University of Southern California Dissertations and Theses

Ageing and mechanical failure of fiber reinforced polymers

(USC Thesis Other)

Ageing and mechanical failure of fiber reinforced polymers

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

AGEING AND MECHANICAL FAILURE OF FIBER REINFORCED POLYMERS

by
Yinghui Hu

A Dissertation Presented to
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MECHANICAL ENGINEERING)

August 2015

Copyright 2015 Yinghui Hu
i

ii

DEDICATION
This manuscript is dedicated to my family, who continue to provide unconditional love
and support in all aspects of my life.

iii

ACKNOWLEDGEMENTS
I would to like thank my supervisor - Dr. Steven R. Nutt for all his support and
guidance through my Ph.D. study. I learned how to conduct research, do experiment and
write paper from him. The first years of my Ph.D. study were difficult, because I did not
have an idea what to do and how to do research. Dr. Nutt guided me through these years,
and gradually I learned the ability of critical thinking and scientific researching. Without
his help and support, I would not have been able to get through the graduate school. So I
want to thank Dr. Nutt from the bottom of my heart.
I would also like to thank Dr. Nikhil K. Kar who was a senior graduate student and
was also my mentor during my first two years of Ph.D. study. He taught me how to do
mechanical experiment and we had successful collaboration. Without his help and
guidance, I would definitely take longer time to learn how to research and would not have
done much work in early years.
I would also like to thank Augustus W. Lang who was an undergraduate student
helping me in later years of my Ph.D. study. Gus did so much work and provided
significant amount of data which strongly supported my research work. His intelligence
and scientific insight during experiment inspired me.
Finally, I would like to thank my colleagues Yunpeng Zhang, Xiaochen Li, Dr. Li
Yuan, Yuzheng Zhang, Dr. Lessa K. Grunenfelder, Sidi Huang for all their support. I
would not have been able to complete this work without the scientific discussions and
collaborations with them.

iv

TABLE OF CONTENTS

Dedication ........................................................................................................................... ii
Acknowledgements ............................................................................................................ iii
Table of Contents ............................................................................................................... iv
List of Tables .................................................................................................................... vii
List of Figures .................................................................................................................. viii
Abstract ............................................................................................................................... x
Chapter 1. Introduction ................................................................................................... 1
1.1 Fiber Reinforced Polymers ....................................................................................... 1
1.2 Environmental Ageing .............................................................................................. 4
1.3 Mechanical Properties .............................................................................................. 6
1.4 Finite Element Method ............................................................................................. 7
1.5 Fatigue Testing ....................................................................................................... 10
Chapter 1 References .................................................................................................... 14
Chapter 2. Hygrothermal Ageing of Polydicyclopentadiene Composites.................. 17
2.1 Abstract ................................................................................................................... 17
2.2 Introduction ............................................................................................................ 18
2.3 Experimental ........................................................................................................... 20
2.3.1 Sample preparation ........................................................................................... 20
2.3.2 Water immersion ageing ................................................................................... 22
2.3.3 Static tension..................................................................................................... 22
2.3.4 Tension-tension fatigue .................................................................................... 22
2.3.5 Acoustic emission ............................................................................................. 23
2.3.6 Dynamic mechanical analysis (DMA) ............................................................. 24
2.3.7 Fracture toughness ............................................................................................ 24
2.3.8 Short-beam shear .............................................................................................. 25
2.4 Results and discussion ............................................................................................ 25
2.4.1 Matrix plasticization ......................................................................................... 25
v

2.4.2 Effect of moisture uptake on fatigue properties ............................................... 31
2.4.3 Asynchronous ageing of fiber, matrix and interface. ....................................... 35
2.4.3 DI water vs. salt water ageing. ......................................................................... 43
2.4.4 Fatigue damage localization. ............................................................................ 45
2.5 Conclusions ............................................................................................................ 46
Chapter 2 Appendix ...................................................................................................... 49
Chapter 2 References .................................................................................................... 51
Chapter 3. Thermal Oxidation Ageing of Polydicyclopentadiene Composites ......... 55
3.1 Abstract ................................................................................................................... 55
3.2 Introduction ............................................................................................................. 55
3.3 Experiments ............................................................................................................. 58
3.3.1 Sample preparation ........................................................................................... 58
3.3.2 Ageing conditions ............................................................................................. 59
3.3.3 Dynamic mechanical analysis .......................................................................... 59
3.3.4 Optical microscope ........................................................................................... 60
3.3.5 Static tensile test ............................................................................................... 60
3.3.6 Short-beam shear test ........................................................................................ 61
3.4 Results and Discussion ............................................................................................ 61
3.4.1 Weight change during thermal ageing .............................................................. 61
3.4.2 Dynamic mechanical analysis .......................................................................... 62
3.4.3 Surface aged layer ............................................................................................ 64
3.4.4 Static tensile property ....................................................................................... 67
3.4.5 Short-beam shear strength ................................................................................ 72
3.5 Conclusions ............................................................................................................. 73
Chapter 3 References .................................................................................................... 75
Chapter 4. Transverse Compression Failure of Unidirectional Composites............. 78
4.1 Abstract ................................................................................................................... 78
4.2 Introduction ............................................................................................................. 78
4.3 Failure Plane Orientation Prediction ....................................................................... 81
4.4 Finite element analysis ............................................................................................ 83
4.4.1 Model Generation ............................................................................................. 83
vi

4.4.2 Matrix Failure Criterion and Progression Rule ................................................ 85
4.4.3 Fiber-Matrix Cohesive Interface ...................................................................... 87
4.5 Experiments ............................................................................................................. 89
4.5.1 Specimen Preparation ....................................................................................... 89
4.5.2 Quasi-Static Compression Test ........................................................................ 90
4.6 Results and Discussion ............................................................................................ 91
4.6.1 Failure Plane Orientation .................................................................................. 91
4.6.2 Stress-Strain Curves ......................................................................................... 96
4.7 Conclusions ............................................................................................................. 98
Chapter 4 References .................................................................................................. 100
Chapter 5. Conclusions and Future Work ................................................................. 104
Appendix Single-Fiber Push-Out Tester .................................................................... 109
A.1 Overview .............................................................................................................. 109
A.2 Apparatus Setup ................................................................................................... 111
A.3 Sample Preparation............................................................................................... 112
Appendix References .................................................................................................. 113
Comprehensive References (Alphabetical) ................................................................. 115

vii

LIST OF TABLES
Table 2.1 Basic properties of cured pDCPD and epoxy neat resin
Table 3.1 Properties of cured pDCPD resin
Table 3.2 Properties of glass fiber

viii

LIST OF FIGURES
Fig. 1.1 Composite core power transmission line
Fig. 1.2 Transverse compression failure of UD composite rod
Fig. 1.3 An example of finite element model
Fig. 1.4 Sketch of a tension-tension fatigue specimen
Fig. 1.5 S-N curve of pDCPD 90° laminate pre-aged condition
Fig. 2.1 Weight change during ageing
Fig. 2.2 tan(δ) curve evolution during ageing from DMA
Fig. 2.3 Fracture toughness of neat cured resin
Fig. 2.4 S-N curves evolution after ageing
Fig. 2.5 Fracture surface of [90° ]4 fatigue samples
Fig. 2.6 Short-beam strength evolution during ageing
Fig. 2.7 Static tensile properties evolution after ageing
Fig. 2.8 Damage evolution of [0° ]2 composites
Fig. 2.9 Accumulated AE energy evolution of [0° ]2 composites
Fig. 2.10 S-N curves evolution in DI water and salt water environments
Fig. 2.11 Fatigue failure of [0° ]2 composites
Fig. 2.12 Optical microscope of 0° polished section
Fig. 2.13 Raman spectrum of PET inclusion
Fig. 3.1 Weight change of samples
Fig. 3.2 DMA curves of neat resin
Fig. 3.3 Surface aged layer of 100 ° C aged resin
Fig. 3.4 Surface aged layer of 150 ° C aged resin
ix

Fig. 3.5 Thickness of oxidized layer evolution
Fig. 3.6 Tensile modulus evolution of 0° composite
Fig. 3.7 UTS evolution of 0° composite
Fig. 3.8 Tensile modulus evolution of 90° composite
Fig. 3.9 UTS evolution of 90° composite
Fig. 3.10 Stress-strain curves of neat resin and composite
Fig. 3.11 SEM image of 90° tensile fracture surface
Fig. 3.12 Short-beam strength degradation versus ageing time
Fig. 4.1 Plastic shear band in hexagonal model
Fig. 4.2 One f = 50% model mesh grid
Fig. 4.3 Interfacial debonding vs. interfacial stress
Fig. 4.4 Digital image correlation pattern of cuboidal sample
Fig. 4.5 Failure plane angle vs. fiber volume fraction
Fig. 4.6 Simulated plastic shear bands
Fig. 4.7 Experimental crack angle and fracture surface
Fig. 4.8 Stress-strain curves of finite element models
Fig. A1 Fiber push-out test of glass fiber composite
Fig. A2 Using 4 μm tip to push glass fiber
Fig. A3 Setup inside SEM chamber

x

ABSTRACT
Fiber reinforced polymers (FRPs) have been widely used in the aerospace industry for
decades. However, the use of FRPs in civilian infrastructure applications is increasing,
particularly in applications such as high-voltage transmission lines, wind turbine blades,
and off-shore oil platforms. In such applications, long-term environmental ageing is a
major concern because these structures, unlike aircraft, are expected to provide decades
of service with minimal inspection or maintenance. This presents a major challenge for
FRPs because ageing has complex effects on mechanical performance, making it difficult
to forecast property retention and predict component lifetime. The overriding focus of the
present work is to determine the effects of different types of ageing on the mechanical
behavior of FRPs. Two FRP systems are considered, each with distinct matrix chemistry
and intended service applications. The first system is a polydicyclopentadiene (pDCPD)
resin with intrinsic hydrophobicity and exceptionally low viscosity, attributes that are
suitable for fabrication of large parts and humid service conditions, such as wind turbine
blades and components for oil and gas production. The second system is a unidirectional
(UD) composite rod intended as the load-bearing core for high voltage power
transmission lines (overhead conductors).
For the first system, pDCPD is an attractive candidate resin for potential use in
corrosive environments because of its intrinsic hydrophobicity, which imparts resistance
to ageing in humid environments. We first investigate the hygrothermal ageing behavior
of pDCPD composites and effects on mechanical behavior, including the fatigue
behavior. Next, we investigate the thermal oxidation ageing of pDCPD resins and
composites to determine effects on mechanical behavior. By investigating the ageing
xi

behavior and resulting degradation in mechanical properties, the water diffusion
mechanism, salt corrosion behavior, and oxidized layer evolution of pDCPD composites
are better understood. These insights provide a basis for lifetime predictions based on
ageing rules and the composite can be optimized to achieve improved performance.
For the second system, UD FRPs have high strength in the longitudinal direction, but
low transverse strength, often resulting in premature damage initiation. The study here
investigates the transverse compression behavior of UD FRPs using both experimental
and finite element simulation. The investigation provides a basis for optimizing the rod
design and improving the characteristics of this new product.

1

CHAPTER 1. INTRODUCTION
1.1 Fiber Reinforced Polymers
Composite materials are comprised of more than one component, each with distinct
characteristics that are combined to achieve synergistic performance. The general
approach to producing composite materials consists of combining reinforcing materials
with a matrix material. The reinforcing material can be fibers, particles, spheres, etc.,
while the matrix materials can be metals, polymers, or ceramics. Polymer matrix
composites (PMCs) consist of a polymer matrix and fiber reinforcement. Polymers are
large organic molecules composed of repeated subunits (monomers), and the carbon
chains act as backbones. Polymers are products from the petrochemical industry, and the
some of the most commonly used matrix resins include epoxy, polycarbonate,
polypropylene, and polyethylene terephthalate. In practical applications, carbon fibers or
glass fibers are most commonly used to reinforce PMCs, and these composites are called
fiber reinforced polymers (FRPs) (also fiber reinforced plastics). Fibers alone are brittle
and cannot be used alone to manufacture structures, but when embedded in polymers,
they can act as the major load-bearing member and prevent brittle fracture, while the
polymer matrix providing load transfer. The major advantages of FRPs include low
density, high stiffness and high strength. Typical disadvantages of FRPs include
difficulties in machining and joining, anisotropy, and relatively high cost, especially for
carbon fiber composites.
Because of the high strength-to-weight ratio, FRPs are widely used in the aerospace
industry [1]. In recent years, FRPs are expanding in civilian infrastructures [2]. One
important civilian application of FRPs is the wind turbine blade. Glass fibers are
2

generally used in wind turbine blade laminates because of low cost relative to carbon
fibers (~10X). Polydicyclopentadiene (pDCPD) is a potential matrix resin for wind
turbine blade applications. The major advantage of the new resin is its intrinsic
hydrophobicity, which makes it suitable for humid service environments (many wind
power plants are built near or in the sea). Compared to traditional epoxy resins, pDCPD
can be formulated with much lower viscosity, which makes it easy to infiltrate large thick
parts, such as the root section of a wind turbine blade. Another advantage is the lower
density of pDCPD, which enhances power efficiency. However, the hygrothermal ageing
and fatigue behavior of pDCPD composites is not well understood and is a major concern
for long-term durability. Thus, studying the water immersion ageing, including seawater
ageing, and the fatigue property evolution after ageing of pDCPD composites are
important for the future application of this material system. This subtopic is discussed in
detail in Chapter 2.
Another application is the new high voltage overhead conductor cables [3]. As shown
in Fig. 1.1, the core of this new type of power transmission line is made of unidirectional
(UD) glass fiber and carbon fiber hybrid composite rod. The advantage of the composite
conductor is low sag at high temperatures, a consequence of the near-zero thermal
expansion of carbon fibers. This low sag allows higher currents and affords increased
ampacity relative to steel reinforced conductors. Cyclic loads are inevitable in service
because of cross-winds and thermal cycles, and thus fatigue behavior is a concern for
long-term durability. Of particular concern is the potential for fatigue damage at the
“dead-ends” where conductors are attached to towers.

3

Fig. 1.1 Composite core power transmission line

Fig. 1.2 shows that during tension-tension fatigue tests, the damage always initiates
within the grip zone in the form of transverse cracking. The reason is that although the
longitudinal strength of UD composites is high, the transverse strength is significantly
lower and become the damage initiation point. Thus, studying the transverse compression
failure of UD composites is important to understand the failure mechanism of this type of
material. And the results findings can help optimizing the design and application of this
new product. This subtopic is discussed in detail in Chapter 4.

4

Fig. 1.2 Transverse compression failure of UD composite rod [3]

1.2 Environmental Ageing
For polymer composites used in civilian infrastructures, long-term durability is a
major concern because these applications are required to last decades with minimal
maintenance. The environmental ageing issue of polymers and polymer composites is a
complex topic and is still an evolving research field [4-8]. Typical ageing factors include
fatigue loading, high temperature, temperature cycling, humidity, freeze-thaw cycling,
ultraviolet radiation, ozone exposure, and salt water corrosion. These environmental
factors can cause polymer oxidation, embrittlement, and cracking. They can also cause
fiber strength degradation and fiber-matrix interface de-cohesion, leading to overall
failure of the composite structure.
For wind turbine blade applications, the ageing factors of greatest concern are
humidity and salt water corrosion [4, 9, 10]. Moisture causes matrix swelling and
accelerates interfacial debonding, leading to the loss of load transfer capability between
5

fiber and matrix. Plasticization of matrix occurs if significant amounts of water molecules
diffuse into polymer networks, which softens the material and decreases the strength and
glass transition temperature (Tg). Glass fiber is sensitive to moisture exposure because of
surface corrosion [11]. The tensile strength of glass fiber can drop by more than 50%
after water immersion.
In hygrothermal ageing research, elevated temperature is often used to accelerate the
diffusion process [5, 6]. Typically, if experiments are conducted at service temperatures,
the required test time is too long for a research project. However, using elevated
temperature has intrinsic problems. First, the relation between the experimental lifetime
at high temperature and the real lifetime at service temperature is difficult to determine.
Although some thermodynamic approaches (Arrhenius equation) and diffusion laws
(Fick’s law) can be referenced to estimate the temperature-life relation, in practice, they
are far often inaccurate because of the complex ageing mechanisms. Second, some ageing
mechanisms such as secondary polymerization may occur at high test temperatures, but
not at service temperatures, especially for ageing at temperatures above Tg. Thus,
although higher temperature can accelerate the ageing process, care must be taken when
selecting the experimental temperature so that it does not cause unrealistic reactions.
For high-temperature applications involving exposure to air, thermal oxidation is a
foremost concern. Previous research on pDCPD oxidation [12] showed that when
pDCPD neat resin samples are aged in air at high temperatures, a thin oxidation layer will
form on the specimen surface, while the inner bulk material is unaffected. This specific
oxidation behavior is called diffusion-limited oxidation (DLO), and it occurs in many
polymers [13]. The thickness evolution and properties of the oxidized layer are important
6

for applications in oxidizing environment. If the oxidized layer is brittle, surface cracks
can initiate due to surface shrinkage or external loads and lead to further oxidation in the
material. As part of the future research plan, an oxidation study of pDCPD is described in
Chapter 3.

