STOCHASTIC BUBBLE
FORMATION AND BEHAVIOR
IN NON-NEWTONIAN FLUIDS
JESSICA REDMON
Submitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
Dissertation Adviser: Prof. Dr. Wojbor Woyczynski
Department of Mathematics, Applied Mathematics and Statistics
CASE WESTERN RESERVE UNIVERSITY
August 2019 CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
Jessica Redmon
candidate for the degree of Doctor of Philosophy⇤
Committee Chair
Dr. Wojbor Woyczynski
Committee Member
Dr. David Gurarie
Committee Member
Dr. Longhua Zhao
Committee Member
Dr. Harsh Mathur
Defense Date
June 3rd, 2019
⇤We also certify that written approval has been obtained for any proprietary ma- terial contained therein. Contents
List of Figures ix
Abstract x
1 Introduction 1
2 Fractional Calculus 5
2.1 The Caputo Derivative ...... 10
2.2 Finite Di↵erence Approximations for Fractional Operators ...... 11
2.2.1 Finite Di↵erence Approximation for the Fractional Derivative 11
2.2.2 Finite Di↵erenceApproximationError ...... 12
3 Burgers Equation 19
3.1 Burger’sequationfromNavier-Stokes ...... 20
3.2 Generalized Burger’s Equation ...... 22
3.3 PowerLawFluids...... 24
3.4 Fractional General Burger’s Equation ...... 27
i 3.5 Time Fractional Generalized Burger’s Equation for non-Newtonian Fluids 31
4 The Rayleigh-Plesset Equation 33
4.1 InNewtonianFluids ...... 33
4.2 InNon-NewtonianFluids...... 38
4.3 Simulations ...... 40
4.3.1 Shear-ThinningFluids ...... 40
4.3.2 Shear-ThickeningFluids ...... 42
5 Bubble Cavitation: Formation and Collapse 43
5.1 Bubble Formation ...... 44
5.1.1 Di↵usionofSeedNulcei ...... 44
5.1.2 InitialRadii...... 51
5.2 Bubble Collapse ...... 52
6 Non-Spherical Bubbles 56
6.1 The Equator-Pole Model ...... 57
6.2 Uncoupling the Di↵erentialEquations...... 61
6.3 Nondimensionalizing the Di↵erentialEquations ...... 63
6.4 Numericalexperiments ...... 66
7 Finite Di↵erence Scheme 77
7.1 Finite Di↵erenceSchemesforBurger’sEquation ...... 77
7.2 Deriving the Finite Di↵erenceScheme...... 78
ii 7.3 NumericalExperiments...... 81
8 Future Work 90
iii List of Figures
3.1 Abdou and Soliman’s Fig. 1: The behavior of u(x,t) evaluated by
variational iteration method versus x for di↵erent values of time with
fixed values d=1, = .125, = .6, ↵ = .4...... 22
3.2 Wei and Borden’s Figure 1: The profiles of the transition layers with
u = 1, u = 1. The thick solid line represents a Newtonian fluid 1 2 (n = 1), the thin solid line is a shear-thickening fluid with n =2and
1 the dashed line represents a shear-thinning fluid with n = 2 ...... 27
3.3 Li, Zhang, and Ran’s numerical solution for 3.25 with p =2,↵ = .25
and d = .05...... 30
3.4 Li, Zhang, and Ran’s numerical solution for 3.25 with p =2,↵ = .75
and d = .05...... 31
4.1 Bubble stretched to 1.1R0 and governed by the Rayleigh-Plesset equa-
tion with no outside forcing. Note the damped oscillations as the bub-
ble returns to equilibrium...... 37
iv 4.2 Bubble stretched to 1.1R0 and governed by the Rayleigh-Plesset equa-
tion with sinusoidal outside forcing. The early behavior, from t =0to
t =1e 4, is dominated by the system’s tendency toward equilibrium with the driving force determining the radius after t =1.5e 4seconds, with a transitional region from t =1e 4tot =1.5e 4...... 38 4.3 Bubble behavior in a power law fluid with n =1.1withinitialradius
90% of the equilibrium radius...... 41
4.4 Bubble behavior in a power law fluid with n =1.2withinitialradius
90% of the equilibrium radius...... 41
4.5 Bubble behavior in a power law fluid with n =1.5withinitialradius
90% of the equilibrium radius...... 41
4.6 Bubble behavior in a power law fluid with n =0.5withinitialradius
90% of the equilibrium radius...... 42
5.1 The figure above from ”The Rheology Handbook” shows viscosity against
shearing stress for Newtonian, shear-thinning and shear-thickening flu-
ids. The solid line is a Newtonian fluid, the dotted line is a shear-
