.
M u....un§‘1.uc2r.
. .a .341»: .«lu‘whb

yuan".

 

53..

t WV...

. c v I.
m. 591.7%

I

.21- f
. .cvhﬁu
{anis
. . 3mm!
. .i..1.. vtlni

“9.1...

dnunru.uluod.. h .
1......mn...»s
. 350.930.“: i"... .

. l.

an. LC.

2-... .1...
..

.uvlfktlluo
o t 319).)...

w. , ‘
'Inj‘11-25LEIHI.‘

. . vuﬂulu...‘>‘-
.v§1v$u.\.1l|

. .. . ‘ «4. our. V3.03”

~ :‘v...11..vV-.. ”yh'ol HM .

,1..Um..U3 unnu: (new...
\.

 

 

 

inf-luv.
V w [I‘- . . . ‘ ﬁr'l‘l t .npovv.x
. .t . . . . II.’ 0. ‘l. .u.“.}.

Js\‘>v t (.mﬂf. . .N . .. \bholln\v.ﬁ||!. .Iu‘iil.

v 1 ti. - .In . :6: .. {I l.\:n.!1|.\

. .uby I 33.0.1“ V#.P . . .. . x ,

. . > 1.5 . . , A

. . . r122... « . . s

l n 1. r! .‘ . .

1 All .‘ A A
. zré \ I...I.I .\..It....oo.|.h.. ..

‘l'HESlS

.
I l
I

’.

/,
,,
.

v..

CHGANS WTYLB

I IIIIII IIIIIIIIIIIII III II IIIIIIIII

1293 01563 9135

 

       

 

III

 

 

 

 

 

 

 

 

This is to certify that the

dissertation entitled

Processing Discontinuous Fiber Polymer Composites:
Fiber Alignment using Electric Fields and
Microstructure-Property Relationships

presented by

Murty N. Vyakarnam

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Chemical Engineering

 

 

 

WZKQM

Major professor

Date 12/2/96

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

~v~r1

v-v -s-x*—-—r -

4‘"—

- V—Y'_v V’Vﬁ-ﬁ

‘

 

 

LIBRARY

Michigan State
University

 

 

 

PLACE N RETURN BOX to roman this chockom from your ncord.
TO AVOID FINES return on or baton duo duo.

 

DATE DUE DATE DUE DATE DUE
30 ﬁg

 

 

 

 

iii—{1.9.2.339
0725 10

 

 

 

 

 

 

 

jﬁI:

 

 

 

 

-—IE:I
—IE::IE
——-I|—_I_——-fj

MSU I. An Affirmatlvo Action/EM Opportunity Imwon

 

 

PROCESSING DISCONTINUOUS FIBER POLYMER COMPOSITES:
FIBER ALIGNMENT USING ELECTRIC FIELDS AND
MICROSTRUCTURE-PROPERTY RELATIONSHIPS

By

Murty Narayan Vyakarnam

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemical Engineering
Composite Materials and Structures Center

1996

Copyright by
Murty Narayan Vyakamam

1996

r1

P?

Cl

AI

 

ABSTRACT

PROCESSING DISCONTINUOUS FIBER POLYMER COMPOSITES:
FIBER ALIGNMENT USING ELECTRIC FIELDS AND
MICROSTRUCTURE-PROPERTY RELATIONSHIPS

By

Murty Narayan Vyakarnam

Discontinuous ﬁber polymer composites are ﬁnding widespread use because of their
case in processability and improvements in performance over un-reinforced polymers.
However, their use is limited to non-structural applications due to the difﬁculties in
controlling ﬁber orientation and ﬁber length reduction during melt processing of long ﬁbers
at high volume fractions. Micromechanics models indicate that the elastic behavior of an
aligned discontinuous ﬁber composite should approach the behavior of a continuous ﬁber
composite if the length of the reinforcement exceeds critical ﬁber length. This provided the
motivation for research in the development of a process that has the potential to manufacture
aligned discontinuous or short ﬁber polymer composites and subsequently investigate the

microstructure-property relationships.

The unique combination of ﬁber alignment using electric ﬁelds in air and the recent
advances in polymer powder processing resulted in a novel, high speed, solvent free Aligned
Discontinuous Fiber (ADF) composite process. A semi-continuous laboratory prototype

ADF process has been developed which has the following unit operations: alignment of

conducting or insulating ﬁbers in air using electric ﬁelds in an orientation chamber; polymer
powder coating/impregnation of ﬁbers, and; compression molding of the powder coated ADF
preform into a composite. It has been found that alignment of dielectric E-glass ﬁbers in air
using electric ﬁelds is an extremely fast process (less than 1 second), but the non-buoyant
nature of the ﬁber motion makes the alignment behavior very complicated and is a balance
of polarization forces due to the electric ﬁelds; hydrodynamic forces due to ﬂuid resistance,
and; the rotational behavior of ﬁbers during free fall. Chopped E-glass ﬁbers of lengths
ranging from 3 to 25 mm have been successfully aligned in air using A.C. electric ﬁelds of

intensities ranging from 300 to 600 KV/m.

The ADF process has been demonstrated for chopped E-glass ﬁber and nylon12
matrix system and the properties of the composites fabricated using the process improve with
an increase in ﬁber alignment and an increase in ﬁber length. Improvements in modulus and
strength values of the ADF composites with ﬁber alignment ranged from 70 to 100 %, when
compared to equivalent composites that were manufactured with random ﬁber orientation.
Micromechanics analysis based on fundamental reinforcement theories, indicated that in the
chopped ﬁber-thermoplastic systems, it is the effective aspect ratio of the ﬁber aggregate

(bundle) that controls the elastic behavior of ADF composites.

To my mother
and

in loving memory of my father.

ACKNOWLEDGMENTS

I would like to express my sincere appreciation and gratitude to Professor Lawrence
T. Dizal for the guidance, vision and support he provided throughout my graduate research
at Michigan State University. I thank him for the many opportunities he provided that helped

me grow as a better researcher and achieve the goals of my education.

I would like to thank all my committee members: Prof. C. Petty, Prof. K. Jayaraman,
Prof. P. Duxbury and Prof. J. Asmussen for their valuable guidance and input from time to
time during my doctoral research. In particular, I want to thank Prof. Petty and Prof.
Jayaraman for their ardent desire to impart knowledge and enhancing my understanding in
the areas of ﬂuid flow and polymer rheology. My thanks are also to Prof. Duxbury and Prof.
Asmussen for the many useful suggestions they provided on harnessing electro-magnetic

ﬁelds for my research.

Many individuals helped me in developing the Aligned Discontinuous Fiber (ADF)
composite process. In particular, I would like to thank the following individuals: John
Brandon of MSU Cyclotron for introducing me to the practical aspects of high voltage

engineering; Mike McLean of the DER machine shop for the timely fabrication of various

vi

C0!

“1

pro

.\lii

Shit:

10 II:

lipa;

components that helped me build the process; Richard Wagner for experimental assistance
in summer '95; and Dr. Serban Peteu for helping me with the high speed video studies. I
want to thank Bob Schweizer of Owens Coming and Philip Chu of Vetrotex CertainTeed for
promptly supplying me with glass ﬁbers of speciﬁc requirements. I would also like to thank
Mike Rich and Brian Rook for their excellent cooperation and assistance provided to me
during these last several years. Finally, I acknowledge the ﬁnancial support provided by the
NSF Center for Low-Cost High-Speed Polymer Composite Processing and Michigan

Materials & Processing Institute.

I was fortunate to have many friends and colleagues, who made my stay at Michigan
State and the Composite Center in particular, intellectually and socially stimulating. I want
to thank Sanjay, Rik, Himanshu, Mark, Brent and Ed for sharing and bearing with me many

up and downs in this phase of life and making it a very memorable one.

Finally, I want to thank my mother for the inspiration and encouragement she
provided me to set higher goals in life. I dedicate this dissertation to her for the many
sacriﬁces she made to offer me good education. I also want to thank my sisters for their love
and encouragement throughout my education. I want to extend special thanks to my wife
Anandita, for her love and friendship ever since we met in graduate school. With she being
a fellow doctoral student, she understood the tribulations of research and provided excellent

support in the hours of crisis.

vii

Ch.

Chap!

TABLE OF CONTENTS

List of Tables .......................................................... xi
List of Figures ........................................................ xii
Chapter 1. Introduction ................................................ 1
Introduction ................................................ 2
Discontinuous Fiber Composites ................................ 3

1.2.1 Fiber Length .......................................... 5

1.2.2 Fiber Orientation ...................................... 7

1.2.3 Fiber Volume Fraction .................................. 8

1.2.4 Polymer Matrix ....................................... 9

1.3 Processing Vs Performance ................................... 11

1.4 Research Motivation ........................................ 15

1.5 Novel Processing Approach ................................... 16

1.6 Dissertation Outline ......................................... 18
References ...................................................... 19
Chapter 2. Review of Powder Processing and Fiber Alignment Techniques ...... 20
PART A: Polymer Powder Processing of Composites .................... 21

2.1 MSU Powder Process ....................................... 22

2.2 Charging of Polymer Powders ................................. 26

PART B: Fiber Alignment Techniques ................................ 28

2.3 Flow Induced Alignment ..................................... 28

2.3.1 Slurry / Hydraulic Based ............................... 28

2.3.2 Spray-up or Pneumatic ................................. 31

2.3.3 Molding and Extrusion ................................ 31

2.4 External Field Induced Alignment .............................. 33

2.4.1 Using Electric Fields .................................. 33

2.4.2 Using Magnetic Fields ................................. 37

2.5 Summary and Discussion ..................................... 37
References ...................................................... 40
Chapter 3. Fiber Alignment Using Electric Fields .......................... 42
3. 1 Introduction ............................................... 43

3.1.1 Electric or Magnetic Fields? ............................ 44

3.1.2 Fiber Dimensions ..................................... 46

viii

3.2 Theory of Fiber Orientation ................................... 48
3.2.1 E-Field Polarization ................................... 49

3.2.2 E-Field Alignment Time ............................... 52

3.2.3 Fiber Hydrodynamics .................................. 55

3.2.3.1 Orientation Behavior of Fibers During Free Settling . . . 55

3.2.3.2 Fiber Motion in Shear Flows ..................... 58

3.2.3.3 Settling Velocity of Fibers ....................... 59

3.3 Experimental Studies ........................................ 61
3.3.1 Experimental Setup ................................... 62

3.3.2 Behavior in DC. Fields ................................ 65

3.3.3 Behavior in AC. Fields ................................ 69

3.3.3.1 Effect of E-ﬁeld Intensity and Fiber Aspect Ratio ..... 70

3.3.4 Fiber Settling Behavior ................................ 82

3.4 Summary of Results ......................................... 86
References ...................................................... 88
Chapter 4. Aligned Discontinuous Fiber (ADF) Composite Process ............ 90
4. 1 Novel Concept ............................................. 91
4.2 ADF Process Design ........................................ 94
4.2.1 Unit Operations ...................................... 94

4.2.2 Fiber Feeder ......................................... 97

4.2.2.1 Fiber Pre-alignment ............................ 99

4.2.3 Electric Field Orientation .............................. 101

4.2.3.1 Orientation Chamber .......................... 101

4.2.3.2 Electrode Design ............................. 102

4.2.3.3 High Voltage Source .......................... 106

4.2.4 Deposition Platform .................................. 108

4.2.5 Powder Coating ..................................... 1 10

4.3 Process Performance ....................................... 112
4.3.1 Fibers ............................................. l 13

4.3.2 Polymer Characterization .............................. 1 14

4.3.3 Operating Conditions ................................. 114

4.3.4 Fiber Orientation Distribution (FOD) Measurement ......... 117

4.3.5 Aligned Vs Random .................................. 121

4.3.6 Concentration Effects ................................. 123

4.4 Process Consolidation ...................................... 128
4.4.1 Process Consolidation Cycle ........................... 130

4.4.2 Consolidation Pressure ................................ 130

4.5 Manufacturing ............................................ 1 33
4.5.1 Processing Rates .................................... 133

4.5.2 High Speed Manufacturing Schemes ..................... 136

4.6 Summary ................................................ 138
References ..................................................... 141

Ch;

Chapter 5. Microstructure-Property Relationships ......................... 142
5.1 Introduction .............................................. 142

5.1.1 Microstructure-Property Relationship Flowchart ........... 145

5.2 Microstructure Descriptors .................................. 147

5.2.1 Fiber Orientation Distribution .......................... 150

5.2.2 Fiber Aspect Ratio .................................. 152

5.2.3 F iber-Matrix Interaction ............................... 154

5.3 Elastic Property Predictions .................................. 156

5.4 Reinforcement Models ...................................... 160

5.4.1 Shear Lag Theory .................................... 161

5.4.2 Eshelby’s Inclusion Approach .......................... 164

5.4.3 Halpin-Tsai’s Relationships ............................ 166

5.5 Experimental Results ....................................... 169

5.5.1 Sample Preparation and Tensile Testing .................. 169

5.5.2 Effect of Fiber Alignment ............................. 169

5.5.3 Effect of Fiber Length ................................ 170

5.6 Model Predictions and Discussion .......................... 170
References ..................................................... 1 77
Chapter 6. Conclusions and Future Work ................................ 179

 

 

I2

Te

Table 1.1

Table 1.2

Table 2.1

Table 3.1

Table 3.2

Table 4.1

Table 5.1

Table 5.2

Table 5.3

Table 5.4

Table 5.5

LIST OF TABLES

Typical Properties of Reinforcing Fibers .......................... 4
Typical Properties of Polymer Matrices ......................... 10
Triboelectric Series of Polymers ............................... 27
Pertinenet Fiber Properties ................................... 45
Behavior of Fibers in DC. Fields .............................. 67
Typical Operating Variables ................................. 1 16
Constituent Properties ...................................... 149
Volume Fraction and Void Fraction of ADF Composites ........... 149

Actual Fiber Orientation Distributions of “Aligned” and “Random” ADF
composites ................................................ 15 1

Fiber Bundle Length and Aspect Ratios ........................ 154

Model Predictions Compared with Experimental Modulus Results. . . 174

xi

 

 

Fi
a.

Fig;

FM
1&1

my,

FIEUR

Figure 1.1
Figure 1.2

Figure 1.3

Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5
Figure 2.6
Figure 3.1
Figure 3.2
Figure 3.3

Figure 3.4

Figure 3.5

LIST OF FIGURES

Concept of critical ﬁber length based on Kelly-Tyson model. ......... 6
Stress transferred to a ﬁber as a function of critical aspect ratio. ....... 6

Performance versus processability trends of polymer composite material

systems. .................................................. 12
Prototype MSU Powder Prepreg Process ........................ 24
MSU High Speed Powder Prepreg Process ...................... 25
ERDE Glycerine process for ﬁber alignment ..................... 30
Budd slurry process for preform manufacture .................... 30
Process to make oriented chopped ﬁber mats using electric ﬁelds ..... 35

Process for alignment of ﬁbers using electric ﬁelds in dielectric ﬂuid . . 35
Forces acting on a ﬁber in the orientation chamber ................ 50
Schematic representation of the orientation state in free falling ﬁbers . . 57
Experimental setup to investigate ﬁber alignment in electric ﬁelds . . . . 63

High speed videographs of free falling E-glass ﬁbers (0.5" long) in the
oreintation chamber recorded at 1000 frames/sec under two conditions.
Top: Fiber alignment in AC. ﬁeld of intensity 400 KV/m. Bottom:
Random ﬁber orientation under no E-ﬁeld. ....................... 71

High speed videograph of free falling E-glass ﬁbers (0.25" long) aligning
in the oreintation chamber in an AC. ﬁeld of intensity 400 KV/m
(recorded at 1000 frames/sec) ................................. 72

xii

 

 

Fl

Figu

Figm
Figun
Figure
Figure
FIBUre

FI§Ure .

 

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Fiber orientation distribution of E-glass ﬁbers settled at the bottom of the
orientation chamber at A.C. ﬁeld intensities of 300, 400 and 500 KV/m
(Top F ODs for 1" ﬁbers; Bottom FODs for 0.25" ﬁbers) ............ 74

High speed videographs showing greater tendency of 0.5" ﬁbers in
attaining a stable horizontal ﬁber orientation state than 0.25" ﬁbers. . . . 76

High speed videographs taken by two cameras showing the natural

tendency of 1" long ﬁbers in attaining a stable horizontal orientation state.

Each frame is a composite shot of the top view (right) and side view (left).
......................................................... 77

Alignment time simulations based on Demetrides and Fuchs models. . . 81

Settling velocities of ﬁbers at observation point B. ................ 84

Settling velocities of ﬁbers at observation point C ................. 85

Novel concept of Aligned Discontinuous Fiber (ADF) composite
processing ................................................ 92

Schematic of the prototype Aligned Discontinuous Fiber (ADF) composite

process .................................................... 95
Fiber feeder cum pre-aligner. .................................. 98
Mechanism of ﬁber pre-alignment. ............................ 100
Electrode edge effect on ﬁber orientation. ....................... 103
Electric ﬁeld near the electrode edge. .......................... 105

Electrode design with a neutralizing ﬁeld for continuous processing. . 107

Circuit diagram for high voltage distribution. .................... 109

xiii

Figt

Fi"

 

’1
LU
s

 

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

SEM micrographs of E-glass ﬁber bundles coated (Top) and uncoated
(Bottom) with nylon-12 powder ............................... 111

Digital images of 1" long chopped E glass ﬁbers settled on the deposition
platform. Left: Aligned in E-ﬁeld of 400 KV/m, Middle: Random with no
E-ﬁeld, Right: Manually aligned ............................... 115

Fiber orientation distribution measurement. ..................... 118

FODs of different ﬁber length E-glass ﬁbers that have aligned in E-ﬁeld of
intensity 400 KV/m. The control case of random orientation is also shown
for comparison ............................................. 122

“Volume of inﬂuence” surrounding a ﬁber under two different orientation
states: random and Pre-aligned. ............................... 126

Non-interacting ﬁber concentration as a function of aspect ratio. ..... 127

Typical consolidation process cycle of a preform with a semi-crystalline
polymer matrix ............................................. 129

Schematic of proposed continuous ADF process incorporating polymer
powder coating; Fiber alignment using electric ﬁelds; Continuous ADF
composite sheet manufacture .................................. 137

Microstructure-property relationship ﬂowchart. .................. 146
“Bundle” reinforcement effect due to in-compatible ﬁber-matrix sizing.155
“Filament” reinforcement effect due to compatible ﬁber-matrix sizing 155

Micro-laminate concept for elastic property determination of ADF
composites ................................................ 1 58

Prediction of modulus for aligned discontinuous ﬁber composites as a
function of the reinforcement aspect ratio. ...................... 168

xiv

Figure 5.6 Effect of ﬁber alignment and ﬁber length on the tensile modulus. . . . . 171

Figure 5.7 Effect of ﬁber alignment and ﬁber length on the tensile strength. . . . . 172

XV

Chapter 1

Introduction
1 .1 Introduction ...................................................... 2
1.2 Discontinuous Fiber Composites ...................................... 3
1.2.1 Fiber Length ................................................ 5
1.2.2 Fiber Orientation ............................................ 7
1.2.3 Fiber Volume Fraction ........................................ 8
1.2.4 Polymer Matrix ............................................. 9
1.3 Processing Vs Performance ......................................... 11
1.4 Research Motivation .............................................. 15
1.5 Novel Processing Approach ......................................... 16
1.6 Dissertation Outline ............................................... 18
References ............................................................ 19

1.1 Introduction

Composite materials are a preferred combination of two or more different materials
or identiﬁable phases which possess synergistic advantages over the performance of
individual constituents. In the case of high performance composites, the reinforcing phase
consists of high modulus ﬁbers and the matrix phase consists of a polymer, metal or ceramic
material. Composite materials are fundamentally different from monolithic materials in
terms of the spatial variation of properties due to the variation in the microstructure of the
constituents which provides tremendous opportunities to utilize these materials in a variety
of situations. Polymer composites in particular, are increasingly replacing traditional
materials in many areas ranging from aerospace, automotive, infrastructure to sporting goods
and bio-medical applications due to their superior weight to stiffness and strength ratios;
resistance to chemical degradation; and better toughness properties. Beyond the superior
properties of ﬁber reinforced polymer composites, the reason for their success and bright
future lies in the fact that unlike monolithic materials, the properties of these materials can
be tailored by varying the microstructural features in particular the ﬁber orientation and ﬁber

length.

The focus of this research is on the development of a process that can manufacture
aligned discontinuous ﬁber composites, where the ﬁbers can be aligned using electric ﬁelds.
This chapter will give an overview of discontinuous ﬁber composites with an emphasis on
factors that inﬂuence both processing and performance. Then the research motivation and

objectives of this dissertation will be presented.

 

 

 

d1:
CO:
ori-
M

(M

the
Perl]
discr

the:

Procr-

[here i

and 1h:

[bilge f

1.2 Discontinuous Fiber Composites

The reinforcing ﬁbers of a composite material can be in a continuous or
discontinuous form and offer a wide range of properties (Table 1.1). In the case of
continuous ﬁber reinforcement the composite can be in the form of a laminate with the
orientation state of the ﬁbers ranging from unidirectional to a lay-up sequence that results in
a quasi-isotropic laminate. Alternatively, continuous ﬁbers can also be in a textile form
(woven, knitted or braided) with a wide range of performance capabilities. Textile or fabric
reinforced composites are popular because of the processing advantages they offer such as
a greater degree of freedom in controlling precisely the ﬁber orientation state in a two
dimensional state. However, the process of weaving and stitching is an expensive step and
the inferior behavior of these composites in compression compared to tension are some of
the drawbacks. When the ﬁbers are discontinuous, there is a signiﬁcant shift in the
performance as well as the processing envelope of these composites. The physical
discontinuity in the ﬁber length acts as a source for discontinuity in the stress-transfer from
the matrix to the ﬁber which results in the reduction of composite mechanical properties.
On the contrary, the discontinuous nature of the ﬁbers offers tremendous advantages in

processing of discontinuous ﬁber composites.

Discontinuous ﬁber composites are also referred to as short ﬁber composites, but
there is no rigorous basis for the distinction between the words “discontinuous” and “short”
and the two words are often used interchangeably'. Certain classes of composites especially

those from injection molding processes are generally referred to as short ﬁber composites,

 

 

 

While '
ﬁber .
00mm

COm p.

 

and 1h.
of P10:
and the
disco“.

perfi‘ct

 

 

Table 1.1 Typical Properties of Reinforcing Fibers 2'3

 

 

 

Fiber Diameter Sp. Modulus Strength
W) (0P2)—
PAN Carbon - A84 8 1.8 248 4.07
PAN Carbon - 1M7 5 1.78 301 5.31
Glass E 10 - 13 2.55 72.4 3.45
Glass S 10 - 13 2.49 86.9 4.3
Aramid“ 11.9 1.45 131 3.62
Polyethylene“ 27 0.97 172 3.0
Boron 140 2.7 393 3.1
Stainless Steel 10 - 50 7.9 190 1.4 - 2.1
B-SiC Whiskers 0.1 - 0.5 3.15 550 - 830 6.9 - 34.5
Carbon Whiskers < 1.0 2.25 145 20.7

 

* Kevlar (DuPont); "Spectra 1000 (Allied Signal)

while whisker reinforced metal matrix composites are generally referred to as discontinuous
ﬁber composites. When the reinforcing ﬁbers are typically in aggregates or bundles the
composites are referred to as chopped ﬁber composites or long discontinuous ﬁber

composites when the length of ﬁber bundles exceed half inch.

Fiber lengths vary anywhere from 0.1 cm to 5 cm in discontinuous ﬁber composites
and the orientation state can vary from completely random to perfectly aligned. Rheology
of processing is a function of the ﬁber aspect ratio, orientation state, the ﬁber volume fraction
and the viscosity of the suspension medium. Incidently, the mechanical performance of the
discontinuous ﬁber composite is also closely related to the above parameters. So, it makes

perfect sense to keep the issues of processing and performance close together in

5

discontinuous ﬁber composites. The microstructural features of discontinuous ﬁber
composites that are critical in processing and performance are introduced in the following

sections.

1.2.1 Fiber Length

The length of the ﬁbers that are used in discontinuous ﬁber composites can vary
widely depending on the process and the type of ﬁbers. Since, length of the ﬁber is an
extremely important parameter in the performance of discontinuous ﬁber composites, a
simple yet a very intuitive concept known as the critical ﬁber length will be introduced at
this stage while detailed analysis will be presented in Chapter 5. According to reinforcement
due to slip, in a representative volume element in a perfectly aligned discontinuous ﬁber
composite, the tensile stress (0c) is transferred from the matrix to the ﬁber through interfacial
shear stress (1) near the ﬁber ends assuming a perfect bonding between the ﬁber and the
matrix (Figure 1.1). This analysis is simple because it assumes the matrix to be ideally
elastic-plastic and hence the interfacial shear stress is a constant value“. The minimum ﬁber
length in which the ﬁber of diameter df can attain maximum allowable stress (0,”) is deﬁned

as the critical length, 1C for an average interfacial shear stress of 1:3,.

- —'- (1)

Force Balance on Representative Volume Element

 

 

 

 

 

 

 

 

 

 

 

+£17- I
Magi}..- “ I
1. Fiber iId‘ oc
.. - m. -. . . ______ Jim-M14 L
l ' L ’I -u.a.L'-“k.'.‘.u--‘-_-ri.‘.;- ‘ _ ._ “mum-mm-wamia
‘ “ * . mama. . V
°r-
55
E
i; (O f)nvg ‘
J
, , , ,, L _ - >
Half Critical Lch Half Critical Length
Variation of Stress Along Fiber Length
Figure 1.1 Concept of critical fiber length based on Kelly-Tyson model.

 

 

0.7

0.6 ,

0.5

Av. sues: I Max. Stress In Fiber

0.9 I

0.8 »

 

 

 

0.4

Figure 1.2

10 100 1000
Critical Aspect Ratio, lllc

Stress transferred to fiber as a function of
critical aspect ratio.

7

Below the critical length, the full reinforcement potential of the ﬁber is not realized but as
the length (1) increases beyond lc it can be easily shown that the average stress level in the
ﬁber (of)av approaches the maximum attainable in a continuous ﬁber, (of)max (Figure 1.1).
The average ﬁber stress (of)av is obtained from the area under the curve in Figure 1.1 and is

given by the following relation 5.

(2)

l
(of)... = (0,)...[1 — (1 — of] ; q = ~5-

From Eq. 1.2, it can be seen that as the length of the ﬁber increases, (of)av approaches (of)mam
(Figure 1.2). From this stress analysis it can be concluded that the performance of

discontinuous ﬁber composites approaches that of continuous ﬁber composites when ﬁber

length exceeds 50 times the critical length and when the ﬁbers are perfectly aligned.

1.2.2 Fiber Orientation

Fiber orientation becomes the dominant factor in inﬂuencing the composite
performance once the ﬁber lengths exceed the critical length. To get an idea of the inﬂuence
of ﬁber orientation, let us consider the more straight forward case of continuous ﬁber
composite lamina at different ﬁber orientations. Piggott6 summarizes the off-axis properties
of composites based on engineering constants, which can be used to predict the composite
properties at an angle 6 from the principal axis. Based on the experimentally determined
engineering constants of longitudinal (El 1), transverse (E22), and shear (GD) moduli and the

major Poisson ratio (v.2) ,the off-axis tensile modulus, E0 is given by:

 

 

2v ' 4
12 . Sin 0
+ (— - —)sm20c0520 +

12 n 22

1

75—6 E (3)
From this relationship it can be shown that there is a tremendous drop in the stiffness
properties of composites as the ﬁbers are mis-aligned from the direction of the applied stress.
In the case of discontinuous ﬁber composites, it is important to note that the effect on
stiffness and strength is expected to be similar and it will be a function of ﬁber aspect ratio

also. One of the objectives of this research is to see the effect of ﬁber orientation distribution

on the elastic behavior of discontinuous ﬁber composites.

1.2.3 Fiber Volume Fraction

The volume fraction of the short ﬁbers in the composite affects both the ﬁnal
mechanical performance of the discontinuous ﬁber composite and the rheology of ﬂow at
the processing stage. Typically, the ﬁber volume fraction (V.) varies between 50 and 75 %
for continuous ﬁber composites, while in discontinuous ﬁber composites it can vary
anywhere from 10 to 60 %. The maximum theoretical volume fraction (meax) that can be

obtained for aligned continuous ﬁbers is:

V
H

ﬁn” % = 0.785 for square array

= 0.907 for hexagonal array

 

 

1.2.4

disllr
01'3513
resins
are b¢
POI}m

Comp)

Shelf II

 

 

9

Higher volume fractions cannot be processed because of the theoretical limitations in
packing ﬁbers in a given volume. The maximum packing volume ﬁaction for discontinuous
ﬁbers that are not perfectly aligned will always be lower than 78.5% and will be a function
of the ﬁber length, ﬁber orientation and ﬁber cross sectional shape and area. As ﬁber
alignment improves, the ability to pack a higher volume fraction of ﬁbers increases. Fiber
alignment thus has the dual effect of allowing higher volume fraction, of ﬁbers as well as
larger aspect ratios, which result in signiﬁcant improvements in the stiffness and strength

properties of the composite.