1.3 Mechanical Properties
Because of the intrinsic anisotropic nature of fibers, FRPs generally behave
anisotropically, in ways determined by the composite layup. Unidirectional (UD) FRPs
feature all fibers in the same direction, and thus show distinct mechanical properties in
the longitudinal and in the transverse directions. The stiffness and strength in the
longitudinal direction can be more than ten times greater than that in the transverse
direction. Thus, when laying up the composites, the longitudinal direction is normally
aligned with the anticipated loading directions.
The mechanical properties of composites are the most important properties for
structural applications, such as the power transmission line and the wind turbine blade.
Studying the mechanical performance and the ageing effects of environments on the
mechanical performance of these products are of great importance for their applications.
Typical mechanical properties of interest include tensile strength, compressive strength,
(interlaminar) shear strength, fiber-matrix interface strength and fatigue strength. For
long-term low-load applications, fatigue performance is also a priority. Failure of FRPs
initiates and propagates in different modes, determined by loading conditions and
composite structures.
7

There are three constituents in FRPs (fiber, matrix and interface), and thus there are
three major failure mechanisms: fiber breakage, matrix cracking, and interface
debonding. Although neat polymers can be ductile or brittle, the matrix in a composite
generally fails by brittle cracking because of the severe stress/strain concentration in the
matrix [14]. Fiber breakage occurs when the tensile load in the longitudinal direction of
the fiber exceeds its ultimate tensile strength. Thus, this type of failure can appear under
longitudinal tensile load and bending load, and is rarely observed under transverse load
and shear load. Fiber breakage requires the highest load and releases the greatest energy
in all three failure modes. Matrix cracking can occur in various loading modes: tensile,
shear or compressive. For ductile polymers, matrix shear bands and ridges can be
observed from the fracture surface. The fiber-matrix interface transfers load between
fiber and matrix. According to previous research [15, 16], impaired interfaces can lead to
significant loss of fatigue strength. The interface is more sensitive to environmental
ageing than fiber and matrix. Therefore, studying the interface property and its evolution
versus ageing time is of great importance for applications in ageing and fatigue
environments. The direct measurement of interface shear strength is achieved using a
single fiber push-out device, which is discussed in Appendix.

1.4 Finite Element Method
Finite element analysis (FEA) (also known as finite element method, FEM) is a
numerical modeling technique to solve partial differential equations (PDEs) or integral
equations. The primary application of FEA today is to solve mechanical problems such as
static/dynamic stress/strain analysis, thermal-mechanical analysis, fluid-structure
8

interaction modeling, impact modeling, coupled mechanical-electrical-magnetic field
modeling, acoustic and shock analysis and etc.
The basic concept of FEA is to mesh the continuum material into small elements (Fig.
1.3), and then all the variables (loads, material properties, displacements, stress, and etc.)
are discretized for computer processing. The basic equation of FEA is Eqn. 1.1:
Ku = P (1.1)
in which u is the displacement vector which is the unknown, P is the load vector and K is
the stiffness matrix. In real problem, the unknown u can be displacement, temperature
and etc.; P can be load, displacement boundary condition, heat flux boundary and etc.; K
is determined by material properties such as elastic modulus, plasticity, thermal
properties and etc. The commercial FEA software today can provide geometric model
building tool, the solver to solve u and finally a tool to display the simulation results.

Fig. 1.3 An example of finite element model [17]

There are several different approaches for modeling of FRPs depending on the type of
the problem and desired results. For large scale and complex shape part modeling, the
results of interest are the overall response of the structure to external load and
9

environment, while the micro scale stress/strain status is neglected. In this type of
modeling, the composite material is usually simulated as homogeneous, and the micro
level fiber-matrix interaction is not modeled because of the unnecessary computation and
computing capability limit.
However, when the micro-level fiber-matrix interaction and microscopic material
failure are of interest, the above macro level approach is no longer suitable and detailed
modeling of the microscopic constituents is required. Because of the large dimensional
scale difference between the constituent level (µ m) and the structure level (m), a limited
area of the microscopic structure (~100 µ m) is usually modeled to represent the whole
composite. This method is called representative volume element (RVE) method and is
widely used in microscopic finite element modeling [18-20].
In RVE modeling of FRPs, each individual fiber is sketched and the material
properties of the constituents are input parameters. The modulus, strength and Poisson’s
ratio of the fiber, and the modulus, Poisson’s ratio, yield strength, flow rule and ultimate
strength of the matrix, and the debonding law of the interface are defined prior to the
analysis. Another issue of the RVE method is that the simulation results can be quite
sensitive to model size. If the model size is too small and there are not enough fibers to
represent the whole composite material, the results can be largely scattered from one
model to another. Thus, model size validation is a required step for every RVE simulation
to make sure that the model is large enough to represent the whole composite.

10

1.5 Fatigue Testing
Fatigue is a major issue for applications involving cyclic loading, such as the wind
turbine blade. Fatigue testing can be classified according to loading mode: tension-
tension fatigue, tension-compression fatigue, compression-compression fatigue and
bending fatigue. Certain specific loading modes must be studied depending on the
loading modes in service. For tensile fatigue properties, a world-wide recognized test
standard (ASTM D3479) is used for FRPs, and thus the test results from different
researchers are comparable. In contrast, results from bending fatigue tests (3-point or 4-
point) are relatively difficult to interpret and there is no widely accepted test standard.
Fig. 1.4 shows a sketch of a tension-tension fatigue specimen. The two ends of the
specimen are tabbed to lessen the stress concentration within the grip zone, preventing
premature and unrealistic failure. Acoustic emission sensors (channel 1 and 2) can also be
used to pick up the acoustic signals generated during the fatigue test. The acoustic
emission technique will be discussed in detail in Chapter 2.

Fig. 1.4 Sketch of a tension-tension fatigue specimen
11

During the fatigue test, the two tabbed ends will be fixed within clamps which are
connected to the load frame. A tension-tension cyclic load is applied to the specimen at a
certain frequency, thus the stress status in the gauge length is cyclic tension-tension. The
stress ratio R is defined in Eqn. 1.2, in which σmin and σmax are the minimum and
maximum stresses in one cycle. Because in a tension-tension fatigue test, both σmin and
σmax are positive, then the range of R is from 0 to 1. Stress amplitude (σamp) is defined in
Eqn. 1.3, while stress range (σrange) is defined in Eqn. 1.4. All these parameters are related
and can be calculated from each other.

min
max
R


 (1.2)
max min
amp
2



 (1.3)
range max min
   
(1.4)

Fatigue loading frequency is defined as the number of load cycles applied in one
second. Loading frequency can have a significant influence on fatigue test results because
of the heating effect [21]. Heating occurs in the specimen for all fatigue tests and
different materials respond differently to loading frequency changes. Generally, higher
loading frequency leads to greater energy dissipation and greater temperature rise. For
certain materials, a slight increase in temperature can significantly decrease the fatigue
life. Thus, during a fatigue test, the specimen temperature should always be monitored to
12

detect if significant heating occurs. And multiple frequencies should be used and
compared to select a proper loading frequency before batch tests.
The most commonly used chart to represent a material’s fatigue property for both
metal and composite is the S-N curve (Wö hler curve). S indicates the maximum cyclic
stress (σmax) while N indicates the number of cycles to failure. Fig. 1.5 shows an example
of a tension-tension S-N curve of pDCPD composite. The ultimate tensile strength (UTS)
for this material is 60 MPa, and the tests were carried out at three load levels: 80% UTS
(S = 48 MPa), 60% UTS (S = 36 MPa) and 50% UTS (S = 30 MPa). And three specimens
were tested at each load level, and thus the replication level (Eqn. 1.5) of this test group is
greater than 0.5, which is sufficient for research purpose according to ASTM E739.

Fig. 1.5 S-N curve of pDCPD 90
o
laminate pre-aged condition

13

total number of load levels
replication 1
total number of specimens

(1.5)

As shown in Fig. 1.5, the S-N curve in a limited stress range generally exhibits linear
behavior in a linear-log plot. Fatigue life increases exponentially as load decreases
linearly. For all fatigue tests, a run-out cycle is generally chosen beyond which the test is
manually stopped regardless of whether the specimen fails or not (essentially setting a
test time limit). The run-out cycle should be selected before all tests based on the
interested loading range or the lifetime range. In Fig. 1.5, the run-out cycle is selected as
10
6
cycles, which covers the medium to high load level (> 50% UTS).
The S-N curve is affected by parameters such as stress ratio R, ageing condition, test
frequency, and ageing time. Because mechanical properties of all components in the
composite degrade after ageing, we expect a drop of the S-N curve after ageing. From the
decrease in magnitude and S-N curve slope change, and by comparing the S-N curves
from different ageing environments and different ageing periods, the fatigue properties
and failure mechanisms of the composites can be understood, leading to design
improvements, lifetime prediction, and facilitating future applications in severe
environments.

14

Chapter 1 References
1. Kumar S, Murthy Reddy KVVS, Kumar A, Rohini Devi G, Development and
characterization of polymer-ceramic continuous fiber reinforced functionally graded
composites for aerospace application, Aerospace Science and Technology
2013;26:185-191.
2. Liao K, Schultheisz CR, Hunston DL, Long-term environmental fatigue of pultruded
glass-fiber-reinforced composites under flexural loading, International Journal of
Fatigue 1999;21:485-495.
3. Kar NK, Hu Y, Barjasteh E, Nutt SR, Tension-tension fatigue of hybrid composite
rods, Composites B 2012;43:2115-2124.
4. Gellert EP, Turley DM, Seawater immersion ageing of glass-fibre reinforced polymer
laminates for marine applications, Composites A 1999;30:1259-1265.
5. Mouzakis DE, H. Zoga, Galiotis C, Accelerated environmental ageing study of
polyester/glass fiber reinforced composites (GFRPCs), Composites B 2008;39:467-
475.
6. Foulc MP, Bergeret A, Ferry L, Ienny P, Crespy A, Study of hygrothermal ageing of
glass fibre reinforced PET composites, Polymer Degradation and Stability
2005;89:461-470.
7. Arun KV, Basavarajappa S, Sherigara BS, Damage characterisation of glass/textile
fabric polymer hybrid composites in sea water environment, Materials and Design
2010;31:930-939.
15

8. Assarar M, Scida D, El Mahi A, Poilâ ne C, Ayad R, Influence of water ageing on
mechanical properties and damage events of two reinforced composite materials:
Flax-fibres and glass-fibres, Materials and Design 2011;32:788-795.
9. Poodts E, Minak G, Zucchelli A, Impact of sea-water on the quasi static and fatigue
flexural properties of GFRP, Composite Structures 2013;97:222-230.
10. Le Gac PY, Choqueuse D, Paris M, Recher G, Zimmer C, Melot D, Durability of
polydicyclopentadiene under high temperature, high pressure and seawater (offshore
oil production conditions), Polymer Degradation and Stability 2013;98:809-817.
11. Schmitz GK, Metcalfe AG, Stress Corrosion of E-Glass Fibers, I&EC Product
Research and Development 1966;5:1-8.
12. Richaud E, Le Gac PY, Jacques V, Thermooxidative ageing of polydicyclopentadiene
in glassy state, Polymer Degradation and Stability 2014;102:95-104.
13. Celina MC, Review of polymer oxidation and its relationship with materials
performance and lifetime prediction, Polymer Degradation and Stability
2013;98:2419-2429.
14. Hu Y, Kar NK, Nutt SR, Transverse compression failure of unidirectional
composites, Polymer Composites 2014.
15. Kotsikos G, Evans JT, Gibson AG, Hale JM, Environmentally enhanced fatigue
damage in glass fibre reinforced composites characterised by acoustic emission,
Composites A 2000;31:969-977.
16. Patel SR, Case SW, Durability of hygrothermally aged graphite/epoxy woven
composite under combined hygrothermal conditions, International Journal of Fatigue
2002;24:1295-1301.
16

17. Abaqus user manual, v6.11, Dassault Systemes 2011.
18. Gonzalez C, LLorca J, Mechanical behavior of unidirectional fiber-reinforced
polymers under transverse compression: Microscopic mechanisms and modeling,
Composites Science and Technology 2007;67:2795-2806.
19. Totry E, Gonzalez C, LLorca J, Failure locus of fiber-reinforced composites under
transverse compression and out-of-plane shear, Composites Science and Technology
2008;68:829-839.
20. Romanowicz M, A numerical approach for predicting the failure locus of fiber
reinforced composites under combined transverse compression and axial tension,
Computational Materials Science 2012;51:7-12.
21. Mivehchi H, Varvani-Farahani A, The Effect of Temperature on Fatigue Strength and
Cumulative Fatigue Damage of FRP Composites, Procedia Engineering 2010;2:2011-
2020.

17

CHAPTER 2. HYGROTHERMAL AGEING OF
POLYDICYCLOPENTADIENE COMPOSITES
2.1 Abstract
We investigated the effects of hygrothermal ageing on the tension-tension fatigue
behavior of unidirectional (UD) glass/polydicyclopentadiene (pDCPD) composites.
Samples were immersed in deionized (DI) water and salt water, and glass/epoxy
composites were used as a benchmark for comparison. Composites of pDCPD showed
less water uptake and superior fatigue performance compared to similarly aged epoxy
composites, a distinction attributed to the intrinsic hydrophobicity of the pDCPD resin.
Superior fiber-matrix interface adhesion in pDCPD composites accounted for the greater
strength retention after ageing. Degradation of fiber and interface were coupled but not
synchronous: glass fiber degradation was determined by ageing time, while interface
degradation depended primarily on moisture level. Salt water influenced the amount of
water absorption, but no salt water corrosion was observed for either composite.

Keywords: Polymer-matrix composites (PMCs); Environmental Degradation; Fatigue;
Acoustic emission.

