2006:60

LICENTIATE T H E S I S

Mechanical properties of flax fibers

and their composites

Edgars SpƗrniƼš Arial

Edgars SpƗrniƼš Times N R

Edgars SpœrniŸš Century

Luleå University of Technology

Department of Applied Physics and Mechanical Engineering
Division of Polymer Engineering

2006:60|: -1757|: -lic -- 06 ⁄60 -- 

Mechanical properties of flax fibers
and their composites
by

Edgars SpƗrniƼš

Division of Polymer Engineering

Department of Materials and Manufacturing Engineering
Luleå University of Technology
S-971 87 Luleå, Sweden

October 2006
 PREFACE

The work presented in this thesis concerns flax fibers as a potential

replacement of synthetic fibers in conventional polymer composites.
The thesis consists of a general introduction and literature review and
two journal papers.
Research nowadays often is a result of team work. Therefore there
is a couple of persons that I would like to acknowledge.
First, I thank my supervisors: Prof. JƗnis VƗrna, Dr. JƗnis Andersons,
Dr. Roberts Joffe and Prof. Vitauts Tamužs. I would like to thank my
co-author Lernart Wallström as well.
Further thanks go to Mr. Vilis Skruls and Mr. Uldis Vilks, research
engineers from Institute of Polymer Mechanics, Riga, Latvia. They
helped with experimental equipment setup for single fiber tensile
tests.
I also thank Dr. Harriëtte L. Bos, who allowed me to use her
illustration figure in this thesis.

Edgars SpƗrniƼš

i
ii
 SUMMARY

Flax fibers, along with a number of other natural fibers, are being
considered as an environmentally friendly alternative of synthetic
fibers in fiber-reinforced polymer composites. A common feature of
natural fibers is a much higher variability of mechanical properties.
This necessitates study of the flax fiber strength distribution and
efficient experimental methods for its determination.
Elementary flax fibers of different gauge lengths are tested by
single fiber tension in order to obtain the stress-strain response and
strength and failure strain distributions. The applicability of single
fiber fragmentation test for flax fiber failure strain and strength
characterization is considered. It is shown that fiber fragmentation
test can be used to determine the fiber length effect on mean fiber
strength and limit strain.
Stiffness and strength under uniaxial tension of flax fiber
composites with thermoset and thermoplastic polymer matrices are
considered. The applicability of rule of mixtures and orientational
averaging based models, developed for short fiber composites, to flax
reinforced polymers is evaluated.

iii
iv
 LIST OF PAPERS

This thesis comprises the following papers:

Paper A
Andersons J., SpƗrniƼš E., Joffe R., Wallström L. Strength distribution of
elementary flax fibres. Composites Science and Technology, 2005 65: p. 693-702.

Paper B
Andersons J., Joffe R., SpƗrniƼš E. Stiffness and strength of flax fiber/polymer
matrix composites. Polymer Composites, 2006 27(2): p. 221-229.

Content of both papers is reported in conferences:

Joffe R., Andersons J., SpƗrniƼš E., Wallström L. Flax fibres for structural
composites. Proceedings of 2nd International Conference on Eco-Composites
EcoComp 2003, 1-2 September 2003, Queen Mary, University of London, United
Kingdom. 10 p.

SpƗrniƼš E., Andersons J., Joffe R., Wallström L. Mechanical properties of

elementary flax fibers and flax-fiber composites. Thirteen International Conference
on Mechanics of Composite Materials, May 16-20, 2004, Riga, Latvia. Book of
Abstracts. p.181.

Joffe R., Andersons J., SpƗrniƼš E., Wallström L. Cellulose-Based Fibers and Their
Polymer Composites: Characterization and Prediction of Properties. Proceedings of
8th International Conference on Woodfiber-Plastic Composites (and other natural
fibers), May 23-25, 2005. Monona Terrace Community & Convention Center,
Madison, Wisconsin, USA. p. 25-36.

SpƗrniƼš E., Andersons J., Joffe R., Wallström L. Mechanical Properties of Flax
Fibres and Composites. Proceedings of International Conference on Structural
Analysis of Advanced Materials ICSAM 2005, 15-17 September 2005, University
“Politehnica” of Bucharest, Romania. p. 47-54.

The conference proceedings are not included in the thesis

v
vi
 TABLE OF CONTENTS

Introduction

1. Motivation of natural fiber applications in polymer composites .....1

2. State of the art ..................................................................................2
2.1. Some facts from history ............................................................2
2.2. Overview of cellulose-based natural fibers...............................2
Structure and Chemical composition ...........................................3
Characterization of mechanical properties...................................5
2.3. Natural fiber composites ...........................................................6
Materials ......................................................................................6
Manufacturing..............................................................................7
Adhesion ......................................................................................8
Performance and applications ......................................................8
Modeling ......................................................................................9
3. Current work ..................................................................................11
3.1. Objectives ...............................................................................11
3.2. Paper A....................................................................................11
3.3. Paper B....................................................................................12
3.4. Conclusions.............................................................................12
4. Future work....................................................................................13
5. References......................................................................................14

Paper A……………………………………………….………..…19
Paper B……………………………………………….…………..47

vii
viii
 Introduction

1. Motivation of natural fiber applications in polymer

composites

Due to the exponential growth of human population on Earth we

face environmental problems more and more. Now, in 21st century, it
is clear that we are paying for advanced technology with ecological
troubles and even disasters sometimes. And it is in our interests to
look for solution. Therefore, from the materials science point of view,
there is growing interest in green, environmentally friendly materials.
If we consider composites, one of solutions can be use of natural
fibers instead of more traditional glass and carbon fibers [1-4]. The
possible advantages of such natural fiber composites (NFC) could be
x Lower pollution level during production;
x Energy necessary for fiber production is lower than that of
glass;
x CO2 neutral: amount of CO2 neutralized during fiber plant
growth is comparable with that emitted during processing;
x Lower cost.
However, all these previous statements should be supported by
quantitative analysis.
x Natural fibers are renewable resources;
x They are biodegradable; however this can be a drawback
during the lifetime of a product;
x Using biodegradable polymers as matrix, we can have totally
recyclable materials.
Natural fibers in composites can compete with synthetic fibers by
x Lower density;
x Healthier in use due to their natural origins;
x Less abrasive to the processing equipment.
Low density is the main point why NFC is interesting in automotive
sector.
Anyway, beside these environmental advantages, we are about to
use natural fibers in composites. The stiffness and strength of fibers
are the basis for the reinforcement, but also the interfacial strength
(adhesion) is important for efficient reinforcement. Therefore these
three parameters are first to be determined and characterized. Let us
have a little look at literature dedicated to natural fibers and their
composites.

1
 Introduction

2. State of the art

2.1. Some facts from history

Natural organic fibers have been around for a very long time, from
the beginning of the life on Earth. The archeological artifacts suggest
that human beings used these materials in fabrics many thousand
years ago.
A direct use of the strength of natural fibers is in lines, ropes and
other one-dimensional products; miscellaneous applications include
early suspension bridges for on-foot passage of rivers and rigging for
naval ships in early times and into the nineteenth century. Many kinds
of textiles, ropes, canvas and paper produced form natural fibers are
in use today.
It may seem surprising, but first natural fiber composites were used
more than 100 years ago. In 1896, for example, airplane seats and
fuel-tanks were made of natural fibers with a small content of
polymeric binders. As early as 1908, the first composite materials
were applied for the fabrication of large quantities of sheets, tubes
and pipes for electronic purposes (paper or cotton to reinforce sheets,
made of phenol- or melamine-formaldehyde resins) [2]. However,
these attempts were without recognition of the composite principles
and the importance of fibers as the reinforcing part of composites.
The use of natural fibers was suspended due to low cost and
growing performance of technical plastics and, moreover, synthetic
fibers. A renaissance in the use of natural fibers as reinforcements in
technical applications began in 90s of 20th century.

2.2. Overview of cellulose-based natural fibers

Different types and examples of natural fibers classified according

to their origin are presented in Figure 1. Asbestos is out of further
consideration in this study due to its cancerogenic nature. In addition,
asbestos does not possess most of advantages mentioned in Section 1.

2
 Introduction

Natural Fibers

Animal Mineral

Wool Asbestos

Silk

Plant or Vegetable

Bast Leaf Seed Wood Grass stem

Flax Sisal Cotton Soft Reed

wood canary
Hemp Banana Coir grass
Hard
Jute Abaka Oil wood Cereal
palm straw
Ramie (rhy,
wheat)
Kapok
Kenaf

Figure 1. Classification of natural fibers according to origin together with several

examples.

Structure and Chemical composition

Generally, plant or vegetable fibers are used to reinforce plastics.
The main polymers involved in the composition of plant fibers are
cellulose, hemicelluloses, lignin and pectin.
Let us consider very popular flax fibers to understand the intricate
structure of plant fibers. The ~1 meter long so-called technical fibers
are isolated from the flax plant for the use in textile industry. These
technical fibers consist of elementary fibers with lengths generally
between 2 and 5 cm, and diameters between 10 and 25 Pm. The
elementary fibers are glued together by a pectin interface. They are
not circular but a polyhedron with 5 to 7 sides to improve the packing
in the technical fiber.

3
 Introduction

hackling

technical fibre elementary fibre,

Ø 50-100 µm plant cell
breaking
scutching Ø 10-20 µm

meso fibril
bast fibre Ø 0.1-0.3 µm
bundle

micro fibril
Ø 1-4 nm
flax stem
Ø 2-3 mm

Figure 2. Composition and built of flax stem. (Illustration by Harriëtte L. Bos.

Reproduced by permission)

What are elementary fibers? They are single plant cells. And
cellulose (C6H10O5)n is a common material in plant cell walls. It
occurs naturally in almost pure form in cotton fiber. Chemical
structure of cellulose monomer is represented in Figure 3. Most of the
elementary fiber consists of oriented, highly crystalline cellulose
fibrils and amorphous hemicellulose. The crystalline cellulose fibrils
in the cell wall are oriented at an angle of about ±10 degrees with the
fiber axis [3, 6] and give the fiber its high tensile strength.

CH2OH OH
O O
OH OH
O O
OH CH2OH
n

Figure 3. Chemical structure of cellulose monomer.

4
 Introduction

Characterization of mechanical properties

The natural organic fibers are basically characterized by the same
parameters and properties as all other fibers. However, due to natural
origin they show much higher variability of the various parameters
than their synthetic counterparts. Chemical composition, crystallinity,
surface properties, diameter, cross-sectional shape, length, strength,
and stiffness vary from fiber to fiber. Moreover, properties depend on
growing (climate), harvesting conditions and processing. This poses
two problems: quality characterization of fibers and difficulties in
application of traditional composite theories.
Usually the single fiber tensile (SFT) test method is used to
measure modulus and tensile strength of natural fibers. It is important
to mention that elementary (flax) fibers have considerably higher
strength than technical fibers [5]. Elementary fibers of flax [5-8],
nettle [6] and wheat straw [9] are tested at single gauge length. It is
verified that Weibull distribution is applicable to approximate
strength distribution of natural fibers [5, 6, 8]. Nonlinearity of stress-
strain response (strain-hardening) is reported [7, 9]. The increase of
the Young’s modulus with strain is explained with a reorganization of
the cellulose fibrils in the direction of the fiber (loading) axis. From
fatigue tests it is established that this effect is irreversible [7].
Another direct method, although less popular, is loop test. Using
this method it is possible to determine both tensile and compressive
strength of fibers. Analyzing fiber failure qualitatively by ESEM [10]
and quantitatively [5] it is obtained that compressive strength of the
flax fibers is about 80% of tensile strength.
Completely different approach is to back-calculate fiber properties
from unidirectional model composite tests. Anisotropy of jute is
studied using this method [11]. In this way it is calculated that the
transverse modulus of fibers is 5-10 times smaller than the
longitudinal modulus. Coefficient of thermal expansion is negative
along fiber, but in transverse direction – positive and comparable with
that of the polymer matrix.
In addition, the effect of heating [9, 12] and moisture [13] uptake of
natural fibers has been studied. It is concluded [12] that during
production of natural fiber reinforced plastics, only short periods of
exposure to high temperatures are allowed. Composite production
temperatures higher than 180ºC have to be avoided in order to
prevent degradation of fibers.

5
 Introduction

Unfortunately, in addition to advantages mentioned in Section 1, we

have several disadvantages of natural fibers:
x Large scatter of all the parameters;
x Properties depend on growing and processing conditions;
x Degradation of properties (moisture, heat, flame);
x Fibers are short; that means lower performance of their
composites;
x Structure is highly inhomogeneous;
x Stress – strain response is nonlinear.
Finally, it is seen from literature that flax fibers have the best
potential to substitute glass in polymer composites. A comparison of
main parameters for flax, hemp, jute and glass fiber is given in
Table 1. Specific properties of natural fibers, especially flax, are
promising.

Table 1. Typical values of E-glass and some natural fibers [1].

