You are on page 1of 96

2016

PASHUPATI DHAKAL

ALL RIGHTS RESERVED


NUMERICAL INVESTIGATIONS OF THE EFFECT OF FILL FACTOR IN AN

INTERNAL MIXER FOR TIRE MANUFACTURING PROCESS

A Thesis

Presented to

The Graduate Faculty of The University of Akron

In Partial Fulfillment

of the Requirements for the Degree

Master of Science

Pashupati Dhakal

August, 2016
NUMERICAL INVESTIGATIONS OF THE EFFECT OF FILL FACTOR IN AN

INTERNAL MIXER FOR TIRE MANUFACTURING PROCESS

Pashupati Dhakal

Thesis

Approved: Accepted:

Advisor Interim Dean of the College


Dr. Abhilash J. Chandy Dr. Eric J. Amis

Faculty Reader Dean of the Graduate School


Dr. Jae-Won Choi Dr. Chand K. Midha

Faculty Reader Date


Dr. Siamak Farhad

Department Chair
Dr. Sergio D. Felicelli

ii
ABSTRACT

Mixing is a complex process in manufacturing where all the component ingredients

are brought together by moving rotors in a closed chamber. Understanding mix-

ing is very important in terms of evaluating the efficiency of the mixer and effect

of various operating parameters such as rotor speed, ram pressure, temperature, fill

factor, etc. With the physical mixing process always being partially filled, it is very

important to determine the effect of fill factor and the best-operating fill ratio that

yields the highest throughput. Mixing also controls the physical, chemical and me-

chanical properties required for downstream processing. Researches on the complex

mixing process are not only limited to increase the productivity of the system but

also to improve the quality of product and repeatability of the operation. Avail-

ability of modern high-performance computing resources and accurate mathematical

models makes computational fluid dynamics (CFD) an important and necessary tool

in understanding some of the complex physical and chemical phenomena associated

with such industrial manufacturing problems. This research presents the calculation

of various flow properties to determine the dispersive and distributive mixing in an

internal mixer. Flow characteristics are calculated in terms of the velocity field and

pressure distribution inside the chamber. Similarly, massless particles have been in-

iii
jected into the flow domain and the statistics of those particles have been used for

determining the distributive and dispersive mixing characteristics for the various set

of operating parameter. For simplicity, this study is conducted on the isothermal

system where the viscous heat generation resulting temperature variation is assumed

to be negligible. A series of 2D and 3D CFD simulations are carried out in a mixing

chamber with various fill factors stirred by counter-rotating rotors. The Volume of

Fluid (VOF) method has been used for capturing the interface between the rubber

and air in partially filled isothermal simulations. Mixing quantities such as length

of stretch, cluster distribution index, and segregation scale have been calculated for

assessing distributive mixing. Dispersive mixing characteristics are assessed based on

the history of shear stress experienced by each particle throughout the entire mixing

cycle.

iv
ACKNOWLEDGEMENTS

I would like to express my sincere gratitude to all the people who have made the

direct or indirect contribution to this project. I am very grateful to my advisor

Dr. Abhilash J. Chandy for providing the opportunity to pursue MS study under

his guidance. He is very friendly and always phenomenal at motivating me on the

academic and research work. I would like to extend the sincere appreciation to the

committee members Dr. Jae-Won Choi and Dr. Siamak Farhad.

I would like to thank The Goodyear Tire & Rubber Company for providing

the research platform for this challenging task. I am very grateful to the whole

Goodyear team for providing valuable suggestion and comments during the course

of this research. Special thanks to my team members Ms. Suma Rani Das and Mr.

Hari Poudyal for being the part of this project.

Finally, I would like to acknowledge my parents, my brother and sister, my

girlfriend and all of my friends, family and relatives for their continuous love and

support throughout my life.

v
TABLE OF CONTENTS

Page

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

CHAPTER

I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.4 Thesis organization . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

II. LITERATURE REVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1 Mixer and mixing mechanism . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Effect of operating parameters . . . . . . . . . . . . . . . . . . . . . 13

2.3 Polymer/Rubber mixing investigations . . . . . . . . . . . . . . . . . 21

2.4 Assessment of mixing quality . . . . . . . . . . . . . . . . . . . . . . 24

III. FORMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2 Multi-phase model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

IV. COMPUTATIONAL DETAILS . . . . . . . . . . . . . . . . . . . . . . . 35

vi
4.1 2D rotor geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.2 3D rotor geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.3 Operating parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.4 Sliding mesh technique . . . . . . . . . . . . . . . . . . . . . . . . . . 38

V. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . 41

5.1 Numerical simulation in 2D geometry . . . . . . . . . . . . . . . . . 41

5.2 Numerical simulation in 3D geometry . . . . . . . . . . . . . . . . . 59

VI. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

vii
LIST OF FIGURES

Figure Page

1.1 The major mixing process in the internal mixer. . . . . . . . . . . . . . 2

2.1 Batch mixer outline showing major parts . . . . . . . . . . . . . . . . . 8

2.2 Two-wing rotor during inside the mixing chamber . . . . . . . . . . . . 10

2.3 Schematic of various mixing steps involved in rubber mixing . . . . . . 12

2.4 Ram positions at different loading (a) overfilled, ram never seats (b)
optimum filled, ram seats in proper time (c) underfilled, ram has no effect 17

2.5 Rotor design based on number of wings (a) 2-wing rotor (b) 4-wing
rotor(b) 6-wing rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6 Shear stress and it’s application time relation. . . . . . . . . . . . . . . 26

2.7 Initial particle position for 2D geometry . . . . . . . . . . . . . . . . . . 29

2.8 Length of stretch illustration . . . . . . . . . . . . . . . . . . . . . . . . 30

4.1 Geometric views of the rotor . . . . . . . . . . . . . . . . . . . . . . . . 36

4.2 (a) Geometry (b) Computational mesh and (c) Grid resolution in
the clearance region between the rotor tip and chamber wall, used
for 2D simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.3 Mesh details across (a) x-z plane (b) x-y plane and (c) Rotor to
chamber wall clearance used for 3D simulations . . . . . . . . . . . . . 39

5.1 Streamlines plot at the end of 10 revolutions (a) on the domain (b)
at the rotor tip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

5.2 Volume fraction of rubber (a) at the beginning (b) at the end of 5
revolutions (c) at the end of 10 revolutions . . . . . . . . . . . . . . . . 44

viii
5.3 Velocity magnitude for (a) 75% (b) 100% fill-factor simulations at
the end of 10 revolutions . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5.4 Pressure plot for (a) 75% (b) 100% fill-factor simulations at the end
of 10 revolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5.5 Probability distribution in %, of the mixing index calculated from


all the computational nodes in the 75% and 100% fill-factor simulations. 47
5.6 Average mixing index for different rotor positions . . . . . . . . . . . . 49

5.7 Maximum shear stress experienced by each particle for an entire


mixing cycle, i.e. 10 revolutions . . . . . . . . . . . . . . . . . . . . . . 51

5.8 Particle distribution for 75% fill-factor simulation (a) at the time
of particle injection (time t = 0) (c) after 2 revolutions (e) after
5 revolutions (g) after 10 revolutions and for the fully filled (b) at
the time of particle injection (time t = 0) (d) after 2 revolutions (f)
after 5 revolutions (h) after 10 revolutions . . . . . . . . . . . . . . . . 53

5.9 Mean length of stretch over 10 revolutions for the 75% and 100%
fill-factor simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

5.10 Probability distribution of the particle pairwise distances, c(r), cal-


culated at the end of mixing cycle, i.e. 10 revolutions . . . . . . . . . . 56

5.11 Cluster distribution index over 10 revolutions for the 75% and 100%
fill-factor simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.12 Particles coded with two colors for 75% fill-factor simulation (a) at
the end of 10 revolutions (c) at the end of 20 revolutions and for
the fully filled (b) at the end of 10 revolutions (d) at the end of 20
revolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

5.13 Cluster distribution index over 10 revolutions for the 75% and 100%
fill-factor simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

5.14 Velocity contours and vectors across the radial cross section at (a)-
(b) one-quarter of rotor height (c)-(d) at half of the rotor height
(e)-(f) three-quarter of rotor height (g)-(h) across axial cross section
for 45% fill factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5.15 Velocity contours and vectors across the radial cross section at (a)-
(b) one-quarter of rotor height (c)-(d) at half of the rotor height
(e)-(f) three-quarter of rotor height (g)-(h) across axial cross section
for 60% fill factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

ix
5.16 Velocity contours and vectors across the radial cross section at (a)-
(b) one-quarter of rotor height (c)-(d) at half of the rotor height
(e)-(f) three-quarter of rotor height (g)-(h) across axial cross section
for 75% fill factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

5.17 Velocity contours and vectors across the radial cross section at (a)-
(b) one-quarter of rotor height (c)-(d) at half of the rotor height
(e)-(f) three-quarter of rotor height (g)-(h) across axial cross section
for 90% fill factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

5.18 Velocity contours and vectors across the radial cross section at (a)-
(b) one-quarter of rotor height (c)-(d) at half of the rotor height
(e)-(f) three-quarter of rotor height (g)-(h) across axial cross section
for fully filled simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.19 Maximum shear stress over 10 revolutions for the five different fill factors 70

5.20 Particle distribution (a) at the time of particle injection (time t = 0)


and (b) at the end of 10 revolutions for three-dimensional geometry . . 71
5.21 Mean length of stretch over 10 revolutions for the five different fill factors 72

5.22 Cluster distribution index over 10 revolutions for the 5 different fill factors 74

5.23 Segregation scale over 10 revolutions for the 5 different fill factors . . . 76

x
LIST OF TABLES

Table Page

4.1 Non-Newtonian Parameters . . . . . . . . . . . . . . . . . . . . . . . . 38

5.1 Percentage of particle broken down for various critical shear stress . . . 69

xi
CHAPTER I

INTRODUCTION

1.1 Background

Mixing is an important step in polymer processing operation as it reduces the gra-

dient or inhomogeneity of the constituents in concentration, density, temperature or

phase of material in bulk. Efficient mixing in mixing chamber is critical in terms of

of meeting the demand of new materials with specific properties that are designed

to meet very stringent and diversified requirements. Mixing not only reduces the

non-uniformity of mixing materials, but is also responsible for breaking down the

filler agglomerate into smaller pieces that promotes the molecular level interaction.

