5 wt %) was eliminated
ties of the solution, relatively smaller values of Re should be due to the non-negligible surface oscillation. Two typical fitting
considered, namely Re ranging from 0.001 to 1 in this study. results, chosen for single bubbles in xanthan gum and PAAm
Within this range, the prolate bubble and the teardrop bubble solutions, are shown in Fig. 8a and b, where the black dots
can also be visualized while the spherical bubble appears only depict the origin profile of the bubble and the red line is the
for low Eo numbers (Eo < 1). fitting profile, with the values of a and b given alongside. It can
be seen that, regardless of whether the bubble has a teardrop or
an ellipsoidal shape, the deviation between the experimental
3.3 Bubble Velocity points and the fitting points is small. The deviation defined in
Eq. (10) is of a magnitude smaller than 3 % for all the cases. To
Concerning the bubble terminal velocity, as shown in Fig. 7, for ensure the validity of the bubble characterization result, fitting
both types of the solutions, it first increases with the bubble data with an error of more than 2 % was eliminated. An exam-
size, then tends to be constant. Although the velocity curves at ple of the averaged fitting results of the bubble shapes in both
different concentrations differ more from each other for xan- PAAm and xanthan solutions under different concentrations
than solutions, this evolution and the final constant velocity do are shown in Fig. 8c and d. It can be observed that under the
not depend on the type of solution or the concentration of the same operational conditions for bubble generation (i.e., the
solute. It could be derived that, in a highly viscous xanthan nozzle size and the air flow rate), the sizes and shapes of the
solution, the bubble rises much more slowly and needs to attain bubbles depend on the type and concentration of the solutes
the final constant velocity at a bigger size. In particular, there (i.e. PAAm and xanthan gum). In addition, the evolution of the
exists a remarkable difference between the curves of the three bubble shapes given by the model is consistent with the origi-
lower concentrations of PAAm solutions and those of the two nal bubble shapes depicted in Fig. 6.
higher concentrations. The reason is that, in the dilute PAAm It can be seen from Fig. 8c and d that, in PAAm solution, the
solutions (0.1–0.5 wt %), the bubble shape is less stable with the size of the bubble slightly changes while the shape becomes
surface oscillation, which leads to a reduction of the viscosity more and more rounded for higher concentrations, namely,
from an oblate bubble to a general
ellipsoidal bubble. In contrast, in
xanthan solutions, the major axis
(vertical) of the bubble increases as
the concentration of xanthan in-
creases, and the bubble deforms
from an oblate ellipsoid to an
oblate teardrop and then turns to
an elongated teardrop.
For a single bubble, a pair of
parameters (a, b) could be obtained
from the fitting result. In Fig. 9, the
relationship between the bubble
Figure 7. Relationship between the bubble rising velocity and the bubble size (Deq = 0.7–7 mm) shape parameters (a and b) is
in (a) PAAm and (b) xanthan gum solutions. depicted for bubbles in both PAAm

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2327

The result of the xanthan gum cases

shows good agreement with this
evolution.
It is found that the relationship
between a and b could be corre-
lated by a power-law function,
which is expressed as below:

a0:85 b ¼ 1 (14)

In the range investigated in this

study (–9 <log Mo <3), the accuracy
of this correlation is acceptable,
with a determination coefficient
r2 > 95 %. It can be seen from Fig. 9
that the curve of Eq. (14) passes the
point (a, b) = (1, 1), which indicates
a perfectly spherical bubble (or an
ellipsoidal bubble when E ≠ 1).
Thanks to Eq. (14), it is possible to
characterize the bubble shape with
only one parameter (either a or b).
In the next section, only the
parameter b is taken into account.

3.5 Prediction of Bubble

Figure 8. Examples of the bubble profiles and the fitting ones by Eqs. (5) and (6) in (a, c) PAAm Shape
and (b, d) xanthan solutions under different concentrations. The bubbles are injected by the
nozzle with a caliber of 1 mm at a flow rate of 10 mL h–1.
It is known from Sect. 3.1 that the
aspect ratio E roughly describes the
bubble shape while the parameters a and b characterize the
details of bubble deformation. As mentioned in the introduc-
tion that a lot of research has been carried out for predicting
the aspect ratio E for bubbles in different liquids, we will focus
on the discussion of the bubble shape parameter b.
Depending on the value of b, the bubble shape changes from
ellipsoidal (without cusp) to teardrop (with cusp). Some exist-
ing research attempts to explain the appearance of the cusp
theoretically and most of the authors support the reason that
the majority of non-Newtonian fluids are composed of long-
chain molecules and possess viscoelasticity [39, 40]. To quantify
the elasticity of the fluid, the Deborah number is utilized to
compare the relaxation time of the fluid to the time scale of
observation [41]:
Figure 9. Relationship between the bubble shape parameters
(a and b) in both the PAAm and xanthan solutions under differ- lU
De ¼ ¼ l_g (15)
ent concentrations. Deq

