1...... ll
.......1.......... .1 .c. a
.13? .

. .1 1.1.11... .11....
1.

1..
a 11
:92“):
1
. .1,
ll l 1 1’12Ih:
:1. “1!: I 54.1“.
1 .1 1
1...... .32.. 12.1"»...
:1 .811: :2. .S. A
95.05 2V5:
211:... fi“. s
1. a. uh 1

1. 531-1

1

.1 .
.5..st :v
11.1.. .4 {RS

“flu... {airwavn

. :1
.1. 5.1-1.5:
. :1. 1:5 .
113‘:

.33 303 1..
fly... :1

.
1.1%».
11-53,

shun-1.5..
. v. 5 |.. 311‘ at
ih’flb‘mnsx 9:1. :1. 1

\ c. .
11211.10 1.

3..
)1 .5939 zili)‘ 1
|\.. 4.513.125...
:33303...‘ t. 1 1
1 1.31.3121: 151-?! .
Gigi .{saxgziuv
5|.5-l‘i:l
.3: $1.11...- . 1..
alsoi‘aianH

.11... .311‘ u
n \ 11‘ i.
I

31-13.
.{11-1 1‘. .1

- 15.1.11:
.1233.

 

4....
\a....« .1!

.135.) . o, . . 11.
111,111.11, 111.11%”? 11.111111111111113?

. Li: I

H15. ‘355

A r -

I
erw- \

This is to certify that the

thesis entitled

THE ALIGNED DISCONTINUOUS FIBER PROCESS: EXTENSION
INTO MANUFACTURING

presented by

Scott Andrew Seymour

has been accepted towards fulfillment
of the requirements for

 

 

PhD degfimin Chemical Engineering

 

 

Major professor

Date 8/25/00

 

0.7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

LIBRARY
Mlchlgan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/Dateouopes-ms

THE ALIGNED DISCONTINUOUS FIBER PROCESS: EXTENSION INTO
MANUFACTURING

By

Scott Andrew Seymour

A THESIS

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

MASTERS OF SCIENCE
Department of Chemical Engineering

2000

ABSTRACT

THE ALIGNED DISCONTINUOUS FIBER PROCESS: EXTENSION INTO
MANUFACTURING

By

Scott Andrew Seymour

The Aligned Discontinuous Fiber (ADF) Process, which was previously
developed at Michigan State University, has been shown to work with several industrial
materials. Chopped glass and carbon fibers of lengths ranging from 0.25" to 1" have
been aligned in electric fields and glass fiber-nylon]2/polypropylenefPP), polyethlyene
terephthalate(PET), and nylon ~12 powder coated ADF preforms have been
manufactured using this process. Carbon fiber/nylon — 12 preforms have also been
manufactured. The parameters that affect the orientation of the fibers in electric fields
are investigated viz. fiber aspect ratio, dielectric nature of the fiber, and the Reynolds
number of the free falling fibers. High speed motion analysis is used to record the events,
and the orientations of the fibers under each condition is measured with image analysis
techniques to get an orientation distribution» function. Significant improvements in
performance, up to 100 % in stiffness and up to 85 % in strength values have been
obtained due to fiber alignment and choice of proper sizing. 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.

ACKNOWLEDGEMENTS

This would not have been possible without the help of several people. I would
like to thank Dr. Drzal for advising me and giving insight on this project. The Michigan
State University Cyclotron were invaluable in the trouble-shooting of the ADF apparatus,
and provided knowledge concerning electric fields. I would also like to thank all of the
CMSC Staff and Chemical Engineering Office for all of their help that they have
provided me with throughout my stay at MSU.

Finally, I would like to thank my fellow CMSC graduate students, Arvind
Krishnaswamy, Shawn Corbin, Tammy Cummings, and the rest for making the MSU

experience fun and not always serious. You all owe me an ice cream.

iii

List of Tables ................................................................................ vi
List of Figures ............................................................................... vii
Chapter 1
l . 1 Introduction ................................................................ 1
1 .2 Composites ................................................................. 2
1.3 Particle Reinforced Composites ......................................... 3
1.4 Fibrous Composites ....................................................... 4
1.4.1 Continuous Fiber Composites .................................. 5
1.4.2 Discontinuous Fiber Composites .............................. 7
1.4.3 Discontinuous Fiber Alignment .............................. 10
1.4.4 Flow Induced Alignment Processes ......................... 10
1.4.5 Miscellaneous Fiber Alignment Techniques ............... 12
1.4.5.1 Electric Field Induced Alignment of
Fibers in External Fields ............................. 13
1.5 Processing of Composites ............................................. 15
1.5.1 Compression Molding ........................................ 15
1.5.2 Thermoforming ................................................ 17
Research Motivation ................................................... l7
. Thesis Objectives and Outline ........................................ 18
Chapter 2
2.1 Materials Introduction ................................................. 20
2.2 Matrices .................................................................. 20
2.2.1 Nylon — 12 ...................................................... 22
2.2.2 Polypropylene ................................................. 22
2.2.3 Polyester Terephthalate (PET) .............................. 26
2.3 Glass Fibers ............................................................. 26
2.4 Carbon Fibers ........................................................... 29
Chapter 3
3.1 Experimental Equipment .............................................. 3 1
3.2 Aligned Discontinuous Fiber (ADF) Process Equipment ......... 31
3.2.1 ADF Transformer ............................................. 32
3.2.2 Vibratory Feeder .............................................. 33
3.2.3 Alignment Chamber .......................................... 34
3.2.4 Miscellaneous Equipment .................................... 37
3.3 Electrical Wiring of ADF Process. ................................. 39
3.4 ADF Preform Processing .............................................. 40

TABLE OF CONTENTS

iv

3.4.1 ADF Preform Consolidation ............................... 42

3.4.2 ADF Composite Sample Preparation ..................... 45
3.4.3 Mechanical Testing of ADF Composite Samples ....... 48
3.4.4 Void and Fiber Fraction Analysis of ADP
Composites ................................................... 49
3.4.5 Manufacturing of the Radii of Curvature ADF
Composites ................................................... 51
Chapter 4
4.1 Results and Discussion .............................................. 54
4.2 Glass Fiber Systems ................................................. 55
4.3 Carbon Fiber Systems ............................................... 68
4.4 Measurement of Fiber Orientation ................................. 75
4.5 Halpin-Tsai/Micromechanics ....................................... 86
4.6 Drapeability Results ................................................. 91
Chapter 5
5. 1 Introduction ........................................................... 96
5.2 Process Description .................................................. 97
5.3 Process Economic Model ........................................... 99
5.4 Current System Evaluation ........................................ 101
5.5 Scale-up Evaluation ................................................ 105
Chapter 6
6.1 Conclusions .......................................................... 1 1 1
6.2 Future Work .......................................................... 113
List of References ..................................................................... 116

LIST OF TABLES

1.3.1 Typical Properties of Commercially Available Reinforcing Media ............. 4
1.4.2.1 Sample Critical Fiber Lengths for Glass and Carbon Fibers ..................... 9
2.2 Properties of Polymer Matrices ..................................................... 21

3.4.2.1 Dimensions of a Type I Specimen .................................................. 45
3.4.4.1 Void Fraction of Early ADF Composite Samples ................................ 50
4.1.1 Physical Properties of F ibers/Matrices Used ...................................... 55
4.2.] Percentage Improvement in Mechanical Properties due to Fiber Sizing ...... 57
4.2.2 Glass Fiber System Results. . . . . . . . . ............................................. 62
4.3.1 Results for Carbon Fiber ADF Composites ....................................... 73
4.4.1 Results of Alignment for ADF Systems ........................................... 77
4.4.2 fp Values for the ADF Composites ................................................. 85

4.5.1 Fiber Bundle Length and Aspect Ratio ............................................ 87

4.5.2 Halpin-Tsai Model Predictions for the Glass F iber/Nylon System ............ 88
4.5.3 Halpin-Tsai Model Predictions for the Glass F iber/PET System ............... 88
4.5.4 Halpin-Tsai Model Predictions for the Glass F iber/PP System ................. 88
4.5.5 Halpin-Tsai Model Predictions for the Carbon Fiber/Nylon System .......... 89
4.6.1 Radius of Curvature Results ......................................................... 93

5.4.1 Laboratory Scale Economic Analysis ............................................. 103
5.4.2 Effect of Labor Cost on Lab Scale ADF Process ................................ 105
5.5.1 Effect of Labor Cost on Industrial Scale ADF Process ......................... 107
5.5.2 Economic Analysis of Industrial Scale Up ....................................... 109

vi

LIST OF FIGURES

1.3.1 Heirarchy of Composites ............................................................. 3
1.4.1.1 Types of Continuous Fiber Composites ............................................ 6
1.4.2.1 Types of Discontinuous Fiber Composites ........................................ 8
1.5.1.1 Temperature and Pressure Profile of Compression Molding .................. 16
2.2.1 SEM of Nylon-12 .................................................................... 23
2.2.2 SEM of Polyproplyene .............................................................. 23
2.2.3 SEM of Polyethylene Terephthalate ............................................... 24
2.2.3.1 Particle Analysis of Ground PET .................................................. 25
2.3.1 Manufacturing of Glass Fiber ...................................................... 27
2.4.1 Schematic of Carbon Fiber Manufacturing Process ............................. 30
3.2.1 ADF Laboratory Process Unit ...................................................... 31
3.2.1.1 ADF Transformer in Housing ...................................................... 32
3.2.2.1 Vibratory Feeder Diagram ......................................................... 34
3.2.3.1 Effect of Non-Neutralizing Electrode ............................................. 36
3.2.3.2 Addition of a Neutralizing Electrode ............................................. 36
3.2.4.1 Linear Slide Motor and Speed Regulator ........................................ 38
3.2.4.2 Infrared Heater ....................................................................... 39
3.3.1 Electrical Wiring Diagram of the ADF Process ................................. 40
3.4.1 Tetrahedron Press ................................................................... 42
3.4.1.1 Vacuum Bag Assembly ............................................................ 43
3.4.2.1 SEM Micrographs of the Sample Specimens .................................... 47
3.4.5.1 Radii of Curvature Molds .......................................................... 51
3.4.5.2 Design of Radii of Curvature Molds ............................................. 52
3.4.5.3 Curved Specimen Compression Testing ......................................... 53
4.2.1 Effect of Fiber Sizing on Tensile Modulus ...................................... 58
4.2.2 Effect of Fiber Sizing on Tensile Strength ....................................... 58
4.2.3 Optical Micrograph of PP and Nylon Sized Fibers ............................. 61
4.2.4 Tensile Modulus Results from the PP/Glass System ........................... 63
4.2.5 Tensile Strength Results from the PP/Glass System ............................ 63
4.2.6 Tensile Modulus Results from the PET/Glass System ......................... 64
4.2.7 Tensile Strength Results from the PET/Glass System ......................... 64
4.2.8 Glass Fiber Systems Tensile Modulus Comparison ............................ 67
4.2.9 Glass Fiber Systems Tensile Strength Comparison ............................ 68
4.3.1 Cross Sections of Glass & Carbon Fibers ....................................... 71
4.3.2 Tensile Strength of Nylon-l2/Carbon Fiber System ........................... 71
4.3.3 Tensile Modulus of Nylon-12/Carbon Fiber System ........................... 72
4.3.4 Comparison of Tensile Strength for Glass and Carbon Fiber ................. 72
4.3.5 Comparison of Tensile Modulus for Glass and Carbon Fiber ................ 73
4.4.1 FOD Image Collection ............................................................. 77
4.4.2 FOD for 0.25” Glass Fibers ....................................................... 79
4.4.3 F OD for 0.50” Glass Fibers ....................................................... 80
4.4.4 FOD for 1” Glass Fibers ........................................................... 81
4.4.5 FOD for 0.25” Carbon Fibers ..................................................... 82
4.4.6 FOD for 0.50” Carbon Fibers ..................................................... 83

vii

4.4.7
4.5.1
4.6.1
4.6.2

5.3.1
5.4.1
5.4.2
5.5.1
5.5.2

FOD for 1” Carbon Fibers ......................................................... 84
Experimental vs. Model Predictions for Nylon/Glass Fiber System. . . . .....90
Compression Testing Apparatus .................................................. 93
Radius of Curvature Results ...................................................... 95
Equipment Sizing Calculation ................................................... 100
Cost Distribution for Lab Scale ADF Process ................................. 102
Laboratory Scale ADF Schematic .............................................. 104
Industrial Scale ADF Schematic ................................................ 108
Cost Distribution for Industrial Scale ADF Process .......................... 1 10

viii

1.1 Introduction

A novel process for producing composite materials was developed at Michigan
State University. The Aligned Discontinuous Fiber (ADF) process was created to fill a
void in composite material selection, one between aligned continuous composites and
random discontinuous composites. Both will be described in further detail in this
chapter.

The ADF process consists of a fiber feeder, an alignment chamber, and a
collection tray. It is important to note that in most parts of this work the term ‘fiber’
refers to a fiber bundle that contains 200 individual fibers for glass and 3000 fibers for
carbon. Fibers are fed into the alignment chamber from the feeder and a preferential
alignment is imparted. The aligned fiber bundles are then collected on a tray that rests

below the alignment chamber. The focus of this research was fourfold:

' To apply the ADF process to a representative number of composite material
systems.

I To modify the aligned discontinuous fiber process for use with conductive carbon
fibers.

' To manufacture and evaluate the mechanical properties of molded ADF parts

I To perform an economic analysis on the ADF process to determine its viability

for industrial scale-up and implementation as a manufacturing process.

This chapter will give a brief overview of composites in general, elucidate the
theory of discontinuous fiber composites and discuss the factors that affect their

processing and properties. Finally, the research objectives will be discussed.

1.2 Composites

A composite material is defined as a solid, which is composed of two or more
constituents having different physical characteristics and in which each substance retains
its identity while contributing desirable properties to the whole. Examples of naturally
occurring composites include bone and wood, while man-made composites such as
concrete and carbon black in rubber have been in use for centuries. A composite material
consists of two components: a reinforcing phase consisting of fibers or particles, and a
matrix phase consisting of plastics, metals, or ceramics. Most composite materials that
have been developed so far have been fabricated to improve mechanical properties such
as strength, stiffness, toughness, and high-temperature performance. In this work, the
main properties that will be discussed will be the composite’s tensile modulus, tensile
strength, and fiber orientation. The material’s tensile modulus is a measure of its
stiffness, tensile strength is a measure of how much force it requires to pull the material
apart, and the fiber orientation is a measure the degree of fiber alignment. It is important
to note that the main difference between composites and pure materials is that composite
materials are on a whole anisotropic (with the exception of uniform particle reinforced
composites). While composites can be engineered to have exceptional properties, they

will not usually exhibit these properties in all directions. This anisotropy arises because

of the variation of microstructures that is inherent in composite materials.] Today these
microstructures can be exploited to design high performance composites that save weight
and space over conventional materials, while providing superior mechanical properties.

1.3 Particle Reinforced Composites

Composites can be divided into two main groups: fiber reinforced and particulate

 

 

 

 

 

 

COMPOSITES
l
PARTICLE COMPOSITES F IBROUS COMPOSITES
l
l l
CONTINUOUS DISCONTINUOUS
WOVEN/ALIGNED RANDOM ALIGNED RANDOM

 

Figure 1.3.1 Hierarchy of Composites

reinforced. Figure 1.3.1 diagrams the hierarchy of composite materials. A particulate is
defined as one that is non-fibrous and generally has no long dimension, with the
exception of platelets. The dimensions of the reinforcement determines its ability to
contribute its properties to the composite, i.e. a round particle would add properties in all

directions, where a fiber would add properties along the fiber direction. Particles are

 

1 Piggott, M., “Load-Bearing Fibre Composites”, Pergamon Pres, 1987

3

 

effective in enhancing the stiffness of composites but do not offer the potential for much
strengthening. For example, hard particles placed in a brittle matrix reduce the strength
due to stress concentrations in the adjacent matrix material. Particles can also be used to
improve the fracture resistance of brittle materials. Adding rubber-like particles to a
brittle matrix improves the fracture resistance by initiating and arresting cracks.2
Particles are also added as fillers to materials, which serve to improve various properties
while reducing cost and without sacrificing the other desirable properties. Table 1.3.1

lists several reinforcing media that are widely used in industry.

Table 1.3.1 Typical Properties of Commercial Reinforcing Media

 

1.4 Fibrous Composites
Fiber-reinforced materials offer a combination of strength and modulus that are

either comparable or better than traditional materials. These materials also exhibit lower

 

2
Agarwal, 8.0.. and Broutman. L.J.. "Analysis and Performance of Fiber Composites", John Wiley and Sons, 1980.

specific gravities, higher strength-weight ratios, and higher modulus-weight ratios. For
these reasons, fiber-reinforced materials have emerged as a fast growing class of
materials used in vastly increasing areas. Fibrous composites can be subdivided into
several groups, two of which will be discussed in detail in this thesis - continuous and
discontinuous fiber composites. Of these two groups, they can both be further subdivided
into two groups, aligned and randomly oriented composites. (Refer to Figure 1.3. 1)

Theoretically, continuous fibers can transmit a load or stress from the application
site to the reaction site via a continuous load path.3 However, this is rarely observed in
practice for two reasons: 1) manufacturing variables make it impossible to produce fibers
with perfect surfaces that will fail at their ultimate tensile strength, and 2) in the ideal
continuous fiber structure the fibers will be stress-free or equally stressed, but this
situation rarely happens in reality. Instead, during processing some fibers will be highly
stressed while others will be unstressed. As the finished composite is loaded to its
ultimate strength, the pre-stressed fibers will support more of the load than the fibers with
the lower built-in stresses. The pre-stressed fibers will fail before the unstressed fibers
and consequentially reduce the ultimate tensile strength of the finished composite. The
role of the matrix in fiber-reinforced composites is not to be a load-can'ying constituent,
but to simply bind the fibers together and protect them.

