MITLibraries
Document Services
Room 14-0551
77 Massachusetts Avenue
Cambridge, MA 02139
Ph: 617.253.5668 Fax: 617.253.1690
Email: docs@mit.edu
http://libraries.mit.edu/docs
DISCLAIMER OF QUALITY
Due to the condition of the original material, there are unavoidable
flaws in this reproduction. We have made every effort possible to
provide you with the best copy available. If you are dissatisfied with
this product and find it unusable, please contact Document Services as
soon as possible.
Thank you.
Some pages in the original document contain pictures,
graphics, or text that is illegible.
BENDING FATIGUE AND CREEP
OF TOUGH MATRIX LAMINATES
by
RIZWAN MAHMOOD GUL
B.Sc (Mechanical Engineering); Peshawar, Pakistan
(1990)
submitted to
The Department of Materials Science and Engineering
in partial fulfillment of the requirement for the degree of
MASTER OF SCIENCE
IN MATERIALS SCIENCE AND ENGINEERING
at the
Massachusetts Institute of Technology
May, 1994
© Massachusetts Institute of Technology 1994
All rights reserved
Signature of Author
Department of Materials Science and Engineering
May 6, 1994
Certified by
Professf Polymer Engineering
Thesis Supervisor
Accepted by
Carl '). Thompson III
Professor of Electronic Materials
Chair, Departmental Committee on Graduate Students
IMASSAiCH iS'T.S
HNSTi
T;' t
2 I.
(,:.
AUG 18 1994
Science
BENDING FATIGUE AND CREEP
OF TOUGH MATRIX LAMINATES
by
RIZWAN MAHMOOD GUL
Submitted to the Department of Materials Science and Engineering
on May 6, 1993 in partial fulfillment of the requirement for the
degree of Master of Science in Materials Science and Engineering
ABSTRACT
A toughened polyester epoxy network (MNS) with a 10 fold increase in
fracture energy (GIc) by the addition of reactive rubber has been developed in
a previous study. The purpose of this study is to characterize the fatigue and
creep behavior of this polymer as a matrix, and to study the general effect of
toughening on these properties. Flexural fatigue (R=-1) is used for the tests.
The reinforcements used are chopped strand mat, glass fabric and carbon
fabric, while MNS is used as the matrix with different rubber contents (0%,
7.5%, 12.5% and 17.5%) and unmodified polyester for comparison. There is
much controversy about the effect of toughening of the matrix on the fatigue
behavior of composites. In this study it has been observed that the fatigue
behavior improves with increasing flexibility provided that the modulus
does not drop too much. The S-N curves are linear on a Log-Log plot and can
be characterized by the value of m (S/So = N-/m). The comparison of the
value of m for different rubber contents with a particular reinforcement
shows that the fatigue performance generally improves with increasing
rubber content; the increase is most obvious in the case of carbon composites.
Damage mechanisms also are discussed.
Creep tests have been performed in flexure, with chopped strand mat
and glass fabric, and MNS as the matrix with different rubber contents (0%,
7.5%, 12.5% and 17.5%). The tests shows that the creep does not increase to a
large extent with increasing rubber content. The performance has been
characterized by Schapery's non-linear constitutive equation. The difference
between the predicted and experimental values at 2188 hrs. ranges from 0.5%8.57% for glass fabric composites and ranges from 1.31-8.76% for chopped
strand mat composites.
Thesis Supervisor:
Title:
Professor Frederick J. McGarry
Professor of Polymer Engineering
2
Table of Contents
Abstract
Table of contents
List of tables
List of figures
Acknowledgment
(1) Introduction
1.1 Fatigue
1.2 Creep
(2) Materials
(3) Specimen
(4) Stress analysis
(5) Fatigue tests
5.1 Test equipment
5.2 Static tests
5.3 Fatigue results
5.3.1 Chopped strand mat composites
5.3.2 Glass fabric composites
5.3.3 Carbon fabric composites
5.4 Discussion of results
5.4.1 S-N behavior
5.4.2 Damage development
5.4.3 Loss of stiffness
(6) Creep tests
6.1 Test results
6.2 Creep characterization
6.3 Discussion of results
(7) Conclusion
(8) Suggestions for further improvements
References
Appendices:
Appendix I: Tables
Appendix II: Figures
3
2
3
4
5
10
11
11
16
18
20
22
25
25
25
27
27
30
33
36
36
38
40
41
41
42
47
49
53
151
54
60
List of Tables
Table
Table
Table
Table
Table
Table
Table
Table
Table
I
:Composition of the composites
54
II : Mechanical properties of the composites tested at 5.08 mm/min.55
III : Mechanical properties of the composites tested at 3378 MPa/sec 56
IV : Value of fatigue characterizing parameters, b and m for chopped
strand mat composites.
28
V : Value of fatigue characterizing parameters, b and m for glass fabric
composites.
31
VI : Value of fatigue characterizing parameters, b and m for carbon
fabric composites.
34
VII: Value of fatigue characterizing parameters, b and m for different
composites.
57
VIII: Relative creep deflection of the composites
58
IX : Value of different terms in Shapery's equation
59
4
List of Figures
Figure
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Normalized S-N curves for unidirectional composites with various
types of fibers, tested in longitudinal tensile fatigue.
60
Normalized S-N curves for cross ply composites with various types of
fibers, tested in reversed flexural fatigue.
60
Strain-N curve showing regions of different damage mechanisms, as
proposed by Talrega.
61
Loading diagram.
22
Flexural fatigue fixture.
61
A typical stress-strain curve for a composite tested in flexure at a rate of
5.08 mm/min.
62
Flexural modulus Vs. % rubber content in matrix for different
reinforcements (form the static tests performed at a ram speed of 5.08
mm/min).
63
Yield strength Vs. % rubber content in matrix for different
reinforcements (from the static tests performed at a ram speed of 5.08
mm/min).
64
Yield strain Vs. % rubber content in matrix for different reinforcements
(from the static tests performed at a ram speed of 5.08 mm/min).
65
A typical load-deflection curve for a composite tested in flexure at a rate
of 3378 MPa/sec.
66
S-N curve for chopped strand mat with unmodified polyester.
67
S-N curve for chopped strand mat with 0% rubber.
68
S-N curve for chopped strand mat with 7.5% rubber.
69
S-N curve for chopped strand mat with 12.5% rubber.
70
S-N curve for chopped strand mat with 17.5% rubber.
71
Strain-N curve for chopped strand mat with unmodified polyester. 72
Strain-N curve for chopped strand mat with 0% rubber.
73
Strain-N curve for chopped strand mat with 7.5% rubber.
74
Strain-N curve for chopped strand mat with 12.5% rubber.
75
Strain-N curve for chopped strand mat with 17.5% rubber.
76
5
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
Comparison of total stress S-N curves for chopped strand mat with
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
77
Comparison of normalized stress S-N curves for chopped strand mat
with unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
78
Comparison of initial strain-N curves for chopped strand mat with
unmodified polyester, and MNS with 0%, 7.5%, 12.5 and 17.5%
rubber.
79
Deflection Vs. No. of cycles curve for chopped strand mat with 7.5%
rubber.
80
Temperature Vs No. of cycles curve for chopped strand mat with 7.5%
rubber.
81
S-N curve for glass fabric with unmodified polyester.
82
S-N curve for glass fabric with 0% rubber.
83
S-N curve for glass fabric with 7.5% rubber.
84
S-N curve for glass fabric with 12.5% rubber.
85
S-N curve for glass fabric with 17.5% rubber.
86
Strain-N curve for glass fabric with unmodified polyester.
87
Strain-N curve for glass fabric with 0% rubber.
88
Strain-N curve for glass fabric with 7.5% rubber.
89
Strain-N curve for glass fabric with 12.5% rubber.
90
Strain-N curve for glass fabric with 17.5% rubber.
91
Comparison of total stress S-N curves for glass fabric with unmodified
polyester, and MNS with 0%, 7.5%, 12.5% and 17.5% rubber.
92
Comparison of normalized stress S-N curves for glass fabric with
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
93
Comparison of initial strain-N curves for glass fabric with unmodified
polyester, and MNS with 0%, 7.5%, 12.5% and 17.5% rubber.
94
Comparison of total stress S-N curves for all chopped strand mat and
glass fabric composites.
95
Comparison of normalized stress S-N curves for all chopped strand
mat and glass fabric composites.
96
Comparison of total stress S-N curves for glass fabric with epoxy,
unmodified polyester, MNS with 0%, 7.5%, 12.5% and 17.5% rubber. 97
6
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
Comparison of normalized stress S-N curves for glass fabric with epoxy,
unmodified polyester, MNS with 0%, 7.5%, 12.5% and 17.5% rubber. 98
S-N curve for carbon fabric with unmodified polyester.
99
100
S-N curve for carbon fabric with 0% rubber.
101
S-N curve for carbon fabric with 7.5% rubber.
102
S-N curve for carbon fabric with 12.5% rubber.
103
S-N curve for carbon fabric with 17.5% rubber.
104
Strain-N curve for carbon fabric with unmodified polyester.
105
Strain-N curve for carbon fabric with 0% rubber.
106
Strain-N curve for carbon fabric with 7.5% rubber.
107
Strain-N curve for carbon fabric with 12.5% rubber.
108
Strain-N curve for carbon fabric with 17.5% rubber.
Comparison of total stress S-N curves for carbon fabric with
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
109
Comparison of normalized stress S-N curves for carbon fabric with
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
110
Comparison of initial strain-N curves for carbon fabric with
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5%
rubber.
111
Comparison of total stress S-N curves for all glass fabric and carbon
112
fabric composites.
Comparison of normalized stress S-N curves for all glass fabric and
carbon fabric composites.
113
Comparison of total stress S-N curves for carbon fabric with epoxy,
unmodified polyester, and MNS with 0%, 7.5%, 12.5% and 17.5% rubber
114
115
Deflection Vs. No. of cycles for carbon fabric with 7.5% rubber.
Photograph of the damage zone in fatigue tests performed on chopped
-116
strand mat with 7.5% rubber.
Photograph of the damage zone in fatigue tests performed on chopped
strand mat with 12.5% rubber.
116
Photograph of the damage zone in fatigue tests performed on chopped
strand mat with 17.5% rubber.
117
7
63
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
Photograph of the damage zone in fatigue tests performed on glass
fabric with unmodified polyester.
117
Photograph of the damage zone in fatigue tests performed on glass
fabric with 7.5% rubber.
118
Photograph of the damage zone in fatigue tests performed on glass
fabric with 12.5% rubber.
118
SEM micrographs of the fracture surface of chopped strand mat with
7.5% rubber.
119
SEM micrograph of the fracture surface of chopped strand mat with
12.5% rubber.
120
SEM micrographs of the fracture surface of chopped strand mat with
17.5% rubber.
121
SEM micrographs of the fracture surface of glass fabric with unmodified
polyester.
122
SEM micrographs of the fracture surface of glass fabric with 7.5%
rubber.
123
SEM micrographs of the fracture surface of glass fabric with 12.5%
rubber.
124
SEM micrographs of the fracture surface of carbon fabric with
unmodified polyester.
125
SEM micrographs of the fracture surface of carbon fabric with 7.5%
rubber.
126
SEM micrographs of the fracture surface of carbon fabric with 12.5%
rubber.
127
SEM micrographs of the fracture surface of carbon fabric with 17.5%
rubber.
128
The deflection Vs. time curve in a creep test on chopped strand mat
with 7.5% rubber.
129
The deflection Vs. time curve in a creep test on chopped strand mat
with 12.5% rubber.
130
The deflection Vs. time curve in a creep test on chopped strand mat
with 17.5% rubber.
131
The deflection Vs. time curve in a creep test on glass fabric with 7.5%
rubber.
132
The deflection Vs. time curve in a creep test on glass fabric with 12.5%
rubber.
133
8
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
modulus Vs. time curve for chopped strand mat with 7.5%
134
modulus Vs. time curve for chopped strand mat with 12.5%
135
modulus Vs. time curve for chopped strand mat with 17.5%
136
modulus Vs. time curve for glass fabric with 7.5% rubber.
137
The creep modulus Vs. time curve for glass fabric with 12.5% rubber.
138
Isochronous stress-strain curve for chopped strand mat with 7.5%
rubber.
139
Isochronous stress-strain curve for chopped strand mat with 12.5%
rubber.
140
Isochronous stress-strain curve for chopped strand mat with 17.5%
rubber.
141
Isochronous stress-strain curve for glass fabric with 7.5% rubber.
142
Isochronous stress-strain curve for glass fabric with 12.5% rubber.
143
Predicted and experimental creep strain Vs. time curve for chopped
strand mat with 7.5% rubber.
144
Predicted and experimental creep strain Vs. time curve for chopped
strand mat with 12.5% rubber.
145
Predicted and experimental creep strain Vs. time curve for chopped
strand mat with 17.5% rubber.
146
Predicted and experimental creep strain Vs. time curve for glass fabric
with 7.5% rubber.
147
Predicted and experimental creep strain Vs. time curve for glass fabric
with 12.5% rubber.
148
Relative creep deflection Vs. stress level (on which the creep test is
performed) for different composite
149
Relative creep deflection Vs. % rubber content in MNS on different
stress levels.
150
The creep
rubber.
The creep
rubber.
The creep
rubber.
The creep
9
Acknowledgements
I acknowledge Prof. F. J. McGarry for his guidance and support during
my stay here. His comments gave me inspiration and encouragement,
leading me to work with the best of my abilities.
I acknowledge my parents for their patience, because I should be
serving them instead of giving them the agony of staying away.
I acknowledge Dr. Mary Chan, Dr. Ramnath Subramaniam, Dr. Peter
Flueler, Maria Raposo, Stephen and Arthur Rudolph for their help and
valuable suggestions. Specially Mary whom I should always be grateful for
teaching me many things and for being a friend in need.
10
Chapter 1
Introduction
Polymer composites are widely used because of their high strength and
stiffness to weight ratios, ease of manufacture and corrosion resistance.
Among the different matrices, epoxy and polyester are the most common.
Usually glass composites are made with polyesters because of their low cost,
but the drawbacks are that they are brittle and have very low fracture
toughness compared to epoxies. Many attempts have been made to increase
the toughness of polyester, but usually they result in a significant loss of
stiffness, strength and chemical resistance [1]. The search to make a tougher
polyester with a modest loss in other properties has led to the development of
an interpenetrating polymer network (I.P.N): a styrene cross linked polyester,
intimately combined on a molecular scale with a rubber toughened epoxy,
which has been named a Molecular Network System (MNS) [2]. The purpose
of this study is to characterize the fatigue and creep behavior of reinforced
MNS composites and to study the effect of toughening the matrix resin on
these properties.
