The ejection of droplets from a bursting bubble on a liquid surface—A

May 8, 2018 - Furthermore, we propose a new dimensionless parameter to describe the region of properties where 'jet' drops will occur. This parameter ...
0 downloads 0 Views 805KB Size
Subscriber access provided by University of Winnipeg Library

New Concepts at the Interface: Novel Viewpoints and Interpretations, Theory and Computations

The ejection of droplets from a bursting bubble on a liquid surface—A dimensionless criterion for ‘jet’ droplets Ryan Mead-Hunter, Monica M. Gumulya, Andrew (A.) King, and Benjamin J Mullins Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.8b00664 • Publication Date (Web): 08 May 2018 Downloaded from http://pubs.acs.org on May 9, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

The ejection of droplets from a bursting bubble on a

2

free liquid surface—A dimensionless criterion for

3

‘jet’ droplets

4

Ryan Mead-Hunter*,†,‡, Monica M. Gumulya⁑, Andrew J.C. King‡,⁑ and Benjamin J. Mullins†,‡

5



6

6845, Australia

7



8



9

University, Perth WA, 6845, Australia

10

Occupation, Environment & Safety, School of Public Health, Curtin University, Perth WA,

Fluid Dynamics Research Group, Curtin University, Perth WA, 6845, Australia Western Australian School of Mines: Minerals, Energy and Chemical Engineering, Curtin



School of Civil and Mechanical Engineering, Curtin University, Perth WA, 6845, Australia

11 12

KEYWORDS

13

Jet Droplets, Bubble Bursting, Computational Fluid Dynamics

14

ABSTRACT

15

This work examines the ejection of droplets from a bursting gas bubble on a free liquid surface,

16

both experimentally and numerically. We explore the physical processes which govern the

17

bursting of bubbles and the subsequent formation of ‘jet’ droplets. We present new relationships

ACS Paragon Plus Environment

1

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 28

1

regarding the dependence of jet drop formation on bubble diameter. Furthermore, we propose a

2

new dimensionless parameter to describe the region of properties where ‘jet’ drops will occur.

3

This parameter, termed the droplet number (Dn), complements existing parameters defining jet

4

drop formation—namely a maximum Ohnesorge number and a maximum Bond number.

5 6 7 8

INTRODUCTION

9

The generation of droplets from bursting bubbles is a commonly known and well-studied

10

phenomenon1-3. This process gives rise to a significant portion of atmospheric aerosols4-7 and has

11

been described as an important process in the enjoyment of a glass of sparkling wine1. Despite

12

being well studied, there are still a number of questions surrounding the role of liquid properties

13

and their contribution to the mechanisms of bubble bursting.

14

Two distinct mechanisms of droplet formation via bubble bursting exist2,3. The first is when a

15

bubble collapses at the liquid surface, giving rise to a liquid column, which breaks into droplets

16

which are ejected vertically with significant momentum—these are termed ‘jet’ droplets (and are

17

shown in Figure 1). The other case is where a liquid film surrounding the bubble, breaks apart.

18

Giving rise to ‘film’ droplets.

19

The formation of ‘jet’ droplets only occurs over a specific range of droplet sizes for a given

20

liquid. Bubbles bursting in water for example will only produce ‘jet’ droplets at diameter of 4.2

21

mm or less8. The relevant properties of the liquid are its viscosity and surface tension, which

22

dictate the nature of the bubble collapse. The range of properties for which ‘jet’ droplets are

ACS Paragon Plus Environment

2

Page 3 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

produced have been described in terms of the dimensionless Ohnesorge number (Oh) and Bond

2

number (Bo). Systems with Oh ≲ 0.037 and Bo ≲ 3, have typically been said to produce ‘jet’

3

droplets8.

4

In recent work, Walls et al.8 identified an additional zone, where systems that would be

5

expected to produce ‘jet’ droplets, based on values of Oh and Bo, do not. This was explored

6

through the use of computational fluid dynamics (CFD) and ascribed to the combined effects of

7

gravity and viscosity.

8 9

The influence of viscosity was also highlighted by Ghabache et al.1 who conducted experiments using 150 to 1000 µm radius droplets, of varying concentrations of glycerol in

10

water. In their work they considered the role of the bubble bursting process in the control of the

11

production of droplets, and noted two regimes. A surface tension dominated regime and a

12

viscosity dominated regime, with the transition between the two occurring at a Morton number

13

(Mo) (Eq. 3) of approximately 3 x 10-8. The process of bubble bursting and the production of

14

‘jet’ droplets may therefore be seen as a function of surface tension, viscosity, gravity and bubble

15

size.