18

2.2 Introduction
The use of polymer matrix composites (PMCs) is expanding to civilian infrastructure,
energy, and marine applications. Despite the significant performance advantages of
PMCs compared to traditional materials, long-term durability remains a major concern in
such applications, particularly those where materials are expected to provide decades of
outdoor service with minimal inspection and maintenance. For example, wind power
turbine blades require resistance to long-term humidity, cyclic temperatures and loads,
UV radiation and seawater ageing, especially in offshore installations, where inspection
and monitoring of structural health pose challenges.
Studies of moisture ageing of PMCs have shown that hygrothermal exposure can
affect fibers, matrix and interfaces in different ways [1-17]. For example, carbon fibers
are reportedly inert to humid environments [12], while glass fibers are sensitive to
moisture exposure [11]. The strength decrease of glass fiber is typically a result of
surface corrosion through an ion exchange mechanism [18]. Similarly, polymers and
interfaces exhibit a wide range of responses to hygrothermal exposure that reflects the
diversity of chemical and structural effects.
Hygrothermal ageing of polymer matrices often involves multiple chemical and
physical mechanisms operating concurrently, and thus presents complex challenges.
Water molecules diffuse into polymer networks and act as a plasticizer when they exist in
a free state. Plasticization softens the matrix and decreases the glass transition
temperature (Tg), modulus and strength [19]. On the other hand, moderate plasticization
also can enhance fracture toughness by impeding crack propagation [20]. In most cases,
water-based plasticization is reversible after drying. On the other hand, for some
19

thermosets after long-term ageing, water molecules can bond strongly with polymer
chains as additional cross-linking, increasing Tg and strength [21].
Additional ageing processes can occur with or without the presence of moisture. For
example, physical ageing is an important issue for polymers during extended high-
temperature exposure. In the structural recovery (relaxation) process, free volume
decreases and polymer chains become more densely packed, which results in
strengthening and shrinkage. Darkening after ageing arises from chemical changes in the
resin, such as oxidation. Oxidation can occur during ageing, although in water
immersion, and is generally limited to a thin surface layer (diffusion limited oxidation,
DLO) if no surface cracking occurs and thus has negligible influence on overall
mechanical properties. Physical ageing and oxidation are generally considered
irreversible.
Hygrothermal ageing of the fiber-matrix interface reportedly causes fatigue strength
degradation of PMCs [13, 15]. Matrix swelling is generally detrimental to the interface
due to the resulting normal tensile stress that facilitates interface separation. Interface
debonding is often observed after water immersion [14], reducing the load transfer
capability between fiber and matrix. The debonded interfaces retain water (“wicking”)
and become capillary diffusion pathways, which in return accelerate the ageing process.
Multiple studies of the effects of hygrothermal ageing on the fatigue behavior of
PMCs have been reported [10-17]. However, further investigation is warranted,
particularly for PMCs based on non-crimp fabrics, which are often used for wind blades.
Polydicyclopentadiene (pDCPD) is a potential matrix material for wind turbine blades,
offshore oil structures, and automotive parts because of the inherent hydrophobicity and
20

the resistance to chemical corrosion [22-25]. Although the fatigue behavior of pre-aged
pDCPD composites has been reported [26, 27], the effect of hygrothermal ageing on the
fatigue behavior has not been systematically investigated. In this study, we report the
effects of ageing in deionized water and salt water environments on the tension-tension
fatigue behavior of UD glass/pDCPD laminates. A conventional glass/epoxy composite is
used as a reference material for comparison, and acoustic emission (AE) is employed to
monitor damage evolution during fatigue tests. The evolution of fatigue behavior with
ageing time is described, and mechanisms involved in fatigue strength degradation are
identified. Results showed that pDCPD composites absorbed less water than epoxy
composites and exhibited superior fatigue resistance. This phenomenon was primarily
attributed to the superior resistance of the hydrophobic pDCPD to water absorption. Both
deionized water and salt water ageing environments influenced the amount of water
absorption, although no salt corrosion effects were observed.

2.3 Experimental
2.3.1 Sample preparation
UD glass/pDCPD laminates and UD glass/epoxy laminates were produced by a
commercial source (Materia, Inc., Pasadena, CA) using common vacuum infusion
processing techniques. The pDCPD resin (Proxima
TM
, Materia, Inc., Pasadena, CA) and
the epoxy resin (Epikote
TM
MGS RIMR 135 resin with RIMH 137 hardener, Momentive,
Inc.) were selected for composite matrices. The pDCPD was formulated using a
ruthenium-based catalyst (Grubbs Catalyst
TM
). This formulation shows favorable
toughness, viscosity, and chemical resistance compared to traditional pDCPD
21

formulations, but the long-term ageing behavior is not yet well understood. The pDCPD
laminates were cured at 30 ° C for 2 hours, and then post-cured at 100 ° C for 30 minutes.
The epoxy laminates were cured at 80 ° C for 8 hours. Table 2.1 shows basic properties of
the two cured resins.

Table 2.1 Basic properties of cured pDCPD and epoxy neat resin
Tg density tensile
modulus
ultimate tensile
strength
tensile
elongation
(° C) (g/cm
3
) (GPa) (MPa)
pDCPD 124 1.05 3.1 73 2.7%
epoxy 101 1.15 2.9 64 3.4%

The laminates were prepared using non-crimp fabric (E-LT 3500, Vectorply, Corp.)
comprised of E-glass (94 wt% PPG Hybon
®
2026 in the warp direction and 6 wt% PPG
Hybon
®
2002 in the weft direction). The properties of the glass fiber are modulus = 82.7
GPa, tensile strength = 2790 MPa, fiber diameter = 17 μm. The glass fibers featured a
polyethylene terephthalate (PET) sizing, which coalesced in matrix-rich regions of the
cured composites. Thus, the fiber sizing had a negligible effect on interface properties.
Analysis of fiber sizing is included in the Appendix.
Two thicknesses of laminates were fabricated: 2-ply laminates were used for 0°
testing and denoted as [0° ]2, while 4-ply laminates were used for 90° testing and denoted
as [90° ]4. The thickness of [0° ]2 laminates was 1.6 mm, while the thickness of [90° ]4
laminate was 3.2 mm. The fiber volume fraction for both laminates was ~58%
(determined by burn-out method).

22

2.3.2 Water immersion ageing
The two ageing environments consisted of (1) deionized water at 60 ° C and (2) 3.5
wt% NaCl solution at 60 °C. The former will be referred as “DI water” and the latter as
“salt water.” Laminate panels were cut to (200×200) mm plates and immersed for ageing.
The plate edges were not sealed during ageing, allowing accelerated diffusion through
fiber-matrix interface. Neat resin samples were also aged for comparison. Temperature
and salinity were monitored and kept constant. Samples were taken out periodically to
measure weight change.

2.3.3 Static tension
Quasi-static tensile tests were conducted following ASTM D3039. The dimensions
of test coupons were (200× 25) mm (length × width), and tabs were used on the two ends
of the sample (50 mm long and 1.6 mm thick). Thus the gauge length of the sample was
100 mm. Fiberglass tabs were bonded to the specimen using epoxy adhesive. Specimens
were pulled to fracture on a load frame (Instron 5567) at a loading rate of 2 mm/min. An
extensometer was used to determine tensile strain values.

2.3.4 Tension-tension fatigue
Tension-tension fatigue tests were conducted in air in a well-ventilated room on a
hydraulic load frame (Instron 8500R-1331) in accordance with ASTM D3479 (for
testing) and ASTM E739 (for data processing). Samples were the same as for static
tensile tests. Load control was implemented using a stress ratio of R = 0.1, a loading
frequency of 10 Hz, and a run-out cycle of 10
6
.
23

Considering the possible effect of sample heating during fatigue tests, which is
sensitive to loading frequency [28], fatigue tests of 10 Hz were compared with 5 Hz at
different load levels using a group of pre-aged samples. No difference was observed in
fatigue life. A temperature rise was observed only when extensive damage occurred near
final rupture, and the maximum temperature rise was less than 3 ° C for both frequencies.
Fatigue fracture surfaces of [90° ]4 samples were examined by scanning electron
microscopy (JEOL JSM-6610), after sputtering gold on the sample surface.

2.3.5 Acoustic emission
Acoustic signals generated by internal damage during fatigue tests were recorded
using an acoustic emission collection system (Physical Acoustics, PCI-2 based AE
systems). Two sensors were attached at the two ends of the gauge length of the sample
using a hot glue gun. The piezoelectric sensors detect transient acoustic waves generated
from a release of localized damage event, recorded as a hit signal. The sensors also can
be used to measure the wave energy and the position of the event, calculated from the
difference in arrival time between the two sensors. Acoustic emission is a useful
nondestructive testing (NDT) technique for composites, as the energy of signals can be
correlated to different damage mechanisms.
Noise from the fatigue testing system was removed using a filter in the AE software.
Thresholds of 70 dB and 60 dB were used for 0° and for 90° fatigue test, respectively.
(The load level in 0° tests was greater and introduced higher noise levels.) Therefore,
some low energy signals, e.g., from matrix and interface cracking below the threshold,
were not captured.
24

2.3.6 Dynamic mechanical analysis (DMA)
Dynamic mechanical analysis (DMA, TA Instruments, Q800) was conducted to
monitor the change in glass transition temperature (Tg) and thermomechanical properties
after ageing. A single cantilever beam sample was used (ASTM D7028), and samples
were cut from 4-ply composite laminates to standard dual-cantilever beam dimensions
(35× 12× 3.2 mm, length×width×thickness). The temperature was ramped from 40 ° C to
160 °C at a rate of 5 °C/min. A loading frequency of 1 Hz and strain amplitude of 2 μm
were used, and the storage modulus, loss modulus and tan(δ) curves were recorded.

2.3.7 Fracture toughness
The fracture toughness of neat polymer samples was measured to determine the
effects of hygrothermal ageing (plasticization or strengthening). Tests were conducted in
accordance with ASTM D5045, and test samples were single-edge-notch beams (SENB).
The dimensions of the SENB sample were (80× 16× 8) mm (length×width×thickness),
determined by several iterations of trial tests to yield valid measurements. Neat polymer
dogbone samples (ASTM D638) were also aged and tested in tension for validation
purposes. The dimensions of the dogbone sample were 70× 10× 4 mm (gauge
length×width×thickness).
The SENB sample was first notched and pre-cracked, then loaded in 3-point bending
(3PB) in failure using a load frame (Instron 5567). The notch and pre-crack were created
after ageing to avoid blunting during ageing. The span length of the 3PB fixture was 32
mm, and the loading rate was 10 mm/min. At least three samples with valid results were
25

tested per test condition to yield reliable data. The load-displacement curve was recorded
to calculate the mode I fracture toughness (K IC).

2.3.8 Short-beam shear
The interlaminar shear strength (ILSS) was evaluated by conducting short-beam shear
(SBS) tests according to ASTM D2344. Samples were cut from 4-ply laminates to
dimensions of (19× 6.4× 3.2) mm (length×width×thickness), and the fiber direction was
along the length direction. Samples were loaded in 3-point bending (3PB) at a loading
rate of 1 mm/min until first failure occurred, and at least five samples were tested per test
condition. The span of the 3PB fixture was set to 12.7 mm, in accordance with the test
standard.

2.4 Results and discussion
2.4.1 Matrix plasticization
Fig. 2.1 shows the weight change for different samples in the two ageing
environments for ageing periods up to one year. The sampling points for Figs. 2.1a, b and
d are 0, 1, 3, 6 and 12 months and they share the same symbol definitions as in Fig. 2.1b,
while sampling points for Fig. 2.1c are weekly. The solid curves in Figs. 2.1a, b and d are
fits to the data.

26

Fig. 2.1 Weight change during ageing

For neat polymer SENB samples (Fig. 2.1a), pDCPD samples exhibited a slight
decrease in weight in the first month, then gradually gained weight over the following 11
months. Saturation was not reached despite ageing for one year. The weight decrease
noted in the first month is attributed to small molecules, such as DCPD monomers and
oligomers, diffusing out of the sample. The subsequent weight increase is caused by
water molecules diffusion into the polymer network (water molecules can exist in a free
27

or bound state with the polymer chains). In comparison, epoxy samples showed much
faster and greater weight gain and approached saturation after 12 months of ageing.
While small molecules may diffuse out of the sample, this effect is more than
compensated by the much faster and concurrent water absorption process. The significant
difference in the amount of water absorbed by pDCPD and epoxy resins is readily
explained by the intrinsic hydrophobicity of pDCPD and the hydrophilic hydroxyl groups
in epoxy networks. The addition of salt to the ageing solution leads to reduced water
absorption for the epoxy samples, which is attributed to osmotic effects [17]. In contrast,
pDCPD samples showed negligible weight change during salt water ageing.
Neat resin dogbone samples (Fig. 2.1b) showed faster water absorption rates than
SENB samples (Fig. 2.1a), indicating the diffusion process for neat polymer samples was
dominated by diffusion through exposed surfaces. The dogbone sample had a larger
surface-to-volume ratio (0.73 mm
2
/mm
3
) than the SENB sample (0.4 mm
2
/mm
3
). The
saturation weight gain of epoxy dogbone samples (~3%) is similar to SENB samples. The
pDCPD dogbone sample did not reach saturation after one year of ageing.
Composite DMA samples (Fig. 2.1c) showed faster weight gain than tensile samples
(Fig. 2.1d), and the rates scale with surface-to-volume ratios (0.85 mm
2
/mm
3
of DMA
sample vs. 0.64 mm
2
/mm
3
of tensile sample). Note that 4-ply laminates (200× 200× 3.2
mm, length×width×thickness) were first aged, then sliced into tensile coupons just before
testing. Thus, the water diffusion mechanism for these laminates was identical to that for
the neat polymer. The one exception - the epoxy composite showed much greater water
absorption when aged in DI water compared to ageing in salt water, and the difference
was greater for the epoxy composites than for the neat polymer samples. This
28

phenomenon can only be explained by an additional mechanisms for water diffusion and
storage operating in the composites - capillary diffusion along debonded fiber-matrix
interfaces [8]. (The possibility of water absorption by voids was discounted because
voids were not detected in microscopic examinations of polished sections of both
composites.) Extensive interface debonding occurred during ageing of epoxy composites
in DI water, but not in salt water. Fig. 2.1c also shows that for epoxy composites aged in
DI water, the weight decreased slightly after reaching saturation. This decrease was
attributed to the leaching of small molecules created by chain-scission [18]. For pDCPD
composites, the amount of water absorption in the two ageing environments was similar,
which is consistent with the ageing of the pDCPD neat polymer (Fig. 2.1a), indicating the
resistance of pDCPD to salt water ageing.
Fig. 2.2 shows the tan(δ) curves for the two composites as a function of temperature
for different ageing times. The Tg of the pDCPD samples increased slightly after two
months of ageing, then remained constant during additional ageing. At the same time, the
peak height and width of the tan(δ) curve decreased monotonically with increasing
ageing, indicating diminishing chain mobility. The slight increase in Tg within the first
two months can be attributed to thermal effects and/or surface oxidation. After this initial
stage, chain mobility of pDCPD decreased slowly, which is typical of physical ageing.
The small amount of absorbed water caused no observable plasticization of pDCPD.

29

Fig. 2.2 tan(δ) curve evolution during ageing from DMA

For epoxy composites (Fig. 2.2b), ageing for two months caused the Tg to decrease
significantly, from 101 ° C to 84 ° C, and the peak of the tan(δ) curve increased and
broadened. These are marked features of plasticization - absorbed water molecules
existed in a free state and acted as a lubricant, facilitating sliding between polymer
chains. Continued ageing for the next ten months caused Tg to increase slightly, albeit to
a much lower value than the pre-aged Tg. This slow increase in Tg can be partly attributed
to physical ageing. Although the increase in peak height and width is not consistent with
physical ageing, the trend of Tg first decreasing and then increasing during ageing has
been reported in multiple studies of epoxy hygrothermal ageing [19-23]. Some
investigators have pointed out that this phenomenon can be caused by water molecules
that strongly bond to polymer chains during long-term ageing [19, 20]. Fig. 2.2b also
shows a small second peak at ~120 ° C (arrow), which might result from this mechanism.
Matrix plasticization resulting from ageing is also manifest by changes in fracture
toughness. Plasticization decreases the matrix strength while simultaneously enhancing
30

fracture toughness [7]. The enhanced fracture toughness can be beneficial to fatigue
performance, because matrix cracking plays an important role in the damage
accumulation process leading to fatigue failure [16]. Fig. 2.3 shows the evolution of the
K IC of neat pDCPD during ageing. Note that the four groups of samples were aged and
tested at the same time points - the plotted data points are artificially offset only to show
symbols clearly. Similarly, an error bar is shown for only one data point (with the
maximum error) for clarity. The K IC of pDCPD decreased slightly (~10%) after ageing
for one year, and the embrittlement was attributed to thermal ageing. The KIC of epoxy,
on the other hand, evolved in a markedly different fashion. The K IC of the un-aged epoxy
was relatively low (about 30% of pDCPD), but increased significantly within the first
three months, and was unchanged between the third and sixth months. After six months
of ageing, the evolution of K IC was consistent with the evolution of water absorption (Fig.
2.1a) and Tg (Fig. 2.2b), indicating that the absorbed water caused matrix plasticization.
However, after one year of ageing, the K IC increased, despite the fact that saturation had
been reached by six months of ageing. The observed increase in K IC occurred
simultaneously with the appearance of the (second) small peak in the tan(δ) curve (Fig.
2.2b), and was tentatively attributed to strongly bonded water molecules. After one year
of ageing, the K IC of the two resin systems was comparable.

31

Fig. 2.3 Fracture toughness of neat cured resin

Summarizing this section, plasticization is the controlling mechanism in
hygrothermal ageing of epoxy, while for pDCPD, plasticization is negligible, and thermal
ageing induces moderate strengthening. These phenomena lead to significantly different
influences on mechanical properties, as described below.