1.1
Fatigue :-
Fatigue is defined as the loss in strength of a material due to repeated
cycles of loading and unloading. This study examines the fatigue behavior
under fully reversed flexure with the ratio of minimum to maximum stress
(R) equal to -1; this is also called bending fatigue or full flexural fatigue. The
reasons for choosing bending fatigue are as follows:
11
(1)
Components and structures made of composites are commonly
subjected to flexural fatigue. For example in the case of boats, the surfaces in
contact with water are subjected to repeated pounding of waves, producing
flexural fatigue. In aircraft the wings and control surfaces are subjected to
wind induced flutter and engine noise producing flexural fatigue [3]. In wind
turbine blades the rotation and changing wind direction produce flexural
fatigue.
(2)
In flexural fatigue the behavior of a composite is matrix dependent.
This is because in flexural fatigue a compression component is involved
which leads to matrix dependent properties. Since the purpose of this work is
to study the effect of toughening of the matrix on the fatigue behavior, it was
logical to use flexural fatigue.
The fatigue of composites is different from that of metals. In metals a
single crack is formed near the end of the sample life and that crack
propagates across the entire cross section causing failure of the specimen. In
composites, fatigue loading causes multiple damage sites and the damage
usually begins to appear at the start of sample life, even on the first loading
cycle [4]. The damage is in the form of matrix cracking, delamination, fiber
breakage and interfacial debonding. Normally, fibers lying transversely with
respect to the stress axis serve as sites for the initiation of debonding followed
by matrix cracking and fiber breakage or delamination [5]. This sequence is
not universal and it may vary widely depending on many factors. In fact, the
fatigue behavior depends on the type of matrix, the fibers, stacking sequence,
the length of fibers, fiber content, void content, the method of processing, the
R value and on other factors. Adding to the complexity is the fact that it may
be difficult even to define the fracture criteria. Fracture in composites does
12
not always mean complete separation of the specimen and in some cases
physical integrity is maintained long after the load carrying capacity of a
specimen has been exhausted. Usually failure is defined either when a surface
crack penetrates across the whole width or when the initial stiffness is
reduced by 20-40% [6].
There is controversy about the effect of rubber toughening of the matrix
on the fatigue behavior of composites, but usually the fatigue behavior
degrades with flexibilizing the matrix. Although it may be expected that an
increase in fracture toughness from the use of rubber particles leads to a
decrease in fatigue degradation rate, this is not what is always observed.
Murakami et al [7] studied the tensile fatigue behavior of neat resins blended
with NBR. He found the fatigue life as measured in the S-N curve decreased
with a increase in rubber content, the slope of S-N curve improved
somewhat. Konur et al [11] reported improved fatigue life in unidirectional
composites when rubber was used to toughen the epoxy matrix. Curtis [10]
observed that in unidirectional composites, toughened epoxies improved the
static strength but the fatigue behavior was degraded, with steeper Strain-N
curves. Mandell
et al
[12] studied the tensile fatigue behavior of
unimpregnated glass strands and of their composites with polyester, epoxy
and rubber modified epoxy, and found that the S-N slope for all of the
composites and for the unimpregnated strands were about the same:
approximately equal to 10. He also plotted the S-N curve slopes for a variety
of materials and showed they were approximately equal to 10. In relation to
increasing the flexibility of the matrix, Hertzberg [8] cited two references for
glass-fabric-reinforced laminates; in one, toughening the epoxy led to better
fatigue life, while in the other, flexibilizing a polyester resin neither
13
improved the fatigue life nor the fracture toughness. Joneja [9] studied the
effect of adding flexible polyester to standard polyester resin on the fatigue
behavior of unidirectional composites. He found the fatigue life and the slope
of the S-N curve decreased with a high quantity of flexible polyester present.
With unidirectional composites loaded parallel to the fibers in tensile
fatigue, high modulus fiber composites are much more fatigue resistant than
low modulus ones: in the S-N curves, the slope of glass fiber composites (low
modulus) is much steeper than that for carbon fiber composites (high
modulus). However when a composite made of off-axis plies is subjected to
tensile fatigue or when a composite with any stacking sequence is subjected to
flexural fatigue (R=-1), all the composites behave in approximately the same
way: the glass composites have the same slope as carbon composites. Studying
the environmental fatigue behavior, Jones et al [13] pointed out that in tensile
fatigue of unidirectional composites the behavior improved with increasing
modulus of the composite [Fig. 1], but he also observed that for +450
laminates, both glass fiber and carbon fiber composites had exactly the same
fatigue life and slope. According to Mandell [14], with unidirectional
composites loaded along the fiber direction, high modulus materials are more
fatigue resistant than low modulus ones, but with off-axis plies or when
loading in compression, the modulus does not affect the fatigue behavior,
which approximately follows the glass composite trend line. Rotem [14]
studied the fatigue behavior of graphite/epoxy laminate for different R values
(R=0.1, 10 and -1) and for two lay-ups (00/900and +450). He found that for
R=0.1 the S-N slopes for +450and 00/900laminates are 9.45% and 1.11%
respectively. For R=10 the slopes for +450and 00/9001laminates are 5.5% and
7.11% respectively. Interestingly however, for R=-1, the slope for +450and
14
00/900laminates are 11.22% and 10.02% respectively, which are similar.
Finally, Osborn [3] tested a variety of specimens with stacking sequences of
+250, +200, +100 and woven cloth, for a variety of fiber types including S glass,
E glass and Graphite, under exactly our testing conditions. He found all these
various specimens followed approximately the same S-N slope [Fig. 2]; since
the only common factor was the matrix, he concluded that the fatigue
behavior was matrix dependent.
To represent the complex nature of the fatigue of composites, Talrega
[5] developed a critical strain model based on Strain-N curves. It characterizes
fatigue strain ranges according to different damage mechanisms. He contrasts
the high strain range in which fiber breakage and interfacial debonding are
predominant, to the low strain range where matrix cracking and interfacial
shear failures control [Fig 3]. His model is a conceptual framework for
comparing fatigue behavior of composites with different constituents and for
characterizing the damage mechanisms [11]. In our study the tests emphasize
Stress-N curves instead of Strain-N curves, but the Stress-N curves can be
transformed by using the initial modulus of the samples.
We tested chopped strand glass mat , glass fabric and carbon fabric
composites, all made with MNS (rubber modified polyester). Glass and carbon
fabric composites made with unmodified polyester also were tested for
comparison. Different fatigue curves are characterized and compared by the
value of b, where b is the slope of S-Log N curve normalized by the single
cycle strength of the composite:
b
S/S - 1
log N
15
Here N is the number of cycles required to break the specimen at stress
S, and S is the one cycle strength of the material. The higher the value of b,
the more rapid is the fatigue degradation. Since the S-N curves are not linear
(indicating a change in failure mechanism), they were also characterized and
compared by the value of m, where 1/m represents the slope of the linear
Log-Log plot and is defined by the following equation:
S =N-m
So
The smaller the value of m, the more rapid is the fatigue degradation.
1.2
Creep :-
Increasingly, composite materials are used in applications, where creep
is important: the increase of deformation with time under constant load. The
magnitude of creep deformation depends on laminate lay-up, stress level,
humidity, temperature and the previous thermal history [27]. Most creep
studies are performed in tension or shear, while the flexural mode, which is
important in many cases, has received little attention. In this study the creep
tests were performed in bending.
Unidirectional composites tested axially in tensile creep show less time
dependent response, because the properties are fiber dominated. When the
matrix properties control, as in a crossply composite in tension or any
composite in compression, significant viscoelasticity becomes evident [16].
Composites both with short and long fibers exhibit non-linear visco-elasticity
[17]. The common ways to describe this non-linear creep are (i) the modified
Boltzmann superposition principle, (ii) the multiple integral representation
16
(iii) the Findlay procedure and (iv) Schapery's constitutive equation [18]. Of
these, the Schapery constitutive equation is most used for composites, and it
is the one used in our study.
17
Chapter: 2
Materials
Three different reinforcements, chopped strand mat, glass fabric and
carbon fabric were used with unmodified polyester and with MNS of varying
rubber contents to form the composites. The rubber contents used were 0%,
7.5%, 12.5% and 17.5% of the total resin weight.
Chopped strand mat has randomly oriented two inch long glass fibers,
of 15 micron diameter. The glass fabric was 181 style with a satin weave
[Boatex], while the carbon fabric was G117 style with a satin weave [Textile
Technologies]. The carbon fiber is a Hercules AS4G6K with a epoxy
compatible sizing, while the glass fiber has a Volan coating. The reasons for
using the woven fabric were easier handling and the more isotropic in plane
properties, compared to uni-directional reinforcement. The MNS resin was
composed of unsaturated polyester (MR13006 from Aristech Corporation),
liquid epoxy (Epon 828 from Shell Chemical), styrene monomer (Aldrich
Chemical) and an amine terminated butadiene acrylonitrile rubber (ATBN
1300x16 from B. F. Goodrich Co.). Catalysts were Ancamine, a tertiary amine
salt and t-butyl peroxybenzoate. A description of the composites is given in
Table. 1.
All the composites are made by the wet, hand lay-up technique. In the
case of glass composites the stacking sequence was the aluminum mold plate
covered with non-porous Teflon sheet, the laminate, a peel ply, a aluminum
press plate, a breather fabric and the Mylar vacuum bag. Alternate layers of
glass mat or glass fabric and resin were applied, and the resin was worked into
18
the fibers with a corrugated roller. The roller reduces voids between the
laminae. Eight layers of glass mat and thirteen layers of glass fabric were used.
When all the layers had been applied, the lay-up was sealed in the vacuum
bag by tape and degassed under a vacuum of 17-18 inches of Hg for about five
minutes. The vacuum removed the entrapped air and voids; the lay-up was
then cured at 100 OC for 10 hrs. under atmospheric pressure. The fabrication
of the carbon composites was the same with a few exceptions. With only one
vacuum bag the laminate was full of voids, so two bags were used to simulate
the autoclave process. To the inner bag a vacuum of 18 inches of Hg was
applied for five minutes to remove the entrapped air and then to the outer
bag a vacuum of 20-25 in of Hg was applied for 10 minutes. This reduced the
void content. Another problem with the carbon composite was poor surface
finish, so a new stacking sequence was used. This was mold plate, non-porous
Teflon sheet, laminate, porous Teflon sheet, bleeder ply, non-porous Teflon
sheet, press plate, breather ply, first vacuum bag, breather ply and the second
vacuum bag. Since all the composites were cured at atmospheric pressure the
void content was expected to be higher than for autoclave or press cures.
19
Chapter: 3
Specimens
The specimen were 5.5 in long by 1 in wide and the support span was 4
in. The thickness ranged from 0.11 to 0.18 in. The size of the specimen was
chosen according to ASTM standard D790 [19], so that it can be considered as a
beam. According to the standard the span should be at least 16 times the
thickness of the specimen; a higher ratio reduces the shear stress in the center
plane of the beam. The ratio in our specimens varied from 22 to 32. According
to the standard, the width should not exceed 1/4 of the span and in our case it
was is exactly 1/4 of the span. Overhang should be sufficient to prevent the
specimen from slipping through the supports, at least 10% of the support
span. In our case the overhang was around 19%. Other considerations in
choosing the specimen size involved effective handling in the testing fixture
and convenience of machining.
The specimens were cut from the plate with a diamond saw, which was
water cooled. The machined surfaces of the cut specimens were then wet
sanded, ground and dried in an oven. The specimens were stored in a
desiccator until testing to reduce moisture absorption effects. In the fabric
composites, we tried to keep a proper alignment of the weave along the
specimen length, but some distortion occurred in fabrication and some
misalignment took place in the cutting process.
The surface of the glass composites was very rough so they were wet
ground and then a thin layer of resin was applied to cover the fibers so
20
exposed, and then cured. This surface finishing did not appear to affect the
fatigue behavior but it was not necessary with the carbon composites.
21
Chapter: 4
Stress Analysis
4.1
Stresses in the specimen
The specimen is considered to be a homogeneous, elastic beam loaded
in three point bending [Fig. 4]. The shear stresses in the center of the specimen
are neglected. The maximum stress in the outer fibers, as defined by ASTM D
790 is given by:
3PL
2bd 2
Where:
a = Stress in the outer fibers at mid span, (Pa)
P = Load at a given point on the load-deflection curve,
L = Support span, (m)
b = Width of beam tested, (m)
d = depth of beam tested, (m)
P Load
L
P/2
Support span
Fig : 4: Loading diagram
The bending modulus of the specimen is calculated as:
E=P
PL3
46d 3b
22
P/2
Where 8 is the deflection in the center of the specimen. The maximum strain
in the outer fibers also occurs at mid span, and is:
=-
E
4.2
Loading rate
Force control, constant amplitude, sine wave loading was used in
testing the specimens. A constant stress was applied to the specimen and the
stiffness degraded due to fatigue. Most polymer composites behave in a viscoelastic manner: their response to a given deformation depends on the loading
rate. Thus if all the specimen are tested at the same frequency the specimens
tested at high stress will experience a higher loading rate than the specimen
loaded to lower stress. This will cause scatter in the results so to reduce this,
all the specimens were tested at the same loading rate but somewhat different
frequencies. The frequency for a particular stress level can be calculated from
the expression for simple harmonic motion. The displacement, S, of a particle
rotating at angular speed co with amplitude A is given as:
S = A Sin ot
The velocity V can be calculated as:
V= Ao Cos cot
Velocity will be maximum at t=O and since o =2 f:
V = 2i7fA
The amplitude A of the sinusoidal loading wave represents stress level, f is
the frequency and V the loading rate. So the frequency can be expressed as:
V
f
2no
23
A constant loading rate is also advantageous because at low stress levels
the test frequency is higher so the time duration of the test is reduced. The
loading rate was chosen as 3378 MPa/sec so our results could be compared to
those of Osborn [3] who used the same rate.
4.3
Failure Definition
As mentioned earlier it can be difficult to locate the failure point in
composites since the material fails progressively as it softens under fatigue.
According to ASTM D 671 [20] fatigue failure is said to occur when the
modulus decreases to 70% of its original value. In the case of the chopped
strand mat composites we observed that cracks on the surface appeared to
penetrate through the entire thickness when the modulus dropped to 60% of
its original value, so we took this condition to define failure: a 40% decrease
in the modulus of the sample.
24
Chapter: 5
Fatigue Tests
5.1
Test Equipment
The fatigue tests were performed on a servo-controlled Instron 8500
machine using a special fixture. The machine has automatic load control to
cope with substantial changes in the specimen stiffness. The deflection is
obtained directly from the machine, to calculate modulus and strain. All the
tests are
performed in a uncontrolled
laboratory atmosphere,
at
approximately 20 OC. A fan was used to avoid any heating of the specimen
from hysterises.
The fixture was specially designed and fabricated for use in reversed
flexural fatigue [Fig. 5]. The frame is made of aluminum but the specimen
support rollers are hardened steel, in bearings. Each support point has a pair
of rollers, one of which can move to accommodate specimens of different
thickness. Nylon bushings on the rollers prevent any sideways drift of the
specimen, while nylon bumper plates keep it from displacing along its length.
The rollers are 17.46 mm in diameter.