16

In addition to Oh, Bo, and Mo, the Weber number (We) has also been used to describe the

17

dynamics of bubble bursting. Ghabache et al.1 use this value to determine a dimensionless jet

18

velocity term (given as WeBo1/2), which clearly demonstrates the surface tension and viscosity

19

driven regimes, as evidenced in Figure 4. The value of the dimensionless terms are defined as

20

follows,

21

We =

ρVtip2 r γ

(1)

Bo =

ρgr 2 γ

(2)

Mo =

gµ 4

ργ

3

(3)

Oh =

µ (4) ργr

ACS Paragon Plus Environment

3

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 28

1 2

where, ρ is the liquid density, r is the bubble radius, γ the liquid surface tension, g acceleration

3

due to gravity, µ the liquid viscosity and Vtip is the tip velocity. The value of tip velocity, as used

4

by Ghabache et al.1 and in this work, is the velocity of the liquid column at the point where it

5

reaches the surface of the bulk liquid.

6 7

Figure 1. Images of the bubble collapse and ejection of droplets in salt water. The 4 frames on

8

the left show the beginning of the collapse of the bubble after it has reached the surface. Moving

9

right we can see the cavity beginning to collapse as it is drawn in from the base as the jet begins

10

to form. The final two frames, are from a camera orientation an and show the ejection of a

11

droplet from the rising liquid column.

12

While the motivations for studying bubble-bursting are many and varied, one application is to

13

the (re-)entrainment of liquid aerosol from coalescing filters11. In previous work9 we have

14

observed this (re-)entrainment and have proposed two possible mechanisms. These have been

15

further explored by Hotz et al.10 who measured the force required to detach a droplet from a

16

fibre, and Wurster et al.11 who identified that the bursting of bubbles was (in most cases) the

17

detachment mechanism for filter re-entrainment.

ACS Paragon Plus Environment

4

Page 5 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

When utilising coalescing media to treat liquid aerosols, the captured liquid has the ability to

2

coalesce and flow within the media. As the filter becomes saturated the liquid moves through the

3

filter to the rear face, where it can drain from the filter. This process is accompanied by the

4

formation of a liquid film on the rear face, though which air must pass. In cases where the

5

saturation level of the filter is high, the air may need to break through the rear liquid film, as

6

process which results in the formation of bubbles, which burst releasing droplets.

7

In this work we simulate the bubble bursting process and the subsequent formation of ‘jet’

8

droplets using experimental data generated as part of this work, and data from Ghabache et al.1

9

for validation. If the key physical processes can be reliably captured then the simulation methods

10

used can be more broadly applied to our coalescing filter case—an application where

11

experiments are considerably more difficult. Through our simulations we will also consider

12

parameters for ‘jet’ droplet formation and attempt to define a parameter to identify the region

13

identified by Walls et al.8 where ‘jet’ droplets do not occur, despite meeting Oh and Bo criteria.

14

The volume-of-fluid (VoF) method has previously been used to simulate bubble motion12,13

15

and to accurately resolve liquid interfaces on small scales14 and will be applied here.

16 17

METHODOLOGY

18

Experimental

19

Experiments were conducted in a clear, liquid filled cylinder with bubbles introduced by

20

passing compressed air through a fine nozzle (of varying size) in the base of the cylinder. The

21

bubble rate was controlled using a multi-stage pressure regulator and fine control achieved using

ACS Paragon Plus Environment

5

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

a manually adjustable needle valve. Bubbles were allowed to rise to the liquid surface and the

2

bursting process was observed with the aid of a high speed camera (Photron FASTCAM SA 3

3

Model 120K-M2) fitted with a macro lens (Nikon AF-S VR Micro-Nikkor 105 mm f/2.8G IF-

4

ED). Only bubbles that clearly produced, or clearly did not produce jet droplets were chosen for

5

further analysis. Bubbles that did not produce jet drops, were measured only for size, bubbles

6

that did produce jet drops were measured for size and tip velocity.

7

Page 6 of 28

Experiments were conducted in (distilled) water, salt water (30 g NaCl/l in distilled H2O),

8

ethanol (99.5%), and 40 and 50% propylene glycol-water mixtures.

9

Results from Ghabache et al.1 were extracted using Plot Digitizer

10

(plotdigitizer.sourceforge.net). These data were produced in a manner similar to that described

11

above, however using glycerol-water mixtures.