2.4.2 Effect of moisture uptake on fatigue properties
Fig. 2.4 shows the S-N curves of the two composites after ageing in salt water. S
indicates maximum cyclic stress while N represents cycles to failure. Data points with
arrows at N = 10
6
cycles indicate run-out samples. The load levels used are 40%-80% of
the ultimate tensile strength (UTS) for the given ageing condition. (Static tensile
properties will be discussed in section 3.2). For both loading directions, the pDCPD
32

composites showed greater fatigue strength, and the slope of the S-N curve decreased
with ageing time, indicating a greater influence of ageing on low-cycle fatigue than on
high-cycle fatigue. Standard deviations of aged samples were generally greater than that
of pre-aged, indicating a greater scatter in mechanical properties from sample to sample
after ageing.

Fig. 2.4 S-N curves evolution after ageing

Fig. 2.4a shows the 0° tension-tension fatigue property evolution. In the pre-aged
condition, pDCPD and epoxy composites displayed similar fatigue properties. However,
after one month of ageing, for pDCPD composites, the high-cycle fatigue life increased,
a phenomenon attributed to post-curing and/or physical ageing of the matrix. In contrast,
for epoxy composites, the low-cycle fatigue life decreased, attributed to fiber strength
decrease, matrix plasticization, and interface degradation. After three months of ageing,
both composites showed fatigue life decrease at all load levels, although pDCPD
composites showed superior strength retention relative to epoxy composites.
33

Fig. 2.4b shows the evolution of fatigue behavior for 90° tension-tension. For the
pre-aged condition, pDCPD composites showed greater fatigue strength than epoxy
composites, indicating superior interface adhesion and/or matrix strength, since fiber
strength has negligible influence on 90° failure. Because the static tensile strength of
pDCPD and epoxy are similar (from dogbone tensile tests per ASTM D638), the superior
fatigue behavior of pDCPD composites is attributed primarily to the superior interface
adhesion. The superior interface adhesion is partly caused by the lower viscosity of the
DCPD resin during manufacture, which facilitates infiltration and promotes fiber-matrix
contact. The trend in S-N curves for 90° direction followed the same trend as 0° fatigue,
suggesting similar damage mechanisms were involved.
The superior post-ageing adhesion of interfaces in pDCPD composites compared to
epoxy composites also is apparent from SEM fractography of 90° fatigue samples. Fig.
2.5 shows the fracture surface of 90° fatigue samples at different ageing conditions. Note
that the fracture morphologies shown in Fig. 2.5 were consistent over the entire fracture
surface and thus are representative for the corresponding ageing condition. Furthermore,
the fractography was consistent for all load levels for each ageing condition. The epoxy
composites (Fig. 2.5b, d, f) showed identical fractography throughout six months of
ageing. Bare fiber surfaces were common, while matrix failure was rare, indicating
relatively weak interface adhesion to the matrix cohesive strength.

34

Fig. 2.5 Fracture surface of [90° ]4 fatigue samples

In contrast, the fracture surfaces of pDCPD composites (Fig. 2.5a, c, e) changed with
ageing time. Pre-aged samples (Fig. 2.5a) and samples aged one month (Fig. 2.5c)
showed a rough surface with matrix shear ridges and an absence of exposed fiber surface,
35

indicating matrix-dominated failure and thus greater interface strength relative to matrix
cohesive strength. After three months of ageing (Fig. 2.5e), both matrix shear ridges and
exposed fiber surface were observed, indicating a mix of interface and matrix failure.
Given that the epoxy matrix absorbed more water (leading to greater plasticization and
strength decrease), we conclude that the interface strength in pDCPD composites was
greater than in epoxy composites throughout the ageing period studied.
The superior interface adhesion in pDCPD composites relative to epoxy composites
prior to ageing arises from multiple factors. The different chemistry of the two resins
leads to different fiber compatibility and thus different interface bond strengths. Also, the
curing cycles of the two resins differ (Section 2.1), and leading to differences in cure
shrinkage, coefficient of thermal expansion and local temperature gradient during
cooling. Thus, the residual stress in the two composites also differs [29].
Figs. 2.4 and 2.5 indicate the important role of the fiber-matrix interface in the UD
composite fatigue performance for both 0° and 90° composites. The interface degradation
is affected by the environmental moisture level, which is determined by the amount of
water absorption in the matrix. Comparing Figs. 2.1 and 2.4, the amount of water uptake
correlated with the fatigue strength decrease, which is consistent with the scenario
described above. Thus, matrix hydrophobicity is critical to preserving the fatigue strength
of PMCs in humid environment application.

2.4.3 Asynchronous ageing of fiber, matrix and interface
The fiber-matrix interface characteristics significantly influence the overall
composite behavior. As shown in Fig. 2.6, a plot of the short-beam strength (SBS)
36

determined by short-beam shear test as a function of ageing time. (Again, only the
maximum error bar is shown for clarity.) The SBS depends mutually on the matrix shear
strength and the interface shear strength, while the dependence on fiber strength is
negligible because fiber breakage generally does not occur in interlaminar shear failure.
The pDCPD composite exhibited a slow and steady decrease in SBS with increasing
ageing time, and the integrity of the fiber interface was well-preserved. The retention of
SBS for pDCPD composites after one year of ageing is 85%, and this trend is consistent
with the hydrophobicity of pDCPD matrix, as discussed previously. In contrast, epoxy
composites showed a significant and continuous decrease in SBS during ageing for one
year, and the SBS after one year of ageing decreased to 40% of the pre-aged value. In
contrast, plasticization of the epoxy matrix saturated after three months, as shown
previously in Figs. 2.1a and 2.3. Thus, the continuous decrease of SBS for epoxy sample
after three months (Fig. 2.6) is attributed directly to the continuous decrease in interface
strength resulting from ageing.

Fig. 2.6 Short-beam strength evolution during ageing
37

Fig. 2.7 shows the static tensile modulus and the UTS evolution during ageing. Note
that the four groups of samples were aged and tested at the same time (data points are
artificially offset to show the error bars). The effect of ageing on modulus is negligible
for 0° composites, while in the 90° orientation, the modulus of pDCPD composite
increased slightly while the modulus of epoxy composite decreased after three months of
ageing. The strength in the 0° direction of the two composites are identical (except for
epoxy aged in DI water, which is discussed in section 3.3 below), and showed significant
decrease after ageing (40% retention after one year). In the 90° direction, the pDCPD
composites showed greater strength retention (70% retention after one year) than epoxy
composites (45% retention after one year).

38

Fig. 2.7 Static tensile properties evolution after ageing

The relationship for composite modulus and constituents moduli follows a simple
rule of mixtures (Eqn. 2.1),
 
 
,0 ,0
,90 ,90
1
1
1
c f m
c f m
E fE f E
f
f
E E E
   






(2.1)
where Ec,0 is the 0° modulus of the composite, Ef,0 is the modulus of the fiber in
longitudinal direction, Em is the modulus of the matrix, and f is the fiber volume fraction.
39

Assuming Ef,0 = 82.7 GPa, Em = 3 GPa and f = 58%, the glass fibers account for more
than 97% of the composite modulus. Eqn. 2.1 also applies to strength, and thus the static
tensile strength of 0° composites is determined by the glass fiber strength. Thus, from the
0° modulus and strength plots in Fig. 2.7, hygrothermal exposure severely degraded the
glass fiber strength, while the glass fiber modulus was retained. The identical strength
values of 0° pDCPD and epoxy composites indicate synchronous degradation of glass
fiber in the two composites.
In 90° composites, the modulus is largely influenced by the matrix. Thus, the slight
increase in modulus of pDCPD composites was attributed to physical ageing of the
matrix, and the decrease in modulus of epoxy composites was caused by the
plasticization of the matrix. Strength, on the other hand, is determined by the strength of
both the matrix and the interface in the 90° direction. Thus, the greater strength retention
of 90° pDCPD composites is attributed to the greater strength retention of the matrix and
interface. Note that the degradation of matrix and interface in the two composites is
asynchronous during ageing.
In 0° composites, the identical static strength degradation of pDCPD and epoxy
composites (Fig. 2.7) is explained by the synchronous degradation of glass fiber. On the
other hand, the much different fatigue behavior (Fig. 2.4a) is explained by the
asynchronous degradation of the matrix and interface, as fatigue failure is largely
influenced by the progressive damage accumulation of the matrix and interface before
final rupture. Indeed, the AE signals recorded during fatigue testing support the
contention of asynchronous matrix and interface degradation, as described next.
40

Fig. 2.8 shows the evolution of sample stiffness and AE signal during fatigue. Fig.
2.9 shows the evolution of sample stiffness and accumulated released AE energy during
fatigue. Sample type and ageing time are marked on each figure. All four composites
were loaded at S = 450 MPa. In each figure, the curve illustrates the evolution of the
secant modulus given by Eqn. 2.2, which provides an indication of sample stiffness:
min
max min
s
S
E





(2.2)
where S and σmin are the maxim um and minimum cyclic stress, while εmax and εmin are the
corresponding maximum and minimum cyclic strain. Strain values are calculated from
fixture displacements, although the relative changes are considered significant. Each
colored dot represents a damage event in the sample captured by acoustic emission, with
different colors indicating different levels of energy released in this event. The right
ordinate of each figure displays the location of AE events in the sample gauge length.
Thus the bottom sensor sits at Position = 0, while the top sensor is at Position = 100 mm.
High-energy signals reportedly represent fiber breakage, while low-energy signals
indicate matrix and interface failure [13, 30, 31]. There is no distinct boundary between
the energy ranges of each damage type, and they overlap appreciably. Here we assume
the red dots signify fiber breakage, while the green and blue dots arise from matrix and/or
interface failure.

41

Fig. 2.8 Damage evolution of [0° ]2 composites

42

Fig. 2.9 Accumulated AE energy evolution of [0° ]2 composites

Comparing the damage evolution process in the two composites as in Fig. 2.8 and
Fig. 2.9, less damage accumulation and fewer low-energy events were observed in epoxy
composites than in pDCPD composites. Furthermore, the damage in epoxy composites
consisted primarily of fiber breakage, while the damage in pDCPD composites included
all three types of fiber, matrix and interface failure. The absence of interface failure in
aged epoxy samples during fatigue arises because massive interface debonding already
occurred prior to fatigue testing. The more extensive interface degradation in epoxy
43

composites is also evidenced by the lower static and fatigue strength values of 90° epoxy
composites (Fig. 2.4 and Fig. 2.7), and by the fatigue fractography of 90° epoxy
composites (Fig. 2.5).
Summarizing the results and discussion above, the fiber and interface showed coupled
but asynchronous degradation, which affected the overall mechanical behavior. The
degradation of glass fiber was determined primarily by ageing time, while the interface
degradation was affected primarily by the moisture level in the matrix. Greater water
absorption by matrix accelerated interface debonding, and debonded interfaces became
capillary diffusion pathways and accelerated water diffusion into the matrix by increasing
the contact area.

2.4.3 DI water vs. salt water ageing
The difference in weight change resulting from ageing in DI water versus salt water
is shown in Fig. 2.1. The pDCPD composites showed slightly greater water uptake after
salt water immersion compared to DI water immersion. This is atypical for polymers and
awaits further investigation. In contrast, epoxy composites showed greater water uptake
after DI water immersion than in salt water, a phenomenon that was caused in part by the
osmotic effect in salt water [2]. However, the large difference in weight gain (>100%) of
epoxy composites in the two environments cannot be fully attributed to these phenomena,
but instead is attributed to differences in the extent of interface debonding. Extensive
interface debonding occurred during ageing in DI water and the debonded interfaces
retained significant amounts of water.
44

A comparison between 0° S-N fatigue curves in DI water and salt water ageing is
shown in Fig. 2.10. Fig. 2.10a shows that the fatigue strength of pDCPD composites
increased after one month and decreased after three months. The S-N curves in DI water
and salt water were identical after three months, which is consistent with the similar
amount of water uptake in the two environments (Fig. 2.1). Fig. 2.10b shows that for
epoxy composites, in salt water ageing, the fatigue strength decreased gradually with
ageing time. In contrast, in DI water ageing, the fatigue strength decreased significantly
after one month, but increased slightly after three months, which could be attributed to
physical ageing of the matrix. The slopes of the three lowermost S-N curves are identical,
indicating damage saturation. Thus, for both pDCPD and epoxy composites, the
evolution of fatigue strength in the two environments was closely related to the amount of
water absorption. Epoxy composites showed significantly greater water absorption in DI
water than in salt water, and thus greater decrease in fatigue strength, while pDCPD
composites showed similar water absorption in the two environments, and thus similar
decreases in fatigue strength. In summary, the salt water environment influenced only the
amount of water uptake - no other effects were observed on the studied composites.

45

Fig. 2.10 S-N curves evolution in DI water and salt water environments

2.4.4 Fatigue damage localization
Damage localization was noticed within the sample gauge length during tension-
tension fatigue of aged composites. Comparing the damage evolution between samples
aged one month (Figs. 2.8a and 2.8b) and three months (Figs. 2.8c and 2.8d), damage was
dispersed over the entire gauge length for the former, while damage was localized in the
latter, where most damage events were clustered at one or two locations. Damage
localization is associated with interface degradation and loss of capacity for fiber-matrix
load transfer. Once damage initiates at a location, stress cannot be transferred to adjacent
areas, and the stress concentration intensifies. Thus, damage accumulates locally, leading
to premature failure.
Further evidence of damage localization is shown in sample images after fatigue
failure (Fig. 2.11). Ageing conditions for both composites from left to right are: pre-aged,
one month aged, and three months aged. Damage localization can be observed as ageing
46

time increases, especially for pDCPD composites. The color change of pDCPD samples
arises from a surface layer (tens of microns thick) caused by oxidation. The oxidized
layer is dense and hard, limiting progressive oxidation into the bulk. Epoxy samples
changed from semitransparent to internally reflective after ageing. However, epoxy neat
resin samples did not show such change after ageing. Thus, the internal reflection is
caused by the reflection of light from the debonded fiber-matrix interfaces.

Fig. 2.11 Fatigue failure of [0° ]2 composites

2.5 Conclusions
The effects of hygrothermal ageing on the static tensile and tension-tension fatigue
behavior of glass/pDCPD UD composite were investigated using acoustic emission
monitoring. A glass/epoxy composite was used as a baseline reference material, and
properties of both 0° and 90° composites were measured. The pDCPD composites
showed less water uptake than the epoxy composites because of the intrinsic
hydrophobicity of pDCPD matrix. The degradation of matrix, fiber and interface
(especially the latter two) was significantly affected by moisture uptake level. Thus,
47

pDCPD composites showed superior mechanical performance under dynamic and quasi-
static loading in both 0° and 90° directions. Fractography analysis revealed that pDCPD
also had greater interface bond strength to glass fibers than the epoxy, contributing to the
high strength retention of pDCPD composites.
We have investigated the effects of hygrothermal ageing on a ruthenium-catalyzed
pDCPD neat polymer and glass fiber composite, using a traditional epoxy resin system as
a benchmark material for comparison. The pDCPD matrix showed strengthening by
thermal ageing and resistance to water absorption and thus maintained much of the
original fiber-matrix interface strength. Consequently, pDCPD composites showed
superior fatigue strength retention after long-term ageing, especially in the high-cycle
fatigue regime. In contrast, epoxy absorbed more water during ageing and was
significantly plasticized. The high moisture level in the epoxy matrix induced severe
fiber-matrix interface degradation. Combined, these factors resulted in significant
deterioration of the fatigue performance of the epoxy composite. For pDCPD composites,
ageing in salt water showed no different effects from ageing in DI water, while salt water
ageing of epoxy composites showed less water absorption compared to ageing in DI
water.
Hygrothermal ageing of PMCs is a major concern for structural durability in humid
service environments. The present study demonstrates that pDCPD composites exhibit
exceptional resistance to hygrothermal environments and as a result, superior retention of
static and dynamic mechanical properties. These characteristics are well-suited to
potential applications in wet and corrosive environments, such as wind energy structures,
offshore oil platforms, and both land and marine vehicles. The intrinsic hot/wet ageing
48

resistance of pDCPD composites indicates potential for high-temperature applications
such as the protective shells of offshore oil risers. The current findings highlight the
importance of matrix hydrophobicity and interface adhesion to hygrothermal ageing
resistance. They can be useful to future investigations of ageing behavior where
mechanical loading, moisture, oxidation and radiation occur simultaneously. However,
certain issues warrant further investigation, particularly the observation that pDCPD
showed greater moisture uptake in salt water, an uncommon phenomenon for polymers.
In addition, direct measurements of interface strength as a function of ageing could result
in more accurate predictions of long-term mechanical behavior, and the insights gained
by such measurements could lead to the design of more durable interfaces.