In polymer processing, to achieve this goal, component materials are agitated using

the rotors in a closed mixing chamber which goes through the different stages of im-

miscible material mixing. Figure 1.1 shows the schematic detailing of the importance

of dispersion and distribution of particles. The knowledge of fluid flow property and

mixing mechanism is needed to evaluate the degree of mixing achieved in each cycle.

The numerical simulation of complex mixing process provides the detailed informa-

tion of mixing quality thereby helping to optimize and improve the overall system

performance.

1
Figure 1.1: The major mixing process in the internal mixer.

2
1.2 Motivation

Because of the multiple operating parameters that can significantly alter the proper-

ties of output product, the understanding of mixing mechanism has always been very

intriguing. Parameters like rotor speed, ram pressure, fill factor, rotor orientation

have a huge impact on the final product and optimization of those parameters can

lead to a significant change in mixing cycle. This research is focused on the study

of the effect of fill factor in an internal batch mixer. A fill factor that is too high

leads to the material being stagnant in most of the region of mixing chamber lead-

ing to the formation of dead spots which do not aid in mixing. A fill factor that is

too low allows material to move freely in the narrow clearance region between rotor

tip and chamber wall leading to insufficient shearing force for filler dispersion. The

results presented from this research work on five different fill factors will be the first

study of this kind and many high-viscous mixing industries are expected to receive

the valuable information required for the optimization of mixing process.

1.3 Objectives

The specific objectives of this research work are summarized in the following points:

• Study the overall flow mechanism in an internal mixer and validate with the

literature and patent claims

• Develop the methodology for numerical simulation of internal batch mixers

3
• Model the geometry and entire manufacturing process of rubber mixing in batch

mixers

• Identify the best combination of operating parameter to improve the quality of

output product and throughput

1.4 Thesis organization

The thesis has been divided into 5 different chapters. Chapter 1 covers the general

background and motivation behind the work. Chapter 2 includes the detail literature

review of the mixer and mixing mechanism. It includes the theory behind different

operating parameters and their effect on mixing. Statistical quantities to calculate

the mixing quality using particle information is also presented along with previous

numerical studies and their outcomes. The detail mathematical modeling is presented

in chapter 3 which includes the continuity and momentum equation along with for-

mulation for multi-phase and particle trajectory. Chapter 4 describes the 2D and

3D geometry with correct dimensions, boundary conditions, and material properties.

The simulation results are presented in chapter 5 where the initial study is performed

on the 2D geometry. The results verified the well-accepted flow phenomena and pro-

vided important insights into the mixing study. The study is extended to five different

fill factors for a 3D geometry. A detailed study of material flow behavior is presented

using the contours vectors plot. Particle based statistics have been calculated to

find the mixing quantities such as maximum shear stress, cluster distribution index,

4
length of stretch and segregation scale for assessing dispersive and distributive mixing

characteristics.

5
CHAPTER II

LITERATURE REVIEW

2.1 Mixer and mixing mechanism

The primary goal of any mixing process is to achieve the coherent homogeneous mass

of constituent material by inducing physical motion. This kind of mechanism occurs

in a sophisticated piece of heavy equipment called internal mixer. Mixing of rubber

compound is laminar, as turbulent mixing is unattanable at the high viscosity of rub-

ber (O(105 ) Pa-s). The mixer should have the unique capability to generate motion

in both axial and radial directions with the material being stretched, compressed,

sheared, folded and kneaded by the action of rotors [1][2].

2.1.1 Internal mixer

The batch operated internal mixer is the most widely used mixing device in polymer

industry because of its flexibility to intake the mixing material in different shapes

and forms. There are two major designs in terms of rotor interaction - intermeshing

and tangential. Adjacent rotors in the intermeshing design interfere with each other

to lock in the groove and operate at the same speed. The tangential rotors on the

other hand, never touch each other and can operate at variable speeds [3].

6
An internal mixer is characterized by a mixer entry chute which acts as the

feeding port for material input. The floating weight of the ram pushes the raw

material downwards to the mixing chamber. The mixing chamber is a lobe-shaped

tank in which two rotors rotate about their axis. The shape of the mixer entry chute

allows the feeding of raw materials in almost any trade form. Rubber bales, the

most widely used trade form for rubbers and polymers, can be fed as one piece in an

industrial-sized mixer. During the mixing process,which happens in a closed chamber,

the ram is at its bottom position and drop door is also closed. At the bottom of the

chamber, a drop door can be opened to discharge the mixed product for downstream

processing. In the drop door or through the side plates a thermocouple is mounted

to record the temperature of the rubber. In the mixing chamber, both rotors and

the drop door are temperature controlled by means of a closed-loop water control.

Underneath the mixer, a downstream device, such as a sheeting extruder with one

or two tangential screws, or two roll mills are placed. The major advantage of the

internal mixer is a high output, relatively short mixing time and highly reproducible

output although heat generation is the major drawback for this kind of machine [4][5].

However, internal mixing operation is expensive because of the batch-wise

production of rubber compounds. This leads to quality variations and batch non-

uniformity. The temperature increase of the compound limits the mixing process

and, therefore, the mixing quality. Eventually, the mixing quality could be so poor

that an extra mixing step, re-milling, is needed. A rheo curve is used to determine

7
Figure 2.1: Batch mixer outline showing major parts [5]

8
the batch to batch consistency and uniformity of the mixing. The variation can be in

terms of viscosity, curing characteristics, and compounds processability as well [5][6].

2.1.2 Mixing mechanism

The Rubber mixing mechanism inside the closed chamber mainly depends on the

nature of feedstock. Generally, rubber is introduced in the form of a bale and other

constituent materials are either solid agglomerates or in liquid or powder form. The

overall mixing mechanism can be divided into following four steps: Subdivision, In-

corporation, Dispersion and Distribution.

2.1.2.1 Subdivision

During subdivision, large scale solid agglomerates are broken down into smaller

aggregates. The viscosity of rubber material decreases due to the molecular dis-

entanglement induced by flow mechanism and also because of increase in tempera-

ture. Flow-induced dis-entanglement of rubber molecules is mainly produced by high

strain rate. However, there is one irreversible viscosity reduction because of chain

scission (mastication), which mainly changes the molecular structure of rubber [3].

2.1.2.2 Incorporation

Incorporation represents the transformation of bulk and separate masses of con-

stituent mixing materials into a cohesive and incompressible mass which can undergo

viscous flow. In this stage, the free surface folding flow that occurs between the

rotors, and the rotor and chamber wall encapsulates the agglomerates. The encapsu-

9
Figure 2.2: Two-wing rotor during inside the mixing chamber [7].

10
lated mass is highly heterogeneous where particles undergo further division utilizing

shear and elongational flow in the rubber. The particulate mass changes into pellets

which are entrapped into the rubber along with void spaces. During the mixing pro-

cess the void space is finally displaced by the rubber which leads to the increment of

the density of rubber mass to a constant value and considered as fully incorporated

mass [3] [4].

2.1.2.3 Dispersion

Dispersion starts after the internal spacing of filler are occupied by rubber, giving it a

perfect solid shape. Dispersion is the process of breakdown of filler agglomerates into

smaller particles by the application of shearing force. Practical studies have shown

that straining is more effective than shearing for dispersion. The combined effect of

high viscosity of rubber and the close clearance in the mixing chamber generates a

large amount of strain during mixing. The probability of rupture of agglomerates

is higher for larger particles but the dispersion process becomes more difficult as

it progresses. The viscosity of rubber material also decreases due to the chemical

interaction with fillers, which reduces the chances of dispersion further. The reduced

viscosity of the rubber and smaller filler size makes it practically impossible to achieve

the full dispersion and a small portion of undispersed filler can be present in the final

product even if the mixing has been carried out for a long time [3].

11
Figure 2.3: Schematic of various mixing steps involved in rubber mixing source:
http://rubbermachineryworld.com/tag/rubber-mixing/

2.1.2.4 Distribution

Distributive mixing is the process of homogenization of the composition to achieve

the uniformity in the quality of mixed products by the transfer of particle from one

point to the other. Mixing in the internal mixer is mostly macro-mixing as the feeding

of filler material is initially concentrated in the small region in the mixing chamber

and it is necessary to distribute them throughout the matrix. Distributive mixing is

generally achieved by two mechanisms called laminar shear mixing and exponential

mixing. Laminar shear mixing is created in the fully filled mixing chamber which

involves close streamline flows. Exponential mixing involves the subdivision and

recombination of streamlines and which is very effective for segregating the flow. The

creation of free surface flow generates the force exchange of material between the

12
closed streamlines . For this reason, most of the practical mixers are operated at

partially filled condition. The driving force for better distribution is rotations of the

material within the mixing chamber, intensity of the material transport and the total

number of rotation applied to the batch [3].

2.2 Effect of operating parameters

The manner in which the mixture has been operating, has a significant effect on pro-

duction cost as well as the quality of the output product[8]. The effect of different

variables of any mixer can be studied to obtain insights into achieving the highest

quality product with the minimum cost. Although the extent of the effect mixing

operation variables is dependent on the type of compounder and their specific pur-

poses, the mixing operation can be effectively analyzed and/or optimized only after

the variables affecting mixing process have been identified and controlled. Some of

the variables are as follows:

2.2.1 Rotor speed

Higher rotor speed generates a higher shear force but the temperature rise is also

very high in such a case. The speed reduces later in the process in order to achieve

the temperature within the desirable range allowing more time for heat transfer.

It is very important to maintain the temperature within the range to prevent the

rubber material from possible degradation. Heat generation and rotor speed relation

vary with the rotor designs as intermeshing rotors have close clearance and better

13
cooling characteristics. However, the feeding intake capacity is significantly reduced

for higher rotor speed. The material on the intake side can have very high viscosity

if it stays on the top of the rotor for too long. There should be a good correlation

between batch weight and rotor speed for getting a well-mixed product [9].