12

Model

13

Simulations were conducted using the OpenFOAM computational fluid dynamics (CFD)

14

software package (OpenFOAM Version 2.1.x, ESI Group). We utilised the volume-of-fluid

15

(VoF) method to resolve the liquid-air interface, coupled with dynamic mesh refinement applied

16

to all cells at the liquid-air interface (defined as cells with a liquid volume fraction above 0.999

17

or below 0.001).

18

For the purposes of simulation, we prepared a rectangular domain, partially filled with liquid

19

and introduced a bubble of air below the liquid surface. The bubbles were allowed to rise due to

20

buoyancy forces, i.e. the bubble had an initial velocity of 0. The simulations were run for

21

sufficient time to allow the bubble to rise to the surface, collapse and eject a column of liquid,

ACS Paragon Plus Environment

6

Page 7 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

potentially producing ‘jet’ droplets. Figure 3 illustrates the latter part of this process where

2

several droplets have been ejected from the collapsing liquid column. As this process is governed

3

by the fluid properties, simulations with different liquids were run for different times.

4

Simulations were considered complete, when the rise of the ejected ‘jet’ droplets stopped (i.e.

5

they began to fall back towards the liquid surface).

6

In order to ensure that domain size was not a factor in the rise of the bubble or its behaviour at

7

the interface, simulations were conducted to assess domain size and level of mesh refinement

8

required. This was critical to ensure that the break-up of the rising liquid column that forms after

9

bubble collapse was resolved.

10

A domain width of 10r was found to be sufficient, with lower values resulting in unrealistic

11

velocity profiles at the interface, possibly due to wall effects or numerical artefacts. An initial

12

mesh comprised of 20 cells per bubble diameter, was found to be sufficient, prior to initial

13

refinement. The mesh is dynamically refined at the liquid interface, with a minimum refinement

14

level of 26 times the number of cells in the refinement region. A fixed limit on the total number

15

of cells in the domain was set at 12 million—a limit which was not reached in any of the results

16

presented. This limit may need to be increased should larger droplets be simulated, however was

17

sufficient for 250, 500 and 1000 µm radius droplets considered in this work. Use of lower levels

18

of refinement, or lower limits on the total number of cells resulted in poor resolution of the liquid

19

column.

20

Given the rapid nature of the bubble bursting process, a sufficiently small time step, and write

21

interval, had to be used to ensure that the behaviour at the interface was properly captured. This

22

was important to allow the tip velocity to be extracted.

ACS Paragon Plus Environment

7

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 28

To allow comparison with the experiments conducted by Ghabache et al.1, a number of

1 2

different water-glycerol mixtures were simulated (water, ethanol and water with varying

3

concentrations of glycerol), which correspond to the range of Morton numbers found to produce

4

‘jet’ droplets as a result of bubble bursting. The properties of the liquids used are given in Table

5

1.

6

Table 1. Liquid properties used in simulations Liquid

Density (ρ) kg/m3

Surface (γ) Tension Viscosity (µ) Pa.s N/m

Water

998

0.072

0.00089

20% Glycerol

1043.5

0.071

0.00135

30% Glycerol

1068.6

0.0695

0.00187

40% Glycerol

1094.8

0.069

0.00272

50% Glycerol

1120.1

0.068

0.00421

55% Glycerol

1134.2

0.0675

0.00568

60% Glycerol

1148.3

0.067

0.00719

65% Glycerol

1162

0.0665

0.01064

70% Glycerol

1175.6

0.066

0.01410

Ethanol

789

0.0224

0.00104

7 8 9 10 11

ACS Paragon Plus Environment

8

Page 9 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Langmuir

RESULTS & DISCUSSION

2

To examine the resolution of the liquid-air interface, the results of simulations were compared

3

visually to the images taken by Ghabache et al.1. For this qualitative validation the simulations of

4

500 µm radius bubbles in water and 50% glycerol solutions were used. The images shown in

5

Figure 2, show the shape of the interface below the bulk liquid surface just before and just after

6

the formation of the liquid column resulting from the collapse of the bubble. These compare

7

favourably to the images shown in Figure 4 of Ghabache et al.1

8

The other important aspect in terms of resolving the interface is the ability to simulate the

9

droplets ejected as part of the bubble collapse process. This is illustrated in Figure 3, for the

10

bubble rising in water case. The expected behaviour (again with reference to the results of

11

Ghabache et al.1 can be seen, in terms of the ejection of the ‘jet’ drops and the ripples on the

12

liquid surface.