1.2.4 Polymer Matrix

Polymers are broadly classiﬁed as thermosets and thermoplastics with further
distinction made between the various thermoplastic polymers based on their degree of
crystallinity. In the ﬁeld of high performance advanced polymer composites, thennosetting
resins have been the predominant choice, whereas a majority of the short ﬁber composites
are based on thermoplastic resins. In terms of global consumption trends, 80 % of the total
polymer shipments are thermoplastic resins, however, only less than 20% of advanced

composites are based on thermoplastic resins.

High strain-to-failure, increased toughness, shorter molding cycles, inﬁnite prepreg
shelf life, recyclability and repairability are some of the advantages of thermoplastics over
thermosets which has caused an increase in their utilization. However, the high melt

viscosities of thermoplastics is a problem when it comes to processing advanced ﬁber

10

Table 1.2 Typical Properties of Polymer Matrices7'8

 

 

Polymer Typei Tg Tm / HDT Tensile Tensile
( °C) (°C) Modulus Strength
(GPg) (MPL
Nylon 66 TP-SC 78 260 1.6 - 3.5 82
Nylon 12 TP-SC 40 175 1.24 35
Polypropylene TP-SC -10 1 76 1 .5 3 1
HDPE TP-SC -20 132 1.23 34
PEEK* TP-SC 143 360 - 400 3.2 100
Polycarbonate TP-A l 50 130 2 .4 65
Epoxy” TS 80 3.38 55 — 130
Polyester TS 6O - 205 2.1 - 3.5 35 - 104
Vinyl ester TS 93 - 135 3 - 3.5 76

 

1' TP - Thermoplastic; SC - Semi-Crystalline; A - Amorphous; TS - Thennoset
"' 1C1; "Shell Epon 1072

composites with continuous ﬁbers at high volume fractions. The advantages with
thermosetting resins has been low shrinkage, excellent ﬁber-matrix adhesion, good chemical
resistance, high thermal resistance and easy processability. On the other hand, the low
viscosity monomeric composition of thermosetting matrices, lends itself to inﬁltration and
coating processes subsequently which is cured to its ﬁnal cross linked state through a

chemical reaction.

From Table 1.2, one can conclude that polymers do not differ drastically when it
comes to their mechanical performance. However, there are substantial differences among

the different types of polymers in terms of thermal and rheological properties that virtually

ll

dictate the mode of composite processing. Thermoplastic polymers are highly viscous while
thermosetting polymers are at a low viscosity when used in their monomeric form. When
it comes to amorphous polymers there is a gradual drop in its mechanical performance above
Tg, while the drop in properties are more sharp in the case of semi-crystalline thermoplastic

polymers and does not occur till the temperature is near the melting temperature TM.

1.3 Processing Vs Performance

If the complete polymer composite materials system is taken into consideration then
performance and processing seem to be opposing trends as shown in Figure 1.3. The
performance of composites improve drastically when one moves from the injection molded
composites to the advanced continuous ﬁber composites. On the contrary, in general, ease
in processing drops when one moves from discontinuous ﬁber composites to continuous ﬁber
systems. Uniaxial orientation of continuous ﬁber reinforced polymer matrix composites
offers the highest attainable stiffness and strength properties. Full advantage of these
properties is taken by designing composite structures where the predominant stresses are
coincident with the ﬁber direction. This leads to a high degree of anisotropy which must be
overcome for practical structural applications by arranging the ﬁbers in plies stacked at
predetermined angles to each other, leading to expensive labor intensive lay-up methods".
To overcome this limitation textile preforms are being used but ﬁber crossover coupled with
limitations on minimum radii of curvature in the ﬁnished part, limit the geometry of ﬁnished

structures to simple shapes. Attempts to overcome this limitation by adopting textile

 

 

Figur

 

12

 

 

 

 

. 0 a r ‘ Q ' W', a. i . L ;.
': ' 'o ﬁﬁrj“ :; 7‘ 7, ’7’
o - § ‘N o' ’of ’

' I I 'I '. .1331.» 7

 

 

 

 

 

 

 

 

 

 

 

 

MOLDED RANDOM ALIGNED CONTINUOUS

 

 

Relative Processability

   

_ lative Perfromance
; trength/Stiffness)

EASE OF PROCESSING -’
PERFORMANCE

 

 

 

 

 

Figure 1.3 Performance versus processability trends of polymer composite
material systems.

 

 

I

 

matr.
shapi-
prep.-
distn'r
[I131 c.
depcr
Unifw
can” '
of the
comp
ofCO,

“imp;

 

 

l3

methods to the handling of ﬁbers, viz. braiding and weaving, to form complex shapes with
dry ﬁbers followed by liquid resin impregnation results in higher cost than other prepreg
forms. In terms of performance, laminate theory suggests that a part made ﬁ'om woven fabric
will exhibit reduced tensile and compressive strengths compared to a part made of
unidirectional materials due to crimping of ﬁbers to some degree in the weave. In the case
of damage tolerance and fracture toughness, woven materials outperform unidirectional

materials.

Fabrication of composites with a random array of short ﬁbers dispersed in a polymer
matrix results in a discontinuous composite material with isotropic properties, that can be
shaped into complex geometries with a high degree of automation. However, structural
properties of these composites can be drastically reduced because of the orientation
distribution of ﬁbers; discontinuity in reinforcement; and the lower volume fraction of ﬁbers
that can be packed into a random array. High performance of discontinuous ﬁber composites
depends on the aspect ratio of the ﬁber (l/d), alignment of ﬁbers, ﬁber-matrix adhesion, and
uniform ﬁber packing. In the case of aligned discontinuous ﬁber composites the end effects
cannot be neglected and the l/d ratio is taken into consideration in predicting the strengths
of the composite but some of the disadvantages of woven fabric are eliminated. An optimum
composite system which would provide superior mechanical performance approaching that
of continuous ﬁber composites, and that which would provide ease in the fabrication of

complex geometrical parts as in the case of short ﬁber composites, is an aligned

ESLii‘II:

comm

l4

discontinuous ﬁber composite due to its high packing of axially oriented ﬁbers (Figure

l.3)'°.

There are two generalizations that should be borne in mind when it comes to

establishing a qualitative relationship between the performance and processing of polymer

composites.

The reinforcing ﬁbers are the dominant phase when it comes to mechanical
performance. The stiffness and strength values of ﬁbers vary a lot but are generally
an order of magnitude higher than that of the typical polymers (Table 1.1 and 1.2).
Hence, the microstructure of the reinforcing phase: the ﬁber orientation distribution,
ﬁber aspect ratio and the ﬁber-matrix adhesion tends to be the controlling factor in
the overall performance of the composite.

On the contrary, the stiffness and strength values of the various classes of polymers
are quite close and there is little variation between the two classes of polymers:
thermoplastics and thermosets. However, they differ widely in the thermal and
rheological behavior (Table 1.2). So, more often that not the selection of the
composite processing technique is dictated by the polymer matrix. For example, low
viscosity monomer impregnation is the choice of prepregging when the matrix is a
thermoset while the choice narrows down to powder prepregging or hot melt
impregnation when the matrix is thermoplastic. Another example is Resin Transfer
Molding which is typically used for thermosetting resins while ﬁber reinforced

injection molding invariably pertains to high melt viscosity thermoplastic resins.

Hem
P01)“
rhen
than

uhhr

asuh
Iheal
Procr
3 h1g1
acnu
mam
aPPIlI
Incite
Can i

dhhii

15

Therefore, processing of a composite material is generally dictated by the behavior

of the polymer.

Hence, the problem of optimizing performance and processing of discontinuous ﬁber
polymer composites is a function of both the reinforcement microstructure as well as the
rheological and thermal behavior of the polymer. The overall ease in processing is
characterized by the speed of manufacturing, productivity, and scope for automation and

ultimately the cost.

1.4 Research Motivation

Controlling the orientation state of discontinuous ﬁber polymer composites has been
a subject of intense research in recent years. Although the signiﬁcance of ﬁber alignment and
the ability to control the orientation state of ﬁbers has been a goal for some time a versatile
processing methodology that can be operated at high speeds of operation and amenable to
a high degree of automation, at the same time does not exist. The motivation of this research
stems from the tremendous need for such a processing technique that can be used in the
manufacturing of aligned discontinuous ﬁber composites that can be used for structural
applications in the automotive, durable goods and infra-structure sectors. It will not only
increase the envelope of utilization of discontinuous ﬁber composites but also provide what
can be termed as microstructure controlled composites where the ﬁber orientation

distribution is controlled based on the ultimate part performance.

16

It is hypothesized that if short discontinuous ﬁbers of lengths exceeding critical ﬁber
lengths can be aligned in electric ﬁelds in air and if these ﬁbers can be suitably coated with
ﬁne polymer powder, then a high speed processing methodology can be established to
manufacture aligned discontinuous ﬁber composites with stiffness and strength properties
approaching those of continuous ﬁber composites. Both the ﬁber alignment in electric ﬁelds
in air and the consolidation of powder coated ﬁbers are rapid steps that will eventually lead
to a high speed processing methodology. Further, such a processing technique will establish
a method to manufacture microstructure-controlled discontinuous ﬁber composites. The
objectives of this research are:

0 Understand the alignment behavior of ﬁbers in air using electric ﬁelds

0 Develop a process that combines ﬁber alignment using electric ﬁelds with powder
coating of ﬁbers to establish a high speed process that can manufacture aligned
discontinuous ﬁber composites

0 Verify the microstructure-property relationships of discontinuous ﬁber composites,
and establish the concept of processing microstructure controlled discontinuous ﬁber

composites.

1.5 Novel Processing Approach

Recent advances in the area of powder coating of ﬁbers at MSU' "'2’” in combination
with the phenomenon of aligning ﬁbers in electric ﬁelds, has paved the way to conceptualize
and develop a novel high speed processing methodology that can manufacture aligned

discontinuous ﬁber composites. Full realization of the stiffness to weight beneﬁts of these

17

composites is possible due to effective ﬁber alignment combined with the ability to pack and
process at higher volume fraction of ﬁbers. Absence of solvents or liquids in the Aligned
Discontinuous Fiber (ADF) process would improve the speed of processing many folds and
make the process environmentally benign. The ﬁber alignment technique of the ADF
process is simple in concept, with the scope for retroﬁtting this technique in an existing
composite sheet or lamination processing unit. Unlike continuous ﬁber random mat
reinforced composites which have poor drapeability and which have problems of
delamination under compression, aligned discontinuous ﬁber composites can be ﬂexible and
can be molded or stamped into complex parts. It is envisaged that ADF composites, with
unique performance and processability capabilities, will expand the range of the applicability

of discontinuous ﬁber thermoplastic composite materials.

Numerous studies have been conducted over the past couple of decades to relate the
microstructure-property relationships of discontinuous ﬁber composites. Unfortunately, in
a majority of the studies there is a variation in the ﬁber aspect ratio or the ﬁber orientation
distribution could not be determined accurately, resulting in inconclusive veriﬁcation of the
validity of the micromechanics models with experimental results. An important contribution
from making composites using the ADF process is to verify the micromechanics models that
are available for discontinuous ﬁber composites with a well characterized composite with
a known ﬁber orientation distribution. This will further the understanding of the interaction
between ﬁber orientation distribution, ﬁber aspect ratio and the ﬁber-matrix interaction

including adhesion. Finally, the ability to control the microstructure goes beyond improving

 

 

1.1

m

16‘

inc
“1
mi
Fir

CI.

18

performance of the composite material in a particular direction. In fact it could potentially
lead to the processing of materials where the thermal, electrical, optical or magnetic

properties can be spatially controlled.

1.6 Dissertation Outline

The dissertation is divided into six Chapters. This chapter set the background and
motivation for research and identiﬁed the problem. Chapters 2 follows with a detailed
review of the literature on powder processing of composites and various ﬁber alignment
techniques. In Chapter 3, the investigation into the behavior of ﬁber alignment in electric
ﬁelds is presented including the theory of ﬁber orientation and the experimental setup used
in the investigations. Chapter 4 outlines the ADF process development in its entirety
including process design, process consolidation and manufacturing schemes. Chapter 5 deals
with microstructure-performance relationships of aligned discontinuous ﬁber composites
using both model predictions from fundamental micromechanics and experimental results.
Finally, conclusions from this work and recommendations for future work are given in

Chapter 6.

 

 

 

J

1J-

10.

H
La)

10.

11.

12.

13.

References

Chou, T. W., Microstructural Design of Fiber Composites, Cambridge University
Press, pp 169-230, 1992.

Milewski, J. V. and Katz, H. S., Handbook of Reinforcements for Plastics, Van
Nostrand Reinhold Co., New York, 1987.

Mallick, P. K. “Fiber Reinforced Composites - Materials, Manufacturing and
Design”, Marcel Dekker, Inc., New York, pp 18-19, 54-65, 1990

Kelly, A. and Tyson, W. R., “Tensile Properties of Fiber-Reinforced Metals:
Copper/Tungsten and Copper/Molybdenum”, J. Mech. Phys. Solids, Vol. 13, p329—
350, 1965.

Agarwal, B. D. and Broutman, L. J., "Analysis and Performance of Fiber
Composites", John Wiley & Sons, 1990.

Piggott, M., "Load Bearing Fibre Composites", Permagon Press, 1980.
Modern Plastics Encyclopedia, McGraw-Hill, New York, 1996.

Bigg, D. M., “Processing of Thermoplastic Composites” in Polymer Rheology and
Processing, Eds. Collyer, A. A. and Utracki, L. A., Elsevier Applied Science,
London, pp 381-406, 1990.

Foley, M. F., "Techno-Economics of Automated Composite Manufacturing
Techniques", SAMPE Quarterly, January 1991.

Chang, I. Y. and Pratte, J. F ., J. Thermoplastic Composite Materials, Vol 4, p.227-
252, 1991.

Drzal, L. T., et. al., “Powder lmpregnation of Advanced Composite Materials”, in
Polymer Powder Technology, Eds. Narkis, M. and Rosenzweig, N., John Wiley &
Sons Ltd., pp 511-530, 1995.

Iyer, S. R., “Continuous Processing of Unidirectional Prepreg”, Ph.D. Dissertation,
Michigan State University, 1990.

Vyakarnam, M. N., “Development of a High Speed Powder Process to Manufacture
Composite Prepreg”, MS. Thesis, Michigan State University, 1992.

19

Chapter 2

Review of Powder Processing and
Fiber Alignment Techniques

PART A: Polymer Powder Processing of Composites .......................... 21
2.1 MSU Powder Process ............................................. 22
2.2 Charging of Polymer Powders ....................................... 26
PART B: Fiber Alignment Techniques ...................................... 28
2.3 Flow Induced Alignment ........................................... 28
2.3.1 Slurry / Hydraulic Based ..................................... 28
2.3.2 Spray-up or Pneumatic ....................................... 31
2.3.3 Molding and Extrusion ...................................... 31
2.4 External Field Induced Alignment .................................... 33
2.4.1 Using Electric Fields ........................................ 33
2.4.2 Using Magnetic Fields ....................................... 37
2.5 Summary and Discussion ........................................... 37
References ............................................................ 40

20

21

The concept of combining polymer powder processing and ﬁber alignment using
electric ﬁelds in air offers a rapid, solvent free, processing technique to manufacture polymer
composites reinforced with discontinuous ﬁbers with a preferred ﬁber orientation. The
literature review will be presented in two parts: Part A will focus on polymer powder
processing; and Part B will focus on ﬁber alignment techniques. The conclusions from this

review will be summarized at the end of the chapter.

PART A: Polymer Powder Processing of Composites

Polymer composite materials are emerging as a potentially revolutionary new material
form capable of vastly increasing the applications for polymers. These materials have an
aerospace history which emphasized high structural performance with cost being a secondary
consideration. Durable goods (transportation, appliance, construction, etc.), infrastructure
(bridges, decking, etc.) and off-shore oil platform applications, promise to increase the use
of polymer composite materials by several times if lower cost and environmentally safe

processing methods can be found to manufacture these materials.

The main objective of any process for the manufacture of polymer composite
materials is to completely surround each individual reinforcing ﬁber (typically having a
diameter of 5 - 20p) with a polymer matrix in the ﬁnal fully densiﬁed condition. Because
of the reactivity or high viscosity of polymer matrices, low temperature methods using

organic solvents have usually been used to manufacture these material forms. This keeps the

 

 

manufact
seriousl)
composit-

free prom

Xi
manufactt
Chapter on
processing
mien he]
Charging c

Impregnati

Va:

10w 10 pm i

 

de‘eIOPcd.

dePOSIted .

 

A nOV’e] m:
speak” or

p nI’ICIDIe r}

 

22

manufacturing costs high. In addition, even low levels of voids (1%) or retained solvent
seriously reduce the mechanical properties of the resulting composite material. New
composite processing employing dry polymer powders offers a low cost, high speed, solvent

free processing technique.

Novel processing methods have been developed at MSU in the last several years to
manufacture continuous ﬁber powder prepreg employing polymer powders. A recent book
chapter on powder processing of composites by Drzal et. al.1 describes the various powder
processing methods and the principles of powder impregnation and consolidation. The
review here will be focussed to a brief description of the MSU Powder process and the
charging characteristics of polymer powders which is key to the powder coating and

impregnation of ﬁbers.

2.1 MSU Powder Process

Various modes of dispersing the polymer powder and impregnating on a spread ﬁber
tow to produce what is a prepreg or towpreg in both wet and dry conditions have been
developed. The powder impregnation processes differ mainly in the way the particles are
deposited on the ﬁbers and the particle-ﬁber forces responsible for holding them in place.
A novel method for spreading a tow of ﬁbers using low frequency acoustic energy from a
speaker or other gas vibrating means was developed where the spreader Operates on the

principle that the pulsating velocity of the gaseous medium drives the spreading of the

23

collimated ﬁber tow into individual ﬁlaments. The spread width can be controlled by the

amplitude of the acoustic energy and the tension in the ﬁber tow.

Creating a uniform and controllable dispersion of ﬁne cohesive polymer powder
using acoustic energy was studied by lyer and Drzalz. It was found that cohesive powders
in the size ranges 5 to 30 um could be aerosolized in a chamber where the powder bed is
subjected to low frequency vibration from an acoustic source. A prototype MSU Powder
prepreg process was developed by lyer et a1.3 where the size of the impregnated particles and
the ﬁbers were of the same orders of magnitude. In this process, spread ﬁber towl2 is passed
through the aerosol cloud that hovers at the top of the aerosolizer (Figure 2.1). Tribocharged
polymer particles electrostatically adhere to the ﬁbers which are later sintered in place (refer
§ 2.2). The rate of deposition is directly controlled by the amplitude of the acoustic energy.
Subsequently, a scaled up version of this process which operates at high speeds was
developed by Vyakarnam and Drzal“, in which the aerosol of polymer powder, generated
acoustically, is entrained into a separate impregnation chamber (Figure 2.2). In this chamber,
spread ﬁbers come in contact with the settling aerosol particles in a counter current fashion.
The size of particles is again critical as it affects both the charge-to-mass ratio acquired and
the ratio of inertial to viscous forces on the particles in suspension. The rate of deposition
can be controlled by the aerosol concentration entrained into the impregnation chamber,
which can be indirectly controlled by the amplitude of the acoustic energy and the aerosol

entrainment rates.

24

 

 

 

Nbroiiors

fﬂpr

~,:::a:7O O

Takoup drum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 “if

Frequency Generator

 

 

 

 

 

 

 

Acoustic Speaker

Figure 2.1 Prototype MSU Powder Prepreg Process.

25

Prepreg Spool

Infra-red Heater

 

 

 

Polymer Powder

 

»:-:-:-: -:-:-:-: Aerosol Generator

nnnnnnnn

arrears; 232533;:

oooooooo
nnnnnnnn
cccccccc
uuuuuuuu
oooooooo
oooooooo
........
--------
nnnnnnnn
........
cccccccc
eeeeeeee
eeeeeeee
oooooooo
cccccccc
oooooooo
nnnnnnnn
oooooooo
........

........
oooooooo
uuuuuuuu
........
nnnnnnnn
oooooooo
........
nnnnnnnn
uuuuuuuu
nnnnnnnn
nnnnnnnn
nnnnnnnn
nnnnnnnn
........
oooooooo
oooooooo
.......
uuuuuuuu
........
........

 

 

 

Acoustic Spreader Fiber Feed 31’001

Figure 2.2 MSU High Speed Powder Prepreg Process.

26

2.2 Charging of Polymer Powders

In most powder processes, polymer particles are either charged by an electric ﬁeld
or acquire charge due to tribocharging. Tribocharging or triboelectriﬁcation is a
phenomenon of surface charging of powder particles that occurs due to the frequent contact
of the powder particles with other surfaces. In the case of electrostatic ﬂuidized beds, ﬁeld

charging of particles takes place due to an imposed electric ﬁeld.

When particles of different work functions are brought into sliding and/or frictional
contact, tribocharging is invariably involved. Work function, is the energy required to knock
off an electron from an atom and is a strong function of the materials’s chemical
composition, polar function groups if present and impurities. Materials can be arranged in
a Triboelectric Series (Table 2.1) depending upon the sign and magnitude of charge acquired
on the surface due to contact and or frictional electriﬁcation. 0310 and Lama6 and others
have separately arranged many polymers in a triboelectric series and found a strong
correlation between the macroscopic property of relative permittivity and the sign and
magnitude of charge of a material. The sign and magnitude of the charge becomes decisive
as the distance between the two materials increases in the triboelectric series. Once a particle
acquires surface charge, the rate at which the redistribution proceeds depends on the
electrical relaxation time of the particle. Relaxation time is a product of particle resistivity
and permittivity7. Most polymers being insulating materials with resistivity greater than 10'3

ohm-cm have relaxation times in minutes or even hours which makes them more likely to

27

Table 2.1 Triboelectric Series of Polymers

 

 

Material Permittivity Sign of Charge
Polyamide 3 - 6 +ve
PPS 3.9 +ve
Epoxy resin 4 +ve
PEEK 3.2 +ve
Epoxy Polyester 3 - 4 +ve
Polycarbonate 2.9 —ve
Polystyrene 2.6 -ve
Polyethylene 2.2 - 2.7 -ve
PTFE 2.1 -ve

 

retain surface charges for long periods of time. It has also been reported that work function
decreases with an increase in particle size which often results in large particles getting
positively charged and small particles getting negatively charged7. This phenomenon often

results in bi-polar charging of polymer powders that have a broad particle size distribution.

The size of the particle has a great effect on the magnitude of charge. The surface
charge on a spherical particle increases with the square of the diameter while the mass
increases with the cube of the diameter. Hence, the charge to mass ratio increases with
decrease in particle size. In polymer powders, the effect of tribocharging has been found to

be more pronounced in the case of size ranges under 50 microns.

28

PART B: Fiber Alignment Techniques

In composite materials reinforced with continuous or discontinuous ﬁbers the spatial
variation in properties can be potentially achieved by varying the ﬁber orientation in a
controlled manner. Fiber orientation can be precisely controlled in the case of continuous
ﬁber composites while controlling the orientation of discontinuous or short ﬁbers is a
challenge. The exponential growth in the use of discontinuous ﬁber reinforced plastics in
the recent years has sparked intense activity in understanding and developing processes that
can control ﬁber orientation and thereby make aligned discontinuous ﬁber composites.
Growing demand has also been experienced for oriented ﬁber preformsi that can be used in
Resin Transfer Molding (RTM) or Structural Reaction Injection Molding (S-RIM) processes
to make large structural parts in automotive and aerospace applications. Further, it is also
envisioned that controlling ﬁber orientation in an attempt to manufacture microstructure
controlled composite materials holds tremendous amount of potential for the advanced
materials of the future. It is expected that material selection and design, part design and
manufacturing will be highly automated and integrated into one task. A review based on
published literature and patents on ﬁber alignment techniques that pertain to manufacturing

of aligned/oriented discontinuous ﬁber composites is presented in this part.

2.3 Flow Induced Alignment
2.3.1 Slurry / Hydraulic Based
ERDE Glycerine Processs: In this process developed by Explosives Research and

Development Establishment (UK), a dilute dispersion of ﬁbers in glycerine is passed via a

29

bafﬂed reservoir through a tapered slit at a constant throughput onto a ﬁlter bed which is
under vacuum (Figure 2.3). Fiber alignment is caused due to the velocity gradients that are
setup between the accelerating streamlined ﬂow and the stationary boundary of the tapered
slit. The viscous drag on the ﬁbers created by the velocity gradients orients the ﬁbers. The
ﬁbers align in the direction of the motion of the liquid and are collected on the ﬁlter bed in
a reciprocating fashion. The carrier liquid is rapidly removed without a loss in alignment
using a ﬁlter surface with low resistance and a high capacity displacement pump. The major
factors that affect the ﬁber orientation in this process are the viscosity of glycerine (which
is a function of temperature); horizontal velocity of the of the orientation hopper; air
pressure; distance between the slit to the ﬁlter bed; and the slit width. Very good ﬁber
alignments were achieved using this process and the process was used by Kacir et alg. in the
preparation of oriented short glass ﬁber mats in order to study the mechanical performance

of aligned short ﬁber composites.

Budd Slurry Process: Another slurry process, popularly known as the Budd Slurry process
(Figure 2.4) for preform manufacture has been developed in the recent years as a means to
manufacture net shape preform for automotive applications. A composite is made from this
preform using the conventional Resin Transfer Molding (RTM) processes. A well agitated
liquid suspension of ﬁbers is used to create a ﬁber mat by either draining the liquid or raising
a ﬁlter bed through the suspension. Some degree of ﬁber orientation can be achieved by
suitable placement of bafﬂes in the reservoir, but the primary focus of this process is not for

inducing ﬁber alignment but to achieve a random orientation distribution. Higher packing

30

 

 

 

 

 

 

 

Feed Recycle :

 

Vacuum pump

Glycerine return

Figure 2.3 ERDE Glycerine process for ﬁber alignment“.

 

 

CV; NOE.
WA'CR L(V(L
Of 4'

ctASS f cars
2' LONG

 

 

'UNCI IO HOLD
V... 00“
IRQ' 00“,
(‘C'I‘OI' 2’)

IO!
(06".00'.ro')

CI. COLA. ruut
' lHrC“

 

 

I/z’ INK auto!
,‘ VANCSKS‘l.SYINS)

SOUAlt SClth
IO'i-Io'wnu “Otis
/" Dd 02%.." OAS!

 

 

  

 

 

woont‘n Inst

 

 

 

“(Tot Olsﬂt

Pun-NIL
(ro'xio‘Ii-)

 

 

 

 

 

 

 

 

 

Figure 2.4 Budd slurry process for preform manufacture ‘°.

31

ﬁ'action of ﬁbers is said to be achievable over typical spray-up techniques. Developmental
efforts are underway to obtain a more uniform variation in the areal weight of the preform
that is manufactured and also establish the cost beneﬁt analysis over the competing spray-up

techniques”).

2.3.2 Spray-up or Pneumatic

For industrial RTM and S-RIM processes, it is essential that the preforrning step be
very rapid to be cost effective. Owens-Coming” has developed a directed preforrning
process with robotic control for industrial RTM and S-RIM preform applications. This
processing methodology is very fast and ﬁber orientation can be controlled by using some
mechanical means like deﬂector plates. Fibers exiting from a hose impinge on a set of
deﬂector plates which are at an angle to each other, and some control is achieved in the
degree of orientation as the ﬁbers build the preform. Typically stiff, thick and long (1 - 3
inches) ﬁber bundles can be effectively oriented by this technique which results in poor
matrix impregnation and inﬂexible ﬁber preforms. Efforts have also been directed at
improving the transport of the ﬁbers, at high throughputs without clogging, using a swirl

ﬂow approach”.

2.3.3 Molding and Extrusion
In the techniques described so far, ﬁbers are oriented in a step separate from the ﬂow
of the resin step. In the molding and extrusion processes the ﬁbers are in a suspension with

volume fractions typically around 10 to 30 %, and aspect ratios around 10 to 30, in the

32

polymer melt that will ultimately form the matrix. In these processes, ﬁber orientation is
ﬂow induced hence the intent is to predict the orientation state of the ﬁbers rather than
controlling it, as the polymer melt ﬂows into a mold cavity. Fiber orientation prediction and
control during typical polymer melt processes like injection molding or extrusion is based
on the same principles except for the difference in the ﬂow regimes from a ﬂuid mechanics
point of view. In these processes, the viscosities of the polymer melts are so high that
conditions of creeping ﬂow are well satisﬁed and hence the problems can be tracked by
analytical solutions to some extent. Advances made in the rheology of ﬁber suspensions is
enabling prediction of ﬁber orientation distribution in molding operations. However, the

effects of ﬁber concentration is still a complex problem and is an area of active research”.