2.2.2 Single speed and variable speed rotors

Mixers that are operating with single motors have uniform speed on all the rotors.

Uniform speed is one of the most peculiar characteristics of intermeshing rotors where

variable speed is impossible because of the geometric constraint. Single speed mixers

are normally used in situations where the mixer is dedicated to a specific product or

group of products when the rotor speed has been determined based on the experience

of producing that product.

Variable speed mixers are more common in the tangential mixer where each

rotor can operate independently. Speed ratio which is the ratio of two rotor speeds,

plays a significant role in the mixing properties. Variable speed rotors gives a better

balance between product quality and productivity because of ability to select a specific

mixing speed. Mixer rpm can be varied during the mixing cyle in order to match the

desired phase angle between the rotors that yield best quality product [8].

2.2.2.1 Fill factor

Fill factor, a very important mixing parameter, is the percentage of filling of mixing

chamber. Both under-filling and over-filling the mixer are not efficient in terms of

mixing performance. If the mixer is under-filled, the ram bottoms immediately (goes

14
to the full down position) as the batch gets inserted into the mixer under minimum

ram pressure. Under this condition, shear mixing is very poor as ram is not providing

enough pressure. On the other hand, if the batch is too large, the ram never gets

to the bottom position. Fed material is stagnant on the throat region, and is never

forced into the mixing chamber. Hence, there is always an optimum fill factor, where

there is enough pressure for shearing action and also, enough void space to allow

material transport between the chambers. It should also be noted that the higher

fill factors do not imply higher product quality, as the material should go through

enough dispersion and distribution of filler particles. On the other hand, although

lower fill factors allow better temperature control and better distribution (in some

cases), it fails to achieve the desired level of dispersion.

For a mixing cycle, the fill factor should be optimal for every step, but it might

not happen in reality. The changes in viscosity because of temperature variation

changes the optimal fill factor. In practice, observation of ram movement and power

requirement during the mixing helps to determine the optimum value of fill factor [8]

[9].

2.2.2.2 Temperature

The rate of heat generation is very high in the mixing chamber as the mechanical

energy from rotors is converted into thermal energy. The temperature inside the

chamber depends on the various other operating parameters, but maintaining an

optimum temperature that is necessary to preserve the material properties is very

15
important. Heat generation is directly proportional to the rotor speed, fill factor,

material viscosity and applied batch pressure. So, it necessary to determine the

optimum combination of all these parameters to control the temperature such that

it stays within the working limit. Mixers are usually cooled with the water jacketing

in the rotor and also in the barrel wall.

2.2.2.3 Ram Pressure

The main purpose of using a ram at the top is to push the fluid (in this case, rubber)

downward during charging and prevent their upward exit during mixing. The use of

higher ram pressure during mixing increases the dispersion quality. It is important

to apply constant ram pressure during the dispersion phase. As soon as the carbon

black incorporation is complete, the ram pressure can be reduced. It is not advised

to use the maximum ram pressure for the entire mixing duration as it leads to rapid

temperature increase as well as wear and tear of the rotor parts. Excessive ram

pressure also leads to the situation where material is moved along with the rotor thus

preventing rotor to rotor exchange of material. On the other hand, inadequate batch

pressure leads to very long mixing time because of very low dispersion. Ram pressure

should be monitored very closely to make sure it is staying on the right position

yielding consistent quality.

The position of the ram corresponds to the amount of ram pressure, which

also indicates whether the used batch weight is correct. For an overfilled mixer, ram

fails to touch the final bottom position and for an underfilled mixer, ram hits it’s

16
Figure 2.4: Ram positions at different loading (a) overfilled, ram never seats (b)
optimum filled, ram seats in proper time (c) underfilled, ram has no effect [10].

17
bottom very quickly. But the ram movement is different depending upon the nature

of rotor mixer. For intermeshing rotors, the intake behavior is controlled by the ram

movement [8] [10].

2.2.3 Rotor Design

Rotor design is the most critical part of the entire mixing process as it determines the

overall flow mechanism inside the mixing chamber. The major function of the rotors

is to disperse the mixing particle and distribute them homogeneously maintaining

the temperature within the desirable limit. The rotors have to be extremely sturdy

to withstand the force generated during the mixing process.

2.2.3.1 2-wing rotors

One of the more primitive designs involves each rotor being equipped with two wings

originating from the two ends of the shaft. One of the wings extends substantially

along the axis than the other short wing. The wings are helically disposed in opposite

direction and the end of the rotors may overlap at some portion of the shaft towards

the center. This kind of rotor is most widely used in mixing industry starting from

polymer to plastic because of their broad area of application. However, 2-wing rotors

are obsolete in most of the new internal mixers because of low productivity [11].

2.2.3.2 4-wing rotors

The productivity is significantly increased from 2-wing to 4-wing rotors because of

two additional wings which facilitate further mixing. So, 4-wing rotors have more

18
rapid dispersion and distribution, but at the same time, heat generation is also very

high. The temperature of the mixing material is higher and so 4-wing rotor-equipped

mixers are designed with a better cooling system. Because of their greater advantage

in terms of material handling and productivity, most of the modern tire and rubber

industry uses 4-wing rotor equipped mixers [9].

2.2.3.3 6-wing rotors

6-wing rotors require higher drive power than the previous two models. It has 3 long

and short wings on each end of the shafts which are set apart by an equivalent angle

in circumferential direction. The basic design feature includes varying wing tip to

chamber side clearances and varying wing tip widths. Because of the higher number

of wings, this model exhibits better mixing efficiency and higher output. Compared

to 2-wing and 4-wing rotor design, viscosity reduction is much faster and mixing time

is shorter while maintaining a good compound quality. This design is excellent for

ingredient dispersion and also for silica mixes [9].

2.2.3.4 Intermeshing and Tangential rotor

Rotors are mounted parallel to each other in their corresponding lobe of the mixing

chamber. In a tangential mixer, the tips of rotors never touch each other and there

is always a constant gap between wing tips so that they can be operated with a

speed differential. Generally, the speed of the rotor moving faster is 10 to 15 percent

higher than the other. This kind of differential speed creates a continuously varying

19
(a) (b)

(c)

Figure 2.5: Rotor design based on number of wings (a) 2-wing rotor (b) 4-wing
rotor(b) 6-wing rotor source http://www.ksbiusa.com/machinery rotors.htm

phase angle between the rotors which improves the mixing. High shear stress and

elongational flow occur at the tip clearance [12].

The intermeshing rotors have a relatively smaller gap and they overlap with

each other, such that they need to be operating at an even speed. Feeding is mostly

assisted by the high-pressure ram force which generates excellent contact between

mixing material and metal surface leading to better a heat transfer coefficient. Al-

though the mixing chamber volume is less as compared to the tangential mixer of the

same size, intermeshing rotors provide larger surface area for cooling. Similarly, the

shearing action takes place not only between wing tip and chamber walls but also

between the rotor tips as they have a narrower clearance [13].

20
2.3 Polymer/Rubber mixing investigations

Internal mixers are widely used in polymer industry to achieve desirable properties

in the final product resulting from a variety of manufacturing processes. Sufficient

and consistent mixing is essential for an effective product performance. The mixing

process is very sophisticated due to the material properties of the polymer particu-

larly viscosity and also due to the several steps involved in the process, which usually

involves a closed rotor-equipped chamber [14]. There are various operating parame-

ters such as mixing time, rotor speed, batch pressure, and cooling rate that have to

be varied depending on the final product type, in order to exploit better, the inter-

actions between the compounding ingredients at the molecular level that can in turn

lead to the formation of a polymer with desirable properties [8]. Practically, a mixing

chamber is always partially filled and the fill factor, which is the percentage of the

volumetric filling, has a significant impact on final product characteristics. If the fill

factor is too high, material remains stagnant in most of the region in mixing chamber

leading to non-uniform mixing, whereas very low fill factors have lots of voids and,

therefore, the compound does not undergo enough shear for an efficient mixing.

There have been quite a few experimental studies investigating the flow be-

havior in the mixing chamber using transparent chamber wall [15, 16, 17]. One of

the very first attempts to study flow behavior in a partially filled mixer using a phys-

ical model was by [18], where the focus was on the flow and mixing behavior of the

region between rotor tip and chamber wall, which really is the most active area of

21
shearing action for tangential mixer. Later, Freakly and Idaris showed that the fill

factor is one of the key variables, which influences the mixing uniformity [19]. In

some of their subsequent studies [20, 21], they used a highly-instrumented internal

mixer with temperature and pressure sensors for studying the effect of rotor speed

and batch temperature on mixing. With regard to numerics, preliminary studies on

the flow behavior of the mixing chamber have assumed the chamber to be fully filled

because this greatly reduces the numerical issues related to convergence and extensive

CPU times that arise due to the presence of the additional air phase. Manas and

co-workers [22, 23, 24, 25] have carried out flow simulations in a fully-filled internal

mixer as well as in twin screw extruders using a commercial software and have cal-

culated mixing quantities related to them. They presented the basic flow behavior

in terms of pressure and velocity contours, and also calculated the effect of rotor

design, material property, and rotor speed. Kim and White [26, 27] investigated in

3D, the flow behavior in a fully-filled mixer for both isothermal and non-isothermal

conditions thereby predicting the temperature rise during the mixing process. Recent

studies have looked at the effects of shear thinning and viscoelasticity on mixing in

a 2D mixer model using ANSYS Polyflow, and have also analyzed the distributive

mixing and mixing efficiency of the mixer [28, 29, 30].

However, very little is known about the flow properties in a partially filled

internal mixer. Literature suggests a fill factor of 0.65-0.85 is generally used, based

on the nature of rotor design and material properties [14]. Numerical studies in

the partially filled mixing chamber were performed by Ghoreishy & Nassehi [31, 32],

22
where they developed a stable numerical code to solve the transient free surface flow.

Another recent study [33] shows the basic differences in flow mechanism between fully

and partially filled chambers. However, the mixing properties were never quantified.

So, there is a need for a more extensive study of mixing in terms of the dispersive and

distributive characteristics in partially-filled mixing, which will be covered extensively

in this thesis.