13

Tip Velocity

14

The ability to resolve the interface is an important aspect of these simulations, however before

15

such simulation can be applied to liquid (re-)entrainment in coalescing filters, we need to

16

conduct a more quantitative analysis. This has been achieved by considering the tip velocity of

17

the liquid column that results from the collapse of the bubble. By extracting this value from the

18

simulations, we were able to evaluate the values of We, Bo and Mo, as used by Ghabache et al.1,

19

and were able to plot the results of our simulations against their experimental results. These

20

values are shown in Figure 4.

21

ACS Paragon Plus Environment

9

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 28

1

2 3

(a)

(b)

(c)

(d)

4

Figure 2. Shape of the interface below the liquid surface, (a) shows the profile after the bubble

5

has reached the surface and 130 µs before the formation of the liquid column. (b) shows the

6

profile 30 µs before formation of the liquid column, and (c) at time 0, where the bubble has

7

collapsed and the liquid column/jet is about to form. (d) shows the profile 30 µs after the bubble

8

collapse, where the liquid column has formed, though is still below the surface of the bulk liquid.

9 10

Figure 3. The bursting of a 1000 µm radius bubble in a glycerol-water mixture, the column/jet of

11

liquid has formed and ejected a droplet.

12 13

ACS Paragon Plus Environment

10

Page 11 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1 2 3

Figure 4. Experimental results from Ghabache et al.1 compared to our simulations. There is reasonable agreement between the simulated values and the experimental values in

4

Figure 4. It is important to note that where we have used bubbles of either 250, 500, 1000 or

5

2000 µm radius in our simulations, Ghabache et al.1 have used bubbles ranging from 150 to 2100

6

µm. It is also evident from Figure 4, that our simulations have correctly captured the two

7

different relationships with Morton number (as illustrated by considering the results either side

8

of Mo = 3 x 10-8).

9

The differences observed between the 250 µm, 500 µm and 1000 µm bubble diameter cases,

10

also appear to be consistent, with the results for 1000 µm bubbles consistently appearing above

11

the results for the 500 µm bubbles, which themselves appear above the results for 250 µm

12

bubbles. A trend easily explained by considering that both We and Bo are a function of bubble

ACS Paragon Plus Environment

11

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 28

1

size. This also explains the vertical orientation of the results of Ghabache et al.1, for liquids with

2

the same value of Mo.

3

There is some question over the ability to capture Vtip accurately, whether in experiment or

4

simulation. In our simulations, we increase the temporal resolution to ensure that the point where

5

the rising liquid column/jet meets the plane of the liquid surface is captured. This is necessary as

6

the velocity of this liquid jet rapidly changes, as the column/jet develops. We must also consider

7

difficulties in determining such values experimentally, as the movement of the liquid column

8

between individual frames, would produce a similar issue.

9

Bubble size and jet dynamics

10

Earlier modelling work, conducted by Boulton-Stone et al.15 observed that larger bubbles

11

produce proportionally wider jets, a trend we have observed in our simulations also. They also

12

observed that smaller bubbles produce faster jets. If we consider the width of the rising jet in our

13

simulations, we observe both wider jets forming with increasing bubble size and well as with

14

increasing viscosity and decreasing surface tension. The full widths at half height of the fully

15

developed jet were extracted from the simulations and are shown in Table 2.

16

Table 2. Full width at half height of the fully developed liquid jet (µm)

17 % Glycerol

30

40

50

55

60

250

40

40

62

88

114

500

54

77

87

92

101

r (µm)

ACS Paragon Plus Environment

12

Page 13 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1 2

If we consider the values in Table 2, we can see that larger bubbles produce wider jets and that

3

an increase in viscosity (increase in glycerol concentration) also produces wider jets. This

4

widening of the jet with increasing viscosity would appear to be evidence of the role of viscosity

5

in inhibiting jet droplet formation. This can also be seen if we consider the number of droplets

6

produced. Smaller bubbles, in water (high surface tension, low viscosity liquids) produce

7

multiple jet droplets, something confirmed by the observations of Hayami and Toba17, where

8

bubbles of less than 1000 µm radius produced multiple droplets and bubbles less than 2000 µm

9

in radius consistently produced droplets. We see a similar pattern in our simulations, where

10

multiple droplets result from the water and lower glycerol concentrations and only single

11

droplets are given off form higher viscosity liquids, before droplet formation ceases at higher

12

viscosities. We also note that the glycerol concentration at which droplet formation decreases

13

reduces with increasing bubble size, confirming the role of gravity in the process.