Fibrous composites, along with some particulate composites (whose particles
contain a long axis) can be described as either random or aligned, depending on their
fiber/particle orientation. There are three types of fiber orientation: one-dimensional,
planar or two-dimensional, and three-dimensional reinforcement. This introduction and

thesis will concentrate on one-dimensional reinforcement primarily. Although aligned

fiber composites exhibit higher mechanical properties than randomly oriented
composites, aligned composites are not necessarily the best choice for every application.
Aligned fiber composites are generally reserved for applications which require high
mechanical properties in a single direction. Whereas randomly oriented composites are

used where a low cost and uniform properties in all directions are required.

1.4.1 Continuous Fiber Composites

Continuous fiber composites are widely used in the aerospace, automotive, and
commercial sectors. Depending upon the application, the fibers can be aligned, random,
2-dimensionally woven, knitted, braided, or woven in a 3-D pattern. In this work we will

deal exclusively with random and aligned.

 

 

 

1 J ,11111/
/

 

 

 

 

 

 

 

/

L 7 / / / / / /
/ J
[ /
Unidirectional Laminate 2-D Isotropic Laminate

 

 

Random Roving

Figure 1.4.1.1 Types of Continuous Fiber Composites

 

3 Piggott, M., “Load-Bearing Fibre Composites", Pergamon Press, 1987

Randomly oriented continuous fiber composites are usually made from
continuous fiber strands called roving or twisted yarn. These fibers are then generally
laid into mats and then molded into whatever shape is required. Applications for these
types of composites can be found in some papers and cardboards, along with many types
of packaging materials. One advantage random continuous fiber composites have over
random discontinuous fiber composites is that they do not tear as easily because of the
greatly increased length of the continuous fiber.

For a one-dimensional reinforced fibrous composite the greatest strength and
modulus are obtained along the fiber direction, with greatly reduced properties transverse
to the fiber direction. To counteract this weakness, continuous fiber composites
manufacturers have borrowed technology from the textile industry to create woven,
knitted, or braided two-dimensionally reinforced preforms. These preforms can then be
impregnated with matrix to form the final composite. Manufacturing of these continuous
fiber composites is more difficult and expensive than discontinuous or particulate
composites, but in certain applications the mechanical property requirements demand that
continuous fiber materials be employed.

Calculating mechanical properties for aligned continuous fiber composites is quite
easily accomplished by using the Rule of Mixtures:

E, = E] V; + E,,, V,,,
Where E is the tensile modulus, V is the volume fraction and the subscripts c. f. and m

stand for composite, fiber, and matrix.

1.4.2 Discontinuous Fiber Composites

Discontinuous fiber composites can be referred to by several different names,
depending upon how they are used: short fiber when injection molded, discontinuous
fiber when employed in reinforced metal matrix composites, chopped fibers when the
fibers are secured in bundles, and long chopped fibers when the fiber bundles are longer
than '/2”. The nature of discontinuous fibers serves as a source of discontinuity in the
stress-transfer from the matrix to the fiber, which results in the reduction of composite
mechanical properties. But, on the other hand, discontinuous fibers offer tremendous
advantages in ease of processing, cost, and can be employed in many applications where

a more isotropic material is required.

 

 

 

 

 

/ l \ \
/_\ | _ x/
/ - - I |\
/> . \I g —\
Random Layer Aligned Layer

Figure 1.4.2.1 Types of Discontinuous Fiber Composites

When designing a part that will be subjected to a complex load field, it is
important to utilize a material that will have good properties in all directions — something
that cannot be attained with continuous fiber composites. (Although bi-directional
laminates and weaves can approach this) Fiber lengths can range anywhere from a few

millimeters to upwards of 2” to 3”. Because of the variety of fiber lengths and types that

can be utilized discontinuous fibers can easily be engineered and processed to have a
wide range of microstructures tailored to the specific load field. An important parameter
when dealing with discontinuous fibers is the critical length, 1c. When using
discontinuous fibers that are shorter than this critical length, the full strength of the fiber
cannot be realized in the finished composite. The following equation governs the average

critical length:

 

where o, is the fiber tensile strength, d is the fiber diameter, and IC is the shear stress at
the boundary of the fiber and matrix. Some sample examples of critical fiber lengths are
given in table 1.4.2.1. The shortest fiber used in this work was 1/8”, well above the

critical length for the thermoplastic polypropylene matrix listed in the table.

Table 1.4.2.1 Sample Critical Fiber Lengths for Glass and Carbon Fibers

 

 

 

 

 

Matrix
E-Glass Polypropylene 0.07 inches
E-G lass Epoxy 0.017 inches
Carbon (AS-4) Polycarbonate 0.030 inches
Carbon (AS-4) Epoxy 0.017 inches

 

 

 

Whereas with continuous fiber composites a simple expression such as the rule of
mixtures can describe the material properties of an aligned composite, the calculations

become more rigorous when short discontinuous fibers are studied. A more rigorous

 

examination of various micro-mechanical models for predicting tensile strength and

modulus in short discontinuous fiber composites will be presented in Chapter 4.
1.4.3 Discontinuous Fiber Alignment

When discussing alignment of fibers, it is important to develop a parameter to act
as a measure of fiber alignment. In this work, the planar orientation parameter, fp, will be

used to describe the fiber orientation distribution. For planar orientation,f,,, is given as:
fl, = 2JN((D)cos (held) —1
0

where N is the fiber distribution function, and (I) is the angle of the fiber from the
reference axis. A more rigorous derivation of f,, for the purposes of this work will be
shown in Chapter 4, along with the Halpin-Tsai equations, which are used to predict
mechanical properties from a micro-mechanical model.

Discontinuous fibers hold several advantages over continuous fibers, which
include rapid, low-cost processibility by compression or injection molding, or by
extrusion, and the ability to mold complex geometries that could not be fabricated using
continuous fibers. Theoretically, discontinuous fibers can also be produced with a lower
number of surface flaws, which would allow them to approach more closely their
theoretical strength. Therefore, engineers and scientists have been trying to develop
processes, which could align short discontinuous fibers on an industrial scale. Below is a

brief summary of prior attempts to tackle this problem.

10

1.4.4 Flow Induced Alignment Processes

Most of these processes are dependent upon flow induced alignment. One of the
first, the ERDE Glycerine Process was developed by the Explosives Research and
Development Establishment (UK).4 This process uses a dilute dispersion of fibers in
glycerine passed through a tapered slit at a constant throughput onto a filter bed, which is
under vacuum. Fiber alignment is achieved via the velocity gradients that are setup
between the accelerated 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 manner. The glycerine is then evacuated quickly through the filter bed
which is under vacuum without disturbing the fiber orientation.

The second process, known as the Budd-Slurry process has been developed to
manufacture net shape preforms for the automotive industry.5 The composite is made
from the shape preform using a conventional Resin Transfer Molding process. This
process is well used in the paper manufacturing industry to make test sheets of papers.
The main component of this process is a large tank that is filled with an aqueous slurry of
fibers. This slurry is positioned above a filter screen that has a vacuum valve on the
backside of it. The mat can be made in two different ways, either raising the filter screen
up through the slurry, or using that vacuum to evacuate the liquid, leaving a fiber mat
deposited on the screen. Slight fiber alignment can be achieved by using a set of baffles

that the fibers pass through as the liquid is evacuated. It should be noted that this process

 

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

11

was not developed for the manufacture of aligned preform mats, but to make high fiber
volume fraction randomly oriented mats.

Whereas the above techniques incorporate the matrix in a separate step, using
injection molding equipment to achieve fiber orientation removes this step by utilizing
the matrix as the orientation medium. Typically, short fibers are used and are premixed
in a molten polymer (typically around volume fractions between 10% to 30% with fiber
lengths less than 1/8”) and injected at high pressure into a mold. Orientation of the fibers
is usually perpendicular to the flow in the core regions of the mold and parallel to the
flow in the wall or skin regions. The phenomenon results from high shear rates near the
wall and extensional flow in the core regions. Several factors can influence the amount
of fiber orientation achieved, which include the design of the mold, the injection pressure,
the temperature of the molten polymer mixture, and the mold temperature.

The last of the techniques to be described in this work that utilizes flow-induced
alignment is extrusion. As with injection molding, the molten polymer mixture is forced
through a die which the design of has a great impact upon alignment. Die geometries can
be engineered to give control over fiber orientation during extrusion. The bulk of this

work has been accomplished at Monsanto.6

1.4.5 Miscellaneous Fiber Alignment Techniques

An alternative to using flow induced fiber alignment was developed at DuPont.

 

: Soh, S.K.. “Slurry Process for Preform Manufacture", Proc. 10'" Annual ASM/ESD ACCE, Dearborn, 1994.
Goettler, L.A.. Leib, R.I.. and Lambright. A.J., Rubber Chem. Technology. 52. 838, 1979.

12

LDF1m stands for long-discontinuous fiber composites, which are made by stretching the
continuous fiber until it breaks.7 These fibers are attached by an adhesive to a tape-
backing which also expands but as the fiber breaks, it retains the original fiber
orientation. By assembling many plies of these tapes, one can manufacture a prepreg of
the desired final part shape then impregnate it with matrix. Mechanical tests have shown
that these composites confirm the fundamental theory for non-continuous fiber
reinforcement describing critical fiber length necessary to achieve efficient average stress
transfer in a composite. (This critical fiber length will be discussed in a later section)
This process has been shown to be a cost-effective solution for manufacturing complex

shaped parts of aerospace structures.

1.4.5.1 Electric Field Induced Alignment of Fibers in External Fields

Studying behavior of fibers in electric fields is not a new concept. The first
published work detailing the effects electric fields have on aligning fibers was published
by J. O Isard in the British Journal of Applied Physics in 1954. Later, in 1958 a
researcher at Stanford University named Demetriades authored the theory for ellipsoidal
alignment in electrical fields for neutrally buoyant fibers with essentially zero Reynold’s
number.8 Thirteen years later, Monsanto obtained the first patent for aligned particles for
producing composites. The next year, the Berol Corporation produced commercial

equipment for the alignment of wood fibers for the production of particle board."‘10

 

7 Chang, I.. and Pratte. J.F.. “LDF Thermoplastic Composites Technology”. Journal of Thermoplastic Composite
yaten'als, Vol 4, pp 227-252, 1991.

Isard, J.O.. “The Orientation of Fibers in Electric Field“. British Journal of Applied Physics, Vol 6, p 176-179. 1955.
9 Jander, M., “Industrial RTM - new developments in molding and performing technologies”, Proc. 10'“ annual ASM/ESD
ACCE, Dearborn, 1991.

13

However, since that time the limitations for developing a industrially viable process were
identified as the time it takes to orient the fibers in the field and the rate at which the
fibers could be deposited.

Two processes were developed that utilized electric fields to attempt alignment of
short fibers. The first was by Talbot and Logan who desired to aligned wood fibers for
the production of pencil stock. The wood fibers were dispersed in air and drawn down a
tower under the influence of a vacuum and a parallel electric field. The fibers then were
collected on a mesh weir for impregnation and consolidation. Talbot and Logan later
teamed with Morrison and Knudson and extended this process for use with chopped glass
fibers.l "'2 The oriented glass mats were manufactured using a complex arrangement of
electrodes embedded in the bottom of the deposition tower. This process did not achieve
high levels of orientation because they did not take advantage of the orientation state of
the free falling fibers as they came under the influence of the electric field.

A second process for aligning fibers in electric fields was developed by Knoblach
at the Massachusetts Institute of Technology.'3 His process produced uniaxially aligned
mats of discontinuous graphite fibers. The fiber bundles were dispersed in a dielectric
fluid and allowed to settle between a parallel plate electric field. Once the fiber bundles
were collected on the deposition plate, the dielectric fluid had to be drained and filtered
off of the fibers, which creates additional steps and also could leave a residue on the
surface of the fibers. Because of these additional steps, this process was not received as

being industrially viable.

 

1° Tucker. C.. and Advani, S.G., “Processing of Short Fiber Systems” in Flow and Rheology in Polymer Composites
Manufacturing, Ed. S.G. Advani, Elsevier Science, 1994.
Isard, J.O., “The Orientation of Fibers in Electric Field”, British Journal of Applied Physics". Vol 6. p 176-179, 1955

12 Talbot, J.W. and Logan, J.D.. US. Patent No. 4,113,812, 1978.

14

1.5 Processing of Composites

For as many types of composites that exist, there are at least an equal amount of
ways of processing them. Fibrous composites can be processed by such means as
compression molding, thermoforming, injection molding, and reactive injection molding.
Compression molding involves the forming of the composite under heat and pressure,
where thermoforming is the shaping of complex geometries under heat and pressure.
Compression molding and thermoforming will be discussed in more detail below.
Injection molding can be employed for both thermosetting and thermoplastic matrices.
For thermoplastics, the matrix and fiber mixture is heated up to form a liquid solution,
which is then injected under pressure into a mold of which shape the final composite will
have. When using thermosetting matrices, injection molding is usually referred to as
reactive injection molding. Generally, the thermosetting matrix will be injected into a
mold that contains a fiber ‘pre-preg’ (a fiber mat without matrix) along with a cross-
linking agent to promote the reaction of the matrix. The finished composite is then

unloaded from the mold.

1.5.1 Compression Molding

Compression molding is presently the most technologically developed and

versatile way to incorporate either continuous or random chopped fibers into a structural

composite. One advantage of compression molding is that it is more rapid that labor

 

13 Knoblach, G.M., “Using Electric Fields to Control Fiber Orientation During the Manufacturing of Composite Materials",
SAMPE Journal, Vol. 25, No 6, p 9, 1989.

15

intensive hand lay-up techniques. However, if a random composite sheet is desired, some
local orientation of fibers could occur during molding. This phenomenon arises from
regions of high resin flow during the compression of the matrix at elevated temperatures.
Typically, a ‘charge’ of material is loaded into the compression molding cavity and then
the mold is closed and subjected to pressure and temperature profile similar to Figure
1.5.1.1. The finished part is then cooled well below the glass transition temperature, Tg,
of the matrix and removed from the mold. It is important to note that the cooling rate of
the material is an important parameter in the properties of the finished composite. For
example, when compression molding polypropylene based composites, an exceptionally
fast cooling rate is required in order to quench any crystallization of the matrix — if high

tensile strength is desired.

Figure 1.5.1.1 Temperature and Pressure Profile of Compression Molding

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PRESSURE ' ‘
TEMPERATURE
' l
l l
i l
i I
l I
i l
HEATING FLOW KINETICS I COOLING
TIME

16

1.5.2 Thermoforming

Thermoforming is a generic term encompassing many techniques for producing
useful plastic shapes from a flat sheet. The first attempt at thermoforming was
accomplished by the ancient Egyptians, who formed tortoise shell (keratin) in hot oil to
manufacture food containers. Much progress has been made since that time, but the basic
process has remained unchanged. Modern thermoforming began about 50 years ago with
the developments in thermoplastic resin chemistry and the invention of the screw
extruder and roll-fed sheet thermoformer. Thermoforming in the automotive industry is
called ‘stamping’, which is why this topic was included in this research. An economic

process for the forming of complex shaped parts would be a great boon to the industry.

1.6 Research Motivation

The controlled orientation of fibers in composites has been intensely studied at
Michigan State University (MSU) over the past several years. Vyakarnam, et al.,
developed a process at MSU that could control the orientation state of short
discontinuous fibers.'4 This advancement opened the door for a processing technique
that can be used for the large scale manufacturing of aligned discontinuous fiber
composites for use in automotive, infra-structural, and durable goods applications. The
ADF process served to advance the utilization of discontinuous fiber composites and

bridge the property gap between continuous and discontinuous fiber composites.

 

1‘ Vyakarnam. M.N., “Processing Discontinuous Fiber Polymer Composites: Fiber alignment using Electric Fields and
Microstructure-property relationships”, PhD Dissertation, MSU, 1996.

17

In order to increase the industrial acceptance of the ADF process, it was necessary to
expand the number of composite systems that structural property data had been collected
for. Also, it was important to expand the use of the ADF process to carbon fibers, which
are used widely in structural applications already. However, the conductive nature of the
carbon fiber had to be addressed when processing in an electric field -— a modification of

the ADF process had to be investigated.