The general purpose of any rubber mixing process is to disperse and distribute

the recipe ingredients into macroscopically homogeneous rubber mixtures so that the

elastomer material formed by vulcanization will later have the entire set of properties

desired by the materials engineer [9]. For instance, filler materials like carbon black

or silica are mixed with rubber to improve various mechanical and chemical prop-

erties. The major mixing mechanisms are dispersive and distributive mixing. Solid

agglomerates incorporated into the rubber matrix may be large and therefore, might

have to be reduced to a smaller length scale. The process of the break-up of filler

particles and agglomerates in the rubber matrix requires a large amount of energy

because of cohesive forces between molecules and typically a very high viscosity [4].

The magnitude of applied shear stress is critical for determining the dispersive mix-

ing ability, which involves breaking down (rupture) of filler agglomerates from their

original size, so that the contact surface between filler particle and rubber material

increases [25]. Studies have also showed that elongational flow is more effective than

pure shear flow for the fluids which have high viscosity ratio and low interfacial ten-

sion [34]. Distributive mixing, on the other hand, is the process of dissemination of

23
these small filler particles throughout the rubber matrix to achieve some degree of

homogeneity [9].

2.4 Assessment of mixing quality

The primary objective of any mixing process is to disperse and distribute fillers as

well as agglomerates in the shortest possible time. The experimental study involves

visualizing the macro and micro-mixing behavior through various methods such as

optical transmission microscopy, optical roughness measurement, reflectometry, elec-

tron microscope to name few. However, there are still doubts about the consistency of

the mixing performance if the sample is taken from the different place in the mixing

chamber. A numerical study, on the other hand, is able to provide the detail informa-

tion, even including the breakup behavior for every single filler particle. The overall

flow pattern in the chamber and mixing mechanism is very helpful in deciding the

best design or operating parameters to maximize the efficiency of the given system.

2.4.1 Dispersive mixing

Rupture and erosion are the two major mechanisms for dispersive mixing as both

of them reduce the size of the agglomerate and increase the surface area of contact.

The filler particle in a rubber matrix, whose motion is induced by the rotor motion

experiences the hydrodynamic force, which breaks down the particle agglomerates.

Rupture is the sudden breakdown of the pellet into multiple pieces. When the ap-

plied hydrodynamic force exceeds the cohesive force between the pellets, rupture is

24
propagated within the cluster. An experimental study has shown that the value of

critical shear stress is inversely proportional to the radius of the pellet [35].

1
τcritical ∝ , (2.1)
R0

The experimental study also concluded that the rupture mechanism resulted in var-

ious fractured pieces, where the radii of multiple pieces had no relation with each

other and in most of the cases there were more than two pieces.

2.4.1.1 Shear stress on a particle

The magnitude of applied shear stress determines whether the particle can be broken

down. The figure below shows the characterization of particle size as the function

of applied forces and time. It can be seen that the filler particle can resist a certain

amount of stress for an indefinite time. A higher shear stress can have a larger impact

and hence is more efficient in terms of dispersion.

A mixing chamber should have regions where the shear stress is higher than

critical, so that good dispersive mixing can be accomplished. In addition, multiple

passes of the agglomerates across the high-shear stress region might be necessary

to break the filler particles in order to achieve the fine level of dispersion [6]. The

majority of the shearing action takes place between the rotor tips, and also in the

clearance between rotor tip and the chamber wall. The dispersion is more difficult

for low viscosity mixtures and slow rotor speed. The shear stress generation for

a given material at given speed is dependent on the tip clearance, as narrower tip

clearance generates higher shear stress. On the other hand, the amount of mixture
25
Figure 2.6: Shear stress and it’s application time relation.

26
material passing through that region decreases with a decrease in clearance. One of

the best ways of comparing dispersive mixing for different fill factors is to compare the

maximum shear stress experienced by the tracer particle for the entire mixing cycle

[12]. The effectiveness of this process can be assessed using the maximum shear stress

distribution in the chamber. By considering the history of shear stress experienced

by the fluid (rubber) particle with time, a cumulative distribution can be calculated

and compared in order to understand the behavior of shear stress experienced by

the material inside the chamber. If the critical value of shear stress is known, such

a cumulative distribution function of maximum shear stress experienced by each

particle determines the fraction of the particles that are broken at least once during

the entire mixing cycle [36]. As described earlier, there is no use of applying shear

stress smaller than the critical shear, however, the probability of particle experiencing

shear stress higher than critical shear increases with the increase in mixing time.

2.4.1.2 Mixing index

To quantitatively analyze dispersive mixing, a global quantity called mixing index,

λM Z , can be employed. Mixing index is defined as [23]:

|D|
λM Z = , (2.2)
|D| + |ω|

where |D| is the magnitude of the rate of deformation tensor and |ω| is the magnitude

of vorticity tensor. The value of mixing index ranges from 0 to 1, where 1, 0.5 and 0,

represent pure elongational flow, simple shear flow, and pure rotational flow, respec-

tively. It was also found that the forces in elongational flow are much higher than
27
the shear flow [34]. Mixing index, λM Z , is a crude way of representing the dispersive

mixing which quantifies the elongational, simple shear and rotational force compo-

nents. A histogram of the mixing index can be calculated from all the computational

cells. However mixing index is frame invariant. Better dispersive mixing is indicated

by a larger percentage of mixing index values ≥ 0.5.

2.4.2 Distributive mixing

Distributive mixing is the process of dissemination of broken filler particle throughout

the entire mixing chamber. Distribution is the key of homogenization of filler agglom-

erates. There are various statistical methods to calculate the degree of distributive

mixing based on particle information that can be injected in a various form during

the mixing.

2.4.2.1 Particle tracking

Massless virtual particles injected during mixing have all the information required for

post-processing and calculating the mixing characteristics. The number of particles

should be sufficient to accurately predict the final particle distribution.

2.4.2.2 Mean length of stretch

An important quantity to consider for assessing distributive mixing is called length

of the stretch. If the initial distance between a particle pair is X0 and their distance

after time t is Xt , length of stretch is defined as [37]:

|Xt |
Lλ = (2.3)
|X0 |

28
Figure 2.7: Initial particle position for 2D geometry

The value of the length of stretch increases as the particles begin to distribute through-

out the mixing chamber. In most of the cases, the initial particle injection is in the

form of the cluster which means the mean length of stretch is smaller at the beginning

of mixing, but this quantity increases with the better mixing.

2.4.2.3 Cluster distribution index

Furthermore, another parameter for evaluating distributive mixing is the discretized

pairwise correlation function given by [38]:

2 X
f (r) = δ(ri + r)δ(ri ), (2.4)
M (M − 1) i

where, M is the total number of particles and f (r) is the correlation coefficient

between the particle pair ranging from distance r − ∆r/2 to r + ∆r/2. If the particle

29
Figure 2.8: Length of stretch illustration

is present in that region, δ(r) is 1, and it is 0 otherwise. This function can be related

to the probability density function as:

Z r+∆r/2
f (r) = c(r) dr. = c(r)∆r (2.5)
r−∆r/2

where c(r) is the probability density function such that area under the curve of c(r)

is always constant and equal to 1.

r=r
X max

c(r)∆r = 1 (2.6)
r=0

The pairwise distribution at the end of mixing cycle, when compared with the ideal

distribution, provides the difference in uniformity of mixing. The ideal distribution

is obtained by distributing the particles throughout the domain uniformly and hence

is solely a function of the mixer geometry. It is assumed that ideal distribution

represents a perfect mixing scenario and is independent of number of particles con-

sidered. The cluster distribution index is an indication of the deviation from this

30
ideal distribution and it can be mathematically expressed as [12]:
R∞
0
[c(r) − c(r)ideal ]2 dr
= R∞ . (2.7)
0
[c(r)ideal ]2 dr

Smaller values of  represent an improvement in distributive mixing. In order to

specifically quantify the deviation of the distribution of particle pairwise distances

from the ideal distribution, the cluster distribution index can be plotted with time.

2.4.2.4 Segragation scale

Segregation scale is the coefficient of correlation with distance between particles in a

sample [39]. Mathematically, the coefficient of correlation Rx is given by,


N
P
(Ci − C̄) (Cj − C̄)
i=1
Rx = (2.8)
N σ2

where, Ci and Cj are the concentrations of ith and jth pair, respectively, and C̄ is the

average concentration of the sample. N is the total number of particle pairs in the

selected bin size, given by N = M (M − 1)/2 where M is the total number of particles

and σ is the standard deviation of the sample [40, 41]. The value of Rx for particle

pairs is integrated from R(0) = 1 to R(r) = 0 ,to obtain the value of segregation scale

for that particular point in time. So, the segregation scale is defined as:

Z r
Ss = R(r) dr. (2.9)
0

Particle concentration is assigned for each particle based on their co-ordinate and the

mixing of two different concentration is observed throughout the mixing cycle.

31
CHAPTER III

FORMULATION

The actual mixing process involves a combination of rubber material and various filler

particles with air occupying the rest of the chamber. Such a process involves complex

interaction between phases, such that the interface tracking is of utmost importance

for accurate prediction of mixing quantities.

3.1 Governing Equations

The mixture material is treated as the homogeneous mass of compound ingredients

which includes polymer, filler, oil, antioxidants etc. Both the working fluid rubber

and air are considered to be incompressible, and are modeled using incompressible

Navier-Stokes equations. The conservation of mass is given by;

∇·V =0 (3.1)

V = uî + v ĵ + wk̂ denotes the velocity vector, and the momentum equation:

DV
ρ = −∇p + ∇ · T + ρg (3.2)
Dt
D
where, Dt represents the material derivative, p is pressure, ρ is the fluid density, g is

the gravitational acceleration. The stress tensor T is given by,

T = 2µD (3.3)
32
where, D is the strain rate tensor. The viscosity, µ, which is the function of local

shear rate, is approximated using the Carreau model as [42] :

µ = µ∞ + (µ0 − µ∞ )[(1 + (γ˙2 )λ2 ](n−1)/2 (3.4)

where, µ∞ is the infinite shear viscosity, µ0 is the zero shear viscosity, γ̇ is the local

shear rate, λ is the relaxation time, and n is the power-law index. The isothermal

assumption of the system eliminates the necessity of temperature dependent viscosity.