Modulus Strength Density Specific Specific
Fibers
(GPa) (MPa) (g/cm3) Modulus Strength
E-glass 72 3530 2.54 28.2 1390
Flax 50-70 500-900 1.4-1.5 ~ 41 ~ 480
Hemp 30-60 300-800 1.48 ~ 30 ~ 370
Jute 20-55 200-500 1.3-1.5 ~ 27 ~ 250

2.3. Natural fiber composites

Most of NFC are short fiber composites with non-homogeneous

length and orientation distributions. It is known that elastic modulus
and the strength of discontinuous fiber composites are moderate
compared to continuous fiber-reinforced composites. But there is
more flexibility in the selection of their fabrication methods. Many
processing methods used for plastics can be applied to the fabrication
of discontinuous fiber composites. This leads to the possibility of
easier mass production.

Materials
To process to composites, natural fibers are typically formed into
some form of fiber mat. Kenaf, hemp, sisal, coir, jute and some other

6
 Introduction

fiber composites have been studied. But the flax fibers appear to have
received more attention.
Although thermal instability of fibers causes restrictions for
matrices, both thermoplastics and thermosets are being used.
Polypropylene (PP, melting temperature 160ºC) is the most popular
thermoplastic matrix, also found to be the best [14] for flax fibers.
There is no distinct favorite among thermosets: epoxies, vinylesters,
polyesters, and other polymers are used.
There are also attempts to use biopolymers [15-21], such as
polyester amide, poly-L-lactic acid and others. The main attention in
this area is focused on manufacturing problems. The properties of
such biocomposites seem very promising too. However they are not
considered in the following review.

Manufacturing
Thermoplastic NFC are manufactured mainly by different extrusion
methods followed by injection [22] or compression molding [23, 24].
Fibers are chopped during process therefore composites have short
fibers (few millimeters at most). Orientation is three-dimensional, but
not necessarily isotropic (depends on method: injection process,
shape of mould etc.).
Besides, natural fiber mat thermoplastic composite plates with
different fiber contents can be manufactured using the film-stacking
method [25, 26]. Thermoplastic pultrusion can also be applied for
continuous process [27].
Compression molding processes is very typical for thermosets [28].
Resin transfer molding [29, 30] and resin infusion are used as well.
These methods ensure relatively longer fibers and more or less
inplane orientation in resulting composite. Since the thermal stability
of the flax fibers may be increased by chemical treatment, then even
autoclave molding technique can be applied [31].
A very important task of NFC manufacturing is to have elementary
fiber, not technical fiber, as the reinforcement in the composite [32].
On the other hand, using technical fibers or textile yarns, it is possible
make long fiber composites with predefined fiber orientation.
However in both cases the basic problem is fiber/matrix adhesion.

7
 Introduction

Adhesion
All the plant fibers are hydrophilic in nature. That is because of
their chemical structure – the hemicelluloses and the pectin are very
hydrophilic [1, 14]. In contrary, many of the common matrix
polymers in composites are largely hydrophobic in nature. Only
thermosets such as phenolformaldehyde and related polymers are less
hydrophobic and are therefore less problematic [1]. This discrepancy
can lead to the formation of ineffective interfaces between the fiber
and matrix. Problem can be solved applying different fiber treatments
(both chemical and mechanical) or modifying chemical composition
of the matrix. Unfortunately, surface treatments have a negative
impact on economical aspect of NFC manufacturing.
There are many publications reporting that properties of PP based
composites are improved using maleic anhydride grafted PP (MAPP)
as matrix additive [8, 22, 25, 32-37]. The same MAPP [38],
acetilation and stearic acid treatment [37] are used for fiber
processing. Adhesion increases also after boiling of the flax fibers
[35, 39].
For epoxy resin, alkali (NaOH), silane (3-aminopropyl-
triethoxysilane), isocyanate (phenyl isocyanate) [31], urea [40] and
other treatments can be employed. Acrylic acid and vinyl trimethoxy
silane of different concentrations are considered for several other
thermosets [8].
Single fiber fragmentation (SFF) test is a common method used to
measure adhesion quantitatively (via critical fiber length or interfacial
shear strength) for synthetic fibers. It is adapted for flax as well [8,
41].

Performance and applications

Usually NFC is compared with glass fiber reinforced polymers of
the same type (matrix; manufacturing method; fiber volume fraction,
length and configuration). Normally these are fiber mat composites.
Composite density must be taken into account, because it is the strong
side of NFC. In other words, specific properties must be compared.
This subject is studied in many papers. The basic results are that NFC
can have very good specific stiffness [24, 42] and reasonable specific
tensile strength [24, 42]. Unfortunately, the impact strength is a
disaster for NFC [24, 25, 43], however it can be upgraded improving

8
 Introduction

adhesion [44]. Flexural strength of NFC is also somewhat lower that

that of GFRP [32].
As expected, flax fiber composites are the best in properties (jute
takes second place). In short, it can be concluded that NFC based on
flax fibers can compete with E-glass based GFRP materials in
stiffness critical structures, whereas for strength and impact critical
applications these materials still need to be optimized further.
Several leading car manufacturers already use flax, banana fibers,
for instance as polymer reinforcement in interior parts (car roofs, door
panels and other). The use of flax fibers in car disk brakes instead of
asbestos fibers, is another example. There is a potential to develop
this tendency further [4, 28].

Modeling
Since we are at the beginning of understanding natural fiber
composites, the most important parameters, as usual, are stiffness and
strength.

Stiffness
A number of theoretical models have been elaborated to model
stiffness of short fiber reinforced composites [45, 46]. Although they
are very different, they all are based on the same basic assumptions:
x the fibers and the matrix are linearly elastic,
x the matrix is isotropic, and the fibers are either isotropic or
transversely isotropic.
x the fibers are axi-symmetric and identical in shape
x the fibers and matrix are well bonded at their interface, and
remain that way during deformation.
It is seen that none of them agrees perfectly with nature of natural
fibers. Let us keep it in mind.
Historically, shear lag models were the first micromechanics
models for short fiber composites. Although classical shear lag
models predict the longitudinal modulus only, they are very popular
due to their algebraic and physical simplicity. Usually they are
implemented by combining the average stress in the fiber with
average matrix stress to construct a modified rule of mixtures. It is
unusual that the fibers in a short fiber composite are arrayed
unidirectionally. Usually some distribution of fiber orientation exists.
Therefore we obtain

9
 Introduction

E KlEKoE E f V f  Em 1  V f ,

where KlE is fiber length efficiency factor, and KoE – orientation

efficiency factor for Young’s modulus calculation. They both can be
calculated having fiber length and orientation distributions.
Another classics are Halpin and Tsai relations:

P 1  9 ˜KV f Pf Pm  1
where K
Pm 1  KV f Pf Pm  1

Initially they were developed for continuous fiber composites.

However they are efficient for short fiber composites as well. Halpin
and Tsai suggested that parameter 9 is related to geometry of
reinforcement. For longitudinal modulus they found it expressed
through aspect ration of fibers 9 2 l d for instance.
However, there is a need to take both fiber length and orientation
distributions into account. It is possible, using laminate analogy
approach [47]. In this method composite stiffness is calculated
“summing” layers with the same fiber length and orientation
according to distribution functions.
Another approach is to use unit cells (matrix bricks with one fiber
inside) in combination with finite element method [48]. Unfortunately
these models include some parameters that are missing for natural
fibers at the moment (fiber Poisson’s ratio and fiber shear modulus,
for instance). This is common problem for other models also [49].
Therefore they are not applicable in this study.

Strength
In comparison to the stiffness, strength theory for short fiber
composites is still under development. Most of the strength models
are also based on the rule of mixtures [50-52] or equivalent laminate
[53, 54] approach.
So in simplest case (considering rule of mixtures), misaligned short
fiber composite strength

V uc KlsKosV uf V f  1  V f V m ,

10
 Introduction

where V uf is fiber strength, but V m – stress in the matrix at the fiber

failure strain. Kls and Kos are fiber length efficiency factor and fiber
orientation efficiency factor for strength calculation respectively. For
all the models, length efficiency factor Kls is related to critical or
ineffective fiber length.
Recently, more complicated models have been advanced [48, 55,
56] that employ numerical modeling of deformation and failure
processes.

3. Current work
3.1. Objectives

As it follows from Section 2, simple and efficient quality control

method for natural fibers is needed. Single fiber tensile test is very
time consuming and tiresome. Single fiber fragmentation test has
been used as an efficient alternative to SFT test for Weibull
distribution parameter determination of synthetic fibers, while for flax
fibers the test is mainly applied for the assessment of fiber/matrix
adhesion. The principal objective of Paper A is to evaluate the
applicability of SFF test for elementary flax fiber strength and failure
strain characterization.
The aim of the Paper B is a systematic evaluation of the
applicability of elementary mechanical models for flax fiber
reinforced polymer matrix composite modulus and strength
prediction.

3.2. Paper A

Elementary flax fiber strength and failure strain distributions at

several gauge lengths are obtained by SFT tests.
It is found that the modified Weibull distribution [57, 58]

ª § l ·J D
§V · º
P V 1  exp «  ¨ ¸ ¨ ¸ »
«¬ © l0 ¹ © E ¹ »¼

11
 Introduction

is preferable for flax fibers considered.

The applicability of single fiber fragmentation test for flax fiber
failure strain and strength characterization is considered. It is shown
that SFF test can be used to determine the fiber length effect on mean
fiber strength and limit strain.

3.3. Paper B

Stiffness of flax fiber composites was modeled using Cox-Krenchel

model. Strength was modeled by two methods: modified Fukuda-
Chou and orientational averaging model for strength proposed by van
Hattum and Bernardo [54]. To verify obtained results, the following
types of flax fiber composites were made.
Flax fiber mats (FFM) produced by FinFlax were used as
reinforcement for polymer composites. The resins used were two
types of vinylesters, modified acrylic resin, polypropylene and maleic
anhydride grafted PP modified polypropylene.
Plates of thermoset natural fiber composites were manufactured by
resin transfer molding, while thermoplastic composite specimens
were produced from compound obtained by co-extrusion of
granulated PP and flax fiber rowing.
Correlation of experimental and theoretical composite
characteristics revealed the degree of sensitivity of the models to
properties of the constituents and their adhesion.

3.4. Conclusions

Strength and failure strain possess the modified Weibull distribution

(i.e. independent exponents characterize strength scatter at a fixed
gauge length and the dependence of mean strength on fiber length).
SFF tests reveal that fiber fragmentation process proceeds in
agreement with the two-parameter Weibull distribution of failure
strain, but the distribution parameters for individual fibers exhibit a
marked scatter. A limited number of SFF tests combined with SFT
tests at a fixed gauge length provide the experimental information
necessary for characterizing both the scatter of fiber failure strain and
strength at a fixed gauge length and the effect of fiber length on the
mentioned parameters.

12
 Introduction

Stiffness and strength under uniaxial tension has been obtained for
flax/PP and flax/PPM composites produced from compound obtained
by co-extrusion of granulated PP and flax, as well as for
FFM/vinylester and FFM/acrylic resin composites manufactured by
resin transfer molding. The rule-of-mixtures relations are shown to
yield acceptable stiffness prediction of both extruded and FFM-based
composites. The sensitivity of a rule-of-mixtures model of strength to
the matrix and adhesion properties apparently depends on the relative
ineffective fiber length; for relatively long-fiber FFM composites the
sensitivity is low and more sophisticated strength models should be
applied.

4. Future work

In order to continue the work described in Paper A, correlation

between parameters of Weibull strength distribution and structure or
morphology of fibers will be sought. This is the subject of a
forthcoming paper.
However, the main goal is to have high performance natural fiber
composites that can be used in load bearing constructions. The ideal
solution is perfectly oriented elementary fibers in the matrix.
Unfortunately it is not possible at the moment. Bet we can improve
quality of technical fibers. There are some good examples already.
Unidirectional flax fiber composites with remarkable properties are
obtained using textile monofilament flax yarn [26]. Could the next
step be cross-ply?
One of possible applications of such high performance natural fiber
composites could be in building industry. It is a relatively new idea to
strengthen and repair concrete or masonry structures with carbon
fiber reinforced polymers. In fact, the only important properties of
composite in such structures are stiffness and strength. Flax fibers
could be better in sound and thermo-insulation. Since possible
amounts of used fibers are very large, the economical effect could be
very great.
This was just one example. There is a great future of high-
performance natural fiber composites in structural parts.