and xanthan gum solutions. The corresponding bubble profiles In our study, the observation time is estimated as equal to
(light gray outline) are also displayed at the positions of values a the characteristic time of the motion (i.e., the equivalent diam-
and b. In this figure, the delicate changes in the bubble shape can eter of the bubble divided by the bubble rising velocity). Since
be visualized and quantified, which is difficult to achieve simply this characteristic time of the motion is relatively short com-
by using the aspect ratio. For PAAm solutions, the data are con- pared with the relaxation time of the fluid, the magnitude of
centrated in the range of 1 < a < 1.2 and 0.8 < b < 1, suggesting the calculated De number in this study may be higher than in
that the bubble is spherical or ellipsoidal, while for the xanthan the literature [42]. Nonetheless, it can still be used to compare
solution, the range of these two parameters is much larger. As the elastic levels for the investigated fluids.
the parameter a increases, the value of b decreases gradually. As Since the cusp could now be quantified by the parameter b,
interpreted in Sect. 2.4.2, the cusp of the bubble appears when the evolution of b is analyzed in terms of the bubble size (V)
b < 0.6 and it becomes sharper when b continues to decrease.

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2328

and various dimensionless numbers (De, Eo, Re), as shown in index n (Eq. 1). After testing several functions, the following
Fig. 10. correlation was proposed to fit the experimental data:
As can be obtained from Fig. 10: (1) For larger bubbles, the
velocity of the bubble increases; then the bubble begins to b ¼ n  0:11De0:19 Eo0:11 þ 0:80 (16)
deform and becomes less rounded when the contribution of
inertial force and buoyancy is stronger than that of the surface This equation is correlated to the result depicted in Fig. 10 as
tension and viscous force. (2) b decreases with higher elasticity b decreases monotonically with high De and Eo numbers and
of the solution in most of the investigated range, except for the shear-thinning levels. The comparison between the param-
some PAAm solutions, which are relatively less elastic. If the eter b predicted by Eq. (16) and the experimental ones is shown
bubble moves too fast, the liquid cannot fill the space in the in Fig. 11. It can be seen that, within the range of this study
rear of the bubble immediately. Hence this space is still occu- (–9 <logMo <3), the prediction quality of parameter b in both
pied by the gas and pressed by the long-chain liquid molecule PAAm and xanthan gum solutions is satisfactory. Relatively
from the side, leading to a little cusp forming in the rear of the large deviations appear at high values of b due to the wobbling
bubble. (3) In the evolution of b – De and b – Re, the evolution of the bubble rising in dilute PAAm solutions. The fitting result
of the value b under three concentrations is quite dispersed. has a coefficient of determination (r2) higher than 95 %.
This result might be caused by the different shear-thinning lev-
els of the solutions as the shear-thinning property is also a rea-
son for the deformation of the bubble [43]. 4 Conclusions
The data in Fig. 10 is too scattered for a general law to be
obtained. Hence, multiple influencing factors are considered Various single air bubbles rising in stagnant non-Newtonian
instead of one. By analogy with the analysis of the aspect ratio, fluids (polyacrylamide, xanthan gum) were investigated by
the investigation was carried out for the evolution of b with dif- high-speed photography. The shapes of the bubbles from
ferent pairs of dimensionless numbers as well as the shear-thin- spherical to inverted teardrop can be characterized with good
ning property, which can be roughly quantified by the power accuracy by parametric equations that contain the aspect ratio

Figure 10. The evolution of parameter b in terms of the bubble size (V) and various dimensionless numbers (De, Eo, Re) in PAAm
and xanthan solutions at different concentrations.