1.7 Thesis Objectives and Outline

When the ADF process was developed, only one type of material was studied. In
this work, results from three different types of reinforcing media, (polypropylene,
polyethylene terephthalate, and nylon-12) and two different types of fibers (carbon and
glass) were explored. The challenge of incorporating conductive (carbon) fibers was also
undertaken. The ADF chamber had to be slightly modified to allow the use of the new
fibers, details of the chamber modifications are addressed in chapter 3. In order to
demonstrate commercial viability, a detailed economic analysis was performed and
complex shaped ADF composites were manufactured. As will be discussed in chapter 4,
the biggest advantage of using discontinuous fibers is the fact that they can be molded
into complex shapes without micro-void formation.

This thesis is divided into six chapters. This first chapter has provided a background
on composite materials and why this research was conducted. The second chapter will
provide a detailed background on the different materials used in this research. Chapter 3

will delve deeply into the experimental equipment and processes used to process and

18

evaluate the ADF composite samples. Chapter 4 will examine the collected data and
discuss the experimental results. The fifth chapter is the economic analysis report that
was generated for the ADF process. Finally, the sixth chapter will be the conclusions

gleaned from this research and the recommendation for future work.

19

2.1 Materials Introduction

This chapter will introduce all of the materials used in this research, how they are
manufactured and then give a detailed description of all of the pertinent material
properties. Three different polymer matrices were used, along with two different fiber
types. Polyethylene terephthalate (PET), polypropylene (PP), and nylon-12 were used
with carbon and glass discontinuous fibers of various lengths. These materials were each
specifically chosen to represent the wide assortment of commercial matrices and fibers

available in industry.

2.2 Matrices

The role of the matrix in a fiber-reinforced composite is to transfer stresses
between the fibers, to provide a barrier against an adverse environment, and to protect the
surface of the fibers from mechanical abrasion. The overall role of the matrix in tensile
load carrying is quite small. However, the selection of the matrix has a major influence
on the shear properties and compressive strength of the finished composite. The interface
between the fiber and composite will also have an influence on the composite properties.
Much work has been done at Michigan State University investigating the use of fiber
coatings (called “sizings”) that increase the adhesive coupling of matrix to fiber. Without
this adhesion, the matrix and fiber are not ideally bonded and the composite properties

will be starkly depressed.

20

There are two types of polymer matrices: thermosetting and thermoplastic. The
main difference between thermosetting and thermoplastic matrices is that when heat is
applied to a thermoplastic matrix, the material will soften and exhibit liquid-like
behavior, ultimately becoming a liquid at temperatures above the melting point of the
material. Thermosetting matrices react chemically to form cross-linked polymer chains
that do not melt with the addition of heat. If enough heat is applied to a thermoset matrix,
however, the material will burn instead of liquefying. The majority of commercial
composites are made with thermosetting matrices, such as epoxies and polyesters, which
provide superior properties at elevated operating temperatures. In the past ten years,

significant progress has been made in synthesizing new thermoplastic polymers that can

Table 2.2 Properties of Polymer Matrices

 

21

approach the elevated temperatures required for commercial use. That fact, coupled with
the gaining public concern over the recyclability of materials, has led to greater industrial
interest in the manufacturing of thermoplastic fiber reinforced plastics. Table 2.2 below
lists common polymer matrices and their properties. Three thermoplastic matrices were
investigated in this research: PET, polypropylene, and nylon-12. SEM’s of all three

matrices are shown on the following pages, Figures 2.2.1 — 2.2.3.

2.2.1 Nylon - 12

Nylon-12 is similar in structure to Nylon 6,6, but does not exhibit the same
strength or stiffness as Nylon 6,6. Thus, its role in commercial composites is extremely
limited. However, nylon-12 is used extensively in women’s cosmetics. Because nylon-
12 is easy to manufacture in small particle sizes, less than 10 microns, and is non-toxic, it
is used as foundation in make-up. Because of the unique design of the Aligned
Discontinuous Fiber process, the polymer matrix is required to be in powder form. The
nylon-12 was supplied by Elf-Atochem under a trade-name of Orgasol 2000NX. The
particle size distribution was centered around 10 microns. The density of the nylon-12

was 1.1 g/cm3.

2.2.2 Polypropylene

Polypropylene has found many uses in the past several years. Employed in

applications ranging from blow-molded plastic sofi drink bottles to carpet fibers, the

22

 

Figure 2.2.1 SEM of Nylon-12

 

Figure 2.2.2 SEM of Polypropylene

23

 

Figure 2.2.3 SEM of PET

recyclability and processing versatility of polypropylene has made it one of the most
popular consumer product materials. The polypropylene (PP) used in this research was
supplied by Micro Powders, Inc. (MPI) under a trade-name of Propylmatte 31. MP1 sells
the Propylmatte 31 as an additive to waxes for increased luster and appearance. The PP
was supplied as a white powder with an average particle size of < 10 microns. As the
SEM shows, the PP powder is not spherical (like the nylon-12), but cylindrical — an
artifact of the grinding required to reduce the particle size. The density of the PP was

0.88 g/cm3.

24

SNOSOIH

N MICE‘OC 2:0

PNUOUIG‘Q-OQB

X MICFOC 1H6
Figure 2.2.3.1 Particle analysis of ground PET

25

121d :8 (II endures
H'S'HIT GI aldues

 

gang/uoag/aunron LafiOdd/d13!a

- HZZBTUNU STDIIXUJ OOTX -
3931.083!”

:[rnprsau noIaa
:Itnpgsau anoqu

so: or :auu 2.6/81/80 = area

00 ‘ 0
uN/OO ' 0

2.2.3 Polyethylene Terephthalate (PET)

PET, also called thermoplastic polyester, was supplied by Morton Thiokol under a
trade name of Corvel Titanitun. The PET arrived as a grayish powder (with 10% TiOz
(by weight) added) with an average particle size of 150 microns. Because the powder
was too coarse for use with the ADF process, it was necessary to have the powder further
reduced in size by an outside grinding company. Union Process, based in Akron, Ohio,
provided the particle reduction services produce PET particles with an average particle
size of 10 microns. The PET (100 grams) was ground dry for 1 hour to develop a thin
flake, then 100 mL isopropyl alcohol (IPA) and 1 mL Synthrapol KB (added to aid in the
grinding) were added and processed wet for 3 hours. The wet slurry was air dried to
remove the liquids added during grinding. A picture of the particle analysis is shown

above in figure 2.2.3.1.

2.3 Glass Fibers

Glass fibers are unique materials that exhibit the familiar bulk glass properties of
hardness, transparency, resistance to chemical attack, stability, and inertness, as well as
fiber properties of strength, flexibility, lightness of weight, and processability. Glass is
manufactured by fusing silicates with soda or with potash, lime, or other metallic oxides.
The molten material is then quenched to prevent crystallization. Glass fibers are
commercially made by melting together all of the raw materials, then formed into the

glass fiber product. The molten glass is extruded through an orifice that is usually

26

  

Furnace

aware
-fi~ fig. a" 333x;_.‘.f-; >4: ».:
.9 . ,j 3x».- _. ,
n, 31,3; .1 'b-‘l'

' .-.\. r - \\‘.'. 43“):
.. “Essa Molton

  

.1

!

  

2

~ I
7 - ' .
.- [or-‘4‘ - *» ._ . _
“a“; > ‘1‘} P 2 '1',“
aw 4"? 3.15391: Glass "‘3
\n\\ i. . .
- Iii-5&3 “- " ’1 ’s g
y. (,1. ‘ V, : A;
at: 1 ' ' ‘*
tr ‘ ,.

Retiner

  

Forehearth

\\\

m

 

 
 
  
 
 

 

I

'. 3‘"-
I‘I ' U

I
$1573.
. _ $ ‘ _
T’l’éibhj“ v . “.tv. 1

,.

 

 

 

 

 

OO

 

 

Winders
l Forehearth j
H
Water Spra Bushing
N @ Filaments
Applicator

 

 

 

 

 

v\'\Gathering Shoe
Strand

 

 

 

 

 

   

Forming \ Spiral

Winder < - 0°"9‘
Transverse
Strand Guide

Direct Draw Collet

Winder

l
Direct Chopper ————>
Direct Chopped Strand

 

 

Figure 2.3.1 Manufacturing of Glass Fiber
(courtesy of Composites - Engineered Materials Handbook, ASM International, Ohio, 1987)

27

between 0.793 to 3.175 mm in diameter, and then rapidly drawing this glass to a fine
diameter. These fine strands are then coated with a ‘sizing’ agent that serves to increase
the adhesion of the fiber to the intended matrix, and to protect them from abrasion-related
damage to the fiber surface. Finally the strands are gathered together to form the finished
product. A detailed diagram of the glass fiber making process is depicted in Figure 2.3.1
above.

There are three main glass fiber types, E-Glass, S-Glass, and C-Glass. E-glass is
a family of glasses with a calciumaluminoborosilicate composition and a maximum alkali
content of 2.0%. E-glasses are used as a general purpose fiber when strength and high
electrical resistivity are required. S-Glass has a magnesium aluminosilicate composition,
which demonstrates high strength and is therefore used in applications where very high
tensile strength is required. S-glass and 8-2 glass fibers have the same glass composition
but different coatings. C-Glass has a soda-lime-borosilicate composition that is used for
its chemical stability in corrosive environments. Therefore it is often used in composites
that contact or contain acidic materials.

The glass fibers used in this research were supplied by both Vetrotex-Certainteed
and Owens Corning Fiberglass. All fibers were E-glass, with a density of 2.55 g/cm3.
The fibers were in bundle form, with an average fiber count of 200 fibers/bundle. They
were chopped into lengths of 1/8”, 'A”, V2”, and 1”. Four different fiber sizings were
investigated: polyethylene, nylon, polypropylene, and PET. The first two sized fibers
were investigated by combining them with the nylon-12 to compare the effects of the

sizing on the physical properties of the composite.

28

2.4 Carbon Fibers

Carbon fibers have been around since Thomas Edison invented the light bulb, but
were not manufactured in great quantities until the late 1960’s. At that time it was
discovered that one could carbonize several fibrous materials into continuous fiber
filaments with relatively low density and high Young’s modulus of elasticity. This fiber
could then be used much as glass fiber had been used: to provide a continuous
reinforcement in various resin systems for the fabrication of structural components.

The two main precursors to carbon fibers are polyacrylonitrile (PAN) and pitch.
PAN is the basis for the majority of carbon fibers commercially available today. These
precursors can be thermally rearranged before thermal decomposition, which allows one
to maintain the same filament configuration. This, along with the lack of surface defects,
makes PAN based carbon fibers stronger than carbon fibers based on any other precursor.
Pitch precursors are based on petroleum asphalt, coal tar, and polyvinyl chloride. Pitch
based fibers are relatively low in cost. The most significant drawback of manufacturing
pitch based fibers is the non-uniformity of fiber properties from batch to batch. This can
be remedied somewhat by pre-mixing the pitch into a mesophase pitch.

There are five main process steps to making a carbon fiber: stabilization,
carbonization, graphitization, surface treatment, and application of surface sizing.
Stabilization is carried out at temperatures below 400 0C. The fibers are stressed to
improve the orientation of the carbon chains. This pre-stress also improves the strength
and modulus of the finished fibers. Carbonization is done at intermediate temperatures,

from 800 — 1200 0C. The fibers are pyrolyzed in inert environments to reduce the

29

Pan Process

 

 

 

Pitch Process Thermoset _ .
Carbonizefl Graphitize H

 

 

 

 

 

 

Spool Epoxy Sizing Surface Treatment

Figure 2.4.1 Schematic of Carbon Fiber Manufacturing Process
(courtesy of Composites — Engineered Materials Handbook, ASM International, Ohio. 1987)

impurity levels and increase crystallinity. Graphitization then heats the fibers to
temperatures in excess of 2000 °C in an inert environment. This step also reduces the
level of impurities and further increases the crystallinity. It is important to note that the
higher the temperature, the higher the modulus of the fibers. Processing of the fiber is
now complete, and the only remaining steps are treatment of the surface to increase
adhesion, add surface groups to aid in bonding with the matrix that is to be reinforced,
and protect the fiber surfaces during shipment and handling.

The carbon fibers used in this research are Hercules-Magnamite 3000 strand tow
(3K Tow) PAN based AS-4 fibers. These fibers were not chopped and sized. Advanced
Composites, L.L.C. of Mamoroneck, NY chopped and sized these fibers to lengths of V4”,

V2”, and 1”. The carbon fiber sizings included nylon, polypropylene, and PET.

30

3.1 Experimental Equipment

This chapter will describe each piece of the experimental equipment in detail.
Several related experiments used during this research will also be described, such as the
process of the composite, the cutting of specimens, fiber and void volume calculations,

and all other various sample preparations.

3.2 Aligned Discontinuous Fiber (ADF) Process Equipment

The Aligned Discontinuous Fiber Process (ADF) consists of four main pieces of

equipment: the high voltage transformer, the alignment chamber, the vibratory feeder,

Schematic of Laboratory Scale Continuous ADF Process

Chopped FibeLr Inlet

 

 

Alignment
High Voltage Source

1.3.
‘ - O - O I - -- -OO- -- - .3-
.a‘.
r's,"

Chamber

Electrodes

Infrared Heater

     

:Powder Depositor ADF Composite Sheet
I i 0

   

Deposition Belt

Figure 3.2.1 ADF Laboratory Process Unit

31

and the deposition platform. Several other minor pieces of equipment are also employed,

such as the infrared heater, the multimeters, the variac, and the high voltage probe.

3.2.1 ADF Transformer

The key component to the ADF process is the Hipatronics RM25 transformer,
which is rated at 25 kV (A.C.) at an input voltage of 110 V. There are two ‘coils’ of
wiring inside the transformer, the first is a dense coil in the very center of the transformer
which initiates the stepping down of the current, and the second much larger coil which

surrounds the center coil and amplifies the voltage. The transformer is housed in a

 

32

silicone oil bath to reduce arcing from the transformer to the surrounding air. Arcing
would occur if the voltage within the transformer exceeded the breakdown voltage of the
surrounding medium, in this case it would indeed exceed the breakdown voltage of air,
but not of the silicone oil, which has a larger breakdown voltage. The output is fed to the
two high voltage electrodes, which can be controlled two ways, first a variac is used to
feed a controllable voltage to the transformer (maximum output of 30 KV), and second,
the high voltage can be taken off of the transformer at 100%, 75%, 50%, and 25% of the
output. This can also be used to vary the voltage of the upper and lower electrodes. It is
important to note that this process is not inherently dangerous, the voltage is stepped up

to a maximum of 30 KV, but the current is stepped down to ~15 mA.

3.2.2 Vibratory Feeder

The second piece of equipment in the ADF process is the vibratory feeder. The
vibratory feeder sits atop the alignment chamber and serves two purposes in the ADF
process, first it will feed the material to the alignment chamber at a constant rate, and
second, it induces a pre-alignment to the fibers. The design of feeder chosen consisted of
a modified vibratory F MC Syntron Magnetic feeder. The slight angle of inclination of
the feeder, coupled with the vibratory motion of the chute, propelled the fibers towards
the opening. The vibration also reduced entanglements and clumping of the fibers. It is
important to note that the feeder did not align the fibers in such a way that the electrode
only preserves the initial alignment. On the contrary, the feeder conveyed the fibers into

the alignment chamber perpendicularly to the electric field.

33

It was found by Vyakarnam, et al., that fibers having a random orientation state
before entering the orientation chamber cannot be easily aligned by the electric field over
small distances 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 at the bottom of the orientation chamber, the fibers will

rebound causing misalignment and disturbing the orientation of the neighboring fibers.

Feed Plate cum Pre-aligner

 

Imam

 

 

 

 

 

 

 

 

 

Slot

Serration

 

Magnetic Vibratory Feeder

Figure 3.2.2.1 Vibratory Feeder Diagram

(reprinted with permission from Vyakarnam)

3.2.3 Alignment Chamber

The alignment chamber consists of two sets of planar copper electrodes housed
between two plexi-glass supports. As the fibers fall into the chamber, they come under
the influence of an aligning torque resulting from the electrical polarization. This torque
induces a charge complex on the ends of the fiber, which serves the purpose of aligning
the fiber with respect to the electrodes and keeping the fiber in the desired alignment until
it reaches the deposition plate.

The upper set of electrodes is used to align the fibers, while the bottom set of
electrodes is used to counteract the resulting electric field generated by the bottom edge
of the electrode. Without this bottom set of electrodes, the electric field is disturbed
resulting in fiber misalignment if operating in a continuous mode. These ‘edge effects’
occur at every boundary of an electrode, where the electric field is greatly concentrated,
so much so that the bottom of the electrodes had to be fit with cylindrical caps to help
dissipate the electric field. By analyzing the strength of the electric field in the main
chamber, and at the uncapped edges, when 25 KV is applied to the high voltage
electrode, the electric field intensity would be 500 KV/m (assuming a 5 cm gap width).
At the edge, which has a thickness of ~0.0625 inches, the electrical field strength would
be ~ 15,000 KV/m, well above the breakdown voltage in air. (3000 KV/m) Without these
caps, the electric field would redirect itself to the closest grounding point (the grounded
electrode) and cause arcing between the plates, or between the plates and a metal object

nearby — which is an unsafe operating condition.