3.2 Multi-phase model

A phase can be defined as the realizable class of material that has inertial response to

and from the flow on which it is embedded. The air and rubber phases are involved

in complex interaction because of huge differences in material properties between the

phases. Flow behavior in partially filled chamber is better understood if the interface

between rubber and air is captured accurately.

Among the various mathematical models, the Euler-Euler approach, treats

different phases as interpenetrating continua. Phase volume fraction is a continuous

function of space and time and the sum of two phases is always equal to one. Eu-

lerian based volume of fluid (VOF) method with higher order discretization scheme

generates a sharp interface between the phases. In the absence of any source/sink

terms for the volume fraction evolution, the formulation of VOF is given by:

∂Cm
+ V · ∇Cm = 0 (3.5)
∂t

where Cm is the volume fraction of mth phase


33
Mixing quantities are calculated based on the particle statistics and for this

purpose, massless particles are injected into the computational domain and their

positions are tracked throughout the simulation. The trajectory can be calculated

using the interpolation scheme from the velocity field given by:

dx
= up (3.6)
dt

where up is the particle velocity.

34
CHAPTER IV

COMPUTATIONAL DETAILS

2D and 3D numerical simulation of rubber mixing in a partially filled internal mixer

are presented here. The working domain is taken from a US patent [7] for 2-wing

internal mixer as shown in the Figure 4.1.

4.1 2D rotor geometry

Although a 2D investigation does not consider axial movement, it provides the knowl-

edge regarding major flow phenomena and mixing mechanisms. The operating pa-

rameters and boundary conditions are taken in such a way that the computational

setup is very close to the actual physical problem. The geometry and corresponding

mesh for solving the discretized equation for 2D simulation is shown in Figure 4.2.

This geometry is a cross section of an actual 3D rotor-equipped chamber, taken in

the middle of chamber, perpendicular to the rotor axis. The phase angle between the

rotor wings is not symmetrical because two wings originate from the two opposite

ends of the shaft. The mesh consists of 200,000, triangular and quadrilateral cells.

The region between rotor tip and chamber wall, which is also the most active region

of shear for tangential mixer, is resolved using 12 layers to capture the sharp gra-

dients in that region. Grid independent studies were carried out to assess the grid

35
Figure 4.1: Geometric views of the rotor

36
(a) (b)

(c)

Figure 4.2: (a) Geometry (b) Computational mesh and (c) Grid resolution in the
clearance region between the rotor tip and chamber wall, used for 2D simulations

requirements for such a problem, and it was found that the mesh with 200,000 cells

and 2500 particles were sufficient to resolve the major flow properties.

4.2 3D rotor geometry

The full 3D geometry of a two wing tangential rotor has 880 mm in length and radius

of 278 mm. The two rotor wings on each shaft are originating from two opposite ends

and rotor wings are twisted, which in turn generates the motion in axial direction in

addition to tangential component. The mesh corresponding to the geometry is shown


37
Table 4.1: Non-Newtonian Parameters

SN Parameter Value Unit

1 Zero shear viscosity 100, 000 Pa-s

2 Infinite shear viscosity 1 Pa-s

3 Relaxation time 10 s

4 Power law index 0.4 -

in Figure 4.3 and based on the grid independence study it is observed that the grid

number of 1 million and particle number of 5200 was sufficient to predict the flow

behavior in the chamber.

4.3 Operating parameters

The mixing chamber is a closed domain assuming that the ram is resting at its bottom

position. Rotors are counter-rotating with left rotor rotating in clockwise direction

as shown in the Figure 4.2 (a). The working material, rubber, was assumed to be

homogeneous fluid with the properties defined by Bird-Carreau model given in the

Table 4.1.

4.4 Sliding mesh technique

The numerical simulation was conducted using a commercial CFD software. The mo-

tion of rotors can be modeled using various mesh displacement methods, of which, the

38
(a)

(b) (c)

Figure 4.3: Mesh details across (a) x-z plane (b) x-y plane and (c) Rotor to chamber
wall clearance used for 3D simulations

39
sliding mesh technique is capable of simulating transient motion without remeshing.

Multiple cell zones are connected with each other through non-conformal interfaces

which can move freely on the pre-assigned path. As the mesh motion is updated in

time, the non-conformal interfaces are updated as well, to reflect the new positions of

each moving zone. Interfaces are such that, the sliding surfaces are always in contact

with each other.

40
CHAPTER V

RESULTS AND DISCUSSION

A series of numerical simulations for various fill factors and geometric conditions are

presented in this thesis. Isothermal calculations are conducted by solving the incom-

pressible Navier-Stokes equations for a homogeneous material that has properties

very similar to a rubber compound. Flow mechanism and mixing performance have

been evaluated using flow contours and other statistical quantities.

5.1 Numerical simulation in 2D geometry

The 2D simulations were conducted for fully-filled (100%) and partially-filled (75%)

in a rotor-equipped chamber. For the partially filled condition, rubber material was

assumed to be at the bottom of the mixing chamber at the beginning of the simulation

as shown in Figure 5.2.

5.1.1 Overall flow mechanism

Figure 5.1 shows the streamlines corresponding to the fully filled simulation in the

mixing chamber. It shows that the material is mostly rotating around the center

of the shaft with rotor wings generating the material movement. Streamlines are

continuously changing with the phase angle between the wings. Flow on the leading

41
edge of the rotor is the summation of pressure flow and drag flow, however, there is

the significant amount of leakage flow in the tip region as shown in the streamlines.

5.1.1.1 Phase volume fraction

For a partially filled simulation, which is a combination of air and rubber, the contour

of volume fraction tracks the interface for rubber and air. Figure 5.2 shows the contour

of volume fraction of rubber for 75% fill factor simulation. For better visualization,

air has been excluded in the plot using the clip function, and hence all the void space

seen in Figure 5.2 represents air. It can also be seen that the rubber is resting at

the bottom at the beginning of the simulation and the transient interface is formed

as the simulation proceeds with time. Also, the region ahead of rotor tip is always

occupied with the rubber material which creates shear necessary for mixing and the

region behind rotor tip is void which facilitates the random motion of rubber.

5.1.1.2 Velocity plot

Instantaneous velocity contours for the two different fill factors at the end of 10

revolutions are shown in Figure 5.3. In the 75% case, contours are shown only for the

rubber phase, with the air phase being made transparent in the visualization. The

maximum velocity in the domain for both the cases is consistent and equivalent to

the maximum tangential velocity of the domain. The presence of air in the mixture is

what makes the two cases significantly different from each other. For the fully-filled

case, rubber is present everywhere in the domain leading to a more streamlined flow.

This is not the case with the partially-filled scenario. In fact, previous experimental

42
(a)

(b)

Figure 5.1: Streamlines plot at the end of 10 revolutions (a) on the domain (b) at
the rotor tip

43
(a)

(b) (c)

Figure 5.2: Volume fraction of rubber (a) at the beginning (b) at the end of 5 revo-
lutions (c) at the end of 10 revolutions

44
(a) (b)

Figure 5.3: Velocity magnitude for (a) 75% (b) 100% fill-factor simulations at the
end of 10 revolutions

studies have indicated that the leading edge of the rotor wing is constantly filled with

mixing material, only if the fill factor is above 0.7. It could also be seen from the

simulation results of the 75% fill factor, that the leading edge always carried mixing

material along with it. In addition, there is a significant amount of void space in the

trailing edge of the rotor wing which allows the rubber material to move freely at

the time of squeezing flow. This kind of regular air voids generates a more random

nature of flow, which potentially facilitates better dispersive and distributive mixing,

in comparison to a fully-filled simulation [19].

5.1.1.3 Pressure plot

Pressure plots for two fill factors are shown in the Figure 5.4. The pressure is maxi-

mum at the leading edge of the rotors, which provides the force necessary for material

movement. In addition, the value of maximum pressure is lower for the partially filled

case, which is because of the presence of the void in the mixing chamber. The pressure

is symmetric about the vertical axis for the fully filled case. However the pressure

45
(a) (b)

Figure 5.4: Pressure plot for (a) 75% (b) 100% fill-factor simulations at the end of
10 revolutions

values on the leading edge are different on each wing of the rotor. The trailing edge

of the rotor has very low-pressure value than that of leading edge, which creates the

adverse pressure gradient leading to the leakage flow also shown in streamlines.

5.1.2 Dispersive mixing

The detailed study of flow properties can be further extended to the study of disper-

sive mixing characteristics which describes the ability of the mixer to breakdown the

filler agglomerates. Injection of massless particles into the simulation domain in the

form of different shapes facilitate the calculation of different statistics for the analysis

of dispersive mixing. The initial injection can be in the form of a cluster or along a

line, based on the kind of parameter to be analyzed.

5.1.2.1 Mixing index and Shear stress

Figure 5.5 shows a histogram of the mixing index calculated from all the computa-

tional cells, for both the fill-factor simulations. Better dispersive mixing is indicated

46
Figure 5.5: Probability distribution in %, of the mixing index calculated from all the
computational nodes in the 75% and 100% fill-factor simulations.

47
by a larger percentage of mixing index values ≥ 0.5. So from Figure 5.5, it is clear

that the fully-filled case has a higher percentage of mixing index values from 0.5 to 1

compared to the partially-filled case. On the other hand, for the mixing index range

of 0-0.5, the partially-filled simulation has a higher percentage. This seems to indi-

cate that a fully-filled chamber exhibits better dispersive mixing characteristics than

a partially-filled chamber, which is contradictory to what has been shown previously

[19]. To explain why the mixing index is not a good indicator of dispersive mixing, it

should be noted that, mixing index is calculated at each computational node and does

not reflect the level of stress or strain. A high value of mixing index means that the

agglomerate going through that region is likely to be broken up assuming the stress

is high enough. However, the mixing index itself does not indicate the probability of

an agglomerate going through that region. The region with high mixing index could

have very little contribution to the dispersive mixing if the probability of agglomerate

going through that region is very low. This will be further confirmed by cumulative

probability of maximum shear stress presented later.