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2329

t [–] numerical variable

Ta [–] Tadaki number
Ux [mm s–1] oscillating velocity
Uy [mm s–1] rising velocity
V [mm3] bubble volume
We [–] Weber number
xi,cal [mm] abscissa by calculation
xi,exp [mm] abscissa by experiment
xC [mm] half-length of the major axis
yA [mm] length of the minor axis

Greek symbols
a [–] bubble shape parameter
b [–] bubble shape parameter
g_ [s–1] shear rate
l [s] relaxation time
m [Pa s] dynamic viscosity
r [kg m–3] density
s [N m–1] surface tension
Figure 11. Comparison result between the parameter b from
correlation (Eq. 16) and b from the measurements in PAAm and
xanthan gum solutions. References

(E) and two shape parameters (a, b). Based on this method, the [1] R. P. Chhabra, Bubbles, Drops, and Particles in Non-
various impacts on the bubble shape can be quantified by Newtonian Fluids, 2nd ed., CRC Press, Boca Raton 2006.
studying the evolution of the three parameters as a function of [2] F. Xu, A. Cockx, G. Hébrard, N. Dietrich, Mass transfer and
the bubble size, the bubble velocity, multiple dimensionless diffusion of a single bubble rising in polymer solutions, Ind.
numbers (Re, Eo, etc.), and the property of the fluids. Within Eng. Chem. Res. 2018, 57 (44), 15181–15194. DOI: https://
the range investigated here (–9 <logMo <–7), it is found that doi.org/10.1021/acs.iecr.8b03617
the cusp of the bubble (quantified by b) is related to the visco- [3] J. Grace, T. Wairegi, T. H. Nguyen, Shapes and velocities of
elasticity as well as the shear-thinning level of the solutions. A single drops and bubbles moving freely through immiscible
simple correlation is then proposed to predict the parameter b. liquids, Trans. Inst. Chem. Eng. 1976, 54, 167–173.
[4] S. Anwar, Lattice Boltzmann modeling of buoyant rise of
Since the parametric equations are able to characterize most of
single and multiple bubbles, Comput. Fluids 2013, 88, 430–
the observed shapes (including cap bubbles), this method can
439. DOI: https://doi.org/10.1016/j.compfluid.2013.09.015
be applied to bigger bubbles or droplets in complex Newtonian
[5] J. Hua, J. F. Stene, P. Lin, Numerical simulation of 3D bub-
and non-Newtonian fluids.
bles rising in viscous liquids using a front tracking method,
J. Comput. Phys. 2008, 227, 3358–3382. DOI: https://doi.org/
Acknowledgments 10.1016/j.jcp.2007.12.002
[6] Z. Yu, H. Yang, L.-S. Fan, Numerical simulation of bubble
The financial assistance provided for F. Xu by the China interactions using an adaptive lattice Boltzmann method,
Scholarship Council is gratefully acknowledged. The federation Chem. Eng. Sci. 2011, 66, 3441–3451. DOI: https://doi.org/
FERMAT is also thanked for its leading-edge material support. 10.1016/j.ces.2011.01.019
[7] L. Böhm, T. Kurita, K. Kimura, M. Kraume, Rising behav-
The authors have declared no conflict of interest. iour of single bubbles in narrow rectangular channels in
Newtonian and non-Newtonian liquids, Int. J. Multiphase
Flow 2014, 65, 11–23. DOI: https://doi.org/10.1016/
Symbols used j.ijmultiphaseflow.2014.05.001
[8] X. Frank, H. Z. Li, D. Funfschilling, F. Burdin, Y. Ma, Bubble
Deq [mm] equivalent diameter motion in non-Newtonian fluids and suspensions, Can.
E [–] aspect ratio J. Chem. Eng. 2003, 81, 483–490. DOI: https://doi.org/
Eo [–] Eötvös number 10.1002/cjce.5450810321
g [m s–2] gravitational acceleration [9] X. Frank, J.-C. Charpentier, Y. Ma, N. Midoux, H. Z. Li,
Mo [–] Morton number A multiscale approach for modeling bubbles rising in non-
n [–] power index/shear-thinning level Newtonian fluids, Ind. Eng. Chem. Res. 2012, 51, 2084–2093.
P [mm] perimeter DOI: https://doi.org/10.1021/ie2006577
r2 [–] determination coefficient [10] N. M. S. Hassan, M. M. K. Khan, D. W. Rackemann, An
Re [–] Reynolds number experimental study of bubble rise characteristics in non-
S [mm2] surface area Newtonian (power-law) fluids, 16th Australasian Fluid