14

While the above indicates that qualitatively we are observing the correct behaviour in our

15

simulations, we can also take a more quantitative approach, by considering the scaling

16

relationships proposed by Gañán-Calvo19. In their work, scaling parameters were introduced and

17

applied to the experimental data of a number of authors, showing that they could all be described

18

by a power law. Here we consider two of the terms described by Gañán-Calvo19 a reduced radius

19

and a reduced velocity, as shown in Eq. 5 and Eq. 6, respectively.  5 

ACS Paragon Plus Environment

13

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 28

6

 1

Where, Rd is the radius of the first jet droplet ejected and lµ and Vµ are expressed as follows;   = 7 

 =   8

2

We can plot our simulated data in the same manner, to illustrate the power law relation. This is

3

shown in Figure 5. Here we see that our simulations conform to the -5/3 power law relation

4

shown by Gañán-Calvo19, indicating that our simulations do correctly reflect the behaviour of the

5

physical system.

6 7 8

Figure 5. Reduced radius and reduced velocities for a subset of our simulated data. The line shows the -5/3 power law described by Gañán-Calvo19.

9

ACS Paragon Plus Environment

14

Page 15 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

Parameters to define jet drop formation

2

The range of Mo value covered by Ghabache et al.1 appears to cover the range of liquid

3

properties where ‘jet’ droplets are produced. Additional simulations on 80% glycerol and a

4

multicomponent oil (Mo > 1 x 10-6) did not produce ‘jet’ drops, which would appear to confirm

5

this. Based on our simulations and the results of Ghabache et al.1, we propose that there is a

6

range of Mo values where ‘jet’ drops may arise as a result of bubble bursting.

7

This range starts with the low value of Morton number of the order of 10-11 (that corresponding

8

to water) and increase to a transition point at approximately 3 x 10-8, where the viscosity

9

dominant regime takes over. ‘Jet’ droplets still form under this regime, however the velocity with

10

which they are ejected appears to decrease, to a point where the collapse of the liquid cavity,

11

does not give rise to droplets. Based on our simulations, and existing literature this region occurs

12

at Morton numbers between 1 x 10-6 and 1 x 10-5.

13

Given that the Morton number is heavily dependent on the surface tension and viscosity of the

14

liquid, it may also be meaningful to consider these two parameters alone. If we choose the

15

viscosity to surface tension ratio µ/γ then our region of transition between the surface tension

16

dominant regime (left of Mo = 3 x 10-8 in Figure 4) and the viscosity dominant regime (right of

17

Mo = 3 x 10-8 in Figure 4) occurs at approximately 0.1, and the region where ‘jet’ droplets are no

18

longer formed corresponds to approximately µ/γ = 0.2. Broader application of such limits would

19

require further simulations of additional liquids, with properties around these limits. It may be

20

the case, however, that a more complete (and non-dimensioned) limit can be found in the form of

21

a dimensionless number.

22

ACS Paragon Plus Environment

15

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2

Page 16 of 28

The droplet number Given that Walls et al.8 observed that bubbles in liquids with Oh ≲ 0.037 and Bo ≲ 3 do not

3

always produce jet droplets, we chose to explore this behaviour with our simulations. As our

4

initial work considered a smaller range of diameters than that of Walls et al.8, we chose to run

5

selected simulations on liquids containing rising bubbles up to radii of 4000 µm. We were also

6

able to also identify this additional region and sought to define it in terms of a dimensionless

7

term, initially considering Mo.

8

While the Morton number can be used to provide some indication of when ‘jet’ drops may form,

9

it does not include a term for bubble diameter, a parameter which is also known to influence ‘jet’

10

droplet formation. The use of a viscosity to surface tension term, described above, also presents

11

the same limitation. As we discussed in the introduction, the formation of jet droplets is related

12

to the surface tension, viscosity, gravity and bubble size, so it makes sense to consider a

13

parameter that contains all these variables. Hence we define such a parameter, the droplet

14

number (Dn), where;

15

Dn = f ( g , r , µ , γ )

(9)

16

Given that a certain value of a ratio of viscosity to surface tension appears to correspond to a

17

regime where ‘jet’ droplets form, it is reasonable to assume that a terms governing the formation

18

of ‘jet’ droplets would include such a term. The Ohnesorge number includes such a term, though

19

however, with γ1/2, and does not itself account for gravity. If we also require the parameters of g

20

and r to be included, it makes physical sense to include them together, as the influence of gravity

21

will ultimately be influenced by the size of the bubble.