Injection Molding: In injection molding, typically, the ﬂow of molten polymer ﬁlled with
short ﬁbers into a thin cold walled mold can be treated as Hele-Shaw ﬂow and the high shear
rates established near the wall aligns the ﬁbers along the direction of the ﬂow while the
extensional ﬂow in the core aligns the ﬁbers transverse to the ﬂow direction. This
phenomenon transforms into a skin and core micro-structure when the melt freezes with the
ﬁbers aligned in the ﬂow direction in the skin and ﬁber transverse to the ﬂow in the core.

The most common ﬂow pattern is from a converging or parallel runner channel through a
narrow gate, which then diverges into a plate and gives a two dimensional ﬂow path.
Convergent and parallel ﬂow induces orientation parallel to the streamlines, but divergent
ﬂows tend to introduce ﬁber orientation normal to the ﬂow direction. The ﬂow pattern is

also inﬂuenced by friction and chilling of the charge at the mold surfaces, which affects the

33

ﬁber orientation distribution along the thickness of a plate type molding, and can result in

a three or ﬁve zone ﬁber orientation distribution.

Within the limitations of molding, a certain degree of control over the orientation
distribution of the ﬁbers can be achieved by optimizing the effects of the following
parameters: (a) design of the cavity; (b) position and form of the runners and gates; (c)

injection rate; (d) temperature of the charge; and (e) mold temperature.

Extrusion: In the case of extrusion, elongational ﬂow in the core region of the extruder die
is a powerful mechanism for orienting ﬁbers in any direction by changing the geometry of
the die. Goettler et al.” of Monsanto were probably the ﬁrst to do extensive studies in

developing practical die geometries to control ﬁber orientation in extrusion.

2.4 External Field Induced Alignment
2.4.1 Using Electric Fields

Studying the physics of ﬁber alignment in electric ﬁelds is referred to as “one of the
longest known effects of electric ﬁeld” by S. Whitehead in a letter in response to an article
entitled “The Orientation of Fibers in an Electric Field” by J. O. Isard that appeared in 1954
in the British Journal of Applied Physics”. So, it can be said that the physics of ﬁber
alignment is well established. Demetriades of Stanford University in 1958 established the
theory for ellipsoidal alignment in electric ﬁelds for neutrally buoyant ﬁbers with essentially

zero Reynolds numbers. The early studies on ﬁber alignment were used as a way of

34

measuring the dielectric constants of ﬁbers. Work in controlling the orientation of ﬁbers for
some end use has also been the subject of continuous investigation over the last seventy
years, initially in the textiles industry and then more recently in the composites industry. In
1971, Monsanto obtained a patent for a process that aligned particles in a resin to produce
a continuous ﬁlament, probably the ﬁrst patent with composite material as the motivation.
The wood ﬁber industry saw the introduction of commercial equipment from the Berol

”"3 in 1972 for orienting wood ﬁbers in an electric ﬁeld for the continuous

Corporation
production of ﬂake and ﬁber board. Recently, Knoblachl6 has published a review of the
problems and potential associated with electric ﬁeld alignment of discontinuous ﬁbers. The
principle limitations in developing a high-speed, low-cost method using an electric ﬁeld for
orienting discontinuous ﬁbers rests in (a) the time it takes to orient the ﬁbers in the ﬁeld and

(b) the maximum rate at which the ﬁbers can be deposited. Three processes developed in

the composite materials area which use electric or magnetic ﬁelds are discussed below.

Alignment of Wood F iber:25 This process aligns wood ﬁbers to produce boards for pencil
stock. Wood ﬁbers were dispersed in air and drawn by a vacuum down a chute in which
parallel electric ﬁeld is established. The aligned ﬁbers are collected on a continuous mesh
belt and later consolidated. The same group, working with Morrison-Knudson '5 later
extended the technology to manufacture oriented chopped glass ﬁber mats where a
complicated array of electrodes are embedded at the bottom of the mat that is being formed
to force orientation of the ﬁbers as they descend (Figure 2.5). One of the primary reasons

why they could not

35

 

Process to make oriented chopped ﬁber mats using

Electric fields”.

Figure 2.5

f.

,
aswmwmtgﬁr’... ._
‘ i. "gnu /.

 
 

i§§‘l§\ .

4
2

A

ililll

 

25

Figure 2.6 Process for alignment of ﬁbers using electric ﬁelds in

dielectric fluid”.

36

achieve the high levels of alignment without the elaborate set of electrodes is because they
did not take advantage of the orientation state of the free settling ﬁbers which were coming
under the inﬂuence of the electric ﬁeld. A binder resin is optionally sprayed on the oriented
chopped ﬁber mat and heated to retain the integrity of the mat. But this mat is just a
precursor to another process of resin transfer molding before becoming a ﬁnal composite

part, which reduces the overall speed of processing of a composite part.

Prepreg Bundle Alignment in a Dielectric Fluid: A recent effort in this area has been the
development of a process by Knoblachl7 at MIT. This process produced uniaxially aligned
mats of discontinuous graphite ﬁbers (Figure 2.6). Prepreg bundles are dispersed in a
dielectric ﬂuid settle through an electric ﬁeld. The bundles are aligned with the ﬁeld before
settling on the bottom of the low viscosity ﬂuid tank, to form a mat of prepreg bundles.
After draining the ﬂuid, the mat is consolidated to form aligned composite plate. The use
of dielectric ﬂuid creates additional steps like draining the liquid from the tank ensuring there

is no loss of ﬁber orientation and the subsequent drying of the wet ﬁber mat.

Whisker orientation: Giacomel'8 developed the concept of whisker alignment, in a
transverse direction between ﬁbers or lamina during the curing cycle of the matrix, using
electromagnetic ﬁelds. The whiskers retain their orientation in the ﬁnal composite form
which improves the structural integrity of the laminate. This approach was more for batch

processing than continuous processing.

37

2.4.2 Using Magnetic Fields

In order for magnetic ﬁelds to align ﬁbers the ﬁbers need to be ferro-magnetic. This
puts a serious limitation on the types of ﬁbers that can be aligned for typical composite
manufacture, since majority of the ﬁbers used in composite materials like glass, carbon and
ararnid ﬁbers are not ferromagnetic. These ﬁbers can be aligned if they are specially coated
with ferro-magnetic materials like nickel etc. Due to this very reason, a continuously
operated process that makes oriented discontinuous ﬁber composites does not exist.
Experimental and analytical analysis were conducted by some groups like Shine and
Armstrong'9 of MIT and Hatta et al20 of Japan to investigate the alignment behavior of
ﬁbers coated with magnetic materials. No attempts were made to develop any continuously

operated process to make aligned discontinuous ﬁber composites using magnetic ﬁelds.

2.5 Summary and Discussion
The ﬁber alignment techniques that have been developed so far can be broadly

classiﬁed as either wet/slurry methods or dry methods.

In the wet methodsmz'23 the ﬁbers are usually in a well agitated liquid suspension and
a ﬁber mat is created by either draining the liquid or raising a ﬁlter bed through the
suspension. Control of ﬁber orientation is limited, but can be achieved to some degree by
guiding vanes or other means like electric ﬁelds when the liquid is dielectric”. The
drawbacks in these processes is the introduction of an additional step of drying the wet mat

which reduces the speed of manufacturing drastically, and secondly the fact that ﬁber mat

38

preform has to be further processed by reaction injection molding or sheet impregnation to
result in a composite part. The mechanical performance of the ﬁnal part is sometimes
lowered due to the presence of voids entrapped during the drying of the wet ﬁber mat.

2536:3738” usually rely on electric ﬁelds or pneumatic means30 to control

Dry methods
ﬁber orientation and are generally faster than the wet processes. The main drawback of the
existing processes is the limited understanding of the dynamics of ﬁber motion which has
resulted in elaborate schemes to create electric ﬁelds by the earlier inventors to orient ﬁbers,
thereby reducing process ﬂexibility in changing ﬁber orientation during a run. Another
major hurdle in the speed of fabrication is the fact that the ﬁber mat preform, formed by these

processes, needs to be further processed by reaction injection molding or sheet impregnation

before a composite part can be made.

All the processes described so far cannot be evaluated on a common platform because
of the differences in terms of the ﬁber lengths that were considered and the ﬂow regimes of
individual situations depending upon the ﬂuid medium. Therefore, caution should be
exercised in comparing the degree of ﬁber orientation from one technique to the other.
Often, the ﬁber alignment technique is treated independently and the beneﬁts of processing
ease is lost when the processing technique is evaluated in totality for manufacturing aligned
discontinuous ﬁber composite. Some of the limitations that exist in terms of the desirables
for composites is:

- ability to align long ﬁbers at high volume fractions

39

- high speeds of ﬁber alignment

- high speeds of the ﬁnal composite manufacture
Recent development of a high-speed MSU Powder Prepreg Process to make composite
materials by Vyakarnam and Drzal“, using an entrained aerosol of ﬁne tribo-charged
polymer particles to coat the ﬁbers, along with the known phenomenon of ﬁber orientation
in electric ﬁelds has created a basis upon which an Aligned Discontinuous Fiber (ADF)
composites process can be developed to potentially overcome all the above limitations. The

details of the ADF process development will be taken up in Chapter 4.

10.

ll.

12.

13.

14.

References

Drzal, L. T., Padaki, S., Vyakarnam, M. N., Femandes, J. H., “Powder
lmpregnation of Advanced Composite Materials”, in Polymer Powder
Technology, Eds. Narkis, M. and Rosenzweig, N., John Wiley & Sons, 1995.

S. R. Iyer and L. T. Drzal, J. of Thermoplastic Composite Materials, Vol 3, pp
325-355 (1990)

S. Iyer, L. T. Drzal, and K. Jayaraman, "Method for Coating Fibers with Particles
By Fluidization in a Gas ", US Patent Nos. 5,102,690; 5,123,373; 5,128,199
(1992)

M. N. Vyakarnam and L. T. Drzal, Apparatus and High Speed Method for
Coating Elongated Fibers, US Patent no. 5,310,582 (1994)

M. N. Vyakarnam and L. T. Drzal, Polymer Processing Society (C onf. Proc. ) pp
31-32 (1992)

C. F. Galo, and W. L. Lama, J. of Electrostatics, Vol. 2, pp 145-150 (1976)
A. G. Bailey, Powder Technology, Vol. 37, p 71-85, (1984)
Bagg, G. E. G., Evans, M. E. N., and Pryde, A. W. H., Composites Vol 1, 1969.

Kacir, L., Narkis, M., and lshai, 0., "Oriented Short Glass-Fiber Composites",
Polymer Engineering and Science, Vol 15, p 525, 532, 1975.

Soh, S. K., “Slurry process for Preform Manufacture”, Proc. 10th Annual
ASM/ESD ACCE, Dearbom, 1994.

Jander, M., “Industrial RTM - new developments in molding and preforming
technologies”, Proc. 10th Annual ASM/ESD ACCE, Dearbom, 1991.

Serban, P. and Petty, C., “Flow control of chopped glass ﬁber bundles through
ﬂexible tubes”, Presented at AIChE Annual Meeting, Miami, 1995.

Tucker, C. L. and Advani, S. 6., “Processing of Short Fiber Systems” in Flow and
Rheology in Polymer Composites Manufacturing, Ed. S. G. Advani, Elsevier
Science, 1994.

Goettler, L. A., Leib, R. 1., and Lambright, A. J ., Rubber Chem. Technol, 52, 838,
1979.

40

15.

16.

17.

l8.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

41

Isard, J. 0., “The Orientation of Fibers in Electric Field”, British Journal of
Applied Physics, Vol 6, p 176-179, 1955.

Knoblach, G. M., "Processing, Electric Fields for Fiber Orientation", in
Encyclopedia of Composite Materials, McGraw-Hill, p. 413-424,1991.

Knoblach, G. M., "Using Electric Fields to Control Fiber Orientation During the
Manufacturing of Composite Materials", SAMPE J., vol. 25, No.6, p9, 1989.

Giacomel, J. A., US Patent No. 4,560,603, 1983.

Shine, A. D. And Armstrong. R.C., “Experimental studies of suspended
ferromagnetic ﬁbers in a magnetic ﬁeld”, Rheologica Acta, Vol. 26, p 162-171,
1987.

Yamashita, S. et. al., “Fiber Orientation Control of Short Fiber Composites:
Experiment”, J. Of Composite Materials, Vol. 23, p 32-41 , 1989.

Bagg, G. E. G., Evans, M. E. N., and Pryde, A. W. H., Composites Vol 1, 1969.

Kacir, L., Narkis, M., and lshai, 0., "Oriented Short Glass-Fiber Composites",
Polymer Engineering and Science, Vol 15, p 525, 532, 1975; Vol 17, p 234, 1977;
Vol 18, p 45, 1978.

Soh, S. K., "Slurry Process for Preforrn Manufacture", Proc. 1011:. Annual
ASM/ESD Advanced Composites Conference & Exposition, Dearbon, 1994.

Knoblach, G. M., "Using Electric Fields to Control Fiber Orientation During the
Manufacturing of Composite Materials", SAMPE J., vol. 25, No.6, p9, 1989.

Talbot, J. W. and Logan, J. D., USPatentNo. 4,113,812, 1978.

Carpenter, C. T. and Stunkard, N. W., US Patent No. 4,111,294, 1978.
Logan, J. D., US Patent No. 4,432,916, 1984.

Talbot, J. W., Peters, T. E. and Logan, J. D., US Patent No. 4, 664, 856. 1987.
Peters, T. E. at al, U. S. Patent No. 5,017,312, 1991.

Ericson, M. L. and Berglund, L. A., Composites Science and Technology, 49 p
121-130,1993.

Vyakarnam, M. N. and Drzal, L. T., "Apparatus and High Speed Method for
Coating Elongated Fibers", US. Patent No. 5,310,582, 1994.

Chapter 3

 

Fiber Alignment Using Electric Fields

3.1 Introduction ..................................................... 43
3.1.1 Electric or Magnetic Fields? .................................. 44

3.1.2 Fiber Dimensions ........................................... 46

3.2 Theory of Fiber Orientation ......................................... 48
3.2.1 E-Field Polarization ......................................... 49

3.2.2 E-Field Alignment Time ..................................... 52

3.2.3 Fiber Hydrodynamics ........................................ 55

3.2.3.1 Orientation Behavior of Fibers During Free Settling ......... 55

3.2.3.2 Fiber Motion in Shear Flows ........................... 58

3.2.3.3 Settling Velocity of Fibers ............................. 59

3 .3 Experimental Studies .............................................. 61
3.3.1 Experimental Setup ......................................... 62

3.3.2 Behavior in DC. Fields ...................................... 65

3.3.3 Behavior in AC. Fields ...................................... 69

3.3.3.1 Effect of E-ﬁeld Intensity and Fiber Aspect Ratio ........... 70

3.3.4 Fiber Settling Behavior ...................................... 82

3.4 Summary of Results ............................................... 86
References ............................................................ 88

42

43

3.1 Introduction

In the proposed Aligned Discontinuous Fiber (ADF) process, ﬁbers are fed from the
top of an orientation chamber which consists of an electrode conﬁguration to align ﬁbers as
they settle under gravity. The ﬁbers are aligned in the direction of the applied electric ﬁeld.
The physics of ﬁber orientation in electric ﬁelds has been previously investigated as pointed
out in § 2.4. However, in almost all the previous studies the experiments were so designed
that the dimensions of the ﬁbers and/or the ﬂuid medium in which they were immersed were
so chosen that the ﬁber motion fell under viscous ﬂow regimes, thereby amenable to
analytical treatment and/or accurate experimental veriﬁcation. The objective of this research
is to align ﬁbers using electric ﬁelds in a manner that would be useful in manufacturing
discontinuous ﬁber polymer composites, whose properties approach those of continuous ﬁber
composites. The phenomenon of ﬁber alignment in air, especially of ﬁbers of dimensions
that are useful in composite materials, offers tremendous technological signiﬁcance by
providing the scope for rapid solvent free processing in a variety of situations. Excepting
some literature in the ﬁeld of aerosol science for ﬁbrous dispersions (sizes ranging from a
micron to several hundred microns), there is little published literature on the orientation
behavior of ﬁbers in air for lengths ranging from a millimeter to 50 millimeters. The
approach taken in characterizing and understanding the ﬁber alignment behavior in electric
ﬁelds is by analyzing the theoretical situations that have been developed so far and making

quantitative or qualitative extensions from them by careful experimental design.

44

E-glass ﬁbers are chosen for developing the ADF process due to their extensive use
in discontinuous ﬁber composites compared to carbon ﬁbers (Table 3.1). Another important
distinction between glass and carbon ﬁbers is in the electrical conductivity which will play
a role in their behavior in electric ﬁelds. Glass ﬁbers are dielectric in nature which will make
it longer to align in electric ﬁelds compared to carbon ﬁbers which are conductive. However,
once the processing principles for dielectric ﬁbers in electric ﬁelds are established, it will be
relatively easy to translate the principles to carbon ﬁbers rather than going vice-versa. Before
the theory of ﬁber orientation is presented, the reason for choosing electric ﬁelds is
discussed. Also the limits to ﬁber dimensions and ﬂow regimes that are required for the

development of a meaningful high speed processing technique are outlined.

3.1.1 Electric or Magnetic Fields?

The ﬁrst question that needed to be addressed in the quest for a rapid ﬁber alignment
technique was which ﬁelds (electric or magnetic) would be most appropriate. Despite the
similarities, the differences have a strong bearing on the utilization of one type of ﬁeld over
the other especially when it comes to polymer composites. Glass, carbon and aramid ﬁbers
which form the bulk of the reinforcements in composite applications cannot be aligned using
magnetic ﬁelds unless they are coated with a ferro-magnetic material. Electric ﬁelds on the
other hand can align any axisymmetric body due to polarization, as long as there is a

dielectric mis-match between the body and the ﬂuid medium.

45

Table 3.1 Pertinent Fiber Properties

 

 

Fiber Sp. Diameter Critical Dielectric
Gr. mm) Lengthﬂgm) gonstgntj;
E-Glass 2.55 10 - 13 400 - 1000 5.9-6.6
AS4 Carbon 1.8 8.0 400 - 600 -
Aramid" 1.45 12.0 800 - 1200 3.6

 

* Approximate ranges with epoxy matrix systems' ; ** Kevlar-49 (DuPont); '1' at 1 MHZ

Secondly, electric ﬁelds are inexpensive and easy to generate. All that is needed to
generate an electric ﬁeld is a high voltage source and an electrode made out of a conducting
material. The geometry and the position of the electrodes can be easily designed to create
a desired E-ﬁeld conﬁguration. In the case of magnetic ﬁelds, large coils are required and

it is more difﬁcult and cumbersome to conﬁgure a desired ﬁeld geometry.

An important distinction between the two ﬁelds lies in the way they breakdown in
a ﬂuid medium. As voltage is increased beyond what is termed a “breakdown” voltage,
electric ﬁelds ionize the molecules and create an avalanche of electrons, which tries to ﬁnd
the nearest positive electrode or a grounding point to discharge. This effect is commonly
referred as the corona discharge. The theoretical breakdown in air occurs at a ﬁeld intensity
of 3000 KV/m over a distance of more than about 1 millimeter”. This value of breakdown
voltage is a strong function of the gas pressure; the gap width between electrodes and

humidity. Under practical conditions in air, it is difﬁcult to operate at greater that 1000

46

KV/m without frequent ﬁeld breakdown. The breakdown voltage establishes an upper limit
on the maximum torque that can be applied to align a body. In the case of a magnetic ﬁeld,
there is no such limitation due to a ﬁeld breakdown. Torques which are several orders of
magnitude higher than what can be obtained from electric ﬁelds in air can be easily obtained
by magnetic ﬁelds in practical situations. Magnetic ﬁelds are the choice when the material

is magnetic like nickel or steel ﬁbers.

3.1.2 Fiber Dimensions

The length of the ﬁbers that will be used in the investigations is determined using the
critical ﬁber length concept that was introduced in §1.2.1. One of the objective of this
research was to determine if the properties of aligned discontinuous ﬁber composites made
by the ADF process will approach the properties of continuous ﬁber systems. This stipulates
that the length of the ﬁbers be at least 10 times the critical ﬁber length. The ﬁber dimensions
that are of interest in discontinuous ﬁber polymer composites are typically larger than those
encountered in electro-rheological ﬂuids, whiskers reinforcements and ﬁbrous aerosols. A
priori it is difﬁcult to determine precisely the critical ﬁber length for any given ﬁber-matrix
system, because of the experimental difﬁculties involved and the variations in the interfacial
adhesion". However, since most polymers have mechanical properties of the same order of
magnitude (see Table 1.2), a reasonable extrapolation from E-glass ﬁber-epoxy data can be

made to estimate the ﬁber lengths that will be needed in this study (Table 3.1)

47

While the optimum length of the ﬁbers for the ADF process should be long enough
so that reinforcement properties approach the properties of continuous ﬁber composites, they
should also be short enough so that shaping of the composite over small radii of curvature
is possible. Based on this reasoning, ﬁbers of nominal lengths ranging from 3 to 25 mm
were selected for this study. These ﬁbers are a few times greater than critical ﬁber length to

greater than 10 times the critical ﬁber length (Table 3.1).

Discontinuous glass ﬁbers are commercially available as chopped ﬁbers. Chopped
ﬁbers are aggregates (bundles) of individual ﬁbers (or ﬁlaments) held together by a sizing.
Commercial glass ﬁbers are always sized to protect the ﬁbers from damage during handling
and also to provide a good interface between the ﬁber and the matrix. The typical length of
chopped ﬁbers ranges from 0.125 to 3 inches and the number of ﬁlaments per bundle can
vary from a few hundred to a few thousand. During manufacturing operations like extrusion
and injection molding, chopped glass ﬁbers are broken down to shorter lengths as well as get
dispersed into individual ﬁlaments due to the shearing action encountered. The reason for
widespread commercial availability of chopped ﬁbers is because individual ﬁlaments in
aggregates or bundles have integrity and are easy to handle in mechanical operations. It is
almost impossible to handle short ﬁbers if they are in ﬁlament form due to the small ﬁber
diameters (10 - 13 microns). The ﬁbers will tend to form fuzzy and entangled masses with

little possibility to be separated and aligned into individual ﬁbers.

48

Another aspect that needs to be considered is the effect of the aggregate nature of
chopped ﬁbers in the alignment behavior as well as the reinforcement behavior. The
effective aspect ratio of the ﬁber bundle is a direct function of the aggregate nature of ﬁbers
i.e the size of the ﬁber bundle. From a micro-mechanics view point, only the length of an
individual ﬁlament is important and not the size of the bundle as long as each ﬁlament is
surrounded by matrix. From a processing viewpoint the bundle size affects the process in
two signiﬁcant ways. First, the number of ﬁlaments in the ﬁber bundle will affect both the
E-ﬁeld alignment times and the hydrodynamics of ﬁber motion since both these are functions
of ﬁber aspect ratio. Secondly the bundle size will affect resin impregnation into the ﬁber
bundle in the consolidation step. Ideally, the ﬁber bundles should have as small a number
of ﬁlaments as possible for bundle integrity and mechanical handling, so that individual
ﬁlaments do not easily separate and create a fuzz. Commercially available chopped ﬁber
bundles typically have about 1000 ﬁlaments or more. Preliminary analysis indicated that
ﬁber bundles of that size may be difﬁcult to align in electric ﬁelds in air, therefore, bundles
with ﬁlaments ranging from 100 to 400 were obtained from Owens-Coming and Vetrotex

CertainTeed.

3.2 Theory of Fiber Orientation

In the ﬁber alignment process three things are taking place, all of which contribute in
effective ﬁber alignment in air using electric ﬁelds.

0 Fiber polarization and alignment in E-ﬁelds.

0 Hydrodynamic resistance.

49

0 Initial ﬁber orientation state, when the ﬁbers are entering the orientation chamber and

the ﬁnal steady-state orientation behavior of free falling ﬁbers.

Each of the above events is discussed separately ﬁrst, but it will be shown later
through experimental results and theoretical analysis that the ﬁber alignment process is in
actuality a combination of all the three events. The purpose of this section is to provide a
theoretical framework on the ﬁber alignment process based on prior studies and extensions
from those results for the cases under study in the dissertation. Experiments that are
conducted to verify some of these concepts and their validity for the ﬁbers that are of interest

in composites manufacture will be taken up in section §3.3.

3.2.1 E-Field Polarization

Conductive as well as dielectric ﬁbers can be aligned in an electric ﬁeld as long as
there is polarization of the ﬁber, which can be achieved when there is a difference between
the dielectric constant of the ﬁber and the medium surrounding it. Conductive ﬁbers are
polarized to a greater degree than non-conductive (dielectric) ﬁbers due to the presence of
free electrons which allow a rapid migration of free charges (of the order of 10“3 seconds)
towards the end. In the case of a dielectric particle only molecular aligmnent can take place.
The force diagram on a single ﬁber orienting in an electric ﬁeld while settling due to the
gravitational ﬁeld in air is shown in Figure 3.1. Electric ﬁelds induce a dipole in the ﬁber
which appears as surface charges at its ends. Columbic forces between the ends and the ﬁeld

causes an electric torque (TE) which causes the ﬁbers to align parallel to the ﬁeld.

50

Fiber

 

 

 

Figure 3.1 Forces acting on a ﬁber in the orientation chamber.

51

This torque is balanced by the hydrodynamic (viscous) drag on the rotating ﬁber (TH).
Consider a neutrally buoyant ﬁber in the ﬂuid medium in the absence of inertial effects, then

the two torques balance.

”=0 (1)

The torque due to an E-ﬁeld of intensity, E0 on a dielectric prolate ellipsoid in a dielectric
ﬂuid is derived using energy arguments5 and the equation is a function of the volume (V) of

the particle and the polarization parameter, WE. When 0 = 0, TE is given by (see Figure 3.1):

TE 2 V WE Sill 2(1) (2)

2
WE = K260E0(q — 1)2F (3)

Polarization parameter, WE depends on the dielectric constant of the ﬁber (KI) and the ﬂuid
medium (K2) and E0. “F” is a factor related to the ratio of dielectric constants (q) of the ﬁber
to the ﬂuid medium (q=r<,/ 1(2); “A” is an elliptical integral constant and “e” is a function of

the aspect ratio (r,3 = a/b) of the ellipsoid.

_ 1.5A — 1
2 [1 + (q — 1><A — 1)] [1 + (q - 1W2] (4)

 

52

 

1+e
l—e) . l
2r2e3 r2 (5)

2e-ln(

 

The balancing between hydrodynamic forces and electric torques on an ellipsoid results in
the rotation of the particle and eventual alignment with the E-ﬁeld. The time taken to align
a ﬁber based on its electrical nature and the electrical ﬁeld intensity is presented the next

section.

3.2.2 E-Field Alignment Time

Demetriades6 derived the equation for the rotation of a prolate ellipsoid, assuming
inertial effects are negligible, particles are neutrally buoyant and Reynolds number is far less
than 1. Mason and co-workers had done some extensive theoretical studies and experimental
observations on the effect of electric ﬁelds on particle motion for a variety of ﬂow situations
and summarized the results in a review article7. It was experimentally found by them that
the analytical model derived by Demetriades to determine the aligmnent time (t) for a
dielectric elliptical body to orient from an initial angle (1),- to (I)f along the electric ﬁeld lines
of intensity Eo holds good under the given assumptions and is given by the following

relationship.

, z *1 [ ln(tan¢f) - ln(tan<i>,) 1
8neosz:P(q,re) (6)

 

53

The polarization function P(q,r,) is a related to the electrical and geometrical nature of the

ellipsoid. For dielectric ellipsoids it is:

(3A - 2)(€I - 1)2Q(re)

 

P ,,> =
W 8n12 + (q — 1)A][(q - 1)A — q] (7)
.5. r .2
q K2 8 b

For conductive ellipsoids, q—~o<, P(q,rc) is given by:

(3A - 2) Q(r,)
811: A(A — 1) (9)

 

P(q.re) =

The parameter Q (re) is related to the elliptical integral function (A) and the aspect ratio of

the ellipsoid (re).

 

.. * I) (10)

Alignment times are directly proportional to the viscosity of the ﬂuid medium and inversely
proportional to the square of the electric ﬁeld strength. For example, a conductive carbon
ﬁber has a dielectric constant of inﬁnity while an insulating glass ﬁber has a value of 6.2.
Dry air has a dielectric constant of 1. Both ﬁbers would be capable of being oriented

although the carbon ﬁber would respond much faster. This relationship (Eq. 6) shows that

54

the alignment time is a very strong function of E-ﬁeld intensity and the polarization function
P. P is in turn related to the ratio of dielectric constants of the ﬁber and the medium and the
aspect ratio. Better ﬁber alignments are possible at higher E0. However, it is to be noted that
this equation can only serve as a ﬁrst approximation to determine the alignment times of the
ﬁbers that can be used in the ADF process. This is because the equation is valid for ﬁber
motions in the viscous ﬂow regime (implying very small ﬁbers or a very viscous medium),
whereas the glass ﬁbers that are being experimented in the setup are relatively large.
Besides, the ﬂuid medium being air results in the ﬁber motions to lie beyond the viscous
ﬂow regime. Mason and co workers7 have veriﬁed this theory for a number of different types
of particles in the colloidal and non-colloidal ranges (largest size tested with this model was

900 pm) under shear ﬂow conditions.