35

The minimum electrode height required for fiber alignment was discovered by
Vyakarnam to be dependent upon two parameters: fiber alignment time, and fiber

settling velocity. A simple correlation was determined:

Minimum Electrode Height: Alignment Time x Settling Velocity

 

 

 

 

 

 

 

 

 

 

 

 

 

Efl'eetofLineerMouon
Stationary [—4
_ \
Electrode Fiber: I .—
_/ _——
\/

/K '\ Forward —->
O— \ — _"
HighVoMem : — -l- — \ __

- _ —

— ._ , *1

\\ \_ __

( F

LinearMotion W ‘—
4/ I
“7
I\I l

 

 

 

Figure 3.2.3.1 Effect of Non-Neutralizing Electrode

(reprinted with permission from Vyakarnam)

Based upon Vyakarnam’s calculations, a height of 50 cm was determined adequate to

align fibers varying in lengths from 1/8” to 1.25”. The width of the alignment chamber

36

was designed to be 25 cm. Therefore, a composite part that was 20 cm in width could be
made with the deposition plate that was chosen.
By changing the spacing of the electrodes, one can effectively control the strength

of the electric field when keeping the input voltage constant. On the laboratory unit, the

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

Effect of Linear Motion
Stati
new _°_""’ H
Elm ‘ \
\/ I , _
/ _ '—"’ _
\ \ z _ _
High Voltage (V1) ’ F \ ’ r
0 / _ Forward '—>
_ — \
o I x — _
W / __
u —’ — _
HF: .... .—
High Voltage (V2) \ — \
—- / _
_/ — _

 

 

 

Figure 3.2.3.2 Addition of a Neutralizing Electrode

(reprinted with permission from Vyakarnam)

gap width can vary between 3 - 15 cm, which resulted in electric field strengths of ~800
KV/m to ~150 KV/m. The strength of the electric field was controlled by a variac unit,

which allowed the operator to adjust the amount of electricity flowing into the

37

transformer. The scale on the unit was 0 — 125 % of the incoming current. A high
voltage probe was used to measure the actual electric field strength. A Fluke 80K-40
probe was employed and connected to a Fluke 45 multi-meter to provide continuous
monitoring of the electric field strength. Normal operating parameters for E-glass fibers
is between 400-500 KV/m, and 60-80 KV/m for carbon fibers, depending on fiber length

used.

3.2.4 Miscellaneous Equipment

Once the fibers have been aligned in the alignment chamber, they deposit onto a
movable platform that is covered with a teflon release cloth and connected to a motorized
slide. The linear slide was driven by a DC. motor which was, in turn, connected to a
speed regulator. The largest uncut ADF preform that could be made was 20 x 35 cm.

The total distance that could be traversed by the slide was 30 inches, at speeds that could

    

F, ' V

Figure 3.2.4.1 Linear Slide motor and $3311 Regulator

38

vary between 10 cm/min and 180 cm/min in both a forward and reverse direction.

The platform can be used for both batch (by remaining stationary) and continuous
operation (by moving back-and-forth under the alignment chamber). The aligned fibers
then are coated with the polymer matrix in powder form, and sintered under an infrared
heater. The particles partially melt and solidify bridging between adjacent fibers and then
holding the fibers in place. As a consequence, the part can be handled without losing the
desired alignment. A common kitchen strainer was used to distribute the matrix (in
powder form) over the aligned fibers. The infrared heater consisted of two infrared
heating elements nested into a metal support frame, which the polymer containing fiber
mat passed underneath at a distance of < 1 inch. Because of the variation of melting

points of the three polymers investigated in

 

Figure 3.2.4.2 Infrared Heater

this work, PET, PP, and nylon-12, different sintering conditions were required. The

nylon-12 and PP were the easiest to melt. Care was required in order not to burn the

39

polymer, which would have an adverse effect on the material pr0perties. The PET was
the highest melting point material and the infrared heater had to be placed very close to

the material to start the melting process.

3.3 Electrical Wiring and Diagram of the ADF Process

The ADF Alignment chamber could be described as one giant capacitor, which is

never allowed to discharge. One electrode is fully grounded, where the other electrode is

 

 

 

VohgeDivida’

 

Figure 3.3.1 Electrical Wiring Diagram of the ADF Process

(reprinted with permission from Vyakarnam)

4O

attached to the high voltage transformer. In the electrical diagram shown above in Figure
3.3.1, the electricity flow, once turned on and fully charged, theoretically is zero, since
there is no direct flow from the high voltage plate to the ground. But in actuality, there is
a small amount of current loss due to corona discharge from the entire surface of the high
voltage plate. The total electrical duty is quite small, making this process quite attractive
to industrial scale-up, which will be discussed in greater detail in Chapter 5.

3.4 ADF Preform Processing

The laboratory scale ADF Process Unit was not capable of manufacturing a full
composite preform at once. Because the matrix was not deposited at the same time as the
fibers, only a finite amount of fibers were able to be deposited before the bottom layer of
aligned fibers would not be incorporated into the perform post sintering. Also, when a
thick layer of fibers were deposited at once, the matrix powder was not able to penetrate
the upper layers and fully secure the fiber alignment of the bottom most fibers. To
alleviate this problem, several ADF ‘layers’ would be deposited and sintered separately.
A finished composite would then be made by assembling a pre-determined amount of
layers and molding them into the final composite shape. The molds used in this research
were 5 inch x 7 inch rectangular “picture frame” molds that were 1 mm (0.03937 in) and
2 mm (0.07874 in) in thickness. To aid in the manufacturing of these ‘layers’, a simple
spreadsheet program was written, called the ADF Composite T 001. One would input the
desired fiber volume fraction, the matrix and fiber densities, the mold volume or
dimensions, and the number of layers desired. The spreadsheet program would then

calculate the amount of fiber and matrix required to make each layer, along with the total

41

weight of the finished composite. After sintering, the preform would then be trimmed on
a paper cutter to roughly 5 x 7 inches. An additional 20 % of material was added to
compensate for the amount of fiber and matrix that was deposited outside of the 5 x 7

inch preform area. An example of the ADF Composite Tool is shown below: (the

 

 

 

 

ADF Composite Tool
Length by Width by Thickness
Finished Composite Size Sinches 7inches 0.07874linches
Composite Volume 2.75 59cubic inches
# of Layers Desired 8
Density of Matrix 1.1g/cm3 PP
Density of Fiber 2.55 g/cm3 Glass
Fiber Volume Fraction 0.5
Weight of Matrix 1.5 l 5745 grams
Weight of Fiber 3.5 l 3773 grams
Weight of Layer 5.0295 1 8 grams
Total Weight of Composite 40.23614 grams

 

 

 

 

example shown below was for a polypropylene and glass fiber composite) All

composites manufactured in this research were calculated to be 50% (by volume) fiber.

3.4.1 ADF Preform Consolidation

Once all of the preform layers were manufactured, subsequent processing was
required to form the finished composite. A Tetrahedron Compression Molding Press was
employed in this research to accomplish this.

Using heat and pressure, compression molding heats the polymer matrix until it is

in a liquid state, then the pressure is applied, which forces the molten polymer into the

42

 

Figure 3.4.1 Tetrahedron Press

5” x 7” mold. The mold is then cooled while still under pressure.

All of this molding was done while a vacuum was drawn on the mold assembly. The top
and bottom platens were lined with Teflon release cloth. The mold rested on the bottom
platen and the preforms were stacked inside of the mold. A Teflon release cloth
‘channel’ was made from the bottom of the vacuum hose port that traveled under the top
platen and rested against the mold. This channel allowed for air to be pumped out of the
mold cavity once the pressure was applied to the platens. Below (in Figure 3.4.1.1) is a

diagram of the mold assembly:

43

 

 

 

 

 

 

 

    

3 Vacuum Hose Port
Mold __ E

E l
Preforms — i

T Down View -— (without
acuum

Platen
Mold

 

Figure 3.4.1.1 Vacuum Bag Assembly

Because of the pressure gradient emanating from the mold to the vacuum hose port, and
the porous nature of the Teflon release cloth, a finite amount of molten matrix was
‘pulled’ out of the mold. To counteract this, an additional 20% of polymer powder was
added to the preform stacks. This addition also drove out any remaining air pockets
within the mold when pressure was applied, which would have led to voids in the
finished product, which plagued this research until it was incorporated into the

processing.

44

 

3.4.2 ADF Composite Sample Preparation

Once the ADF Composite was manufactured, it needed to be prepared for testing.
The material properties that were of interest were the tensile strength and tensile
modulus. According to ASTM D63 8-96 “Standard Test Method for Tensile Properties of
Plastics” subsection 6.1.3 Reinforced Composites, “the test specimen for reinforced

composites shall conform to the dimensions of the Type I specimen.”

Width of Narrow Section 0.50 inches

 

 

 

 

 

 

Length of Narrow Section 2.25 inches
Width Overall 0.75 inches
Gage Length 2.0 inches

 

Table 3.4.2.1 Dimensions of a Type I Specimen

The 5” x 7” x 1 mm sample was cut into six ‘slices’ along the long axis of the composite,
(which was along the aligned fiber direction) using the diamond tipped F elker saw. The
Felker saw kept the sample cool during the cutting with a water spray, but this exposure
to water did not have any detrimental effects on the final properties of the specimens
because the samples were allowed to dry thoroughly before testing was done. These six

slices then had to be shaped into the “dog bone” structure. A tool called the Tensilkut

45

 

was employed to accomplish this. The Tensilkut was quite similar to a carpenter’s router,
it was mounted upside down on a metal deck, and the abrasive die poked through a hole
in the deck. A prior researcher in the group, Ed Drown, had several steel “molds” made
for use with the Tensilkut. One of them was designed to the Type I specifications, which
was then used for this research. These molds would hold one of the ADF slices and
protect the interior section of the piece, while grinding away the unwanted center
sections. After this mechanical grinding away of the slice to make the narrow section,
the microstructure of the edges was quite disturbed. Various grades of sand paper were
used to smoothen the sidewalls of the dog bone. An added benefit of the sand paper was
that the frictional heat that was generated also helped anneal the damaged matrix. The
sanded dog bone structures were then stored at room temperature for 48 hours prior to
testing.

One important difference between the technique used in this research and that
used prior by Murty Vyakarnam, was that his samples were cut using a high-speed laser
on the MSU campus. Due to lack of availability of the equipment, it was decided to use
the Tensilkut for the processing of the samples in this research. It is important to note
that the microstructure that is created on the sidewalls of the samples is quite different
when these techniques are compared. Figure 3.4.2.] contains six SEM’s of different
sample preparations. The first row of two show the “as cut” by the Tensilkut, these show
extensive surface damage from the mechanical abrasion. The second row of two are

Vyakarnam’s samples that were cut using the MSU High-Speed Laser. These samples

46

 

A 1... .I
13.1... . . .
ate-”3w Rep.

a

J;

t...

 

Figure 3.4.2.1 SEM Micrographs of the Sample Specimens

show considerable annealing of the matrix surface, with small amount of fibrous residue
with sand paper. Some of the surface damage was annealed by the frictional heat, and
there is less fibrous residue. The strength difference between the composites made by
Vyakarnam and those made in this research can be explained by this coupon preparation

difference.

3.4.3 Mechanical Testing of ADF Composite Samples

Testing of the properly conditioned samples was accomplished using an MTS in
the Composite Materials and Structures Center Laboratory. The MTS was employed in
tension mode using the laser extensometer to measure the strain on the sample. The laser
extensometer used reflective tape placed on the gage section of the sample coupon to
reflect the laser signal back into the machine to measure how far the sample had been
‘stretched’ before failure. This tape was applied to the sample using a special ‘mold’ that
held the sample coupon, while tape was applied at specific locations on the gage area of
the coupon. A straining rate of 0.02” per minute was used for sample testing, and the
1000 lb load cell was used for this testing. Plots of the stress versus strain were made by
the MTS computer. A representative sample of this plot will be presented in Chapter 5.
The tensile strength was computed by dividing the load at failure by the cross sectional

area of the sample. Tensile modulus was computed by the MTS computer by measuring

48

it? 51

13382

W.

5.5L.

1161

C011

H

:01.

the slope of the stress/strain curve generated by the machine. The slope up until failure
was used for this calculation.
For mechanical property comparison purposes, the testing technique used in this

research was copied directly from the prior researcher.

3.4.4 Void and Fiber Fraction Analysis of ADF Composites

Two of the critical parameters when manufacturing composites are void fraction
and fiber volume fraction. Anything over 3-4% void in a composite will cause poor
mechanical performance, as the stress will not be uniformly transferred across the
composite. In the case of fiber volume fraction, it is important to know the value of this
parameter if any accurate prediction of the mechanical properties is required. All
composites used in this research were targeted to be ~50 % +/- 2% fiber by volume.

The methods used to compute these values were similar. In the case of fiber
volume fraction, the sample was weighed and placed in a furnace to burn the matrix off.
The glass fiber composites were burned off at a temperature of 600 0C for 4 hours. The
fibers were weighed and the void fraction was calculated by determining how much of
the sample (by volume) was matrix. Of course, the sizing of the fiber bundles was also
burned off, but for the purposes of this research, the weight/volume of the sizing was
assumed to be negligible. The carbon based composites had to be handled with more
care than the glass — because the carbon fibers would burn off with the matrix if the
fumace temperature was too high. Since glass fibers are thermally stable in an oxygen

containing atmosphere to temperatures in excess of 1500 DC, this was not a problem. The

49

oxidation of the carbon fiber could have been avoided by using an inert atmosphere, such
as N2, but matrix combustion rate would have been affected as well. A furnace
temperature of 400 0C was then used for 6 hours to completely remove the matrix, while
keeping the carbon fibers intact.

The void fraction was computed similarly to the fiber volume fraction, since the
fiber volume fraction was needed to compute the void fraction. The void fraction
samples were carefully cut to be 0.50” by 0.50” squares with sanded edges to make the
square as uniform as possible. The squares were then measured and weighed. The
volume of the square was then calculated. An additional volume measurement was
conducted in which the sample was immersed in a 100 mL graduated cylinder. This
measurement was not as accurate as the volume calculated by the square’s geometric
measurements, but it served as a ‘sanity’ check on the volume. The samples were then
loaded into the oven and the matrix was baked off just as with the fiber volume fraction.
Once the fiber volume fraction was known, the computed composite volume was

measured and compared to the composite volume by the geometric measurements. The

Void Fraction Analysis

 

 

 

 

Aligned 0.25” 6%
Random 0.25” 11%
Aligned 0.50” 10%
Random 0.50” 9%

 

 

 

 

Table 3.4.4.1 Void Fraction of Early ADF Composite Samples

50

difference in these two measurements was the void volume fraction of the composite.
Voids were a large problem in the beginning of this research. The early ADF composites
that were made contained an unacceptable amount of voids.

It was determined that the voids were being formed during the compression molding step,
specifically when the matrix was molten and pressure applied. Under a microscope, it
was determined that the void fraction was weighted more heavily on the top surface of
the composite. It was theorized that since the Teflon release cloth was porous, the molten
matrix would be pushed out of the bottom of the mold when the pressure was applied. To
counteract this effect, an additional 20 % (by weight) of matrix was added to the top of
the preform stack in the mold prior to vacuum bag assembly. This additional matrix
would compensate for the loss through the Teflon release cloth, and also mechanically

force the voids out of the composite during consolidation.
3.4.5 Manufacturing of the Radii of Curvature ADF Composites

The third goal of this research was to manufacture ADF composites to test the
‘drapeability’ of the ADF composites. These curved composite were made using special
molds that were manufactured by the Michigan State University Metal Shop. Three
different radii of curvature molds were constructed, 0. 125”, 0.25”, and 0.50”. A diagram

of the molds is shown below:

51

 

Side View Back View

‘7

     

 

 

 

Figure 3.4.5.1 Radii of Curvature Molds

The ADF preforms would be hand placed on the bottom piece of the mold, centered with
respect to the radius of curvature. The top piece would then be placed on top of the
preforms and inserted in the vacuum bag. Careful sample preparation was required to
minimize shifting of the mold pieces during vacuum bag assembly and positioning of the
sample inside of the tetrahedral press. A Teflon release cloth sheet was placed between
the metal parts and ADF preforms.

The radii of curvature of the molds were calculated by using 25% of the
circumference of the appropriately sized circle. Additional length was added to each half

of the composite to make the sides 7”. This method is illustrated below:

   

Circle of Radius ‘X'

Figure 3.4.5.2 Design of Radii of Curvature Molds

52

For design consideration, a composite thickness of 1 mm was used. In order to keep the
top piece of the mold from compacting the curved portion of the composite, metal ‘stops’
were placed under the top piece of the mold. These stops ensured that the very tip of the
curved portion was exactly 1 mm in thickness. Besides the mold preparation difference,
the processing of the radii of curvature composites was identical to that of the flat
composites. Sample coupon preparation was slightly different, since these samples were
simply cut with the F elker saw into 1” strips, as there is no ASTM method for testing of
these type of coupons. Testing of the coupons was conducted on the MTS machine in
compression. Instead of grips, the flat disks were used top and bottom, with the curved
coupon centered between the two plates. These plates were then compressed until failure

of the sample.