ACS Paragon Plus Environment

16

Page 17 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

Based on these considerations we propose the following expression for Dn,

2

Dn =

3

rgµ 2

γ2

(10)

Upon evaluating this term for our simulations on 0.5, 1 and 2 mm diameter bubbles, we note

4

that there appears a value of Dn, above which ‘jet’ droplets no longer occur. This threshold

5

occurs when Dn ≈ 0.000126, with values above this not appearing to produce ‘jet’ droplets. The

6

implication here is that jet droplets only occur at very low values of Dn. This appears to reflect

7

the role of viscosity—in Eq. 10 and in the physical system. If we consider the viscosities of the

8

water glycerol mixtures (see Table 1) the values are low for pure water and increase with

9

increasing glycerol concentration, effectively covering 2 orders of magnitude. The pattern we

10

have observed, and which is reflected in Figure 6, is that higher viscosities (for bubbles of the

11

same size, as indicated by Mo) tend not to produce jet droplets, which is certainly the case for the

12

70% glycerol solution. Given that for a fixed droplet radius the value of viscosity shows the

13

greatest range, it makes sense that this parameter has such a strong influence on the value of Dn.

14

With increasing viscosity there is an increase in the value of the numerator in Eq. 10, which

15

results in an increase in the value of Dn. Moving us towards the threshold value. If we consider

16

different size bubbles in a liquid of the same viscosity we will see a similar pattern with the value

17

of the numerator increasing as a function of r. This in agreement with the work of Walls et al.8,

18

who noted the role of viscosity and gravity in limiting the formation of jet droplets.

ACS Paragon Plus Environment

17

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 28

1 2

Figure 6. The droplet number for our experiments and simulations, as well as selected data from

3

Walls et al.8. The dashed vertical lines indicate two potential ‘threshold’ value of Dn, which will

4

be discussed further in text. The three points to the left of the first vertical line, correspond to 0.5

5

mm bubbles, for which Oh < 0.037, and would not be expected to produce jet droplets.

6 7

ACS Paragon Plus Environment

18

Page 19 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Langmuir

Closer inspection of our parameter reveals that it may also be expressed in terms of a function

2

of Bo and Oh, where;

3

 =  !ℎ

(8)

4

In order to further explore the use of this parameter we consider the results of Walls et al.8 and

5

additional simulations conducted for larger bubbles. Our results and selected results from Wall et

6

al.8 (only results where ‘jet’ droplets occurred) are shown in Figure 6. The figure also includes

7

lines for Dn = 0.000126 and Dn = 0.000398, which based on our results present potential

8

threshold values for droplet formation. It should be noted that the 3 data points to the left of the

9

Dn = 0.000126 line, correspond to Oh values of less than 0.037, so are not expected to produce

10

jet droplets. Our intention therefore, is not to provide a single value which dictates jet droplet

11

formation, rather an additional term that works with the established values of Oh and Bo. What

12

we propose is that bubbles in liquids with Oh values above 0.037 and Bo values below 3 would

13

then also require a Dn number below a certain value. The intent of this term is to account for the

14

region where jet droplets would be expected to occur, based on Oh and Bo values, however do

15

not.

16

Given that the majority of our works considers smaller droplets, we chose to further explore

17

the suitability of our new dimensionless number, though consideration of larger droplets. This

18

also presented the opportunity to test different liquids which may influence the value of Dn for

19

which jet droplets occur. As Dn is a function of viscosity and surface tension, this is a possibility

20

that should be considered. Using two different concentrations of water-propylene glycol

21

solution, we tested bubbles larger than 1500 µm in radius, and observed cases where jet droplets

22

did occur and did not occur; and evaluated Dn for these. We found that bubbles with Dn >

ACS Paragon Plus Environment

19

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 28

1

0.000398 did not produce jet drops. Indicating that a limiting value of Dn exists for these cases

2

though the value differs slightly. Given that the viscosity to surface tension ratio of water-

3

propylene glycol mixtures is approximately twice that of water-glycerol mixtures of the same

4

concentration, this difference in terms of limiting Dn value may make sense, though it may also

5

be the case, that further experimental measurements and/or simulations are required to find a

6

single value. A second line for Dn = 0.000398 is given in Figure 6.

7

Our simulations and experiments can be seen to agree relatively well with the region defined

8

by the Dn lines in Figure 6. With our bubble simulations falling left of the Dn = 0.000126

9

vertical line, and our experiments with water-propylene glycol solutions sitting left of the Dn =

10

0.000398 line, for cases where jet drops were observed (black diamond and black circles in

11

Figure 5). There is however some variation in terms of the results of Walls et al.8 and where they

12

sit in relation to the line for Dn = 0.000126. The values which are close could simply be

13

explained by the value of 0.000126 being approximate, however some values are further

14

removed from the Dn = 0.000126 line. For these values there are two possible explanations.