49

Chapter 2 Appendix
Both pDCPD and epoxy composites were reinforced with glass fibers with PET
sizing. The UD composites were first cut and polished in 0° directions, then examined
microscopically. Fig. 2.12 shows the matrix-rich region between three adjacent fiber tows
in a pDCPD composite. Unidentified inclusions (marked with arrow) appear in light
contrast in the matrix-rich region. These inclusions showed similar morphology in 90°
cross-section (not shown here), and were thus equi-axed.

Fig. 2.12 Optical microscope of 0° polished section

Further investigation by micro-Raman (Renishaw in Via Raman Microscope)
determined that these inclusions originated from the PET fiber sizing and developed
50

during processing. Fig. 2.13 shows the Raman spectrum of an inclusion. The laser
wavelength was 532nm (green), the laser power was 10% of 0.2W, the exposure time
was 10 seconds, and the scan range was 100cm
-1
to 3200 cm
-1
. Spectra from pDCPD and
epoxy composites were identical.

Fig. 2.13 Raman spectrum of PET inclusion

In Fig. 2.13, the wavenumber and relative amplitude of each peak are identical to the
PET reference spectrum [32], confirming the identity of the inclusions. Because PET is
relatively insoluble in pDCPD and epoxy resins, these PET inclusions originate from the
PET fiber sizing on glass fibers. Thus, because the sizing disappeared from fibers, the
glass fibers and matrix in both composites were in direct contact.
51

Chapter 2 References
1. Gellert EP, Turley DM, Seawater immersion ageing of glass-fibre reinforced polymer
laminates for marine applications, Composites A 1999;30:1259-1265.
2. Mouzakis DE, Zoga H, Galiotis C, Accelerated environmental ageing study of
polyester/glass fiber reinforced composites (GFRPCs), Composites B 2008;39:467-
475.
3. Foulc MP, Bergeret A, Ferry L, Ienny P, Crespy A, Study of hygrothermal ageing of
glass fibre reinforced PET composites, Polymer Degradation and Stability
2005;89:461-470.
4. Arun KV, Basavarajappa S, Sherigara BS, Damage characterisation of glass/textile
fabric polymer hybrid composites in sea water environment, Materials and Design
2010;31:930-939.
5. Assarar M, Scida D, Mahi AE, Poilane C, Ayad R, Influence of water ageing on
mechanical properties and damage events of two reinforced composite materials: Flax-
fibres and glass-fibres, Materials and Design 2011;32:788-795.
6. Arif MF, Meraghni F, Chemisky Y, Despringre N, Robert G. In situ damage
mechanisms investigation of pa66/gf30 composite: effect of relative humidity,
Composites B 2014;58:487-495.
7. Park Y, Kweon J, Choi J, Failure characteristics of carbon/BMI-Nomex sandwich
joints in various hygrothermal conditions, Composites B 2014;60:213-221.
8. Chateauminois A, Chabert B, Soulier JP, Vincent L, Hygrothermal ageing effects on
the static fatigue of glass/epoxy composites, Composites 1993;24:547-555.
9. Pauchard V, Chateauminois A, Grosjean F, Odru P, In situ analysis of delayed fibre
52

failure within water-aged GFRP under static fatigue conditions, International Journal
of Fatigue 2002;24:447-454.
10. Kellas S, Morton J, Curtis PT, The effect of hygrothermal environments upon the
tensile and compressive strengths of notched CFRP laminates, Composites 1990;42:41-
62.
11. Vauthier E, Abry JC, Bailliez T, Chateauminois A, Interactions between hygrothermal
ageing and fatigue damage in unidirectional glass/epoxy composites, Composites
Science and Technology 1998;58:687-692.
12. Liao K, Schultheisz CR, Hunston DL, Long-term environmental fatigue of pultruded
glass-fiber-reinforced composites under flexural loading, International Journal of
Fatigue 1999;21:485-495.
13. Kotsikos G, Evans JT, Gibson AG, Hale JM, Environmentally enhanced fatigue
damage in glass fibre reinforced composites characterised by acoustic emission,
Composites A 2000;31:969-977.
14. McBagonluri F, Garcia K, Hayes M, Verghese KNE, Lesko JJ. Characterization of
fatigue and combined environment on durability performance of glass/vinyl ester
composite for infrastructure applications, International Journal of Fatigue 2000;22:53-
64.
15. Patel SR, Case SW, Durability of hygrothermally aged graphite/epoxy woven
composite under combined hygrothermal conditions, International Journal of Fatigue
2002;24:1295-1301.
16. Meziere Y, Bunsell AR, Favry Y, Teissedre JC, Do AT, Large strain cyclic fatigue
testing of unidirectional carbon fibre reinforced epoxy resin, Composites A
53

2005;36:1627-1636.
17. Poodts E, Minak G, Zucchelli A, Impact of sea-water on the quasi static and fatigue
flexural properties of GFRP, Composite Structures 2013;97:222-230.
18. Schmitz GK, Metcalfe AG, Stress Corrosion of E-Glass Fibers, Industrial and
Engineering Chemistry Product Research and Development 1966;5:1-8.
19. DeNeve B, Shanahan MER, Water absorption by an epoxy resin and its effect on the
mechanical properties and infra-red spectra, Polymer 1993;34:5099-5105.
20. Alessi S, Conduruta D, Pitarresi G, Dispenza C, Spadaro G, Accelerated ageing due to
moisture absorption of thermally cured epoxy resin/polyethersulphone blends. Thermal,
mechanical and morphological behavior, Polymer Degradation and Stability
2011;96:642-648.
21. Zhou J, Lucas JP, Hygrothermal effects of epoxy resin, Polymer 1999;40:5505-5522.
22. Muhlebach A, Schaaf PA, Hafner A, Setiabudi F, Thermal stability and degradation of
hydrocarbon metathesis polymers, Journal of Molecular Catalysis A: Chemical
1998;132:181-188.
23. Dimonie D, Dimonie M, Munteanu V, Iovu H, Couve J, Abadie MJ. Nature of the first
exothermic peak on the DSC curve of linear polydicyclopentadiene obtained with a
special catalytic system, Polymer Degradation and Stability 2000;70:319-324.
24. Barnes SE, Brown EC, Corrigan N, Coates PD, Harkin-Jones E, Edwards HGM,
Raman spectroscopic studies of the cure of dicyclopentadiene (DCPD), Spectrochim
Acta A 2005;61:2946-2952.
25. Le Gac PY, Choqueuse D, Paris M, Recher G, Zimmer C, Melot D, Durability of
polydicyclopentadiene under high temperature, high pressure and seawater (offshore
54

oil production conditions), Polymer Degradation and Stability 2013;98:809-817.
26. Samborsky DD, Agastra P, Mandell JF, Fatigue trends for wind blade infusion resins
and fabrics, AIAA SDM 2010 Wind Energy Session, Paper 2820.
27. Agastra P, Mandell JF, Testing and simulation of damage growth at ply drops in wind
turbine blade laminates, SAMPE 2010, Paper 398.
28. Mivehchi H, Varvani-Farahani A, The effect of temperature on fatigue strength and
cumulative fatigue damage of FRP composites, Procedia Engineering 2, 2010:2011-
2020.
29. Kravchenko OG, Li C, Strachan A, Kravchenko SG, Pipes RB, Prediction of the
chemical and thermal shrinkage in a thermoset polymer, Composites A 2014;66:35-43.
30. Kar NK, Hu Y, Barjasteh E, Nutt SR, Tension-tension fatigue of hybrid composite rods,
Composites B 2012;43:2115-2124.
31. Kempf M, Skrabala O, Altstadt V, Acoustic emission analysis for characterisation of
damage mechanisms in fibre reinforced thermosetting polyurethane and epoxy,
Composites B 2014;56:477-483.
32. Kuptsov AH, Zhizhin GN, Handbook of Fourier transform Raman and infrared spectra
of polymers, Elsevier 1998, ISBN 0444826203, page 155.

55

CHAPTER 3. THERMAL OXIDATION AGEING OF
POLYDICYCLOPENTADIENE COMPOSITES
3.1 Abstract
The thermal ageing and oxidation behavior of pDCPD is investigated and their effects on
thermal and mechanical properties are studied. Both ageing temperatures below (100 ° C)
and above (150 ° C) glass transition temperature (Tg) are used and both air and nitrogen
environments are applied. Thermal properties are studied by dynamic mechanical
analysis (DMA) while mechanical properties are studied by static tensile test and short-
beam shear test. Surface aged layer is studied using optical microscope. Results show that
a thin surface oxidized layer is formed on air aged samples. This oxidation layer is brittle
and causes the decrease in strength and failure strain of neat resin sample. Fiber-matrix
interface degradation is significant under 150 ° C ageing environment while interface is
well maintained at 100 ° C, evidenced by the analysis on tensile and shear failure.
Mechanical properties of pDCPD resin are stable throughout 25 weeks of ageing at both
ageing temperatures with and without oxygen, indicating the superior thermal ageing
resistance of this new resin.

3.2 Introduction
The thermal ageing and oxidation behavior of polymer composites is the major issue
of concern for application in elevated temperature environments, especially when
exposed to oxygen [1-10]. Polydicyclopentadiene (pDCPD) as a new resin for these
potential usage has shown corrosion resistance in hygrothermal environments, but its
response in thermal oxidizing environment has not been well understood [11]. Chemical
56

reactions of polymers during thermal-oxidative ageing mainly contains post-curing,
secondary polymerization, carbonyl growth, chain scission, and volatile material release
[12]. So far as reported, oxidation and diffusion affect mechanical behavior, but not vice
versa. Tiny amount of oxidation can significantly change mechanical properties due to
chain scission.
Some researchers pointed out that there are two competing mechanisms in thermal
oxidation [13]. One is the cross-linking reaction or secondary polymerization caused by
the high temperature. The additional cross-linking causes molecular weight increase,
strengthening and Tg increase. This mechanism occurs throughout the thickness of the
sample because it is a thermal effect. The competing mechanism is the chain-scission
reaction caused by oxidation. Chain-scission causes molecular weight decrease, chain
mobility increase and Tg decrease.
However, for many polymers, oxidation is only limited within a thin surface layer
because the diffusion limited oxidation (DLO) effect. The DLO effect is caused by the
greater oxygen consumption rate than the oxygen diffusion rate of the polymer. Thus, the
thickness of the oxidation layer is determined by the ratio of oxidation reaction rate to
oxygen diffusion rate. Many polymers show low O2 permeability thus thin oxidized layer
(10-100 µ m). This thin oxidized layer prevents further oxygen diffusion into the material
and acts as a protection shell if the oxidized material is condense and solid. However,
mode I crack may occur in the oxidized layer because of its greater shrinkage than bulk
material.
Oxidation behavior of composite is quite different from neat resin and is much more
complex. According to [14-18], because of matrix shrinkage after oxidation and the
57

thermal expansion coefficient difference between fiber and matrix, stress concentration
occurs at fiber-matrix interface, and thus facilitates interface debonding. The matrix-rich
region shows the greater stress concentration because of the greater shrinkage. Matrix
shrinkage also increases with conditioning time and oxygen pressure.
For unidirectional (UD) composites, the oxidation behavior differs in the transverse
direction and in the longitudinal direction [18]. In the transverse direction, if the fiber
volume fraction is low enough that the fibers are not in contact with each other, the
oxidation behavior is similar to neat resin and oxidation only occurs in a thin surface
layer. However, the diffusion rate is slower than neat resin because fibers hinder oxygen
diffusion [19, 20]. However, if the fiber volume fraction is high and the fibers are in
contact with each other, the fiber-matrix interface can debond because of matrix
shrinkage and become oxygen diffusion paths because the interfaces are connected as a
network. In the longitudinal direction, if no interface debonding occurs, the diffusion
behavior is similar to neat resin because the diffusion path is not blocked by fibers.
However, when interface debonding occurs and the debonded interfaces become the main
diffusion paths, the process can be fast. In this case, UD composites can show
significantly different ageing behaviors when the fiber ends are sealed or not.
In this study, we investigated the thermal ageing behavior of pDCPD polymer and its
glass fiber composites at different temperatures and in different environments. The
thermal and mechanical properties were monitored by dynamic mechanical analysis,
static tensile test, short-beam shear test and optical microscope inspection. Results show
that this new formulation of pDCPD resin is resistant to thermal ageing below or above
its glass transition temperature, but the fiber-matrix interface of its glass fiber composites
58

is damaged at high-temperature ageing. Oxidation is limited within surface layer, which
is a typical diffusion limited oxidation behavior.

3.3 Experiments
3.3.1 Sample preparation
Both pDCPD (Proxima, Materia, Inc.) neat resin and glass fiber composites are
prepared to study the thermal ageing effects on them. The neat resin panels are cured at
30 ° C for 2 hours, and then post-cured at 100 ° C for 30 minutes. Two dimensions of
panels are prepared: 8 mm thick panels are used for surface aged layer study, while 4 mm
thick panels are later cut into dogbone-shape tensile test specimens (ASTM D638 type I).
Unidirectional (UD) glass fiber laminates are produced using common vacuum infusion
processing technique (Materia, Inc.). Non-crimp fabric (E-LT 3500, Vectorply, Corp.)
comprised of E-glass (94 wt% PPG Hybon
®
2026 in the warp direction and 6 wt% PPG
Hybon
®
2002 in the weft direction) are used. Fiber volume fraction is ~58% from burn-
out method. Two dimensions of composite laminates are prepared: 2-ply (1.6 mm thick)
laminates are used for 0° testing while 4-ply (3.2 mm thick) laminates are used for 90°
testing and shear testing. Table 3.1 and Table 3.2 show the basic properties of cured
pDCPD and the glass fiber, respectively.

Table 3.1 Properties of cured pDCPD resin
Tg Density Tensile modulus Ultimate tensile strength Tensile elongation
(° C) (kg/m
3
) (GPa) (MPa)
124 1050 3.1 73 2.7%

59

Table 3.2 Properties of glass fiber
Modulus (GPa) Tensile strength (MPa) Average fiber diameter (μm)
82.7 2790 17

3.3.2 Ageing conditions
Three different ageing conditions are used: 100 ° C in nitrogen, 100 ° C in air, and
150 ° C in air. These two temperatures are selected to study the different ageing effects
below and above Tg (the Tg of un-aged pDCPD is ~130 ° C). Nitrogen ageing environment
is used to compare the ageing effect with and without oxygen. Samples aged in air are
placed in air-circulated temperature-controlled ovens, while samples aged in nitrogen are
sealed in glass containers and these containers are placed in temperature-controlled
ovens. Weight change of both neat resin and composite samples are monitored using
specific groups of 8 mm neat resin blocks and 4-ply laminate blocks. Total studied ageing
period is 25 weeks, while a group of samples are taken out from ovens and got thermal
and mechanical properties tested at time points of 0, 1, 4, 9, 16 and 25 weeks. These time
points are selected because diffusion process is assumed Fickian and properties are
expected to show linear relationship with square root of time.

3.3.3 Dynamic mechanical analysis
Dynamic mechanical analysis (DMA) (Q800, TA Instruments) is conducted on neat
resin samples to study the thermal properties evolution. Single cantilever beam sample is
cut from center of 8 mm thick resin block, thus surface aged layer’s influence is
neglected. The dimension of single cantilever beam is (35× 12× 3) mm
(length×width×thickness). According to ASTM D7028, during the test, temperature is
60

ramped from 50 ° C to 180 ° C at a rate of 5 ° C/min. Displacement control is used with the
maximum strain of 0.1% and a loading frequency of 1 Hz. Time, temperature, storage
modulus, loss modulus and tan(δ) are recorded during the test.

3.3.4 Optical microscope
Color change of the surface aged layer is observed using optical microscope
(Keyence VHS). Optical microscope is better option than scanning electron microscope
(SEM) for this observation because contrast in color cannot be observed under SEM. The
8 mm thick neat resin block is first cut into (10× 10× 8) mm sample, and then the sample
in mounted and polished in the cross-section direction before observation.