——.l l l l l 1

Upper Plate ——>

 

 

Specimen P

Lower Plate I

Applied Force _, T T T T T T

Figure 3.4.5.3 Curved Specimen Compression Testing

53

4.1 Results and Discussion

This chapter will delve into the results obtained during this research, touching
on four main areas: glass fiber, carbon fiber, drapeability, and the corresponding

Halpin-Tsai predictions.

i _ l+g77V,
I)»: 1— 771/]
where:
f!— -1
_ P...
—— +
P... ,-
C = 2(61/13)

a and b are defined as the two dimensional parameters associated with a
reinforcement, most commonly length and width. The other parameters include the
fiber volume fraction (Vf) and the material and fiber property being estimated. (Pf and
Pm) The key parameters that will be presented and discussed are tensile strength,
tensile modulus, and fiber orientation.

The first section will describe the results from the three glass fiber systems
studied, polypropylene, nylon-12, and polyester terephthalate. Nylon-12 was
evaluated as an ADF matrix based on prior research. The nylon systems served as a
direct comparison to the prior results (i.e. a sanity check that the ADF was
functioning properly) and also as a baseline to compare prOperties with the carbon

fiber based systems. The other two matrices used in this research were chosen for

54

several factors, which included: high mechanical properties and widespread
commercial availability.

The second section will show the results from the corresponding carbon fiber
systems, the third section will detail the amount of orientation that was imparted to
the fibers during ADF processing, the fourth section will describe the Halpin-Tsai
relationships and how well the equations predicted the properties of the ADF
composites, and the final section will talk about the drapeability results from the glass
fiber system.

Below is a table listing the physical parameters of interest of the pure

materials used in this research. The first three listings are the matrices, the final two

are the fibers.

 

Table 4.1.1 Physical Properties of Fibers/Matrices Used

4.2 Glass Fiber Systems

A total of four glass fiber systems were investigated — two nylon-12, PP, and

PET. The two nylon-12 systems were nylon sized fibers and polyester (PE) sized

55

fibers. For the PP and PET systems, they were sized with PP and PET compatible

materials. It is important to note that the different sizings for the nylon-12 systems
were investigated to determine how much of an influence on mechanical properties
the sizing agent had. Below are the results of the two nylon systems.

In the beginning of this research, the amount of alignment imparted to the
fibers was minimal. In order to determine what process parameter had changed since
the prior researcher had worked at Michigan State, several detailed equipment checks
were conducted. Most of the initial checks revolved around the transformer and
voltage meters. However, one process variable that was thought to be insignificant
was the location of the apparatus. When the CMSC moved from the Research
Complex to it’s new home in the Engineering Building, the experimental conditions
Changed. The new building’s air conditioning facilities maintained the ambient air at
a much drier level. This lack of humidity was found to have a large processing effect.
In order to combat the lack of humidity, all of the glass fibers were kept in a humidity
Chamber were the relative humidity was kept at a level of around ~89-92%. It was
determined that the sizing agent on the fibers absorbed the water vapor, which greatly
increased the electrical polarization and aided alignment in the electric field.
However, it was still quite difficult to maintain the electric field at a strength of >450
KV/ In for sustained periods due to the dryness of the laboratory air. In a production
facility, it is imperative that the raw materials and electrodes be housed in a

controlled environment to provide process stability.

56

It was thought that the fiber sizing may have an influence on the mechanical
properties of a finished composite. A hydrophilic sizing may aid fiber alignment
more than a hydrophobic one, and the effect of humidity on the ADF process was

studied. To quantify the effect the sizing had on the final properties, the two nylon-
12 systems were directly compared. For the random fiber systems, the effect was
much greater on the tensile strength, as a ~30% tensile strength improvement
(averaged over the three fiber lengths) was noticed, where the tensile modulus only
increased ~4%. (the small increase for the modulus is most likely attributed to the
conflict of results between the random 0.25” fiber PE sized system — and the random
0.5” PE sized system) On the aligned systems, a 14% increase was seen for the
tensile strength and a ~7% increase was observed for the tensile modulus. The results
are briefly summarized in Table 4.2.1. The results suggest that the fiber sizing does
not have a large impact on the mechanical properties of the composite. Although the
random results show a large increase, due to the inherent nature of the random
composite, the result most likely is not significant. However, the 14% increase in

tensile strength and the 7% increase in tensile modulus for the aligned sample may be

Significant.

Strength Modulus

    

Aligned Random Aligned Random

 

 

 

 

O.25”(*) 21% 37% 3.3% 44%
l\()-5” 7.6% 38% 6.4% 17%
“\1” 14 % 13% 10% 9.3%
firm 14% 29% 6.7% 4.1%

 

 

 

 

 

 

Table 4.2.1 Percentage Improvement in Mechanical Properties due to Fiber
Sizing

57

-O—NytonlNylon Sized Glass ~Alimod
+NyleniPE Sized Glass - Aligned
+NylonINylon Stzod Glass - Mn
+NyloanE Sized Glass - Rm

Tensile Modulus (Mpsl)

 

O 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.2.1 Effect of Fiber Sizing on Tensile Modulus

+NyloniNylon Sized Glass- NW
+NyloniNylon Sized Glass - Rum
+NyloanE Sized Glass - Ahmed
+Ny|oanE Sized Glass - Rnldom

Tenslle Strength (pal)

 

O 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.2.2 Effect of Fiber Sizing on Tensile Strength

58

For direct comparison purposes, the 0.25” data points for the nylon sized fibers were
estimated by interpolating between the 0.125” and 0.5” properties. (0.25” nylon sized
fibers were not available for this research, so 0.125” fibers were substituted.)

The improvement in mechanical properties that was due to the fiber alignment
was similar to the results of the prior work. Tensile strength for the nylon-sized fibers
was increased by 64% for the 0.125”, 56% for the 0.50”, and 70% for the 1” fibers.
For the PE-sized fibers, 94% improvement for the 0.25”, 100% for the 0.50”, and
70% for the 1” fibers. The tensile modulus was increased 88% for the 0.125” nylon
sized system, 80% for the 0.50”, and 61% for the 1” system. For the PE sized system,
the increase was 53% for the 0.25”, 99% for the 0.5”, and 60% for the 1” system.
These results indicate that a high degree of alignment was achieved, which will be
described in more detail in section three.

The average tensile strength improvement due to ADF processing was 63%
for the nylon sized system and 88% for the PE sized system. The tensile modulus
was increased 76% for the nylon sized and 70% for the PE sized system.
Unfortunately, no statistically significant improvement due to fiber sizing
compatibility can be proven from the above results. Although it can be shown that
the overall mechanical properties are increased for the nylon compatible system, the
aVlfil‘age tensile strength improvement for the PE sized system is 25% more than the
nylon-sized system. This result indicates that the greatest effect on mechanical
pr Openies is the degree of alignment. Fiber sizing compatibility, although desirable

When designing a composite part for optimum performance, has a secondary effect on

59

the overall mechanical properties. Thus, a small increase in fiber alignment would
have more of an effect on the mechanical properties compared to a large increase in
compatibility between fiber and matrix.

When comparing results from this research to the work of Vyakarnam, er al.,
it is important to note that there is a 10% different in fiber volume fraction. This
research used 50% fiber, where 40% fiber was used for the prior work. Even with
less fiber, Vyakarnam was reporting mechanical properties that exceeded the results
in this research. Of the two mechanical properties, the largest difference was
observed between the tensile strength of Vyakarnam’s samples and those used in this
research, the difference between the tensile moduli was markedly less. As was
discussed in Chapter 3, the reason for this disparity was determined to be the
microstructure on the edge of the finished test coupons — laser cut vs. mechanically
grinded.

The polypropylene sized fibers used in this research were supplied by
Vetrotex Certainteed (who supplied all of the E-glass fibers). One of the problems
encountered working with these fibers was that the sizing agent was not as
structurally sound as the nylon and PET compatible sizings. Below are two optical
micrographs comparing nylon-sized fiber bundles with representative polypropylene
fiber bundles. (the nylon sized fiber bundle is on top, the PP sized bundle is on the
bottom) One can immediately notice that some of the bundles have broken down into
individual filaments. This made the processing of these fibers through the vibratory

feeder quite tedious and time consuming. Another problem encountered with the PP

60

system was the matrix. The PP supplied by Micron Powders, Inc., had a molecular

weight (weight average) of ~3,000 g/mol. Because of this low molecular weight (and

 

‘x I .
Figure 4.2.3 Optical Micrograph of Polypropylene and Nylon Sized Fibers

6|

comm 3. _. comm v.0
ccoN moo comm med
comp who cc 5 mm. P

£955. 3:82 £0.5ch 328$

58:3. 8:22
3393230528

082 m; 84.8 «N
82 I 089 EN
85 3 88 a:

mic «.282» _ can» 568$
599% 3:33 Sucoem. «:33:
SEE 8:22

859mm
«8.9m 532222

9.28: £29m 8a.“. «use «.3 2.3
can a... an: 9:29, $8 a 8358 __<

5:: a 2%; .3 8:05 ages: so. 92 8353 Ga 552

coocw mmN co Pom mcé
cocmw cod comNN 5N6
comp P cc. F coccp mad

£895 838: £895 838:
Eoocum been:

3302.9.

cocmp vc. _. ccomN mod

coco F om. _. cccx. P mod

88 2.0 83 :1
«.63 «.232» can» «.232»
5896 «:38: £995 838:

585: . 8.5.2

85832
320 m 5.58;:

:w

:80°

:36

.9F

coon-O

:mNd

:mufic

62

+PPIGlass - Random
1+PPIGIass - Aligned

Tenslle Modulus (Mpsl)

 

0 0.25 0.5 0.75 1 1.25
Flbsr Length (Inches)

S

'+PPIGlsss - Random
+ PPIGlsss - Aligned

Tsnslls Strongflt (psi)
(.0

 

2000
1000
0
0 0.25 0.5 0.75 1 1.25
Fiber Length (Inches)

Figure 4.2.5 Tensile Strength Results from the PP/Glass Systems

63

+PETIGIass - Random
+PET/Glass - Aligned

Tensile Hodulus (IPsI)
M
(II

    
    

0 0.25 0.5 0.75 1 1.25
Fiber Length (Inches)

Figure 4.2.6 Tensile Modulus Results from the PET/Glass Systems

20000

15°00 +PET/Glass - Random

+PETle -Allgned

Tensile Strength (psi)

10000

0 0.25 0.5 0.75 1 1.25
Flber Length (Inches)

Figure 4.2.7 Tensile Strength Results from the PET/Glass Systems

therefore short polymer chain length), the final consolidation of the composite part
without large-scale crystallization was impossible. The Tetrahedral press had the
ability to cool the platens in excess of 60°C/minute, but the PP would still become
crystallized and brittle. To help prevent the crystallization from occurring, quenching
the sample in an ice bath was attempted. This procedure still did not prevent the
crystallization. Eventually it was decided to test the PP samples as is, realizing that
all tensile strength data would be significantly depressed, while tensile modulus data
would still be meaningful. The results for the PP/glass system are presented in
Figures 4.2.4, 4.2.5 and Table 4.2.2, along with the results for the rest of the glass
systems.

When comparing the alignment effect on the tensile modulus, the 0.25”
system showed a 117% improvement, the 0.50” system showed a 161%
improvement, and the 1” system showed a 141% improvement. For the tensile
strength, the improvement was 75% for the 0.25”, 95% for the 0.50”, and 49% for the
1” system. The average improvement added by ADF processing for the PP system
was 140% for tensile modulus, and 73% for tensile strength.

The highest mechanical properties for all of the glass fiber systems was
exhibited by the PET/Glass fiber systems. This was expected, due to the higher
intrinsic properties of the PET matrix. One important difference between the PET
matrix used in this research and the nylon and PP was that the PET contained 10%
Titanium Dioxide (by weight) as a filler. Normally the role of fillers in matrices is to
reduce the amount of matrix in a given sample, therefore reducing the raw material

cost to the manufacturer. They also are used as a plasticizing material for

65

exceptionally rigid materials. Typically the mechanical properties of the filled matrix
will not be as large as that of the unfilled matrix. However, the mechanical
properties of the PET system used in this research do not appear to be affected in that
fashion, as they were higher than what was expected. A possible explanation might
be that the compression molding and/or the sample preparation caused a beneficial
alteration of the microstructure. The titanium dioxide might also have concentrated
around the fiber and acted as a stress/load conduit that more efficiently transmitted
the load. Either way, the mechanical properties were higher than expected but were
consistent across the sample set and no obvious explanation exists at this time.

When comparing the alignment effect on the tensile modulus, the 0.25”
system showed a 54% improvement, the 0.5” system showed a 64% improvement,
and the 1” system showed an 84% improvement. For the tensile strength, the
improvement was 39% for the 0.25”, 73% for the 0.50”, and 88% for the 1” system.
The average improvement added by ADF processing to the PET system was 67% for
tensile modulus, and 67% for tensile strength.

For all of the glass fiber systems, the ADF processing added at least a 50%
improvement in mechanical properties. The results also indicate that this
improvement increases with increasing fiber length. There are several possible
reasons for this. First, as will be shown in section 4.4, the longer fibers were aligned
to a greater degree. When processing the smaller fibers, they would be aligned as
they fall, but when they hit the deposition platform there would be some bouncing of
fibers, thus causing disruption of the alignment. The longer fibers would also exhibit

this bouncing effect, but to a lesser degree). As the fiber length increased, the

66

individual fiber itself became more ellipsoidal than cylindrical. This change in cross-
sectional geometry had a profound impact on the fluid mechanics. As a result, the
longer fibers were more ‘buoyant’ during the fall through the alignment chamber and
thus hit the deposition platform with less momentum per unit weight, causing less
bouncing and ‘dis’-alignment. The other possible explanation would be an argument
for the minimum effective fiber length — the longer fibers would theoretically act
more like continuous fibers (on a micro-mechanical scale) than the 0.25” and 0.5”

fibers.

Effect of Fiber Length on Tensile Modulus of
E-Glass Fiber ADF Composites

 

+Nylon/Nylon Sized Glass ~ Aligned
+Nylon/PE Sized Glass - Aligned
+Nylon/Nylon Sized Glass — Random
+Nylon/PE Sized Glass - Random
+PPIGlass - Random
+PPIGlass - Aligned
—O—PETlGlass - Random
——PETIGIass - Aligned

Tensile Modulus (Mpsi)

 

 

 

0 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.2.8 Glass Fiber Systems Tensile Modulus Comparison

67

Effect of Fiber Length on Tensile Strength of E-Glass Fiber
ADF Composites

35000 .

 

30000 —O—NylonlNylon Sized Glass- Aligned
+NylonlNylon Sized Glass - Random
+Nylon/PE Sized Glass - Aligned
+NyloanE Sized Glass - Random
+PPIGlass - Random
+PPIGIass - Aligned
--+—PETIGIass - Random
——PETIGIass - Aligned

25000
20000
1 5000

1 0000

Tenslle Strength (psi)

5000

 

 

 

O 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.2.9 Class Fiber Systems Tensile Strength Comparison

4.3 Carbon Fiber Systems

The primary motivation of this research was to develop a process in which to
align electrically conductive fibers, such as carbon fibers. In theory, the conductive
fiber would act as an electrical ‘bridge’ between the electrodes, thus shortening the
effective electrode spacing. Since the effective distance between the electrodes was
shortened, the localized electric field strength would be increased, thus causing more
electrical discharges (arcing) between the electrodes.

The carbon fibers used in this research were obtained from Hexcel. These
fibers were PAN based-A84 supplied in a 3K tow (continuous). The fibers were sent

out for sizing and chopping to Composite Materials, LLC, based in Mamoraneck,

68

NY. The fibers were coated with a nylon-12 compatible sizing. It is important to
note that where the glass fibers were ellipsoidal in cross-section, the carbon fibers
were perfectly circular. The carbon fibers were also much more dense as a bundle —
weighing on average 10X more on a bundle basis.

When carbon fibers were first used with the ADF process (before any
modifications to the apparatus) it was discovered that the carbon fibers would line up
much like iron filings subjected to a magnetic field when they were landed on the
deposition platform. There was also considerable arcing between the plates as the
fibers fell. However, when arcing occurred during processing of the glass fibers, the
fuse in the variac power supply would be blown. With the carbon fibers, the fuse did
not blow but instead the electrical field strength was decreased. This drainage of the
electrical field continued until the local instability corrected itself, i.e. the fiber fell
onto the deposition plate. Another phenomenon observed was that when the carbon
fibers passed underneath the electrodes, (when processing in continuous mode) the
field had a strong enough ‘pull’ on the fibers to lift one of the ends up to the
electrode. This effect, combined with the fact that the fibers would form ‘chains’
lining up between the electrodes, proved to be a difficult problem. Once the chain
was completed, with fibers spanning from the bottom of one electrode, across the
deposition platform, and up to the opposite electrode, the field would arc and actually
burn the Teflon release cloth underneath. These initial experiments indicated that to

be successful:

69

1) The electric field strength had to be < 100 KV/m (4 — 5X less than

for glass fibers)
2) The electrode spacing had to be increased.
3) The spacing between the deposition plate and the upper electrode

had to be increased.
4) A padded layer had to be added under the Teflon release cloth on

the deposition platform.