thickening fluid, and the dashed line is a shear-thinning fluid...... 48
5.2 Non-Newtonian Di↵usionProcesses ...... 49
5.3 StandardBrownianmotionforaNewtonianfluid...... 49
5.4 Di↵usion for a shear-thickening fluid on the left, and its two-dimensional
projection on the right...... 50
v 5.5 The left figure is di↵usion for a shear-thinning fluid next to its two-
dimensional projection on the right...... 50
5.6 Figure 4.2 from Brennan’s ”Fundamentals of Multiphase Flows”. ”The
nucleus of radius, R0,entersalow-pressureregionatadimensionless
time of 0 and is convected back to the original pressure at a dimension-
less time of 500. The low-pressure region is sinusoidal and symmetric
about 250.” [12] ...... 54
6.1 Evolution of bubble radius and size of the deformation with initial
radius equal to R0, the equilibrium radius, and initial deformation size
2 10 R0. Thisfluidhasthesamepropertiesaswater...... 67
6.2 Evolution of bubble radius and size of the deformation with initial
radius equal to R0, the equilibrium radius, and initial deformation size
1 10 R0. Thisfluidhasthesamepropertiesaswater...... 68
6.3 Evolution of bubble radius and size of the deformation with initial ra-
dius 1.05R0 where R0 is the equilibrium radius and initial deformation
3 size 10 R0. The theoretical fluid used is 100 times more viscous than
water,withthesamesurfacetension...... 69
6.4 Evolution of bubble radius and size of the deformation with initial ra-
dius 1.05R0 where R0 is the equilibrium radius and initial deformation
2 size 10 R0. The theoretical fluid used is 100 times more viscous than
water,withthesamesurfacetension...... 70
vi 6.5 Evolution of bubble radius and size of the deformation with initial
radius 1.1R0 where R0 is the equilibrium radius and initial deformation
3 size 10 R0.Thetheoreticalfluidusedis5timesmoreviscousthan
water,withthesamesurfacetension...... 71
6.6 Evolution of bubble radius and size of the deformation with initial
radius 1.1R0 where R0 is the equilibrium radius and initial deformation
1 size 10 R0.Thetheoreticalfluidusedis5timesmoreviscousthan
water,withthesamesurfacetension...... 72
6.7 Evolution of bubble radius and size of the deformation with initial ra-
dius 1.05R0 where R0 is the equilibrium radius and initial deformation
2 size 10 R0.Thetheoreticalfluidusedis10timesmoreviscousthan
water,withthesamesurfacetension...... 73
6.8 Evolution of bubble radius and size of the deformation with initial
radius 1.2R0 where R0 is the equilibrium radius and initial deformation
2 size 10 R0.Thetheoreticalfluidusedis10timesmoreviscousthan
water,withthesamesurfacetension...... 74
6.9 Evolution of bubble radius and size of the deformation with initial
radius 1.1R0 where R0 is the equilibrium radius and initial deformation
3 size 10 R0.Thetheoreticalfluidusedis5timesmoreviscousthan
water,withthesamesurfacetension...... 75
vii 6.10 Evolution of bubble radius and size of the deformation with initial
radius 1.1R0 where R0 is the equilibrium radius and initial deformation
1 size 10 R0.Thetheoreticalfluidusedis5timesmoreviscousthan
water,withthesamesurfacetension...... 76
7.1 This is a shear-thinning fluid. For smooth initial conditions, the fluid
velocity propagates smoothly toward an equilibrium solution. Note the
bowing of the steady state for choices of ↵ further from 1...... 82
7.2 For smooth initial conditions, the shear-thickening fluid velocity prop-
agates smoothly toward an equilibrium solution. Note that the bowing
of the steady state for choices of ↵ further from 1 does not seem to
depend on fluid type, only ↵...... 83
7.3 With step function initial conditions, we see bubbles occur under cer-
tain conditions. For larger values of ↵,theshear-thickeningfluidresists
breaking to form bubbles, but as ↵ decreases, the sharp changes in ve-
locity indicative of bubble formation begin to occur...... 84
7.4 For larger values of ↵,thereisagainresistancetofluidrupture,and