3.3.5 Static tensile test
Static tensile test is conducted on both neat resin and composite samples. Test
standards used are ASTM D638 for neat resin and ASTM D3039 for composite. Loading
rate for neat resin samples is 5 mm/min while for composite samples is 2 mm/min.
Tensile strain data is collected using an extensometer (Instron 2630-109) until sample
rupture. Load, time, displacement and strain are recorded during test. Tests of neat resin
samples and 90° composite samples are conducted on a tabletop load frame (Instron
5567), while tests of 0° composite samples are conducted on a floorstanding load frame
(Instron 5585H).

61

3.3.6 Short-beam shear test
Short-beam shear (SBS) test is a test method to measure interlaminar shear strength
of composite laminate, because samples generally fail in the mode of interlaminar shear
in this test. According to ASTM D2344, 4-ply composite sample is used for this test and
the dimension of SBS sample is (19.2× 6.4× 3.2) mm (length× width× thickness). Fiber
direction is along the length direction. Sample is loaded in a three-point bending fixture
until first load drop occurred. The span length is 12.8 mm. Loading rate is 1 mm/min.
Load, time and displacement are recorded during test. The load at the first load drop
during test is used to calculate the short-beam strength (SBS).

3.4 Results and discussion
3.4.1 Weight change during thermal ageing
Fig. 3.1 shows the weight change of neat resin and composite samples during
thermal ageing. Fig. 3.1 shows that for 100 ° C air aged samples, the weight
monotonically increased during 25 weeks of ageing. This is because of the surface
oxidation - addition of oxygen element in the oxidized layer added weight to the sample.
Samples aged in 100 ° C nitrogen showed monotonic weight decrease during 25 weeks of
ageing. Because oxidation cannot occur in nitrogen environment, this phenomenon is a
result of volatile molecules diffusion out from samples. For 150 ° C air ageing condition,
the situation is complex: the weight change is a combination of increase and decrease.
Reasonable explanation is that both oxidation and volatile diffusion are significant at this
temperature and they compete with each other. Oxidation effect is fast and short-term
while volatile diffusion is more significant at longer ageing time.
62

Comparing data of neat resin and composite, composite samples show greater weight
gain, or in other words, less weight loss. This is easy to understand: glass fiber is stable to
high temperature and change only occurs in matrix. Because composite only has 40%
volume fraction of matrix while neat resin has 100%, composite has less volatile
molecules per unit volume/weight. In contrast, oxidation is a surface effect and has no
difference between neat resin and composite. Therefore, the combined effect is composite
samples have less molecules diffusion out thus less weight loss compared to neat resin.

Fig. 3.1 Weight change of samples

3.4.2 Dynamic mechanical analysis
Fig. 3.2 shows the loss modulus curves from three ageing conditions after 25 weeks.
The Tg of un-aged Proxima resin from loss modulus peak is also marked on the figure
(~130 ° C). Fig. 3.2 shows that after 9 weeks of ageing, sample aged in 100 ° C air shows
slight decrease in Tg, while sample aged in 100 ° C nitrogen shows greater decrease in Tg.
63

These decreases can be interpreted as a plasticization effect. Small molecules from chain-
scission reaction during thermal ageing act as plasticizers can cause decrease in glass
transition temperature. Sample aged in air shows higher Tg than nitrogen aged, because
its oxidation layer has a higher Tg than bulk resin. Oxidized layer is dark and is more
densely packed, thus the molecular mobility is lower. Different from the 100 ° C aged
samples, sample aged in 150 ° C air shows significant increase in Tg. Secondary
polymerization occurs at high temperature and leads to increase in Tg.
After 25 weeks of ageing, samples aged in all three environments showed increase in
Tg, which is caused by secondary polymerization after long-term thermal ageing. Among
the three environments, 100 ° C nitrogen environment has the lowest Tg compared to
100 ° C air and 150 ° C air, because oxidized layer has higher Tg than un-oxidized
material. And higher ageing temperature in air lead to more severe oxidation and thicker
oxidation layer, causing higher overall glass transition temperature of the whole sample.

Fig. 3.2 DMA curves of neat resin
64

3.4.3 Surface aged layer
Fig. 3.3 and Fig. 3.4 show the surface color change during ageing for different
ageing conditions. Fig. 3.3 shows that for 100 ° C air aged sample, there is a distinctive
dark surface layer, while 100 ° C nitrogen aged sample does not show. This dark surface
layer is a result of oxidation and thus called oxidized layer. The thickness of oxidized
layer (TOL) is an important property for thermal ageing of polymer. The evolution of
TOL with ageing time is plotted in Fig. 3.5. The TOL versus square root of time relation
is not linear, indicating a non-Fickian diffusion process. The TOL tends to reach a
saturated value after long-term ageing, which is a result of faster oxygen consumption
rate compared to oxygen diffusion rate. This phenomenon is common is surface oxidation
of polymers and is called “diffusion limited oxidation (DLO)”.
Another difference between 100 ° C air aged and 100 ° C nitrogen aged samples is
that the thickness of color change layer of nitrogen aged sample is greater than that of air
aged sample. This is a result of the blocking effect from the oxidized layer. Color change
of nitrogen aged sample is surface carbonization and small molecules diffusion out
through surface, and is only caused by high temperature and irrelevant with oxidation.
The oxidation layer on air aged sample blocked the diffusion process thus thinner color
change layer is observed.

65

Fig. 3.3 Surface aged layer of 100 ° C aged resin

Fig. 3.4 shows the surface color change of 150 ° C air aged sample. Compared to
100 ° C aged samples, 150 ° C aged sample shows greater degree of darkening and thicker
color change layer. Also the thickness of oxidized layer is difficult to measure because
there is no distinct color contrast. The total color change thickness is ~2 mm.
66

Fig. 3.4 Surface aged layer of 150 ° C aged resin

67

Fig. 3.5 Thickness of oxidized layer evolution

3.4.4 Static tensile property
Figs. 3.6-3.9 show the static tensile modulus and ultimate tensile strength (UTS)
evolution of 0° and 90° composites. For 0° composite, modulus shows slight decrease of
7% compared to un-aged condition after 25 weeks of ageing. No difference between
100 ° C and 150 ° C ageing is observed. UTS of 100 ° C aged decreases by 2% while UTS
of 150 ° C aged decreases by 13% after 25 weeks. The stableness of 0° modulus indicates
the stability of glass fiber to thermal ageing, because 0° modulus of composite is
dominated by fiber modulus. Stableness of the UTS of 100 ° C air aged samples indicates
the intact of fiber-matrix interface, while the 13% decrease of UTS of 150 ° C air aged
samples indicates some extent of degradation of fiber-matrix interface because 0° UTS is
affected by both fiber and interface.

68

Fig. 3.6 Tensile modulus evolution of 0° composite

Fig. 3.7 UTS evolution of 0° composite

69

Fig. 3.8 and Fig. 3.9 show complex evolution of the modulus and UTS of 90°
composite. Since 90° mechanical properties are influenced by both matrix and fiber-
matrix interface. The complexity of the evolution of modulus and UTS indicates complex
change in matrix and interface. The moduli of 100 ° C air aged and 150 ° C air aged
samples are close (decrease by 4%) after 25 weeks. The stableness of 90° modulus
indicates the stability of matrix modulus because fiber modulus is regarded constant
during ageing. The UTS of 100 ° C air aged samples shows 7% increase while the UTS of
150 ° C air aged samples shows 8% decrease after 25 weeks of ageing. The difference in
the evolution of the UTS from two ageing conditions can be explained by the competition
between matrix strengthening and interface degradation. For 100 ° C ageing condition,
interface strength is maintained while for 150 ° C ageing condition, interface is degraded.
This conclusion is consistent with the analysis from 0° composite.

Fig. 3.8 Tensile modulus evolution of 90° composite
70

Fig. 3.9 UTS evolution of 90° composite

Fig. 3.10 shows the difference between 100 ° C and 150 ° C ageing for composites.
Curves of aged samples are artificially offset for clear display. After 16 weeks of ageing,
150 ° C air aged sample shows plasticity in early stage (at ~25 MPa), but 100 ° C air aged
sample does not show this behavior and its curve is still similar to un-aged. This early
plasticity of 150 ° C aged composite can be caused by two possible factors: the early
plasticity of matrix, or the massive interface debonding at early stage. Later discussion
will reveal the latter is the real cause.
Fig. 3.10 also shows the evolution of stress-strain curve of neat resin sample. After
16 weeks of ageing, air aged samples show early failure while modulus is stable, which is
an indication of embrittlement. This embrittlement of air aged sample is caused by
surface oxidation because oxidized layer is brittle compared to un-aged polymer. And the
71

strength and failure strain decrease between 100 ° C and 150 ° C aged are close, indicating
similar degree of surface ageing. In contrast, nitrogen aged sample does not show
embrittlement and its strength and failure strain are close to un-aged values.
Comparing the stress-strain curves of neat resin and composites, the early plasticity
of 150 ° C aged composite is not caused by the degradation of matrix, but caused by the
deterioration of interface bonding. This observation is consistent with previous
discussion: ageing at 150 ° C causes significant interface damage while ageing at 100 ° C
has little influence on interface.

Fig. 3.10 Stress-strain curves of neat resin and composite

Fig. 3.11 shows the fracture surfaces of 90° tensile failure. The fracture surface of
100 ° C aged sample shows matrix shear ridges while no exposed fiber is observed. In
contrast, fracture surface of 150 ° C aged sample consists of both matrix shear bands and
72

exposed fiber surface, which is an indication of inferior interface bonding strength
compared to 100 ° C aged sample.

Fig. 3.11 SEM image of 90° tensile fracture surface

3.4.5 Short-beam shear strength
Fig. 3.12 shows the short-beam strength evolution versus ageing time. As is
consistent with the trend of tensile strength evolution, the 150 ° C aged samples show
greater decrease in SBS (20% after 16 weeks) than 100 ° C aged samples (1% after 16
weeks). Short-beam strength can be interpreted by a combined strength of matrix shear
strength and interface shear strength. And from previous discussion, interface strength
degradation is the main cause for the decrease in SBS.

73

Fig. 3.12 Short-beam strength degradation versus ageing time

3.5 Conclusions
We investigated the thermal ageing and oxidation behavior of polydicyclopentadiene
(pDCPD) polymer and its unidirectional glass fiber composites. Two temperatures of
100 ° C and 150 ° C are used while two atmospheres of air and nitrogen are used. Thermal
properties are measured by dynamic mechanical analysis. Mechanical properties are
measured by static tensile test and short-beam shear test. The surface oxidized layer is
monitored by optical microscope. Results show that the mechanical properties of pDCPD
neat resin and composites are stable during 25 weeks of ageing in these environments.
Glass transition temperature increased after long-term ageing, which is a typical
secondary polymerization phenomenon in thermal ageing. High temperature (150 ° C)
showed significant damage to fiber-matrix interface compared to low temperature
(100 ° C) ageing. Evidences are the early plasticity of 90° tensile test of the composite and
74

the exposed fiber surface of the fracture surface. Thus, interface is one aspect
manufacture should pay attention to when using pDCPD composites in elevated-
temperature environments.
Thermal ageing of polymers and composites is a key issue for high-temperature
applications, e.g., oil risers. Our study shows the investigated new formulation of pDCPD
resin is stable exposing to the studied thermal ageing environments. One problem is the
fiber-matrix interface degradation at high ageing temperature, which needs further
research on how to strengthen this weak link. Future work includes measuring fiber-
matrix interface strength directly by single-fiber push-out test method. We also need to
study other thermal ageing temperatures and combining loading with temperature (creep)
to obtain an overall view of the properties of this pDCPD polymer under thermal attack.

75

Chapter 3 References
1. Schieffer A, Mairea JF, Lé vê queb D, A coupled analysis of mechanical behaviour
and ageing for polymer-matrix composites, Composites Science and Technology
2002;62:543-549.
2. Park SJ, Seo MK, Lee DR, Studies on the mechanical and mechanical interfacial
properties of carbon-carbon composites impregnated with an oxidation inhibitor,
Carbon 2003;41:2991-3002.
3. Abu-Sharkh BF, Hamid H, Degradation study of date palm fibre/polypropylene
composites in natural and artificial weathering: mechanical and thermal analysis,
Polymer Degradation and Stability 2004;85:967-973.
4. Colin X, Verdu J, Strategy for studying thermal oxidation of organic matrix
composites, Composites Science and Technology 2005;65:411-419.
5. Guisandez J, Tiemblo P, Gó mez-Elvira JM, Change of thermal and dynamic-
mechanical behaviour of a metallocene isotactic polypropylene during low-
temperature thermo-oxidation, Polymer Degradation and Stability 2005;87:543-553.
6. Zenkiewicz M, Rauchfleisz M, Czupryńska J, Polański J, Karasiewicz T, Engelgard
W, Effects of electron-beam irradiation on surface oxidation of polymer composites,
Applied Surface Science 2007;253:8992-8999.
7. Olivier L, Ho NQ, Grandidier JC, Lafarie-Frenot MC, Characterization by ultra-micro
indentation of an oxidized epoxy polymer: Correlation with the predictions of a
kinetic model of oxidation, Polymer Degradation and Stability 2008;93:489-497.
76

8. L. Olivier, Baudet C, Bertheau D, Grandidier JC, Lafarie-Frenot MC, Development
of experimental, theoretical and numerical tools for studying thermo-oxidation of
CFRP composites, Composites A 2009;40:1008-1016.
9. Lafarie-Frenot MC, Grandidier JC, Gigliotti M, Olivier L, Colin X, Verdub J,
Cinquin J, Thermo-oxidation behaviour of composite materials at high temperatures:
A review of research activities carried out within the COMEDI program, Polymer
Degradation and Stability 2010;95:965-974.
10. Awaja F, Nguyen MT, Zhang S, Arhatari B, The investigation of inner structural
damage of UV and heat degraded polymer composites using X-ray micro CT,
Composites A 2011;42:408-418.
11. Richaud E, Le Gac PY, Verdu J, Thermooxidative ageing of polydicyclopentadiene in
glassy state, Polymer Degradation and Stability 2014;102: 95-104.
12. Celina MC, Review of polymer oxidation and its relationship with materials
performance and lifetime prediction, Polymer Degradation and Stability
2013;98:2419-2429.
13. Middleton J, The effect of ozone on polymer degradation in Polymer Core Composite
Conductors, Polymer Degradation and Stability 2013;98:436-445.
14. Gigliotti M, Grandidier JC, Lafarie-Frenot MC, The employment of 0/90
unsymmetric samples for the characterization of the thermo-oxidation behaviour of
composite materials at high temperatures, Composite Structures 2011;93:2109-2119.
15. Gigliotti M, Olivier L, Vu DQ, Grandidier JC, Lafarie-Frenot MC, Local shrinkage
and stress induced by thermo-oxidation in composite materials at high temperatures,
Journal of the Mechanics and Physics of Solids 2011;59:696-712.
77

16. Gigliotti M, Grandidier JC, Lafarie-Frenot MC, Assessment of chemo-mechanical
couplings in polymer matrix materials exposed to thermo-oxidative environments at
high temperatures and under tensile loadings, Mechanics of Materials 2011;43:431-
443.
17. Gigliotti M, Pannier Y, Minervino M, Lafarie-Frenot MC, Corigliano P, The effect of
a thermo-oxidative environment on the behaviour of multistable [0/90] unsymmetric
composite plates, Composite Structures 2013;106:863-872.
18. Haque MH, Upadhyaya P, Roy S, Ware T, Voit W, Lu H, The changes in flexural
properties and microstructures of carbon fiber bismaleimide composite after exposure
to a high temperature, Composite Structures 2014;108:57-64.
19. Vu DQ, Gigliotti M, Lafarie-Frenot MC, Experimental characterization of thermo-
oxidation-induced shrinkage and damage in polymer–matrix composites, Composites:
Part A 2012;43:577-586.
20. Vu DQ, Gigliotti M, Lafarie-Frenot MC, The effect of thermo-oxidation on matrix
cracking of cross-ply [0/90] composite laminates, Composites A 2013;44:114-121.

78

CHAPTER 4. TRANSVERSE COMPRESSION FAILURE OF
UNIDIRECTIONAL COMPOSITES
4.1 Abstract
The mechanical response of unidirectional composites subject to uniaxial transverse
compressive loads was measured and analyzed by finite element simulation. Consistency
in failure plane orientation was observed when comparing simulated matrix shear band
angle to measured crack angle. A model based on hexagonal packing of fibers was
proposed and the shear band angle was shown to depend on the fiber volume fraction.
The effects of strong and weak fiber-matrix interfaces were considered using models with
randomly distributed fibers for a valid statistical analysis. The results of these models
showed that the composite compressive strength increased with the fiber loading for the
strong interface case, while the strength was independent of the fiber loading for the
weak interface case because of interface debonding.