The padded layer had to be added to account for the increased mass of the carbon
fiber bundles. Without the pad, the carbon fibers would bounce upon impact and all
imparted alignment would be lost. One added benefit of processing the carbon fibers
was that the heat capacity of the carbon was higher than the glass, therefore the
sintering of the matrix powder onto the deposited mat was accomplished with much
less time.

When looking at the results of the carbon fiber system as shown in Figures
4.3.2 —— 4.3.5, it is easy to see that the ADF processing had a significant improvement
on the mechanical properties. When comparing the alignment effect on the tensile
modulus, the 0.25” system showed a 175% improvement, the 0.50” system showed a
149% improvement, and the 1” system showed a 60% improvement. For the tensile
strength, the improvement was 35% for the 0.25”, 100% for the 0.50”, and 15% for
the 1” system. The average improvement added by ADF processing to the
carbon/nylon system was 128% for tensile modulus, and 50% for tensile strength.

When analyzing these results, there appears to be a diminishing return on mechanical

70

property improvement as the fibers became longer — which is opposite of the glass
fiber system trends. However, there is a logical explanation for this. As was
discussed prior in this section, the cross-sectional geometry difference between the
glass and carbon fiber tows played a large role in the behavior when processed with
the ADF apparatus. Below, in Figure 4.3.1 is a diagram of the cross-sections of the

glass and carbon fiber tows:

   

Glass Fiber Carbon Fiber

Figure 4.3.1 Cross Sections of Glass and Carbon Fiber Bundles

 

+Ny|onlCarbon , Random
+Nylon/Carbon - Aligned

Tensile Strength (psi)
:3
O
o
O

 

 

 

 

0 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.3.2 Tensile Strength of Nylon-12/Carbon Fiber

71

+Nylonlei‘bon - Rel-idem
7; +Nylonlerbon - Aligned

Tensile Modulus (Mpsi)
N
0|

  
  

 

0 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

      

25000
j 20000
5
§ 15000
i:
as) +C-‘bonlNylon Random
3 10000 +GlsselNylonRsndan
. 3 +CarbonINyionAugnod
‘ g... +GlsselNyion Aligned
5000
l
0 , . . 7.7.; -- «1.11 .. . :. . :w: "a . . . '
0 0.25 0.5 0.75 1 1.25
Fiber Length (inches)

Figure 4.3.4 Comparison of Tensile Strength for Glass and Carbon Fiber ADF
Composites

72

 

 

 

 

f5
0.
E.
g +CarbonlNylon Random
é +GlasslNylon Random
2 +GlassINyion Aligned
2 +CerbonlNylon Aligned
'7:
C
O
'—

0 0.25 0.5 0.75 1 1.25

Fiber Length (inches)

Figure 4.3.5 Comparison of Tensile Modulus for Glass and Carbon Fiber ADF
Composites

Nylon/Nylon Carbon Fibers
Nylon-Sized

Aligned Random
Modulus Strength Modulus Strength
(Mpsi) (psi) (Mpsi) (psi)

0.25" 3.47 6300 1.26 4700
0.50" 4.1 1 13400 1.65 6700
1 " 4.97 25600 3.11 22200

All samples at 50% volume fiber +/- 3%

Table 4.3.1 Results for Carbon Fiber ADF Composites

73

The combination of the increased mass of the carbon fiber bundles and the more
aerodynamic shape of them resulted in the velocity at impact with the deposition plate
being greater than that of the glass fibers. Whereas one could visually see the 1”
glass fibers ‘fluttering’ down the alignment chamber, the 1” carbon fibers would fall
at a much faster rate. This resulted in the need for a compliant plate to reduce the
amount of ‘dis-alignment’ upon contact. Since the 0.25” and 0.50” fibers had less
momentum than that of the 1” fibers, they would be oriented to a larger extent —
resulting in a larger tensile modulus value. Another important difference between the
carbon and glass fibers was that the sizing material applied to the bundles appeared to
be thicker or of a stronger material. While processing, some glass fibers would
appear ‘fuzzy’, which under an optical microscope would appear as individual
filaments that had penetrated the sizing material. One could release the contents of
the glass fiber bundles by rolling them between ones fingers. The carbon fibers, on
the other hand, had a much greater structural integrity of the sizing agent. This lack
of ‘fuzziness’ allowed the carbon fibers to pack more efficiently overall than the glass
fibers, although the bouncing problem was still an issue. Also, the shorter carbon
fibers would make more efficient use of the space on the deposition platform as
compared to the longer carbon fibers. As is true in real life, the smaller the object, the
more efficiently the object can be packed. Since all of the ADF composites processed
in this research were made from several ADF mats, e.g., eight mats were used for
each composite, the weight of fibers going into the manufacture of each mat was
constant. The only variables were then the length and number of fibers. The longer

fiber ADF mats were made with less fibers than the shorter fiber ADF mats. Since

74

the carbon fibers were stiff, the 1” ADF mats were much thicker than those of the
0.25” and 0.50” fibers. This increased thickness was a direct result of the inefficient
packing of the fibers on the deposition plate Due to less parallel alignment for the
longer fibers and more crossing of the fibers.

When comparing the mechanical results for the carbon and glass fiber nylon
systems, they agree well with the physical property data. The carbon fibers have a
tensile modulus that is roughly 3X higher than that of glass, and the glass fibers have
a slightly larger tensile strength. In figures 4.3.4 — 4.3.5, one can see that the results
track these trends. An interesting result was observed that as the fiber length
increased, the amount of improvement in material property decreased compared to the
shorter fiber lengths. Whereas the 0.25” and 0.5” systems increased 175% and 149%,
respectively, for tensile modulus, the 1” fibers only increased 60% compared to the
random sample. When analyzing the fiber orientation diagrams in the next section,
they show that the longer carbon fibers also were aligned to a greater degree. The
only plausible explanation was that the random 1” carbon/nylon fiber system had an
usually large tensile modulus. Overall, the carbon fiber systems increased 128% for
tensile modulus and 50% for tensile strength, which shows that the ADF process adds

value to the raw material.

4.4 Measurement of Fiber Orientation

The key parameter of any ADF composite is how aligned the fibers are.

Without knowing how well the fibers are aligned, it would be impossible to predict

75

the mechanical properties using micromechanical models and determine if the ADF
process equipment was functioning at its optimal conditions. A measurement of the
alignment is known as the fiber orientation distribution (FOD). Typically, the fiber
orientation distribution is obtained by a measurement of the orientation of a
statistically significant number of fibers by cross sectioning the composite. The cross
section is then analyzed and all non-circular fiber cross sections are reconstructed
using the aspect ratio of the ellipse that is formed. This technique can be very tedious
and inaccurate, so a new technique developed earlier by a previous researcher was
used.. The technique measures the fiber orientation in an indirect way. A series of
images of fibers being deposited on the deposition plate were collected using a
Panasonic CCTV (Model WV 1410) camera, which was connected to a PC outfitted
with a frame-grabber board. Because glass fibers will reflect light preferentially
when it is incident at an angle, utmost care was taken to ensure that the lighting in the
room was uniform across the deposition plate. A statistically significant number of
fibers, 1000, were recorded for each F OD. On average, a larger number of images
were required for the larger fibers. The Global Lab Image Analysis software package
was used to digitally filter and sharpen the images of the fibers to enhance the edge
detection and improve the contrast of the fibers compared to the background (Teflon
release cloth). The “particle classify” routine in the software was used to identify the
fibers and their orientation with respect to a pre-entered guide. It is important to note
that fibers that were touching each other or otherwise connected in some fashion were

unable to be classified by the analysis package, therefore all fiber orientations are for

76

less than monolayer coverage of fibers. This means that any ‘dis-alignment’ that

occurs in normal processing of fibers

Global Lab image . Automatic and Manual
Analysis

f, = 2 ln(¢)cos’(¢)d¢—I

D

Figure 4.4.1 F OD Image Collection

(reprinted with permission from Vyakarnam. e! 01.)

falling on top of other fibers is not included in the FOD. These FOD’s can then be

0.50" 0.25"
Carbon Carbon

1" Glass ' 1" Carbon

 

 

+/- 20°
74% 72% 71% 68% 66% 65%

 

 

+/- 30°
82% 80% 79% 77% 76% 78%

 

 

 

 

 

 

 

All results +/— 3%

Table 4.4.1 Results of Alignment for ADF Systems
described as ‘best’ case. All of the FOD data for all six of the systems studied is

reported and a summary of the alignment is presented above in a table. Table 4.4.1

77

 

above lists the percentages of fibers oriented within a specified angle. For the glass
fibers, the trend is clear that increasing fiber length slightly increases the amount of
orientation that one can impart. The carbon fibers also exhibit this trend except for
the 1” and 'A” cases when looking at all fibers within +/- 30°. The carbon fibers also
appear to not be aligned as well, which makes sense. The process for aligning the
carbon fibers has not been fully optimized. A careful investigation into a pad
material which absorbs the most energy from the fibers upon impact could increase
the fiber alignment percentage. A clear indicator that the carbon fibers are still
bouncing when they hit the deposition plate is that while the data points for the+/- 10°
is around 10 % lower for carbon, the total amount within +/- 30° is roughly the same.
This same total within 30° suggests that an equivalent amount of alignment is
imparted to the carbon fibers, but the bouncing disrupts the alignment slightly.

One can describe the orientation state of a short fiber by two angles, 0 and 0. The
transformation matrix from the original (composite) coordination system of the fiber

to that of the three principle axes can be obtained by T below:

sin 6 cos ¢ sin 6 sin ¢ cos 6
T(3 D) = cos6 cos ¢ cos6 sin (15 — sin 6
— sin ¢ cos ¢ 0

For planar orientation, the transformation tensor simplifies to the following form:

78

Fiber Orientation Distribution for 1I4" Glass Fibers

30% .

25%

20%

.Aligned

. 15%
3: . Random

10%

, 5%

 

Angle

L._. .77

Figure 4.4.2 FOD for 0.25” Glass Fibers
(+/- 2 °/.)

79

. Fiber Orientation Distribution of 1I2" Glass Fibers

30%

25%

20%

IAIigned

‘ as 15%
IRandom

10%

5%

 

l 0% E;

Figure 4.4.3 FOD for 0.50” Glass Fibers
(+/- 2%)

80

' Fiber Orientation Distribution for 1" Glass Fibers

25%

20%

IAIigned

15% .Random

5%

 

0% . , , .
§A°§§x9§59>°°~°r909i9¢969699
Angie

Figure 4.4.4 FOD for 1” Glass Fibers

(+/- 2%)

Fiber Orientation Distribution for 1I4" Carbon Fibers

i 18%
16%
1 14%

12%
I Aligned

10% , .Random

8%
l 6%,
4%.

2%.

 

a°s>°>°n9°q9v96°0°
Angie

Figure 4.4.5 FOD for 0.25” Carbon Fibers
(+/. 2%)

82

25%

j 20%

15%

IAIigned
I Random

10%

it
fit
->

 

Figure 4.4.6 FOD for 0.50” Carbon Fibers
(+/- 2%)

83

 

Fiber Orientation Distribution for 1" Carbon Fibers

IAIigned

I Random

. . . in, «realist

(tori.

 

6.966699

99:19:~°°~°q9

59:05:99»?

Angle

bers

Carbon F

Figure 4.4.7 F 0D for 1”

(+/- 2%)

84

cos¢ sin¢ 0
T(2D) = — sin¢ cos¢ 0
0 0 1

Because the alignment imparted by the ADF process completely planar in nature, the
two-dimensional transformation will suffice. For planar orientations, the orientation
state of a composite can be described by a single parameter,f,,, which is the
trigonometric average of the fiber distribution obtained from the above tensor. For

symmetrical FOD’s (like the ADF process) f,, is given as:
f,, = 2<c052¢> — 1

Table 4.4.2 gives the values of f,, for the ADF composites:

Ahgmd 0.65 0.64 0.62 0.58 0.57 0.58

 

 

R d
an °m 0.03 0.08 0.07 0.04 0.11 0.09

 

 

 

 

 

 

 

Table 4.4.2 fp values for the ADF composites

(all results +/- 3%)

The results gathered in this work agree well with the results from the prior
researcher. In the case of fiber mats, Vyakarnam found that for all the four fiber
lengths, the orientations of about 70% of the fibers laid between +/- 20° and nearly
80~90% of the fibers between +/- 30°. The results from this work were roughly 70%

between 20° and slightly over 80% between 30°. Fiber orientation parameter results

85

 

 

also agree with the prior work — for 0.25”, 0.50” and 1” aligned Vyakarnam recorded

values of 0.63, 0.63, and 0.73, respectively.

4.5 Halpin-Tsai/Micromechanics

The three most important parameters that affect the mechanical properties of
discontinuous fiber composite systems (for a constant fiber volume fraction) are l)
the fiber orientation distribution (FOD) 2) fiber aspect ratio and 3) fiber-matrix
interaction. Fiber-matrix interaction can be best described as the amount of matrix
impregnation of the fiber bundle. This parameter depends greatly on the fiber sizing
and the interfacial adhesion between the matrix and the fiber sizing. While the first
two parameters play a pivotal role in the modulus predictions, the last parameter is
critical for the strength and fracture behavior. The role of the fiber volume fraction is
well understood and has a linear effect relationship to the mechanical properties as
expected from the Rule of Mixtures.

All ADF samples produced during this research were targeted for 50% fiber
volume fraction (+/- 3 %). Experimental matrix burn-off analysis of test coupons
verified this. Void content was comparable to samples made by the prior researcher.
Whereas commercially available continuous fiber composites are usually < 1% void,
the average void content in these ADF composites was ~ 4%. Higher void content
could influence tensile strength values, but tensile modulus should not be affected.

The degree of impregnation of the matrix into the chopped fiber bundle plays

an important role in determining whether the fiber or bundle will be the dominant

86

 

 

 

factor. Vyakarnam investigated whether the reinforcement aspect ratio was more

bundle or fibrous in nature. When using compatible fiber sizings with the matrix,

good impregnation can be achieved, as seen in micrographs shown in Chapter 3.

Incompatible fiber sizings, such as the PE-sized glass fibers, result in lower

impregnation of the fiber bundle and therefore lower mechanical properties.

Fiber aspect ratio is calculated by dividing the length of the fiber by the

diameter. Bundle aspect ratio is slightly more complicated to calculate:

Lb

ab:—

d

where b indicates the dimensions of the bundle. The following table lists all of the

aspect ratio for glass and carbon fibers.

 

 

 

 

 

 

 

 

 

Glass Fibers Carbon
Nominal Fibers
Length(mm) Filament Effective Filament Effective
Aspect Ratio Bundle Ratio Aspect Ratio Bundle Ratio
6.3 485 28.5 870 7.9
12.7 977 57.5 1750 16.0
25.4 1990 115 3500 32.0
Table 4.5.1 Fiber Bundle Length and Aspect Ratios

From the table above, one can see how the carbon fibers have a higher filament aspect

ratio and a smaller bundle ratio, due to the smaller fiber and the larger number of

filaments (i.e. 3000) that make up each bundle. The glass fibers, on the other hand,

had on average ~ 200 filaments per bundle.

87

 

 

     

    

fp

  

Exp.
Result

Halpin

Aligned

  

Tsai

Aligned

Model

Random

   

  

Predictions

Random

 

  

 

 

 

 

 

 

 

 

 

 

 

 

(inches) Aligned/ Allgned/ ‘F [lament ' ‘Bzmdlc ' ‘Fi/amcnl ' ’Bzmdle '
Random Random
0.25" 06210.03 1.841120 2.15 1.63 1.07 0.85
0.50" 0.64/008 2.331120 2.40 2.05 1.10 0.99
1" 065/007 2.657164 2.71 2.19 1.08 1.01
Table 4.5.2 Halpin-Tsai Model Predictions for the Glass Fiber/Nylon —12

Fiber
Length

        

    

System

fp

  

Exp.
Result

  

Halpin

Aligned

  

Tsai

Aligned

  

Model 1’

Random

redictions

Random

 

  

 

 

 

(inchc‘S) AIIgned/ Aligned/ 'Fi/unwnl ' ‘BumI/c ' ‘Fi/umcnl ' ‘Bzmdlc '
Random Random
0.25“ 0.62003 2.83/184 3.37 2.89 1.79 1.65
0.50“ 0.64/008 3.27/200 3.8-1 3.16 2.34 1.92
1" 0.65 "0.07 4.33/34 4.46 3.67 2.84 2.33

 

 

 

 

 

 

 

 

 

Table 4.5.3 Halpin-Tsai Model Predictions for the Glass Fiber/PET System

Fiber
Length

        

fn

Exp.
Result

   

Halpin

Aligned

Tsai

Aligned

Model

Random

 
     

Predictions

Random

 

 

 

(inches) Allgncd/ Al'g°°°/ ‘Fi/umcn/ ' 'Bum/lc ' ‘Filumcnl ’ 'Bunu’lc '
Random Random
0.25“ 0.62003 1.65/0.76 1.86 1.71 0.86 0.81
0.50“ 0.64008 2.48/095 2.57 1.85 0.94 0.87
1“ 065/007 3.4/1.41 3.22 2.28 1.1 1 0.99

 

 

 

 

 

 

 

 

Table 4.5.4 Halpin-Tsai Model Predictions for the Glass Fiber/PP System

88

 

    

  

Exp.
Result

 

Halpin

Aligned

Tsai

Aligned

  

 

Model

Random

Predictions

Random

 
    

 

 

 

 

 

 

 

 

 

 

 

 

 

(inches) Allgned/ Aligned/ ‘Filumcm ' ‘Bzmdle ' ‘Filumcm ' ‘Bundle '
Random Random
0.25" 0.58/004 3.471126 3.91 3.10 1.35 1.15
0.50“ 0.571011 4.111165 4.35 3.76 1.55 1.42
1“ 0.581009 4.97131 1 4.63 4.32 7.18 1.89
Table 4.5.5 Halpin-Tsai Model Predictions for the Carbon Fiber/Nylon-lZ

System

The Halpin—Tsai equation is given below:

P _1+g77V,

P " 1—27V,

I"

where:

P are the property to be calculated, subscripts f and m are fiber and matrix, and a, b
are geometric constants of the reinforcement.