5.2
Static Tests
Static three point bend tests were performed at two different loading
rates on Instron 4505 and Instron 8500 machines. First they were done at a
cross-head rate of 5.08 mm/min, to determine the yield strength, yield strain,
and modulus of all the composites, as presented in Table 2 (a typical stress25
strain curve is shown in Fig. 6). Three specimens were tested for each
material and fracture was due to fiber breakage. The results show that
modulus and strength decrease with increasing rubber content, except for the
carbon composites where the absence of this trend may be due to varying
quality and fiber content of the different laminates [Fig. 7]. In general the yield
strains and stresses of the MNS composites are greater than with the
unmodified polyester, but the yield strain does not vary much with
increasing rubber content [Fig. 8 & 9].
Because the fatigue tests were done at a rate of 3378 MPa/sec, the static
strength parameters measured at the same rate are shown in Table 3. A typical
load-stroke curve is shown in Fig. 10. Comparing Table 2 and Table 3 it is clear
that both modulus and strength increase appreciably with rate. With chopped
strand mat the average increase in strength is around 110%, with the glass
fabric it is 80% and with the carbon fabric it is 50%. Comparing Fig. 6 and Fig.
10 it can be observed that with the higher loading rate the load-deflection
curve is straight and there are no knees, as observed at the low rate because of
ply breakage.
26
5.3
Fatigue Tests
5.3.1 Chopped Strand Mat Composites
Fatigue tests were done with unmodified polyester and four different
rubber contents: 0%, 7.5%, 12.5% and 17.5%. At each stress level three
specimens of the 7.5% rubber content were tested to observe the scatter. As
seen in Figurel3, the scatter was modest, suggesting that both the procedure
and the specimens were consistent. The S-N curves are presented in Figures
11-15 and the
-N curves, based on the initial modulus, are shown in Figures
16-20. As shown in Table 4 the loss of strength per decade of cycles (parameter
b) on a linear scale, for the laminates with unmodified polyester, 0%, 7.5%,
12.5% and 17.5% rubber is 14.74, 15.36, 13.27%, 13.58% and 13.67% respectively.
There is a change in slope at 104 cycles. Table 3 show the b value before and
after 104 cycles. The value of m (the inverse of the drop in strength per decade
of cycles on a Log-Log plot) for the laminates with unmodified polyester, 0%,
7.5%, 12.5% and 17.5% rubber is 7.86, 7.44, 6.96, 7.43 and 7.59 respectively. The
comparison of the value of m for laminate with 7.5% rubber, with the
laminates with unmodified polyester and 0% rubber show that the
degradation rate increases with the addition of rubber (the smaller the value
of m the more rapid is the degradation). But as we further increase the rubber
content the degradation rate decreases. The overall decrease and increase in
degradation rate is small. From the values of b it is clear that before 104 cycles,
the degradation rate increases with increasing rubber content (the larger the
value of b the more rapid is the degradation), which may reflect the lower
static strength trend with increasing rubber content. The decrease in slope
after 104 cycles is large and this may signal the approaching endurance limit.
27
In Fig. 21 the S-N curves with different rubber contents are compared and in
Fig. 22 the normalized curves are presented. The
-N curves are shown in
Fig. 23. It can be seen that the rubber level has little effect on the fatigue
strength behavior but the strain-number of cycles performance is improved
by the rubber.
b = S/SO -1
logN
Specimen
Log(N)
Log(S/So)
for whole
upto 104
after 104
life, %
cycles, %
cycles, %
14.74
15.36
13.27
13.58
13.67
17.22
18.83
19.46
18.89
21.39
6.25
5.93
4.97
5.14
4.31
for whole life
Chopped Strand mat
Unmodified polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
7.86
7.44
6.96
7.43
7.59
Table : 4 :- Value of fatigue characterizing parameters, b and m for chopped
strand mat composites.
As seen in Figures 60-62 the damage zone can be observed by whitening
of the specimens and this increases, reaches a maximum, and then decreases
as the cycles increase. The maximum occurs at around 104 cycles and this
suggests that the change in slope there is due to a change in damage
mechanism. Microgrpahs taken of the side (free edge) of the fracture surfaces
are shown in Fig. 66, Fig. 67 and Fig. 68 for laminates made of chopped strand
mat with 7.5% ,12.5% and 17.5% rubber. The fracture surface was polished in
the longitudinal direction, then coated with a thin layer of gold and examined
in the S.E.M. (Cambridge Instruments) using back scattered electrons. The two
28
micrograph for each laminate show the top and bottom edges. The primary
mode of failure is similar in all three cases: the crack starts at the surface due
to iterfacial debonding or matrix cracking in the resin rich area, then
propagates inward by matrix cracking and some fiber breakage. Sometimes the
crack moves longitudinally, because of the toughness of the matrix, until it
finds a weaker location or until it reaches the end of the strand.
Since interfacial debonding is the dominant mode of failure, we have
modified the Talrega diagram [Fig. 3] as shown in Figures 31-35. The top
horizontal band determined from the static strain to failure represents fiber
breakage, and the sloped data refers to transverse fiber debonding and matrix
damage. The lower horizontal band is the limiting strain for the onset of
transverse fiber debonding,
d.b.. This was taken to be 0.12%, as obtained by
plotting the numbers of cycles for the onset of debonding against the
maximum cyclic strain and recommended by Talrega for short fiber and cross
laminate composites. According to this interpretation, the £-N curves for all
the laminates tested show that failures are based on matrix cracking and
transverse fiber debonding; this is consistent with the fracture mechanisms
evident in the micrographs.
The increase in deflection with cycles was observed for a few specimens
as shown in Fig. 24 (In all of these cases the failure was defined as a 40%
decrease in modulus). The curves can be divided into three stages. In the first
(I) the modulus drops rapidly, then reaches a steady state (II). In the final
stage, (III), near the end of fatigue life, again the modulus decreases rapidly.
29
In a few specimens, the temperature increase was monitored [Fig. 25];
increases of only a few degrees centigrades were found and these were judged
to be inconsequential.
5.3.2 Glass Fabric Composites
Fatigue tests were performed on laminates with unmodified polyester,
and with 0%, 7.5%, 12.5% and 17.5% rubber contents. Two specimen of the
7.5% rubber laminate were tested on each stress level to confirm that the
scatter was not large. The S-N curves are shown in Figures 26-30 and
-N
curves, based on the initial modulus, are shown in Figures 31-35. The value
of b [Table 5] for laminates with unmodified polyester, 0%, 7.5%, 12.5% and
17.5% rubber is 13.49%, 13.21%, 13.94%, 12.99% and 13.59% respectively. Here
the change in slope occurs around 105 cycles instead at 104 cycles as was the
case with the chopped strand mat. Table 5 show the b value before and after
105 cycles. The b values for the different composites are nearly the same;
adding 7.5% rubber increased it somewhat, but with 12.5% rubber it decreased.
One interesting point to note is that the unmodified polyester composite,
after 10 5 cycles, degrades at a faster rate. than the MNS composites. This
means that the latter have better long term behavior. The decrease in slope
for the glass cloth composites after 105 cycles is not as large as for the chopped
strand composites which means that the endurance limit is farther away.
The value of m for laminates made with unmodified polyester, with
0%, 7.5%, 12.5% and 17.5% rubber is 9.03, 8.77, 8.09, 9.14 and 9.55 respectively,
which shows the same trends as the value of b. The S-N curves based on
30
stress are compared in Fig. 36 and the normalized curves are shown in Fig. 37.
The MNS composites perform better at low cycles but the curves converge at
high cycles. The difference was much less in the normalized curves. The
-N
curves for the different composites are compared in Fig. 38 where the
advantage of the MNS system is clear. This correlates with the yield strains of
these laminates [Table. 2].
b
S/So- 1
_
for whole
upto 105
after 105
life, %
cycles, %
cycles, %
Unmodified Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
13.49
13.21
13.94
12.99
15.67
15.23
15.01
14.65
7.61
6.29
7.13
6.69
9.03
8.77
8.09
9.14
MNS 17.5% Rubber
13.59
14.93
3.20
9.55
Specimen
Log(N)
Log(S/So)
logN
for whole life
Glass Fabric
Table : 5 :- Value of fatigue characterizing parameters, b and m for glass fabric
composites.
All of the glass composites are compared in Fig. 39 and the normalized
S-N curves are shown in Fig. 40. The fabric performs better than the chopped
mat, but the differences become smaller in the normalized data. In the latter,
at low cycles the fabric is superior, but the two types converge at high cycles.
Since we did not test glass fabric/epoxy composites in our study, data from the
literature were used for comparison. Osborn [3] performed such tests under
exactly the same condition and Fig. 41 shows that the MNS curves are above
31
the epoxy and the unmodified polyester curves. On a normalized basis [Fig.
42], all the composites show similar behavior.
The size of the damage zones, revealed by whitening of the specimens
[Fig. 63-65] varies with the number of cycles: at first it increases, reaches a
maximum and then decreases as the cycles increase. This is most obvious in
the 7.5% rubber composites. The maximum zone occurs around 105 cycles,
where the change in slope of the S-N curve is seen. Micrographs taken of the
sides of the fracture surfaces are shown in Fig. 69, Fig. 70 and Fig. 71 (The two
micrographs for each specimen show the top and bottom surface of the edge).
They show the primary mode of failure in the MNS composites is different
from the unmodified polyester ones: interfacial debonding and matrix
cracking, while in the case of unmodified polyester the fracture is
delamination in the center caused by the interlaminar shear stress. In the
MNS composites, the crack appears to start at the surface by interfacial
debonding of transversely oriented fibers or by breakage of cross ply strands. It
then extends inward due to matrix cracking, breaking cross ply strands in the
process. When the crack encounters longitudinal ply strands, either it breaks
them or passes around them. In the unmodified polyester laminate while
there is some damage on both the surfaces the primary cause of failure is the
delamination in the center.
The Talrega concept has been applied to the glass fabric results, as
shown in Figures 31-35. The data suggest that the failures are due to
transverse fiber debonding and matrix cracking (with delamination in case of
the pure polyester): this is consistent with the fracture mechanisms evident
in the micrographs.
32
5.3.3
Carbon Fabric Composites
Fatigue tests were performed on laminates based on unmodified
polyester and 0%, 7.5%, 12.5% and 17.5% rubber. The S-N curves for each are
shown in Fig. 43-47 and
-N curves are shown in Fig. 48-52. The value of b
[Table 6] for laminates made of unmodified polyester and with 0%, 7.5%,
12.5% and 17.5% rubber is 12.23%, 12.22%, 11.35%, 12.38% and 9.18%
respectively. At 105 cycles there is change in slope, but with very few points
beyond 105, only the b up to 105 cycles can be reported accurately, as shown in
Table 6. The values of b show that rubber modified composites degrade at a
slower rate as compared to unmodified polyester and 0% rubber composites.
Generally the performance improves with increasing rubber content, except
in the case of the 12.5% material which has a lower yield strain [Table 3]. The
values of m for unmodified polyester, and with 0%, 7.5%, 12.5% and 17.5%
rubber are 11.37, 11.07, 12.69, 11.66 and 17.79, respectively, which shows
explicitly that the fatigue performance improves with increasing rubber
content. The S-N curves for different composites are presented in Fig. 53 and
normalized in Fig. 54 (The low strength of the 17.5% rubber laminate may be
due to a higher matrix content or poor quality of manufacture). These
normalized curves for the carbon composites show a behavior different from
glass composites; in the latter, all the curves were nearly the same (see Fig.
37). The
-N curves are presented in Fig. 55. The curve for the 7.5% rubber
material is best and if the yield strains are considered (Table 2), it is seen the
performance increases directly with increasing yield strain.
33
Log(N)
Log(S/So)
b = S/S O - 1
logN
for whole
upto 105
after 10 5
life, %
cycles, %
cycles, %
Unmodified Polyester
12.23
15.00
11.37
MNS 0% Rubber
12.22
12.908
11.07
MNS 7.5% Rubber
11.35
12.24
12.69
MNS 12.5% Rubber
12.38
13.45
11.66
MNS 17.5% Rubber
9.18
9.97
17.79
Specimen
for whole life
Carbon Fabric
Table : 6 :- Value of fatigue characterizing parameters, b and m for carbon
fabric composites.
The carbon and glass fabric composites are compared in Fig. 56 and Fig.
57. With respect to stress, the trend is as follows: carbon with 7.5% rubber is
the top, followed by carbon with 12.5% rubber, carbon with 0% rubber, carbon
with 17.5% rubber, carbon with unmodified polyester, glass with 7.5% rubber,
glass with 12.5% rubber, glass with 0% rubber, glass with unmodified
polyester and glass with 17.5% rubber. For the normalized data the following
trend is observed: carbon with 17.5% rubber, carbon with 7.5% rubber, carbon
with 12.5% rubber, carbon with 0% rubber, carbon with unmodified polyester
and then all the rest. In Fig. 58, a plain weave carbon with epoxy composite
(reported by Chou et al [21]) tested in cantilever fashion performs better than
the other composites, but it appears to degrade at a much higher rate: its slope
is steeper.
Micrographs of the side of the fracture surface are shown for laminates
of carbon cloth and unmodified polyester [Fig. 72], with 7.5% rubber [Fig. 73],
34
12.5% rubber [Fig. 74] and 17.5% rubber [Fig. 75]. The two micrographs for each
material show the top and bottom surfaces at the edge. The primary mode of
failure combines interfacial debonding and matrix cracking and there is much
more debonding in the carbon composites than in the glass ones. The carbon
fibers have an epoxy compatible sizing so the fiber may not adhere well due to
the large amount of polyester present in the MNS. Apparently the damage
starts at the surface due to interfacial debonding, then propagates inward by
matrix cracking, interfacial debonding and infrequent fiber breakage.
Talrega diagrams are presented in Figures 48-52, and the curves show
that failure is due to interfacial debonding and matrix cracking, which is
consistent with the micrograph study.
The decrease in modulus as represented by an increase in deflection
was observed for a few specimens, as shown in Fig. 59. One sample was tested
until it reached the maximum deflection the test fixture could handle, while
the other two were taken to a 40% decrease in modulus. All the curves are
similar to those observed for chopped strand mat: they can be divided in three
main stages. In the first (I) the modulus drops rapidly and reaches a steady
state second stage (II). Finally, near the end of the fatigue life the modulus
deteriorates rapidly (III).
35
5.4
Discussion of results
5.4.1
S-N behavior
Improved fatigue performance in composites with higher modulus
fibers has been widely observed [11,14]. Since the modulus of the MNS
composites decrease with increasing rubber content [Table 2], it could be
expected that the fatigue performance might degrade. This was not observed.
According to Mandell [14], in short fiber glass composites, if the matrix is
fatigue resistant, it is possible that the behavior may be fiber dominated. This
was observed both for the chopped strand mat and the glass fabric/MNS
composites. When the normalized curves for different glass mat composites
are compared [Fig. 15], all superpose. Similarly when the normalized curves
for the glass fabric composites are compared [Fig. 26], all of them tend to
follow the same line. Since the only common factor in these cases is the glass
fibers, it can be concluded that the fatigue behavior was fiber dominated and
was limited by the performance of the fibers. Matrix variations did not change
this.