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201800690 by UAEH - Universidad Autonoma del Estado de Hidalgo, Wiley Online Library on [06/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 2330

Mechanics Conference (AFMC), Gold Coast 2007, http:// [26] K. Sugihara, T. Sanada, M. Shirota, M. Watanabe, Behavior
acquire.cqu.edu.au:8080/vital/access/manager/Repository/ of single rising bubbles in superpurified water, Kagaku
cqu:2807 (Accessed January 31, 2017). Kogaku Ronbunshu 2007, 33, 402–408.
[11] N. M. S. Hassan, M. M. K. Khan, M. G. Rasul, D. W. Racke- [27] D. Legendre, R. Zenit, J. R. Velezcordero, On the deforma-
mann, Bubble rise velocity and trajectory in xanthan gum tion of gas bubbles in liquids, Phys. Fluids 2012, 24, 043303.
crystal suspension, Appl. Rheol. 2010, 20, 65102. [28] G. Kelbaliyev, K. Ceylan, Development of new empirical
[12] L. Zhang, C. Yang, Z.-S. Mao, Numerical simulation of a equations for estimation of drag coefficient, shape deforma-
bubble rising in shear-thinning fluids, J. Non-Newton. Fluid tion, and rising velocity of gas bubbles or liquid drops,
Mech. 2010, 165, 555–567. DOI: https://doi.org/10.1016/ Chem. Eng. Commun. 2007, 194, 1623–1637. DOI: https://
j.jnnfm.2010.02.012 doi.org/10.1080/00986440701446128
[13] D. Dekée, R. P. Chhabra, A photographic study of shapes of [29] L. Liu, H. Yan, G. Zhao, Experimental studies on the shape
bubbles and coalescence in non-Newtonian polymer solu- and motion of air bubbles in viscous liquids, Exp. Therm.
tions, Rheol. Acta 1988, 27, 656–660. DOI: https://doi.org/ Fluid Sci. 2015, 62, 109–121. DOI: https://doi.org/10.1016/
10.1007/BF01337462 j.expthermflusci.2014.11.018
[14] D. Dekée, P. J. Carreau, J. Mordarski, Bubble velocity and [30] W. Myint, S. Hosokawa, A. Tomiyama, Shapes of single
coalescence in viscoelastic liquids, Chem. Eng. Sci. 1986, 41, drops rising through stagnant liquids, J. Fluid Sci. Technol.
2273–2283. DOI: https://doi.org/10.1016/ 2007, 2, 184–195. DOI: https://doi.org/10.1299/jfst.2.184
0009-2509(86)85078-3 [31] H. Yan, J. Zhao, Experimental study on shape and rising
[15] D. Funfschilling, H. Z. Li, Effects of the injection period on behavior of single bubble in stagnant water, J. Cent. South
the rise velocity and shape of a bubble in a non-Newtonian Univ. Sci. Technol. 2016, 47, 2513–2520.
fluid, Chem. Eng. Res. Des. 2006, 84, 875–883. DOI: https:// [32] R. L. Davidson, Handbook of Water-Soluble Gums and
doi.org/10.1205/cherd.01229 Resins, McGraw-Hill, New York 1980.
[16] N. Dietrich, G. Hebrard, Visualisation of gas-liquid mass [33] R. B. Bird, P. J. Carreau, A nonlinear viscoelastic model for
transfer around a rising bubble in a quiescent liquid using polymer solutions and melts–I, Chem. Eng. Sci. 1968, 23,
an oxygen sensitive dye, Heat Mass Transfer 2018, 54 (7), 427–434. DOI: https://doi.org/10.1016/0009-2509(68)87018-6
2163–2171. DOI: https://doi.org/10.1007/s00231-018-2297-3 [34] P. J. Carreau, I. F. MacDonald, R. B. Bird, A nonlinear visco-
[17] N. Dietrich, K. Wongwailikhit, M. Mei, F. Xu, F. Felis, A. elastic model for polymer solutions and melts–II, Chem.
Kherbeche, G. Hébrard, K. Loubière, Using the ‘‘Red Bottle’’ Eng. Sci. 1968, 23, 901–911. DOI: https://doi.org/10.1016/