13
 Introduction

5. References

1. Lilholt H. and Lawther J.M. Natural organic fibres. In: Comprehensive

composite materials, editors: Kelly A. and Zweben C. Vol. 1. New York:
Pergamon Press; 2000.
2. Bledzki A.K. and Gassan J. Composites reinforced with cellulose based
fibres. Progress in Polymer Science, 1999. 24: p. 221-274.
3. Joshi S.V., Drzal L.T., Mohanty A.K. and Arora S. Are natural fiber
composites environmentally superior to glass fiber reinforced composites?
Composites: Part A, 2004. 35: p. 371-376.
4. De Bruijn J.C.M. Natural Fibre Mat Thermoplastic Products from a
Processor's Point of View. Applied Composite Materials, 2000. 7: p. 415-
420.
5. Bos H.L., van den Oever M.J.A. and Peters O.C.J.J. Tensile and
compressive properties of flax fibres for natural fibre reinforced
composites. Journal of Materials Science, 2002. 37: p. 1683-1692.
6. Davies G.C. and Bruce D.M. Effect of Environmental Relative Humidity
and Damage on the Tensile Properties of Falx and Nettle Fibers. Textile
Research Journal, 1998. 68(9): p. 623-629.
7. Baley C. Analysis of the flax fibres tensile behaviour and analysis of the
tensile stiffness increase. Composites: Part A, 2002. 33: p. 939-948.
8. Joffe R., Andersons J. and Wallström L. Strength and adhesion
characteristics of elementary flax fibres with different surface treatments.
Composites: Part A, 2003. 34: p. 603-612.
9. Hornsby P.R., Hinrichsen E. and Tarverdi K. Preparation and properties of
polypropylene composites reinforced with wheat and flax straw fibres. Part
I Fibre characterization. Journal of Materials Science, 1997. 32: p. 443-
449.
10. Bos H.L. and Donald A.M. In situ ESEM study of the deformation of
elementary flax fibres. Journal of Materials Science, 1999. 34: p. 3029-
3034.
11. Cichocki Jr. F.R. and Thomason J.L. Thermoelastic anisotropy of a natural
fiber. Composites Science and Technology, 2002. 62: p. 669-678.
12. Van de Velde K. and Beatens E. Thermal and Mechanical Properties of
Flax Fibres as Potential Composite Reinforcement. Macromolecular
Materials and Engineering, 2001. 286(6): p. 342-349.
13. Stamboulis A., Baillie C.A. and Peijs T. Effects of environmental
conditions on mechanical and physical properties of flax fibers.
Composites: Part A, 2001. 32: p. 1105-1115.
14. Van de Velde K. and Kiekens P. Thermoplastic polymers: overview of
several properties and their consequences in flax fibre reinforced
composites. Polymer Testing, 2001. 20: p. 885-893.
15. Jiang L. and Hinrichsen G. Flax and cotton fiber reinforced biodegradable
polyester amide composites, 1. Die Angewandte Makromolekulare
Chemie, 1999. 268(4649): p. 13-17.

14
 Introduction

16. Jiang L. and Hinrichsen G. Flax and cotton fiber reinforced biodegradable
polyester amide composites, 2. Die Angewandte Makromolekulare
Chemie, 1999. 268(4650): p. 18-21.
17. Williams G.I. and Wool R.P. Composites from Natural Fibers and Soy Oil
Resins. Applied Composite Materials, 2000. 7: p. 421-432.
18. Nishino T., Hirao K., Kotera M., Nakamae K. and Inagaki H. Kenaf
reinforced biodegradable composite. Composites Science and Technology,
2003. 63: p. 1281-1286.
19. Oksman K., Skrifvars M. and Selin J.-F. Natural fibres as reinforcement in
polylactic acid (PLA) composites. Composites Science and Technology,
2003. 63: p. 1317-1324.
20. Keller A. Compounding and mechanical properties of biodegradable hemp
fibre composites. Composites Science and Technology, 2003. 63: p. 1307-
1316.
21. Plackett D., Andersen T.L., Pedersen W.B. and Nielsen L. Biodegradable
composites based on L-polylactide and jute fibres. Composites Science and
Technology, 2003. 63: p. 1287-1296.
22. Karmaker A.C. and Youngquist J.A. Injection Molding of Polypropylene
Reinforced with Short Jute Fibers. Journal of Applied Polymer Science,
1996. 62: p. 1147-1151.
23. Mohanty A.K., Wibowo A., Misra M. and Drzal L.T. Effect of process
engineering on the performance of natural fiber reinforced cellulose
acetate biocomposites. Composites: Part A, 2004. 35: p. 363-370.
24. Oksman K. Mechanical Properties of Natural Fibre Mat Reinforced
Thermoplastic. Applied Composite Materials, 2000. 7: p. 403-414.
25. Garkhail S.K., Heijenrath R.W.H. and Peijs T. Mechanical Properties of
Natural-Fibre-Mat-Reinforced Thermoplastics based on Flax Fibres and
Polypropylene. Applied Composite Materials, 2000. 7: p. 351-372.
26. Madsen B. and Lilholt H. Physical and mechanical properties of
unidirectional plant fibre composites - an evaluation of the influence of
porosity. Composites Science and Technology, 2003. 63: p. 1265-1272.
27. Van de Velde K. and Kiekens P. Thermoplastic pultrusion of natural fibre
reinforced composites. Composite Structures, 2001. 54: p. 355-360.
28. Magurno A. Vegetable fibres in automotive interior components. Die
Angewandte Makromolekulare Chemie, 1999. 272(4751): p. 99-107.
29. Rouison D., Sain M. and Couturier M. Resin transfer molding of natural
fiber reinforced composites: cure simulation. Composites Science and
Technology, 2004. 64: p. 629-644.
30. Sèbe G., Cetin N.S., Hill C.A.S. and Hughes M. RTM Hemp Fibre-
Reinforced Polyester Composites. Applied Composite Materials, 2000. 7:
p. 341-349.
31. George J., Ivens J. and Verpoest I. Mechanical properties of flax fibre
reinforced epoxy composites. Die Angewandte Makromolekulare Chemie,
1999. 272(4747): p. 41-45.
32. van den Oever M.J.A., Bos H.L. and van Kemenade M.J.J.M. Influence of
the Physical Structure of Flax Fibres on the Mechanical Properties of Flax
Fibre Reinforced Polypropylene Composites. Applied Composite
Materials, 2000. 7: p. 387-402.

15
 Introduction

33. Van de Velde K. and Kiekens P. Influence of Fibre and Matrix

Modifications on Mechanical and Physical Properties of Flax Fibre
Reinforced Poly(propylene). Macromolecular Materials and Engineering,
2001. 286(4): p. 237-242.
34. Van de Velde K. and Kiekens P. Development of a Flax/Polypropylene
Composite with Optimal Mechanical Characteristics by Fiber and Matrix
Modification. Journal of Thermoplastic Composite Materials, 2002. 15: p.
281-300.
35. Van de Velde K. and Kiekens P. Effect of material and process parameters
on the mechanical properties of unidirectional and multidirectional
flax/polypropylene composites. Composite Structures, 2003. 62: p. 443-
448.
36. van den Oever M.J.A. and Bos H.L. Critical Fibre Length and Apparent
Interfacial Shear Strength of Single Flax Fibre Polypropylene Composites.
Advanced Composites Letters, 1998. 7(3): p. 81-85.
37. Zafeiropoulos N.E., Williams D.R., Baillie C.A. and Matthews F.L.
Engineering and characterisation of the interface in flax
fibre/polypropylene composite materials. Part I. Development and
investigation of surface treatments. Composites: Part A, 2002. 33: p. 1083-
1093.
38. Cantero G., Arbelaiz A., Llano-Ponte R. and Mondragon I. Effects of fibre
treatment on wettability and mechanical behaviour of flax/polypropylene
composites. Composites Science and Technology, 2003. 63: p. 1247-1254.
39. Van de Velde K. and Kiekens P. Effect of Flax/PP Panel Process
Parameters on Resulting Composite Properties. Journal of Thermoplastic
Composite Materials, 2003. 16: p. 413-431.
40. Hepworth D.G., Vincent J.F.V., Jeronimidis G. and Bruce D.M. The
penetration of epoxy resin into plant fibre cell walls increases the stiffness
of plant fibre composites. Composites: Part A, 2000. 31: p. 599-601.
41. Zafeiropoulos N.E., Baillie C.A. and Hodgkinson J.M. Engineering and
characterisation of the interface in flax fibre/polypropylene composite
materials. Part II. The effect of surface treatments on the interface.
Composites: Part A, 2002. 33: p. 1185-1190.
42. Wambua P., Ivens J. and Verpoest I. Natural fibres: can they replace glass
in fibre reinforced plastics? Composites Science and Technology, 2003.
63: p. 1259-1264.
43. van den Oever M.J.A., Bos H.L. and Molenveld K. Flax fibre physical
structure and its effect on composite properties: Impact strength and
thermo-mechanical properties. Die Angewandte Makromolekulare Chemie,
1999. 272(4760): p. 71-76.
44. Gassan J. and Bledzki A.K. Possibilities to Improve the Properties of
Natural Fiber Reinforced Plastics by Fiber Modification - Jute
Polypropylene Composites -. Applied Composite Materials, 2000. 7: p.
373-385.
45. Tucker III C.L. and Liang E. Stiffness predictions for unidirectional short-
fiber composites: Review and evaluation. Composites Science and
Technology, 1999. 59: p. 655-671.

16
 Introduction

46. Fukuda H. and Takao Y., 1.13 Thermoelastic Properties of Discontinuous

Fiber Composites, in Comprehensive Composite Materials. 2000. p. 377-
401.
47. Fu S.-Y. and Lauke B. The elastic modulus of misaligned short-fiber-
reinforced polymers. Composites Science and Technology, 1998. 58: p.
389-400.
48. Dong M., Schmauder S., Bidlingmaier T. and Wanner A. Prediction of the
mechanical behaviour of short fiber reinforced MMCs by combined cell
models. Computational Materials Science, 1997. 9: p. 121-133.
49. Meddad A., Azaiez J., Ait-Kadi A. and Guénette R. Micromechanical
Modeling of Tensile Behavior of Short Fiber Composites. Journal of
Composite Materials, 2002. 36(04): p. 423-441.
50. Kelly A. and Tyson W.R. Tensile properties of fibre-reinforced metals:
copper/tungsten and copper/molybdenum. Journal of the Mechanics and
Physics of Solids, 1965. 13: p. 329-350.
51. Fu S.-Y. and Lauke B. Effects of fiber length and fiber orientation
distributions on the tensile strength of short-fiber-reinforced polymers.
Composites Science and Technology, 1996. 56: p. 1179-1190.
52. Lauke B. and Fu S.-Y. Strength anisotropy of misaligned short-fibre-
reinforced polymers. Composites Science and Technology, 1999. 59: p.
699-708.
53. Halpin J.C. and Kardos J.L. Strength of Dicontinuous Reinforced
Composites: I. Fiber Reinforced Composites. Polymer Engineering and
Science, 1978. 18(6): p. 496-504.
54. van Hattum F.W.J. and Bernardo C.A. A Model to Predict The Strength of
Short Fiber Composites. Polymer Composites, 1999. 20(4): p. 524-533.
55. Zhu K. and Schmauder S. Prediction of the failure properties of short fiber
reinforced composites with metal and polymer matrix. Computational
Materials Science, 2003. 28: p. 743-748.
56. Nguyen B.N. and Khaleel M.A. A mechanistic approach to damage in
short-fiber composites based on micromechanical and continuum damage
mechanics descriptions. Composites Science and Technology, 2004. 64: p.
607-617.
57. GutƗns J.A. and Tamužs V.P. Scale effect of the Weibull distribution of
fibre strength. Mechanics of Composite Materials, 1984. 20(6): p. 1107-
1109 (in Russian).
58. Watson A.S. and Smith R.L. An examination of statistical theories for
fibrous materials in the light of experimental data. Journal of Materials
Science, 1985. 20: p. 3260-3270.

17
 Introduction

18
 Paper A

Strength distribution of elementary flax fibres

J. Andersons1, E. SpƗrniƼš1, R. Joffe2, L. Wallström2
1
Institute of Polymer Mechanics, University of Latvia, Aizkraukles iela 23, RƯga LV-1006, Latvia
2
Division of Polymer Engineering, Luleå University of Technology, SE-971 87 Luleå, Sweden

Composites Science and Technology 65 (2005) 693-702

19
20
 Composites Science and Technology 65 (2005) 693-702

Strength distribution of elementary flax fibres

J. Andersons1, E. SpƗrniƼš1, R. Joffe2, L. Wallström2
1
Institute of Polymer Mechanics, University of Latvia, Aizkraukles iela 23, RƯga LV-1006, Latvia
2
Division of Polymer Engineering, Luleå University of Technology, SE-971 87 Luleå, Sweden

Abstract. Flax fibres, along with a number of other natural fibres, are being
considered as an environmentally friendly alternative of synthetic fibres in fibre-
reinforced polymer composites. A common feature of natural fibres is a much
higher variability of mechanical properties. This necessitates study of the flax fibre
strength distribution and efficient experimental methods for its determination.
Elementary flax fibres of different gauge lengths are tested by single fibre tension
in order to obtain the stress-strain response and strength and failure strain
distributions. The applicability of single fibre fragmentation test for flax fibre
failure strain and strength characterization is considered. It is shown that fibre
fragmentation test can be used to determine the fibre length effect on mean fibre
strength and limit strain.