Furthermore, the average value of mixing index can be calculated by weighing

the corresponding parameter for each element by the surface area of the element itself

[23]. The average value of λM Z for two fill factors is shown with respect to the rotor

position in Figure 5.6. The values do not change with time, at all positions, for the

fully-filled case, whereas that is not the case for the partially filled case. At 75%

fill, the average mixing index values at all positions increase from around 0.43 to 0.5

from 2 to 10 revolutions. This is because the overall flow field generated by the rotor

48
Figure 5.6: Average mixing index for different rotor positions

motion in the fully-filled case does not change from one revolution to the other, but

in the partially-filled simulation, the void spaces create a more random motion of the

rubber material, which in turn leads to changes in the average mixing index with

time. In addition, it should be noted that the average value of λM Z at the end of 10

revolutions is around 0.5 in both cases indicates that the majority of the mixing is

due to shearing.

49
The mixing index alone is not sufficient to determine whether the dispersive

mixing is efficient. For instance, has the particle experienced enough shear stress to

break down into smaller agglomerate? More specifically, it has to be considered in

conjunction with the shear stress. Since shear forces are the driving mechanism for

the breakdown of agglomerates, it is important to know the shear stress experienced

by each particle for the entire mixing cycle.

Figure 5.7 shows the cumulative distribution of the maximum shear stress

experienced by each particle, after 5 revolutions and 10 revolutions, for both the 75%

and 100% fill factor cases. It can be seen that for both fill factors, the curve shifts

to the right as the simulation time increases indicating that the amount of material

experiencing larger values of shear stress increases with time. It is also clear that this

shift is higher for the partially-filled case compared to the fully-filled case. This leads

to the general conclusion that dispersive mixing is better in a 75% filled chamber

than a fully-filled chamber for the same duration of mixing. Analysis of this quantity

usually has to be accompanied with a knowledge of the critical shear stress needed to

break down the rubber compound agglomerates. However, in this study, the rubber

compound is treated as bulk homogeneous material, when in reality it is made of

multiple components such as carbon black, sulphur, silica, etc., and hence, the exact

value of the critical shear stress is unknown. Nevertheless, for any value of critical

shear stress, 75% fill-factor would be more effective if the mixing is compared for

the same period of time. For instance, if the critical shear stress was 75 kPa, at

the end of 10 revolutions, around 30% of the material in the 75% fill-factor case has

50
Figure 5.7: Maximum shear stress experienced by each particle for an entire mixing
cycle, i.e. 10 revolutions

experienced a shear stress > 75 kPa, whereas for the fully-filled case, barely 10% of

material experienced shear stress greater than critical. In other words, if we assume

75kPa as the critical shear stress for a particular set of filler material, a 75% fill factor

is almost three times as effective in dispersive mixing as a fully-filled scenario.

51
5.1.3 Distributive mixing

A good spatial distribution of the broken agglomerates is the representation of good

mixing in the context of distributive characteristics. The dynamics of distributive

mixing is quantified by considering various other parameters such as length of stretch

and cluster distribution index. These quantities can be numerically calculated by

tracking the motion of a set of massless/fictitious particles in the domain. Although

they are fictitious, their locations are entirely determined by the velocity field, specifi-

cally by integrating the latter over time. Figure 5.8 illustrates a typical instantaneous

particle distribution at the time of injection and at various revolutions for both fully

filled and 75% fill-factor case.

5.1.3.1 Mean length of stretch

The mean length of stretch is the ratio of distances between any particle pairs at

the beginning of the simulation and at the time, t, respectively. For any type of

particle injection in the form of cluster, it is expected that the process of distributive

mixing leads to an increase in the value of the mean length of stretch. Figure 5.9

shows the temporal evolution of mean length of stretch for 10 revolutions in both fill

factors. As the initial particle injection was a square cluster of particles in the region

between two rotors, the mean length of stretch keeps on increasing as the cluster

breaks up. Higher values of mean length of stretch imply a higher dissemination

of the initial cluster of particles, thereby indicating better distributive mixing. It

can be seen from the figure that the mean length of stretch increases for both cases

52
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.8: Particle distribution for 75% fill-factor simulation (a) at the time of
particle injection (time t = 0) (c) after 2 revolutions (e) after 5 revolutions (g) after
10 revolutions and for the fully filled (b) at the time of particle injection (time t = 0)
(d) after 2 revolutions (f) after 5 revolutions (h) after 10 revolutions

53
Figure 5.9: Mean length of stretch over 10 revolutions for the 75% and 100% fill-factor
simulations

54
and continues to increase with oscillations. These oscillations are due to stretching,

folding and reorientation phenomena in a dynamic mixer. It should also be noted

that the maxima and minima of these curves are separated by exactly one revolution.

Among the two fill factors, the mean length of stretch for 75% fill factor is consistently

higher throughout the simulation reflecting better distributive mixing.

5.1.3.2 Cluster distribution index

The deviation of particle distribution from ideal distribution is represented by cluster

distribution index. Figure 5.10 shows probability distribution of particle pairwise

distances, i.e. c(r), for two fill factors at the end of 10 revolutions along with the

ideal distribution. It can be seen that the curve of ideal distribution has two peaks, at

approximately 0.5 m and 0.9 m, due to the nature of rotor and geometry. The fully-

filled curve has a high peak at low particle pairwise distances of approximately 0.1

m, indicating that the particles are still in clusters and not very well distributed. On

the other hand, the 75% curve is more aligned with the ideal distribution, thereby

indicating a better distribution. In order to specifically quantify the deviation of

the distribution of particle pairwise distances from the ideal distribution, the cluster

distribution index over the time is shown in Figure 5.11. The value is higher at the

beginning of the simulation, but decreases subsequently with time, as the particle

distribution approaches the ideal distribution. It can be seen from the figure that the

75% fill factor not only reaches the lowest value of cluster distribution index faster

55
Figure 5.10: Probability distribution of the particle pairwise distances, c(r), calcu-
lated at the end of mixing cycle, i.e. 10 revolutions

but also maintains the lowest value throughout the simulation indicating that it has

better distributive mixing.

5.1.3.3 Segregation scale

Particles are color coded into lower and upper half as shown in the Figure 5.12 at the

end of 10 revolutions. The simulation was conducted for 10 additional revolutions. As

is clear from the figure, there are still some segregated regions in the mixer indicated

by the separation of colored particles. Both fill factors exhibit the breakage of initial
56
Figure 5.11: Cluster distribution index over 10 revolutions for the 75% and 100%
fill-factor simulations

57
(a) (b)

(c) (d)

Figure 5.12: Particles coded with two colors for 75% fill-factor simulation (a) at the
end of 10 revolutions (c) at the end of 20 revolutions and for the fully filled (b) at
the end of 10 revolutions (d) at the end of 20 revolutions

58
interface between the two types of particle indicated by red and blue color. However,

the final mixing result at 20 revs is significantly different for the two fill factors in

the region between the rotors. There are two big blocks of unmixed red particles on

each chamber for fully filled case as shown in Figure 5.12 (d). In the 75% fill factor

case, the particles are very well mixed and the segregation is very low.

The evolution of segregation scale with time for two fill factors are shown in

the Figure 5.13. Initially, particles are separated into lower and upper half leading

to a high value of segregation scale. For the first 2 revolutions, the curve is identical,

however, it decreases rapidly for 75% fill factor as the particles move with the more

random flow pattern generated in partially filled mixing chamber. The segregation

scale decreases with each revolution even for the fully filled case because the particles

in the narrow clearance region are able to mix although there is no any mechanism

for random flow. Figure 5.13 concludes that distributive mixing for the partially filled

is better as it maintains the lowest value of segregation scale throughout the entire

mixing cycle.

5.2 Numerical simulation in 3D geometry

3D numerical simulation of rubber mixing in a 2-wing rotor-equipped geometry are

presented here. The fill factors considered are 45% ,60%, 75%, 90% and 100%. Mixing

performance of each corresponding fill factors has been compared between them to

determine the most efficient fill factor in terms of dispersive and distributive mixing.

59
Figure 5.13: Cluster distribution index over 10 revolutions for the 75% and 100%
fill-factor simulations

60
5.2.1 Overall flow mechanism

The flow behavior for 3-D geometry can be better described using cross sections in

the radial and axial directions. Three different cross sections are considered in the

axial direction at (1) 1/4, (2) 1/2 and (3) 3/4 from one of the rotor end. Similarly,

the axial cross-section is taken along the centerline of the rotor axis.

5.2.1.1 Velocity magnitude and Vectors

The velocity magnitude and vectors for the different cross sections of a 2-wing rotor

geometry are shown in Figures 5.14 - 5.18. A volume fraction cutoff criteria of 0.5 is

employed so that one can visualize the contours only in the region occupied by rubber.

For all fill factors, with the left rotor rotating in counter-clockwise direction, the flow

mechanism is the combination of pressure flow and drag flow. An interesting feature

in the flow behavior is the availability of rubber material on the leading edge of the

rotor. For the 45% fill factor, it can be seen that most of the material is attached

to the rotor shaft leading to plug flow. The maximum velocity is consistent with

the tangential velocity at the rotor tip (0.6 m/s). The vector diagram shows that

majority of the flow is following the rotor motion in a circular path around the center

of rotation. However, there is a material exchange between the chamber. At different

cross sections as shown in the figure, the direction of material movement varies along

the axial direction as the relative position of rotor wing changes. For the cross section

at 1/4 of the rotor height, the left wing is pushing mixture material into the right

wing at that instant (at the end of 10 revolutions). The rotor geometry itself is acting

61
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.14: Velocity contours and vectors across the radial cross section at (a)-(b)
one-quarter of rotor height (c)-(d) at half of the rotor height (e)-(f) three-quarter of
rotor height (g)-(h) across axial cross section for 45% fill factor

62
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.15: Velocity contours and vectors across the radial cross section at (a)-(b)
one-quarter of rotor height (c)-(d) at half of the rotor height (e)-(f) three-quarter of
rotor height (g)-(h) across axial cross section for 60% fill factor
63
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.16: Velocity contours and vectors across the radial cross section at (a)-(b)
one-quarter of rotor height (c)-(d) at half of the rotor height (e)-(f) three-quarter of
rotor height (g)-(h) across axial cross section for 75% fill factor
64
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.17: Velocity contours and vectors across the radial cross section at (a)-(b)
one-quarter of rotor height (c)-(d) at half of the rotor height (e)-(f) three-quarter of
rotor height (g)-(h) across axial cross section for 90% fill factor
65
(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 5.18: Velocity contours and vectors across the radial cross section at (a)-(b)
one-quarter of rotor height (c)-(d) at half of the rotor height (e)-(f) three-quarter of
rotor height (g)-(h) across axial cross section for fully filled simulation
66
as a single wing rotor because the wing originating from the opposite end of the shaft

terminates ahead of this plane. The plane taken at the 1/2 of axial dimension shows

two rotors being the mirror image of each other. However, the wings of each rotor are

not symmetrically placed. The material motion is mostly pushed downward because

of the rotation direction and the rotor geometry. A significant amount of material is

squeezed in the region between the two rotor tips as they rotate. The flow properties

at 3/4 of rotor height are opposite of the properties at 1/4 of the rotor height because

of the relative rotor orientation. The material movement is from the right chamber

to the left chamber in the region between the two rotors.