The Halpin-Tsai equations were used to calculate two extremes of aspect
ratios. The ‘filament reinforcement’ was defined as compatible fiber sizing with the

matrix, and the ‘bundle reinforcement’ was defined as incompatible fiber sizing with

89

the matrix. The tensile modulus was calculated with the Halpin-Tsai equations for
both cases. Two fiber orientation states were also measured, varying from random (fp
values ~ 0.10) to ‘aligned’ (fp values ~ 0.6) The calculated modulus values were used
to obtain the modulus of the actual ADF composite by integrating the effect of fiber
orientation using the F OD. From the results on the two tables above, it appears that
the filament reinforcement approach predicts the values far better than the bundle

approach, which is contrary to the results in the prior work.

 

+ Aligned Filan‘ent

+ Aligned Bundle

+ Random Filament

+ Random Bundle

+Aligned - Experimental
Result

Tensile Modulus (mpsi)

+ Random - Experimental
Result

 

0 0.5 1 1.5
Fiber Length (inches)

 

 

 

Figure 4.5.1 Experimental vs. Model Predictions for Nylon/Glass Fiber System

The bundle approach underestimates the modulus for all of the glass and
carbon fiber systems. The filament approach tends to overestimate the modulus for
most cases, except for the 1” carbon fiber results, which it underestimates. The
results make sense from the standpoint that all of the fiber systems studied in this

work, except for the PE-sized glass fibers, had compatible sizings with the matrices.

90

4.6 Drapeability Results

In industrial use, aligned discontinuous fiber composites compete against
continuous fiber composites. In order to be commercially competitive, there must be
some advantage that attracts interest, such as case of processing, cost, or some type of
improved mechanical/physical property. ADF composites, when made from the same
type of fiber as continuous fiber composites, cannot match the strength of their
continuous fiber counterparts. However, ADF composites do possess two advantages
— they can be formed into complex geometric shapes and they are easier to process.
Since the fibers in the ADF preform are already ‘broken’, they have the mobility to
conform to a complex shape without the formation of voids where the continuous
fiber would have broken. ADF composite preforms can be manufactured in large
volume quite easily, and then be stamped into complex shapes with the same case —
something that continuous fiber composites cannot do. Thus, it was decided that
these attributes would be experimentally verified for the ADF process. The work
conducted was the first of it’s kind and several processing metrics were invented to
deal with this new property called ‘drapeability’. The most important was how to test
this parameter, which will be described later in this section.

The processing of the samples for the drapeability work was the same as for
the regular ADF systems. Eight ADF mats were processed for each composite
sample under identical processing conditions. The molds for this work were

described in Chapter 3. Three molds were made with a 0.125” radius of curvature,

91

 

one with 0.25”, and one with 0.50”. The composites mats were hand-laid into the
mold, with an additional amount of matrix added to the top to counter the losses of
matrix through the sides of the mold. The entire assembly was then vacuum bagged
and processed with the Tetrahedron press. The only processing difference between
the shaped and flat ADF composites was that additional heating time was required for
the shaped composites — due to the differences in the size of the mold.

A sample preparation technique had to be developed for this new test, as
ASTM did not have a standard method for testing the curved specimens. The
samples were first cut with the Felker saw so that the length of the sample was 2
inches from the middle of the curved section. Once the 2-inch gauge length was cut,
the piece was then sectioned into 6 samples, which then had the sides sanded with
consecutively finer grits of sandpaper. Once the samples had been polished and
allowed to dry at room conditions overnight, they were tested using the same
equipment as the tensile specimens. However, the equipment was modified to use
compression instead of tension. Large cylindrical load plates were attached to the

load points, with the curved sample resting in the center of the lower disc.

Upper Load Plate >

 

 

l

Specimen

 

 

 

Lower Load Plate i l

 

 

Figure 4.6.1 Compression Testing Apparatus

 

Compression was then applied at a rate of I/2”/min.

Fiber Length (all \alues lbs/111’)

 

 

 

 

 

0.25” Aligned 0.50” Aligned 1” Aligned
0.125” 44 23 54
0.25” 33 49 60
0.50” 20 57 82

 

 

 

 

 

 

Table 4.6.1 Radius of Curvature Results

(all results +/- 4 %)

The data above is considered to be the strength of the curved ‘joint’. The
same equation for tensile strength computation was used for the compression data.
Where the strength was equivalent to the compressive force divided by the cross
sectional area of the joint As one can see from the table, the 0.25” fiber exhibited the
largest compressive strength at the shortest radius of curvature. As the ROC
increased, the same length of fiber had a more difficult ‘bridging’ the ROC gap. For
example, at 0.25” and 0.50”, the 0.25” fiber was not able to fully ‘bridge’ the gap and
thus exhibited lower strength numbers. The same phenomenon is true for the 0.5”
and 1” fibers. From this preliminary research, if the ROC study was expanded to

0.75”, l”, and 1.5”, we would observe a ‘peaking’ of compressive strength for the

93

0.50” and 1” fibers, much like what was seen for the 0.25” fibers in the 0.125” to
0.25” progression. The 0.25” fiber ‘peaked’ at the 0.125” ROC, where the 0.50” and
1” fibers both became stronger as the ROC increased.

The drapeability testing done in this research was quite basic. The data
suggested several trends discussed above, but the magnitudes of the mechanical
properties are not indicative of the potential strengths of the ADF composites when
stamped into complex shapes. More work is needed in this area, such as comparisons
of randomly oriented samples with the aligned samples and the measurement of the

properties of a neat resin coupon of the same configuration.

94

 

1 100.00

90.00
I 80.00
70.00
60.00
......1/ " Aligned

+1] " Aligned
_._ 1" Aligned

50.00

40.00

Force at Break (lbsIan)

30.00

20.00

1 10.00 .

 

0.00

6" 0“ \0’

Radius of Curvature (in)

Figure 4.6.2 Radius of Curvature Results

95

5.1 Introduction

An economic analysis of the Aligned Discontinuous Fiber Composites Process
(ADF) has been conducted to estimate the costs associated with this process. This
evaluation broke the cost per pound of finished product down into three main categories,
capital costs, labor costs, and material costs.

The initial phase of the analysis evaluated the current system. This system has the
capacity to produced ~18,000 pounds of composite preform per year. At this rate, the
cost per pound of composite comes out to be $6.51/lb (without material costs). However,
because of the limited laboratory scale apparatus currently in use, greater speeds and
throughput are expected upon industrial scale-up. This should reduce the cost of
composite significantly.

The second phase of the analysis evaluated a scaled up production ADF process.
This scaled up system has the capacity to process 1,000,000 pounds of preform per year.
Analysis shows that the cost per pound of preform approaches ~$0.35/1b.

The ADF process has the potential to be cost effective composite production
process. The process requires state-of-the-art off-the-shelf equipment with low power
requirements, a minimum of moving parts and operates under principles, which can easily
be scaled up without new machine design requirements. The ADF process appears to
have the speed and reliability, which make the process desirable for large-scale industrial

production.

96

5.2 Process Description

The process being evaluated produces aligned discontinuous fiber (ADF)
composites. The initial step of the process uses a vibratory feeder to feed discontinuous
bundles of reinforcing fibers (e.g. glass or carbon) into the alignment chamber. The
feeder also “pro-aligns” the bundles into a planar orientation perpendicular to the electric
field and also controls the rate at which the fiber bundles are fed to the alignment
chamber. The fiber bundles then pass through the high voltage alignment chamber where
they come under the influence of an electrical field, which produces a polarization torque
on fibers oriented at an angle to the field, which causes them to align. This alignment
chamber consists of two parallel plate copper electrodes housed within a plexi-glass
support structure. An important feature of this process is that since the electrodes are
parallel in alignment, one can scale up the process according to theory in both the x- and
y-directions. The aligned bundles then deposit onto a moving veil, which runs at a
controlled speed predetermined by the thickness required. This veil then passes under a
powder dispersion apparatus, which coats the fibers with the desired amount of
thermoplastic matrix in powder form. Alternatively, the powder can be coated on
individual fibers with the MSU high-speed polymer powder process and then chopped
into short bundles. After being coated with the powder, the preform sheet passes under
an infrared heater, which sinters the polymer particles to the fiber bundles and ensures
that the desired alignment is not disturbed during subsequent processing/handling. The
sintered preform sheet can then be carried by the veil to a take-off apparatus which can

roll the preform into drums for packaging and delivery.

97

 

The electric field strength is monitored via a high voltage probe that is connected
to the high voltage plate. The high voltage probe is in turn connected to a multimeter
voltage readout. Control of the electric field strength can be accomplished two ways, by
altering the setting on the variac unit (which controls the input voltage to the
transformer), or by changing the separation distance of the electrodes. The parallel plate
geometry simplifies the mathematics, yielding a straightforward relationship between the
electric field strength and the electrode gap.

E = V/d
Where E is the electric field strength, V is the input voltage, and d is the electrode
separation. Control of the fiber orientation can be maintained by periodic examination of
the fiber orientation diagram (FOD), which is measured by a video camera/computer tool.

When working with E-glass fiber composites it is important to have control over
the environmental conditions. Earlier in this research it was noticed that the humidity of
the workspace played an important role. Humidity of ~50-60% is sufficient to reduce the
breakdown voltage of the equipment sufficiently to prevent excess discharges. On a
worker’s safety note — the electric field utilizes a high voltage, however the amperage is
stepped down to the ~15 milliamp range which is well under the heart stopping threshold.
Although one would not want to “bridge” the electrodes, it would not do any serious

damage to the individual.

98

 

5.3 Process Economic Model

In order to effectively estimate the costs associated with different levels of
production, a cost and economic model was developed using a spreadsheet running under
Microsoft Excel. The main elements of this model are available in Tables 5.4.2 and
5.5.1. The model takes into account all of the usual costs associated with operating a
process, i.e. real estate, HVAC, utilities, materials, etc. However, since the cost of
producing the preform is going to depend to a large extent on the cost of the composite
fiber and matrix, it was decided to compute the cost per pound of the material,
independent of the material costs so that the evaluation would only take into account the
process.

The costs were broken down into two main categories, capital and variable costs.
The capital costs were broken down into the major components of the system. The
estimates used for these major components were derived based upon costs incurred in
building the current bench top system or through quotes from industrial vendors. The
useful life of the process was assumed to be three calendar years. This was assumed
because of the nature of the equipment. The capacitor is operating at constant voltage;
the conveyor and its components are off-the-shelf items, and the electrodes are static and

are not degraded by the process.

99

 

1,000,000 lbs/year Example

 

1) 1,000,000 lbs /( 80% of 365 days/year (using 80% uptime) * 24 hours/day) =
143 lbs/hour preform
2) Next, we chose 50% fiber volume fraction:
Checking densities: 1.94 * 10'3 lb/cm3 Polypropylene
5.62 * 10'3 lb/cm3 E-Glass Fiber
The ratio of densities = 5.62/1.94 => 2.89

Now it is a simple algebraic equation:

2.89X + X = 143 lbs/hour preform
3.89X = 143 lbs/hour
X = 36.8 lbs/hour

3) Therefore:
Flowrate of fiber = 2.89X => 106 lbs/hour E-Glass
F lowrate of matrix = X => + 37 lbs/hour Polypropylene

For a total of 143 lbs/hour preform

Figure 5.3.1 Equipment Sizing Calculations

100

The variable costs included the utilities, matrix, fibers, and labor. A labor rate of
$15/hr was assumed for this process. This rate was applied to production by dividing the
$15 figure by the pounds produced per hour by one operator. Utilities were computed by
measuring the amount of electrical usage in an hour and multiplying by the cost of
electricity per hour. Finally, the material costs were computed by first determining the

desired volume fraction, 50% fiber (by volume) in this work, and then back-calculating

the weight per hour from the total production for one year. An example of this

calculation can be found in Figure 5.3.1.

5.4 Current System Evaluation

The current bench-top system deposits chopped fiber bundles (200-400
filaments/bundle for E-glass) onto a moving deposition plate at a rate of 2 lb/hr, resulting
in a yearly production value of ~1 8,000 lbs/yr.(schematic of process in shown in Figure
5.4.2) The preform is 8 inches wide and would be wound up onto a roll for shipping and
handling. The system also has a relatively small footprint, it only would take up ~200 ft2
of floor space. One operator can easily supervise this process and possibly more, which
would reduce the overall cost of the process. The ability to use one operator to run
multiple systems has a major effect on the costs as seen in Table 5.4.2. Results from the

economic model are listed in Table 5.4.1

101

6395 e5. 23m .3 6.. 5:2.an :5 3.6 «sec

 

 

 

 

5:880 5360 ”626.0 0:82.". ”6:255 $6.950

”99mm .62”. EoEoom

uses. 566633 ”.3

”8.23.0 coEmoaoo

”defiance;

$005
.0982". 33:51

 

n=0 950:5

 

£2.on
.3820 EoE:o__< 6580: Lo Emcee...

”50.0 0660.2". coach ”mates oom=o> :91

gs: mn< o_aom.nu._ 2.: ..2 mumoo *0 Caz—5:35

 

 

102

..NKnm. _. P»
afimtop 5
v _..m¢m.mmmw

«00550. 5

2.8.9»
52.8.2»

505553
50$ 5.09

.8898...»
6.58.8.8

29.5.2»
$80 6.8:;

“.36.. + $6.88.). o? 980 .90.»
8.6.5.6: Q; 880 .68»
8800 some; .65...

50.? 2mm
mm P. a
00 ..w

8.8...»
.985. 888..»
.8680 2.6.6:: 8.8»
.__o 9.8...» 8.8»
.58.. 8.8»
8.2»
5660.53. “.3 8.28
.3520 5668.2 8.88
.383...» ©. ”esteem 8.2»
8.8.»

.98.“. .8685 8.8»
2.8.2 8.8360 8.8»
“55.0 88.60 8...: 8.8m»
OmfitoE< 0mm>-m

omofimmw

ommd a
coin»
com»

0013
000.3
0mm»
8 5
on»
om»
omnfim
com. 8
on»
com»

on 5
ommw

oedema
3.00

$80 .8060

31.8% 382.com 23m roughened Sam 039—.

£80 .3028: 22. 2:» Ed 628m P
mace .6588? 2»: .9500 .00 800
8800 .90... so... Edam?

6.80 5.686.226 2.8.8888 .22. 8.85.8.0
02.808 .0960

:23... m.

noose... .5855
"Seam :30
22.52 £35.:
“26396; 8....

”8...; 89.8 8.... 363.88.» 00 .6 3.28

668%» 68.5.6.0 8663.88.» .66.“. .o 5.88

H.922 53.20036 .260 26.0 5.85.

H5.88.0 5:80.000 EEK 68.x 26.0 52.0
826.0 2.2.9.0

658.2. 03.0 83 E288”.
$5826....

Sagas: oE:_o> .090
”320 oom=o> :9...
US$5.29...

022. :5 “6.220 .0 gm

magmanoo 863090200520 86.0
6.68 96.. - 8.6053 8890 89.0 385.585 3.5.?

103

ease—aim ma bow-weenaa aim charm

:om cosioaoo

0 n

Hoonm 0:80:80 an? 09.8009 .6250”

Logo: BEN—Ma:

 

 

mevoboofi .

..M \.

' llll-.Iel'l.-‘.'..
3.50m om8.o> emf
0:253:40

8: .2600Eovoom

 

6.5 66.0 886.6

8880 mQ< 8525.80 0.8m b08883 .«o ouafionom

104

Table 5.4.2 - Effect of Labor Cost on Laboratory Scale ADF Process Economics

Systcms/

 

Operator
Added

Cost/lb. $6.51 $3.72 7 $2.79 $2.32 $2.05

 

After reviewing the analysis developed for the current system, it is apparent that it is
highly cost prohibitive. At ~$6/lb added to the material cost during processing, the price
of the ADF composite preform product would be very high compared to other processes
and therefore noncompetitive. On the positive side, the total capital investment for
setting up a pilot-scale line is relatively low. It is important to note that the main factor
contributing to the high costs in the lab scale is the capital costs. The low-volume of
material produced coupled with the expensive pieces of equipment combine to push the
$/1b upwards. Figure 5.4.1 shows the distribution of the various costs in the laboratory

scale process.