Carbon fibers have excellent fatigue properties, in contrast to glass [11].
Because of this, the fatigue behavior of carbon composites will be matrix
dominated. Figure 41 illustrates this: the tougher the matrix the better the
fatigue resistance . As the strain to failure in a static test increases, the fatigue
resistance improves.
Fig. 57 presents the normalized data for all the fabric based composites
tested. The carbon fabric composites have improved fatigue behavior
36
compared to glass fabric composites. This is a marked difference from the
observation made by Osborn [Fig. 3], in which all the composites had the same
normalized S-N curve. Similarly according to Mandell [14], with off axis plies
or in compression loading the high modulus carbon composites tended to
follow the glass trend line. In our study, because of yarn crossovers, weave
distortions and the presence of the compression component, the behavior
was expected to be similar to Osborn's: i.e. carbon composites performing the
same way as glass. The observation that the carbon composites performed
better than glass, however, is significant. In unidirectional composites, carbon
performs better than glass, but both perform similarly in off-axis laminates
and in compression loading. The off-axis plies fail at low strains and develop
cracks before the plies in the dominant stress direction fail. With the MNS
the strain to failure increases, so the fatigue resistance of the carbon fibers
operates and gives a different slope from the glass composites. The fact that an
increasing strain to failure improves the fatigue behavior has been observed
previously. Hayashi et al [22] found with aramid fiber/epoxy composites
(£f=1.5%) the rate of decrease of the maximum bending moment was slower
than that for carbon fiber/epoxy composites (f=0.8%).
Similarly Newaz [23]
found that a glass/epoxy composite had better fatigue performance compared
to a glass/vinyl ester one, especially at high cycles because of the higher strain
to failure of the glass/epoxy.
Because of the change in slope of the S-N curves it was difficult to fit a
straight line (S/S o = 1 - b log N), but when the data were plotted on a Log S-Log
N basis a linear relationship is obtained: (S/S o = N -Ym). Hence m is a fatigue
parameter and as its value increases it denotes improved fatigue
performance. For both chopped strand mat and glass fabric composites the
37
fatigue improvement required 12.5% rubber, and the 7.5% rubber did not
show it, when we compare the performance with unmodified polyester. Since
the glass fibers are Volan coated, this effect may be present because of better
adhesion of glass fibers with polyester, as compared to the adhesion with
MNS. With further increase in rubber content (17.5%)
the fatgiue
performance improves in both types of glass composites. With the carbon
fabric, the m value showed the fatigue improvement with increasing rubber
content. If Strain-N curves are examined (Figs. 23, 38 and 55) are examined, it
will be seen that a consistent behavior prevails: the higher the strain to
failure in the static test at 3378 MPa/sec, the better is the fatigue resistance.
Thus the use of the rubber modification to toughen the matrix and to increase
its ability to plastically deform, also provides better fatigue resistance in the
fiber reinforced composites.
5.4.2 Damage development
Reversed flexural fatigue represents the most severe loading condition,
because of the compression component involved. In compression, local resin
and interfacial damage lead to fiber instability which is more severe than the
fiber isolation mode which occurs in tensile loading [11]. The damage
development as observed from the micrographs for chopped glass mat [Fig.
66-68], glass fabric [Fig. 69-71] and carbon fabric [Fig. 72-75] is similar, except for
the glass fabric/unmodified polyester composite. Usually the damage starts on
the surface with fiber/resin debonding of the transverse fibers. With
increasing number of cycles, matrix cracking develops and some fiber
breakage occurs. As the cycling continues, interfacial debonding and matrix
cracking develop in strands oriented in the stress direction and the damage
38
moves toward the center of the thickness. Crack propagation through a fiber
causing its breakage seems to follow the explanation presented by Joneja [9]: as
interfacial debonding propagates along the fiber in the matrix which has low
strength and modulus but high ductility, a high stress gradient develops at
the microcrack tip and this searches out a weak segment of the fiber. Such
explains the poor fatigue performance of some of the flexible matrix
composites previously cited in section 1.1. This poor performance may be due
to the large modulus drop which accompanies the increased flexibility for
these polymers; in the MNS resins the modulus does not decrease so much
and it causes the fatigue performance to improve instead of degrading.
The improvement in fatigue performance of the MNS composites is
due to the delay of interfacial debonding and matrix cracking in high cycle
fatigue. As observed by Newaz [23] in low cycle fatigue, damage is in the form
of matrix cracking, interfacial debonding and fiber breakage, while in high
cycle fatigue the damage is predominately matrix cracking and interfacial
debonding. Since fiber breakage dominates low cycle fatigue, the normalized
S-N curves overlap in the low cycle regime, but the difference in performance
become more apparent at high cycles. This is most apparent in the case of
carbon composites [Fig. 54]. The change of damage mechanism with
increasing number of cycles explains the change in the slope of the S-N curve
and the maxima of the damage zone as observed by whitening of the glass
composites [Fig. 60-65]. Since the whitening is due to matrix cracking, the
maximum in the damage zone can be explained as follows: at high stresses
and in static tests the damage is mainly in the form of fiber breakage so the
damage zone is small. As the stress reduces, matrix cracking increases causing
the damage zone to reach its maximum value. With a further reduction in
39
stress, interfacial debonding becomes more pronounced and the damage zone
decreases.
With carbon fiber composites, as observed by SEM micrographs Fig. 7275, the interfacial debonding is much more pronounced, because the carbon
fibers have an epoxy compatible sizing, and therefore do not adhere as well to
the MNS because of the large amount of polyester present. Since the fatigue
performance improves with adhesion [24], the fact that the carbon composites
perform much better than glass composites with a less steep slope [Fig. 57] is
quite remarkable.
5.4.3 Loss of Stiffness
In fiber reinforced composites the failure is progressive, resulting from
a gradual accumulation of various types of damage. The damage growth is
accompanied by a loss in stiffness: deflection increases when tested at a
constant stress level. These effects are shown in Fig 24 and Fig 59, (which also
note the percent of the ultimate tensile strength at which the tests were
performed). The stiffness loss is divided into three stages. In the first there is a
rapid decrease caused by matrix and interface damage, which reaches a
saturation level at the end of the stage. In the second, the modulus remains
nearly unchanged but the hysterises loss rapidly increases [8]. In this stage,
cracks interact and start to spread to transverse plies. In the final stage the
damage in the longitudinal and transverse directions coalesce and there is
some fiber breakage. This final stage is characterized by a very high damage
growth rate, such that a 40% loss in stiffness and complete fracture occurs at
approximately the same number of cycles [Fig. 59].
40
Chapter: 6
Creep Tests
6.1
Test Results
When a continuous fiber reinforced composite is loaded along the fiber
direction the properties are nearly time independent, but when the same
composite is loaded in compression or in flexure, and in the case of off-axis
laminates or discontinuous fiber composites, the properties usually are time
dependent. Since our composites have time dependent or visco-elastic
properties it is important to characterize their long term performance, so that
structural components could be designed with confidence. Therefore flexural
creep tests are performed on chopped strand mat with 7.5%, 12.5% & 17.5%
rubber and on glass fabric with 7.5% & 12.5% rubber composites. The tests are
performed at EMPA (Swiss Federal Laboratories for Materials Testing and
Research), in Zurich, Switzerland [26], using ASTM 2990-90 standard [25] with
three point flexural loading. Three specimens were tested at a constant stress
level of 16.7% of the ultimate flexural strength (UFS) and one sample each
was tested at 30%, 45% and 60% of the flexural strength. The total time for
which the specimens were tested was 120 days (2188 hrs ). All the tests were
carried out in a laboratory atmosphere at 23 oC and 50% relative humidity
(The effect of physical aging has not been investigated in this study). Only two
failures were observed, one in chopped strand mat with 7.5% rubber (1200
hours) and the other in chopped strand mat with 17.5% rubber (400 hours),
both at a stress level of 60% of ultimate.
41
The mid span deflection was measured as a function of time. The
deflection-time curves for the chopped strand mat are shown in Fig. 76-78 and
for the glass fabric are shown in Fig. 79-80. The creep modulus as a function of
time is shown in Fig. 81-85.
The creep behavior of the composites at different stress levels is
compared on the basis of their RCD (relative creep deflection) value, where
RCD is defined as:
RCD%
tot -
XlOO
50
Where
6 tot
is the deflection at the total time t=2188 hrs and 6o is the
initial deflection. The RCD values are given in Table. 8.
6.2
Creep Characterization
Since the service life of composite applications runs to years, it is
impractical (and expensive) to do testing over that period [27]. Therefore some
form of characterization is required to predict the long term properties of the
material. Since the design usually is based on stiffness, the technique should
be able to predict the long term modulus of the specimen. A visco-elastic
material can be characterized in two ways, linear and non-linear, depending
upon which dominates. In linear characterization, the constitutive equations
are functions of time only, while in the case of non-linear characterization
the constitutive equations are functions of time and stress. In linear behavior,
the isochronous curves (stress-strain curves at a particular time) are straight
lines. The isochronous curves for different times (t=1 min., 28 days and 120
42
days) are shown for chopped strand mat (Fig. 86-88) and for glass fabric (Fig.
89-90). These show that chopped strand mat composites behave in a nonlinear way at all three times, especially at high stresses; the non-linearity
increases with increasing rubber content and time. The woven fabric
composites behave in a linear way, at low stresses and times (It is common to
observe linear visco-elastic behavior in short term tests, but the behavior may
become non-linear for longer times at the same stress levels [18]. The larger
the deformation or stress to which a visco-elastic material is subjected, the
larger is the non-linearity [28] ). As is clear from the isochronous curves, and
from other considerations, the non-linear characterization
is more
appropriate for the creep of the MNS composites.
Several
different
characterization:
techniques
are used
for
non-linear
creep
the Multiple integral equation-a rational mechanics
approach; the Findlay procedure-a semi-empirical technique; the Schapery
equation-an irreversible thermodynamics approach; and the modified
superposition principle [28]. The Multiple integral representation is
impractical if strong non-linearities are present, while the modified
superposition principle and the Findlay procedure are completely empirical.
On the other hand the Schapery constitutive equation is based on principles
of thermodynamics and is most used for composites so this is the approach
taken in our study. A shortcoming is that it does not provide any physical
insight into the behavior observed. Its general form is:
£(t) = goAo(t) + gJAA(
-
0
where
= (t)= jd
0
ac
43
' ) dl(t)]
(1)
and
Nf = ()=d0 ao
where A is the immediate elastic response while AA is the delayed
response or creep compliance and
go,
gl,
g2
and ao are functions of stress.
These stress dependent properties have specific thermodynamic significance.
Changes in go, g and
g2
reflect third and higher order dependence of Gibb's
free energy on the applied stress and a
arises from similar high order effects
in both entropy production and free energy [29]. Creep characterization using
the Schapery equation usually requires creep and recovery tests. Since in our
study only creep tests have been performed, we cannot compute g, and g2
individually; instead we compute the product of the two. Thus six parameters
are required instead of seven to characterize the visco-elastic response. The
creep compliance, AA, of the composites is often approximated by using a
power law of the form:
AA(^) = A x14n
Where n is a material constant and go,
g,
g2
and a are approximated by the
following expression.:
g' =
Sinh//
or
O,
Sinh
,
Where o is a material constant and g' represent all of the functions of
stress. Since in our tests a constant load was applied in a single step, eq. (1) can
be integrated to yield:
44
D(t) =
_(t)
= goAo +
ggAt n
n
a
(2)
At low stresses the material behaves linearly and all the functions of
stress are equal to one, i.e., go=glx g2=aa=l thus Eq. (2) reduces to a linear
form:
(3)
D(t) = (t) = Ao + A x t
In our study the low stress level was 16.7% of the ultimate strength and
the constants A0 , A and n are calculated through a least squares fit between
the experimental data (at 16.7%) and equation (3) using the software
"MinSquare" and then the average values of the constants are calculated.
Assuming that A, Aand n are the same at the high stresses (30%, 45% and
60%), the other constants go, g,
g2
and a are calculated by fitting equation (2)
by the least square method to the experimental data. The stress dependence of
these quantities is then calculated for each composite. The values of the terms
in eq. (2) [go, g,,
g2
a,
A0, A and n] for different composites are shown in
Table 9.
The fitted curves and the experimental points for creep strain vs. time
are compared, for chopped strand mat, in Fig. 91-93, and for glass fabric, in Fig.
94-95. These show reasonable congruence, so the analytical expressions could
be used for the predictive purpose; the largest error or difference is about 9%.
While it would be tempting to extrapolate the predictions to a longer time
periods, there is no demonstrated justification for doing so. This reservation
45
becomes even more important when it is recognized that no consideration of
failure or fracture is implicit in the predictions.
46
6.3
Discussion of results
Two variables were used in the creep tests: rubber content and stress
level. The following observations can be made about the effect of these two
on the RCD values [Table. 8]. For both composites the RCD values are fairly
independent of the stress level [Fig. 96]. Up to 45% of ultimate strength the
RCD is independent of rubber content [Fig. 97], but at higher stresses the
chopped strand mat shows a higher RCD with increasing rubber content, as
also shown by the breaking of the specimens. This observation is consistent
with that by Hertzberg [8]: the creep rate increases in direct proportion to the
volume of rubber present in a neat resin. The RCD values for the glass fabric
composites are smaller than the chopped strand mat composites, because of
the presence of continuous fibers in the direction of the load. Overall both the
chopped strand mat and the glass fabric composites do not creep excessively,
which confirms that the MNS composites can be used safely in structural
applications for long service periods.
The creep in a composite arises from two processes: viscoelastic
deformation of the matrix, which may be reversible, and damage, which is
irreversible. The damage is due to matrix cracking, interfacial separation and
fiber fracture. At high stresses these can be expected to be prevalent. That is
why the RCD values sharply increase at 60% of ultimate strength [Fig. 96],
leading to failure in some cases (Since the chopped strand mat with 17.5%
rubber has the lowest strength it is expected to have the highest amount of
damage, hence its failure after 400 hours).
47
The isochronous curves [Fig. 86-90] show the laminates behave in a
non-linear viscoelasitc manner. Usually the long time response is more nonlinear than the short time behavior, thus the non-linear characterization is
used in this study.
48
Chapter: 7
Conclusions
7.1
Fatigue
(1) The static tests (at 5.08 mm/min) show that when the rubber content in
MNS is increased from 7.5% to 17.5%, the modulus decreases 11%-37% and
the strength decreases 12%-44%. The yield strain of the MNS composites is
larger, compared to the unmodified polyester (32%-60%), but it does not
change appreciably with increasing rubber content. The static tests at the stress
rate at which the fatigue tests are performed (3378 MPa/sec) show higher
modulus and strength; the average increase in strength (50%-110%) depends
on the type of reinforcement.