1. Introduction

Ecological concerns have resulted in a renewed interest in natural

materials, and such issues as recyclability and environmental safety
have become increasingly important for the introduction of new
materials and products. Structural polymer composites are
traditionally utilizing man-made fibres (such as glass or carbon
fibres) as reinforcement, but environmental issues have generated a
considerable interest in natural fibres. Plant fibres such as flax, hemp,
sisal and kenaf are under consideration as environmentally friendly
and relatively low-cost alternatives for glass fibres in structural
engineering composites [1-3]. However, their use currently is mainly
confined to reinforcement in compounded thermoplastic products
similar to glass-mat-reinforced thermoplastics. These natural fibre
mat thermoplastics are used in the automotive sector, in interior and
exterior components [4, 5].
Flax fibres used as reinforcement are located in the bast of the flax
plant. The fibres are typically extracted by retting followed by
mechanical processing (scutching and hackling). The mechanical
properties of the obtained fibres are affected by the natural variability
in plant and by the processing stage and damage sustained during

21
 Paper A

processing [6-8] and thus have considerable scatter. Therefore a

simple and efficient method for fibre strength evaluation is needed.
The mechanical tests of elementary flax fibres [6, 8-10] revealed that
fibre strength is reasonably well approximated by the two-parameter
Weibull distribution. Single fibre fragmentation (SFF) test has been
used as an efficient alternative to single fibre tensile (SFT) test for
Weibull distribution parameter determination of synthetic fibres,
while for flax fibres the test is mainly applied for the assessment of
fibre/matrix adhesion [10-13]. The principal objective of the present
study is to evaluate the applicability of SFF test for elementary flax
fibre strength and limit strain characterization.

2. Experimental
2.1. Materials

Flax fibres delivered by FinFlax Oy (Finland) were used. The fibres

had been enzyme retted for 22h. The fibres were stored and tested at
ambient conditions. The resins used in single fibre composite (SFC)
specimens were vinyl ester (VE) and unsaturated polyester (UP).

2.2. Single fibre tensile tests

2.2.1. Specimen preparation

Single fibres were carefully manually separated from the bundles.

Fibre ends were glued onto a paper frame according to the
preparation procedure described in ASTM D 3379-75 Standard.
During mounting the specimens were handled only by the paper
frame. Fibre length outside the frame (gauge length) was 5, 10, or 20
mm. Upon clamping of the ends of the paper frame by the grips of the
test machine, frame sides were carefully cut in the middle.

2.2.2. Test setup

The tests were carried out on a electromechanical tensile machine

equipped with mechanical grips. Load-displacement curve was
recorded during the test. Upper grip of the machine was attached
through a hinge and thus allowed to self-align. All tests were
displacement controlled with the loading rate of 0.5 mm/min.

22
 Composites Science and Technology 65 (2005) 693-702

Since the fibres were not pre-stretched before the test, there was an
initial displacement before load was actually applied to the fibre. The
amount of this displacement was defined as an interval from the
beginning of the test until the point at which load increase is
observed. It was discounted later on during data processing.

Fig. 1. Polygonal shape of the flax fibre.

Although flax fibre cross-section has a polygonal shape (see Fig. 1)

and fibre thickness varies somewhat along the fibre [14], we treated
each fibre as perfectly round and having a constant diameter in order
to simplify analysis. Fibre diameter was evaluated from optical
observations under microscope as the average of five apparent
diameter measurements taken at different locations along the fibre.

2.3. Fibre fragmentation tests

2.3.1. Specimen preparation

Although single fibre fragmentation test is not strictly standardized,

there are certain techniques that are established and widely used on
synthetic fibres. The techniques may vary depending on particular
experimental setup, but generally they include the following steps:
single fibre separation from bundle, fibre pre-stretching (only to
straighten the filament), mounting of the fibre on the frame and
casting of the resin in the mould with the frame (see e.g. [15]). Our
experimental procedure applied for synthetic fibres [16, 17] includes
strain measurement by extensometer, therefore certain length of the
samples is required. Hence specimen preparation procedure for the
short fibres had to be modified in order to obtain specimens with
length that would allow use of extensometer. The short flax fibres
were modified by attaching fiber extensions on both ends of the

23
 Paper A

filament, see Fig. 2. Thin fishing line of 90 Pm diameter was used as

fibre extensions. Once “extended” fibre is obtained it can be handled
as any other long fibre. Further specimen preparation is according to
the procedure described above (it also includes slight fibre pre-
stretching in order to straighten the filament).
25-35 mm

Flax fiber
50-150 mm
Fiber extension

Fast drying glue

Fig. 2. Modification of flax fibre.

25-30 mm
2 mm
Side view

100-110 mm
3-4 mm
Front view

Clamping area (end tabs) Fiber Matrix

Fig. 3. SFC specimen.

SFC specimens were prepared by mounting the extended fibres on a

1 mm thick steel frame, using double-sided adhesive tape. The frame
was then placed between two flat Teflon-coated aluminum mould
plates, separated by spacers of 2 mm thickness and provided with a
silicon tube sealing. Mylar film was attached to the mould in order to
obtain smooth and transparent surface. After the resin had solidified,
the mould was placed in the oven for post-curing. VE was post-cured
at 50ºC for 2h and 80ºC for 5h, UP was post-cured at 50ºC for 2h.
The plates were cut and polished into specimens with the dimensions

24
 Composites Science and Technology 65 (2005) 693-702

shown in Fig. 3. Visual inspection revealed that the SFC specimens

thus obtained were of uniform quality, with well-aligned fibres and
no discernible bubbles or voids.
The effect of fibre extensions on the stress state in a SFC and on
SFF test results is discussed in Appendix.

2.3.2. Test setup

The SFF test was performed in a MINIMAT miniature mechanical

test machine from Polymer Laboratories Ltd (UK). The test machine
was mounted on the x - y table of a Zeiss optical microscope. Load
was measured by the MINIMAT’s built-in load cell (1000N) and the
displacement was registered by the electronic unit of the tensile stage.
Fragmentation of the fibres was observed during the loading.
Extensometer was used to measure applied strain. In order to avoid
pausing the machine to count fibre cracks, the test was carried out at
rather low loading rate of 0.1 mm/min. Loading was stopped if the
specimen failed, or when the fragmentation saturation level was
achieved, namely, no new fibre breaks appeared with strain increase
by 0.5%. During the experiment the data were transferred to the PC.
In order to measure the fibre diameter, digital pictures of the fibres
were made before the loading. Images were made by the CCD camera
attached to the microscope and then transferred to the PC for further
processing. As in the case of single fibre tests fibre diameter was
evaluated from analysis of digital images as the average of five
apparent diameter measurements taken along the fibre.

3. Single fiber tensile test results

3.1. Stress-strain response

Not only limit stress and strain, but also the actual shape of the
stress-strain curve was found to vary among fibres, ranging from
linear elastic to markedly strain-hardening. (Note that similar
variability in mechanical response was also observed for hemp fibres
[18]). A typical stress-strain diagram of an elementary flax fibre is
shown in Fig. 4. The apparent variation of tangent modulus with
strain confined mostly to the initial, small strain part of the diagram
(reported for flax fibres in e.g. [14, 19]) is attributed to the orientation

25
 Paper A

stress

Hn
strain
Fig. 4. Typical stress-strain curve of flax fibre.

of the fibrils along the axis of the fibre under load. (This phenomenon
is irreversible in that upon unloading, subsequent reloading is linear
elastic up to the previously achieved stress level with the modulus
equal to the maximum modulus achieved during the previous load
cycle [14].) At larger load/strain values, fibre response becomes
linear, and we use the linear part of diagram for Young’s modulus
calculation. There is a marked scatter of the measured modulus, E,
values as seen in Fig. 5. The probabilities here and in the following
are estimated as median ranks assigned to the measured modulus
values at each gauge length using the following approximation

i  0.3
P
m  0.4

where i is the i -th number in ascendingly ordered modulus data and

m is the sample size (i.e. number of modulus measurements
performed at the given gauge length). The average value and standard
deviation of the modulus and non-linear strain increment H n are
provided in Table 1. In view of the pronounced scatter, these results
are in rough agreement with the estimated modulus value of about 80
GPa based on UD composite tests [20, 21] for ArcticFlax fibres
produced by FinFlax.

26
 Composites Science and Technology 65 (2005) 693-702

1.0

0.8

0.6
P

0.4

0.2
10 mm
20 mm
0.0
0 20 40 60 80 100 120 140

E, GPa
Fig. 5. Young’s modulus distribution.

150
10 mm
20 mm

100
E, GPa

50

0
0 10 20 30 40

d, Pm
Fig. 6. Flax fibre modulus as a function of fibre diameter.

27
 Paper A

Table 1. Flax fibre deformation characteristics

Fibre Average Standard Average Standard
length, modulus deviation non-linear deviation
mm E , GPa sE , strain s Hn ,
GPa increment %
Hn , %
10 69 20 0.32 0.42
20 64 21 0.41 0.49

Fibre modulus slightly decreases with diameter increase as noted in

[10, 14, 21], but the effect is small and largely overshadowed by the
scatter, Fig. 6.

3.2. Strength distribution

The experimental fibre strength distribution yielded by SFT tests is

shown in Weibull coordinates in Fig. 7. The Weibull distribution

ª l DV º
§ V ·
PV 1  exp « ¨¨ ¸¸ » (1)
«¬ l 0 © EV ¹ »¼

parameters from SFT tests were determined by the maximum

likelihood method (MLM). In Eq. (1), DV is the shape parameter, EV
– scale parameter of the Weibull distribution, l designates fibre
length and l0 – reference length. The parameter values at the gauge
lengths tested are summarized in Table 2 (here in Eq. (1) and below
the reference length is chosen as l0 1mm ). It is seen that the two-
parameter Weibull distribution approximates the experimental data at
each of the gauge lengths reasonably well. Deviations from Weibull
distribution are mostly confined to the weakest fibres and can be
tentatively related to the damage done to the fibres in the specimen
preparation process.

28
 Composites Science and Technology 65 (2005) 693-702

2 l = 5 mm

0
ln(-ln(1-P))

-2

-4

300 500 700 1000 1500 2000

a) V, MPa

2 l = 10 mm

0
ln(-ln(1-P))

-2

-4

300 500 700 1000 1500

b) V, MPa

29
 Paper A

2 l = 20 mm

0
ln(-ln(1-P))

-2

-4

200 300 500 700 1000 1500 2000

c) V, MPa
Fig. 7. Strength distribution of flax fibres at 5 (a), 10 (b), and 20 mm (c) gauge
length. Solid lines – MLM approximation by two-parameter Weibull distribution
Eq. (1), dashed lines – by modified Weibull distribution Eq. (3).

Table 2. Weibull distribution parameters of flax fibre strength obtained by SFT tests
Data Gauge Number of Shape Scale
reduction length, specimens parameter parameter
method mm DV EV , MPa
MLM 5 90 2.9 1870
10 70 3.0 2080
20 58 2.5 2800
Eq. (2) 5…20 - 5.2 1430

The dependence of the average fibre strength on the gauge length is

plotted in Fig. 8 in logarithmic scale. The error bars correspond to the
average strength standard deviation estimated as sV m , where sV
and m are standard deviation of the fibre strength and the number of

30
 Composites Science and Technology 65 (2005) 693-702

1100

1000

900
V!, MPa

800

700

600

5 10 20

l, mm
Fig. 8. Average fibre strength as a function of gauge length obtained by SFT tests.
Solid line - approximation by Eq.(2), dashed line - prediction by SFF tests, Eq.(17).

2000
10 mm
20 mm

1500
V, MPa

1000

500

0
0 50 100 150

E, GPa
Fig. 9. Flax fibre strength as a function of fibre modulus.

31
 Paper A

tests at the corresponding gauge length. It follows from Eq. (1) that
the average strength is a power function of gauge length

1 DV
V E V l l0 * 1  1 DV (2)

By approximating the data with Eq. (2) as shown in Fig. 8,

estimates of the Weibull distribution parameters DV 5.2 and
EV 1430MPa are obtained. The Weibull shape parameter derived
applying Eq. (2) is considerably higher than that obtained from SFT
at a fixed length and amounting to ~2.9 for the gauge length interval
considered. This discrepancy in parameter values characterizing the
fibre strength scatter at a fixed gauge length and the average strength
dependenceon the fibre length suggests that the modified Weibull
distribution [22, 23]

§ § l · J V ª V º DV ·
PV 1  exp¨  ¨¨ ¸¸ « » ¸ (3)
¨ © l0 ¹ ¬ E V ¼ ¸
© ¹

is preferable to (1) for the flax fibres considered. The distribution (3)
reconciles the mismatch of the fibre strength scatter at a fixed gauge
length (characterized by DV ) and the average strength dependence on
the fibre length (governed by the exponent, J V DV )

J V DV
V E V l l0 * 1  1 DV (4)

It is suggested in [23, 24, 17] that the distribution (3) reflects the
fibre-to-fibre strength parameter variation in a batch of fibres, each of
which has the Weibull two-parameter strength distribution (1). The
parameters of the distribution (3) determined by MLM from strength
data in 5 to 20 mm gauge length interval are as follows: J V 0.46 ,
DV 2.8 , EV 1400MPa . The slope and location of the
corresponding plots in Weibull coordinates, Fig. 7, are in good
agreement with the SFT test data. Consequently, the modified
Weibull distribution adequately describes both gauge length
dependence of strength and the strength distribution at a fixed gauge
length for the flax fibres considered.