5.5 Scale-up Evaluation

The industrial scale ADF model is based on 1,000,000 lbs of preform per year.
The preform would measure 36” across, which would allow for the stamping of relatively
large, complex parts. Based on simple capacitor theory, scale-up of the ADF process
would be relatively simple. A schematic diagram of the industrial process is shown in

Figure 5.5.1. Since there is negligible charge flow between the electrodes, the

105

 

transformer is used only to sustain the electric field. Based on this fact, one large
transformer could power an entire process line. Scale-up of the electrodes is
accomplished quite easily in the x- and y-directions. Although in this analysis a 36” size
preform was chosen, one can easily fabricate a much larger preform for the stamping of
very large, complex parts, e. g. car doors, undercarriage battery holders for electric
vehicles, etc. The thickness of the preform mat can easily be controlled by the spacing
and number of ADF Alignment chambers. For example, using a wider spacing between
the electrodes will increase the amount of fiber that can be processed in that chamber.
The reason for this is that there is a critical fiber density, which when exceeded causes
misalignment of fibers. This misalignment happens because a large amount of fibers
come into contact and ‘clump’ together, becoming harder to align as they fall. When the
electrodes are moved farther apart, this fiber density is decreased, allowing for larger
fiber flows. However, increasing the electrode gap will require a larger input current
because of the electric field strength relation discussed in Section 5.2.

The economic results for the 1,000,000 lb/yr process are listed in Table 5.5.2.
Having one operator operate the industrial scale ADF Process adds $0.35 to each pound
of material processed. Adding multiple ADF lines under the control of single operator
greatly improves the economics, as seen in Table 5.5. 1. As one can see, the labor cost.
Table 5.5.1 Effect of Labor Cost on Industrial Scale ADF Process Economics

Systems/

 

Operator
Added

Cost/lb. $0.35 $0.30 , $0.28 $0.28 7 $0.27

 

106

greatly affects the economics, which means that the equipment used in the ADF process
is relatively inexpensive

Compared to the laboratory scale, the industrial scale price is fairly independent
of the capital costs. As Table 5.5.1 shows, the labor cost has an affect to a certain degree,
but does not in any way drive the economics. The relationship between the various costs

and comparison with the bench scale model can be seen in Figure 5.5.2.

107

ouufiosom Etna—5:— hQ< “.mfi PEME

EEO @393. wocaoi 0:89:00

20m $3250 23H 3:09.00

 

 

 

 

 

£20 0320M couohi
530m “$2.35 I- - - - - -
55809 536m» .iI‘
585:0 Eufiam__<i+ . . :8 E5 mm? 29%

baqsm om§o> swam

9:8 as an? 32>? E 8 =5 2% is: 838; 5a

:5 ":2 28m 353220 Osmaofim

108

omfiomfin 5
omfimmdwwa
«Qmmodofimw

fife—aum 3.5365 .8» 838m wag—«5‘ 382.com «Wm 939—.

:33 + $3292 9; 230 .92
8.8.222 o? 880 .80...
8800 28:85 .82.
88.888» ”88 :28
883.8» 8.88.88» :83? e: ”8... 5“.

8.88.8» 8.08.8» 8888858:
8.89m» 8.892» ”382.. m.
8.8» 8.88» 8822 8.25::
8.88.8 8.88.8» uo<>z
6288. 5 88m...» 8686» 868.03 888w .mom
5288 £28: 8.8» 8.85 ”28322 88
8.8.2» “=0 888 8.8.. .» 8.8»..8 8888 .838
59:88.» :83 8.8 F. F» 8.88» :88”. bombs
5888» .3220 .8892 8.2» 8.8» 88.2, 88% 8.:
858.3» .8858 an.» 8.8» 8.88» 8888 «8.965
88 88.888883 8.8» 8.83 ”885 2.258
8.8 F» 8.88 888; 85.2.0.8:
8:888» ”288 8880 8.8» 8.8.» 8828.2
8388.8 "88.2 8.838 8.8» 8.88» “32a 8§o> 8.:
9988.0 .» “.86 88.8 :83 8.88» 8.88 8.5283
.88 288>
980 .588

888500 82893883.... 88.6
28w 8:83... I §Eocoom 889a Sen. macs—5:085 3:92

mamoo .thME 93 n_\emm.o
$80 .3886? 9:2.”

$82. 8080 F

5888 :82 $8.83

@803. mm 506

9:53. mmood
EBB Pvmvdm
EB. «.00..

59E AXbm

22; 58

109

$38.... .3... .2585 2:8 883.85 .80 8.8 28:.

 

 

 

 

 

8.28%

”50.0 08.06.”. 00:3 E9555 u.n.<

2080000 50300

           
     

”99mm. .83.

:9me m.

batsmm.
.mpEmco EmECQZ
535008.905.

Es ".04. 28» 8.8295 .2 280 B 5:39:85

 

 

110

6.1 Conclusions

The ADF process was successfully modified to accommodate alignment and
processing of conductive fibers. ADF composites have superior properties to random '
orientation composites. An investigation into the thermoforming behavior and properties
of ADF composites was also conducted. The ADF process has been shown to be capable
of being utilized with any discontinuous fiber to produce rapid pre-forms that can be
shaped into complex geometries.

An additional question is the economics of large-scale commercialization of the
process. An economic analysis was performed on the ADF process. The result of the
analysis was that ADF processing adds a cost of $0.35/lb at an annual production of
1,000,000 lb/year. With the economic analysis complete, all barriers to industrial scale-

up for manufacturing have been removed.
Conclusions of this work are summarized below:

1) Results from the prior researcher were duplicated for nylon-12/ glass fiber
ADF composites. The alignment of sized fibers was found to be sensitive to
changes in the relative humidity of the processing environment. Allglass
fibers were housed in a humidity chamber (relative humidity > 90%) prior to
alignment. Carbon fibers did not exhibit this phenomenon due to the large
intrinsic conductivity of the fiber. All fiber systems studied showed a
mechanical property improvement of at least 70% for both strength and

modulus.

lll

2)

3)

4)

5)

6)

The ADF process was modified to process conductive (i.e. carbon) fibers.
The key modifications were to further separate the electrodes, add a compliant
pad to the deposition plate to reduce fiber bouncing, and increase the distance

between the upper and lower set of electrodes.

Data was collected for alternate polymer matrix ADF composites. PET and

PP/E-Glass ADF composites were manufactured and tested.

The fiber orientation distribution of the ADF composites was shown to be, ~

70% of fibers were aligned +/- 20° and nearly 80% within +/- 30°.

E-glass fibers were oriented to a higher degree than the carbon fibers. The
fiber orientation parameter, fp, of the glass fiber ADF composites was between
0.62 — 0.65, while the carbon fiber ADF composite, fp was between 0.52 —
0.58. This indicates that the carbon fiber process has not been fully optimized

and further work needs to be done in this area to address this.

A study of the drapeability of the ADF composites was conducted. Three
molds of various radii of curvature were constructed to investigate how well
ADF composites can be formed into complex shapes. Molds with radii of
1/8”, Ma”, and '/2” were made for use with nylon/glass fiber ADF composites.

Results indicate that fiber length should be longer than the radius of curvature

ll2

to get the strength benefit of ADF composites. Based on this result,
commercial use for complex shaped parts should use a distribution of fiber

lengths to maximize part strength.

7) Micromechanical models were employed to predict the mechanical properties
of finished composites. The Halpin-Tsai equations were used with two
different reinforcement types, filament and bundle. It was found that
‘filament’ reinforcement predicted the composite properties better than that of

bundle reinforcement.

8) A detailed economic analysis was completed on the ADF process, for both the
laboratory scale (18,000 lbs/year ADF composite), and industrial (1,000,000

lbs/year). Cost added per pound for laboratory scale processing was $6.51/lb,

while the industrial scale reduced the cost to $0.35/lb.

6.2 Future Work

During this work, several off-shoots were discovered that were worthy of

investigation at a future date. Below are summary of these ideas:

1) Examine the conductivity of fibers sizings and how they affect alignment

of fibers. Of the three different sizings on the fibers used in this research.

2)

3)

4)

5)

it appeared that the most hydrophilic sizing was the easiest to align in the

process chamber.

Further work needs to be done on the drapeability ofADF composites.
The results here should be compared to random composites to see the
benefit of ADF processing. Also, carbon fibers should be investigated to
see whether their increased stiffness will allow them to be processed in
this fashion. More mold geometries should be investigated beyond the

single ‘L’ shape.

Process limitations, such as fiber flow through the chamber needs to be
studied to see where the drop-off in alignment occurs. There will be a
finite density of fibers in the chamber that when exceeded results in a

degradation of orientation.

Construct an industrial proto-type, which would consist of several
alignment chambers in series to build-up the preform thickness. What

affect would this have upon the final orientation?

Investigate alternative strategies of aligning fibers in the z-axis for use in
adhesives. If fibers could be aligned perpendicular to the adhesive joint, a
higher bonding strength and rigidity would be observed. One could take

advantage of the phenomenon where the fiber was pulled upwards to stand

ll4

6)

7)

on end by the electrode passing over it to make the fibers ‘stand’ up. An
adhesive sheet passing below the electrode could then ‘capture’ this

alignment for future consolidation in an adhesive tape.

Create a laboratory unit with more than one pair of electrodes in the
processing chamber. By alternating power through the various sets of
electrodes, a multi-directional laminate structure could be manufactured

with discontinuous fibers.

Investigate advanced fiber/matrix systems used in the aerospace industry
with the ADF process. While the critical parts of the aircraft will still
need to constructed of the continuous fiber version of the advanced
system, various non-critical parts could be fashioned from ADF versions

to save weight and cost.

ll5

List of References

 

Agarwal, B. D. and Broutman, L. J ., “Analysis and Performance ofFiber (..‘ompositef’,
John Wiley and Sons, 1980.

Bigg, D. M., “Mechanical Property Enhancement of Semicrystalline Polymers — A
Review”, Polymer Engineering and Science, Vol 28, pp 830 — 841, I988.

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.

Bigg, D. M., Hiscock, D. F ., Preston, J. R. and Bradbury, E. J., “High Performance
Thermoplastic Matrix Composites”, Journal of Thermoplastic Composite
Materials, Vol I, pp 146 — 160, I988.

Bigg, D. M. and Preston, J.R., “Stamping of Thermoplastic Matrix Composites”,
ANTEC, pp. 1450 — 1453, 1989.

Bozarth, J. M., Gillespie, J. W. and McCullough, R. L., “Fiber Orientation and Its Effect
upon Thermoelastic Propreties of Short Carbon Fiber Reinforced
Polyetheretherketone”, ANTEC, pp 568 — 572, 1986.

Caprino, G., Tagliaferri, V., Covelli, L., “The Importance of Material Structure in the
Laser Cutting of Glass Fiber Reinforced Plastic Composites”, Journal of

Engineering Materials and Technology, Vol I 17, pp 133 - 138, 1995.

Chang, I. And Pratte, J. F., “LDF Thermoplastic Composites Technology”, Journal of
Thermoplastic Composite Materials, Vol 4, pp 227 — 252, 1991.

Chou, T.W., “Microstructural Design of Fiber Composites”, Cambridge University
Press, 1992.

Clarke, D. R., Suresh, S. and Ward, 1. M., “An Introduction to Composite Materials”,
Cambridge University Press, 1996.

Composites - Engineered Materials Handbook, ASM International, Ohio, 1987.
Davies, P, Manson, J, Roulin, M., Kausch, H., Echalier, B. and Jaxquemet, R., “Structure
and Properties of a Stampable Thermoplastic Composite”, Journal of

Thermoplastic Composite Materials, Vol 4, pp 285 — 298, 1991.

Florian, John, “Practical Thermoforming”. Marcel Dekker, I996.

ll6

Friedrich, K. and Scherer, R., “Incremental Deformation Studies During Matched-Die
Forming of Continuous Carbon Fiber Reinforced Polypropylene Laminates”,
Composite Material Technology 111, pp 163 — 175.

Giles, H. F. and Reinhard, D. L., “Compression Molding of Polypropylene Glass
Composites”, 36th International SAMPE Symposium Proceedings, pp556 — 571,
1991.

Gruenwald, G., “Thermoforming: A Plastics Processing Guide”, Technomic, 1987.

Guell, D. C. and Papathanasiou, T. D., “Flow-Induced Alignment in Composite
Materials”, Woodhead Publishing, 1997.

Haplin, J .C. and Kardos, J. L., “Strength of Discontinuous Reinforced Composites”,
Polymer Engineering and Science ", Vol 18, pp 496 — 504, 1978.

Halpin, J. C. and Kardos, J. L., “The Halpin-Tsai Equations: A Review”, Polymer
Engineering and Science, Vol 16, pp 344 — 352, 1976.

Ikegawa, N., Hamada, H. and Maekawa, 2., “Effect of Compression Process on Void
Behavior in Structural Resin Transfer Molding”, Polymer Engineering and
Science, Vol 36, pp 953 — 962, 1996.

Kacir, L., Narkis, M. and Ishai, O., “Oriented Short Glass-Fiber Composites — l -
Preparation and Statistical Analysis of Aligned Fiber Mats”, Polymer Engineering
and Science, Vol 15, pp 525 — 531, 1975.

Lavengood, R.E., “Strength of Short-Fiber Reinforced Composites”, Polymer
Engineering and Science, Vol 12, pp 48 — 52, 1972.

Lee, J. H., Vogel, J. H., “An Investigation of the Formability of Long Fiber
Thermoplastic Composite Sheets”, Journal of Engineering Materials and
Technology, Vol 117, pp 127 - 132, 1995.

Lee, J. L. and F ong, L., “Preforming Simulation of Thermoformable Fiber Mats using
Geometric Modeling Approach”, Science and Engineering of Composite

Materials ”, Vol 4, pp 133 — 142, 1995.

Mallick, P.K., “Fiber Reinforced Composites — Materials. Manufacturing, and Design”,
Marcel Dekker, 1988.

Moore, H. D. and Kibbey, D. R., “Manufacturing: Materials and Processes”, Grid
Publishing, 1982.

Muzzy, J. D., Colton, J. S. and Wu, X., “Thermoforming of High Performance
Thermoplastic Composites”, Polymer Composites, Vol 1 1 pp 280 — 285, 1990.

117

Nangrani, K., Wenger, R. and Neel, P., “Effect of Moisture on Mechanical Properties of
Glass Reinforced Polyethylene Terephthalate”, ANTEC, pp 79 —— 83, 1986.

Okine, R. K., “A Formable Advanced Thermoplastic Composite Sheet Based on an
Excess Filament Length Concept”, Journal of Thermoplastic Composite
Materials, Vol 5, pp 152 — 165, 1992.

Okine, R. K., “Analysis of Forming Parts from Advanced Thermoplastic Composite
Sheet Materials”, Journal of Thermoplastic Composite Materials, Vol 2, pp 50 —
76, 1989.

Okine, R. K., Edison, D. H. and Little, N. K., “Properties and Formability of an Aligned
Discontinuous Fiber Thermoplastic Sheet”, Journal of Reinforced Plastics and
Composites, Vol 8, pp 70-91, 1990.

Piggott, M., “Load-Bearing Fibre Composites”, Pergamon Press, 1987.
Schwartz, M. M., “Composite Materials Handbook”, McGraw-Hill, 1992.

Schwartz, RT. and Schwartz, H.S., “Fundamental Aspects of Fiber Reinforced Plastic
Composites”, Interscience, 1968.

Snyder, D. L., Strong, B. and Lawson, J. S., “Postforming of Thermoplastic Composite I-
Beams”, Journal of Thermoplastic Composite Materials, Vol 4, pp 2 - 19, 1991.

Soh, SK. and Tseng, A.A., “Processing of Polymers and Polymeric Composites”,
ASME, 1990.

Sturman, P. C. and McCullough R. L., “Static Field Electrical Conductivity of Dilute
Fiber Suspensions”, Journal of Applied Physics, Vol 72, pp 2883 — 2888, 1992.

Suemasu, H., Friedrich, K. and Hou, M., “On Deformation of woven fabric-reinforced
thermoplastic composites during stamp-forming”, Composites Manufacturing.
V015, pp 31 — 39, 1994.

Throne, John, “Thermoforming”, Hanser Publishers, 1987.

Tsuji, N., Springer, G. S. and Hegedus, I., “The Drapeability of Aligned Discontinuous
Fiber Composites”, Journal of Composite Materials, Vol 31, pp 428 — 465, 1997.

Vyakarnam, M., “Processing Discontinuous Fiber Polymer Composites: Fiber Alignment

using Electric Fields and Microstructure Relationships Ph.D. Dissertation.
Michigan State University, 1996.

118

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

lllll11111111111111IIIZIIJIUIIZIIIHSIMIIWII