Similarly, contours of velocity magnitude and velocity vectors are considered

on the axial cross-section along the centerline to study the material movement in

that direction, as it is generated by the unique rotor design and is very critical for

the overall mixing. Very interestingly an unique type of flow behavior is generated,

as it can seen that the flow on either side of the rotor wing is parallel to the direction

of the rotor axis, but opposite to each other. In the region between the two rotor

wings, the material gets squeezed around the center of the chamber, because the two

inner wings from two rotors force the material in opposite direction. The direction

of the force is not parallel to the rotor axis but somehow inclined. It was also found

that the direction flips twice on every revolution showing the material moving back

and forth in an axial direction.

As the fill factor increases, the amount of raw material on the leading edge

of the rotor wing increases. The experimental study has shown that the leading

67
edge of rotor wing is always occupied by mixture material if the fill factor is greater

than 0.7[19]. The simulation results proves that argument with very little amount of

rubber material in the leading edge of rotor wing for 60% fill factor. As it can be

seen for the 75% fill factor, the amount of material is higher and keeps increasing for

higher fill factor. The overall flow direction and velocity magnitude follow a similar

pattern for all the fill factors. For higher fill factors like 90% and 100%, the flow

pattern is more streamlined as shown in the Figure 5.17 - 5.18.

5.2.2 Dispersive mixing

Based on the finding of two dimensional study, shear stress is very critical for assesing

the dispersive mixing characteristics. The three dimensional study has been focused

on the finding the maximum shear stress which will be described in detail here.

5.2.2.1 Maximum shear stress

The cumulative distribution function of the maximum shear stress experienced by

a particle over its entire mixing cycle is shown in the Figure 5.19. The plot shows

the probability of filler material passing at least once through the given critical shear

stress. The shear stress plot can be divided into three different regions. In the region

with low critical shear stress (<50,000 Pa), higher fill factor seems to do better in

terms of dispersive mixing as almost all of the corresponding filler particles are broken.

In the mid-critical shear region (<100,000 Pa), all the fill factors seem to do better as

most of the particles are broken for all of the fill factors. However, for higher critical

shear stress (>150,000), which corresponds to the particle with the smaller radius,

68
Figure 5.19: Maximum shear stress over 10 revolutions for the five different fill factors

the 75% fill factor seems to display the best dispersive mixing characteristics among

all of the fill factors.

This can further be confirmed with the data presented in the Table 5.1.

Considering a filler particle having critical shear stress 50,000 Pa, One out of five

particles will not be broken down for 45% fill factor. However, that number improves

for the same value of critical shear stress if the fill factor is increased. Very small

portion of the particle remains intact for 60% and 75% fill factor, and all the particles

will be broken down in the 90% and fully filled case. As the diameter of filler particles

decreases, the critical shear stress value increases. Considering the relatively smaller

69
Table 5.1: Percentage of particle broken down for various critical shear stress

Shear stress 50kPa 100kPa 150kPa

45% 80.1 28.8 5.9

60% 96.2 24.27 6.27

75% 94.82 27.9 9.07

90% 100 26.53 3.57

100% 100 31.55 3.76

filler particle with critical shear stress 150,000 Pa, it is found that more than 9 percent

of particles experience the shear stress higher than that for 75% fill factor. This value

is much lower for higher fill factors in comparison to other partially filled cases as

shown in Table 5.1.

5.2.3 Distributive mixing

Similar to the approached used in the two dimensional study, distributive mixing

properties have been calculated based on the statistical quantities.

5.2.3.1 Mean length of stretch

The higher the value of the mean length of stretch, the more effective is the distribu-

tive mixing. Figure 5.21 shows the temporal evolution of the mean length of stretch

for the different fill factors. Depending on the initial injection of the particles, which

in the current study is carried out along a line, it is expected that the mean length of

stretch will have the lowest value initially since the particles are close to each other

70
(a) (b)

Figure 5.20: Particle distribution (a) at the time of particle injection (time t = 0)
and (b) at the end of 10 revolutions for three-dimensional geometry

and will keep increasing as the simulation proceeds. All the fill factor cases display

this kind of behavior and in addition, oscillations are present in their evolution, which

represents the stretching and unstretching of the material. The amplitude of these

oscillations decreases with time for all the cases which means particle distribution is

getting more uniform. If the simulation is performed for very long time, the mean

length of stretch will reach the plateau indicating the maximum stretch possible. The

Figure 5.21 shows that, for all the fill factors, there are major oscillations at the be-

ginning of the simulation which disappear as the simulation proceeds. 75% fill factor

case has the most steady and the highest value after 3 revolutions and furthermore,

maintains it throughout the simulation. However, other fill factors have more oscil-

lations especially in the beginning, although all of them are converging towards the

same value at the end of the simulation. It should be noted that, it is difficult to infer

71
Figure 5.21: Mean length of stretch over 10 revolutions for the five different fill factors

a ranking with regard to the effectiveness of distributive mixing from the prediction

of mean length of stretch due to the presence of oscillation.

5.2.3.2 Cluster distribution index

Cluster distribution index shows the dissemination of initial cluster of the particle.

Particles injected at the beginning are very close to each other which resembles the

bulk of agglomerate. For this simulation, particles are injected in the region between

the two rotors. Although studies have shown that the cluster distribution differs

72
based on the initial cluster position [29]. Figure 5.22 shows the evolution of cluster

distribution index for all five fill factors considered in the simulation. It is expected

that the partially filled chamber will perform better when it comes to distributive mix-

ing because of the randomness of flow, however, the cluster may remain unaffected

under those conditions. Figure 5.22 shows that 45% fill factor has the highest cluster

distribution index among all demonstrating a possibility of plug flow that does not

break down the initial cluster and rather carries along with its rotational flow. This

is followed by the fully filled case in terms of cluster distribution index which further

affirms the observation that the streamlined flow is bad for mixing. On the other

hand, 75%, 60%, and 90% show fairly better cluster distribution index in comparison

to the other two fill factors. The 75% fill factor case has the lowest value of cluster

distribution index among all of the fill factors and that value is attained in a signifi-

cantly lower time. Furthermore, this value is maintained throughout the simulation,

thus showing that it is the best fill factor among all based on this parameter.

5.2.3.3 Segregation scale

Segregation scale is the measurement of the size of the region of homogeneous con-

centration. The initial particles are color coded similar to the result presented for the

2-D study. Figure 5.23 shows the evolution of segregation scale for the different fill

factors with time. Firstly, the initial segregation scale shown in the Figure 5.23 was

based on the particle position, where one half of the numbers of particles are assigned

a concentration of 0 and the other half 1. Secondly, lower segregation scale values

73
Figure 5.22: Cluster distribution index over 10 revolutions for the 5 different fill
factors

74
imply better distributive mixing. All the temporal evolutions of the segregation scale

display oscillations initially, but within 4 revolutions, these oscillations are damped

and the values continuously decrease till the end of the simulation. From the figure,

it can be concluded that the 90% and 100% filled cases, which have the highest segre-

gation scale, almost throughout the simulation, possibly have the worst distributive

mixing characteristics although their ranking is different from the expectation that

fully filled would have the worst distributive mixing characteristics. Among the rest

of cases, it can be seen that 45% has a smaller value towards the beginning of the

simulation, however, 60% and 75% surpass that value after 5 revolutions. All the

fill factors exhibit decreasing segregation scale over the time, which represents the

blocks of homogeneous concentration being crushed into smaller ones. At the end of

the simulation, both 75% and 60% fill factor have the smallest value of segregation

scale displaying their better distributive characteristics.

75
Figure 5.23: Segregation scale over 10 revolutions for the 5 different fill factors

76
CHAPTER VI

CONCLUSIONS

Two-dimensional numerical simulations of 2-wing rotor mixing of rubber compounds

are conducted using a commercial CFD code for both partially-filled and fully-filled

mixing chambers, which correspond to 75% and 100% fill-factors, respectively. The

primary objective of this study was to compare the dispersive and distributive mixing

characteristics between the fully-filled and partially-filled cases. Dispersive and dis-

tributive mixing characteristics are critical in evaluating and improving the mixing

efficiency and consequently the throughput of mixers used in manufacturing, specifi-

cally in tire manufacturing processes. Firstly, flow patterns were presented for both

the cases in order to understand the differences with respect to material movement.

Mixing index, which has been commonly used in previous literature to assess the dis-

persive mixing, was investigated for both the simulations here and it was found that

this quantity was unable to predict the ranking of dispersive mixing performance. An

explanation for this behavior is that mixing index does not reflect the magnitude of

stress or the probability of particle (representing filler agglomerates) passing through

the local region corresponding to the mixing index. Both these factors are important

for dispersion assessment.

77
Massless particles were injected into the domain and their positions were

tracked by employing an integration of the velocity fields. The positions of these par-

ticles were used to calculate various statistics related to the quantification of mixing.