The theoretical principles underlying the alignment of air borne particles in an electric
ﬁeld have also been studied by F uchss, where the hydrodynamic resistance provided by air
is neglected. Therefore, torque resulting from shear forces associated with laminar ﬂows is
neglected. Further, Fuchs considers most airborne particles to be conductive because of the
moisture adsorbed on the surfaces when the particles are exposed to ambient conditions of
relative humidities greater than 30 %. The equation for angular or rotational mobility of
prolate ellipsoids was derived and the alignment time (after simpliﬁcation for large aspect
ratios) is given as:

_ 71
I _
1.51teoE: cos¢sind> (1 1)

55

In this equation, notice there is no dependence on the aspect ratio of the ﬁber. When a ﬁber
is aligned from (1),: 85° to (I)f = 5°, the time to align any particle is 4 x 10'5 s and l x 10'5 s
for an E ﬁeld intensity of 250 and 500 KV/m respectively. If the conduction effect is
neglected then the particle is treated as a dielectric and the alignment time increases several
orders of magnitude to around a second. Lilienfeld", veriﬁed the alignment of ﬁbers that
were up to 20 microns in length in air and found reasonable agreement with Fuchs’s

contention that air borne particles (with RH > 30%), behave like conductive particles.

3.2.3 Fiber Hydrodynamics
3.2.3.1 Orientation Behavior of Fibers During Free Settling

Does a ﬁber fall vertically or horizontally during settling in an unbounded ﬂuid
medium? In other words do ﬁbers have any tendency to fall in a preferred orientation state?
Answers to these questions will be ascertained by considering the case of a single ﬁber in a
dilute situation, since it is quite difﬁcult to predict the orientation state of a suspension of
ﬁbers with ﬁber-ﬁber interactions. An insight into these questions will be very useful in
understanding the alignment behavior of ﬁbers especially because of the non-buoyant nature
of the ﬁbers. Qualiﬁed arguments are made which are supported by both theoretical and

10,11,12

experimental investigations The orientation behavior of ellipsoidal particles is

analyzed in the following three ﬂow regimes based on Reynolds number (Re) of the particle.

d V
Re : f P
T1 (12)

 

56

Where (If is the ﬁber diameter (or effective bundle diameter), V is the ﬁber velocity, where
p is the ﬂuid density and n is the ﬂuid viscosity. Figure 3.2 mimics the orientation behavior
of ﬁbers in two dimensions as they settle in the different ﬂow regimes. This ﬁgure depicts
initial orientation state (vertical, horizontal and oblique at 45°) of the ﬁber when it is released

and the ﬁnal orientation state of the ﬁber that is attained at steady state.

- Viscous Regime (Re < 1): When an ellipsoidal particle or ﬁber is falling very
slowing through a resisting medium then the orientation state of the body remains
unchanged and maintains the initial state of orientation in which it was released.
This holds good for any initial state of orientation (see Figure 3.2). This is because
the resultant torque due to viscous forces acts through the center of the ellipsoid
which nulliﬁes any coupling effect. Certainly, the ellipsoid should be perfectly
symmetrical for this phenomenon to hold good. This is the regime where creeping
ﬂow conditions apply and most of the analytical modeling work pertains to this

regime.

0 Intermediate Regime (Re ~ 1): As Re gets greater than 0.1, the nature of the
ellipsoidal motion begins to change. The particles start to orient themselves in such
a way that the drag forces are balanced with the gravitational forces and they occupy
an orientation state which has the maximum resistance to free fall. In other words
a ﬁber will tend to orient itself with its long axis normal to the direction of free fall.

The orienting force increases with increase in Re until a stable orientation is

57

 

 

Initial State

 

 

 

 

 

 

 

 

 

 

Re<0.1 0.1<Re<25 Re>25
Viscous Regime Intermediate Regime Inertial Regime

Figure 3.2 Schematic representation of the orientation state in free falling ﬁbers.

58

achieved at some value of Re between 10 to 100. This kind of stability in the
orientation state is also in agreement with the theoretical predictions one gets from
potential ﬂow conditions. Under these conditions, a couple acts on an ellipsoidal

particle to orient normal to the ﬂow direction.

- Inertial Regime (Re > 1): As the Re value increases above one and passes through
the stable motion regime, there lies a critical Re where vortex shedding starts to take
place and starts to get difﬁcult to predict the orientation state of a ﬁber. This can be
anywhere from a Re value of 25 to 100. At Re values beyond this region, turbulent
effects also play a role and it once again becomes very difﬁcult to predict accurately

the orientation state.

3.2.3.2 Fiber Motion in Shear Flows

In a recent book chapter by Tucker and Advani”, a review of the current research that
surround ﬂow induced ﬁber orientation prediction in short ﬁber composite processing
pertaining to molding operations has been presented. Although, the ﬁber alignment and
motion in the ADF process is not in the viscous ﬂow regimes, prior research in this area
provides useful insights into the behavior of ﬁbers in the process developed. Jeffery'4 solved
the problem of the motion of a single ellipsoidal particle in a shear ﬁeld and found that
ellipsoidal particles align in the direction of shear ﬂow and transverse to ﬂow direction in

elongational ﬂows. Further, Jeffery’s equation predicts that a single particle in a simple

59
shear ﬂow (vl = ze; v2=v3=0) will undergo a periodic rotation, which is a strong function

of aspect ratio of the particle.

2n 1
T:__r+_
C(e )

re (13)

Where T is the time period of rotation and rc is an equivalent aspect ratio. Jeffery’s equations
have been experimentally veriﬁed'5 and found that the predictions of this model to be
accurate when all the assumptions are met. An important conclusion that one can draw from
the above relationship is that as the aspect ratio (length) of the ﬁber increases, the period of
rotation increases in a shear ﬂow. Due to the large difference in the densities between glass
ﬁbers and air, the ﬁber motion is far from neutrally buoyant and shear ﬂow situations arise,
because of the translation and rotation of the ﬁber during free fall. If we apply the
phenomenon of ﬁber motion in shear ﬂow as predicted by Jeffery, then it may be concluded
that the stability of the ﬁber motion increases as the length of the ﬁber increases. And
stability of ﬁber motion improves ﬁber alignment for non-buoyant particles. Based on these
arguments it will be shown in § 3.3.3.1 that 1" long ﬁbers align better than 1/4" long ﬁbers

in the ADF process.

3.2.3.3 Settling Velocity of Fibers
The dimensions of the ﬁbers under investigation (a few millimeters to 50 millimeters

in length) are beyond the region where Brownian effects are encountered, hence these effects

60

can be safely neglected from the present analysis. Extensive analytical treatment and
experimental work has been done to determine drag on non-spherical particles when they
settle in the viscous regimes. For non spherical particles, orientation and the shape factor
become critical in the determination of the drag coefﬁcient. There are two limiting
conditions for the state of ﬁber orientation during the settling of a cylinder: (a) cylinder axis
parallel to the direction of ﬂuid ﬂow (Vertical case), and (b) cylinder axis normal to the
direction of ﬂuid ﬂow (Horizontal case). Analytical expressions for drag force F, (with
subscript V and H representing the vertical and horizontal cases respectively) have been
obtained for ellipsoids by Oberbeck in 1876. Batchelor'° offers a model to predict the drag
on cylinders based on slender body theory. The drag force of a cylinder settling under the
extreme cases of horizontal and vertical orientations can be used to estimate the settling
times of ﬁbers of various lengths and diameters. Dynamic shape factors k are obtained using
the simple relations which follow, where "a" is the diameter and "L" length of the cylinder

11 11

respectively; re is the aspect ratio; 6. = ln(L/2 re); and the subscript "ve" stands for

equivalent sphere terms.

1/3

 

__F_.
II F... W (14)

e.+0.307e:2 3
. 2:: Lu — + 0.4266
I 11 [ 1-e/2 ]

“I:
II

<—:+0.307e2
H 47”]14” [W + 0.11963] (15)

h“J
11

61

Using the above equations with d,c = a(1.5 re)”, the shape factors for the cylinder become

6 +0.307t—:2

e +0.307e2
1+ /2

k, = (2/3)4’3rf’3 [ + 0.4266]

kH = 2(2/3)4’3rj’3 [ + 0.1196] (16)

Fibers under investigation can be treated as slender bodies, and it would be sufﬁcient for the
present study to use the appropriate relations developed above to determine the drag on a
ﬁber. The hydrodynamic drag on the ﬁber can be estimated if the settling velocity of the
ﬁber is known. In section § 3.3.4 the settling velocities of the ﬁbers which are

experimentally determined are presented.

3.3 Experimental Studies

In the presentation on the theory of ﬁber orientation in electric ﬁelds, various aspects
were discussed that affect the alignment process. However, it was realized that there were
several limitations in the prior studies (as discussed below) which made it important to
experimentally verify the alignment behavior of ﬁbers in air before designing a process that

can manufacture aligned discontinuous ﬁber composites.

0 First, the ﬁber alignment time derivations were based on small ellipsoidal particles
in creeping ﬂows. The dimensions of the ﬁbers that are going to be used in

composite manufacture are at least an order of magnitude larger than the ﬁbers that

3.3.1

62

were experimentally veriﬁed by the models discussed above by previous

investigators.

Secondly, the alignment behavior was characterized for neutrally buoyant particles.
In the case where ﬁbers like glass or carbon are to be aligned in air, the density mis-
match between the ﬁbers and the ﬂuid medium (air) is extremely large (~ 103). So,
it is expected that rotational motions of the ﬁbers in the three dimensions will play

a role in the ﬁber alignment process.

Thirdly, all the models were derived for the case of a single ﬁber aligning in a ﬂuid
medium. In real processing situations, ﬁber-ﬁber and ﬁber-wall interactions will
play a major role in the effectiveness of the alignment process. This is an extremely
complicated problem to be solved from ﬁrst principles, hence experimental studies

were conducted which shed some light on this aspect.

Experimental Setup

The critical aspect in the ADF process prototype development has been the ability to

control ﬁber orientation and determine the various parameters that inﬂuence it. An

experimental setup was built to study the orientation behavior of ﬁbers as they settle due to

gravitational force in electric ﬁelds with air as the ﬂuid medium (Figure 3.3). A variety of

glass ﬁbers of lengths ranging from 0.125" to 1" have been investigated in this setup. The

forces that are encountered by the ﬁber are shown in Figure 3.1. The key parameters

63

 

 

 

 

 

 

High Speed
#3 Video Camera

 

 

 

 

1:
Fiber Feed Point E

 

 
 
  

High Speed
Video Camera

I} :33 I

Observation Points

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Adjustable. , _ . *' ’ "

Electrodes K 1 1 El P BM
Motion Analyzer

I-lipatronics RM25 E F'be ' i
High Voltage .ionu Deplosifion III-III..." Glow L31)
Image Analysis
Transformer Plate
Data Processing

Figure 3.3 Experimental setup to investigate ﬁber alignment in electric ﬁelds.

64

investigated are the ﬁber alignment times, ﬁber settling behavior and ﬁber orientation

distributions (FODs) as a function of E-ﬁeld intensity and ﬁber geometric dimensions.

The experimental setup comprises of an open ended rectangular chamber with two
vertical parallel copper electrode plates of dimensions 50 cm x 20 cm. These electrodes are
supported by an insulating ﬁberglass framework. The electrodes are mechanically held in
a groove provided in a set of insulating (phenolic) brackets. The insulating framework and
other ﬁxtures are fastened by plastic screws to avoid any distortion in the electric ﬁeld due

to the presence of conducting material.

The setup is so designed that the gap width between the electrodes can be easily
varied from 3 to 10 cm in steps of 1 cm. The electrode gap width can be varied to account
for large ﬁbers and to control E-ﬁeld intensity. The ﬁbers are fed from the top in a
controlled fashion and deposited on a bottom plate. The bottom plate on which the ﬁbers are
deposited can be located at two different heights (B and C), so that experiments can be

performed at two different electrode heights of 25 and 50 cm.

A high speed video camera and Kodak EktaPro EM high speed motion analyzer is
used to study the electro-dynamics of ﬁber motions. Recording up to 12,000 frames per
second can be achieved by this system but a recording rate of 1000 frames per second was
found to be adequate for this study. The recordings are stored in a video cassette recorder

and printed on a thermal paper frame by frame. A provision was established to download

65

the images from the high speed motion analyzer into a computer through a frame grabber
board. These individual digital images are further processed using Global Lab Image

analysis software'7 to determine ﬁber orientations or ﬁber translations.

The high voltage source was generated using a RM 25 Hipatronics transformer. In
view of the high voltages that this transformer would generate and the potential for
breakdown in air, it was not operated in air. The transformer is housed in a special tank that
contains silicone oil, which is a dielectric ﬂuid that has a greater resistance to voltage
breakdown than air. A variac was used to control the electric ﬁeld intensity (E) between the
parallel plate electrodes. The electrode gap width was changed when it was felt necessary

to maintain at least a 1:3 ratio between the ﬁber length and the gap width.

3.3.2 Behavior in D.C. Fields

The behavior of ﬁbers in electric ﬁelds was observed for the ﬁrst time in this study
using D.C. ﬁelds. The observations were made in the experimental setup that was connected
to a high voltage source generated from Hipatronics 60 KV D.C. power supply. The
objective was to bracket the conditions of the electric ﬁelds under which the ﬁbers would
align. The gap width of the electrodes and the voltage was controlled to give a give a net
variation in ﬁeld intensity from 130 KV/m to 600 KV/m. The polarity of the high voltage

electrode was tried with both —ve and +ve settings.

66

The behavior of four different types of glass ﬁbers were tested by dropping them from
the top of the orientation chamber in a random fashion. One was commercial grade chopped
ﬁbers (PPG 3830 of length 0.19 inches) which are meant to be used in molding operations
and the other three types of ﬁbers were hand cut to half inch length from continuous ﬁber
rovings (see Table 3.2). General conclusions were arrived on the behavior of these glass

ﬁbers in D.C. ﬁelds by observations as noted in Table 3.2.

The general conclusion obtained from these scouting experiments was that glass
ﬁbers came under a strong electrophoretic force in D.C. ﬁelds, which disturbed any
alignment achieved. The ﬁbers got charged in the presence of high intensity ﬁelds and
thereby got attracted to either the + ve or - ve electrode based on the sign of the charge the
ﬁbers acquired (Table 3.2). There seemed to be some relation between the type of sizing and
the type of charge it acquired. This is apparent from the fact that the second and third type
of ﬁbers got attracted towards the -ve electrode while the fourth one moved towards the

ground plate.

This ﬁrst set of experiments, were encouraging in terms of the fact that E-ﬁelds were
having an effect on the ﬁbers of dimensions that are of interest for composite materials and
there seemed to be a potential for ﬁber alignment. In other words electrical torques could
be seen in action. The electrophoretic behavior of the ﬁbers was however, complicated
because it was not clear if the glass ﬁbers were charged prior to their entry into the

orientation chamber or if they got charged the moment they came under the inﬂuence of the

67

Table 3.2 Behavior of Fibers in D.C. Fields

 

 

 

Fiber Type Observations

Chopped glass ﬁber Appeared too heavy and little effect of the ﬁeld could be

strands (3/ 16") seen. Fell at the bottom of the chamber without any
alignment.

Water sized glass ﬁber Came under electrophoretic effect and the ﬁbers driﬁed to
strands (‘/2") the high voltage plate (which was at - 45 KV). This drifting
occurred near the top of the chamber.

Polyester sized glass Effect of the ﬁeld was observed. Drift was towards the -ve
ﬁber strands (‘/2") electrode.

Minimally sized glass Strongly driﬁed towards the electrode but this time towards
ﬁber strands (‘/2") the ground (or away from the -ve electrode)

ﬁeld. If it is assumed that the drift is due to the charging of the ﬁbers as they come under the
inﬂuence of the ﬁeld, then this phenomenon can be explained by the nature of dielectric

polarization.

Glass ﬁbers being dielectric in nature and with the possibility of surface moisture and
ionic moieties on the surface, it can be expected that the dielectric constant is a strong
function of the frequency of the electric ﬁeld. Depending on the frequency (f) of the ﬁeld,
polarization of dielectrics takes place in two stages. First, the moment the body comes under
the inﬂuence of the ﬁeld, instantaneous polarization due to electronic and atomic effects

(electronic polarization) takes place. Later on the slower processes due to dipole orientation

68

and/or ionic migration (lossy polarization) takes place. With sufficient time the polarization

of a dielectric material will reach saturation levels's'”.

In the case of D.C. ﬁelds (f = 0), saturation polarization will be attained i.e both
electronic as well as dipole polarizations will take place and this results in a strong electrical
torque acting on the ﬁber. The strong electrophoretic effect observed in D.C. ﬁelds can be
attributed to the strong electrical torque acting on the ﬁber combined with the negligible
viscous resistance offered by air. If the time to attain lossy polarization is far less than 1/f,
then it is expected that saturation polarization will be attained. However, if the time to orient
the dipoles is greater than l/f, then there will be no polarization due to the movement of
permanent dipoles. Under A.C. ﬁelds, with the ability to control the frequency, f, it is
possible that above a certain frequency the degree of polarization can be lesser than that
obtained with D.C. ﬁelds. This can turn out to be more suitable for an effective ﬁber

alignment process, especially when the ﬂuid resistance is very low.

Some general conclusions were drawn from this set of experiments:
0 Glass ﬁbers of dimensions that are of interest in the composite materials have the

potential to be aligned in electric ﬁelds in air as the medium.

- Fibers come under a strong electrophoretic force in the D.C. ﬁelds and thereby drift.
So, it may be concluded that AC. ﬁelds would provide a better choice to align ﬁbers.

The frequency of the AC. ﬁeld at which optimum alignment is possible can be

69

estimated from the polarization times of the ﬁber material and its surface
characteristics. Initial calculations indicated a low frequency of 60 Hz should work
for this application, since polarization times of these ﬁbers are estimated to be of the
order of 10" sec”. Secondly, the strategy was to investigate with AC. ﬁelds at a
frequency of 60 Hz for the alignment purposes since A.C. ﬁelds of that frequency are

the most easiest and inexpensive to generate.

o It was also noted that the mode of ﬁber feeding had a signiﬁcant effect on the
alignment process. For example, when ﬁbers are in clumps and or entangled, they
fall through the orientation chamber with hardly any effect of the ﬁeld. Therefore,

for effective ﬁber alignment, the ﬁbers should be as free as possible.

3.3.3 Behavior in A.C. Fields

The majority of the experimental work was conducted with AC. ﬁelds at 60 Hz.
This frequency of the AC. ﬁeld is considered low enough to neglect magnetic ﬁeld effects5
and the relationships presented in § 3.2.1 and § 3.2.2 for the electrostatic ﬁeld cases can be
applied here as well. The effectiveness of the electric ﬁeld on the behavior of orientation
was observed at three stages: A, B and C (Figure 3.3). It is important to note from Figures
3.1 and 3.3, that in the orientation chamber, the E-ﬁeld is normal to the parallel plate
electrodes i.e. the E-ﬁeld lines are oriented along axis “2". In the case of a perfect E-ﬁeld
ﬁber alignment 0 = 90° and (I) = 90°. The likelihood of this kind of a ﬁber alignment

occurring under practical condition is remote and possible only when the ﬁber is neutrally

70

buoyant. Glass ﬁber alignment in air is a non-buoyant case and it is possible that a ﬁber gets
aligned in the E-ﬁeld with (I) = 90° yet 0 may not equal 90°. It will become apparent in the
following sections that the effectiveness of ﬁber alignment on the deposition plate improves

when 0~ 90°.

A high speed video camera with Kodak's Ektapro motion analyzer was used to study
the electrodynamics of the ﬁbers. Glass ﬁbers of lengths ranging from 0.125 to 1 inch and
E-ﬁeld intensities ranging from 0 to 600 KV/m were used in this study. It was observed that
an electric ﬁeld intensity of greater than 300 KV/m was capable of orienting these dielectric
ﬁbers. As expected, it was observed that with an increase in the ﬁeld intensity the degree of
alignment increased. High degree of ﬁber alignments were recorded at the bottom of the
orientation chamber for all the ﬁber lengths investigated. One such set of videographs is
shown in Figure 3.4, where 0.5 inch long glass ﬁber bundles are aligning in an E-ﬁeld of
intensity 400 KV/m. Note the random orientation state of ﬁbers in the top videograph, where
there is no E-ﬁeld. Similarly, Figure 3.5 shows 0.25 inch long glass ﬁbers aligning in an B-

ﬁeld of intensity 400 KV/m.

3.3.3.1 Effect of E-ﬁeld Intensity and Fiber Aspect Ratio
Ultimately, the alignment of ﬁbers that settle on the deposition platform is of primary
concern. So, the angles of the ﬁbers that settled on the deposition platform were measured

with respect to the electric ﬁeld direction and an orientation distribution was plotted. The

71

 

Figure 3.4 High speed videographs of free falling E-glass ﬁbers (0.5" long) in the
orientation chamber recorded at 1000 frames/sec under two conditions.
Top: Fiber alignment in A.C. ﬁeld of intensity 400 KV/m. Bottom:
Random ﬁber orientation under no E-field.

72

 

Figure 3.5 High speed videograph of free falling E-glass ﬁbers (0.25" long) aligning
in the orientation chamber in an A.C. ﬁeld of intensity 400 KV/m
(recorded at 1000 frames/sec).

73

details of the measuring technique for determining the ﬁber orientation distribution (F OD)
are given in § 4.3.4. The ﬁber orientation distributions as a function of ﬁeld intensity and
aspect ratio is shown in Figure 3.6 for 0.25 and 1 inch long ﬁber cases. From the orientation
distributions, it may be concluded that 1 inch long ﬁbers align better than the 0.25 inch ﬁbers
at each of the E-ﬁeld intensities 300, 400 and 500 KV/m. The FOD for 0.25 inch ﬁbers
improves (width of the distribution narrows) with an increase in the ﬁeld intensity from 300
to 500 KV/m whereas the improvement of alignment in 1 inch long ﬁbers is small. This data
seems to contradict the relationships presented in § 3.2.2, where it was shown that the time
required to align a ﬁber increased with ﬁber length (also see Figure 3.9). In other words for
the same height of the orientation chamber and the ﬁeld intensity, the 1 inch long ﬁbers
should align to a lesser degree than the 0.25" ﬁbers and subsequently have a broader FOD.
Secondly, the improvement in the degree of alignment as Eo increased from 300 to 500
KV/m does not seem to follow a square relationship (see Eqs. 6 and 11). These results led
to the investigation into the inﬂuence of hydrodynamics of ﬁber motion on the overall ﬁber

alignment process in electric ﬁelds in air.

From a series of high speed experiments it was concluded that ﬁber alignment
improves as 0 tends to 0° during the ﬁber motion in the orientation chamber. This is due to
two reasons. First, from the high speed videography it was observed that the impact energy
absorption characteristics of the deposition plate was found to be a critical source for
misalignment of ﬁbers after they got aligned in the electric ﬁeld. Moreover, the bouncing

of the ﬁbers on impact with the deposition plate decreases as the angle 0 tends to 90°.

74

F008 of 1" E-Glaee Fibers

 

0.35

0.3

 

 

 

 

 

g 0.2
a
0
50.15
0.1
0.05
o .
"9° 4” 3° Angle 8r Deviation” 6° 9°
FODs of 0.25" E-Glass Fibers
035
0.3
0.25
5 0.2
a
U
.E 0.15
0.1
0.05

 

 

 

 

Angle of Deviation

Figure 3.6 Fiber orientation distributions of 1" and 0.25" E-glass ﬁbers settled at
the bottom of the orientation chamber at A.C. ﬁeld intensities of 300
(I), 400 (O) and 500 (A) KV/m.

75

Secondly, it was observed that the degree of alignment in the orientation chamber and on the
deposition plate was dependent on the angle 0 of the ﬁber at the entry into the electric ﬁeld.
This implies that as the ﬁbers tend to fall horizontally, the effectiveness of the E-ﬁeld to
orient the ﬁbers to the desired angle improves. Infact, the mis—alignment of ﬁbers that seems
to persist despite increasing the ﬁeld intensity from 300 to 500 KV/m may be partly

attributed to this phenomenon.

Under the same ﬁber feeding conditions for both the 0.25" and 1" long ﬁbers at the
top of the orientation chamber, it has been found that the 1" long ﬁber tends to attain a stable
horizontal position by the time it is half way through the chamber whereas the 0.25" ﬁber
does not attain a stable motion. This can be evidenced from the set of high speed
videographs shown in Figures 3.7 and 3.8 taken in the absence of electric ﬁeld. Figure 3.7
is a set of videographs taken at the bottom of the orientation chamber showing 0.25" ﬁbers
settling. It is clear that although there is some natural tendency by the ﬁbers to occupy a
horizontal plane a good number of them are also in a three dimensional random orientation
state. Whereas, the tendency of the 1" ﬁber to occupy a horizontal plane is so predominant
that it was recorded in a series of videographs as shown in Figure 3.8. Each frame is split
into two to record the images from the two cameras that were used. One camera was placed
on the top of the orientation chamber (right side of the frame)while the other was placed
infront of the orientation chamber (left side of the frame). See Figure 3.3 for details. In the

ﬁrst frame, a 1" long ﬁber is released vertically. By following the motion of the ﬁber which

76

 

Figure 3.7 High speed videographs showing a greater tendency of 0.5" ﬁbers in
attaining a stable horizontal ﬁber orientation state than 0.25" ﬁbers.

77

.Eo: 33> 2.7. new 9.3.: 33> no. 2: no .27. osmoaﬁoo a m_ 2.5.: scam .033 55352“. 353.5: «353
a “5E5...” :_ 82E wee. .._ mo 351:8 .953: a... 9:32? 2:0an 95 .3 ecu—5 wig—Meow? .5on :3:

. J... . .....

   

en 2:3..—

 

 

78

is recorded at 500 frames per second, it is evident that within a short period of time, 0.082

seconds from frame 316 to frame 357, the ﬁber occupies a horizontal plane.

The reason for better alignment of 1 inch long ﬁbers over 0.25 inch ﬁbers can be
partly explained by the above observations. And the reason for better stability in the ﬁber
motion as the length increases can be directly explained using Jeffrey’s model for ﬁber

motion in shear ﬂows (as discussed earlier in § 3.2.3.2).

3.3.3.2 Alignment Time

The ﬁber alignment time can be determined in the case of a neutrally buoyant particle
situation by measuring the time it takes to rotate from an initial angle (1) 1 to a ﬁnal angle (I)f
with respect to the E-ﬁeld lines. Although it seems very simple for experimental
determination, in the ﬁber/air case that is under investigation, the ﬁber motion is not
neutrally buoyant and therefore the ﬁber alignment in the “2" direction is coupled with ﬁber

motion in the “3" direction (Figure 3.1).

0 With the suspension medium being air, the stability of ﬁber motions and the effect
of external drafts could not be avoided. Combined with this, the total settling and
alignment times for the ﬁbers under investigation used were less than 0.5 seconds,
which requires the use of the high speed video camera and the cumbersome lighting

arrangements to record the events.

79

- Ideally, in order to record both the alignment and drift of the ﬁber in the orientation
chamber two high speed video cameras are needed. One camera can traverse down
the orientation chamber at approximately the same speed at which the ﬁber is falling.
A second camera at the top of the orientation chamber which has a dynamic
focussing capability to track the ﬁber as it is falling. This camera will record the drift
in the ﬁber translation. Although this kind of an arrangement was not available, a
wealth of information was obtained by high speed videography using at least one and
sometimes two cameras in ﬁxed positions.

0 Glass ﬁbers are translucent and reﬂective in light at certain angles of incidence. This
nature of the ﬁbers combined with their small dimensions makes it very difﬁcult to
get good images of the ﬁber as it is translating and rotating. Optimum bright lighting
arrangements had to be made from the front and the top of the orientation chamber
to get the ﬁber in each frame, which made experimentation quite difﬁcult especially
when the electric ﬁelds also had to be used.

0 The gap width of the electrodes could not be spaced wide apart so that there are no
wall effects during ﬁber motion. For most of the experiments a minimum E value
of 300 KV/m was needed to orient the ﬁbers, this put a limitation on the maximum
gap width that can be used because of the limitation in the maximum voltage that can

be generated by the transformer.

Therefore, an alternative approach was used that combines experimental observations

with model predictions for alignment times and arrive at an estimate for ﬁber alignment time

80

for the glass ﬁbers. Both Fuchs and Demetriades models were used in conducting
simulations for alignment times of glass ﬁber bundles that were used in the experimentation
and subsequent development of the ADF process. Since the ﬁbers were aggregates of
ﬁlaments, an equivalent aspect ratio had to be determined before using the models for ﬁber
alignment. Microscopic examination of the chopped ﬁber bundles indicated that they are
elliptical in cross section and the individual ﬁlaments are packed in almost a square array
instead of a hexagonal array. In order to translate the bundle size into an equivalent ﬁber
diameter, it is considered that the elliptical cross section of the bundle can be transformed
into a circular cross section with all the ﬁbers close packed in a square array. Aspect ratios
for all the E-ﬁeld and hydrodynamic computations are based on this equivalent bundle
diameter, dbc. For a given number of ﬁbers per bundle (N R,) the dbc is given by the following

relationship.

d

:2 =
f 7‘ (17)

Both Demetriades and Fuchs models give similar trends in alignment times excepting the

curves are slightly offset and the model by Fuchs predicts longer times (Figure 3.9).