4.2 Introduction
Unidirectional (UD) composites generally exhibit high stiffness and tensile strength
in the longitudinal direction, although these same properties are much lower in the
transverse directions. The overall compressive failure of UD composites in many cases
initiates from transverse cracking [1]. The crack angle under uniaxial compression, which
is the angle between the failure plane and the plane perpendicular to the loading axis, is
generally 50° -60° for UD composites with polymer matrix [2-6]. The neat polymer
matrix also shows a failure plane angle greater than 50° [7-9] but is usually several
79

degrees lower than the crack angles of composites. This difference stems from the
presence of fibers and will be considered in the present study.
Several theories address the failure of unidirectional composites [10-15] and attempt
to describe the development of the failure plane. Mohr-Coulomb theory [10] assumes
compressive failure occurs in the form of shear failure along a specific plane, and the
orientation of the failure plane is determined by friction angle and cohesion, both intrinsic
material properties. The failure plane angle range is 45° -90° from Mohr-Coulomb theory.
Christensen’s theory [11] assumes the failure plane orientation is where the normal
plastic strain increment is zero, and yields a failure plane angle range of 45° -60°. Puck’s
theory [12] predicts that, the failure plane occurs where the angle-dependent effort
function reaches a maximum. The transverse failure plane angle and strength are affected
not only by transverse properties, but also by the longitudinal stress state. All of the
above theories generally treat composites as a continuum, and do not account for
microscopic failure mechanisms involving fiber-matrix interactions, which is critical to
understanding composite failure mechanisms.
Several recent simulation studies have attempted to address micro-level failure
mechanisms [6, 16-22]. For example, LLorca et al. [6, 19, 20] investigated transverse
properties of UD composites using the representative volume element (RVE) method by
finite element simulation. Random distributions of fibers were simulated, and the
influence of matrix and interface properties on the overall composite behavior was
studied. In contrast, Blassiau et al. [16] studied the load transfer mechanism in UD
composites using models with regularly packed fibers. The effect of fiber volume fraction
and interface debonding on the load transfer and overall mechanical response was
80

investigated. Correa et al. [17, 18] studied the fiber-matrix interface crack growth in UD
composites under transverse compression, using a single fiber model. Ghassemieh et al.
[21] investigated the influence of fiber loading on stress concentration, matrix failure and
interface debonding, and simulations were based on a model with hexagonally packed
fibers. Kok et al. [22] studied the effect of fiber volume fraction, temperature and
interface adhesion on matrix yielding and fracture, using finite element models with
quadratically and hexagonally packed fibers. However, simulations listed above did not
address the issue of failure plane orientation, and the effect of fiber volume fraction was
not studied using randomly distributed fibers. In any event, fiber distributions in
composites are never regularly packed, and thus accurate simulations must feature
statistically random spatial distributions of fibers.
Here, we employ a micromechanics analysis to show how the fiber volume fraction
(f) and interface strength influence failure plane angle, compressive strength, strain
concentration and interfacial debonding in UD composites loaded in transverse
compression. We build RVE models of different fiber volume fraction levels, and
investigate the effect of fiber volume fraction on the failure plane orientation. The models
were built with randomly distributed fibers. For each condition, a quantity of models with
different fiber distributions was computed to obtain a statistically valid result. Simulation
results show consistency with experimental data: failure plane angle increases with fiber
volume fraction, depending on interface status. Experiments using samples with f = 70%
were conducted, and finite element models of f = 10%, 30% and 50% were employed.
For comparison, experimental data of f < 70% from literature [2-6] was used. Simulation
81

data for random RVE of f > 50% was not available and is reported because of numerical
convergence difficulty.

4.3 Failure Plane Orientation Prediction
As mentioned, failure plane angle of UD composites under transverse compression is
generally greater than that of neat matrix, although they are both in the range of 50-60° .
The difference can be explained by the distortion of the matrix shear band caused by the
presence of fibers. As f increases, the fiber distribution approaches hexagonal packing,
which is the densest possible arrangement. Fig. 4.1 shows a cross-section of a UD
composite with regular hexagonal packing of fibers, in which both the fiber diameters (d)
and the distance between adjacent fibers (D) are uniform.

Fig. 4.1 Plastic shear band in hexagonal model

A dimensionless parameter ξ = d/D is used to describe the areal density of fibers: ξ =
0 indicates a neat matrix condition, while ξ = 1 indicates the maximum theoretical fiber
82

volume fraction. The relationship between ξ and the fiber volume fraction (f) can be
derived from the geometric relationship as (Eqn. 4.1):
2
23
f

 
(4.1)
in which
max
91%
23
f


when ξ = 1.
As shown in Fig. 4.1, when this model undergoes compression in the y direction (the
x direction is assumed to be unconstrained), the matrix shear stress reaches a maximum
and forms a shear band along the bold line [4], which is the common tangent of fibers #1
and #2. This plastic shear band can be the most likely potential failure plane, and the
angle θ can be calculated as (Eqn. 4.2):
1 1 1
/ 2 2
sin sin sin
6 6 6
3 / 2 3 3
d
f
D
   


  

   
     

   

   

(4.2)
However, Eqn. 4.2 is not sufficient to limit the entire range of θ, because other factors
become critical when f approaches minimum and maximum. The lower bound of θ is
restrained by the neat matrix failure angle θmatrix when f approaches 0 and the fibers are
sparse and cannot effectively deflect the shear band. The higher bound of θ is restrained
by the hexagonal packing of fibers. When f approaches a critical value and the shear band
trace contacts fibers #3 and #4 in Fig. 4.1, the shear band is locked at 60° by the
surrounding fibers. This critical f is 68%, back-calculated from Eqn. 4.2. Applying these
two constraints, the final form of the f-θ relation is (Eqn. 4.3):
 
matrix
o1
o
for 0
30 sin 0.61 for 68%
60 for 68%
f
ff
f



 

  




(4.3)
83

However, the above relation is based on regular hexagonal packing of fibers, although
in practice, fiber distributions in composites are random. Thus, experiments and
simulations of samples with randomly distributed fibers are required to validate this
prediction, as described below. Note that the higher bound of 60° predicted by Eqn. 4.3 is
consistent with Christensen’s continuum based theory [11], implying a possible intrinsic
connection between the micro-level and macro-level models.

4.4 Finite element analysis
4.4.1 Model generation
Micro-level 2D RVE models with randomly distributed fibers were generated using a
code developed by the authors (Matlab, MathWorks Inc.), and static transverse
compression was simulated using commercial software (Abaqus, Dassault Systemes).
First, a sequential random algorithm [23] was used in the code to generate randomly
distributed parallel fibers. In this algorithm, individual fibers are generated sequentially at
random locations, and if a newly generated fiber overlaps with an existing one, it is re-
generated until there is no overlap. Compared to the Monte Carlo algorithm, the
sequential algorithm is simpler and faster, although the greatest f is generally limited to <
60%. Second, geometric information of the fibers such as center location and diameter
was passed into a script (Python). The script can be executed in finite element software
(Abaqus) to generate the parts and assembly of fibers and matrix. The final step is
defining material properties, interface and contact parameters, meshing the model, and
running the analysis in finite element software (Abaqus).
84

Three fiber volume fraction levels (10%, 30% and 50%) were simulated. For each
fiber volume fraction level, 10 iterations of models with different random fiber
distributions were generated and computed for a statistical analysis. For each iteration,
simulations were performed twice with the interface strength set to strong and weak,
respectively, to study the influence of the interface on the overall mechanical response.
Fig. 4.2 shows one f = 50% model with an inset to reveal the mesh detail. The model size
was (100 × 100) μm, and the fiber diameter was 10 μm with 0.5 μm standard deviation.
Models with f = 10% and 30% have the same model size and fiber dimension. The
statistical distribution of fiber diameters was determined from SEM images of polished
cross-sections of the samples. For both fiber and matrix, the mesh size was 1 μm, and
linear triangular plane strain elements (CPE3) were used. Interface nodes of fiber and
matrix are allowed to overlap at the beginning of simulation to achieve interaction.
Uniaxial transverse compression was simulated in the vertical direction using general
static analysis step. The two side surfaces were constrained to be vertical and straight
during compression, but no force or displacement boundary conditions were applied. We
assumed a fiber modulus of 72.3 GPa and a Poisson’s ratio of 0.22, values that are typical
for E-glass fibers, and the fibers were assumed to be isotropic.

85

Fig. 4.2 One f = 50% model mesh grid

4.4.2 Matrix failure criterion and progression rule
According to Mohr-Coulomb theory, failure under uniaxial compression occurs by
shear failure along a specific plane. The shear strength (τ) on the failure plane is affected
by the normal stress component (σ) on it. Eqn. 4.4 describes the Mohr-Coulomb failure
criterion [24]:
tan c     (4.4)
in which c is the cohesion and  is the friction angle, both of which are intrinsic
properties of the material. The failure plane angle θ is given below (Eqn. 4.5):
1
22






(4.5)
For epoxy, the measured failure plane angle is 50° -60° [7-9], and the friction angle
can be back-calculated from Eqn. 4.5. A friction angle of 14° was used in the models,
corresponding to a failure plane angle of 52° . The failure plane angle for a neat epoxy
resin is reportedly 52° [7], although the mechanical properties for the polymer matrix in a
composite can be quite different from that of the neat resin. Thus, this input value for the
86

angle is chosen only for illustration purposes and for comparison with the failure plane
angle of composites. A cohesion stress of 45 MPa was assumed, based on published data
[7-9], and the compressive strength was calculated to be 115 MPa from Mohr-Coulomb
theory. A modulus of 3.1 GPa was implemented and the matrix resin was assumed to be
isotropic.
The failure progression rule (flow rule) from [25] was adopted, which describes
material’s behavior after the failure criterion (Eqn. 4.4) was reached. In this flow rule, the
flow potential G for the yield surface is a hyperbolic function in the meridional stress
plane (Eqn. 4.6):
   
22
tan tan
mw
G c R q p       (4.6)
where
   
     
2
22
2 2 2 2
4 1 cos 2 1
3 sin
6cos
2 1 cos 2 1 4 1 cos 5 4
mw
ee
R
e e e e e


   


       

and
 
3
cos 3
r
q





3 sin
3 sin
e






in which ε is the meridional eccentricity, ψ is the dilation angle, q is the Mises stress, p is
the equivalent pressure stress, and r is the third invariant of deviatoric stress. A value of
0.1 was used for ε, which defines the rate at which the hyperbolic function approaches the
asymptote. The dilation angle ψ, which defines the material volume expansion during
shearing, is always less than or equal to the friction angle  , and ψ > 0 indicates material
87

expansion while ψ < 0 indicates contraction during shearing. When ψ =  , the plastic
strain increment is always normal to the yield surface (i.e., associated flow), a condition
that applies for metals but may not apply to polymers. A dilation angle of 12° was chosen
for the present study to simulate the general case of non-associated flow of polymers. A
comparison was made between 12° and 14° dilation angles for the same model, and
results were only slightly different, indicating that the final results are not sensitive to
dilation angle. Both the failure criterion and progression rule of the matrix were
implemented using the material definition module in finite element software (Abaqus).

4.4.3 Fiber-matrix cohesive interface
A cohesive interface [6, 19, 20] was used to describe the fiber-matrix debonding. Fig.
4.3 shows the relationship between the interfacial stress (t) and the debonding
displacement (δ).

Fig. 4.3 Interfacial debonding vs. interfacial stress

For a 2D problem, there are two interfacial stress components - normal (tn) and shear
(ts), which are related to normal and shear separations δn and δs, respectively. There are
88

two stages in the debonding process. In the first stage (initial loading), δ increases
linearly with t (Eqn. 4.7), and no damage occurs. The second stage begins when t reaches
a maximum (i.e., the debonding strength, t
0
). From this moment, damage ensues, and t
starts to decrease while δ starts to increase, albeit more quickly. The significant increase
in δ appears as debonding occurs. A dimensionless parameter D is introduced to describe
the damage level (Eqn. 4.8) [24], and D = 0 implies no damage, while D = 1 implies
complete debonding. The shaded area under the δ-t curve in Fig. 4.3 represents the total
interfacial energy J.
n nn ns n
s sn ss s
t K K
t K K


     
  
   

     
t Κ δ (7)
 
 
f0
f0
D
  
  



(8)
We used an interfacial stiffness Knn = Kss = 1× 10
8
GPa/m [6] for both normal and
shear directions, and assumed no coupling between the two directions (Kns = Ksn = 0 in
Eqn. 4.7). The interfacial energy (J) was set to 100 J/m
2
[26]. For the strong interface
case, fibers surfaces were tied to the matrix, and the cohesive interface was not used,
ensuring debonding did not occur. For the weak interface models, the cohesive interface
was implemented, with an interface strength of
0
n
t =
0
s
t =30 MPa, which is 70% less than
the matrix compressive strength (115 MPa). The weak interface can be regarded as a
degraded interface, such as often results from ageing. The magnitude of glass fiber-epoxy
interface strength was reported to be 10-100 MPa [27-30].
89

Post-debonding contact between fiber and matrix was enabled using “hard contact”
in the normal direction and frictionless contact in the shear direction [24]. “Hard contact”
behavior can be described by Eqn. 4.9:
0 for 0 (open), and
0 for 0 (closed)
ph
hp
 



(4.9)
in which p is the contact pressure between two surfaces, and h is the “overclosure” (the
interpenetration of the surfaces). Thus, for “hard contact”, extrusion of the two surfaces
into each other is forbidden, and contact stress can only be compressive. The frictionless
assumption in the shear direction can lead to some underestimation of overall stiffness
and strength, and there is no reliable measured data of the interface friction that can be
used in the model. The cohesive debonding rule and post-cracking contact of the interface
were implemented using contact definition module in finite element software (Abaqus).

4.5 Experiments
4.5.1 Specimen preparation
Unidirectional glass fiber (E-glass 366) composite rod with an epoxy matrix (Lindoxy
190) was manufactured using a hot pultrusion process (Composite Technology
Corporation, Irvine, CA), yielding rods with diameter of ~9 mm. The fiber volume
fraction was ~70%, measured by cutting and polishing a cross-section of the rod sample
and viewing under scanning electron microscope (JEOL JSM-6610 SEM). The rod was
further cut using a diamond saw (Struers Minitom) into cuboidal samples in the
dimensions of height×width×thickness = (6× 3× 4) mm. The height-to-width ratio was 2:1
in accordance with conventional compression tests. Surfaces of the cuboidal samples
were then polished to ensure that they were smooth and flat, and to ensure that the sample
90

was in strict cuboidal shape to prevent rotation and shear during compression test.
Finally, a black-and-white speckle pattern was applied to the front surface for accurate
strain measurement by digital image correlation (DIC), as shown in Fig. 4.4.

Fig. 4.4 Digital image correlation pattern of cuboidal sample

4.5.2 Quasi-static compression test
Cuboidal samples were transversely loaded in compression until fracture occurred
using an electric-motor driven load frame (Instron 5567). The loading rate was low (1
mm/min) to ensure a quasi-static stress state in the sample and to minimize the strain rate
effect. 16 samples in total were prepared and tested. During testing, load was recorded,
and images of the DIC pattern on the front surface were recorded at select time intervals.
The images were post-processed using DIC analysis software (VIC-2D, Correlated
Solution, Inc.) to obtain accurate strain values. Crack angles were also measured from
microscopic images after test, and fractography was performed using SEM (JEOL JSM-
6610 SEM).