32
 Composites Science and Technology 65 (2005) 693-702

Both axial stiffness and strength of flax fibres is imparted by the

cellulose microfibrils, therefore it appears plausible that these
quantities are correlated. Fig. 9 shows that a weak correlation is
indeed present.

3.3. Failure strain distribution

Failure strain distributions were determined only for 10 and 20 mm

flax fibres. The corresponding Weibull plots are given in Fig. 10. The
parameter values of the two-parameter Weibull distribution for failure
strain

ª l DH º
§ H ·
PH 1  exp « ¨¨ ¸¸ » (5)
«¬ l 0 © EH ¹ »¼

obtained by MLM are provided in Table 3.

For linear elastic fibres with negligible modulus scatter, the relation
between failure strain and strength distributions is trivial – both share
the same shape parameter DV DH , and the scale parameter of the
strength distribution is given by the product of fibre modulus and the
corresponding strain distribution parameter, E V EE H . Flax fibre
stress-strain curve obtained by SFT typically possesses an initial non-
linear zone as shown schematically in Fig. 4. Hence, flax fibre
strength is related to its modulus, failure strain, and the non-linear
strain increment, H n , as follows:

V E H Hn (6)

Table 3. Weibull distribution parameters for failure strain of flax fibres

Test Gauge Number of Shape Scale
method length, specimens parameter parameter
mm DH EH, %
SFT 10 68 2.7 4.15
20 56 2.8 4.79
SFF ~24 21 6.9 2.42

33
 Paper A

2 l = 10 mm

0
ln(-ln(1-P))

-2

-4

0.7 1 2 3 4

a) H, %

2 l = 20 mm

0
ln(-ln(1-P))

-2

-4

0.4 0.7 1 2 3

b) H, %
Fig. 10. Failure strain distribution of flax fibres at 10 (a) and 20 mm (b) gauge
length. Solid lines – MLM approximation by two-parameter Weibull distribution
Eq. (5).

34
 Composites Science and Technology 65 (2005) 693-702

Note that all the fibre parameters entering Eq. (6) vary considerably
between fibres. However, both failure strain and non-linear strain
increment are virtually uncorrelated with the fibre modulus.
Therefore, Eq. (6) remains valid also for average values of the
parameters:

V E H  Hn (7)

Thus Eq. (7) relates mean failure strain and fibre strength;
accounting for mean stress expression Eq. (4), failure strain as a
function of fibre length is given by

J V D V
H E V l l0 * 1  1 DV E  Hn (8)

The prediction by Eq. (8) complies with SFT test data, Fig. 11.

2.0

1.8

1.6
H!, %

1.4

1.2

5 10 20

l, mm
Fig. 11. Average failure strain as a function of gauge length obtained by SFT tests.
Solid line – Eq. (8), dashed line – Eq. (16) based on SFF results.

35
 Paper A

4. SFF tests
4.1. Analysis of fragmentation data

During the initial fragmentation stage when fibre break interaction

is negligible the average fragment length, l , of a fibre with Weibull
failure strain distribution (5) is given by

D H
ªH f º
l l0 « » (9)
¬ EH ¼

where H f is the fibre strain, H f

H a  H r , composed of the
mechanical strain applied to the SFC, H a , and residual fibre strain,
H r . A typical fragmentation diagram of elementary flax fibre is
plotted in Fig. 12. The presence of a pronounced initial fragmentation
stage with power-law dependence of average fragment length on fibre

15
10

5
l!, mm

3
2

0.5

0.3

2 3 4 5

Hf , %
Fig. 12. Typical fibre fragmentation diagram showing the dependence of average
fragment length on fibre strain and the approximation of the initial part of the
diagram employed for Weibull parameter evaluation.

36
 Composites Science and Technology 65 (2005) 693-702

0
ln(-ln(1-P))

-2

-4

3 5 9 15

DH
a)

0
ln(-ln(1-P))

-2

-4

2 3 4

EH, %
b)
Fig. 13. Weibull plot of fibre failure strain distribution, Eq. (5), parameters DH (a)
and EH (b) obtained by SFF tests.

37
 Paper A

strain in agreement with Eq. (9) corroborates the applicability of

Weibull distribution (5) for failure strain of individual fibres.
We estimated the residual fibre strain using fibre failure strain data
obtained by SFT tests at a fixed gauge length so that the average fibre
failure strain at the given gauge length derived from SFF tests is
equal to the value obtained from SFT. The following iterative
procedure was applied. An initial value for H r was assigned, and the
Weibull distribution parameters for each SFC specimen were
determined approximating the initial linear part of the fragmentation
diagram in double logarithmic coordinates by (the logarithm of) Eq.
(9) as shown in Fig. 12. Then the average failure strain for each SFF-
tested fibre at a length l * , H i l * , was estimated using the obtained
1 D i
Weibull parameters as Hi l * E i l * l0 * 1  1 D i . The
calculated failure strain values for SFF-tested fibres were averaged
producing the fragmentation test estimate for average fibre failure
1 n
strain at gauge length l * , H l * ¦ H i l * , where n is the
ni1
number of SFC tested. The obtained H l * value is compared with
average fibre failure strain determined by SFT tests at the given
gauge length, and the residual strain estimate adjusted accordingly.
Such iterations are repeated until H r value at which H l * is equal
to the average failure strain yielded by SFT test is found. Using
average fibre failure strain at l* 10mm gauge length as a
benchmark, the residual strain was found to be H r 0.41% in the
case of VE matrix and H 0.31% for UP matrix.
Failure strain distribution (5) parameters obtained by SFF tests vary
considerably between fibres as the Weibull plots for the parameters
reveal, Fig. 13. The average values of the parameters are given in
Table 3. The coefficient of variation for shape parameter, kD ,
amounts to 0.30 and for scale parameter, k E , to 0.13. It is interesting
to note that a roughly comparable variability of Weibull distribution
parameters was also obtained by SFF tests of E-glass fibres [17],
namely kD 0.27 and k E 0.10 .

38
 Composites Science and Technology 65 (2005) 693-702

4.2. Modified Weibull distribution of fibre failure strain

It has been shown by numerical simulation [24] that the batch

strength distribution of fibres, each of which has Weibull strength
distribution (1) with the same shape parameter but randomly varying
scale parameter (the latter also being Weibull-distributed), closely
approximates the modified Weibull distribution (3). Flax SFF tests
suggest that failure strain of individual fibres has Weibull distribution
(5), and the variability of the scale parameter also agrees with
Weibull distribution as revealed by the linearity of the corresponding
Weibull plot, Fig. 13b. However, the scatter of the Weibull shape
parameter values is not negligible, Fig. 13a. Therefore the
approximate relations derived in [24] for parameters of the modified
Weibull distribution are not applicable here.
Similar fibre-to-fibre variability of Weibull distribution parameters
was also obtained in SFF tests of glass fibres, the strength of which
complied with the modified Weibull distribution [17]. We apply the
method proposed in [17] to estimate from SFF tests the parameters
the modified Weibull distribution of failure strain

§ §l J DH ·
·HªH º
PH 1  exp¨  ¨¨ ¸¸ « » ¸ (10)
¨ © l0 ¹ ¬ EH ¼ ¸
© ¹

Assuming that (i) the difference in Weibull parameter values

obtained by SFF tests stems solely from the variation in fibre strength
distribution parameters among fibres and (ii) the set of Weibull
parameters determined by SFF tests fully describes the interfibre
strength variability, we obtain [17]

1 n 1 D i
H ¦ E i ˜ l l0
ni1
* 1 1 Di (11)

­1 n 2  2 Di
sH ® ¦ Ei ˜ l l0 * 1  2 Di 
¯n i 1
2
1 (12)
º ½°
n 2
ª1 1 D i
 « ¦ E i ˜ l l0 * 1  1 Di » ¾
¬n i 1 ¼ °¿

39
 Paper A

where n is the number of fibres tested by SFF, and D i , Ei are the

Weibull parameters of the i -th fibre. Further, equating the ratio of
standard deviation and average failure strain provided by Eqs. (12)
and (11) at l l 0 to the coefficient of variation for limit strain
according to Eq. (10), we obtain relation for the shape parameter of
distribution (10)

sH * 1  2 DH
2
1 (13)
H l l0
* 1  1 DH

Equating average strength and its derivative with respect to fibre

length based on Eq. (11) and on distribution (10) result in

H l l0
EH * 1  1 DH (14)

d H JH
 EH * 1  1 DH (15)
dl l l0
l 0D H

Solving Eq. (13) for DH and determining EH and J H from Eqs. (14)
and (15) yields the numerical values for distribution (10) parameters
based on SFF test results as follows: J H 0.79 , DH 4.97 ,
EH 2.5% . The average failure strain as a function of fibre length
according to Eq. (10)

J H DH
H E H l l0 * 1  1 DH (16)

is plotted in Fig. 11 by dashed line. It is seen that the length

dependence for fibre failure strain yielded by Eq. (16) agrees
reasonably well with the SFT experimental data, although failure
strain variability at fixed fibre length is underestimated.
Combining Eqs. (7) and (16), we obtain for average fibre strength

J H DH
V E E H l l0 * 1  1 DH  H n (17)

40
 Composites Science and Technology 65 (2005) 693-702

The relation (17) is reproduced in Fig. 8 against SFT experimental

data. It is seen that Eq. (17) based on SFF tests can rather accurately
predict the length effect of fibre strength. Thus a limited number of
SFF tests combined with SFT tests at a fixed gauge length can
provide the experimental information necessary for characterizing the
scatter of fibre failure strain and strength and the effect of fibre length
on the mentioned parameters.

5. Conclusions.

Elementary flax fibre strength and failure strain distributions at

several gauge lengths are obtained by SFT tests. Strength and failure
strain possess the modified Weibull distribution (i.e. independent
exponents characterize strength scatter at a fixed gauge length and the
dependence of mean strength on fibre length). SFF tests reveal that
fibre fragmentation process proceeds in agreement with the two-
parameter Weibull distribution of failure strain, Eq. (5), but the
distribution parameters for individual fibres exhibit a marked scatter.
A limited number of SFF tests combined with SFT tests at a fixed
gauge length provide the experimental information necessary for
characterizing both the scatter of fibre failure strain and strength at a
fixed gauge length and the effect of fibre length on the mentioned
parameters.

Acknowledgements

Undergraduate students Henrik Lindberg, Borsini Serena, Isidori

Tatiana and graduate student Philippe Lingois were involved in this
work.
This study is partially financed by the Swedish Royal Academy of
Sciences.
Part of this study has been carried out with financial support from
the Commission of the European Communities granted to the
GROWTH Project GRD1-1999-10951 "ECOFINA". The study does
not necessarily reflect the views of the European Commission.
Flax fibres for this study were supplied by partner in ECOFINA
project, FINFLAX. Fibre surface treatment was performed by partner
in ECOFINA project, University of Basque Country.

41
 Paper A

Appendix. Verification of the extended fibre SFF test method

The intended role of fibre extensions is to hold the fibre in place

while the resin solidifies. Once the resin has cured, the perturbation of
the fibre stress due to the presence of extensions is expected to extend
only by a limited distance, so called stress transfer length, from the
fibre ends. This distance, a function of fibre and resin moduli and
typically an order of magnitude larger than fibre diameter, is to be
excluded from monitoring for fibre cracks during fragmentation test.

Matrix

Fiber

Extension
line

Fig. A1. Finite element model of the extended fibre SFC.