(2) The slope of the S-N curves change around 104 cycles for chopped strand
mat composites and around 105 cycles for woven fabric composites. The
damage zone, which is evident by the whitening of the specimens in the glass
composites, increases as the cycles increase, reaches a maximum, and then
decreases as the cycles increase further. The maximum occurs at about 104
cycles for the chopped strand mat and at about 105 cycles for the glass fabric.
This suggests the change in slope is due to a change in damage mechanism. It
is proposed that in the low cycle range the damage is dominated by fiber
breakage but with an increasing number of cycles, matrix cracking becomes
more dominant, causing the maxima in the damage zones. With further
cycles, interfacial damage becomes dominant, causing the decrease in damage
zone.
(3) Generally the damage progression is similar for all the composites studied
except for glass fabric with unmodified polyester. Usually it starts on the
49
surface with debonding along the transverse fibers. Matrix cracking then
develops, and as the cycling continues, debonding and matrix cracking
develop in the strands oriented in the stress direction, and the damage moves
to te center of the thickness, causing some fiber breakage along its way. The
Talrega diagram also shows the damage mechanisms are matrix cracking and
debonding. In the case of the glass cloth with the unmodified polyester, the
failure is mainly due to shear delamination in the center. The reason for this
may be that the fibers are Volan coated, which is more compatible with the
unmodified polyester, and thus they adhere better, so the matrix breaks.
(4) Although the S-N curves are not linear, when the data are plotted on a
Log-Log plot the curves become so. Then the data are characterized by the
value of m, where 1/m represents the slope of the linear (Log-Log) plot,
(S/S o = N Ym). The values of m for the different composites show that the
fatigue performance improves with increasing rubber. For chopped strand
mat the increase is in the range of 6.7%-9% when the rubber content is
increased from 7.5% to 17.5%; for glass fabric the increase is 13%-18% when
the rubber content is increased from 7.5% to 17.5% and for carbon fabric the
increase is in the range of -8% to 40% when the rubber content is increased
from 7.5% to 17.5%.
(5) Since all the normalized S-N curves for the glass composites overlap, it
appears the fatigue behavior is fiber dominated. In the carbon composites the
behavior is matrix dominated, which is obvious from the trends in the
normalized S-N curves for the different matrices. This is consistent with the
fact that carbon fibers are more fatigue resistant than glass.
(6) Generally with off-axis laminates and when the loading has a compressive
component, both carbon and glass composites perform similarly. Because of
this, and because of fiber misalignment, weave distortion and yarn
50
crossovers, the performance of the carbon and glass composites was expected
to be similar. However, with the MNS matrix, due to its increased strain to
failure and a modest decrease in modulus, the performance of the carbon
composite was better than glass.
(7) With the chopped strand mat the change in slope after 104 cycles is large
and it may represent an approaching endurance limit.
(8) The initial strain-N curves show the fatigue performance improves with
increasing rubber content for any reinforcement.
(9) The stiffness loss, represented by deflection vs. number of cycles, is divided
into three stages. In the first, there is rapid decrease in stiffness; in the second
it remains nearly unchanged; in the final stage, again it drops rapidly, and the
40% loss in modulus (our fracture criterion) and complete failure occur at
approximately the same number of cycles.
7.2
Creep
(1) Flexural creep tests were performed only on chopped strand mat and glass
fabric composites, with two variables: rubber content and stress level. The
comparison of relative creep deflection shows the values are independent of
the stress level in most cases. Increasing the rubber content in the woven
fabric composites shows the RCD values are independent of rubber content,
while the same is true with the chopped strand mat composite at low stress
levels, but at higher stress the chopped strand mat with 17.5% rubber has a
higher RCD value.
(2) The isochronous curves show that the laminates exhibit non-linear
viscoelastic behavior.
51
(3) The Schapery constitutive equation represents the time-dependent creep
behavior of these composites quite well; relatively small divergencies are
apparent after about 1000 hours of creep.
52
Chapter: 8
Suggestions For Future Work
(1) Fatigue testing of the unreinforced MNS matrix with different rubber
contents.
(2) Testing the composite specimens to higher (107) cycles to confirm the
presence or absence of an endurance limit.
(3) Testing MNS composites in fatigue with a variety of different rubber
compositions.
(6) Testing the composites in creep at more stress levels to find more accurate
functions for the stress dependent parameters in Schapery's equation.
(7) Creep recovery tests should be performed to find all the parameters in
Schapery's equation and to provide further information about the damage
accumulation in creep.
53
Appendix: I
Pure
MNS 0%
MNS 7.5%
MNS 12.5%
MNS 17.5%
Polyester
Composition
Resin 1
Polyester
60.44 %
51.2%
47.33 %
44.77 %
42.21 %
Epoxy
0 %
15.3%
14 %
13.43 %
12.66 %
Styrene
39.6 %
33.5%
31 %
29.3 %
27.63 %
Rubber
0%
0%
7.5 %
12.5 %
17.5 %
Chopped
strand mat
51.90%
51.99%
46.89 %
47.71 %
43.58 %
Glass Fabric
60.09%
60.76%
62.04 %
63.57%
55.18%
(ATBN)
Glass Content 2
1. Weight percent based on the total weight of resin.
2. Weight percent based on total weight of composite.
Table: 1 :- Composition of resin and Glass content in the composite
54
Specimen
Flexural Modulus
Yield Stress
E MPa
a
MPa
Yield Strain
s m/m
Chopped Strand mat
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
9336(±615)
9491(±325)
11031 (386)
9688 (+275)
7358 (589)
256.31(±35.79)
268.21(+19.79)
293.52 (+9.90)
256.46 (+13.22)
200.23 (±26.76)
.0303(±.0036)
.0308(+.0022)
.0363 (+.0005)
.0357 (+.0008)
Glass Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
16148 (355)
14979(+1322)
18318 (+515)
16354 (184)
11576(+269)
278.55 (+25.80)
366.22(±23.42)
422.07 (±14.26)
371.40 (+17.40)
234.84(+1.96)
.0195 (+.0031)
.0284(+.004)
.0277 (+.0012)
.0286 (±.0018)
.0227(±.002)
Carbon Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
39452 (+2546)
43803(+222)
35515 (+1655)
49014 (+547)
30999 (4071)
365.84 (+29.58)
650.87(+33.18)
529.45 (+60.04)
582.74 (±13.02)
422.51 (87.49)
.0096 (+.001)
.0150(±.0008)
.0154 (+.0024)
.0127 (+.0006)
.0145 (+.0004)
.0345 (.0016)
Table: 2 :- Mechanical properties of all the composites tested in three point
bending at a rate of 5.08 mm/min. Numbers in parenthesis are standard
deviations.
55
Specimen
Initial Modulus
E MPa
Flexural Strength Strain to failure
ca
MPa
E m/m
Chopped Strand mat
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
13362
13881
15679
13645
11011
488.06
503.88
572.56
572.19
518.06
.0365
.0363
.0365
.0419
.0471
Glass Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
23469
21975
27248
22960
14927
564.74
672.88
730.53
649.91
408.56
.0241
.0306
.0268
.0283
.0274
Carbon Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
44616
47847
49518
54241
40217
643.67
820.82
840.33
867.59
532.76
.0144
.0172
.0170
.0160
.0133
Table : 3 :- Mechanical properties of all the composites tested in three point
bending at a loading rate of 3378 MPa/Sec (Loading rate on which fatigue
specimen are tested).
56
b
Log(N)
Log(S/So)
S/S O - 1
logN
for whole
life, %
upto 104
cycles, %
after 104
cycles, %
for whole life, %
14.74
15.36
13.27
13.58
13.67
17.22
18.83
19.46
18.89
21.39
6.25
5.93
4.97
5.14
4.31
7.86
7.44
6.96
7.43
7.59
for whole
life, %
upto 105
cycles, %
after 105
cycles, %
Glass Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
13.49
13.21
13.94
12.99
13.59
15.67
15.23
15.01
14.65
14.93
7.61
6.29
7.13
6.69
3.20
Carbon Fabric
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
12.23
12.22
11.35
12.38
9.18
15.00
12.908
12.24
13.45
9.97
Specimen
Chopped Strand mat
Pure Polyester
MNS 0% Rubber
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
9.03
8.77
8.09
9.14
9.55
11.37
11.07
12.69
11.66
17.79
Table : 7 :- Value of fatigue characterizing parameters, b and m for different
composites.
57
Specimen
RCD %
RCD %
16.7% of f
30% o
Chopped Strand mat
MNS 7.5% Rubber
MNS 12.5% Rubber
MNS 17.5% Rubber
27.80
28.14
28.80
Glass Fabric
MNS 7.5% Rubber
MNS 12.5% Rubber
16.10
17.97
RCD %
f
RCD %
45% of cf
60% of
26.32
26.32
31.61
27.52
23.55
28.04
tf=1200h
28.64
tf=400h
15.52
17.03
13.72
16.17
13.96
22.68
f
Table: 8 :- Relative creep deflection after 120 days at different stress levels. tf
shows the time of failure in hours.
58
Ao
A
n
go
gl*g2
a
Sinh o/227.30
sinh o/198.84
c/109.57
a/227.30
a/198.84
sinha/109.57
o/218.40
0/49.40
Composite
Chopped
strand mat
7.5% Rubber
12.5% Rubber
7.5146E-5
7.2108E-5
2.0580E-5
2.7202E-5
0.087059
0.074309
sinha/133.34
a/133.34
17.5% Rubber
sinh a/218.40 sinha/49.40
1.0793E-4
2.9203E-5
0.088642
sinho/140.78 sinha/120.20
a/67.61
o/140.78
a/120.20
sinh a/67.61
4.5350E-5
8.1880E-6
0.072685
sinh a/408.18
Glass fabric
7.5% Rubber
o/408.18
12.5% Rubber
4.8976E-5
8.0848E-6
0.086216
sinh c/231
o/231
a/156.91
sinho/201.32 sinho/156.91
/201.32
sinh O/193.74
a/87.61
a/193.74
sinh a/87.61
Table: 9 :- Values of different terms in Schapery's equation.
59
Appendix: II
100
50
.d
E
0
z
0
0
2
4
8
6
Log life
Fig. 1: Normalized S-N curves for unidirectional composites with various
types of fibers. tested in longitudinal tensile fatigue 111.
IWU
·
I
I
90 ........
..........
.
o..e
o
III
.
x+
80
"t......
Eat
70
.. 11 ... - .........-
60
4.)
. . .;. . . ~oa ;............
50
Un
o-
40
.j
30
20
10
A
10C
o - NEMA Grade G-10 E-glass
+ - E-glass re-Pre-Preg, +/-10 Degx
x - E-glass Wet Wind, +/-10 Deg
* - E-glass Wet Wind, +/-25 Deg
*- S2-glass Wet Wind, +/-10 Deg
,- S2-glass Wet Wind, +/-25 Deg
. .
+
..
.......
O
ox+
AJ
x
+
;
I.........
.....
...-
I ..............-
El.* :
AP ................-
.- Graphite Wet Wind, +/-25 Deg .
[i- Graphite Wet Wind, +/-20 Deg
I
101
I
I
102
10 4
103
105
I
106
Number of Cycles
Fig. 2: Normalized S-N curves for cross ply composites with various
types of fibers. tested in flexure fatigue [31.
60
107
/fibre breakage,interfacial debonding
tax
Et
matrix cracking,
interfacial shear
failure
fatigue limit of matrix
Cm
-- ,
,,,,,,,,,\
log N
Fig. 3: Strain -N curve showing regions of different damsge mechanisms
as proposed by Talrega [5].
A
attached to
load cell
Instron driver
Fig. 5: Flexural fatigue fixture.
61
600
400
Cn
2
0.
L
U,
200
0.02
Strain
(mm/mam)
Fig : 6: A typical stress-strain curve for a composite
62
I
50000
.
unmodified poy eCart )on fabri I
I
400001
unmodified polyester
o
230000
Glass fabric
unmodified polyester
b20000
0.
4I
10000 unmodified polyester
'
Chopped strand mat
0
0
10
5
15
% Rubber content in matrix
Fig: 7: Flexural modulus Vs. % rubber content in matrix for different
reinforcements (form the static tests performed at a ram speed of 5.08
mm/min).
63
20
700 Carbon fabr ic
\-
600 aa
500
-
unmodified polyester
.,
Ub
z;
YI
400 -
P4
"0
~abric
~~~~~Glass
^
unmodified
unmodified polyester
Th
--4
300-
I'd
a,
I
200 -
unmodified polyester
A
Chopped strand mat
100
_
0
0
10
5
15
15
% Rubber content in matrix
Fig: 8: Yield strength Vs. % rubber content in matrix for different
reinforcements (from the static tests performed at a ram speed of
5.08 mm/ min).
64
20
0.04
.
*-
Chopped strand mat
:,_..unmodified polyester
0.03
e!0.02 -
4
-
A
Glass fabric
unmodified polyester
Carbon fabric
Id
'
'
-,
~~~~~~
tD
0.01
'4*-
unmodified polyester
_
0.00
0
10
5
15
15
% rubber content in matrix
Fig: 9: Yield strain Vs. % rubber content in matrix for different
reinforcements (from the static tests performed at a ram speed
of 5.08 mm/min).
65
20
2000
1800
E]
1600
a
1400
Z
0
1200
1000
O-j
B
E
800
El
El
600
El
El
400
El
200
.1
0
j
0
I
I
5
10
I
15
15
20
25
30
Deflection, mm
Fig: 10: A typical load-deflection curve for a composite (carbon fabric/MNS
with 7.5% rubber) tested in flexure at a rate of 3378 MPa/sec.
66
I
rare\
500 i
I
tt
400
rn
I~~~~~~
bD
o4
VS
.E.
-a1
ebo
300-
0~~~~
200-
0~~~
0
El~~~~~~El~ ~ ~
0~~~
0
100
_____I__
A
UI
10
0
101
_ _______
I0,2
_ ________
..
102
10
_________
'''I
3
10
_ ________
.10
4
10 5
_ ________
..106
106
________
...
10 7
No. of cycles to failure
Fig: 11: S-N curve for choDped strand mat/unmodified polvester. R=-1
67
600-
500
v
400-
(IW
(
300Il
200-
El~~~~
100 -
_
0
________
_
______·
_
_ _____·
10.
100
10'
103
102
_
a
El~~~~
____1_1·
...0
104
1
...
10
5
·
I····
· ____1
10...
10
6
10 7
No. of cycles to failure
Fig: 12: S-N curve for chopped strand mat/MNS with 0% rubber, R=-1
68
I
-·h
600 I
II
500
cn
2Q,
400-
to
r.
--
300
mm
2
-
200 0
02
m
2
100 -
2
E] E
_ ________
0..1I0
0
Io
1C)
]
2
________ _ _ _______ _________ _ _______
_ ________ ______I
.. I
.102
..106
103
102
106
105
10 3
10 4
No. of cycles to failure
Fig: 13: S-N curve for chopped strand mat/MNS with 7.5% rubber, R=-1
69
700 I
600
tt
en
C4
500
400 -
.