the graph only begins to feature the sharp changes in u indicative of
bubble formation at ↵ =0.5...... 85
7.5 For smaller ⌘,thepowerlawindex,weseethefluidbreakingforall
values of ↵ rather than only for small alpha...... 86
viii 7.6 For this shear-thinning fluid, we see both increased resistance to fluid
rupture and a slower decay to a steady state...... 87
7.7 Here we see clear fluid rupture for all values of ↵ and for ↵>1, we see
the break continue indicating the bubble’s continued existance. . . . . 88
7.8 Finally, we see as with ⌘ = .5, bubble formation for all values of ↵.We
also see more bubbles staying formed...... 89
ix Stochastic Bubble Formation and Behavior
in Non-Newtonian Fluids by
JESSICA REDMON
Abstract
Conditions for bubble cavitation and behavior in non-Newtonian fluids have numerous applications in physical sciences, engineering and medicine. Non-Newtonian fluids are a rich, but relatively undeveloped area of fluid dynamics, with phenomena from di↵u- sion to bubble growth just beginning to receive attention. In the course of examining bubble cavitation, it became apparent that the random particle motion responsible for determining potential bubble formation had not been researched. As cavitation bub- bles collapse, they deform into a variety of non-spherical shapes. Due to the complex dynamics and the radial focus of current equations on bubble behavior, no accepted model has yet emerged. This work explores the behavior using numerical methods on both fluid and bubble models to examine this system from di↵erent prospectives, culminating in a time-fractional, power-law Burger’s type equation showing bubble formation under these conditions.
x Chapter 1
Introduction
For my doctoral research, I have been most interested in the work pertaining to non-Newtonian and bubbling fluids, though I have also spent time examining more general properties of non-Newtonian fluids and the behavior of non-spherical bubbles in Newtonian fluids. I have primarily been examining bubble formation and collapse in non-Newtonian fluids by way of a time-fractional Burger’s equation. The models developed for examining these phenomena have been studied individually, but have not previously been combined to describe this system.
Over the past couple of decades, research in nonlinear dynamics has taken an interest in fractional di↵erential equations. The foundation of fractional calculus has created an entire class of di↵erential equations that can be used to describe phenom- ena whose current behavior depends on time or space in a longer or shorter reaching way. This class of equation is becoming popular in a variety of applications in fluid
1 mechanics, biology, finance and physics, to name a few [1, 5, 7]. Increasing attention has gone into researching methods to approximate or solve these equations, from trial equations, to homotopy perturbation, and a plethora of numerical methods including
finite di↵erence methods [6]. I will be focusing on using a finite di↵erence approach to examine bubble formation under a modified fractional Burger’s equation applying to non-Newtonian fluids.
The generalized Burger’s equation without a fractional derivative has been studied intently for several decades, giving rise to considerable literature on many interesting applications [2, 7, 8]. The development of the time- and space-fractional derivative opened up a new type of Burger’s equation that include force through the boundary
[1, 4, 5, 6], creating a more accurate system for various Newtonian fluid flows. For non-Newtonian fluids, the generalized power-law Burger’s equation has been devel- oped, and traveling wave solutions discovered and studied [2]. However, this leaves many more complex systems unexplored.
Non-Newtonian fluids are a rich, but relatively undeveloped area of fluid dynamics, with phenomena from di↵usion to bubble growth just beginning to receive attention.