In order to illustrate the principal features of stress transfer, we

performed linear elastic analysis of the SFC under thermomechanical
loading by finite element code ANSYS. In order to simplify analysis,
Plane83 2-D 8-node axi-symmetric elements (10000 elements in
total) were used to model SFC geometry of coaxial fibre and

42
 Composites Science and Technology 65 (2005) 693-702

extension embedded in a resin cylinder; schematic picture of the

model and example of the mesh are shown in Fig. A1. Symmetry
conditions are enforced along the top edge of the model and vertical
( y ) axis, which is the line of axial symmetry. The thermoelastic
properties of the constituents were chosen as for glass fibre and VE
matrix: fibre modulus E f 73GPa , Poisson’s ratio Q f 0.22 ,
coefficient of thermal expansion CTEf = 5·10-6 1/ºC; matrix modulus
Em 3GPa , Poisson’s ratio Q m 0.35 , coefficient of thermal
expansion CTEm= 6·10-5 1/ºC. The radius of the extension is taken
four times larger than fibre radius, r f , and that of the matrix cylinder
is 40 r f ; the length of the extension line is 40 r f , the fibre length is
160 r f . The stress state in the SFC was calculated for thermal and for
mechanical loading separately. The thermal loading is due to the
temperature drop from a stress-free post-cure temperature to room
temperature. The mechanical loading is accomplished by applying a
uniform y-direction displacement to the bottom (i.e. y 0 ) line of the
model corresponding to the nominal applied strain of 1%. The
calculated axial strain at the symmetry axis of the model ( x 0 ) in
the extension and the fibre is shown in Fig. A2 as a function of the
relative distance form the bottom edge of the model ( y 0 ). It is seen
that the perturbation zone of the fibre strain extends for a distance of
about 50 r f from the fibre/extension joint for the range of extension
mechanical characteristics studied. (Note that the slight dependence
of the calculated far-field fibre strain under mechanical load on the
extension modulus, Fig. A2b, is due to the rather limited cross-section
area of the resin in the model as compared to the actual dimensions of
the SFC’s tested, Fig. 3.)
In order to verify the applicability of the modified SFF test method,
we performed a limited number of extended-fibre SFF tests for an E-
glass fibre previously tested by the ordinary SFF method involving
long fibres as reported in [17]. Short glass fibres of ca 26 mm length
were cut from continuous filaments separated from the glass fibre
bundle. The fibre extensions were attached and SFC specimens
prepared following the routine described in Section 2.3.1; VE matrix
was used. Fragmentation testing and data analysis was carried out in
exactly the same way as described in [17] for long-fibre SFCs. The
average values of the Weibull shape and scale parameters, based on

43
 Paper A

0.0
Thermal loading, 'T = 800C
Eext= Em CTEext= CTEm
-0.2
Eext= Ef /2 CTEext= CTEf
Eext= Ef CTEext= CTEf
axial strain, %

-0.4

-0.6

-0.8 extension perturbation

zone

0 50 100 150 200

normalised distance, y/rf

a)
2.0
Mechanical loading, H = 1 %

Eext=Em Qext=Qm
1.5 Eext=Ef/2 Qext=Qf
Eext=Ef Qext=Qf
axial strain, %

1.0

0.5

extension perturbation
zone
0.0
0 50 100 150 200

normalised distance, y/rf

b)
Fig. A2. Axial strain distribution in the extension and the fibre under thermal (a)
and mechanical (b) loading. Dotted line shows the position of extension/fibre joint.

44
 Composites Science and Technology 65 (2005) 693-702

four SFF tests, amounted to 8.0 (1) and 3150 (120) MPa respectively
(here and below the number in brackets is the standard deviation of
the corresponding average value). Twelve long-fibre SFF tests
yielded the average shape parameter of 8.6 (.7) and scale parameter
3090 (90) MPa [17]. It is seen that the fibre strength distribution
parameters obtained by the modified SFF test agree within
experimental scatter with those obtained by the ordinary SFF
procedure.

References

1. Lilholt H, Lawther JM. Natural organic fibres. In: Kelly A, Zweben C,

editors. Comprehensive composite materials, vol. 1. New York: Pergamon
Press; 2000.
2. Bledzki AK, Gassan J. Composites reinforced with cellulose based fibres.
Progress in Polymer Science 1999;24:221–274.
3. Joshi SV, Drzal LT, Mohanty AK, Arora S. Are natural fiber composites
environmentally superior to glass fiber reinforced composites?
Composites: Part A 35;(2004):371–376.
4. Magurno A. Vegetable fibres in automotive interior components. Die
Angewandte Makromolekulare Chemie 1999;272:99–107.
5. de Bruijn JCM. Natural fibre mat thermoplastic products from a
processor’s point of view. Applied Composite Materials 2000;7:415-420.
6. Davies GC, Bruce DM. Effect of environmental relative humidity and
damage on the tensile properties of flax and nettle fibers. Textile Research
Journal, 1998;68:623-629.
7. Van de Velde K, Baetens E. Thermal and mechanical properties of flax
fibres as potential composite reinforcement. Macromolecular Materials
and Engineering 2001;286:342-349.
8. Bos HL, van den Oever MJA, Peters OCJJ. Tensile and compressive
properties of flax fibres for natural fibre reinforced composites. Journal of
Materials Science 2002;37:1683-1692.
9. Van de Weyenberg I, Ivens J, Verpoest I. Parametric study of the
relationship between the fibre and the composite properties of flax fibre
reinforced epoxy. Proceedings of ECCM9, Brighton; June 2000.
10. Joffe R, Andersons J, Wallström L. Strength and adhesion characteristics
of elementary flax fibers with different surface treatments. Composites:
Part A 2003;34:603–612.
11. Van den Oever MJA, Bos HL. Critical fibre length and apparent interfacial
shear strength of single flax fibre polypropylene composites. Advanced
Composites Letters 1998;7:81-85.
12. Zafeiropoulos NE, Baillie CA, Matthews FL. A study of transcrystallinity
and its effect on the interface in flax fibre reinforced composite materials.
Composites: Part A 2001;32:525–543.

45
 Paper A

13. Zafeiropoulos NE, Baillie CA, Hodgkinson JM. Engineering and

characterisation of the interface in flax fibre/polypropylene composite
materials. Part II. The effect of surface treatments on the interface.
Composites: Part A 2002;33:1185–1190.
14. Baley C. Analysis of the flax fibres tensile behaviour and analysis of the
tensile stiffness increase. Composites: Part A 2002;33:939–948.
15. Drzal LT, Herrera-Franco PJ, Ho H. Fibre–matrix interface tests. In: Kelly
A, Zweben C, editors. Comprehensive composite materials, vol. 5. New
York: Pergamon Press; 2000.
16. Andersons J, Joffe R, Hojo M, Ochiai S. Fibre fragment distribution in a
single-fibre composite tension test. Composites: Part B 2001;32:323-332.
17. Andersons J, Joffe R, Hojo M, Ochiai S. Glass fibre strength distribution
determined by common experimental methods. Composites Science and
Technology 2002;62:131-145.
18. Eichhorn SJ, Young RJ. Composite micromechanics of hemp fibres and
epoxy resin microdroplets. Composites Science and Technology
2004;64:767 –772.
19. Hornsby PR, Hinrichsen E and Tarverdi K. Preparation and properties of
polypropylene composites reinforced with wheat and flax straw fibres. Part
I Fibre characterization. Journal of Materials Science 1997;32:443-444.
20. Oksman K. Mechanical properties of resin transfer molded natural fiber
composites. In: The Fifth International Conference on Woodfiber-Plastic
Composites, 1999, pp. 97-103.
21. Oksman K. High quality flax fibre composites manufactured by the resin
transfer moulding process. Journal of Reinforced Plastics and Composites
2001;20:621-627.
22. Gutans JA, Tamuzh VP. Scale effect of the Weibull distribution of fibre
strength. Mechanics of Composite Materials 1984;(6):1107-1109 (in
Russian).
23. Watson AS, Smith RL. An examination of statistical theories for fibrous
materials in the light of experimental data. Journal of Materials Science
1985;20:3260-3270.
24. Curtin WA. Tensile strength of fiber-reinforced composites: III. Beyond
the traditional Weibull model for fiber strengths. Journal of Composite
Materials 2000;34(15):1301-1332.

46
 Paper B

Stiffness and strength of flax fiber/polymer

matrix composites
J. Andersons1, R. Joffe2, E. SpƗrniƼš1
1
Institute of Polymer Mechanics, University of Latvia, Aizkraukles iela 23, LV-1006 RƯga, Latvia
2
Division of Polymer Engineering, Luleå University of Technology, SE-971 87 Luleå, Sweden

Polymer Composites 27 (2006) 221-229

47
48
 Polymer Composites 27 (2006) 221-229

Stiffness and strength of flax fiber/polymer

matrix composites
J. Andersons1, R. Joffe2, E. SpƗrniƼš1
1
Institute of Polymer Mechanics, University of Latvia, Aizkraukles iela 23, LV-1006 RƯga, Latvia
2
Division of Polymer Engineering, Luleå University of Technology, SE-971 87 Luleå, Sweden

Abstract. Flax fiber composites with thermoset and thermoplastic polymer

matrices have been manufactured and tested for stiffness and strength under
uniaxial tension. Flax/polypropylene and flax/maleic anhydride grafted
polypropylene composites are produced from compound obtained by co-extrusion
of granulated polypropylene and flax fibers, while flax fiber mat/vinylester and
modified acrylic resin composites are manufactured by resin transfer molding. The
applicability of rule-of-mixtures and orientational averaging based models,
developed for short fiber composites, to flax reinforced polymers is considered.

1. Introduction

Flax fibers are increasingly being studied and used as a

reinforcement of traditional thermoplastic (see e.g. [1-15]) and
thermoset [16-23], as well as biopolymer [24-25] matrix composites.
In order to experimentally determine the most efficient utilization
conditions of flax fiber properties, different flax fiber reinforcement
types (i.e. rowing, mat) and treatments are considered. Effect of fiber
treatment on the stiffness and strength of polypropylene (PP) [1, 10]
and epoxy [17, 19, 20] composites has been studied extensively. The
reinforcing fiber length [3, 7, 21], fiber content [3, 7, 17, 18, 21, 23],
fiber retting degree [13, 20] and decortication stage [4, 20] are also
factors affecting the mechanical properties of a composite. It would
obviously be helpful to characterize fiber and interface properties by
dedicated tests and then apply theoretical models for composite
property prediction instead of producing and testing composites to
evaluate the overall effect of each particular treatment. However,
comparatively few attempts to relate theoretically the mechanical
properties of a flax fiber composite to those of the fiber, matrix, and
their interface have been made [3, 4, 7, 15, 26]. Variations of rule-of-
mixtures relationships were found useful for modulus prediction [26]
and longitudinal and transverse strength and modulus correlations [7]

49
 Paper B

of unidirectionally reinforced composites as well as modulus and

strength prediction for random flax/PP composites [3, 4]. Approaches
to composite stiffness prediction explicitly accounting for the
variability in fiber properties have also been proposed [15, 26].
The aim of the present study is a systematic evaluation of the
applicability of elementary mechanical models for flax fiber
reinforced polymer matrix composite modulus and strength
prediction. Using commercial, well characterized in terms of axial
mechanical properties [28] and adhesion [29] enzyme-retted flax
fibers, thermoplastic matrix composites were produced by extrusion
and thermoset composites by resin transfer molding (RTM).
Correlation of experimental and theoretical composite characteristics
revealed the degree of sensitivity of the elementary models to
properties of the constituents and their adhesion.

2. Experimental
2.1 Materials

Flax fiber mats (FFM) produced by FinFlax were used as

reinforcement for polymer composites. The resins used were as
follows; thermoset resins: vinylesters XZ92485 and Momentum 411-
350 by DOW (denoted as VE1 and VE2), modified acrylic resin (AR)
Modar 835S by Ashland which can be cured as unsaturated polyester.
(Note that Modar 835S was mistakenly referred to as unsaturated
polyester (UP) in [28-30]). Thermoplastic: Adstif 770 ADXP Basell
polypropylene and maleic anhydride grafted PP (MAPP) modified
polypropylene (PPM).

2.2 Composites Manufacturing and Testing

Plates (of 3 mm thickness) of thermoset natural fiber composites

(NFC) were manufactured by RTM and postcured for 5h at 80º C
after gelation. Rectangular specimens were cut from the plates for
tensile tests. Some of the specimens were cut and polished after
mechanical tests to evaluate the morphology of the composites.
Micrographs in Fig. 1 suggest that the fibers (and fiber bundles)
exhibit relatively uniform distribution of orientations within the
specimen plane.

50
 Polymer Composites 27 (2006) 221-229

a)

b)

51
 Paper B

c)
Fig. 1. Top (a), edge (b), and cross-section (c) micrographs of a flax
mat/thermoset matrix NFC specimen.

Thermoplastic composite specimens were produced from

compound obtained by co-extrusion (Krupp Werner & Pfleiderer
ZSK 25 WLE twin screw extruder) of granulated PP and flax fiber
rowing. Temperature in the extruder at the polymer entrance point
was 180°C and it was raised to 190-200°C after the 2/3 of the path
along the length of the extruder screw was passed. Afterwards
compound was once more heated in the oven to 190-200°C and
pressed in the press (pressure of 15-25 tons, temperature of the press
§ 40°C) under stiff profile that allowed to produce rectangular and
dog-bone shaped specimens with constant thickness. Fig. 2 shows
representative micrographs of top, edge, and cross-section views of a
rectangular specimen. Although a fraction of the fibers appears
randomly distributed, there is distinct evidence of a preferential
direction in the fiber orientation, namely along the specimen length-
wise direction.