V)
bo
C:
el
.a
300
;X2
03
200 -
X
El
0
0
0
100-
-
1()0
·
I
lw
I
10 1
·
· ·
l.l.
I
I
' ' ' ' 'I '1l
W WWI
3
102
''
10 4
.. .
10 5
rr
rapt
T
10 6
T
T
TT
TT
10 7
No. of cycles to failure
Fig: 14: S-N curve for chopped strand mat/MNS with 12.5% rubber, R=-1
70
700 '
600 I
J
500
CD
._
a
400 -
bo
a:
z;
.,..
300
Crr
In ~
200 -
~
~
n
E
a
[][][
100 -
04
10
El
El
_
..10
________
10I
_
_ _______
_
________
'..I
10 3
102
_
_______
_______
aw
10 4
roll1
10 5
_
·
a________
XrT
10 6
1
______
.* 1
*ww
107
No. of cycles to failure
Fig: 15: S-N curve for chopped strand mat/MNS with 17.5% rubber, R=-1
71
0.04
1
I
J
t\Composite static failure strain (Fiber breakage)
0.03
-
Matrix cracking nd
o
* 0.02
interfacial debonding
a0)
·
oI
/
Q~
< 0.01
Ln
o
Limiting strain for transverse fiber deonding 0.12% A
0.00
3
------------------------------------------------~~,\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\~~~~~~~
I.
. I.
LO° 10 1
l
102
I
103
. . I... I., .
10 4
I II
r1~~l1...- I .~r· 71~~
10S
1 6
7
No. of cycles to failure
Fig: 16: Strain-N curve for chopped strand mat/unmodified polyester, R=-I
72
0.04
II
I1111)11(11111111111111111111111111111111
Composite static failure strain (Fiber breakage)
0.03 -
-.
.,-I
0.02 4)
4-.
0
Er
0
ad
Matrix cracking and
interfacial debonding
0
r
[]
0.01 -
El
] 0
0
El
Limiting strain for transverse fiber debonding 0. 12%
------------------------------------------------10 4
10
10
I1
.
I . ..
I
... I
0.00JII t,\\\\\\\\\\\\\\\\\\\\\\\\\,,\\\\\\\\\
10
2
10 3
10 o
10 1
'
'
I
.
..
. 111..1
.I
.
11
..
I
5
1 .rI...
rr
6
1
07
No. of cycles to failure
Fig: 17: Strain-N curve for chopped strand mat/MNS with Oo rubber, R=-I
73
0.04 I L
.
(111(1(111111111111111111111111111111111
Composite static failure strain (Fiber breakage)
0.03 -
E
-4
-4
4
-4
Matrix cracking and
interfacial debonding
0.02 -
0
Cd
X,
Rr
4'
S:L
°c
0
0.01 -
0
Q0
E
Limiting strain for transverse fiber debonding. 0. 12%
0.00I 10
---I
o
10 I
-I
10 3
10 2
10 4
10 5
6
10s
10 7
No. of cycles to failure
Fig :18 : Strain-N curve for chopped strand mat/MNS with 7.5% rubber, R=-I
74
0.05
I
I
0.04
------ --------
-------------lllTllllIlllll
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
Composite static failure strain (Fiber breakage)
1:
0.03
*_l
Matrix cracking and
interfacial debonding
x¢d
-4
-P.
-4
9c
0
0.02'
-0-
a
AEE
0.01
E]
El
]
Limiting strain for transverse fiber debonding. 0.12%
0.00
10°
10 i
10 2
10
3
10 4
10 5
10 6
10 7
No. of cyvles to failure
Fig: 19: Strain-N curve for chopped strand mat/MNS with 12.570 rubber, R=-I
75
0.05
-
II
I
IiiIIiiIIIIiIIIIiII1I1I1IIt
11111111111IIIIIIIIIIIIIIIIIIIIIIIIIIIIII
II
IIIIIIIIIIIIII
Composite static fail ure strain (Fiber breakage)
0.04
E
0.03 -
E;
Matrix cracking and
interfacial debonding
I-
.4
)
cd
0
[]
0.02 -
Q
-4
z:
E
El[]
1 r~
0.01 -
E
El
Limiting strain for transverse fiber debonding. 0.12%
0.00
10
10 I
102
10 3
10
10 5
10 6
10 7
No. of cycles to failure
Fig: 20: Strain-N curve for chopped strand mat/MNS with 17,5T%rubber, R=-I
76
700
600
500
vr4
400
a
bEdI
>i
300
200
100
0
0
2
4
6
Log (No. of cycles to failure)
Fig: 21: Comparison of total stress S-N curves for Chopped strand mat
with unmodified polyester, with 0%. 7.5%, 12.5% and 17.5% rubber
77
8
1.2
1.0
0.8
a
Lq
I
0.6
Z
0.4
0.2
0.0
0
2
4
6
8
Log (No. of cycles to failure)
Fig: 22: Comparison of normalized stress S-N curves for Chopped strand
mat with unmodified polyester. with 0%, 7.5%, 12.5% and 17.5% rubber
78
0.05
0.04
-,
-,4
0.03
a)
-
,,I,,
0.02
-~,,
.<
0.01
0.00
0
2
4
6
Log (No. of cycles to failure)
Fig: 23: Comparison of initial strain-N curves for all Chopped strand
mat with unmodified polyester, with 0%, 7.5%, 12.5% and 17.5% rubber.
79
8
11
A
10
987;
6-
C
J
5-
+
O
+
MEN
IIIII
+StageStage
MNNNM
I
Stage II
4-
30% of UTS
* 32.5% of UTS
3Stage I
2-
O-
0
v.
.
o
38.5% of UTS
38.5% of UTS
*
44.5% of UTS
+
44.5% of UTS
-
1000
2000
3000
4000
No. of cycles
Fig: 24: Deflection Vs. No. of cycles curve for chopped strand mat/MNS
with 7.5% rubber
80
30
.
28 .
26-
0
moons
ONE
B0
0 0 0 0 0 0 0 0 0
0 0
E]E3
0
24-
20
32.5% of UTS;Frequency=2.888 Hz
* 38.5% of UTS;Frequency=2.437 Hz
0
22
a
20
0
I
I
I
1000
2000
3000
I
X
4000
No. of cycles
Fig: 25: Temperature Vs. No. of cycles curve for chopped strand mat/MNS
with 7.5% rubber
81
800
700 -
600
I
I
500
V)
cn
400 -
.
300 0
0
200 -
0
El []~0
100 -
01C)0
...
10
10I
10 3
102
..
104
10...
105
106
No. of cycles to failure
Fig: 26: S-N curve for glass fabric (181)/unmodified polyester, R=-1
82
107
700 I
600 -
(t
500-
en
v)
u)
400 -
to
.
300-
0
C.,
r
200-
a
0
0
100
_ _ _______
0
10 °
101
_ _ _______
_ ________
102
10
_ _ ______
_ ____I__
...0
...
,
3
10
4
105
_ ________.,I
10
_ _ _____
6
No. of cycles to failure
Fig : 27: S-N curves for glass fabric (181)/MNS with 0% rubber, R=-1
83
10 7
--800 -
700
600
500
()
v)
400-
U
E
0
X
b
a:
0 X
300 -
0
0
0
200 -
B
00
[r
100 -
0
_________
-
0
1C
)
..
10I
_
_ _______
_
_ _______
..
10 3
102
_
________
...I
_
_ _______
..
10
104
5
105
_________
_
_______
106
No. of cycles to failure
Fig: 28: S-N curve for glass fabric (181)/MNS with 7.5% rubber, R=-1
84
10 7
800 I
700
600
500-
.
400El
I.
C)
0~~~
300-
0
a
0
200
-
00~~
m~~
El~~~~~~ El0
El~~~~~
100-
0
_ _____II·...
100
10
-
· __1___·
.102
· _1_1····
_ _ 1·___
... I
10 3
102
10
· · ____·1-...
4
10 5
____11__
...
106
__ 1_1___
10 7
No. of cycles to failure
Fig: 29: S-N curve for glass fabric (181)/MNS with 12.5% rubber, R=-1
85
500
.
I
400 X
t%
)
300-
C,
4-,
0
0
Q)
0
200 -
0
0
a
0 []
[]~~~
100
_____1__
0
100
10
_ ________
_ _______
10 3
102
________
_ ________
10 4
10
_________
..106
________
I06
10 5
No. of cycles to failure
Fig: 30: S-N curve for glass fabric (181)/MNS with 17.5% rubber, R=-1
86
10 7
0.03 I
I
I
1-10.02
i
1lI[WI1Ifl ItfWLII[
ltftd~IfftlIF
Composite static failure strain (Fiber breakage)
-
4
Matrix cracking and
delamination
o4
=4
I:;
El
·,
·.
0
0.01
£
<c
0E
0
X-
0
0
Limiting strain for transverse fiber debonding. 0.12%
0.00 -1
10°
I
10 I
\~\
\\\\ \\\`
\
102
\
~
N
\,~\\~,\\
10 3
\
\\\\
10'
10 l
10'
No, of cycles to failure
Fig: 31: Strain-N curve for glass fabric/unmodified polester, R=-1
87
10 7
0.04
E
0.03
lllllnllllmT1l[lllnTmIlTllllrmmT1lllllll
`C
1-
omposite static failure strain (Fiber breakage)
9
4
Cd
.-4
0.02 ·1)
Matrix cracking and
interfaical debonding
El
$14
101
G
0.01 -
Ea
E3
El
El
l
Limiting strain for transverse fiber debonding A
~NN~S~K\N>\NK~\ N\\
NT,\\\Ž
N\\\\\\K
K
I
0.00 --------------------------------------------------10
10 6
10 2
10 3
10 4
10 5
10 0
No. of cycles to failure
Fig: 32: Strain-N curve for glass fabric/MNS with Oo rubber, R=-I
88
10 7
0.03
........................ .
Composite static failure strain (Fiber breakage)
0.02 q
Matrix cracking and
interfacial debonding
-4
s
-a
El
E
in
rq
a)
i-a
0
3
~z " m
ra
0.01 -
[:P
P-s
E
O
n
'2rE
a 1
Limiting strain for transverse fiber debonding. 0. 12%
0.00
. . . ..- I . I
O
Lo
10
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\N
.
7
.
. . I. .11
10
3
.
I
. I . .11
10
.
I
.
.. 11
o5
I
.1
I .
1
.1
I I I I I
10
No. of cycles to failure
Fig: 33: Strain-N curve for glass fabric/MNS with 7.5% rubber, R=-1
89
1(0
U.UJ -
I1111111111111111 11111111111 111 1111 111111
iiii11111111IIIIIIIIIIIIIIIII111II11111
Composite static failure strain (Fiber breakage)
0.02
Matrix cracking and
interfacial debonding
4,---
4l
0.01
E339r
Limiting strain for transverse fiber debonding. 0.12%
0.00 -
100o
o
102
103
O
10 5
10 6
No. of cycles to failure
Fig: 34: Strain-N curve for glass fabric/MNS with 12.5% rubber, R=-I
90
107
0.03
IComposite static failure strain (Fiber breakage)
i
0.02 -
-
Matrix cracking and
interfacial debonding
ad
oa
-,
El
[
.
R 0.01
o 13
¢-
[]
El
Limiting strain for transverse fiber debonding 0.12% *
~~~~~....
0.00 1
....
10 O
101
10 2
10 3
I
,
..
.
10+
N
.
10 S
..
...
106
10
No, of cycles to failure
Fig: 35: Strain-N curve for glass fabric (181)/MNS with 17,5% rubber, R=-1
91
800
,
600
60C
an
o
--
400
0
-.
200
0
0
4
2
6
Log (No. of cycles to failure)
Fig: 36: Comparison of total stress S-N curves for glass fabric with unmodified polyester, MNS with 0%, 7.5%, 12.5% and 17.5% rubber, R=-1
92
8
1.2
1.0
0.8
da
ro
0.6
Z
0.4
0.2
0.0
0
2
4
6
Log (No. of cycles to failure)
Fig: 37: Comparison of normalized stress S-N curves for glass faric with
un-modified polyester, MNS with 0%, 7.5%. 12.5% and 17.5% rubber. R=-1
93
8
0.04
- 0.03
-,- 0.02
0)
4
0
0.01
(.00
0
2
4
6
Log (No. of cycles to failure)
Fig: 38: Comparison of initial strain-N curves for glass fabric with unmodified polyester, MNS with 0%, 7.5%. 12.5% and 17.5% rubber, R=-1
94
8
800
· Chopped strand mat with unmodified polyester
o Chopped strand mat with 0% rubber
A Chopped strand mat with 7.5% rubber
· Chopped strand mat with 12.5% rubber
N Chopped strand mat with 17.5% rubber
3 Glass fabric with unmodified polyester
A Glass fabric with 0% rubber
[ Glass fabric with 7.5% rubber
Glass fabric with 12.5% rubber
+ Glass fabric with 17.5% rubber
l
600
B [
5
4.
tn
r.O400'
to
z
Wl
1
A
0
o++
N
·
A
t
MA
El
+*
200
[]
o
A
m%o
+
H aAO+"+
A8
KFcr94
,
' "'
n
1I
IL
v
10
0
I
I
E I
T
I-
[
101
..I
l1·Z}
0 2r0
10
_
rI
2
10
3
.a·
j
10
r
r1041
4
10l
-
.
h
_l·R
--16
10 6
I ______·
10 7
No. of cycles to failure
Fig: 39: Comparison of total stress S-N curves for all Chopped strand mat
composites and Glass fabric composites
95
Chopped strand mat with unmodified polyester
o Chopped strand mat with 0% rubber
A Chopped strand mat with 7.5% rubber
1.2
Chopped strand mat with 12.5% rubber
Chopped strand mat with 17.5% rubber
1.0
fn
(/:
Glass fabric with unmodified polyester
Glass fabric with 0% rubber
Glass fabric with 7.5% rubber
·
m Glass fabric with 12.5% rubber
+ Glass fabric with 17.5% rubber
0.8 -
W
Q)
N
Cq
+
0
00+
+
0.6 -
.--
% A0
0
0
0
0.4 -
M"AO
A
0.2
n..u
I
13
A~~~~~
-
0.0 1I 1
100
,
As
·
.
I
10
·
B · """
'..
2
10.I
'
10
·
102
10 3
10 4
10''
..106.
6
10
10 7
No. of cycles to failure
Fig :40: Comparison of normalized stress S-N curves for all Chopped strand mat
composites and Glass fabric composites
96
800 I
.
0
o
I
I
600
us
N
II
II
a
+
*
II
.
V)
cn
(V
.
.
*
b
Glass fabric with unmodified polyester
Glass fabric with 0% rubber
Glass fabric with 7.5% rubber
Glass fabric with 12.5% rubber
Glass fabric with 17.5% rubber
Glass fabric with unmodified epoxy [Ref. 3]
I
N
H
400 -
&
M
N
._
7:5
*E
N
N
0
N
O
+
+
4t
#
N
EP
oM
200 -
+
+
+
0
0
+
0
-
I
100
----11....-- . -1 -----.