The height of the electrodes was based on the alignment time computations which
was estimated to be about 0.5 seconds for a ﬁber bundle of length 1 inch, the ﬁber bundle
will get aligned in an electric ﬁeld of intensity ranging from 250 to 500 KV/m. Considering

that the settling velocities of glass ﬁber bundles was around 1 m/s (refer §3.3.5), the height

81

 

 

   
 

 

 

 

 

IO
’
—_— .—-"" uni—.—
’
‘I' ’I-
I d
l / “" '- ' ‘

A

m

V

G)

0.1

E

l-1

H

5 I,

go 00] .. _Demetriades (SOOKV/m)

it: Demetriades (250KV/m)

- - - Fuchs (500 KV/m)
— _Fuchs (250 KV/m)
0.001 ..
0.0001 * t ‘
0 10 20 30 40 50

Fiber Length (mm)

Figure 3.9 Alignment time simulations based on Demetriades and Fuchs models.

 

 

ofmc
onent
ahgnn

invest

34l4

ofme
501118 1

bundh

82

of the electrode is designed to be 50 cm. Observations were made at different heights of the
orientation chamber using a series of high speed video recordings. High degree of ﬁber
alignments were recorded at the bottom of the orientation chamber for all the ﬁber lengths

investigated.

3.3.4 Fiber Settling Behavior

Slow motion analysis of the high speed video was used to assess the settling behavior
of the ﬁbers. The objective here being to compare and see if the settling behavior follows
some of the theoretical predictions that were discussed in § 3.2.3. The motion of the ﬁber

bundle is also qualitatively characterized (see Figure 3.1 for the coordinate system) as:

Planar - bundles fall with a rotation in the horizontal plane (1-2);
Horizontal - the bundles with almost no inclination in the 1-2 plane;
Vertical - the bundles fall with no inclination in the 1-3 plane;
Tumble - when the motion cannot be classiﬁed under any of the above.
0 The ﬁbers tend to fall Horizontal under terminal conditions. In the presence of an

A.C. electric ﬁeld (400KV/m), the ﬁber motion tends to be more Horizontal with
almost no Planar motion. There was no signiﬁcant difference in the terminal settling
velocities of ﬁber bundles with and without the electric ﬁeld.

0 The impact energy absorption characteristics of the deposition plate has been

identiﬁed as a critical source for misalignment of oriented ﬁbers.

83

0 Terminal settling velocities of 0.25, 0.5, and 1 inch long ﬁber bundles is about the
same indicating slender body behavior.
0 Reynolds Nos. indicate that the ﬁber bundles fall in the intermediate regime. Inertial

effects have to be considered as the bundles are not settling under creeping ﬂow.

Settling Velocity: From the high speed video recordings, the ﬁber motion was digitized
frame by frame and ﬁbers which had motions that were close to either “vertical” or
“horizontal” in orientation were selected. The velocity of the translation of the centroid of
the ﬁber with respect to a reference frame, was measured from a series of images and an
average value is calculated. This kind of measurement was made for three different ﬁber
lengths tested and the settling velocities are plotted as a function of the ﬁber length and the
type of motion that is characterized. Figure 3.10 shows the spread of the data points when
the ﬁber motion is recorded at observation point B, while Figure 3.11 shows the results at
the point C (the bottom of the orientation chamber). It has been found that there is little or
no increase in the settling velocity of the ﬁbers as they move from region B to C, indicating
terminal conditions have been attained. A careful analysis of the spread in the settling
velocity data shows that at both points B and C the values range from 1.0 to 1.6 m/s. The
orientation state of the ﬁber has a very strong inﬂuence on the settling velocity. The settling
velocity of the ﬁber in a vertical orientation is the highest and the settling velocity in a planar
or horizontal orientation is the lowest. From the relations given in § 3.2.3.3 for slender
bodies it can be seen that as the aspect ratio of a ﬁber increases the settling velocity tends to

level off. It may be concluded that the aspect ratios of the glass ﬁber bundles under

Settling Velocity (rule)

84

 

 

 

 

 

 

 

 

2
1.8 II
1.6 I) .
O
1.4 g A
1.2 0 I
I
O ‘ I
1 ..
0.8 .. 9
0.6 «I
.Planar
0'4 II I Horizontal
‘Tumble
0.2 4» .Vertical
0 . ¢ ; ¢
0 0.25 0.5 0.75 1 1.25

Chopped Fiber Length (inch)

Figure 3.10 Settling velocity of ﬁbers (Observation Point B).

85

 

 

 

 

 

 

 

 

2
1.8 4r
16 J)- .
I
E 1 2
.9 ’ i I . Q
g 1 H
> O
2 0.8 <
s 0.6 «
oPlanar
0.4 n I Horizontal
‘Tumble
0.2 l .Vertical
0 4 t + 1
O 0.25 0.5 0.75 1 1.25
Chopped Fiber Length (inch)

Figure 3.11 Settling velocity of ﬁbers (Obseravation Point C).

86

investigation (lengths ranging from 0.25" to 1") were long enough to reach the asymptotic

values.

3.4 Summary of Results

E-glass ﬁbers of lengths ranging from 0.125 to 1 inch have been successfully aligned
using alternating current (A.C.) electric ﬁelds with intensities ranging from 300 to 600
KV/m. A.C. ﬁelds were preferred over D.C. ﬁelds to avoid electrophoretic effects which
disturbed the ﬁber alignment process. Based on the polarization characteristics of dielectric
materials, A.C. ﬁelds with a frequency of 60 Hz was chosen and found to be adequate for

effective ﬁber alignment in the ADF process.

The ﬁber alignment in air using electric ﬁelds is extremely fast and the estimates for
E-glass range between 0.25 to 0.5 seconds for the ﬁbers lengths investigated (ranging from
0.125 to 1 inch). It is very interesting to note that the theoretical predictions from
Demetriades model for a 1 inch long ﬁber bundle predicts 0.25 s and 0.06 s at 250 and 500
KV/m respectively while Fuchs model predicts 1.3 s and 0.3 s for the same situation. So
despite the fact that Re < 1 does not hold good for the ﬁber motion in this study, the models
(especially Demetriades) serves to give a reasonably good estimate of alignment times for

a ﬁrst order approximation.

A very interesting settling behavior has been observed in this study. The orientation

state of the ﬁbers as they approach near the bottom of the orientation chamber tends to

87

become horizontal, the trend being more predominant as the ﬁber length increases from 0.25
to 1.0 inch. Based on the settling velocity and the Reynolds numbers (10 - 25 )that were
calculated this behavior made perfect sense. This tendency of ﬁber settling has some unique
beneﬁts to the ﬁber alignment phenomenon, since it brings stability in the ﬁber motion
during free fall. The alignment behavior of the ﬁbers in air is more complicated than it is
in a more viscous ﬂuid medium due to the non-buoyant nature of ﬁber motion. With air as
the ﬂuid medium, the viscous resistance is small and gravitational and the rotational forces
become important. Fiber alignment becomes a strong function of not just the E-ﬁeld
strength but also on the stability in the ﬁber motion. Although it is predicted (in the case of
buoyant situation) that the alignment time increases with ﬁber aspect ratio, it has been found
that longer ﬁbers have better alignments for the same electric ﬁeld intensity due to their more

stable motion during free fall.

10.

ll.

12.

13.

14.

15.

16.

References

Communication by Mike Rich based on unpublished data, 1995.
Glor, M., “Electrostatic Hazards in Powder Handling”, John-Wiley & Sons, 1988.

Meek, J. M. And Craggs, J. D., “Electrical Breakdown of Gases”, John Wiley &
Sons, 1978.

Herrera-Franco, P. J. And Drzal, L. T., “Comparison of methods for the
measurement of ﬁber/matrix adhesion in composites”, Composites, Vol. 23, 1992.

Stratton, J. A., Electromagnetic Theory, McGraw-Hill, pp 215-217, 1941.
S. T. Demetriades, J. Chem. Phys, Vol 29, p. 154, 1958.

Arp, P. A., Foister, R. T., and Mason, S. G., "Some electrohydrodynamic effects
in ﬂuid dispersions", Advances in Colloid and Interface Science, Vol 12, p 295-
356, 1980.

Fuchs, N. A., The Mechanics of Aerosols, The Macmillan Co., 1964.

Lilienfeld, P., “Rotational Electrodynamics of Airborne Fibers”, J. of Aerosol
Science, Vol. 16, No. 4, p 315-322, 1985.

Jayaweera, K. O. L. F. and Mason, B. J ., “The behavior of freely falling cylinders
and cones in a viscous ﬂuid,” .1. Fluid Mech., 22, pp 709. 1965.

Pettijohn, E. and Christiansen, E., Chem. Eng. Prog., 44, pp 157, 1948.

Newsom, R. K. and Bruce, C. W., “The dynamics of ﬁbrous aerosols in a
quiescent atmosphere”, Phys. Fluids, 6 (2), pp 521, 1994.

Tucker, S. G. and Advani, S. G., Processing of short ﬁber systems, Flow and
Rheology in Polymer Composites Manufacturing, Ed. Advani, S. G., Elsevier
Science, 1994.

Jeffery, G. B., “The motion of ellipsoidal particles immersed in a viscous ﬂuid,
Proc. R. Soc., London, A 102, pp161, 1922.

Jayaweera, K. O. L. F. and Mason, B. J ., “The behavior of freely falling cylinders
and cones in a viscous ﬂuid,” .1. Fluid Mech., 22, pp 709. 1965.

Batchelor, G. K., J. Fluid Mech., Vol 44, p 419, 1970.

88

16.

17.

18.

19.

89
Global Lab Image Analysis Software Manual, 1995.

Bartnikas, R. and Eichhom, R. M., “Engineering Dielectrics Volume IIA,

Electrical Properties of Solid Insulating Materials: Molecular Structure and
Electrical Behavior”, ASTM, 1983.

Morton, W. E. and Hearle, J. W. 8., “Physical Properties of Textile Fibers”, The
Textile Institute (UK), 1993.

P. Lilienfeld, “Rotational Electrodynamics of Airborne Fibers”, J. of Aerosol
Science, Vol. 16, No. 4, p 315-322, 1985.

Chapter 4

 

Aligned Discontinuous Fiber (ADF)
Composite Process"

4. 1 Novel Concept ................................................... 91
4.2 ADF Process Design .............................................. 94
4.2.1 Unit Operations ............................................ 94

4.2.2 Fiber Feeder .......................... . .................... 97

4.2.2.1 Fiber Pre-alignment .................................. 99

4.2.3 Electric Field Orientation .................................... 101

4.2.3.1 Orientation Chamber ................................ 101

4.2.3.2 Electrode Design ................................... 102

4.2.3.3 High Voltage Source ................................ 106

4.2.4 Deposition Platform ........................................ 108

4.2.5 Powder Coating ........................................... 1 10

4.3 Process Performance ............................................. 112
4.3.1 Fibers ................................................... 113

4.3.3 Operating Conditions ....................................... 114

4.3.4 Fiber Orientation Distribution (FOD) Measurement ............... 117

4.3.5 Aligned Vs Random ........................................ 121

4.3.6 Concentration Effects ....................................... 123

4.4 Process Consolidation ............................................ 128
4.4.1 Process Consolidation Cycle ................................. 130

4.4.2 Consolidation Pressure ...................................... 130

4.5 Manufacturing .................................................. 1 33
4.5.1 Processing Rates .......................................... 133

4.5.2 High Speed Manufacturing Schemes ........................... 136

4.6 Summary ...................................................... 138
References .................................. . ......................... 141

 

‘ Filed for US Patent (1996).

90

91

4.1 Novel Concept

Recent advances in the area of polymer powder coating of ﬁbers at MSU"2'3 in
combination with the physics of aligning ﬁbers in electric ﬁeld, has paved the way to
conceptualize and develop a novel high speed processing methodology that can manufacture
aligned discontinuous ﬁber composites. Full realization of the stiffness to weight beneﬁts
of these composites is possible due to effective ﬁber alignment combined with the ability to
consolidate at higher volume fraction of ﬁbers. Absence of solvents or liquids in the Aligned
Discontinuous Fiber (ADF) process improves the speed of processing many fold and makes
the process environmentally friendly. The ﬁber alignment technique of the ADF process is
rapid and simple in operation and design, and potentially adaptable to integration into
existing composite sheet or lamination processing. Unlike continuous ﬁber random mat
reinforced composites which have poor drapeability and which have problems of
delamination under compression, aligned discontinuous ﬁber composites are ﬂexible and
can be molded or stamped into complex parts. In this chapter, the details of the process that
was designed, built and operated are presented. The continuous manufacturing schemes that

can be derived from this approach are also presented at the end of the chapter.

The concept of the ADF process is very versatile and can lead to three possible routes
for manufacturing discontinuous ﬁber polymer composites (Figure 4.1).
0 The ﬁrst one (shown by a solid line) can be a semi-continuous process where the
ADF preform impregnated with polymer powder (at volume fractions > 50 %) is

stacked to a desired thickness and consolidated under a temperature and pressure

92

.239:— ozmenﬁoo Aha—«Q .83”...— nsoaatueoma Baum—«V no Eco—:8 .25 Z ﬁe 25w:—

Aaquuoaogé \llll

 

   

 

 

 

 

 

 

 

 

 

 

 

58382 Al a\ 22-922 v.1
858me /III...\\ /
a
z
I
a _q gonna—moan: J.
’ 833m
/ " 8E>_om _
z I l \I I I l l
x
I x
358 non—.8
A . Lsaeamm , £3... 938:. x \ I L. I
3:888 8:3:088 suns“: 4|. 353 v
mg :88 «do S< hon—E / I l \
\ 8323895 /
x .8958 cocoon—
\ copra—om .89m
x
x
\
Amongaoﬁaé \\\III/ \
0:53.352 wan:— / x

\

825 3025.80 I l l I \

93

cycle to obtain a composite. This approach is well suited for the high melt viscosity
thermoplastic polymers. It reduces the time and pressure requirements during the
consolidation step since the polymer is already in place and has to ﬂow only locally.

0 The second approach (top dotted line), also well suited for thermoplastics, consists
of the continuous manufacturing of ADF composite in sheet form. Blanks cut from
this sheet can be thermoformed to a desired part, using high speed manufacturing
techniques. It is envisioned that the continuous manufacturing of ADF composites
will be a direct outcome of the principles established in the ﬁrst approach.

0 The third option (bottom dotted line) has potential for use with thermosetting resins.
Here, the amount of powder coating is kept to a minimum (volume fractions < 10 %),
so that sintered particles act as binders that hold the ADF preform together in a semi
consolidated state. This preform is then injected with a low viscosity cross-linking
resin via Reaction Injection Molding (RIM) or Structural Reaction Injection Molding
(S-RIM) process to form an ADF composite. The resin gels and forms a solid
network with time and temperature (depending on the cure kinetics). This resin
injection approach is well suited for thermosetting polymers and fabricating large

parts.

Having identiﬁed the potential features of this novel concept, the focus of this
research is to design, fabricate and develop a prototype Aligned Discontinuous Fiber (ADF)

composite process that achieves to establish the following objectives:

94

0 High speed processing capability for manufacturing aligned discontinuous ﬁber
polymer composites, using electric ﬁelds in air.
0 Improvement in mechanical properties of discontinuous ﬁber composites with ﬁber

alignment over random discontinuous ﬁber composites.

4.2 ADF Process Design

The prototype Aligned Discontinuous Fiber (ADF) composite process that has been
developed is shown schematically in Figure 4.2. This prototype ADF process is designed
with the dual capability of operating under batch mode when the deposition platform is
stationary; or semi-continuous mode when the platform repetitively travels back and forth.
Chopped ﬁbers are fed from a specially designed ﬁber feeder in to the orientation chamber.
Fibers are aligned in this chamber by the electric ﬁelds and settle on the moving deposition

platform. The speed controllable linear platform mimics the motion of a moving belt.

4.2.1 Unit Operations
For the purposes of understanding and designing the ADF composite process it was
necessary to ﬁrst identify the unit operations and establish the design parameters focusing

on the principles of processing rather than the rate of manufacturing (see Figure 4.2).

95

.339:— ozuonﬁoo Aha—«é 85m usesnauooﬂn conw=< unbouoa on. .«o 3352—5 Nd. 95w:—

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\

 

 

 

 

 

    

 

025 5.8::
03.59550
5:825 \ “8on

 

          

1'-

..oEZG away—P»

 

gusto—m 8a: 8:525 ooh—om away—3» gum!

 

Suwanee“— ..Fs Stock 385::

wit—oz :emmmoEEeU Til

23E c2325

96

The unit operations identiﬁed for the ADF process are:

Fiber alignment using electric ﬁelds
Polymer powder impregnation of ﬁbers

Consolidation of polymer powder coated aligned discontinuous ﬁber preform

As shown in Figure 4.1, the polymer powder impregnation and the subsequent sintering

operation can take place either before or after the ﬁber alignment step. In order to keep the

development of this process focussed to ﬁber alignment in electric ﬁelds, the powder coating

step is taken up after the ﬁber alignment step. The following design features have been

identiﬁed as critical in developing the ADF process.

Method for pre—alignment of ﬁbers using a suitable feeding system to control transfer
into the orientation chamber.

An orientation chamber with the necessary residence time for ﬁber alignment and the
ﬂexibility to provide the desired electric ﬁeld intensity and conﬁguration.
Identiﬁcation of the effect of ﬁber alignment during dynamic deposition conditions
when the platform is set in motion.

Method for controlled impregnation of ﬁbers with tribo-charged polymer powder.
Determination of the ADF production and processing rates as a function of the
maximum ﬁber concentration in the orientation chamber.

Determine sensitivity of the degree of ﬁber orientation in the ADF preform mat on

the mechanical performance of the ADF composites.

Design criterion was established for each of these features.

97

4.2.2 Fiber Feeder

The role of ﬁber feeding was critically evaluated for effective ﬁber alignment in
electric ﬁelds. It has been observed that ﬁbers having a random orientation state before
entering the orientation chamber cannot be easily aligned by the E-ﬁeld due to the tumbling
nature of the ﬁber motion. This tumbling motion and random orientation state is also
detrimental to the ﬁnal degree of alignment because once the ﬁber hits the moving platform
(belt or veil) at the bottom of the orientation chamber, the ﬁbers will rebound and get mis-
aligned. Fiber feeding poses another problem due to potential entanglement and clogging
of the feeder, which results in uneven and intermittent ﬁber delivery. A number of designs
of short ﬁber feeders were considered which were simple and inexpensive; which had the
potential to feed the ﬁbers with minimum clogging problems; and which had the scope to

provide some preferred orientation to the ﬁbers as they entered the orientation chamber.

The ﬁnal design consisted of a modiﬁed vibratory FMC Syntron Magnetic feeder
(Model FTOC). This feeder was chosen to feed the short ﬁbers, with or without powder
impregnation, into the orientation chamber (Figure 4.3). The vibratory motion provided by
the feeder coupled with the inclination of the feed chute to the horizontal plane provides
forward motion to the ﬁbers. The vibratory mechanism also reduces entanglement and
clogging of ﬁbers. The mass ﬂowrate of the ﬁbers can be controlled by regulating the
amplitude of the vibratory feeder. The feeder has a V-grooved chute, through which material
ﬂows downward. This type of an arrangement could not feed the ﬁbers into the orientation

chamber with a preferred ﬁber “pre-alignment”.

98

98:23-93 53 860$ hog—E Md 95w:—

uoﬁoom b98£> 032%va

8&2?on 5:0 8.2m pooh

 

SO .5an a Man

cementum

«2m

 

 

ﬁg

 

 

 

 

 

 

 

 

 

 

cocwzaoum :30 33A pooh

99

As discussed earlier in § 3.4, the most desired conﬁguration for “pre- alignment” is
the ﬁbers entering the orientation chamber in a horizontal plane (0 ~ 0). A unique pre-
alignment system was developed and incorporated in the feeding system such that ﬁbers fall
with their long axis parallel to the deposition platform at the bottom of the orientation

chamber.

4.2.2.1 Fiber Pre-alignment

The pre-aligner consists of a rectangular plate mounted on the vibratory mechanism
which propels the ﬁbers in one direction (Figure 4.4). The plate with raised corrugations at
predetermined locations on the surface forces the ﬁbers to align parallel to the corrugations
as they are transported forward. When ﬁbers are fed from one end of the plate in a random
fashion, the pre-aligner changes the orientation of the ﬁbers by the time they traverse the
plate and then pass through a set of slots into the orientation chamber as shown in
schematically in Figure 4.4. The corrugations were fabricated by embedding 0.0625 inch
diameter stainless steel rods onto the plate. Corrugations on this plate are located at 2.0, 1.5,
1.0 and 0.5 inch spacings so that any ﬁber that is less than 2.0 inches long pre-aligned
parallel to the corrugations as the ﬁbers move towards the exit slots. If a ﬁber is greater than
2.0 inches, then it is possible that it can bridge over the corrugations without pre-alignment
during its traverse. Because of the pre-alignment of the ﬁbers as they exit through the slots
in the plates and into the orientation chamber, a predominantly planar orientation state is

maintained as depicted in Figure 4.4.

100

.Eoﬁuwm—aéa .85— .? Emma-£82 9.9 03w:—

l .. l I 580m 2: $55 3a... .5

I 85m now—$88 ins—m:

I
. i“
_ as: 538-5 8:0 .88.."

.25 .8832?

 

 

 

 

gm“ :EOﬁEOO
w..m1.ot.m.oL.m.oLsm.o_+:m..or Fwd .IEH ... gnaw +-..-..c.m.- .
V
\ \

$8 $3: .. \ \
_ .. .

v3. 585 as, 8:875

/.l

 

 

 

 

 

 

 

 

 

 

 

 

 

5 8n;

101

4.2.3 Electric Field Orientation

Electric ﬁeld orientation forms the key step in the ADF process. The orientation
chamber and the electrodes were designed based on the principles of ﬁber alignment
described in Chapter 3. The alignment of freely falling ﬁbers in electric ﬁelds is further
complicated when the ﬁbers settle on a moving platform. The good alignment that is
achieved when the ﬁbers reach the bottom of the orientation chamber in a batch mode is
totally mis-aligned when the linear platform is set in motion. During the development of the
semi-continuous ADF process, it was found that the edge effects of the electrodes play a
critical role in the ﬁber alignment process. Special features had to be incorporated in the

electrode design to overcome edge effects.

4.2.3.1 Orientation Chamber

The ﬁber orientation chamber must have the necessary residence time for ﬁber
alignment and the ﬂexibility to provide the required electric ﬁeld intensity and ﬁeld
conﬁguration. The minimum electrode height can be calculated from the product of the two

parameters: (a) ﬁber alignment time; and (b) ﬁber settling velocity.

Min. Electrode Height = Alignment Time x Settling Velocity (1)

These parameters have been estimated from the experiments on the prototype for glass ﬁbers
in lengths ranging from 0.125 to 1.0 inches. Based on this calculation a height of 50 cm was
chosen for the electrode height as explained in Chapter 3. This orientation chamber is

rectangular in cross-section with an open top and bottom, and fabricated of insulating

102

materials. The parallel plate electrodes are mounted on the inside faces. The height of the
orientation chamber is 50 cm and the electrode gap width can be varied from 3 to 15
centimeters. The width of the ADF preform that is formed by the process was based on the
size of the composite test panel that would be needed to determine composite mechanical
properties. So the width of the orientation chamber was set at 25 cm, which produces an

ADF preform of width 20 cm in this semi-continuous process.

4.2.3.2 Electrode Design

A parallel plate electrode conﬁguration, which creates E-ﬁeld lines that are parallel
between the electrodes is used in the process for alignment of ﬁbers in the direction of the
moving platform. Depending on the electric ﬁeld intensity and the ﬁber aspect ratios, the
ﬁbers tend to align along the ﬁeld lines and deposit on the moving veil, located on the
platform. In the batch mode when the deposition platform is stationary, the E ﬁeld
distortion at the edge of the electrode at the bottom of the orientation chamber did not affect
the alignment of the ﬁbers in any signiﬁcant way (Figure 4.5). In the continuous mode of
operation, the ﬁbers that were already on the deposition platform would get mis-aligned as
they passed under the high voltage electrode in the reverse direction. There was literally a
“sweeping action” and the ﬁber alignment would be completely lost. Edge effects are
created at the bottom of the electrodes because the edges have small radii of curvature and
the tendency of the E-ﬁeld to redirect itself toward the closest grounding point (Figure 4.6).

The proximity of the deposition platform compared to the opposite ground electrode

103

 

 

 

 

 

 

 

 

 

_ _/_

\I/

_ \_\

IV om8>od
W\ \

l l I
II. I I
Al Recon

 

 

 

 

/
.l. 98:3
§on Hausa co Seem

dot—3.8!: 8:: no “coho onto 23.—82m mi 253...—

aouo—Z 80:3

 

l/

 

A\

 

 

Cc om§o> emf

104

provides a good source for grounding if the deposition platform is grounded. This will occur
when the height between the electrode edge and the deposition platform, h is smaller than
the electrode gap width, d. This edge effect is compounded because of the sharp radius of
curvature at the bottom of the electrodes. A region of very high ﬁeld intensity can be formed
which can be many times the intensity in the regions far from the electrode edges. The E-

ﬁeld intensity (E) at the two different locations is given by the following relationships.

Far Field: E = 21;: where d = electrode gap width
Near Edge: E ~ I; where r = radius of the edge

2r (2)
Near Edge: E ~ % where h = electrode gap height

When a voltage of 25 KV is applied to the high voltage electrode with an electrode
gap width of 5 cm, the E-ﬁeld intensity (E) far from edge is 500 KV/m. At the same time,
at the edge of the electrode which has a thickness of 0.0625 inch, the intensity will be about
15,000 KV/m. This is much higher than the far ﬁeld intensity and any ﬁber near such a ﬁeld
is bound to get distorted. A corona discharge often takes place at this point because these

intensities are higher than the breakdown voltages of 3000 KV/m in air.

105

  
  

 

d
‘ ’ - , >
_ _ _ a _ _ _
High Voltage(V) _______ _
Electrode Thickness, 2r ———————
/ / 5: ————— A
/ 152‘ \ \\ ‘h
g l I \ . j}
Linear Motion
El =V/d
E2=V/2r or V/h

E2 >> E1

Figure 4.6 Electric ﬁeld near the electrode edge.

106

To overcome this edge effect, the edge of the electrode plate was so designed that the
radius of curvature is much greater, which will reduce the intensity. A conductive brass
cylindrical rod was retroﬁtted to the edge of the plate which provided it with an effective
radius of curvature of 0.25 inch (Figure 4.7). This decreased the degree of mis-alignment of
ﬁbers due to the reduction in the ﬁeld intensity near the edge to 3900 KV/m, but mis-
alignments could not be completely eliminated. A more effective solution to the problem
of edge effects was found in a novel electrode design which created a neutralizing ﬁeld just
below the edge of the high voltage electrode (Figure 4.7). The neutralizing ﬁeld is created
by the set of electrodes that are placed directly below the deposition substrate. The high
voltage of this neutralizing electrode was derived from a voltage divider. As a consequence,
the electrode edge effects were completely eliminated from the process when the platform
was set in motion both in the forward as well as the reverse directions. With this
modiﬁcation, the ﬁbers that were once aligned on the mat did not get mis-aligned even at

relatively high speeds of the deposition platform.

4.2.3.3 High Voltage Source

The high voltage needed for the ADF process is generated using a high voltage (HV)
Hipatronics transformer (Model RM25). A.C. ﬁelds are created by linking the secondary
output of the transformer to parallel plate electrodes made of copper. The electric ﬁeld

intensity can be controlled by two means: (a) by controlling the line voltage using a variac

107

$3239:— msoaaznoo he 20: wag—2:8: a 5?: 560—. cuckoo—m Ev ouswmm

 

 

 

 

 

 

 

 

 

 

 

L
J

888E
$8332

 

at ows_o> if
F 0 iv

.G
coco: 833

 

 

 

 

 

 

 

I ll \
l \
/
IIV 0805M
l I \
I I l
I \
/ I
All “custom
I l \
l \
/
I 5235

8%: 5&4 co comm

 

 

 

 

 

‘1 v
l. .4
03m 3055.5
/
\ JO
23 ”33> BE
n\ no
/
\ /
\
c, l
. A! ,, - 088%.
e2:

108

thereby controlling the input voltage on the primary windings of the transformer; or (b)
changing the gap width of the parallel plate electrodes. For a primary input voltage of 1 10V,
the transformer has a rated output of 25 KV at 15 mA. The gap width of the electrodes can
be varied from 3 to 15 cm, thereby creating a variation in electric ﬁeld intensity from 833
KV/m to 167 KV/m. The voltage is monitored using a high voltage Fluke 80K—40 probe

whose signal is sent to a digital Fluke 45 multi meter.