In the course of examining bubble cavitation, it became apparent that the random particle motion of the fluid responsible for determining potential bubble formation had not been researched. I built a model for this process from basic principles based
2 on fundamental particle movement in a similar manner to a derivation for the di↵u- sion equation. While it has not been rigorously tested, initial results follow expected behavior.
Recent interest in bubble growth in non-Newtonian fluids has yielded a new model similar to the Rayleigh-Plesset equation for shear thickening fluids, but derived using more readily accessible fluid properties like coe cient of consistency and shearing stress so that it may be implemented for engineering and manufacturing applications without much, if any, modification [2]. As there were no assumptions in the deriva- tion process specific to shear-thickening fluids rather than any non-Newtonian fluid,
Iappliedthemodeltoshear-thinningfluidswithgreatsuccess.
Another area that has been of longstanding interest is the behavior of non- spherical bubbles. As cavitation bubbles collapse, they deform into a variety of non-spherical shapes that determine the shape and direction of any resulting shock waves [15]. However, because of the complex dynamics and the radial focus of current equations on bubble behavior, no accepted model has yet emerged. As the model for this is new and this is the first successful implementation, and because of the complex dynamics, it is currently intended for fixed viscosity, or Newtonian, fluids.
Numerical methods are often employed in the study of fractional Burger’s equation
3 and generalized Burger’s equation [1, 4, 5, 9] due to the nonlinear nature of the equa- tion. Interesting cases have been found for non-Newtonian fluids, bubbling fluids and parameter choices for the time-fractional Burger’s equation separately [1, 2, 4, 6, 8, 9].
Finite di↵erence approximations have become increasingly popular for solving both fractional di↵erential equations and wave and Burger’s type equations [1, 3, 6]. Re- cently Li, Zhange and Ran developed a linear finite di↵erence scheme for a generalized time fractional Burger’s equation, sometimes used to model bubbling fluids, that is globally stable, providing a groundwork for my own research into non-Newtonian bubbling systems.
4 Chapter 2
Fractional Calculus
This chapter will focus on laying the groundwork for the fractional derivative defini- tions I am using. I will show why it is applicable here, a finite di↵erence approximation for the Caputo derivative and end with a proof to show viability.
Fractional calculus is a generalization of classical di↵erentiation and integration developed to study di↵erential equations of fractional order [24]. There are many es- tablished ways to define a fractional derivative of order ↵ that are di↵erent, but all are considered correct as long as they approach the classical derivatives as ↵ approaches an integer [24, 11]. The first was derived by Liouville in 1832 for the exponential function, followed by a fractional derivative for a power function by Riemann in 1847
[24]. Some of the more common definitions include [11]:
5 1 dn t x(⌧) Riemann-Liouville D↵x(t)= d⌧ t (n ↵) dtn (t ⌧)↵ n+1 Z0 1 t dn x(⌧) Caputo D↵x(t)= d⌧ t (1 ↵) dtn (t ⌧)↵ Z0 n 1 (k) k ↵ t (n) ↵ x (0)t 1 x (⌧) Gr¨unwald-Letnikov Dt x(t)= + ↵+1 n d⌧ (k +1 ↵) (n ↵) 0 (t ⌧) Xk=0 Z 1 Riesz D↵x(t)= t 2cos ⇡↵ (n ↵) 2 n t d x(⌧) n 1 x(⌧) n ↵ n+1 d⌧ +( 1) ↵ n+1 d⌧ dt (t ⌧) t (t ⌧) Z 1 Z
where n is the ceiling of ↵.
While no geometric interpretation of the fractional derivative has been verified, un- like integer order integrals and derivatives, it is generally accepted that the fractional derivative acts as ”memory” for the system [24, 32]. Fractional di↵erential equations have been increasingly employed to model complex systems in di↵erent research ar- eas and engineering applications with nonlinear behavior or long term memory. It is particularly useful for fatiguing or viscoelastic systems [10, 11, 24]. Many problems in electromagnetics, acoustics, viscoelasticity, electrochemistry and material science are also better described by fractional order di↵erential equations [4]. Fractional dif- ferential equations have already been applied to a variety of solids and fluids [24, 23], but work in elastic solids suggests suitability to non-Newtonian fluids.