52
 Polymer Composites 27 (2006) 221-229

a)

b)

53
 Paper B

c)
Fig. 2. Top (a), edge (b), and cross-section (c) micrographs of an extruded flax/PP
matrix NFC specimen.

Tensile specimens were of rectangular shape, 250 mm long and 25

mm wide. In order to prevent failure of the specimens in the gripping
area, very fine metal mesh was applied on the clamped ends (50 mm
long) of each sample. Tensile tests were carried out in an Instron
1272 tensile machine with a 25 kN load cell. The experiment was

54
 Polymer Composites 27 (2006) 221-229

performed in displacement-control mode at a stroke rate (i.e. cross-

head displacement rate) of 2 mm/min. During loading, the
longitudinal displacement was measured by an MTS extensometer.
All output data (strain, displacement of cross-head, and load) were
collected by an acquisition system and transferred to the PC.

3. Elementary models of short fiber composite stiffness and

strength
3.1 Stiffness

A number of theoretical models for prediction of the elastic

properties of short-fiber reinforced composites (SFC) have been
elaborated differing in accuracy and complexity (for recent reviews,
see e.g. [31, 32]). The relatively simple Cox-Krenchel model was
found to yield good agreement with experimental modulus values for
a range of glass fiber lengths and volume fractions [33], and also to
perform acceptably for random flax/PP composites [3, 4]. Composite
modulus E is related to fiber and matrix moduli, E f and Em , and
fiber volume fraction Q f by a rule of mixtures type of relationship:

E K oEKlE E f Q f  1  Q f E m (1)

where K oE is orientation factor and K lE is fiber length efficiency

factor. For reinforcing fibers of length l , the latter is given by

tanh El 2
KlE 1 (2)
El 2

where

1 2Gm
E
rf E f ln R r f

and rf stands for fiber radius, Gm is matrix shear modulus, and R

relates to the interfiber spacing in the composite. The ratio R r f can

55
 Paper B

be expressed as R r f K R Q f , where the numerical factor K R

depends on fiber geometrical packing, being equal to S 4 for square
packing [31].
If the reinforcing fiber length is variable with the distribution
density given by h l ,

f
1 § tanh El 2 ·
K lE ³ ¨¨©1  ¸lh l dl . (3)
l 0
El 2 ¸¹

Fiber orientation factor K oE is determined by the fiber orientation

distribution (see eg. [34, 35]). Neglecting transverse deformations,
fiber orientation factor K oE value for random in-plane orientation of
the fibers (that would approximate fiber arrangement in flax fiber
mat) is 3 8 , while for random three-dimensional fiber orientation
(approximating fiber arrangement in short-fiber extruded composite)
KoE 1 5 .
Application of more complicated models of SFC stiffness (e. g. [36,
37]) is hampered for flax fibers by the lack of experimental data
concerning transverse and shear properties of the fibers. In fact,
apparently the only natural fiber that has been exhaustively
characterized in terms of thermoelastic anisotropy is jute [38].

3.2 Strength

In comparison to the elastic properties, strength theory for SFC is

still under development. Most of the SFC strength models are also
based on the “rule of mixtures” [39-42] or equivalent laminate [43,
44] approach, i.e. are basically engineering approximations. Recently,
more complicated models have been advanced [45-48] that employ
numerical modeling of deformation and failure processes.
The strength of a SFC can be expressed as [39-42]

V uc K sV uf Q f  1  Q f V m (4)

56
 Polymer Composites 27 (2006) 221-229

where V uf is fiber strength, V m is stress in the matrix at the fiber

V uf Em
failure strain (for linear elastic constituents, V m ), and K s
Ef
denotes the fiber efficiency factor. The latter can be decomposed as
K s Kls ˜Kos [40, 41], where fiber length efficiency factor K ls and
orientation factor K os have similar interpretation as for modulus
calculation. (Note that strain concentration factor is also likely to
contribute to K s [32, 49]). Fiber length efficiency factor is estimated
as [39]:

­1  l c 2l l t l c
K ls ® (5)
¯ l 2l c l  lc

where

V uf r f
lc (6)
W

and W is interface shear strength (IFSS). Approximating experimental

strength data for a range of fiber lengths and volume fractions in
glass/PP short fiber composites, a best-fit value of 0.2 for K os was
found [50]. This fitted parameter value apparently incorporates both
fiber orientation and fiber efficiency factors. For flax/PP, the fitted
parameter K os amounted to 0.075 [3] indicating thus considerably
smaller reinforcement efficiency by flax as compared to glass fibers.
Models [40-42] provide means for determining K s , but at the cost
of introducing other intrinsic material parameters, critical-zone (or
process zone) width in Fukuda and Chou [40] model or bridging
stress and strength of an oblique fiber in Fu and Lauke [41, 42]
model. We consider the simplest critical-zone model by Fukuda and
Chou [40], modified according to the suggestions of Jayaraman and
Kortschot [35] (see Appendix). The fiber strength as a function of its
length is evaluated according to the modified Weibull distribution
leading to

J D
V uf l E V l l0 * 11 D (7)

57
 Paper B

The Weibull strength distribution parameters entering Eq. (7) are

presented in [28]. By combining Eqs. (6) and (7) [44], we obtain for
the critical length

D J D
§ EV * 1  1 D r f ·
lc ¨ ¸ (8)
¨ Wl J D ¸
© 0 ¹

The critical zone width ln is chosen equal to the critical fiber length
l . Then for random two-dimensional fiber orientation assumed for
c

fiber mat reinforced composite, it follows from Eqs. (A 4), (A 5) that

the strength is given by

Q f f V uf l l§ § 2
· 2 ·

V uc ¨ 3cos 1 lc  lc ¨ 3  2 ¨§ lc ¸· ¸ 1  ¨§ lc ¸· ¸ u
4S l³ l ¨ l l ¨© © l ¹ ¸¹ © l ¹ ¹¸
c © (9)
§ l ·
u ¨1  c ¸ h l dl  1 Q f V m
© 2l ¹

Analogically, for the strength in case of a random three-dimensional

fiber orientation assumed for extruded SFC, it follows from Eq. (A
4), (A 6) that

5
Qf f
V uf l l ª § l c · º§ l c ·
V uc ³ «1  ¨ ¸ »¨1  ¸h l dl  1  Q f V m (10)
5 lc
l «¬ © l ¹ »¼© 2l ¹

For application of more complicated models of SFC that provide

also strength under combined loading [41, 42, 44] and/or complete
stress-strain diagrams accounting for gradual damage accumulation
and matrix non-linearity [45-48], more detailed information on fiber
and interface properties is needed than is typically available for
natural fibers.

4. Results and discussion

Experimental modulus values for flax mat/thermoset matrix NFCs

are shown in Fig. 3 compared with polymer matrix modulus. It is

58
 Polymer Composites 27 (2006) 221-229

seen that flax fiber volume fraction of 30% provides significant

stiffness increase over pure polymer as expected. Theoretical
modulus values obtained by Eq. (1) are also presented in Fig. 3.
Elementary flax fiber modulus E f 69GPa and the average diameter
d f 16ȝm for FinFlax fibers [28] are taken here and in subsequent
calculations. Since the average fiber length in flax mat is evaluated at
about 20 mm that exceeds considerably the stress transfer length of
the fiber, length efficiency factor in flax mat composites was assumed
equal to 1. Due to the relatively uniform planar arrangement of fibers
in the mat, Fig. 1, fiber orientation factor for random in-plane
orientation, KoE 3 8 , was applied. It is seen that the Young’s
modulus of the composite is slightly but consistently overpredicted.
This is likely to be due to the presence of fiber curvature in the mat
(as seen in Fig. 1a) that makes the composite somewhat more
compliant than the assumed perfectly planar arrangement of straight
fibers.

12
matrix
composite
predicted
10

8
E, MPa

0
FFM / AR FFM / VE1 FFM / VE2

Fig. 3. Experimental and predicted modulus of flax fiber mat/thermoset matrix

composites.

59
 Paper B

The dependence of extruded flax/PP modulus on fiber volume

fraction is shown in Fig. 4. MAPP modification of the PP matrix has
no discernible effect on composite modulus. The distribution of flax
fiber length in extruded flax/PP composites is given in Fig. 5 (after
[51]). The dependence of modulus on fiber volume fraction estimated
according to Eqs. (1), (3) is plotted in Fig. 4 by a solid line. Lacking
precise fiber orientation distribution measurements, we employ fiber
orientation factor for random three-dimensional fiber orientation,
KoE 1 5 , as a rough approximation, notwithstanding the preferential
fiber orientation suggested by Fig. 2. The theoretical dependence of
Young’s modulus on the fiber volume fraction agrees reasonably well
with the test data.

6
flax / PP
flax / PPM
5

4
E, MPa

0
0.0 0.1 0.2 0.3

Qf
Fig. 4. Experimental (markers) and predicted (solid line) modulus variation with
fiber volume fraction for extruded flax/PP composites.

For fiber mat reinforced composites, the critical fiber length

calculated using the IFSS values obtained in [29] is negligible in
comparison to the average fiber length; monodisperse fiber length
distribution is assumed. Eq. (9) leads to the following expression for
composite strength

60
 Polymer Composites 27 (2006) 221-229

3
V uc Q f V uf l  1  Q f V m (11)
8

The flax fiber strength distribution parameters needed for the mean
fiber strength at a given length, V uf l , evaluation according to Eq.
(7) are as follows: D 2.8 , EV 1400MPa , J 0.46 (and
l0 1mm ) [28]. The prediction by Eq. (11) is plotted against
FFM/thermoset NFC strength in Fig. 6.

1.6

1.4

1.2

1.0
-1
h(l ), mm

0.8

0.6

0.4

0.2

0.0
0 1 2 3 4 5 6

l, mm
Fig. 5. Fiber length distribution in extruded flax composite.

The apparent IFSS of flax fiber and PP matrices are reported in [4,
52] and constitute 8 MPa for PP and 12 MPa for PP/MAPP. The
critical fiber length lc evaluated by Eq. (8) equals 1.17 mm for
flax/PP and 0.82 mm for flax/PPM composites, respectively. The
critical length values lie close to the mode of extruded fiber length
distribution density, Fig. 5. The theoretical strength values calculated
by Eq. (10) are compared to the experimental results for extruded
flax/PP and PPM at different fiber volume fractions in Fig. 6. It is
seen that the experimental and predicted strength values are

61
 Paper B

reasonably well correlated, but the strength is consistently

overestimated by Eqs. (10), (11). Therefore we introduce an empirical
fiber efficiency factor K s* modifying the fiber contribution term in Eq.
(A 4) and hence in Eqs. (9) – (11). The value of K s* 0.63 yielded the
best fit to extruded flax/PPM strength data. Figs. 5 and 6 show that
the same K s* value provides reasonably good accuracy of strength
prediction also for extruded flax/PP and FFM/thermoset NFCs.

120
experimental strength, MPa

100

80

60

40

20

0
0 20 40 60 80 100 120

theoretical strength, MPa

Fig. 6. Experimental strength vs. theoretical prediction for extruded flax/PP and
PPM at Q f 0.13 (Ƒ), 0.2 (¨), 0.29 (¸) and for flax mat/thermoset matrices,
Qf 0.3 (ż). The slope of the dashed line is 1 that would correspond to equal
experimental and predicted strength.

The adhesion of flax and PP is apparently insufficient to provide

enhancement of the NFC strength above that of the polymer, while
addition of MAPP leads to a notable strength increase, Fig. 7. The
higher IFSS due to MAPP, reflected in the decrease of the critical
fiber length, is relatively well translated into NFC strength increment
by the modified Fukuda and Chou model. By contrast, the model
predicts virtually the same strength for FFM/AR and FFM/VE2

62
 Polymer Composites 27 (2006) 221-229

NFCs, Fig. 8, although their IFSS differ by a factor of two [29]. This
is apparently due to the fact that IFSS enters the strength model only
via the critical fiber length. Therefore the average fiber length in the
flax mat, being large in comparison with lc , makes the effect of
variations in lc insignificant. Hence, in order to better reflect the role
of adhesion on the strength of relatively long fiber NFC, a more
sophisticated model is needed.

45
flax / PP
flax / PPM
PP
40

35
Vc , MPa

30

25

20
0.0 0.1 0.2 0.3

Qf
Fig. 7. Experimental (markers) and predicted (solid lines) strength variation with
fiber volume fraction for extruded flax/PP composites.

Although the detailed information on the anisotropic elastic and

strength properties of the fibers required by most of the strength
models is largely missing for flax, relatively rough estimates can still
be made. Consider an orientational averaging model for strength
proposed by van Hattum and Bernardo [44]. The model represents
misaligned SFC as an assembly of unidirectionally reinforced (UD)
plies. Tsai-Wu strength tensor [53, 54] of the SFC is obtained by first
averaging limit strains of the mentioned UD plies over all directions
weighted by fiber orientation distribution function, and then
transforming the limit strain tensor to strength tensor by means of

63
 Paper B

SFC compliance tensor (obtained by orientational averaging of UD

stiffness tensors). UD composite stiffness is derived using Halpin-
Tsai relations [55], longitudinal strength – by the rule of mixtures,
while transverse and shear strength is approximated by that of the
matrix in the case of perfect adhesion [44].