15

Firstly, the manner in which the radius is determined differs between our work and the work of

16

Walls et al.8. In our work we us the radius of the imposed bubble in the simulations, which is

17

essentially the bubble equivalent radius used by Georgescu et al.16 and in a number of previous

18

experimental works2,15,16,17. In contrast Walls et al.8 have used the cap radius, i.e. the radius when

19

the bubble reaches the surface. At this point the bubble takes on a shape closer to an ellipsoid

20

and therefore has a larger radius than a sphere of equivalent volume. This means that the values

21

from Walls et al.8 shown in Figure 6 have effectively been shifted to the right. Had an equivalent

22

radius been determined based on the dimensions of a sphere with the same volume been

23

provided the results would be shifted to the left. Secondly, the methodology employed by Walls

ACS Paragon Plus Environment

20

Page 21 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

et al.8 is not fully described. The values for the properties given glycerol mixtures given by Walls

2

et al.8 vary from the values we have used in our simulations, in particular the value of surface

3

tension. This would influence the results of the simulations. It may be that these were evaluated

4

at different temperatures, however this information is not provided.

5

CONCLUSION

6

We have simulated the ejection of jet droplets from bubble bursting in liquids, with a range of

7

parameters. Using the VoF method and dynamic mesh refinement, we were able to capture the

8

key elements of the physical processes of bubble bursting and ‘jet’ droplet formation. The results

9

compare favourably with existing literature. Our simulations have provided refinement to

10

existing experimental data, to show a bubble size influence on jet drop formation. We have also

11

proposed a dimensionless term governing the formation of ‘jet’ droplets, which complements

12

existing limits, Oh ≲ 0.037 and Bo ≲ 3. Our term, Dn, provides the final boundary condition to

13

the system at a values of approximately 0.000126 and 0.00398. It is hoped that this parameter

14

will prove useful to those studying bubble bursting processes, or applications which involve the

15

bursting of bubbles to produce droplets.

16 17

AUTHOR INFORMATION

18

Corresponding Author

19

*Email: [email protected]

20

Author Contributions

ACS Paragon Plus Environment

21

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 28

1

The manuscript was written through contributions of all authors. All authors have given approval

2

to the final version of the manuscript.

3

ACKNOWLEDGMENT

4

This work was supported by resources provided by the Pawsey Supercomputing Centre with

5

funding from the Australian Government and the Government of Western Australia, and funded

6

by an Australian Research Council (ARC) Linkage Grant (LP140100919). We would also like

7

to acknowledge Marion Arzt for assistance with some experiments and Stefan Wurster for useful

8

discussions on the topic.

9 10 11 12 13 14 15 16 17 18

REFERENCES 1) Ghabache, E., Antkowiak, A., Josserand, C. & Seon T. On the physics of fizziness: How bubble bursting controls droplets ejection. Phys. Fluids 2014, 26, 121701. 2) Knelman, F., Dombrowski, N. & Newitt, D.M. Mechanism of the bursting of bubbles. Nature 1954, 173, 261. 3) Woodcock, A.H., Kientzlen, C.F., Arons, A.B. & Blanchard, D.C. Giant condensation nuclei from bursting bubbles. Nature 1953, 172, 1144-1145. 4) Andreas, E.L., Edson, J.B., Monahan, A.P., Rouault, M.P. & Smith, S.D. The spray

19

contribution to net evaporation form the sea – A review of recent progress. Bound.- Layer

20

Meteor. 1995, 72, 3-52.

ACS Paragon Plus Environment

22

Page 23 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2 3 4 5 6 7

Langmuir

5) Resch, F.J. Darrozes, J.S. & Afeti, G.M. Marine liquid aerosol production from bursting of air bubbles. J. Geophys. Res.: Oceans 1986, 91, 1019-1029. 6) Spiel, D.E. More on the births of jet drops from bubbles bursting on seawater surfaces. J. Geophys. Res.: Oceans 1997, 102, 5818-5821. 7) Spiel, D.E. On the births of jet drops from bubbles bursting on seawater surfaces. J. Geophys. Res.: Oceans 1995, 100, 4995-5006. 8) Walls, P.L.L, Henaux, L. & Bird, J.C. Jet drops from bursting bubbles: How gravity and

8

viscosity couple to inhibit droplet production. Phys. Rev. E 2015, 92, 021002.