10
1
. - ------.
10 3
102
. ---.----
10 4
- ----105
106
107
No. of cycles to failure
Fig 41 : Comparison of total stress S-N curve for glass fabric (7688) with epoxy
-[Ref. 31 to the S-N curves for vlass fabric (181) with unmodified olvester, MNS
with 0%, 7.5%, 12.5% and 17.5% rubber, R=-1
-
97
1.2
(I
C Glass fabric with unmodified polyester
o Glass fabric with 0% rubber
N Glass fabric with 7.5% rubber
A Glass fabric with 12.5% rubber
+ Glass fabric with 17.5% rubber
* Glass fabric with unmodified epoxy [Ref.
1.0
U
.
U
On
.,
*
0.8
U
,]
.4-
10
N
0.6 -
# +
A
N
.-
a
N
N
+
*
+
in[[[
-+0,%
0.4
N ##ze5I
0.2
"
A.,o
1
0.0
-1
100
_.
10 1
10'
1022
10
103
104
10 5
10
106
107
No. of cycles to failure
Fig 42 : Comparison of normalized stress S-N curve for glass fabric (7688)
with epoxy [Ref. 3] to the S-N curves for glass fabric (181) with unmodified
polyester. MNS with 0%, 7.5%. 12.5% and 17.5% rubber. R=-1.
98
700
?
I
I
600
500
-
400
-
300
E~~~~la
B
200
100
.I
0-
10I0
·
10 1
10 3
102
104
· TV
10 5
No. of cycles to failure
Fig: 43: S-N curve for carbon fabric/unmodified polyester, R=-1
99
i
i[
i
106
.
1000
l
800
U
U)
V)
U)
600
-e
r.
.,.
1:$
zp
0
0 0
400
0
[][]~
El D
[]~~~
200-
1 · 1···1·
0
10
0
10 1
·
_r711·
--
10 3
102
· · · 1_·1_·..
104
_ _ ______
_ _ ______
.. I
10
.'
.
5
No. of cycles to failure
Fig: 44: S-N curve for carbon fabric/MNS with 0% rubber. R=-1
100
106
1000
Ill
I
800
I
1
ca
M
rn
(n
600
bo
I
:
'7z
Q)
[]
400 -
I1
El
%
El~~~
200
0 I
10 0
II
__ _____ _ _ _____
101
'
.. ._~~~~~~~~
...
__
102
10 3
. .
104
10
105
10
· w.106
10
W
.
.
6
No. of cycles to failure
Fig: 45: S-N curves for carbon fabric/MNS with 7.5% rubber, R=-1
101
....
10 7
1000
I
I
800
(t
600
bb
r.0
[
400
[]
[] []
[]
200
_........
________
0
100
101
_ ________
..102
_ _______
103
102
_ ________
_ ________
10 4
105
_ ________
..106
_ _______
106
No. of cycles to failure
Fig: 46: S-N curve for carbon fabric /MNS with 12.5% rubber, R=-1
102
10 7
1000900
-
800-
CZ
7nn00
,
m
600
Un
I1
to 500
.
r:- 400
B
El
300 -
X
0
rn%
200 100 -
0100
_ _______
...
10I
_ _______
...102
_ _ _______
10 3
102
10 3
_ ________
..
_ _______
104
10I
10
________
_ _______
106
No. of cycles to failure
Fig: 47: S-N curve for carbon fabric/MNS with 17.5% rubber, R=-1
103
107
0.016
2
0.014
I.............
..... ..
.............. - --
*
I
.........
Composi te static failure strain (Fiber breakage)
0.012-
Matrix cracking and
interfacial debonding
0.010 -
.
.4
i. 0.008 E
/l
4-,
E
0.006-
x
4
0.004 -
Limiting strain for transverse fiber debonding. 0.12%
0.002 1
-------------------------------------------------
0.000II 1i
10
_II
o
'lo
''
10 1
1I
1
2
·
.
-
10 3
10
,
.
10 4
I ·
·
,1I·
10 5
10 6
No, of cycles to failure
Fig :48: Strain-N curve for carbon fabric/unmodified polyester, R=-I
104
0.020
I
i
I
Z 0.015
-
t l lIl I{llhl l lffi
. ..............
IIIIItII
I-
.....
Composite static failure strain (Fiber breakage)
.4
Matrix cracking and
interfacial debonding
-d
El
z 0.010
/
0
P4
03
0
0oEl3
0.005
E
-
Limiting strain for transverse fiber debonding
CE
A
.
...
.....................................----
I
N1\
I
0.000I
10 0
10
1
10
10 3
10 +
10 5
10 6
No, of cycles to failure
Fig: 49: Strain-N curve for carbon fabric/MNS with 0% rubber, R=-1
105
0.020
IIIIIIIIIII
II IIIII IIIIIIIIIII
IIIII~I~IIi
IIrI~IIIII
I~IrIIII~II
I ~IIIIIIIIII
E 0.015
Composite static failure strain (Fiber breakage)
Matrix cracking and
interfacial debonding
*1-
II
/
-4
m[]
I
E
s
]
J.005
¢6
Limiting strain for transverse fiber debonding. 0.12%
0.000
*1.
............................................
!
10 O
10 I
10
10 2
10 4
10
5
10 6
No. of cycles to failue
Fig :50: Strain-N curve for carbon fabric/MNS with 7.5% rubber, R=-I
106
0.020
0.018
i
0.016 I
IllllI)IIIIIIllllll Tm lllii
iiIIIIIIIIIIIIIIII
I1lllI II~IIIII
Composite static failure strain (Fiber breakage)
-0.014
Matrix cracking and
interfacial debonding
0o.o12
" 0.010
-4
+,0.008
[]
El
x
20.006
0.004
0.002
Limiting strain for transverse fiber debonding. 0.12% A
0.000II a~\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\x
.. I
10 0
10 i
10 2
10 3
10
4-
10 5
10 6
No. of cycles to failure
Fig :51: Strain-N curve for carbon fabric/MNS with 12.5%rubber, R=-I
v
107
0.016
-
Composite static failure strain (Fiber breakage)\
0.014
o0.012
Matrix cracking and
interfacial debonding
.. 0.010
-4
1a
- 0.008
..di
o0.006
J.004
0
0.002
Limiting strain for transverse fiber debonding. 0.12% m
0.000II
10 C
I
10 j
10
:\\\\\\\\\\\\\\\\\\\\
I7T
T1 TI1
10
I
T II
10
I
4
T
I
1
10 I
1
I II
10 E
No. of cycles to failure
Fig: 52: Strain-N curve for carbon fabric/MNS with 17,50% rubber, R=-I
108
1000
800
a
600
.4
as
400
200
0
0
2
4
6
Log (No. of cycles to failure)
Fig: 53: Comparison of total stress S-N curves for carbon fabric with unmodified polyester MNS with 0%, 7.5%, 12.5% and 17.5% rubber, R=-1
109
8
1.2
1.0
% 0.8
0)
Z 0.6
0.4
0.2
0
2
4
6
8
Log (No. of cycles to failure)
Fig: 54: Comparison of normalized stress S-N curves for carbon fabric with
unmodified polyester, MNS with 0%, 7.5%, 12.5% and 17.5% rubber, R=-1
110
0.02
ri
- 0.01
4
0
R
0.00
0
2
4
6
Log (No. of cycles to failure)
Fig: 55: Comparison of initial strain-N curves for carbon fabric with unmodified polyester, MNS with 0%, 7.5%, 12.5% and 17.5% rubber, R=-1
111
8
· Glass fabric with unmodified polyester
o Glass fabric with 0% rubber
Glass fabric with 7.5% rubber
A
10UUa'!
'U')').
* Glass fabric with 12.5% rubber
N Glass fabric with 17.5% rubber
° Carbon fabric with unmodified polyester
A Carbon fabric with 0% rubber
a Carbon fabric with 7.5% rubber
* Carbon fabric with 12.5% rubber
__
+ Carbon fabric with 17.5% rubber
800
1
; ;
(
..
600
1r
a)
A
a
A
-1
400'
+
MAEE
*A
0
A*
+
a
+
L
n l
96
A
A
A++
s*
200
0
O
N .l·k
N
'.
I
A.
xA
4
11~·7
1177711 17177rr · 711111 1 ~71M 1 lll
O'
100
101
10 3
102
10 4
10
5
106
No. of cycles to failure
Fig: 56: Comparison of total stress S-N curves for all Glass fabric
and Carbon fabric composites.
112
1
Glass
o Glass
A Glass
* Glass
" Glass
fabric with unmodified polyester
fabric with 0% rubber
fabric with 7.5% rubber
fabric with 12.5% rubber
fabric with 17.5% rubber
0
1.2
j
..
1.0
to
(o 0.8'
°
Crhnn
fhric with
ilnmnAificle
nnulvsfPr
UYVILUY·LI·L·LI·
UIUIIVUI·U
YVL·ULL
A
[
*
+
Carbon fabric with 0% rubber
Carbon fabric with 7.5% rubber
Carbon fabric with 12.5% rubber
Carbon fabric with 17.5% rubber
I
II
+
+
N
(x
0.6
0.6 *
o
;n
.
+
[]~O
0.4
aA
*q
a.
a
n
u. /
o
0.0
100
10 1
10 3
102
10 4
10 5
10
6
107
No. of cycles to failure
Fig : 57 : Comparison of normalized stress S-N curves for all glass fabric
and Carbon fabric composites
113
1000
800
(z
4
us
t;
I
600
Q)
b)
.z
I:
400
200
0
0
2
4
6
8
No. of cycles to failure
Fig: 58 : Comparison of total stress S-N curve for plain weave carbon fabric
with unmodified epoxy [Ref. 21] to the S-N curves for carbon fabric with
unmodified polyester, MNS with 0%. 7.5%, 12.5% and 17.5% rubber. R=-1.
114
max. in it
0.50
a 48% of UTS
* 48% of UTS
H 51% of UTS
0.40-
S
0.30
0
*
4U
U
w
N
0
40% decre ase
in modulus
1.1-
0.20
N
0 ~,
N
.'IMMr
N r
"IM.".
M
;
e",,..m
i
Stage II
Stage II
0.10-
Stage I
__ __ _ ______
0.00
10
1
__I ______I
_ _ _______ I
_ ________
105
102
No. of cycles
Fig : 59: Deflection Vs. No. of cycles for carbon fabric/MNS with 7.5%
rubber laminate
115
-_- -__-;
--_1
._-
_--_-;
-
li___
_i__
_
---;I--=L;--=;·--=-=--r;-;i-
1
· _-1--- i
:=
-
=
=_-1.
- -
-1_·
·
..-;--:-----'L
Fig. 60: Photograph of the damage zone in fatigue tests performed on
chopped strand mat/MNS with 7.5% rubber.
Fig. 61: Photograph of the damage zone in fatigue tests performed on
chopped strand mat/MNS with 12.5% rubber.
116
=
.
i
Fig. 62: Photograph of the damage zone in fatigue tests performed on
chopped strand mat/MNS with 17.5% rubber.
I
Fig. 63: Photograph of the damage zone in fatigue tests performed on
glass fabric/unmodified polyester.
117
Fig. 64 : Photograph of the damage zone in fatigue tests performed on
glass fabric/MNS with 7.5% rubber.
i-----?,-----
------ 1
Fig. 65 : Photograph of the damage zone in fatigue tests performed on
glass fabric/MNS with 12.5% rubber.
118
(a)
(b)
Fig. 66 : SEM micrographs of the fracture surface (longitudinal view) of
chopped strand mat with 7.5% rubber (a) Top surface (b) Bottom surface.
19
_I
_
(a)
- -
(b)
Fig. 67: SEM micrographs of the fracture surface (longitudinal view) of
chopped strand mat with 12.5% rubber (a) Top surface (b) Bottom surface.
120
Ilr
-
-
I
--
-
-
(a)
(b)
Fig. 68 : SEM microarar)hs
of thp fratrfilrPo
-____
-v- ....... ,
. t...
-
ClrFsro
.. L tit.[ ..
t.bL
(If'%"I r; #,,"A; -L.t~t.
.11cl1
I - --
VIeWtLi)l
N,., IC
chopped strand mat with 17.5% rubber (a) Top surface (b) Bottom surface.
121
(a)
(b)
Fi. 69: SEM microravDhs of the fracture surface (longitudinal view) of
glass fabric with unmodified polyester (a) Top surface (b) Bottom surface.
122
-- -
(a)
i
(b)
Fig. 70 : SEM micrographs of the fracture surface (longitudinal view) of
glass fabric with 7.5% rubber (Two different specimens are shown).
123
Fig. 71 : SEM micrograph of the fracture surface (longitudinal view) of
glass fabric with 12.5% rubber composite (Top surface).
1 24
(a)
*
,
-
X
-I-
_
-
1
1
(b)
Fig. 72: SEM micrographs of the fracture surface (longitudinal view) of
carbon fabric with unmodified polvester (a) Top surface (b) Bottom surface.
125
(a)
(b)
Fig. 73 : SEM micrographs of the fracture surface (longitudinal view) of
carbon fabric with 7.5% rubber (a) Top surface b) Bottom surface.
126
(a)
(b)
Fig. 74 : SEM micrographs of the fracture surface (longitudinal view) of
carbon fabric with 12.5% rubber (a) Top surface (b) Bottom surface.
127
(a)
(b)
Fig. 75 : SEM micrographs of the fracture surface (longitudinal view) of
carbon fabric with 17.5% rubber (a) Top surface (b) Bottom surface.
128
12
10
CZ 8
t.
Ck
a
t
6
a4
d
4
x
2
0
.01
.1
1
10
100
1000
10000
Time, hrs
Fig: 76: The deflection Vs. time curve in a creep test on chopped strand
mat with 7.5% rubber.
129
12
154 MPa
10-
e
t%
115 MPa
._
!
'x3
,..a
t%
x-
-
" -- K -
-
-
-
77 MPa
u
4-
>/
43 MPa
- -k
0
_
.01
_
_______
.1
--
_
_
______·
f-----
_
___·I·_
1
10
1I
·
· · 1__·
100
·
·····
..............
1000
·
1
···
T_1
10000
Time, hrs
Fig: 77: The deflection Vs. time curve in a creep test on chopped strand
mat with 12.5% rubber.
130
12
.
120 MPa
10-
t=400 hr
d
ti
tC
90 MPa
(n
90 MPa
._,
o
.
x
u
W
1---
60 MPa
4-
Q)
CZ
33 MPa
0
''
.01
.
....
.1
.....
1
1
10
1
· _
100
·
·
1000
1
1
_I
10000
Time, hrs
Fig: 78 : The deflection Vs. time curve in a creep test on chopped strand
mat with 17.5% rubber.
131
... IU
. '\
253 MPa
8190 MPa
r.
0
6
127 MPa
r.6
o
4
.4.
Ui
70 MPa
F=
.
_ R
A
---
2
_ _ _____
0
.01
____I
_ _______
.. i
.1
. . . . ....