120
matrix
composite
rule of mixtures
100 orientational averaging

80
Vc , MPa

60

40

20

0
FFM / AR FFM / VE1 FFM / VE2

Fig. 8. Experimental and predicted strength of flax fiber mat/thermoset matrix

composites.

Regarding transverse and shear stiffness of flax fiber, we follow the

assumption made in [7] that for flax and jute, both being
lignocellulosic fibers, the mechanical behavior is qualitatively similar
and the degree of anisotropy of stiffness properties is approximately
equal. Therefore the ratios of transverse and shear moduli to the axial
fiber modulus for flax fiber are taken equal to the corresponding
ratios for jute fiber at room temperature [38] (i.e. although the moduli
of flax and jute fibers differ considerably, their ratios are assumed
equal). Longitudinal tensile strength of a UD layer is estimated by the
modified Fukuda and Chou model, while transverse and in-plane

64
 Polymer Composites 27 (2006) 221-229

shear strengths are evaluated by the micromechanical relations of [55]

assuming that the former is governed by the matrix strength while the
latter – by fiber and matrix adhesion characterized by IFSS. The
predicted FFM/AR and FFM/VE2 NFC strengths, based on the van
Hattum and Bernardo model with the assumptions discussed above
and IFSS values taken from [29], are presented in Fig. 8. It is seen
that the orientational averaging model, despite the approximate nature
of the material properties used as input, captures the difference
between FFM/AR and FFM/VE2 strength considerably better than the
rule of mixtures approach. Such an enhanced sensitivity to matrix and
adhesion characteristics apparently stems from the implicit
accounting for the SFC failure modes due to matrix fracture and fiber
debonding inherent in the orientational averaging model [44].

5. Conclusions

Stiffness and strength under uniaxial tension has been obtained for
flax/PP and flax/PPM composites produced from compound obtained
by co-extrusion of granulated PP and flax, as well as for
FFM/vinylester and FFM/acrylic resin composites manufactured by
resin transfer molding. The rule-of-mixtures relations are shown to
yield acceptable stiffness prediction of both extruded and FFM-based
composites. The sensitivity of a rule-of-mixtures model of strength to
the matrix and adhesion properties apparently depends on the relative
ineffective fiber length; for relatively long-fiber FFM composites the
sensitivity is low and more sophisticated strength models should be
applied.

65
 Paper B

Acknowledgements

Part of this study has been carried out with financial support from
the Commission of the European Communities granted to the
GROWTH Project GRD1-1999-10951 "ECOFINA". The study does
not necessarily reflect the views of the European Commission.
Flax fibres for this study were supplied by partner in ECOFINA
project, FINFLAX.
Authors also would like to thank Dr. Lennart Wallström for
arrangements concerning manufacturing of composites.
E. SpƗrniƼš gratefully acknowledges the support of ESF.

Appendix. Modification of Fukuda-Chou strength theory.

Jayaraman and Kortschot [35] have noted an inconsistency in the

way the load born by the short fibers is calculated for SFC modulus
[34] and strength [40] estimation, and also provided the corrected
model for modulus [35]. Based on the statistical derivations presented
in [35], it is straightforward to correct also the strength model [40]. A
basic part of the analysis is calculation of the force sustained by the
fibers crossing a scan line (arbitrary line normal to the applied load in
a thin rectangular specimen) [35]. This involves finding the number
of fibers of length l and orientation T that cross the scan line,
determining the load-direction component of the force such fibers
carry at failure, and finally integrating over fiber length and
orientation to find the total force sustained by the fibers. The number
of fibers of length between l and l  dl and orientation between T
and T  dT  that cross the scan line is [35]

Ac l
N l ,T Vf cosTh l g T dldT (A 1)
Af l

where h l and g T are fiber length and orientation distribution

density correspondingly, l denotes the average fiber length, A f
stands for fiber cross-section area, and Ac – composite specimen
cross-section normal to the applied load. The force carried by a fiber
in the load direction (i.e. normal to the scan line) is [40]

66
 Polymer Composites 27 (2006) 221-229

­ A f V 0 l cos 3 T cosT t l n l
F l ,T ® (A 2)
¯ 0 cosT  l n l

where ln is the critical zone width and V 0 l is related to fiber

strength V uf and critical fiber length lc as follows

­ 1  l c 2l V uf l t lc
V0 l ® (A 3)
¯ l 2l cV uf l  lc

fS 2
The total force is given by FT ³ ³ N l ,T F l ,T dTdl . Therefore the
0 0
corrected expression for tensile strength of SFC by the rule of mixture
[40] is as follows

f T0
V 0 l l cos 4 T
V uc Qf ³³ h l g T dTdl  1  Q f V m (A 4)
ln 0
l

where T 0 cos 1 l n l .
Let us consider SFC with fibers of random two-dimensional
orientation. Then g T 2 S and the integral in (A 4) is given by

V 0 l §¨
f § 2· 2 ·
1 1 l n l ¨ 3  2§¨ l n ·¸ ¸ 1  §¨ l n ·¸ ¸h l dl
³ 3 cos  n (A 5)
4S ln
l ¨¨ l l ¨
© © l ¹ ¸¹ © l ¹ ¸¸
© ¹

For random three-dimensional orientation, g T sin T [40] and the

integral in (A 4) reduces to

5
1 V 0l ª § ln · º
f

5 l³ l «¬ © l ¹ »¼
«1  ¨ ¸ » h l dl (A 6)
n

67
 Paper B

References

1. G. Cantero, A. Arbelaiz, R. Llano-Ponte, and I. Mondragon. Compos. Sci.

Tech., 63, 1247 (2003).
2. T. Czvikovszky, H. Hargitai, I. Ràcz, and G. Csukat. Nucl. Instrum.
Methods Phys. Res. B, 151, 190 (1999).
3. S. K. Garkhail, R. W. H. Heijenrath, and T. Peijs. Appl. Compos. Mater.,
7, 351 (2000).
4. M. J. A. van den Oever, H. L. Bos, and M. J. J. M. van Kemenade. Appl.
Compos. Mater., 7, 387 (2000).
5. M. J. A. van den Oever, H. L. Bos, and K. Molenveld. Die Angewandte
Makromolekulare Chemie, 272, 71, (1999).
6. K. Oksman. Mechanical properties of natural fibre mat reinforced
thermoplastic. Appl. Compos. Mater, 7, 403 (2000).
7. B. Madsen and H. Lilholt. Compos. Sci. Tech., 63, 1265 (2003).
8. Nechwatal, K.-P. Mieck, and T. Reußmann. Compos. Sci. Tech., 63, 1273
(2003).
9. K. van de Velde and P. Kiekens. Compos. Struct., 54, 355 (2001).
10. K. van de Velde and P. Kiekens. Macromol. Mater. Eng., 286, 237 (2001).
11. Stamboulis, C. A. Baillie, S. K. Garkhail, H. G. H. van Melick, and T.
Peijs. Appl. Compos. Mater, 7, 273 (2000).
12. C. N. Singleton, C. A. Baillie, P. W. R. Beaumont, and T. Peijs.
Composites Part B, 34, 519 (2003).
13. K. van de Velde and P. Kiekens. Compos. Struct., 62, 443 (2003).
14. S. Wong, R. Shanks, and A. Hodzic. Compos. Sci. Tech., 64, 1321 (2004).
15. J. Biagiotti, S. Fiori, L. Torre, M. A. López-Manchado, and J. M. Kenny.
Polym. Compos., 25, 26 (2004).
16. J. Gassan. Composites Part A, 33, 369 (2002).
17. J. George, J. Ivens, and I. Verpoest. Die Angewandte Makromolekulare
Chemie, 272, 41 (1999).
18. D. G. Hepworth, D. M. Bruce, J. F. V. Vincent, and G. Jeronimidis. J.
Mater. Sci., 35, 293 (2000).
19. D. G. Hepworth, J. F. V. Vincent, G. Jeronimidis, and D. M. Bruce.
Composites Part A, 31, 599 (2000).
20. van de Weyenberg, J. Ivens, A. De Coster, B. Kino, E. Baetens, and I.
Verpoest. Compos. Sci. Tech., 63, 1241 (2003).
21. van Voorn, H. H. G. Smit, R. J. Sinke, and B. de Klerk. Composites Part
A, 32, 1271, (2001).
22. W. Thielemans and R. P. Wool. Composites Part A, 35, 327 (2004).
23. M. Baiardo, E. Zini, and M. Scandola. Composites Part A, 35, 703 (2004).
24. G. I. Williams and R. P. Wool. Appl. Compos. Mater, 7, 421 (2000).
25. Oksman, M. Skrifvars, and J.-F.Selin. Compos. Sci. Tech., 63, 1317
(2003).
26. O’Donnell, M. A. Dweib, and R. P. Wool. Compos. Sci. Tech., 64, 1135
(2004).
27. Lamy and C. Baley. J. Mater. Sci. Lett., 19, 979 (2000).
28. Andersons, E. SpƗrniƼš, R. Joffe, and L. Wallström. Compos. Sci. Tech.,
65, 693 (2005).

68
 Polymer Composites 27 (2006) 221-229

29. R. Joffe, J. Andersons, and L. Wallström. J. Mater. Sci., 40, 2721 (2005).
30. R. Joffe, J. Andersons, and L. Wallström. Composites Part A, 34, 603
(2003).
31. L. Tucker III and E. Liang. Compos. Sci. Tech., 59, 655 (1999).
32. H. Fukuda and Y. Takao. Thermoelastic properties of discontinuous fiber
composites. In: Kelly A, Zweben C, editors. Comprehensive composite
materials, vol. 1. New York: Pergamon Press; 2000.
33. J. L. Thomason and M. A. Vlug. Composites Part A, 27, 477 (1996).
34. H. Fukuda and K. Kawata. Fibre Sci. Tech., 7, 207 (1974).
35. Jayaraman and M. T. Kortschot. J. Mater. Sci., 31, 2059 (1996).
36. S.-Y. Fu and B. Lauke. Compos. Sci. Tech., 58, 389 (1998).
37. S.-Y. Fu, X. Hu, and C.-Y. Yue. Compos. Sci. Tech., 59, 1533 (1999).
38. F. R. Cichocki Jr. and J. L. Thomason. Compos. Sci. Tech., 62, 669 (2002).
39. A Kelly and W. R. Tyson. J. Mech. Phys. Solids, 13, 329 (1965).
40. H. Fukuda and T.-W. Chou. J. Mater. Sci., 17, 1003 (1982).
41. S.-Y. Fu and B. Lauke. Compos. Sci. Tech., 56, 1179 (1996).
42. B. Lauke and S.-Y. Fu. Compos. Sci. Tech., 59, 699 (1999).
43. J. C. Halpin and J. L. Kardos. Polym. Eng. Sci., 18, 496 (1978).
44. F. W. J. van Hattum and C. A. Bernardo. Polym. Compos., 20, 524 (1999).
45. Y. Termonia. J. Polym. Sci. B, 32, 969 (1994).
46. Dong, S. Schmauder, T. Bidlingmaier, and A. Wanner. Comput. Mater.
Sci., 9, 121 (1997).
47. K. Zhu and S. Schmauder. Comput. Mater. Sci., 28, 743 (2003).
48. B. N. Nguyen and M. A.Khaleel. Compos. Sci. Tech., 64, 607 (2004).
49. L. Mehan and L. S. Schadler. Compos. Sci. Tech., 60, 1013 (2000).
50. J. L. Thomason, M. A Vlug, G. Schipper, and H. G. L. T. Krikor.
Composites Part A, 27, 1075 (1996).
51. K. Oksman, R. Långström, B. Nyström, and K. Joseph. The influence of
fibre microstructure on mechanical properties and on fibre breakage
during the extrusion process of natural fiber reinforced thermoplastic
composites. TR SICOMP 01-016, 2002.
52. M. J. A. van den Oever and H. L. Bos. Adv. Compos. Lett., 7, 81 (1998).
53. S. W. Tsai and E. M. Wu. J. Compos. Mater., 5, 58 (1971).
54. S. W. Tsai and E. M. Wu. A General Theory of Strength for Anisotropic
Materials. Technical report AFML-TR-71-12, 1972, p. 61.
55. J. C. Halpin and J. L. Kardos. Polym. Eng. Sci., 16, 344 (1976).
56. M. Skudra. Micromechanics of failure of reinforced plastics. In: Handbook
of Composites, Vol. 3. Failure Mechanics of Composites, Eds. C. G. Sih,
A. M. Skudra, Elsevier Science Publishers B. V., 1985, pp. 1-69.

69