Figure 4.8 shows the circuit diagram of the voltage distribution to the various
electrodes. The variac provides the ﬁrst control on primary voltage of the HV transformer.
The secondary from this HV transformer is sent to a voltage divider. The top electrode
receives the full voltage of the secondary, while the bottom electrode receives a fraction of
the voltage of the top electrode in four possible steps: 100, 75, 50 and 25 %. The settings are

determined based on the processing conditions.

4.2.4 Deposition Platform

The aligned ﬁbers with or without powder impregnation are deposited on a Teﬂon
release ﬁlm placed on a sliding deposition platform in the laboratory prototype. The forward
and backward motion of the deposition platform on the linear slide simulates a continuous
manufacturing operation. If ﬁber alignment is not disturbed during the forward and reverse
motion, it can safely be assumed that the ﬁbers will not be disturbed in the event of exposure
of the aligned ﬁbers to subsequent ﬁelds of the multiple orientation chambers. 80 the

principles that are developed with this prototype can be directly translated to the

109

 

 

 

 

 

 

__ﬂmnnd
o
A.C. Input 0 25 KlL 15 mA
120V, 60 Hz
25 KV Hipatronics Transformer
Top Elec trodes
Voltage Divider
I Bottom Electrodes

 

Figure 4.8 Circuit diagram for high voltage distribution.

110

manufacturing situation as discussed in § 4.5.4. The linear slide is driven by a D.C. gear
motor connected to a speed regulator. Controllable forward and reverse motion results in the
building up of an uniform aligned discontinuous ﬁber preform with the desired thickness.
The largest uncut size of the ADF preform that can be formed is 20 x 35 cm, which allows
physical property measurements to be conducted. The total traverse length of the platform
is 30 inches. The forward and reverse motion is controlled by a speed controller and the
speed of the platform can range from 10 cm/min to 180 cm/min. An operating speed of 50

cm/min was chosen for the prototype ADF process.

4.2.5 Powder Coating

Provision must be made to incorporate a step for controlled impregnation of ﬁbers
with tribo-charged polymer powder. The mode of impregnating polymer matrix on ﬁbers
is by bringing in contact an aerosol of ﬁne tribo-charged polymer powder and ﬁbers, to get
a uniform coating of particles on the ﬁber. Acoustic aerosolization is an effective means to
generate an aerosol of ﬁne polymer powders of sizes less than 30 microns“. This aerosol can
be entrained through a nozzle where the ﬁbers are introduced, similar to the idea used in the
high speed MSU Powder Prepreg Process developed by Vyakarnam and Drzals. The amount
of matrix or binder that gets deposited on the ﬁbers is controlled by controlling the particle
concentration ﬂux in the aerosol, which is directly related to the matrix volume fraction in
the composite. Figure 4.9 shows electron micrographs of E-glass ﬁbers coated with 10

micron size nylon 12 particles. It is important to note that the uniform distribution of

 

Figure 4.9 SEM micrographs of E-glass ﬁber bundles coated (Top) and uncoated
(Bottom) with nylon-12 powder.

112

polymer particles around the ﬁbers requires the polymer to only ﬂow locally over small
distances, thereby reducing the polymer melt ﬂow times during the compression molding
step“. This improves matrix impregnation, minimizes void formation during consolidation
and improves the mechanical performance of the part. Powder impregnation makes

compression molding of the ADF preform into a ﬁnal composite part a rapid step.

Since the focus of the ADF process was more on the development of the ﬁber
alignment technique, a continuous mode of impregnating discontinuous ﬁbers in an aerosol
of ﬁne polymer powder was not developed speciﬁcally for the ADF process. Such a system
can however, be easily developed based on the earlier work on powder processing. For the
present investigations, polymer powder is introduced after the ﬁber alignment step by
uniformly spraying a measured quantity of the powder on the ADF preform. The mass of
powder that is required is calculated as per the volume fraction of the ﬁnal composite. The
ADF preform with the polymer particles is passed under an infra-red heater to sinter the
particles in place. A conventional oven was also used to sinter the nylon 12 particles on the

glass ﬁbers making the ADF preform handleable.

4.3 Process Performance

First, the ﬁbers and the polymer matrix used in the process development are characterized
for their physical and thermal properties that are useful in determining the operating
conditions of the process. Later the process operating principles and performance will be

taken up.

113

4.3.1 Fibers

E glass chopped ﬁbers of nominal lengths 0.125, 0.25, 0.5 and 1 inch were supplied
by Owens-coming. The number of ﬁlaments per bundle was approximately 250. The length
range and the number of ﬁlaments that were used in demonstrating the ADF process was

based on the ﬁber alignment behavior in electric ﬁelds.

Length Distribution: The actual length of 100 ﬁbers was measured that were picked
randomly from the lot. The length distribution was plotted and it was found that chopping
method used by Owens-Coming was extremely efﬁcient and all the ﬁbers were within 5 %
of the nominal length. The lengths of the smaller length ﬁbers were also measured at random
and found to close to the nominal value. Based on these observations, the nominal length
of the chopped ﬁber was used for all computation and felt that it was not necessary to include

a length distribution.

Aspect Ratio Distribution: For each of the ﬁber length measured, its mass was also
weighed. It was found that the mass distribution was not as sharp as the length distribution.
The number of ﬁlaments for each of the ﬁbers was also determined by a simple mass balance
and found that there is variation in the number of ﬁlaments per bundle. The mean value was
227 instead of the nominal number of 250 that was reported (see Table 5.4). Since all the
discontinuous ﬁbers were obtained from the same rovings, this mean value for the number

of ﬁlaments is used in all computations.

114

4.3.2 Polymer Characterization

The nylon-12 (polyamide) polymer powder was obtained from Atochem, Inc. The mean
particle size of the powder was determined using Malvem Particle Size analyzer and the
Image analysis techniques and was found to be 10 microns. The thermal transitions of this
semi-crystalline polymer were evaluated using a Differential Scanning Calorimetry (DSC).
The melting point of this polymer has a sharp onset starting from 171 °C and the glass

transition temperature could not be detected clearly but is reported to be 40 °C.

4.3.3 Operating Conditions

The schematic of the ADF process given in Figure 4.2 can be referred to again for this
section. Chopped ﬁbers are fed from the specially designed ﬁber feeder in to the orientation
chamber. Fibers are aligned in this chamber by the electric ﬁelds and settle on the moving
deposition platform. The alignment of 1 inch long E glass ﬁbers in electric ﬁeld of intensity
400 KV/m, is shown by the digital images of the ﬁbers that have settled on the deposition
platform (Figure 4.10). The images of the ﬁbers when there is no E-ﬁeld and the best efforts
to align manually using a pair of tweezers are also shown in the ﬁgure for comparison. The
ﬁbers are then sprayed uniformly with the polymer powder of a measured quantity. The
speed controllable linear platform mimics the motion of a moving belt. The prototype ADF
process is designed with the ﬂexibility of operating under batch mode when the platform is
stationary; or semi-continuous mode when the platform is set in linear motion. The aligned

ﬁbers deposit on a Teﬂon release ﬁlm which is placed on the top of the linear platform.

115

62.9..“ £2.52 3ng .Eoeé 2. a? 52:5. 3:52 .33: 2:. ca 22.; 5 3&5.

3qu diets—a neEmono—e on. .8 3.33. E25 awn—mi c2523 uno— :~ no women: .835 :3. 953%

 

116

Table 4.1 Typical Operating Variables

 

 

Variable Operating Range
Velocity of Deposition Platform 50 cm/min
E-Field Intensity (Top) 300 - 500 KV/m
Neutralizing E-Field Intensity (Bottom) 300 - 500 KV/m
Electrode Gap width (d) 4 - 8 cm
Electrode gap height (h) l - 2 cm
Fiber Feeder Amplitude 3 - 4 on dial
Oven Temperature for Sintering 200 °C
Processing Ratei' 900 g/hr (2 lbs/hr)

 

'1' Estimated from semi-continuous operation data.

To demonstrate the operation of the prototype ADF process, a model composite
system was chosen: chopped E-glass ﬁber and nylon-12 polymer matrix. Based on the ﬁber
alignment behavior in the process and the polymer sintering temperature, the optimum
operating conditions were obtained to operate the ADF process in a semi continuous mode
to produce an ADF preform. The operating variables and their ranges in a typical processing

situation are given in Table 4.1.

117

4.3.4 Fiber Orientation Distribution (FOD) Measurement
One of the objectives of developing the ADF process was to demonstrate the
feasibility of making ADF composites with controllable ﬁber orientation distributions

(FODs). The orientation distribution of the ﬁbers is needed in the ADF process for three

reasons:

0 Serve as a feedback on the effectiveness of the ﬁber alignment process.

0 Determine the effect of various parameters on the degree of ﬁber orientation.

- Predict the mechanical properties of the ADF composites using micromechanics
models.

Typically, the ﬁber orientation distribution in a composite is obtained by a
measurement of a statistically signiﬁcant number of ﬁber orientations by sectioning the ﬁnal
composite and reconstructing the orientation of individual ﬁbers by the aspect ratio of the
ellipse that is formed. This is a very tedious technique, and numerous approximating
techniques have to be used in these methods and often bias is introduced against those which
are normal to the plane of observation”. Because of the many short comings in the existing
measuring techniques", a protocol has been developed for evaluating the ﬁber orientation

distribution in the ADF process, which can be used for both the above mentioned purposes.

The measurement step consists of three steps (Figure 4.11). Typically once the

processing is done for the manufacture of an ADF composite, its FOD is obtained in an

118

 

 

Global Lab Image . Automatic and Manual
[ Analysis ) ( Measurements )
T i
E Computation ]

f,, = 2 jn(¢)cos2(¢)d¢—1

—90

 

 

 

 

 

 

Figure 4.11 Measurement of ﬁber orientation distribution.

119

indirect way by image analysis. The method utilized in obtaining the POD consists of
running the ADF process once again under identical conditions but this time not with the
intention of making a composite but recording a series of images of the ﬁbers that are
deposited on the Teﬂon release ﬁlm using a Panasonic CCTV (Model WV 1410) camera
which is connected to a computer with a frame grabber board. Recording a proper image
was of utmost importance, because this is the step where bias could be introduced in this
technique. For example, glass ﬁbers shine and reﬂect light when incident from an angle.
If the lighting is not uniform over the region of interest, it is very possible to record ﬁbers
that are oriented in a certain direction only which shine more prominently than the other
ﬁbers that are in the other directions. This kind of an artifact ﬁnally results in showing that
ﬁbers are aligned in a particular direction. The error in this kind of a biased image gets
worse when the digital image is subjected to ﬁltering operations. So before recording a
series of images it was always checked to see if ﬁbers in all the directions were being

recorded.

A number of images of the ﬁbers that are deposited on the Teﬂon release ﬁlm are
recorded to give a statistically signiﬁcant F OD, which typically contains the orientations of
approximately 1000 ﬁbers. Typically, the longer the ﬁber length the more the number of
images have to be recorded to get signiﬁcant number of ﬁbers. Global Lab image analysis
soﬁware is used to treat the digital image to a few ﬁltering operations which sharpen the
images of the ﬁbers for edge detection and also improve contrast of the ﬁbers against the

background. The “Particle Classify” routine in this software can be trained to identify ﬁbers

120

and their orientations automatically. This feature reduces the time for data analysis
drastically. However, in the case when there is excessive bridging or networking of ﬁbers,
this feature is not very effective and the ﬁbers have to be individually identiﬁed to determine
their orientation. Once the ﬁber orientations are logged, the pertinent data can be exported
to any spread sheet environment like Microsoft Excel. All the statistical analysis is

performed at this stage for the determination of the FOD.

The force ﬁeld that the ﬁbers encounter in the orientation chamber is depicted in
Figure 3.1. If the ﬁber orientation at this stage needs to be described then it can be done
by a three dimensional description (using (1) and 0). So the unit vector p. of the ﬁber

describing the orientation in the three axes is given by:

p1 = sinBcosd)
p2 = sinBsind) (3)
p3 = cosB

Assuming the orientation of ﬁbers is 3 even function and not biased in one direction, the
ﬁber orientation distribution function, N (0.4)) may be described by the following

trigonometric relationship.

N(0.(I)) = Acosmﬂ sin”d) (4)

where A is chosen to satisfy the normalization condition of the distribution. Although, the
formulation of FOD function will be similar to the one that will be used in the orientation

chamber, there are certain differences in the assumptions. First, it can be safely assumed in

121

the case of ADF preform that the ﬁbers are in a state of planar orientation (2D) and only (I)
is required to determine the F OD and 0 is 90". In the case of thin ADF samples (noting that
length of ﬁbers are far greater than the thickness of the sample) a digitized image of the
ﬁbers on the mat will suffice in determining the FOD. Therefore, in the case of a 2D (planar)

orientation, the F OD function , N (d) ) will simplify to

N(d)) = Apcosmd) (5)

The Planar Orientation Parameter fp will be used to reﬂect the FOD in a given condition.

j; = 2 f:N(d>)cos’"¢ dd) — 1 (6)

In the FOD measurement for composites, it has been carefully veriﬁed that the F OD of the
ADF preform is not disturbed during the subsequent consolidation processing of the
composite. This was veriﬁed using a few tracer ﬁbers with known orientations before
consolidation and checking their orientation state after the consolidation. Therefore, it may
be concluded that the FOD obtained by this method is a true representation of the F OD in the

ADF composite.

4.3.5 Aligned Vs Random

Using the above method a series of FODs were obtained for each of the chopped
glass ﬁber mats that were produced under the two conditions: randomly oriented and aligned
in an E-ﬁeld of 400 KV/m cases (Figure 4.12) at the optimum operating conditions as

speciﬁed in Table 4.1. As expected, the orientations of the ﬁbers in the random case for all

122

 

 

30 1’ ORendom I
u_ IE-l'leld

 

 

 

 

   

 

 

0 Random

25 4» .56.” "2" Flbers

 

 

 

 

   

Figure 4.12 FODs of different ﬁber length E-glass ﬁbers that have aligned in E-ﬁeld
of intensity 400 KV/m. The control case of random is also shown for
comparison.

123

the four ﬁber lengths indicate a uniform spread of orientation distribution, without any
signiﬁcant bias in any direction. The unbiased ﬁber orientation distribution of the control
random cases serves to establish that there is no extraneous induced alignment in the process
other than that caused by the E-ﬁeld. In the case of aligned ﬁber mats, for all the four ﬁber
lengths, the orientations of about 70% of the ﬁbers lie between 3:20" and nearly 80 - 90% of
the ﬁbers between i30°, indicating a high degree of ﬁber alignment in the direction of the

E-ﬁeld.

4.3.6 Concentration Effects

The rate of processing is limited by the concentration effects in the orientation
chamber. If the concentration of the ﬁbers is increased from the very dilute case to a
concentrated case, the probability of ﬁbers touching each other starts to increase. As the
probability of ﬁber interactions increase the probability of a bridging phenomenon, taking
place increases in the orientation chamber. The bridging phenomenon not only causes a high
degree of mis—alignment but also has the detrimental effect of shorting the ﬁelds in the case
of conductive ﬁbers. Infact, concentrated ﬁber suspension also reduces the effective
permittivity of the suspension. An interacting concentration of ﬁbers will result in a high
probability of arcing or even subsequent breakdown in the electric ﬁeld. This phenomenon
of "ﬁbration" has been looked into by Okagawa and Mason'0 for electric ﬁelds, and by
F ermigier and Gastll in the case of magnetic ﬁelds, for spherical and axisymmetric particles
in both colloidal and non-colloidal systems. However, the size range of particles studied by

these authors is much smaller compared to the dimensions of the ﬁbers useful in composite

124

manufacture of this study. It is also known that beyond a certain concentration of ﬁbers and
particles, the electric ﬁeld lines will be distorted, thereby lowering the efﬁciency of

alignment.

If C,— is deﬁned as the theoretical maximum of the non-interacting concentration of
ﬁbers, then it is related to the aspect ratio of the ﬁber and the ﬁber orientation state. The
lowest Cr is obtained when the ﬁber orientation state is a 3-dimensional random state,
implying a ﬁber will have freedom to rotate with respect to all the three axes. However, in
the ADF process, the ﬁbers are fed in a predominantly planar fashion using the pre-aligner
(Figure 4.4). Figure 4.14 shows the “volume of inﬂuence” surrounding a ﬁber under the
two extreme orientation states of the ﬁbers under consideration. This intuitive analysis into
the effect of orientation state on the maximum non-interacting concentration of ﬁbers has a
direct bearing on the way the ﬁbers should be fed into the orientation chamber. It can be
easily shown by the following relationships that the predominantly planar orientation state
of the ﬁbers increases Cf compared to the random state.

Random Orientation: In this case, the ﬁbers are in a complete state of three dimensional
randomness and have the freedom to assume any orientation state. A conservative estimate
of the maximum concentration of the ﬁbers that is non-interacting can be obtained by
considering each ﬁber to have a “sphere of inﬂuence”. This sphere of inﬂuence has a
diameter equal to that of the length of the ﬁber. Taking the volumetric ratio of the ﬁber to
that of the “sphere of inﬂuence” of diameter equal to the length of the ﬁber (1) we get the

relationship for Cf.

 

-’ _1_ 3 '12 (7)

Pre-aligned Orientation: The unique design of the ﬁber feeder plate ensures that the ﬁbers
are pre-aligned before they enter the orientation chamber. First, the ﬁbers are in a
predominantly, planar orientation state. Secondly, the ﬁbers are also pre-aligned parallel to
the slots (Figure 4.3) which makes them pack better. For the purposes of making a
conservative estimate of the concentration of ﬁbers in a given volume, it is assumed that the
ﬁber can subtend up to an angle of 300 (see Figure 4.14). A “cylindrical volume of
inﬂuence” can be visualized and calculated with a diameter equal to 1/2 (= 2 x 1/2 sin 30°)

and height equal to the ﬁber length. l.

 

L121 2
C‘ I 4 I i]—
4 2

Figure 4.15 is a plot of Cf as a function of the ﬁber aspect ratio for the two orientation states.
From these relationships, it may be seen that for a chopped ﬁber bundle of say an aspect ratio
100, the limiting non-interacting concentration is 0.00015 for the random case while it is
0.0004 for the pre-aligned case. Therefore it may be concluded that by the unique design of
the ﬁber feeder, the theoretical maximum non-interacting concentration of ﬁbers (Cf) in the

orientation chamber has been increased by more than 2.5 times. This increase in the Cr value

126

Spherical Volume oflnﬂuence Cylindrical Volume oflnﬂuence

 

Length. 1

Figure 4.14 “Volume of inﬂuence” surrounding a ﬁber under two
Different orientation states: Random and Pre-aligned.

Max. Non-interacting Concentration

127

 

0.1 <

0.01 '1

0.001 «

0.0001

0.00001 4»

   

 

0.000001

—0— Cf-Rando-mf .
1 -ﬁ— Cf-Prealigned .

 

 

Aspect Ratlo

Figure 4.15 Non-interacting ﬁber concentration as a
function of aspect ratio.

128

results in an increase in the production rate proportionally. However, it is also important to
point out here, that this concentration effect on the electric ﬁeld and its distortion is
unknown. Although, production rate will be increased by this effect, it is unknown if the

degree of ﬁber alignment will remain the same.

4.4 Process Consolidation

Consolidation is the processing step in which the ADF preform is compression
molded under temperature and pressure to get a fully densiﬁed and essentially void free
composite. In the thermoplastic powder coated chopped ﬁber preform, the ﬂow of the
polymer matrix is local and there is no bulk ﬂow of the molten polymer because of the
uniform distribution of the polymer particles on the surface of the ﬁber bundles (see Figure
4.9). This factor is a distinct advantage for the high melt viscosity thermoplastic polymers
that are subjected to ﬂow with low penneabilities in ﬁbrous composite beds. This results in
faster processing times and lower operating pressures in the consolidation process. A typical
consolidation process cycle is shown in Figure 4.13. Consolidation primarily involves the
heating of the preform from the ambient temperature (Ta); applying pressure after the
temperature of the polymer reaches the processing temperature (Tp) which is typically above
the melt temperature (Tm) for semi-crystalline polymers. The part is held at Tp under
pressure till the ﬂow of the polymer is complete and autohesive strength is developed. It is
subsequently cooled under pressure till the temperature falls below the glass transition

temperature of the polymer (Tg) to obtain a fully densiﬁed composite.

129

a...
m...

5.52: coahzm o:...39€o..=8m a 5.3 5.8.9:. a .e 0.93 9.39:. 5.32333 .33.»...

 

  

 

  

 

All 08....
"IIJIIImHM—molwwlIIITIIIIIIHIIIIIIII IIIIIIII n
- -
025000 W moﬁng m 304... W
oggaaoh

 

 
  

 

 

E... 2&3

a...
m...

a...
a...

130

4.4.1 Process Consolidation Cycle

ADF mats that were produced in the process were carefully transferred to a
compression molding caul plate, without disturbing the ﬁber orientation distribution. A
known number of plies were stacked up depending upon the ﬁnal thickness of the composite
part desired. Vacuum assisted compression molding was performed in an instrumented
Carver Press to make a part with minimum voids. A differential scanning calorimetry studies
were conducted on the nylon-12 matrix to determine the thermal transitions and the
crystallization behavior before developing the optimum consolidation cycle. It was also
veriﬁed that the FOD of the ADF preform was not disturbed during the consolidation
process.

Typically eight plies of the ADF preform of sizes 5" x 7"are stacked in an open mold
with dams on all four sides to prevent resin ﬂow out. The preform is heated in Carver press
to 200°C and held at that temperature for 10 minutes under 100 psi pressure. Vacuum was
applied as soon as the preform temperature reached 100 °C, so that all the entrapped air as
well as the moisture if any, will be removed from the system. Vacuum is pulled off during
the cooling stage after the temperature drops below the melting point of the polymer, i.e 175
°C. The cooling is controlled by the water ﬂow rate and a steady cooling rate is maintained.

Pressure is maintained till the temperature of the resin drops below Tg.

4.4.2 Consolidation Pressure
Recently, a consolidation model has been developed for continuous ﬁber powder

prepregs by Padaki”, incorporating the various phenomena that occur during the

131

consolidation process. The results from this study are used to arrive at some of the
processing parameters for consolidating ADF preforms. Though several of the features of
this model are applicable for powder coated ADF preforms, caution was exercised in

adopting the results for ADF preforms in view of some differences in the material form.

The consolidation process is most commonly treated as a load distribution problem
where the pressure applied to the composite preform (p) is balanced by the average resin
pressure (Pr) and the effective stress in the ﬁber network (0). The resin pressure is estimated
by modeling a squeezing ﬂow of a particle between two ﬂat surfaces. The force, F that is
needed to squeeze a particle from an initial height, ho to a ﬁnal height, h in time, t is given

by:

 

8na4 1 1
F: ___
t [122 I22) (9)

Where, a can be a characteristic dimension of the particle which is equal to ho. The ﬁnal
height, his determined to be the inter-ﬁber distance in a fully densiﬁed composite of a given

volume fraction.

The average effective stress (0) in the ﬁber network has been modeled by
Gutoswski”, as a deformation of the ﬁber network that occurs during any consolidation

process.

 

 

 

4 (10)
- 1

 

Vf is the varying ﬁber volume fraction of the lay-up that is being consolidated from an initial
volume fraction of V0 and Va is the maximum possible volume fraction that is possible for
the ﬁber architecture. As is called the “spring constant” which is a function of the ﬁber
bending stiffness (E) and a geometric parameter, [3. [3 is typically the span length to span

height ratio for the ﬁber beam network.

5 B4 (11)

As has been determined for AS4 carbon ﬁbers (E = 34 x 10" psi) to be to be 0.125 resulting
in [3 = 225. [3 value for chopped glass ﬁbers is not known, but certainly will be much smaller
than continuous ﬁber unidirectional preforms, and will also be dependent on the ﬁber

orientation.

This model describes the phenomenon when an initial load is applied it is carried by
the polymer matrix. As the polymer is squeezed the volume fraction of the ﬁber network

increases and the stiffness of the ﬁber network increases. It is to be noted here that the

133

typical volume fraction of discontinuous ﬁber composites is less than 50 % while for the
continuous ﬁber composites it is over 50%, indicating substantially lower pressures are

needed in consolidating ADF preforms.

4.4.3 Void Fraction

Each panel that is consolidated is tested for volume ﬁaction and void fraction using
a bLu'n off test. At two small pieces of the composite weighing approximately 1 gm. are cut
from random locations of the panel and the actual volume is measured by water
displacement. These samples are then burnt off in a mufﬂe furnace at 600 °C for 3 hrs and
the weight of the ﬁbers remaining is measured. The volume fraction of ﬁbers and the void

content is then determined (see Table 5.2).

4.5 Manufacturing

The focus of this chapter so far has been on the development of the ADF process -
its design, operation and performance. Now the issues involved in the making of ADF
preforms is discussed. First, the processing rates for the prototype process are derived and
later the high speed methodology of the ADF process and the continuous manufacturing

schemes are taken up.

4.5.] Processing Rates
The overall processing rate in terms of mass of ADF preform produced per time (P)

is a function of the mass ﬂux in the orientation chamber that is settling on the deposition

134

platform per unit time and the dimensions of the orientation chamber. If the ﬁber settling
velocity is 8,, the critical concentration of the ﬁbers in the orientation chamber is Cf (vol/vol),
the area of cross section of the orientation chamber is A0c , and the density of the ﬁbers is p,

then P is given by the following relationship.
P = P CfStAoc (12)

If the ﬁber feed rate is precisely known then that value can be used in arriving at Cf by a
simple mass balance. The rate at which the thickness (T) of the ADF preform grows under

batch mode is given by:

 

_d_T : Cf S,
dt A06 (13)

The thickness of the ADF preform (T) that is formed when the deposition platform is set in

motion at a velocity, V is given by:

V W (14)

Equations 11 and 12 are useful in predicting the thickness of the ADF preform that is formed

under the “batch” and “continuous” modes respectively.

135

The velocity of the deposition platform (V) plays a direct role in the rate of
processing. However, the velocity of the platform cannot be increased without consideration
to its effect on the ﬁber orientation distribution and the secondary operations like the
sintering times for the polymer particles. With the electrode modiﬁcations that incorporated
a neutralizing ﬁeld, it was observed that at the velocities of interest (around 50 cm/min) the
deposition platform velocity did not have any discernable effect on the ﬁber orientation
distribution. It can be expected however, at much higher velocities, air currents would
introduce mis-alignments. The velocity of the platform/belt will be a function of the
downstream processing like polymer particle sintering and heat transfer times needed when

continuous operation of the ADF composite sheet is the product.

The principle of ADF processing has been established for the case where there is only
one orientation chamber. If high production rates are needed, it can be easily seen that it can
be achieved by using multiple number of orientation chambers in series. So the scale up
factors for ADF process are:

- area of cross section of the orientation chamber (Am)

- number of orientation chambers (Noe).

The production rate (P) in the case of a multiple orientation cell system will be given

by the following relationship (see also § 4.5.1):

P = NOC p CfS, ADC (15)

136

4.5.2 High Speed Manufacturing Schemes
The ADF processing methodology offers a solution to the problem of reducing the

cycle time in the fabrication of an aligned discontinuous structural composite and lends itself

for a highly automated process, due to the following three reasons:

(a) a simple and rapid orientation technique using electric ﬁelds;

(b) controlled powder impregnation of the matrix gives immense ﬂexibility in the ﬁber
volume fractions that can be obtained and eliminates the additional step of resin
transfer molding as the case would have been if a preform is the ﬁnal product of the
process; and

© by incorporating a technique to increase the concentration of ﬁbers that can be

processed in the orientation chamber, thereby increasing the production rate.

As pointed out in the beginning of this chapter, the concept of ADF processing can
be applied in a variety of ways. Just the ﬁber alignment itself can be incorporated in any
continuous preform or composite sheet manufacturing process. The incorporation of the
polymer powder provides still greater beneﬁts in terms of high speeds of solvent free
processing. Figure 4.16 shows the schematic of a continuous ADF process that can be
conceived based on the principles developed in this study. Alternatively, a moving veil with

a layer of the polymer matrix ﬁlm can be passed through the orientation chamber. The

137

4:59:35:— 82? 31.2383 ma 2855.80 $.22.— 92329 win: u=oE=w=a hoar—
mwazaco .3632— ..oEbe: mac—22:85 8395 mn< 28:52.8 congea— uc oﬂaaoaom 5.9 «Sm:—

 

 

 

 

 

 

     

 

 

 

2ﬁ. :8
30.3 :3 05:09 . .
.HWHMHWMHWHWWWHOHMHWmam...
Beam ‘\\\\\\\ mm mmmmmmmmmmmmmm mm.
0:89:00 mmm mmm ”mm mm“
.5... . 5...: m. i mm 3 ea 533%.,
2. ages; ﬂ WEN” mm
.m .3... s... .8...
38.6 8585 mm mm E E
- mwmm .8883» 3881‘.