experiment for the visualization and the fast characterization 0009-2509(68)80024-7
of gas-liquid mass transfer, J. Chem. Educ. 2019, 96 (5), [35] N. Dietrich, Étude locale et expérimentale des phénomènes
979–984. DOI: https://doi.org/10.1021/acs.jchemed.8b00898 interfaciaux, Ph.D. Thesis, Vandoeuvre-les-Nancy, INPL,
[18] A. Kherbeche, M. Mei, M.-J. Thoraval, G. Hébrard, 2008. http://www.theses.fr/2008INPL078N (Accessed on
N. Dietrich, Hydrodynamics and gas-liquid mass transfer September 20, 2016).
around a confined sliding bubble, Chem. Eng. J. 2019, in press. [36] J. W. Harris, H. Stöcker, Handbook of Mathematics and
DOI: https://doi.org/10.1016/j.cej.2019.04.041 Computational Science, Springer, New York 1998.
[19] D. W. Moore, The rise of a gas bubble in a viscous liquid, [37] R. Clift, J. R. Grace, M. E. Weber, Bubbles, Drops, and
J. Fluid Mech. 1959, 6, 113–130. DOI: https://doi.org/ Particles, Dover Publications, New York 1978.
10.1017/S0022112059000520 [38] S. Amirnia, J. R. de Bruyn, M. A. Bergougnou, A. Margaritis,
[20] T. Tadaki, S. Maeda, On the shape and velocity of single air Continuous rise velocity of air bubbles in non-Newtonian
bubbles rising in various liquids, Chem. Eng. 1961, 25, biopolymer solutions, Chem. Eng. Sci. 2013, 94, 60–68. DOI:
254–264. DOI: https://doi.org/10.1252/ https://doi.org/10.1016/j.ces.2013.02.032
kakoronbunshu1953.25.254 [39] S. M. Barnett, A. E. Humphrey, M. Litt, Bubble motion and
[21] T. D. Taylor, A. Acrivos, On the deformation and drag of a mass transfer in non-Newtonian fluids, AIChE J. 1966, 12,
falling viscous drop at low Reynolds number, J. Fluid Mech. 253–259.
1964, 18, 466–476. DOI: https://doi.org/10.1017/ [40] R. Zenit, J. J. Feng, Hydrodynamic interactions among bub-
S0022112064000349 bles, drops, and particles in non-Newtonian liquids, Annu.
[22] R. M. Wellek, A. K. Agrawal, A. H. P. Skelland, Shape of Rev. Fluid Mech. 2018, 50, 505–534. DOI: https://doi.org/
liquid drops moving in liquid media, AIChE J. 1966, 12, 10.1146/annurev-fluid-122316-045114
854–862. DOI: https://doi.org/10.1002/aic.690120506 [41] M. Reiner, The Deborah number, Phys. Today 1964, 17, 62.
[23] I. A. Vakhrushev, G. I. Efremov, Interpolation formula for DOI: https://doi.org/10.1063/1.3051374
computing the velocities of single gas bubbles in liquids, [42] R. P. Chhabra, Non-Newtonian fluids: An introduction, in:
Chem. Technol. Fuels Oils 1970, 6, 376–379. DOI: https:// Rheology of Complex Fluids (Eds: J. M. Krishnan, A. P. Desh-
doi.org/10.1007%2fBF01171684 pande, P. B. S. Kumar), Springer Science & Business Media,
[24] L.-S. Fan, K. Tsuchiya, Bubble Wake Dynamics in Liquids New York 2010, 3–34. https://link.springer.com/chapter/
and Liquid-Solid Suspensions, Butterworth-Heinemann, 10.1007/978-1-4419-6494-6_1
Boston, MA 1990. [43] M. Warshay, E. Bogusz, M. Johnson, R. C. Kintner, Ultimate
[25] F. Raymond, J.-M. Rosant, A numerical and experimental velocity of drops in stationary liquid media, Can. J. Chem.
study of the terminal velocity and shape of bubbles in Eng. 1959, 37, 29–36. DOI: https://doi.org/10.1002/
viscous liquids, Chem. Eng. Sci. 2000, 55, 943–955. DOI: cjce.5450370107
https://doi.org/10.1016/S0009-2509(99)00385-1

Chem. Eng. Technol. 2019, 42, No. 11, 2321–2330 ª 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com