9

9) Mullins, B.J., Mead-Hunter, R., Pitta, R., Kasper, G. & Heikamp, W. Comparative

10 11

performance of philic and phobic oil-mist filters. AIChE J. 2014, 60, 2976-2984. 10) Hotz, C., Mead-Hunter, R., Becker, T., King, A.J.C., Wurster, S., Kasper, G. & Mullins,

12

B.J. Detachment of droplets from cylinders in flow using an experimental analogue. J.

13

Fluid Mech. 2015, 771, 327-340.

14

11) Wurster, S., Meyer, J., Kolb, H.E. & Kasper, G. Bubbling vs. blow-off – On the relevant

15

mechanism(s) of drop entrainment from oil mist filter media. Sep. Purif. Technol. 2015,

16

152, 70-79.

17 18 19 20

12) Tomiyama, A., Zun, I., Sou, A. & Sakaguchi, T. Numerical analysis of bubble motion with the VOF method. Nucl. Eng. Des. 1993, 141, 69-82. 13) Xu, Y.G., Ersson, M. & Jonsson, P. Numerical simulation of single argon bubble rising in molten metal under a laminar flow. Steel Res. Int. 2015, 86, 1289-1297.

ACS Paragon Plus Environment

23

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2 3 4 5 6 7 8 9 10 11 12

Page 24 of 28

14) Mead-Hunter, R., King, A.J.C. & Mullins, B.J. Plateau-Rayleigh instability simulation. Langmuir 2012, 28, 6731-6735. 15) Boulton-Stone, J.M. & Blake, J.R. Gas bubbles bursting at a free surface. J. Fluid Mech. 1993, 245, 437-466. 16) Georgescu, S.C., Achard, J.L. & Canot, E. Jet drops ejection in bursting gas bubble process. Euro. J. Mech. B: Fluids 2002, 21, 265-280. 17) Hayami, S. & Toba, Y. Drop production by bursting of ari bubbles on the sea surface (I). J. Oceanogr. Soc. Jpn. 1958, 14, 145-150. 18) Toba, Y. Drop production by bursting of air bubble on the sea surface (II). J. Oceanogr. Soc. Jpn. 1959, 15, 121-130. 19) Gañán-Calvo, A.M. Revision of bubble bursting: universal scaling laws of top jet drop size and speed. Phys. Rev. Lett. 2017, 119, 204502.

13 14 15 16 17 18

ACS Paragon Plus Environment

24

Page 25 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

Supplementary Information

2

If we consider the variables that govern the system, we have;

3

ρ

4

g

m

5

r

m

6

µ

kg

7

γ

kg

8

Bucking ham Pi yields 5 parameters -3 units = 2 variables in a valid non-dimensional system.

9

Conventional set

10

kg

m-3

m-1

s-2

s-1 s-2

Selecting σ, ρ, r as repeating variables; Π =  $ % & ' (

11

Linear system (using primary variables) )(: + + - = 0 .: − 3- + 1 + 2 = 0 3: − 2+ − 22 = 0

12

Solution gives Π =

&  = -

Π =  $ % & '  4 13

Linear system (using primary variables) )(: + + - + 5 = 0 .: −3- + 1 − 5 = 0 3: −2+ − 5 = 0

14

Solution gives

ACS Paragon Plus Environment

25

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Π =

 6&

Page 26 of 28

= !ℎ

1

Our chosen set

2

If we choose different repeating variables, we get an alternate system. Π =  $ (% & ' 

3

Linear system (using primary variables) )(: + + 2 = 0 .: - + 1 − 2 = 0 3: − 2+ − 2- − 2 = 0

4

Solution gives Π =   (/ &/  =

5

From our manuscript  = Π =

6

6&( 

&(  

For the second system we have Π =  $ (% & ' 4

7 8

Linear system (using primary variables) )(: + + 5 = 0 .: - + 1 − 35 = 0 3: − 2+ − 2- = 0

9

Solution gives (&  Π = = 

10

This would suggest that the system can be described by Bo, Oh and Dn.

ACS Paragon Plus Environment

26

Page 27 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

1

Further iterations

2

Changing the repeating variables again, yields two further solutions  8 ( 9 =

3

1 :

And  1 =  (& 

4

For completeness

5 6

There are only 5 possible non-dimensional number form this set. The four considered so far are Bo, Oh, Mo and Dn. The final is (& 8 

7 8 9

Which appears to be a dimensionless radius. We have not considered the tip radius, which would likely bring the Weber number into play. It would appear that Bo, Oh and Dn are sufficient to describe the system.

10 11

ACS Paragon Plus Environment

27

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 28 of 28