1
........
10
I
100
.
.
.. ...
.
1000
Time, hrs
Fig : 79 The deflection Vs. time curve in a creep test on glass fabric
with 7.5% rubber.
132
10000
10
-
222 MPa
8
tC
167 MPa
r.
111 MPa
't
,.
4-
.4-
62 MPa
-·
0
-
.01
I
.1
·- v- MMWW-Y
·
·
...
1
..
___. . I.
_. ________r..
10
100
_ 1__.
a
.
1000
Time, hrs
Fig: 80: The deflection Vs. time curve in a creep test on glass fabric
with 12.5% rubber.
133
..
10000
12000
11000100009000-
W8000 u
7000 -
~ 6000 A
49 MPa
&45000-
U 4000
49 MPa
-
3000
-
2000
-'-
49 MPa
m
88 MPa
;
132 MPa
-*-
176 MPa
1000
0
(,I
...
I
.01
..1
.1
-- .I1
---- -.--
--
1
-
--
.
10
.
.
.
.I
100
.
..
r..
.
-- I
1000
Time, hrs
Fig: 81: The creep modulus Vs. time curve for chopped strand mat
with 7.5% rubber.
134
l.
1
10000
_____
12000)
11000
10000
9000
ad 8000
V> I VV
. 6000
0o
a^-
43 MPa
i5000-
:*
43 MPa
l-
43 MPa
U 4000 -
3000
-
'
77 MPa
bf
115 MPa
a3
154 MPa
20001000-
0.01
'
'
I ' *1
I
.1
'
'-1
I
1
10
.
.. I1
1))
.
.
..
... .....
10(0
Time, hrs
Fig: 82: The creep modulus Vs. time curve for chopped strand mat
with 12.5% rubber.
135
10000
---9(K)
I
8000
7000
dw6000-
5000-
I
0to
0 4000
-
U 3000
-
---33 MPa
=
33 MPa
33 MPa
a
*
2000-
60 MPa
90 MPa
120 MPa
1000-
0*
.01
·
.... I
.1
..
'1 ' I
.....
1
....' '
10
...
100
I
.
1000
Time, hrs
Fig: 83: The creep modulus Vs. time curve for chopped strand mat
with 17.5% rubber.
136
10000
Ps
_
_
_II
_
-
20000
18000 nnn~/~/
,4000
12000 I&
*
E 8000
MPa
70 MPa
-70
'- 170 MPa
m 1 127 MPa
U
6000-
N
4000-
a
1 190 MPa
253 MPa
2000 0
·
.01
.· '
'".1.. I.."
.1
.1
l
1
10
100
.
1000
......
10000
Time, hrs
Fig: 84: The creep modulus Vs. time curve for glass fabric with 7.5% rubber.
137
Af---
1800016000 14000 2000
oo0000 0
62 MPa
62 MPa
8000
---
U
60 0 0
62 MPa
-
'
4000 -
~
111 MPa
*-
167 MPa
--f-
222 MPa
2000 n-
.01
·
..
· . · l.
.1
.1
·.
..
1
..
...·
10
.
.··.
· iI'l
100
...
.
....
i
1000
11
10000
Time, hrs
Fig: 85: The creep modulus Vs. time curve for glass fabric with 12.5% rubber.
138
200
150
r
co
rn
100
,-4
L;
50
0
0.0
0.5
1.0
1.5
2.0
2.5
Creep strain (%) at various time
Fig: 86: Isochronous stress-strain curve for chopped strand mat with
7.5% rubber at various times.
139
3.0
200
150
4)
--
100
-
L14
50
0
0.0
0.5
1.(
1.5
2.0
2.5
Creep strain (%) at time t
Fig: 87: Isochronous stress-strain curve for chopped strand mat with
12.5% rubber at various times.
140
3.0
__
d 100
ai
50
0
0.0
0.5
1.0
1.5
2.0
2.5
Creep strain (%) at time t
Fig: 88: Isochronous stress-strain curve for chopped strand mat with
17.5% rubber at various times.
141
3.0
300
250
200
vr
0
150
x
%A
L-
100
50
0
0.0
0.5
1.0
1.5
Creep strain (%) at time t
Fig: 89: Isochronous stress-strain curve for glass fabric with 7.5% rubber
at various times.
142
2.0
300
250
200
G)
Pz
4
vr
fA
x
%A
150
X
I
-l
100
50
0
0.0
0.5
1.(
1.5
Creep strain (%) at time t
Fig : 90: Isochronous stress-strain curve for glass fabric with 12.5% rubber
at various times.
143
2.0
Experimental data on 49 MPa
N
* Experimental data on 49 MPa
A Experimental data on 49 MPa
· Experimental data on 88 MPa
N Experimental data on 132 MPa
__ _1_________^_
___^__ __1____^
ExperlIenIaldil uarad Uoi
/
Ivll-ad
3
0
Predicted curve on 49 MPa
Predicted curve on 88 MPa
.
-
-1_
- - 2.
__
_ a_ -
1
1 L
T
-
redicted curve on 13L TVlra
2 Predicted curve on 176 MPa
0
Ue
. I~
04
~
n
1
0
-1
.01
...1
.1
- ----....
... 1
..-.--...
...1
1
10
1()0
....
1000
- --.---.-
. ------.
10000
Time, hrs
Fig: 91 : Predicted and experimental creep strain Vs. time curve for
chopped strand mat with 7.5%/, rubber.
144
100000
Experimental data on 43 MPa
* Experimental data on 43 MPa
A Experimental data on 43 MPa
° Experimental data on 77 MPa
W
[
3.
'-v
.
--.
4- A -r-
1 1 I M In
° Experimental data on 154 MPa
Predicted curve on 43 MPa
'
Predicted curve on 77 MPa
Predicted curve on 115 MPa
Preidcted curve on 154 MPa
02
P4
o
0
-
.01
.1
1
1
1()0
1(0(
10000
Time, hrs
Fig: 92: Predicted and experimental creep strain Vs. time curve for
chopped strand mat with 12.5% rubber.
145
100000
Experimental data on 33 MPa
0
* Experimental data on
A Experimental data on
· Experimental data on
0 Experimental data on
Exynrimpntal data on 120 MPa
°
3I
33 MPa
33 MPa
60 MPa
90 MPa
nF
Predicted curve on 33 MPa
Predicted curve on 60 MPa
cirvp nn qn MPa
Prdirtl
Predicted curve on 120 MPa
o/
0
a)
U
k.
n
or~~
o~~~
~
C)
~
-
~
C
O~~~
10
I-
.
.t
.0 1
~
_ ___·_·_1
.1
·
-- #-
C
-
I
3·
011
0
_Ell
·~-
0-
0~
B0
I
C
_ ____·_1
...1 _ _ _______
1
1()
_ _ ______·
1...
10()
_ _ _______
....
1000 _ _______
1000
_ ______
10000
Time, hrs
Fig: 93: Predicted and experimental creep strain Vs. time curve
for chopped strand mat with 17.5% rubber.
146
100000
Experimental
* Experimental
A Experimental
° Experimental
" Experimental
3
OE
primen
__I
___________
2.0
-Predicted
-
tal
curve
data on 70 MPa
data on 70 MPa
data on 70 MPa
data on 127 MPa
data on 190 MPa
_I__
do
on
___
70
2Zn NMAT)P
__ __ _
___
MPa
Predicted curve on 70 MPa
Predicted curve on 127 MPa
Predicted curve on 190 MPa
Predicted curve on 253 MPa
1.5 30 [[
0
E
0
n3
3
1.0 -
U
0.5
2
p 21I
a
...1 . ---..--0.0 I I - .-- 1-...
.01
.1
1
-.----
1..
,.1I
10
:
100
1000
-------.
- -----
10000
Time, hrs
Fig: 94: Predicted and experimental creep strain Vs. time curve for
glass fabric with 7.5% rubber.
147
100000
Experimental data at 62 MPa
* Experimental data at 62 MPa
A Experimental data at 62 MPa
· Experimental data at 111 MPa
N
Experimental data at 167 MPa
0
2.(
0
1-4
a)
U
1.0
0.0
I¢.
.
10
1
100
1000
10000
Time, hrs
Fig: 95: Predicted and experimental
creepstrain Vs. time curve for
glass fabricwith 12.5% rubber.
148
100000
40
.
:
Chopped strand mat with 7.5% r ibber
6-
Chopped strand mat with 12.5% rubber
°
Chopped strand mat with 17.5% rubber
30.. ..
u
a)
tzi
a)
U
20 -
10 -
Glass fabric with 7.5% rub ber
'
m-----
0-
10
I
20
1
·
I
1
-
1
.
40
30
·
(lcc,
n
A-t
IL.
·
'I
,hr
U
'
50
LI
A Yr;
4h
VVLL
1'
LL.
rl/ ,,
IL
hr
-,
LC1
·
I
60
Stress level (% of ultimate strength)
Fig 96: Relative creep deflection Vs. stress level (on which the
creep test is performed) for different composites
149
o/ /
70
40 I
I
Chopped strand mat tested on a stress level of
16.7% of UFS
6 - 30% of UFS
*
45% of UFS
30 C
0
,-f
.4-
U
cr_:4)
-'
20
4
0
a>
0
.0-
Cd
4
(D
Glass fabric tested on a stress level of
10
0
-
6
·
8
·
1
1
10()
·
12
12
·
'
16.7% of UFS
m
30% of UFS
N
45% of UFS
*
60% of UFS
·
1
14
14
·
16
16
% Rubber in MNS
Fig: 97: Relative creep deflection Vs. rubber content in
MNS at different stress level.
150
18
References
1. Subramaniam, R., FlUieler, Peter H., McGarry, F. J., "Composite based on
Molecular Network system".
2. McGarry, F. J. , Subramaniam, R. , "Molecular Network System : A
Toughened Crosslinked Polyester System".
3. Osborn, Jason W. , "The effect of full flexural fatigue on the flexural
strength composites", Master of Science Thesis, MIT (1991).
4. Kim, Ran Y. , "Fatigue Behavior", Composite Design 4th Ed. (Stephen W.
Tsai, ed.), p 19-1, Think Composites (1988).
5. Talreja, Ramesh, "Fatigue of Composite Materials", Technomic Publishing
Company (1987).
6. Oldyrev, P. P., "Construction of Fatigue curves in hard high-cycle loading
of reinforced plastics", Mechanics of Composite Materials, v. 23, n. 3, Nov
1987, p 319-325.
7. Murkani, R. , Noriaki, S. , Kusumato, N. and Yoshiaki, M. , "Tensile
Fatigue Characteristic of Polymer Blends" Kobunshi Robunshu, Eng. Ed.,
Vol. 5, n 3 (1976).
8. Hertzberg, Richard W. , Manson, John A. , "Fatigue of Engineering
Plastics", Academic Press Inc (1980).
9. Joneja, Surendra K. , "Matrix Contribution to Fatigue Behavior of Glass
Reinforced Polyester Composites", Journal of Reinforced Plastics and
Composites, Vol. 6, n. 4 (1987).
10. Curtis, P. T. , "The Fatigue Behavior of Fibrous Composite Materials",
Journal of Strain Analysis, Vol. 24, No. 4 (1989).
11. Konur, O., Matthews, F. L., "Effect of the properties of the constituents on
the fatigue performance of composites: a review", Composites, Vol. 20, No. 4
(1989).
12. Mandell, J. F., Huang, D. D., McGarry, F. J., "Tensile Fatigue Performance
of Glass Fiber Dominated Composites", Research Report R80-4, MIT (1980).
13. Jones, C. J. , Dickson, R. F. , Adam, T. , Reiter, H. and Harris, B. , "The
environmental fatigue behavior of reinforced plastics", Proc. Royal Soc.
London, A 396, p. 315-338 (1984).
14. Mandell, J. F. , "Fatigue behavior of short fiber composite materials"
Fatigue of Composite Materials Vol. 4 (Ken L. Reifsnider, ed.), p. 231-337,
Elsevier Science Publishers (1991).
151
15. Rotem, A. , "The Fatigue Behavior of Composite Laminates under
Various Mean Stresses", Composites Structures, Vol. 17, p 113-126 (1991).
16. Nguyen, D. H. and Ogale, A. A. , "Compressive and Flexural Creep
Behavior of Carbon Fiber/PEEK composites", Journal of Thermoplastic
Composite Materials, p. 83-99, Vol. 4 (1991).
17. Zhang, S. Y., Xiang, X., Y., "Creep Characterization of a Fiber Reinforced
Plastic Material", Journal of Reinforced Plastics and Composites, p. 1187-1194,
Vol. 11 (1992).
18. Ward, I. M. , "Mechanical Properties of Solid Polymers", Second Edition,
John Wiley & Sons (1983).
19. ASTM D 790-90 Flexural Properties of Unreinforced and Reinforced
Plastics and Electrical Insulating Materials, Annual Book of ASTM Standards
(1990).
20. ASTM D 671 Flexural Fatigue of Plastics by Constant-Amplitude-of-Force,
Annual Book of ASTM Standards (1990).
21. Chou, Shen, Chen, Hong-Chu and Lai, Che-Chun, "The fatigue properties
of weft-knit fabric reinforced epoxy resin composites", Composites Science
and Technology 45, p. 283-291 (1992).
22. Hayashi, I. , Asanuma, K. , Niwa, K. , Fukano, M. and Maruyama, N.,
"Mechanism of Damage Propagation in CFRP/AFRP Hybrid Laminates under
Fatigue Flexural Loading", JSME International Journal, Series I, Vol. 34, No. 2
(1991).
23. Newaz, Golam M. , "Influence of Matrix Material on Flexural Fatigue
Performance of Unidirectional Composites", Composites of Science and
Technology 24, p. 199-214 (1985).
24. Shih, G. C. , Ebert, L. J., "The Effect of the Fiber/Matrix Interface on the
Flexural Fatigue Performance of Unidirectional Fiberglass Composites",
Composites Science and Technology 28, p. 137-161 (1987).
25. ASTM D 2990-90 Standard Test Methods for Tensile, Compressive, and
Flexural Creep and Creep-Rupture of Plastics, Annual Book of ASTM
Standards (1990).
26. Flueler, P. , Motavalli, M. , "Creep Behavior of MNS-Composites
Reinforced with Chopped Glass Mats and Woven Glass fabrics", EMPA Nr.
148465, September 1993.
152
27. Tuttle, M. E. , Brinson, H. F. , "Prediction of the Long-Term Creep
Compliance of General Composite Laminates", Experimental Mechanics,
36(1), p. 89-102 (1986).
28. Dillard, D. A. , "Viscoelastic behavior of laminated composite materials",
Fatigue of Composite Materials Vol. 4 (Ken L. Reifsnider, ed.), p. 339-384,
Elsevier Science Publishers (1991).
29. Schapery, R. A. , "On the Characterization of Nonlinear Viscoelastic
Materials", Polymer Engineering and Science, Vol. 9, No. 4, p. 295-310 (1969).
153