The history of the maximum shear stress experienced by each particle was analyzed

by considering the cumulative distribution of the same. The cumulative distribution

function of the maximum shear stress showed that a larger percentage of the material

in partially-filled chambers experienced high shear stresses (possibly larger than crit-

ical) in comparison to fully-filled, thereby pointing to the superior dispersive mixing

capabilities of the former.

Furthermore, distributive mixing was assessed using length of stretch, PDFs

of pairwise distances and cluster distribution index. The temporal evolution of length

of stretch showed that the 75% fill-factor had higher mean length of stretch through-

out the simulation than the 100% fillfactor. PDFs of pairwise distances demonstrated

the closer alignment of the PDF of the 75% case with the ideal distribution. And

finally, the cluster distribution index of the 75% fill-factor case decreased the fastest

and was lesser in magnitude throughout the simulation. All these variables showed

consistently that 75% fill-factor had better distributive mixing that fully-filled. This

kind of study helps to understand the flow properties for various fill factor and even-

tually can be used for finding the optimum fill factor.

Furthermore, 3D CFD simulations of rubber compound mixing in a counter-

rotating rotor-equipped mixing chamber were conducted at five different fill factors:

45%, 60%, 75%,90% and 100% (or fully-filled), to understand the effect of fill factor on

78
dispersive and distributive mixing characteristics. Several quantities were calculated

and presented here for this purpose. Velocity vectors across various planes are shown

in order to analyze material movement, which directly affects mixing. To assess

dispersive mixing, cumulative probability density function of maximum shear stress

was calculated. In addition, distributive mixing characteristics are assessed with

the help of parameters such as segregation scale, mean length of stretch and cluster

distribution index.

Instantaneous velocity vectors that were representative of the overall flow

patterns for the different fill factor cases indicated that the 45% fill factor resulted

nearly in plug flow, whereas the rest of the fill factors produced velocity gradients

between the rotor and the chamber wall. Also, random air spaces that were visible

in the 60% and 75% cases throughout the chamber, which favor a more disorderly

motion of the material, and hence better mixing, ceased to exist in the 90% and

100% cases. Dispersive mixing was evaluated based on the maximum shear stress

experienced by particle over the entire mixing cycle and it was found that the dis-

persion capacity differs with the value of critical shear stress. For very low critical

shear stress (corresponding to higher agglomerate diameter), higher fill factor per-

form better. However, 75% fill factor performed best when the critical shear value is

higher than 150 kPa. With regard to distributive mixing, among all the fill factors,

the 75% case showed the fastest decrease in segregation scale and the highest length

of stretch. The knowledge gained from such CFD simulations of partially-filled and

fully-filled mixing chambers with a focus on the importance of fill factor, specifically

79
in terms of determining the dispersive and distributive mixing efficiency, can help in

further improving the design of the manufacturing process in the tire industry.

80
BIBLIOGRAPHY

[1] Edward L Paul, Victor A Atiemo-Obeng, and Suzanne M Kresta. Handbook of


industrial mixing: science and practice. John Wiley & Sons, 2004.

[2] Nobuyuki Nakajima. Science and practice of rubber mixing. iSmithers Rapra
Publishing, 2000.

[3] Sadhan K De and Jim R White. Rubber technologist’s handbook, volume 1.


iSmithers Rapra Publishing, 2001.

[4] John M Funt. Mixing of rubber. Smithers Rapra Technology, Limited, 2009.

[5] Anil K Bhowmick. Current topics in elastomers research. CRC press, 2008.

[6] Chris Rauwendaal. Mixing in polymer processing, volume 23. CRC Press, 1991.

[7] Narku O Nortey. Two-wing non-intermeshing rotors of increased performance for


use in internal batch mixing machines, December 22 1987. US Patent 4,714,350.

[8] Frank J Borzenski. Overview of variables affecting batch mixing in a tangential


mixer. Rubber World, 6(219):8, 1999.

[9] Andreas Limper. Mixing of Rubber Compounds. Carl Hanser Verlag GmbH Co
KG, 2012.

[10] Richard F Grossman. The mixing of rubber. Springer Science & Business Media,
1997.

[11] Peter R Wood. Mixing of vulcanisable rubbers and thermoplastic elastomers,


volume 10. iSmithers Rapra Publishing, 2004.

[12] Ica Manas-Zloczower. Mixing and compounding of polymers: theory and practice.
Carl Hanser Verlag GmbH Co KG, 2012.

[13] Juergen W Pohl. Trends in rubber mixing. Rubber world, 217(6), 1998.

81
[14] PK Freakley. Rubber processing and product organisation, 1985.

[15] Kyonsuku Min and James L White. Flow visualization of the motions of elas-
tomers and molten plastics in an internal mixer. Rubber chemistry and technol-
ogy, 58(5):1024–1037, 1985.

[16] Kyonsuku Min and James L White. Flow visualization investigations of the
addition of carbon black and oil to elastomers in an internal mixer. Rubber
chemistry and technology, 60(2):361–380, 1987.

[17] Ryszard Brzoskowski, Kazuhisa Kubota, Kiho Chung, James L White, Fred-
erick C Weissert, Nobuyuki Nakajima, and Kyonsuku Min. Experimental and
theoretical study of the flow characteristics of rubber compounds in an extruder
screw. International Polymer Processing, 1(3):130–136, 1987.

[18] James M MacKelvey. Polymer processing. Wiley, 1962.

[19] PK Freakley and WY Wan Idris. Visualization of flow during the processing of
rubber in an internal mixer. Rubber Chemistry and Technology, 52(1):134–145,
1979.

[20] PK Freakley and SR Patel. Internal mixing: a practical investigation of the flow
and temperature profiles during a mixing cycle. Rubber chemistry and technology,
58(4):751–773, 1985.

[21] PK Freakley and SR Patel. Internal mixing: A practical investigation of the


influence of intermeshing rotor configuration and operating variables on mixing
characteristics and flow dynamics. Polymer Engineering & Science, 27(18):1358–
1370, 1987.

[22] Jing-Jy Cheng and Ica Manas-Zloczower. Hydrodynamic analysis of a banbury


mixer 2-d flow simulations for the entire mixing chamber. Polymer Engineering
& Science, 29(15):1059–1065, 1989.

[23] JJ Cheng and I Manas-Zloczower. Flow field characterization in a banbury mixer.


International Polymer Processing, 5(3):178–183, 1990.

[24] H-H Yang and I Manas-Zloczower. 3d flow field analysis of a banbury mixer.
International Polymer Processing, 7(3):195–203, 1992.

[25] Ica Manas-Zloczower. Analysis of mixing in polymer processing equipment. Rhe-


ology Bulletin, 66(1):5–8, 1997.

82
[26] JK Kim, JL White, K Min, and W Szydlowski. Simulation of flow and mixing
in an internal mixer. International Polymer Processing, 4(1):9–15, 1989.

[27] JK Kim and JL White. Non-newtonian and non-isothermal modeling of 3d-flow


in an internal mixer. International Polymer Processing, 6(2):103–110, 1991.

[28] Robin K Connelly and Jozef L Kokini. The effect of shear thinning and dif-
ferential viscoelasticity on mixing in a model 2d mixer as determined using fem
with particle tracking. Journal of Non-Newtonian Fluid Mechanics, 123(1):1–17,
2004.

[29] Robin K Connelly and Jozef L Kokini. Examination of the mixing ability of
single and twin screw mixers using 2d finite element method simulation with
particle tracking. Journal of food engineering, 79(3):956–969, 2007.

[30] Maureen L Rathod and Jozef L Kokini. Effect of mixer geometry and operating
conditions on mixing efficiency of a non-newtonian fluid in a twin screw mixer.
Journal of Food Engineering, 118(3):256–265, 2013.

[31] V Nassehi and MHR Ghoreishy. Simulation of free surface flow in partially filled
internal mixers. International Polymer Processing, 12(4):346–353, 1997.

[32] V Nassehi and MHR Ghoreishy. Modeling of mixing in internal mixers with long
blade tips. Advances in Polymer technology, 20(2):132–145, 2001.

[33] Jinpeng Liu, Fanzhu Li, Liqun Zhang, and Haibo Yang. Numerical simulation of
flow of rubber compounds in partially filled internal mixer. Journal of Applied
Polymer Science, 132(35), 2015.

[34] Ica Manas-Zloczower and Hongfei Cheng. Analysis of mixing efficiency in poly-
mer processing equipment. In Macromolecular symposia, volume 112, pages
77–84. Wiley Online Library, 1996.

[35] Véronique Collin, Edith Peuvrel-Disdier, B Alsteens, Vincent Legat, T Aval-


osse, S Otto, and HM Metwally. Numerical and experimental study of disper-
sive mixing of agglomerates. In Society of Plastics Engineers Annual Technical
Conference 2006, ANTEC 2006, volume 2, pages Pages–908. Society of Plastics
Engineers, 2006.

[36] Hongfei Cheng and I Manas-Zloczower. Study of mixing efficiency in knead-


ing discs of co-rotating twin-screw extruders. Polymer Engineering & Science,
37(6):1082–1090, 1997.

83
[37] Julio M Ottino. The kinematics of mixing: stretching, chaos, and transport,
volume 3. Cambridge university press, 1989.

[38] Tang H Wong and Ica Manas-Zloczower. Two-dimensional dynamic study of


the distributive mixing in an internal mixer. International Polymer Processing,
9(1):3–10, 1994.

[39] PV Danckwerts. The definition and measurement of some characteristics of


mixtures. Applied Scientific Research, Section A, 3(4):279–296, 1952.

[40] Alena Kukukova, Joelle Aubin, and Suzanne M Kresta. Measuring the scale of
segregation in mixing data. The Canadian Journal of Chemical Engineering,
89(5):1122–1138, 2011.

[41] Alena Kukukova, Joelle Aubin, and Suzanne M Kresta. A new definition of
mixing and segregation: three dimensions of a key process variable. Chemical
engineering research and design, 87(4):633–647, 2009.

[42] R Byron Bird and Pierre J Carreau. A nonlinear viscoelastic model for polymer
solutions and meltsi. Chemical Engineering Science, 23(5):427–434, 1968.

84

You might also like