6 Fractional di↵erential equations (FDE) have been appearing more frequently in both research and engineering applications. The fractional derivative has applications to many physical problems such as ?frequency dependent damping behavior of mate- rials, motion of a large thin plate in a Newtonian fluid, creep and relaxation functions for viscoelastic materials, the PI Dµ controller for the control of dynamical systems, etc. [4] Di↵erential equations of fractional order also describe phenomena in fields like electromagnetics, acoustics, viscoelasticity, electrochemistry, and material science. In
Bagley and Torvik?s review of work done in this field before 1980, they demonstrate that the frequency dependence of the damping materials is well described by models of half-order fractional di↵erential equations.
Mainardi [41] and Rossikhin and Shitikova [42] published an overview of frac- tional derivatives generally for solid mechanics, and particularly for the modeling of viscoelastic damping. Magin [43] published a critical review in three parts for the ap- plications of fractional calculus to bioengineering. Further applications of fractional derivatives in other fields, as well as related mathematical tools and techniques can be found in [44, 45, 46, 47, 48], while applications of fractional derivatives in elec- trochemical processes [36, 37], dielectric polarization [38], colored noise [39],..., and chaos [40] have been demonstrated by other authors.
Fractional calculus has been used to model both physical and engineering pro-
7 cesses that are best described by FDEs, so a reliable and e cient technique for solv- ing FDEs is necessary. Atanackovic and Stankovic [49] analyzed lateral motion of an elastic column this is fixed at one end and loaded at the other in terms of a sys- tem of FDE. Shawagfeh [50] used an Adomian decomposition method for non-linear
FDE. Daftardar-Gejji and Babakhani [52] also used Adomian decomposition method to obtain solutions of a system of FDEs, and explored the convergence of the method.
Daftardar-Gejji and Babakhani [51] studied the existence, uniqueness and stability of solutions to a system of FDE, and presented an analysis of the system. They also later showed an iterative method of solving nonlinear functional equations. Finally,
Momani [54] used Adomian decomposition method to present nonperturbative ana- lytical solutions of both space and time-fractional Burger?s equations. [4]
More recently, the fractional calculus was successfully applied to non-di↵erentiable problems arising in the areas of solid mechanics [68], heat transfer and wave prop- agation [69], di↵usion [70], hydrodynamics [71], vehicular tra c flow [72] and other topics [6]. A fractional calculus approach has recently been used to describe wave propagation in non-local media, sub-and super- di↵usion and in various generalized
Burger’s equations. Sapora, Cornetti and Carpinteri modeled wave propagation in an elastic media where disturbances e↵ect non-local points. They considered displace- ment of a one dimensional, Eringen bar with length L and endpoints a and b [23].
Reimann-Liouville fractional derivates of order 1 <↵<2describetheinteractions
8 between nonadjacent points. Using this model, they were able to analyze wave prop- agation in this one dimensional elastic media and found the non-local interactions resulted in deformations of the wave shape which were more pronounced if only one end of the bar was fixed[23].
D. del-Castillo-Negrete, B. A. Carreras and V. E. Lynch used a fractional oper- ator in a reaction di↵usion system to model the front dynamics of a superdi↵usive media like plasma and other sheer-thinning non-Newtonian fluids. They examined the system at the particle motion level to determine that a fractional operator in the space was appropriate for super di↵usion. Previous studies in reaction-di↵usion systems exhibiting both subdi↵usion and superdi↵usion including [27] ,[28] and [5] used a temporal fractional derivative. At the particle level, superdi↵usive fluid par- ticles did not move according to a stochastic Gaussian process, as is necessary for the use of Laplacian operators, but instead were better described by ↵ stable L´evy distributions with 1 <↵<2. In particular, they analyzed the Fisher-Kolmogorov equation
F ( )= (1 )
with a finite domain: 0