03.; avenue—m mmEmmmmm.mmmmmmmmm.ummmm

Scam—n-29uovoom ; ,, .8953

Babe.—

Bonm @380 BvBom
oﬁnoz

5:: Eu 8.326

138

deposited ﬁbers can then be covered with another layer of the polymer ﬁlm and heat treated
to form a handleable mat or laminate with the desired ﬁber orientation. The thickness of
such a material can be built by a series of orientation chambers and later compression
molded, typical of a sheet lamination process to form a composite sheet. The consolidation
cycle needed to process these composites will be the same as described above for the same

polymer matrix.

4.6 Summary

A semi-continuous prototype process has been developed that can manufacture
aligned discontinuous ﬁber composites using electric ﬁelds. The key steps in the process are:
ﬁber alignment in air using electric ﬁelds; powder coating/impregnation of ﬁbers and;
consolidation of the powder coated ADF preform into a composite. The ADF process has
been demonstrated for E-glass ﬁber and nylon12 matrix system and can be easily extended

to many other ﬁber-matrix combinations.

The design of the orientation chamber for controlling ﬁber orientation in air has two
unique features that counter the edge effects of the electric ﬁelds when the deposition
platform is set in motion during the continuous processing of ADF preforms.

- Create an equal and opposite ﬁeld under the high voltage electrode by placing

another electrode under the moving slide/mat/veil. This neutralizes the edge

effects of the high voltage electrode, which other wise disturbs the ﬁber

139

orientation of the ﬁbers that are aligned by the E-ﬁeld and that are deposited
on the mat.

- Designed the edges of the electrodes which are closer to the moving mat with
a conductive cap which has a cylindrical cross-section with a relatively large
radius of curvature. The cylindrical edge reduced the effective E-ﬁeld

intensity at the edge, thereby reducing the edge effects and minimizing B-

ﬁeld breakdowns.

Developed a unique vibratory ﬁber feeder and pre-aligner. Fibers with or without
powder impregnation can be fed into the orientation chamber without entanglement. Fibers
falling from the feeder pre-aligner plate tend to fall with a planar orientation state instead of
a three dimensional random state which makes them align faster in the direction of the
electric ﬁeld. This feeding mechanism also has the potential to increase the theoretical

critical processing concentration of ﬁbers that can be used in the orientation chamber.

The ﬁber orientation distribution in the ADF preform as well as the orientation
chamber was obtained using digital image analysis techniques. Under the prototype ADF
processing conditions, for all the E-glass ﬁber lengths nearly 70 % of the ﬁbers of the ﬁbers
were aligned i 20° and nearly 80 - 90 % within :h 30", indicating a high degree of ﬁner
alignments using electric ﬁelds. Further, improvements in alignments can be made based on

the understanding gained on the alignment behavior of ﬁbers in air.

140

Polymer powder coating of the ﬁbers and thereby controlling the matrix volume
fraction in the ﬁnal composite is one of the biggest advantages of the ADF process. The
combination of powder coating with electric ﬁeld alignment makes the process a potentially
high-speed, low—cost process to make aligned discontinuous ﬁber composites. This step
eliminates the additional step of resin transfer. The technology of powder coating is well
established in the group and extensively patented and hence can be easily incorporated in the

ADF process developed.

The prototype process in its current conﬁguration has a rated capacity of
manufacturing 2 lbs/hr of ADF preform. Conventional scale up factors i.e. increasing the
width of the orientation chamber from the existing 20 cm to the desired width and using
multiple orientation chambers to build up ADF preform rapidly result in direct increases in
production rates. These scale-up factors can be directly applied to the process without

affecting the principle of operation in any way.

10.

ll.

12.

13.

References

Drzal, L. T., Padaki, S., Vyakarnam, M. N., Femandes, J. H., “Powder
lmpregnation of Advanced Composite Materials”, in Polymer Powder
Technology, Ed.. Narkis, M. and Rosenzweig, N., John Wiley & Sons Ltd., pp
511-530, 1995.

lyer, S. R., “Continuous Processing of Unidirectional Prepreg”, Ph.D.
Dissertation, Michigan State University, 1990.

Vyakarnam, M. N., “Development of a High Speed Powder Process to
Manufacture Composite Prepreg”, MS. Thesis, Michigan State University, 1992.

Iyer, S. R. and Drzal, L. T., Powder Technology, 57(2), p 127-133, 1989.
Vyakarnam and Drzal, US. Patent 5,310, 538, 1994.

Padaki, S. and Drzal, L. T., "Development of a Process and Consolidation Model
for Powder Prepreg Composites", Proc. 1 Oth. Annual ASM/ESD Advanced
Composites Conference & Exposition, Dearbom, 1994.

Fisher, G. And Eyerer, P., “Measuring Spatial Orientation of Short Fiber
reinforced Thermoplastics by Image Analysis”, Polymer Composites, Vol. 9, No.
4, p 297-307, 1988.

O’Connell, P. A. And Duckett, R. A., Comp. Sci. & Tech., Vol . 42, p 329-347,
1991.

Liang, E. W., Poslinski, A, J. and Stokes, V. K., “Fiber Orientation in
Discontinuous Fiber Composites: An Overview, General Electric Report No.
93CRD136, 1993.

Okagawa, A. and Mason, S. G., J. Colloid Interface Sci, Vol 47, p 568, 1974.

Fermigier, M. and Gast, A. P., "Structure Evolution in a Paramagnetic Latex
Suspension", Journal of Colloid and Interface Science, Vol 154, No. 2, 1992.

Padaki, S., “A Process and Consolidation Model for Polymer Powder lmpregnated
Composite Tapes”, Ph.D. Dissertation, Michigan State University, 1995.

Gutowski, T. G., Morigaki, T., and Cai, Z., “Consolidation of Laminate
Composites”, Journal of Composite Materials, Vol. 21, pp. 172-188, 1987.

141

Chapter 5

 

Microstructure-Property Relationships

5. 1 Introduction .................................................... 143
5.1.1 Microstructure-Property Relationship Flowchart ................. 145
5.2 Microstructure Descriptors ........................................ 147
5.2.1 Fiber Orientation Distribution ................................ 150
5.2.2 Fiber Aspect Ratio ........................................ 152
5.2.3 Fiber-Matrix Interaction ..................................... 154
5.3 Elastic Property Predictions ........................................ 156
5 .4 Reinforcement Models ............................................ 1 60
5.4.1 Shear Lag Theory .......................................... 161
5.4.2 Eshelby’s Inclusion Approach ................................ 164
5.4.3 Halpin-Tsai’s Relationships .................................. 166
5.5 Experimental Results ............................................. 169
5.5.1 Sample Preparation and Tensile Testing ........................ 169
5.5.2 Effect of Fiber Alignment ................................... 169
5.5.3 Effect of Fiber Length ...................................... 170
5.6 Model Predictions and Discussion ................................ 170
References ........................................................... 177

142

143

5.1 Introduction

Discontinuous or short ﬁber composites pose a rich array of microstructural
combinations that oﬁen makes them very attractive for such a wide range of applications.
However, it is also this variation in microstructural features that makes it difﬁcult to describe
the thermo-elastic behavior of these materials using simple models'. These materials can be
can be anything from ﬁber reinforced injection molded polymer composites to whisker
reinforced polymer/metal matrix composites or chopped glass ﬁber Sheet Molding
Compound (SMC) composites where the mechanical performance of these materials is
enhanced by the presence of ﬁbers. In a majority of the processes that are used to fabricate
discontinuous ﬁber composites, there is generally little or no control on the ﬁber orientation
distribution (F OD) during the processing stage, and more often than not the FOD of these
composites is not known, which makes the prediction of properties unreliable. The
development of the Aligned Discontinuous Fiber (ADF) composite process using electric
ﬁelds, has the potential to control the orientation state of ﬁbers during the processing stage
of these composites. The ability to fabricate discontinuous ﬁber composites with
microstructure that is known a priori provides a unique opportunity to both investigate the
fundamental microstructure-property relationships and evaluate the ability of the ADF
composites. This study has two objectives.
0 First it serves to validate the guiding principle that motivated the development of the

ADF processing method i.e. the elastic performance of discontinuous ﬁber

composites approach that of continuous ﬁber composites when the length of the

ﬁbers are greater than critical ﬁber length and when they are aligned in the direction

144

of the stress. Some of the questions that are addressed in this study are: What
reinforcement architecture is important in chopped ﬁber-thermoplastic systems? Is

it the individual ﬁber aspect ratio or the bundle (aggregate) aspect ratio?.

- Secondly, this study will provide insight into the elastic behavior of the chopped ﬁber
- thermoplastic systems. This class of composites is gaining popularity for semi-
structural applications in the automotive and durable goods industry because of their
improved energy absorption characteristics; environmental ﬁ'iendliness due to
recyclability, repairability and solvent free processing; and ﬁnally due to high speed
manufacturing. Most of the composite reinforcement theories and their experimental
veriﬁcation have focussed on either continuous or single ﬁber cases in thermosetting
resins due to their extensive usage in the composites materials area so far. The
effect of the aggregate (bundle) nature of the ﬁbers on mechanical properties of
discontinuous ﬁber reinforced composites is not well understood"2'3. Despite, the
extensive use of discontinuous ﬁber composites and the extensive data on properties,
very few studies exist, with the exception of Kacir et. a14 and Piggott et. als. that

combine theoretical property predictions with accurate experimental validation.

Interestingly, the aspect ratio of ﬁbers which played such a signiﬁcant role in the
alignment behavior in electric ﬁelds also seems to be a crucial factor in the reinforcement

behavior. So, it was all the more logical to investigate the microstructure-property

145

relationships of discontinuous ﬁber composites and arrive at conclusions that will have a

direct bearing on both the processing and performance of discontinuous ﬁber composites.

5.1.1 Microstructure-Property Relationship Flowchart

In this chapter, composite reinforcement theories are analyzed and modeled for the
discontinuous ﬁber composites, to predict the modulus of these composites as a function of
the ﬁber orientation distribution, ﬁber aspect ratio and the ﬁber-matrix interaction. These
predictions are then compared with experimentally determined tensile properties. The
overall approach in modeling the microstructure-property relationship is depicted in the
ﬂowchart shown in Figure 5.1. Computations are performed with input from the ﬁber-matrix
constituent properties and the ﬁber volume fraction. The microstructure descriptors: ﬁber
orientation distribution (FOD) and ﬁber aspect ratio are individually evaluated and also
become inputs to the reinforcement models. The ﬁber orientation distribution in the
composite is represented by a statistical distribution function. Modiﬁed shear lag, modiﬁed
Eshelby’s elastic solution and Halpin-Tsai models are analyzed and evaluated for the effect
of ﬁnite ﬁber aspect ratio on the modulus of perfectly aligned discontinuous ﬁber
composites. The properties from the material direction can be transformed to the three
principal directions by invoking the transformation matrix. Property transformation is
integrated over the thickness of the composite using the ﬁber orientation distribution (FOD)
function. The computations in the preceding steps leads to the prediction of the elastic

properties of the ADF composites. The computed values are then compared with

146

  
     

 

 

 

.362
o>an_

9.2—9.3: aim—8:20.. area—Eatosuoeameom—z ﬁm «Sum...—

8383. .282

 

wEwEo>< omboE=_o>
, can cases—.832“; coca—5:0
.3 weaken—8m 862.800 860.5

cosmocto>
.8585qu

 

 

 

+

25323— Bahia—m:
ho
333 8582

 

 

 

 

.5
wﬁ 32% BEBE

>

 

 

 

 

5:082:— 5:32-837.—
otmm 8on1..
:ozzﬂbma 225:3.0 Se..—

 

.252
832—88832

 

Sci—om

 

 

 

 

.352 0.. 39:32:22

.3;

83.3.3.5 E05550

 

 

 

147

experimentally determined values. A comparison between the experimental results and the

model predictions will be presented at the end of the chapter.

It should be noted that the term “ADF composite” used in this chapter refers to any
discontinuous ﬁber composite that is made using the ADF process described in Chapter 4.
For micromechanics purposes, these composites will be further qualiﬁed by the terms
“Aligned” and “Random” representing the cases when ADF composites were made using the
E-ﬁelds to align the ﬁbers and without using the E-ﬁeld respectively. Further, the Aligned
ADF composites should not be considered as perfectly aligned discontinuous ﬁber
composites. In fact they have some misalignment of ﬁbers and the true ﬁber orientations are

described by the F OD.

5.2 Microstructure Descriptors

Three principal features of the microstructure that dictate the ﬁnal mechanical
properties of discontinuous ﬁber-matrix composite system for a given ﬁber volume fraction,
are: (a) ﬁber orientation distribution (b) ﬁber aspect ratio and © ﬁber-matrix interaction. The
term ﬁber-matrix interaction includes the effect of ﬁber sizing and the interfacial adhesion
and also the ﬁber wet-out or resin impregnation inside the ﬁber bundle in case of aggregate
reinforcement. In fact, the ﬁrst two factors play a major role in the modulus predictions
compared to the third factor, which becomes critical in the case of strength and fracture

behavior". The effect of ﬁber volume fraction on discontinuous ﬁber composites can be

148

easily extrapolated once the effect of the microstructural features are well understood.

Modulus and strength values tend to be linear functions of the ﬁber volume fraction.

To keep the experimental part of the study focussed to the inﬂuence of F OD and the
aspect ratio, the ﬁber volume fraction is ﬁxed at 40 % and the same constituents; E glass
ﬁbers and nylon-12 matrix, whose properties are shown in Table 5.1 are used throughout the
study. Each of the test panels that were fabricated were veriﬁed for ﬁber volume fraction
using a burn out test. At least two samples were tested from each panel and the ﬁber volume
fraction was found to be around 40 % with a i: 2% variation (Table 5.2). The burn out test
also gave the void content in the composite sample. It was typically around 2 to 3 % or

slightly higher in a few cases.

The void content seemed to be on the higher side when compared with good
continuous ﬁber composites which are typically under 1 %, but this value seemed reasonable
for compression molded chopped ﬁber thermoplastic composites because of the difﬁculties
in matrix impregnation into the chopped ﬁber bundles. Studies7‘8 relating the effect of void
content on the reduction in mechanical properties, indicate that the void content does not
have a signiﬁcant effect on the modulus values as long as it is typically 3 % and under.
However, strength values could be more inﬂuenced by the presence of voids. Optimization

of processing conditions should further reduce the void content.

Table 5.1 Constituent Properties

149

 

 

Property E-Glass Nylon-12
Tensile modulus, EIl (Msi) 10.5 0.19
Shear Modulus, G,2 (Msi) 4.3 0.07
Poisson Ratio, v12 0.22 0.33

 

Table 5.2 Volume Fraction and Void Fraction of ADF Composites

 

 

 

 

 

 

 

 

 

 

Fiber length Orientation Fiber Volume Void fraction
(in) Fraction (%) (%)
0.125 Random 41.3 2.6
0.25 Random 41.7 3.0
0.5 Random 42.5 2.9
1.0 Random 38.0 3.5
0.125 Aligned 41.9 3.2
0.25 Aligned 40.7 2.7
0.5 Aligned 44.6 2.7
1.0 Aligned 42.5 4.3

 

 

 

 

 

150

5.2.1 Fiber Orientation Distribution

One of the objectives of developing the ADF process was to demonstrate the
feasibility of making ADF composites with controllable ﬁber orientation distributions
(FODs). Typically once the processing is done for the manufacture of an ADF composite,
its FOD is obtained in an indirect way by image analysis. The method adopted in obtaining
the FOD consists of running the ADF process once again under identical conditions but this
time not with the intention of making a composite but recording a statistically signiﬁcant
number of images using a Panasonic CCTV. A number of images of the ﬁbers that are
deposited on the Teﬂon release ﬁlm are taken to give a statistically signiﬁcant FOD. The
Global Lab image analysis software is used to process the digital image, identify ﬁbers and
determine their orientations in each image. It has also been carefully veriﬁed that the FOD
of the ADF mat is not disturbed during the subsequent processing of the composite.
Therefore, it may be concluded that the F OD obtained by this method is a true representation

of the FOD in the ADF composites (Table 5.3).

The general three dimensional orientation state of short ﬁber in a composite can be
described by the angle (3 and (I). The transformation matrix from the original coordinate
system of the ﬁber to that of the three principal directions can be obtained by T (3D) as

deﬁned below.

sinBcosd) sinBsind) c086
T (3D) = cosBcosd) cosOsind) -sin6 (1)
-sin([) cosd) 0

 

 

151

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m; m 85 86 o; 85 _ N; H Ed _ m; e
8a :2 :3 5m .33 com m2 2..“ 2
83 8:. Ed NS. Re mom 23 2:. 3
Ed 85. a: NE. mg 8... Rd mm... 8
m3 m3 :2 3m 33 3... a: 8... mm
cod .3; Nam 2% Nam as 8._ 2% 8
m3 :4. 8.... mom Sm mam N3 88 mm
88 com 8.4 mg m? 8... 8.4 9.8 mm
8.x 8:. 8.8 88 3.2 New 8.2 Rs 2
8.: 88 2.2 88 ~22 Mac 8.8 as m
8.8 com 8.8 a? 8.2 mg 3:. a; m-
3.: 8.” 2.: :8 8.2 8.8 3.2 8.5 2-
S... 8:. 2.2 OS 3.8 cam 82 8.8 2.
23 com mum 88 5.4 3.8 wow as 2.
m3 8? 9% NE N2 88 N3 New 2.-
2.. 2... N3 88 e: 28 8; m2. 8.
a _ ._ 3.8 gm 3.... :.N 8... :3 N3. 8-
one 98 a; 88 82 RM o2 am... we-
and mom .8; Km of am... c: Q...” 3.
g 88.2 3 888m $3 88? 3 588m 2% 88:... Aéeoncé g 88.2 3 588m a -1882
L a; ii a: .884 .25

 

 

 

 

8:8..an an... .8353... E... ..8..u=<.. 8 22.3.85 5.2.5826 he... 8.882 nm 2...?

 

152

For planar orientation, the elements of the rotational transformation T (2D) simplify to the

following form.

 

 

cosd) sind) 0
T (20) = -sin<[) cosd) 0 (2)
O O 1

It is pertinent to point out here that the ﬁbers in the ADF composites lie in a planar state and
hence the two dimensional transformation matrix (2D) will sufﬁce. This assumption is valid
when the length of the ﬁbers is greater then the thickness of the composite. For a planar (2D)
orientation of ﬁbers, the orientation state can be characterized by a single parameter fp,
which is a trigonometric average of the distribution obtained from the above transformation,

when the FOD is symmetrical.

fp = 2<cosz¢> - 1 (3)

Table 5.3 gives the FODs and the fp values of the ADF composites that were fabricated.

5.2.2 Fiber Aspect Ratio

A statistically signiﬁcant number of ﬁber lengths and their weights were measured
for each type of ﬁber that was supplied and used for making ADF composites. The ﬁber
length distribution was found to be very narrow in each case and hence the nominal length
was used in all subsequent computations. The mean number of ﬁlaments per bundle were

determined from the weight and length measurements of typically 100 ﬁber bundles of each

153

length and type. It was found that there was a variation in the number of ﬁlaments per
bundle and was not as sharp a frequency distribution as was the case with ﬁber length. The
ﬁber length and the mean number of ﬁlaments in each type of ﬁber bundle that was
investigated in this study is given in Table 5.4. In the chopped glass ﬁber systems, the
reinforcement aspect ratio is unknown. It is not clear if it is just the ratio of bundle length
to the bundle diameter or if it is the ratio of the length of the bundle to the diameter of
individual ﬁlament. It is postulated that it will depend on the degree of impregnation into
the ﬁber bundle. Good impregnation is achieved when there is a good wet out between the
ﬁber and the matrix and when an optimum time-temperature-pressure consolidation cycle
is used. It is also felt that good thermoplastic resin impregnation into the ﬁber bundles can
be achieved only when some shearing action. The compatibility of the sizing and the resin
impregnation was investigated for the polyester sized and the nylon sized cases as shown in
the micrographs (Figures 5.2 and 5.3). The ﬁber bundles distinctly remain as bundles in the
ADF composite fabricated with polyester sized ﬁbers due to poor wet out (Figure 5.2), while
in the case of nylon sized ﬁbers there is a good dispersion of the ﬁbers and the nylon12

matrix has impregnated the individual ﬁber bundle (Figure 5.3).

As explained in § 3.3.3.1, the aspect ratio of the chopped ﬁber transformed into an

equivalent ﬁber bundle aspect ratio.

db.» 4
a =—= —N 4
9 df 1t bf ()

154

Table 5.4 Fiber Bundle Length and Aspect Ratios

 

 

 

 

 

Nominal Filament Effective
Length (mm) Aspect ratio Bundle Ratio
3.2 246 14.5
6.3 485 28.5
12.7 977 57.5
25.4 1985 1 14.9

 

 

 

 

 

Note: Supplied by Owens Corning with polyester compatible sizing; average ﬁlament diameter - 13 um
and the average effective bundle diameter is 221 um based on a square packing of an average number of 227
ﬁlaments per bundle.

In the event of an aspect ratio distribution, the root mean square of the effective aspect ratio
(ac)is to be taken:

WW (5)

<aez> = faezp(a)da

The chopped ﬁbers used in the ADF process had a very tight distribution in lengths. The
number of ﬁlaments that were present in the ﬁber bundle were also consistent. Therefore,
it was assumed that there was no signiﬁcant distribution in the ﬁber length and hence, mean

ﬁber lengths are used in the determination of aspect ratios.

5.2.3 Fiber-Matrix Interaction
These ﬁber-matrix interactions are quite complicated to describe in simple
quantitative terms and hence in the present study they will be dealt in qualitative terms. Two

extremes cases are generated by choosing two sizing chemistries. The ﬁrst one is a polyester

155

 

Figure 5.2 “Bundle” reinforcement effect due to in-compatible fiber-matrix
sizing.

 

Figure 5.3 “Filament” reinforcement effect due to compatible ﬁber-matrix sizing.

156

compatible sizing which is certainly not the optimum one for nylon 12 matrix and the other
one is a nylon compatible sizing. The intent is not to relate the performance of discontinuous
ﬁber composites with different adhesion levels as was done in a very comprehensive study
for continuous ﬁber composites by Madhukar and Drzal“, but to make observations on the
effect of sizing chemistry on the modulus and strength values. Microscopic examination of
the composites made from these different sizings indicated a more “aggregate” or “bundle”
phenomenon in the polyester compatible sizing (Figure 5.2) and the ﬁbers are more

“dispersed” in the nylon compatible case (Figure 5.3).

5.3 Elastic Property Predictions
The generalized Hooke’s law for the stress (o)-strain (6) relationship for any material

is given by:

C”. e].
E. = S. 0,
a J

(6)

Where, C”- is the stiffness matrix and SD» is the compliance matrix and [S] = [C]". Fiber
reinforced composite materials are usually highly anisotropic in nature and may require up
to 21 elastic constants instead of the 2 elastic constants that are required for isotropic
materials to relate stress and strain in the body. The engineering properties like tensile
modulus, shear modulus and poisson’s ratio in the three principal directions are obtained

from the compliance matrix, [C].

157

The problem of the elastic behavior of discontinuous ﬁber composites can be
simpliﬁed by considering it to be equivalent to a series of orthotropic laminates of perfectly
aligned discontinuous ﬁber composite plies stacked up in a series with different ﬁber
orientations (Figure 5.4). The sequence of ﬁber orientation will be a weighted function of
the ﬁber orientation distribution in the composite. By invoking the laminate approach, the

implicit assumption is that plane stress conditions apply as deﬁned below.

021 =1 =0 (7)

The assumption of planar orientation holds good as long as it is known that there is no tilting
of the ﬁbers “out of plane” and when the length of the ﬁber is sufﬁciently longer than the
thickness of the composite. Both of these conditions are satisﬁed in the ADF composites that
are fabricated in this process. This results in further reduction in the number of elastic
constants that are needed to characterize the elastic behavior in the plane to ﬁve, as indicated

in the in the e - 0 relationship below.

11 12

f6 - “S S
E = S 822

0.
O 02 (8)
Yiz O O Seoutlz‘

L . .

I2

 

 

 

 

 

 

As shown in Figure 5.4, now the off axis properties have to be determined. A transformation

matrix (T) is used to relate the stresses in 1-2 direction with the x-y direction.

158

 

.nozneaﬁeo 93 he gang—Fauna 5.2.9:— 8.3.0 .Su 2.3—So 935555532 1m «.5»:—

38980 Ba..— 385885 853 3.8%».—

 

 

 

 

 

 

 

 

 

 

 

159

 

 

 

 

 

 

r 1 . . .
Ox [ cosz¢ sin2¢ —25in¢cosd> 01
0y = sinzd) c0324) 2sin¢cosd> 02 (9)
30'. _sin(bcos¢ -sind)cos¢ c0324) - sin2d>4 -1312‘

The transformation matrix, T also transforms the strain in the 1-2 direction to x-y direction.
On further simpliﬁcation of these matrix operations to determine the engineering properties
in the 1-2 direction as a ﬁanction of the x-y direction", we get the expression for modulus (E)

as function of the engineering properties (El, E2 and Gl2 ) in the material direction.

4 2v ' 4
—1— = C—OS—gl + (% - —'2)sin2¢cosz¢ + sm (1)

E E (1,, El E2

x l

 

(10)

Now, the important thing is to determine the engineering properties El, E2 and G,2
as a function of the reinforcement aspect ratio. Fundamental composite reinforcement
models will be analyzed and used in the prediction of these engineering properties for the
perfectly aligned discontinuous ﬁber composites and then predict the properties of ADF
composites and later compare with experimental results. Since volumetric averaging over
the thickness of the composite may be applied in the case of a uniform strain, the modulus
of the composite Ec is ﬁnally obtained by integrating the expression for Ex with the ﬁber

orientation distribution.

E. = f E,(<b) n(<b) dd> (11)
0

160

The planar ﬁber orientation distribution function n(¢) gives the relative distribution of ﬁbers
with respect to the external coordinate axis, where the reference axis (x) can be the machine

direction in the ADF process.

5.4 Reinforcement Models

In an effort to model the thermo-elastic properties of discontinuous ﬁber composite
system, various composite reinforcement theories are evaluated for their applicability to
composites reinforced with ﬁnite aspect ratio ﬁbers and variable orientation state dispersed
in a thermoplastic matrix. Shear lag, Eshelby’s approach and Halpin-Tsai semi-analytical
relationships are the three most used reinforcement models in composite mechanics. The
applicability of one model over the other in different situations is still not well understood
especially in the case of chopped ﬁber - thermoplastic matrix systems. Hence, all three
models were considered in this analysis. For each of the models discussed below, Ec is
computed as a function of the aspect ratio. The overall strategy is to obtain the
microstructure-property relationships of discontinuous ﬁber composites as a function of the

ﬁber aspect ratio, and ﬁber orientation distribution.

The Rule of Mixtures (ROM) models generally give good predictions in the case of
composites with “perfect” bonding between the ﬁber and the matrix. A priori one cannot
assume its applicability to short or discontinuous ﬁber composites because the bounds are
too far apart for any practical utility and do not take into account the ﬁber aspect ratio or the

ﬁber orientation distribution.

161

Voigt2 bound (upper bound) is obtained by applying a constant strain to a composite
consisting of unifome distributed continuous ﬁbers in a matrix. The composite property

P is a simple weighted summation (parallel reaction) of the constituent properties.

P = ZPiVi
E = £ij + 5me

C

(12)

Reuss2 bound (lower bound) is obtained by applying a constant stress to a composite
consisting of uniformly distributed continuous ﬁbers in a matrix. The reciprocal of the
composite property P is a weighted summation of the reciprocal of constituent properties

(series reaction).

V.
I) Z _.l.
2 P’ 13
i. : _I/_f + _I/_”l ( )
EC Ef Em

or a ﬁber volume fraction of 40 %, Ec for the E-glass ﬁber and nylon-12 matrix system, the
bounds will lie between 4.3 Msi (upper) and 0.3 Msi (lower). The futility of using the rule
of mixtures for the prediction of modulus of discontinuous ﬁber systems can be immediately

appreciated when these wide bounds are compared with experimental results given in § 5.5.

5.4.1 Shear Lag Theory

Shear lag theory was originally proposed by Cox”), where the tensile stress is

transferred from the matrix to the ﬁber through shear stresses. Here both ﬁbers and the

162

matrix behave elastically and the interface transfers the stress from the ﬁbers to the matrix
without yielding or slip. Perfect interfacial bonding is assumed and the axial strain of the
composite is taken as that of the matrix in the far ﬁeld. Using this approach the stiffness of
the composite and the stress for the onset of matrix yielding or interfacial sliding can be

calculated with the relationship given below.

tanh(ns)] + E V

E =E 1-
c jyf[ "S
2E

where n2 = m s =

E,(1 + vm)ln(1/Vf)

 

(14)

 

In the Kelly-Tysonll slip model, which is a special case of the shear-lag theory it is assumed
that the matrix shear modulus is a constant which results in a constant interfacial shear stress.
The stress transfer relationship gets simpliﬁed as explained in § 1.2.1 (see also Figure 1.1).
Based on this slip model, the concept of critical ﬁber fragmentation length and its
experimental determination has been developed by Drzall2 and others. Matrix plasticity is
applicable to well bonded reinforced metals and frictional sliding to reinforced polymers and
ceramics. In the case of polymer matrices the slip-elastic formulation is presented by
Piggottl3 seems to explain the behavior better. Further modiﬁcations to the shear-lag model
are introduced by taking into account the effect of the stress transfer across the ﬁber ends for
discontinuous ﬁber composites. This factor is important when small aspect ratio are being
dealt or small ﬁber/matrix stiffness ratios, for example in the case of injection molded
composites or whisker reinforced metal matrix composites. This factor is incorporated in

the present analysis.

163

One of the chief drawbacks of the original shear-lag model has been the source of
error that is introduced due to neglecting the transfer of normal stresses across the end of the
ﬁber. There have been several attempts to correct this situation. The recent solution
presented by Clynel4 is not only simple but also intuitively logical and hence will be
considered here along with the original Cox’s model. The modiﬁcation is brought about by
introducing an expression for the ﬁber end stress (06) . Since the value of 0,, must lie
between the peak stress in the ﬁber (0(0) and the far ﬁeld stress in the matrix (am), the

following expression is proposed.

0 : f0 mo (15)

For the modiﬁed Shear Lag model, taking into account the ﬁber end stress transfer, we get

 

 

E — E '
EC : E/V/[l _ (f m) tanh(rzs)]+ Eme
- Ef ns (16)
E l—sech ns + E
where Em' = f [ 2( H m

The modiﬁed shear-lag model by Clyne has been tested for whisker reinforced metal matrix
composites and found that the modulus values are closer to experimental values than those
predicted by the original shear-lag model. This is primarin due to the small aspect ratios of
the reinforcement (aspect ratio of whiskers is typically < 10) where the modiﬁcation has the

maximum effect.

164

5.4.2 Eshelby’s Inclusion Approach

Eshelbyls solved for the elastic ﬁeld in and around an ellipsoidal inclusion and also
the associated strain energy of this system. The original model assumes an ellipsoidal
inclusion with a uniform non-elastic strain, embedded in an elastic body. In the case of an
ellipsoidal inclusion in a matrix of stiffness tensor, CM, the stress (0,) within the inclusion

can be obtained using Hooke’s law in terms of the elastic strain, (eC - 6T ).

0l 2 CM (EC — ET) (17)

For a shape change (ET) in the inclusion due to the imposed external stresses, all that is
required is the knowledge of the ﬁnal constrained strain,eC. Eshelby found that EC can be
obtained from ET using a tensor which is known as the Eshelby “S” tensor. This tensor is a

function of the inclusion aspect ratio and the Poisson’s ratio and is given by

EC = S 6T (18)

With the constrained shape known, the inclusion stress can be evaluated as

ol = CM (S - 1) e7” (19)

The S tensor allows for the calculation of the uniform stress and strain within the inclusion
without having to look at the complicated stress ﬁeld in the matrix itself. Numerous

advances have been made to account for ﬁber-ﬁber interactions in composites by Mori-

Tanaka'6 and others‘m.

165

Taya and Arsenaultl9 compared Shear-lag and Eshelby models for the prediction of
modulus in short whisker reinforced metal matrix composites (MMC). The focus of their
study was to compare the two models for aspect ratios less than 10. The rigorous solution
of Eshelby’s approach gives better predictions of the elastic properties in the event of small
aspect ratio inclusions (<10) and when the modulus mis-match between the stiff ﬁbers and
the matrix is small (E/Em <10). This implies, this approach is more applicable in the case
of whisker reinforced metal matrix composites than typical polymer matrix composites,
where both these conditions are satisﬁed. In the case of short ﬁber reinforced polymer
composites, say injection molded composites, the ﬁber aspect ratios are typically between
10 and 50 and Ef / Em > 20. At these aspect ratios the modulus values start leveling off and
oﬁen the computational effort of this approach does not merit usage of this model among the

users of polymer composites.

Eshelby’s approach was also considered by McGee and McCullough20 to predict the
elastic properties of Sheet Molding Compounds (SMC). The random nature of the ﬁbers in
SMCs creates a grainy micro-structure of ellipsoids (aspect ratios <5) where each ellipsoid
is an aggregate of well aligned ﬁlaments. The problem is similar in some respects to that of
the ADF composites, where the chopped ﬁbers remain as ﬁber bundles even after
consolidation. However, the aspect ratios are considerably higher in the case of ADF
composites compared to the SMC composites (see Table 5.3). In SMC composites the
ﬁbers curve and get entangled creating grainy microstructure which can be modeled as

ellipsoids of aspect ratios typically less than ﬁve. A computer program called SMC that was

166

developed at the University of Delaware is used in this study to predict the composite
modulus at different ﬁber aspect and different ﬁber orientation parameters based on
Eshelby’s approach. The ﬁber orientation distribution is considered symmetrical and
characterized by a convenient trigonometric average parameter “fp” deﬁned in the following

fashion:

fp = 2<cosz¢> - 1 (20)

When fp =1, the ﬁbers are perfectly aligned and when fp = 0, they are perfectly random. An
fp value of 0.5 implies the ﬁbers are reasonably aligned. The SMC program is used to
determine the modulus Ec of the composite, using the fp value and the aspect ratio of the

reinforcing ﬁber.

5.4.3 Halpin-Tsai’s Relationships

Halpin-Tsaiz' equations are obtained by reducing rigorous elastic solutions to simpler
analytical forms where the ﬁber geometries are taken care through the use of some empirical
factors. The relations that are pertinent for the estimation of composite modulus or any

property, Pare given below where and b are geometrical dimensions of the reinforcement.

 

E . :33
Pm I - n19
— (1)/Pm) _1 C— 2a (21)
n (P/Pm) +c b

167

The relations have been veriﬁed for injection molded composites where the ﬁber aspect
ratios are typically greater than 10 and it has been found that experimental results are close
to predictions obtained from these relations. Halpin and Kardos22 have also predicted the
properties of short ﬁber composites based on these relations and using micro-laminate

analogy.

Figure 5.5, shows a simulation of the composite modulus (Be) as a function of the
aspect ratio, using the Shear lag models and the Halpin-Tsai relationships and the SMC
program. Note these results are for the perfectly aligned cases. In all the models, the
modulus values increase sharply in the beginning and reach an asymptotic value as soon as
the aspect ratios reach around 100. Notice how the incorporation of the stress at ﬁber ends
in the shear-lag model has shifted the curve upwards and reaches the asymptotic value
sooner. In the simulations of perfectly aligned discontinuous ﬁbers composites itself there

is quite some variation in the prediction of each model.

In the real ADF composite, the mis-alignment of ﬁbers has to be taken into account.
It is needless to say how an accurate FOD is very important in predicting the properties of
the actual composite. In the next section the elastic properties of the actual composites are
predicted by incorporating the actual ﬁber orientation distributions of individual composites

and compared with experimental results.

.23.. 32.2. 3088.859. 2: .3 .3325.— a 3 352.58 .83..— 2.3-5.88% gauze 3.. 2:35.: .3 329:3...— m.m 9.5!...

 

168

 

 

28¢ 581
89 8. 9 F
.i . a o
- - .. no
533.2316!
cm: 5%: le ,
cvmvcaixlll, .. F
:Fmvoczolll .
25:816.:
. m...
u
Ililllliilxl m
i N m
m
a
t mN m
i. 0
Av m.”
.. 4
1| I \

 

mi

 

169

5.5 Experimental Results
5.5.] Sample Preparation and Tensile Testing
The mechanical performance of ADF composites was characterized by determining
the tensile properties using ASTM D638. A series of ADF composites were prepared using
the ADF process using the ﬁbers supplied by Owens-Coming (see Table 5.1) with lengths
ranging from 0.125 to 1". For each ﬁber length, two panels were fabricated - one is the
“Aligned”case operated at an B ﬁeld intensity of 400 KV/m and the other is the “Random”
case without any B ﬁeld. FODs are determined for each of these cases and are given in Table
5.3. Chopped glass ﬁber-nylon12 ADF composite panels (4.75"x7"x0.0625") having a
constant ﬁber volume fraction of approximately 40%, were cut into dog bone specimens
using a C02 laser source of 360 Watts operated at a cutting speed of 30"/min. The specimens
were then carefully cleaned from the edges to avoid any crack initiation during the test.
Uniaxial load was applied on the dog bone specimens at a cross head speed of 0.2"/min in
a standard UTS machine. This unit was also equipped with a laser extensiometer, which
provided the strain on the sample. Tensile modulus and tensile strength values were obtained

from each specimen tested and ﬁve specimens were tested in each panel.

5.5.2 Effect of Fiber Alignment

The tensile modulus and tensile strength of the ADF composites are plotted as a
function of the ﬁber length in Figure 5.6 and Figure 5.7 respectively. As expected there was
a signiﬁcant improvement in the tensile properties of the ADF composites when the ﬁbers

were aligned, compared to the base line case (with the same ﬁber volume fraction) where the

170

ﬁbers are randomly oriented. For each of the ﬁber lengths, the improvements in modulus
values were nearly 70 to 100 %, while the improvements in the strength values ranged

between 60 to 90 % over the non-aligned case and the un-reinforced nylon-12 matrix.

5.5.3 Effect of Fiber Length

The tensile properties of the ADF composites increases with ﬁber length as one
would expect from the predictions of composite reinforcement theories. Both tensile
modulus and strength values increase signiﬁcantly with length when the ﬁbers are aligned
and the increasing trend seems to go beyond a ﬁber length of 1 inch. In the random ﬁber
case, the modulus values seem to level off once the ﬁber length reaches 0.5", indicating that
ﬁbers longer than this length would not improve the stiffness of the composite in any

signiﬁcant way. A similar trend was observed for the strength curve but to a lesser extent.

5.6 Model Predictions and Discussion

Based on the critical ﬁber length postulation that was presented in Chapter 1, the
properties of aligned discontinuous ﬁber composites should approach the properties of
continuous ﬁber composites when the length of the ﬁber is about ten times the critical ﬁber
length (say about 0.5 inches). Indeed, the experimental results show signiﬁcant
improvements in properties (Figures 5.6 and 5.7). However, the modulus and the strength
values do not seem to level off once the ﬁber length exceeds 0.5 inches, indicating the
performance levels of continuous ﬁber composites have not been completely achieved. So

the question that remains unanswered is, what is the actual reinforcement length? Secondly.

Tensile Modulus (Mp3!)

171

 

  

 

 

 

 

3.000

+Random

1+Aligned
2.500 «~
2.000 ..
150° W H
1.000 1. {/r
0.500 ..
0.000 . . s s

0 0.25 0.5 0.75 1 1.25

Glass Fiber Length (Inch)

Figure 5.6 Effect of ﬁber alignment and ﬁber length on the tensile modulus.

1’th Strength (pl)

172

40000

 

191725135
35°“ “ -+.. AWL

 

 

 

 

o 0.25 0.5 0.75 1 1.25
can: Flbor Length (Inch)

Figure 5.7 Effect of ﬁber alignment and ﬁber length on the tensile strength.

173

what is the effect of ﬁber mis-alignment on the properties in the aligned ﬁber cases?
Thirdly, how good are the predictions from the reinforcement models that were elaborated

in § 5.4.

These questions can be answered by looking at the predictions obtained from the
fundamental composite reinforcement theories which also take into account the actual FOD
of the composites. Based on the above formulations and the experimental data obtained
computations were performed to determine the modulus of the composite, Ec at the two
extremes of the reinforcement aspect ratio. These two extremes of the aspect ratios are the
limits of interest based on the observations made from the micro-structural micrographs ize
“ﬁlament reinforcement “ and “bundle reinforcement” (Figures 5.2 and 5.3). Table 5.4
shows the details of the model predictions for the actual ADF composites that were
fabricated and tested. Based on the chopped ﬁber length, the two different aspect ratios are
considered: “ﬁlament” aspect ratio and “bundle” aspect ratio. Relationships presented in §
5.4 are then used to obtain Ec for the two different aspect ratios. These modulus values for
the “perfectly aligned composites” are then used to obtain the modulus of the actual ADF
composite by integrating the effect of ﬁber orientation using the FOD. Notice for each ﬁber
length, two ﬁber orientation states were tested: aligned and random. The ﬁber orientation

state was also characterized by the term “fp” to get the predictions from the SMC program.

174

Table 5.5 Model Predictions Compared with Experimental Modulus Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

"Aligned" ADF Composites
Fiber Length fp ] Expt. Result Halpin-Tsai (Msi) Eshelby (Msi)
(in) Msi (Std. Dev.) "Filament” "Bundle" "Filament" "Bundle"
0.13 0.74 1.58 (0.13) 2.37 1.48 3.22 3.22
0.25 0.63 1.77 (0.11) 2.16 1.66 2.92 2.89
0.50 0.63 2.08 (0.28) 2.38 2.04 2.93 2.72
1.00 0.73 2.6 (0.30) 2.4 2.21 3.25 2.59
"Random" ADF Composites
Fiber Length fp ] Expt. Result Halpin-Tsai (Msi) Eshelby (Msi)
(in) Msi (Std. Dev.) "Filament" "Bundle" "Filament" "Bundle"
0.13 0.18 0.93 (0.19) 1.17 0.87 1.86 1.86
0.25 0.12 1.10 (0.15) 1.15 0.97 1.97 1.96
0.50 0.10 1.16 (0.02) 1.12 1.01 2.00 1.87
1.00 0.02 1.32 (0.14) 1.07 1.02 2.08 1.71

 

 

 

175

Halpin-Tsai relations under predict the modulus for small aspect ratio ﬁbers and this
has been evident in the earlier comparisons between Shear lag and Halpin-Tsai relations also
(Figure 5.5). A closer look at the predictions made by the Eshelby technique indicates that
it over predicts and secondly the effect of aspect ratio also seems to decrease much faster
than any of the other techniques. When the ﬁber bundle effect is considered, the predictions
made by the Halpin-Tsai method are closer to the experimental data (within the experimental

scatter).

Based on the experimental results and the theoretical predictions it can be concluded
that the chopped E-glass ﬁber - nylon12 composite system that was studied, the
reinforcement microstructure is an aggregate of individual ﬁlaments and that the effect of
reinforcement is a function of the bundle aspect ratio. Since, the critical ﬁber length concept
of the shear-lag theory was based on the assumption of an individual ﬁlament surrounded by
matrix on all sides, it cannot be directly applied to this situation when the reinforcement
microstructure is an aggregate of individual ﬁlaments and hence was not computed for
comparison with experimental values (Table 5.4). Further work is required in modifying the
shear-lag concept for reinforcements that are aggregates of ﬁlaments. However, the
reinforcement trend that was shown in Figure 1.2 as a function of ﬁber length as predicted
by the critical ﬁber length concept is evidenced in the experimental results shown in Figure
5.6. This proves that the underlying concept of critical ﬁber length exists even in the case

of “bundle reinforcement”.

176

Finally, the following conclusions can be summarized from this study of the
microstructure-property relationships.

0 The properties of the discontinuous ﬁber composites improve with increase in ﬁber
alignment and an increase in ﬁber length.

0 In the chopped ﬁber-thermoplastic systems, the effective reinforcement aspect ratio
is a function of the bundle size and ﬁber-matrix interaction which includes resin
impregnation in to the bundle and the sizing compatibility.

- The effective reinforcement aspect ratio approaches the ﬁlament aspect ratio only
when the ﬁber-matrix interaction is optimized and the effective reinforcement aspect
ratio approaches the bundle aspect ratio when there is no resin impregnation in to the

ﬁber bundle.

10.

11.

12.

13.

References

Carman, G. P. and Reifsnider, K. L., “Micromechanics of short-ﬁber composites”,
Composite Science and Technology, 43, pp 137-146, 1992.

Eduljee, R. F. and McCullough, R. L., “Elastic properties of composites”, in
Material Science and Technology - A Comprehensive Treatment, Vol 13, Ed.
Chou, T. W., VCH, 1993.

Ericson, M. L. And Berglund, L. A, “Processing and Mechanical Properties of
Oriented Preformed Glass-Mat-Reinforced Thermoplastics”, Composites Science
and Technology, Vol. 49, p121-130, 1993.

Kacir, L., lshai, 0. And Narkis, M., “Oriented Short Glass-Fiber Composites. IV.
Dependence of Mechanical Properties on the Distribution of Fiber Orientations”,
Polymer Engineering and Science, Vol 18, No. l, p45-52, 1978.

Piggott, M. R., Ko., M. and Chuang, H. Y., “Aligned Short-Fier Reinforced
Thermosets: Experiments and Analysis Lend Little Support for Established
Theory”, Composites Science and Technology, Vol 48, p291-299, 1993.

Madhukar, M. S. and Drzal, L. T., “Fiber-Matrix Adhesion and its Effect on
Composite Mechanical Properties: 11. Longitudinal (0°) and Transverse (90°)
Tensile and Flexure Behavior of Graphite/Epoxy Composites”, J. Of Composite
Materials, Vol. 25, p 958-991, 1991.

Tang, J. M., Lee, W. 1., Springer, G. 8., “Effects of Cure Pressure on Resin Flow,
Voids and Mechanical Properties”, J. Of Composite Materials, Vol. 21, pp 421-
440, 1987.

Harper, B. D., Staab, G. H., Chen, R. S., “A Note on the Effects of Voids upon the
Hygral and Mechanical Properties of AS4/3502 Graphite/Epoxy”, J. Of
Composite Materials, Vol. 21, pp 280-289, 1987.

Piggott, M. R., “Load Bearing Fiber Composites”, Pergamon, 1980.

Cox, H. L., " The elasticity and strength of paper and other ﬁbrous materials",
British J. of Applied Physics, 3:72, 1952.

Kelly, A. And Tyson, W. R., “Tensile Properties of Fiber-Reinforced Metals:
Copper/Tungsten and Copper/Molybdenum”, J. Mech. Phys. Solids, Vol. 13,
p329-350, 1965.

Herrera-Franco, P. J. And Drzal, L. T., “Comparison of Methods for the

Measurement of F iber/Matrix Adhesion in Composites”, Composites, Vol. 23,
No. 1, 1992.

Piggott, M. R., “Short Fiber Polymer Composites: A Fracture Based Theory of
Fiber Reinforcement”, J. of Composite Materials, Vol 28, No. 7, 1994.

177

14.

15.
16.
17.

18.

19.

20.

21.

22.

178

Clyne, T. W., " A Simple Development of the Shear Lag Theory Appropriate for
Composites with a Relatively Small Modulus Mismatch", Materials Science &
Engineering, A122 (1989) 183-192.

Eshelby, J. D., “Proc. Roy. Soc. Lond., A241, 376, 1957.
Mori, T. and Tanaka, K., Acta Metall., Vol 21, 571, 1973.

Christensen, R. M., “Mechanics of Composite Materials”, John Wiley and Sons,
New York, 1979.

Mura, T., “ Micromechanics of Defects in Solids”, Nijhoff, The Hague, 1987.

Taya, M. and Arsenault, R. J ., “A Comparison Between a Shear Lag Type Model
and an Eshelby Type Model in Predicting the Mechanical Properties of a Short
Fiber Composite”, Scripta Metallurgica, Vol. 21, p 349-354, 1987.

McCullough, R. L., " Micro-models for Composite Materials-Particulate and
Discontinuous Fiber Composites" in Micromechanical Materials Modeling, 1989.

Halpin, J. C. And Kardos, J. L., “The Halpin-Tsai Equations: A review”, Polymer
Engineering and Science, Vol 16, No. 5, 1976.

Halpin, J. C., and Kardos, J. L., “Strength of Discontinuous Reinforced
Composites: I Fiber Reinforced Composites”, Polymer Engineering and Science,
Vol 18, No 6, 1978.

Chapter 6

 

Conclusions and Future Work

The problem of processing discontinuous ﬁber polymer composites with ﬁber
alignment in a preferred direction was solved by a unique combination of ﬁber alignment
using electric ﬁelds in air and polymer powder processing, which resulted in a rapid, solvent
free processing technique. The development of the prototype Aligned Discontinuous Fiber
(ADF) composite process was a result of the integration of the fundamental understanding
and investigations into the following aspects:

0 Behavior of ﬁber alignment in air using electric ﬁelds and

- Microstructure-property relationships of discontinuous ﬁber composites.

The conclusions from this study are summarized below.

1. A semi-continuous prototype process has been developed that can manufacture
aligned discontinuous ﬁber composites using electric ﬁelds. The key steps in the
process are: alignment of conducting or insulating ﬁbers in air using electric ﬁelds;
powder coating/impregnation of ﬁbers, and; consolidation of the powder coated ADF
preform into a composite. The ADF process has been demonstrated for E-glass ﬁber

and nylon12 matrix system and can be extended to other ﬁber-matrix combinations

179

180

E-glass ﬁbers of lengths ranging from 0.125 to 1 inch have been aligned using
alternating current (A.C.) electric ﬁelds with intensities ranging from 300 to 600
KV/m. A.C. ﬁelds were preferred over D.C. ﬁelds to avoid electrophoretic effects
which disturbed the ﬁber alignment process. Since the polarization times of the
ﬁbers are very rapid, A.C. ﬁelds with a frequency of 60 Hz was found to be adequate

for effective ﬁber alignment.

Fiber alignment in air using electric ﬁelds is extremely fast. The alignment times
for dielectric ﬁbers like E-glass range between 0.1 to 0.5 seconds for the ﬁbers
lengths investigated (ranging from 0.125 to 1 inch). This rapid alignment makes the

ADF processing technique very attractive for high speed processing.

The alignment behavior of the ﬁbers in air is more complicated than it would be in
a more viscous ﬂuid medium due to the non-buoyant nature of ﬁber motion in air.
A balance exists between the electrical torque due to the polarization of ﬁbers;
hydrodynamic forces due to ﬂuid resistance and rotational torque due to resultant
shearing action. With air as the ﬂuid medium, the viscous resistance is small and
gravitational and the rotational forces become important. Fiber alignment is a strong
function of E-ﬁeld strength, aspect ratio and the stability in the ﬁber motion.
Although it is predicted (in the case of buoyant situation) that the alignment time

increases with ﬁber aspect ratio, it has been found that longer ﬁbers have better

181

alignments for the same electric ﬁeld intensity due to their more stable motion during

free fall.

The ﬁnal design of the chamber for controlling ﬁber orientation in air has two
unique features that counter the edge effects of the electric ﬁelds when the deposition
platform is set in motion during the continuous processing of ADF preforms.

- A ﬁeld equal in intensity and opposite in direction has been added under the
high voltage electrode by placing under the moving slide/mat/veil. This
neutralizes the edge effects of the high voltage electrode, which otherwise
disturbs the ﬁber orientation of the ﬁbers that are aligned by the E-ﬁeld and
that are deposited on the mat.

- A special design for the edges of the electrodes which are closer to the
moving mat has been developed. A conductive cap which has a cylindrical
cross-section with a relatively large radius of curvature has been used to
reduce the effective E-ﬁeld intensity at the edge, thereby reducing the edge

effects and minimizing E-ﬁeld breakdowns.

Developed a unique vibratory ﬁber feeder and pre-aligner. Fibers with or without
powder impregnation can be fed into the orientation chamber without entanglement.
Fibers falling from the feeder pre-aligner plate tend to fall with a planar orientation
state instead of a three dimensional random state which makes them align faster in

the direction of the electric ﬁeld. This feeding mechanism has the added beneﬁt of

182

increasing the theoretical critical processing concentration of ﬁbers that can be used

in the ﬁber orientation chamber.

The ﬁber orientation distribution in the ADF preform as well as the orientation
chamber was obtained using digital image analysis techniques. Under the prototype
ADF processing conditions, for all the E-glass ﬁber lengths nearly 70 % of the ﬁbers
of the ﬁbers were aligned i 20° and nearly 80 - 90 % within :1: 30°, indicating a high
degree of ﬁner alignments using electric ﬁelds. Further, improvements in alignments
can be made based on the understanding gained on the alignment behavior of ﬁbers

in air.

Polymer powder coating of the ﬁbers and controlling the matrix volume fraction in
the ﬁnal composite is one of the biggest advantages of the ADF process. The
combination of powder coating with electric ﬁeld alignment makes the process a
potentially high speed-low cost process to make aligned discontinuous ﬁber
composites. This step eliminates the additional step of resin transfer. The
technology of powder coating is well established in the group and extensively

patented and hence can be easily incorporated in the ADF process developed.

The properties of the discontinuous ﬁber composites improve with an increase in
ﬁber alignment and an increase in ﬁber length. Improvements in modulus of the

ADF composites ranged from 70 to 100 %, when compared to equivalent composites

10.

11.

183

that were manufactured without the electric ﬁeld. Similar improvements in strength
values have also been noted. The improvement in properties due to a combination
of alignment and increasing the length from 0.125 to 1 inch increases the property by

nearly 300 %.

Discontinuous ﬁber composites with known ﬁber orientation distribution and ﬁber
aspect ratio had been fabricated using the ADF process. This provided a unique
opportunity to apply fundamental micromechanics analysis to compare theoretical
predictions with experimental results. Also, it provided the basis to verify one of the
primary objective of the ADF process development - the properties of ADF
composites should approach the properties of continuous ﬁber composites as the
length of the ﬁbers increase and as the ﬁbers are aligned in the direction of the stress.
This microstructure-property relationships also integrates processing and
performance in the ADF process because the ﬁber aspect ratio is a critical parameter

in the ﬁber alignment process itself.

Analysis from microstructure-property relationships indicated that in the chopped
ﬁber-thermoplastic systems, the effective reinforcement aspect ratio is a function of
the bundle size and ﬁber-matrix interaction which includes resin impregnation into
the bundle and the sizing compatibility. Further, it may be concluded that the
effective reinforcement aspect ratio approaches the ﬁlament aspect ratio only when

the ﬁber-matrix interaction is optimized and the effective reinforcement aspect ratio

184

approaches the bundle aspect ratio when there is no resin impregnation in to the ﬁber

bundle.

12. Fundamental ﬁber reinforcement theories were used to predict the properties of the
ADF composites that were fabricated in the process with a known ﬁber orientation
distribution and aspect ratio. The Halpin-Tsai model was found to predict the
properties reasonably well, while the model based on Eshelby’s approach over
predicts the modulus. The shear-lag approach could not be directly applied because
of the aggregate nature of the ﬁbers. In all the models the predictions were closer
when the reinforcement aspect ratio was taken to be the bundle aspect ratio rather

than ﬁlament aspect ratio.

Future Research

The work presented in this dissertation establishes the concept and operating
principles of a processing technique for manufacturing aligned discontinuous ﬁber
composites. It is envisaged that this kind of approach will eventually lead to high speed,
highly automated, solvent free, manufacture of microstructure controlled material forms,
where material performance based on the microstructure is taken care right at the
manufacturing step. High speeds of fabrication are possible and amenable to CAD/CAM
robotic manufacturing techniques. This will especially prove to be very effective in the case
of a complex lay-up sequence often needed in a large part. Based on the Computer-Aided

Design, one can program the ﬁeld directions in the orientation chamber and the ﬁber and

185

matrix powder feed rate to lay up powder impregnated ﬁbers in the desired location with the

desired ﬁber orientation with the orientation chamber being moved by a robotic arm.

In this project the focus was on controlling the ﬁber orientation for reinforcement,
but the concept can be exploited for a variety of other ﬁmctions properties like electrical
conductivity and the thermal properties by a suitable selection of the ﬁbers and the matrix.
A polymer composite ADF material form with controlled electrical conductivity may ﬁnd
applications in electro-magnetic shielding. A polymer composite ADF material form with
preferred ﬁber orientation may also become very critical in bio-materials especially in the
area of body implants. This ADF material form also has great applications in the automotive,

sports and recreation industries where light weight structural materials are needed.

Some remaining speciﬁc issues that need to be investigated for this processing approach are:

0 A comprehensive model needs to be developed for characterizing the
electrodynamics of ﬁber motions incorporating inertial effects for the ﬁber
dimensions that are of interest to composites manufacture.

- The limitations on the processing concentrations of ﬁbers need to be established
based on the E-ﬁeld distortion effects due to ﬁber-ﬁber interactions in the orientation
chamber.

- Investigate if the processing rate can be increased by applying a vacuum, through the
orientation chamber. If such a measure is adopted, how will the ﬁber alignment

dynamics change?

 

186

Investigate the effect of multiple orientation chambers in series and study the effect
of ﬁber mis-alignment if any, as the ADF preform builds up in thickness.
Investigate, hybrid ﬁber systems where a mixture of ﬁbers of different dielectric
and/or geometrical features are fed simultaneously through the orientation chamber
to get different ﬁber orientation distributions in different directions.

Engineer different type of electrodes to create ﬁeld conﬁgurations to manufacture
unique material forms and potentially 3D orientation structures.

Study the stamping or thennoforrning behavior of ADF composites and characterize
stampability of these materials as a function of ﬁber length, ﬁber orientation and

polymer rheology.