91

4.6 Results and discussion
4.6.1 Failure plane orientation
In Fig. 4.5, we sum the θ vs. f data taken from experiments, from measured values in
the literature [2, 5-7], and from simulations in the present work. Each simulation data
point represents the average of the ten iterations of models, and the standard deviation is
less than 6% for all simulated failure plane angles. The approach for determining the
angle is consistent. In the low range of f, experimental data are not available, since f >
50% for most composites. Both strong- and weak-interface models at f = 10% show θ =
52° , and this value is the same as the input neat matrix failure plane angle, confirming the
negligible role of fibers in the transverse direction. In the medium range of f, reported θ
values [2, 6] and models with strong interface agree with the hexagonal model prediction,
while weak-interface models show failure plane angles of ~52° at all fiber volume
fraction levels. Therefore, Eqn. 4.3 is applicable when matrix yield dominates the failure.
In the high range of f, simulation data are not available because of the numerical
convergence difficulty in the fiber loading model. All measured and reported values [5]
are 60° , in accordance with the prediction of Eqn. 4.3. Whether the failure is interface-
dominated or not is unclear for these data points.
92

Fig. 4.5 Failure plane angle vs. fiber volume fraction

Fig. 4.6 shows the simulated plastic strain contours for different fiber volume
fractions and different interface conditions. Figs. 4.6 (a) and (b) feature the same f = 10%
model, while Figs. (c) and (d) are from the same f = 30% model, and Figs. (e) and (f) are
from the same f = 50% model. A strong interface was used in Figs. (a), (c) and (e), while
a weak interface was used in Figs. (b), (d) and (f). Each model at a specific fiber volume
fraction level represents only one of the ten iterations, each with different fiber
distributions. For f = 10% (Figs. (a) and (b)), plastic shear bands form round fibers. Small
strain concentrations appear in regions near the interface. Multiple shear bands form in
the matrix, and they all show an angle of 52° , which is identical to the neat matrix failure
angle. No difference is observed for strong and weak interfaces. Thus at f = 10%, the
fiber content is so low that it has a negligible effect on the composite mechanical
behavior, and the composite behavior is similar to the neat resin. In contrast, for f = 30%
93

and 50% models, shear bands are random and their orientations are much influenced by
the packing of fibers.
To determine the failure plane for these models, the location of the maximum plastic
strain (ε
pl
max) is marked in each figure, and the shear band showing the maximum plastic
strain in the model is regarded as the potential failure plane. The angle of this plane was
measured as the angle of the centerline of the shear band, as shear bands generally have
widths equal to the spacing between neighboring fibers. In this way, the failure plane for
a single model was determined. However, the failure plane angle θ showed slight but
consistent variance from different models because the arrangement of fibers for each
model was random. The value of θ from the ten iterations of the models at each fiber
volume fraction level was summarized to yield a valid statistical result and to minimize
the artificial measurement error. The standard deviation was less than 6% for each fiber
volume fraction level. (This issue is addressed later.) Note that the angles marked in Figs.
(c, d, e, f) are only for these specific models and do not represent the average from ten
iterations. For Figs. (d) and (f), one fiber was magnified to show the interface debonding
state. The polar coordinate α was constructed regarding α = 0 at the horizontal position.

94

Fig. 4.6 Simulated plastic shear bands

The deviation effect of fibers on matrix shear band is clearly illustrated in Fig. 4.6,
which confirms the assumption in the hexagonal model. As fibers become more densely
packed, the arrangement gradually shifts from random packing to hexagonal packing, as
hexagonal packing is the most efficient arrangement. Thus, the hexagonal model is more
applicable to high fiber loading cases. Shear bands in strong-interface models are
95

continuous and smooth, while the shear bands in weak-interface models are disconnected
and random. This phenomenon is caused by the state of the interface: for the intact
interface model, the stress field is continuous everywhere in the model, while for the
damaged interface model, the stress is discontinuous across the interface, causing the
discrete appearance of the strain field. Thus the influence of fibers on matrix is
diminished in damaged interface models, and their failure plane angles deviate from the
prediction in section 4, since interface debonding is not accounted for in the hexagonal
model.
Fig. 4.7 shows the measured crack angles for one cuboidal sample at the same
location at different magnifications. The crack angle is consistent from the millimeter
scale to the micrometer scale. The rightmost image in Fig. 4.7 is the same scale as the
RVE models. Thus, the failure plane angles from experiment and simulation are
comparable in length scale. The average crack angle from 16 samples was 60° with a
standard deviation of 5%. This crack angle at f = 70% is consistent with reported values
[5] and with the Eqn. 4.3 prediction. The image of the fracture surface is inset in the
leftmost figure. Both bare glass fiber surface and matrix fracture are shown, indicating a
combined failure of matrix and interface. Whether the damage is initiated/dominated by
matrix shear or interface debonding is difficult to determine from the images.

96

Fig. 4.7 Experimental crack angle and fracture surface

4.6.2 Stress-strain curves
Fig. 4.8 shows the stress-strain curves of models and experiment. Each curve
represents the average of all the models/samples at the same fiber volume fraction level.
The “+” mark on the experimental curve indicates the point of brittle fracture, which
occurs at a failure strain < 2%. The standard deviation is less than 4% for both
experiment and simulation curves. Note that the experimental curve may not be
comparable with simulated curves because (1) the experimental curve is for a sample
with f = 70%, while the simulated curves are for models with f = 10%-50%, and (2) the
mechanical properties input of the fiber, matrix and interface in the models are from
literature data and not from measured values.
97

Fig. 4.8 Stress-strain curves of finite element models

From Fig. 4.8, modulus in transverse direction increases with f as predicted by a
simple rule-of-mixtures (Eqn. 4.10).
11
c f m
ff
E E E

 (4.10)
in which Ec, Ef and Em are the modulus of the composites, fiber and matrix, respectively.
For f = 10%, no difference was observed between strong and weak interfaces, because
fibers are sparse and have negligible effect on the overall behavior. For f = 30% and 50%,
yielding of the weak-interface models occurs earlier than the strong-interface models.
This is caused by interfacial debonding before matrix yield, since the interface strength is
70% less than the matrix yield strength.
98

For all weak-interface models in Fig. 4.8, the maximum stress that can be reached is
~115 MPa, the matrix compressive yield strength. Thus, when the interface strength is
significantly less than the strength of the matrix and when massive interface failure
occurs before matrix yield, the strength of the composites is determined by the matrix,
regardless of the fiber content. For the strong-interface models in Fig. 4.8, strength
increases with f, a result of the deviatoric stress level change with f [2,22]. As reported
previously [2, 22], when the fiber volume fraction increases, under external transverse
compression load, the stress state in the matrix tends to be tri-axial instead of uniaxial.
Thus, the deviatoric stress (shear stress) in the matrix decreases as f increases, and leads
to a higher compressive strength because matrix failure is triggered by shear failure along
plastic shear bands.

4.7 Conclusions
The effects of fiber volume fraction and interface strength on the uniaxial transverse
compressive behavior of UD composites were studied through experimental
measurements and finite element simulations. A model featuring hexagonal packing of
fibers was proposed to predict the failure plane angle of composites. Predictions were
consistent with published data for composites with f > 50% (data for composites with f <
50% are not available). The predicted failure plane angle for f > 68% was fixed at 60° ,
although this prediction awaits validation by direct observation and simulation. The
model predictions also were consistent with finite element analyses assuming strong
interfaces, although the predictions were not consistent with weak-interface finite
element models because debonding was not taken into account.
99

Analysis of stress-strain curves shows that composites become more brittle as the
fiber volume fraction increases, a consequence of strain concentration. Although the neat
resin can be ductile, the failure of composites is generally brittle because the strain
magnification factor in the matrix can be up to 10× in the plastic range. The compressive
strength values determined from finite element simulations indicate a strength increase
with increasing fiber loading for strong-interface models. The mechanism responsible for
this phenomenon is the deviatoric stress decrease with increasing fiber loading (provided
the interface is intact). However, for the weak interface condition, interfacial debonding
occurs prior to matrix yield, leading to behavior similar to the unreinforced matrix,
regardless of fiber volume fraction.
Insights into the micromechanical effects of fibers and interfaces are critical to
understanding deformation and failure mechanisms in UD composites. Using the
methods presented here, crack angle and strength can be estimated before testing, and the
failure criteria can be used to guide design. The failure plane angle range and failure
modes from our investigations are consistent with those of Christensen’s continuum-
based theory: failure is ductile and the failure plane angle is low at low fiber volume
fractions, while failure is brittle and the failure plane angle is high at high fiber volume
fractions. This consistency suggests a possibility of linking the two scales to build
accurate multi-scale failure models.

100

Chapter 4 References
1. Kar NK, Hu Y, Barjasteh E, Nutt SR, Tension-tension fatigue of hybrid composite
rods, Composites B 2012;43:2115-2124.
2. Collings TA, Transverse compressive behaviour of unidirectional carbon fibre
reinforced plastics, Composites 1974;5:108-116.
3. Daniel IM, Luo J, Schubel PM, Werner BT, Interfiber/interlaminar failure of
composites under multi-axial states of stress, Composites Science and Technology
2009;69:764-771.
4. Hsu SY, Vogler TJ, Kyriakides S, Inelastic behavior of an AS4/PEEK composite
under combined transverse compression and shear, International Journal of Plasticity
1999;15:783-806.
5. Huang YK, Frings PH, Hennes E, Mechanical properties of Zylon/epoxy composite,
Composites B 2002;33:109-115.
6. Gonzalez C, LLorca J, Mechanical behavior of unidirectional fiber-reinforced
polymers under transverse compression: Microscopic mechanisms and modeling,
Composites Science and Technology 2007;67:2795-2806.
7. Fiedler B, Hojo M, Ochiai S, Schulte K, Ando M, Failure behavior of an epoxy
matrix under different kinds of static loading, Composites Science and Technology
2001;61:1615-1624.
8. Quinson R, Perez J, Rink M, Pavan A, Yield criteria for amorphous glassy polymers,
Journal of Materials Science 1997;32:1371-1379.
9. Yee AF, Du J, Thouless MD, Toughening of epoxies, Polymer Blends, 2000.
10. Paul B, A modification of the Coulomb-Mohr theory of fracture, Journal of Applied
101

Mechanics 1961;06:259-268.
11. Christensen RM, DeTeresa SJ, Failure plane orientations for transverse loading of a
unidirectional fiber composite, International Journal of Solids and Structures
2003;40:7055-7062.
12. Puck A, Schurmann H, Failure analysis of FRP laminates by means of physically
based phenomenological models, Composites Science and Technology 1998;58:1633-
1662.
13. Hashin Z, Failure Criteria for Unidirectional Fiber Composites, Journal of Applied
Mechanics 1980;47:329-334.
14. Tsai SW, Wu EM, A general theory of strength for anisotropic materials, Journal of
Composite Materials 1971;5:58-80.
15. Pinho ST, Davila CG, Camanho PP, Iannucci L, Robinson P, Failure models and
criteria for FRP under in-plane or three-dimensional stress states including shear non-
linearity, NASA/TM-2005-213530.
16. Blassiau S, Thionnet A, Bunsell AR, Micromechanisms of load transfer in a
unidirectional carbon fibre–reinforced epoxy composite due to fibre failures. Part 1:
Micromechanisms and 3D analysis of load, Composite Structures 2006;74:303-318.
17. Correa E, Mantic V, Paris F, Numerical characterisation of the fibre-matrix interface
crack growth in composites under transverse compression, Engineering Fracture
Mechanics 2008;75:4085-4103.
18. Correa E, Mantic V, Paris F, A micromechanical view of inter-fibre failure of
composite materials under compression transverse to the fibres, Composites Science
and Technology 2008;68:2010-2021.
102

19. Canal LP, Segurado J, LLorca J, Failure surface of epoxy-modified fiber-reinforced
composites under transverse tension and out-of-plane shear, International Journal of
Solids and Structures 2009;46:2265-2274.
20. Totry E, Gonzalez C, LLorca J, Failure locus of fiber-reinforced composites under
transverse compression and out-of-plane shear, Composites Science and Technology
2008;68:829-839.
21. Ghassemieh E, Nassehi V, Prediction of Failure and Fracture Mechanisms of
Polymeric Composites Using Finite Element Analysis, Polymer Composites
2001;22:528-554.
22. Kok JMM, Meijer HEH, Deformation, yield and fracture of unidirectional composites
in transverse loading 1. Influence of fibre volume fraction and test-temperature,
Composites A 1999;30:905-916.
23. Segurado J, LLorca J, A numerical approximation to the elastic properties of sphere-
reinforced composites, Journal of the Mechanics and Physics of Solids 2002;50:2107-
2121.
24. Abaqus user manual, v6.11, Dassault Systemes, (2011).
25. Mené trey Ph, Willam KJ, A triaxial failure criterion for concrete and its
Generalization, ACI Structural Journal 1995;92:311-318.
26. Zhou XF, Wagner HD, Nutt SR, Interfacial properties of polymer composites
measured by push-out and fragmentation tests, Composites A 2001;32:1543-1551.
27. Koenig JL, Emadipour H, Mechanical characterization of the interfacial strength of
glass-reinforced composites, Polymer Composites 1985;6:142-150.
28. Liao YT, A study of glass fiber-epoxy composite interfaces, Polymer Composites
103

1989;10:424-428.
29. Rao V, Drzal LT, The dependence of interfacial shear strength on matrix and
lnterphase properties, Polymer Composites 1991;12:48-56.
30. Favre JP, Auvray MH, Cheneau-Henr P, Galiotis C, Vlatias C, Paipetis A, Pegoraro
M, Severini F, Di Landro L, Yuan L, Fiber/matrix mechanical interaction in carbon
fiber/bismaleimide model composites, Polymer Composites 1996;17:937-947.

104

CHAPTER 5. CONCLUSIONS AND FUTURE WORK
The hygrothermal ageing and thermal oxidation ageing of a new formulation of
polydicyclopentadiene (pDCPD) polymer and composites were investigated in this work.
This new formulation of pDCPD is manufactured using a ruthenium-based catalyst and
shows considerable advantages over traditional polymers in chemical-ageing resistance,
impact resistance, fracture toughness and etc. The potential applications of this new
polymer are applications in corrosive environments, e.g., wind turbine blades, protection
shell of offshore oil risers. However, a thorough understanding of the long-term ageing
behavior of this new polymer is needed because these civilian infrastructures are required
to last decades with minimal maintenance. In this study, we investigated two most
common ageing mechanisms in service - hygrothermal ageing and thermal oxidation
ageing, and studied the effects of ageing on the thermal and mechanical properties of the
pDCPD polymer and composites, including glass transition temperature, static tensile
strength, shear strength, tension-tension fatigue properties and etc. The failure procedure
and how ageing affects the mechanical behavior was analyzed and discussed, and ways of
strengthening this material to ageing are suggested. The goal of this dissertation was to
improve the understanding of the long-term ageing effects on mechanical properties of
this new polymer from a scientific standpoint, while producing results that can be applied
to service conditions.
In Chapter 2, the hygrothermal ageing behavior of pDCPD polymer and its glass
fiber composite was studied, and the effects of ageing on tensile fatigue properties were
investigated. Hygrothermal ageing is a common ageing mechanism for wind turbine
blade application. In this study, a traditional epoxy polymer for making wind turbine
105

blade was used as benchmark material. Two ageing conditions of both deionized water
and salt water were used at an elevated temperature to accelerate the ageing process. The
pDCPD polymer and composites showed significantly less water absorption compared to
epoxy sample, thanks to its intrinsic hydrophobicity from molecular level. Thus,
plasticization of matrix was observed in epoxy samples but not in pDCPD samples.
Because of the less water uptake in pDCPD, its matrix showed less swelling and the
strength of fiber-matrix interface was well maintained over months of ageing. In contrast,
the fiber-matrix interface in epoxy composites showed significant degradation and thus
the fatigue resistance decreased significantly over a relatively short ageing period
compared to pDCPD composites. The degradation of the strength of glass fiber in the two
composites was synchronous and was primarily dependent on ageing time. In contrast,
the degradation of fiber-matrix depended mainly on water absorption level in the matrix.
The difference between deionized water ageing and salt water ageing was only the
difference in the amount of water absorbed, while no salt corrosion effect was observed
for both resin systems.
In Chapter 3, the thermal ageing and oxidation of pDCPD polymer and glass fiber
composites were investigated. Both air and nitrogen environments were used to study the
effect of thermal ageing with and without oxygen. Both ageing temperatures below and
above the original glass transition temperature of pDCPD were used to study the effects
of ageing temperature. A thin (~100 μm) surface oxidation layer was observed in air aged
samples but not in nitrogen aged samples. And this oxidation layer was increasing to a
saturated value, indicating a typical diffusion limited oxidation behavior. The oxidation
layer was brittle compared to un-aged inside bulk resin and caused the embrittlement of
106

the whole neat resin tensile samples after long time (16 weeks) of ageing. At initial stage
of ageing, un-cured DCPD monomers and oligomers diffused out from the sample and
caused the initial weight loss, especially for nitrogen aged samples. Chain-scission also
occurred and small molecules generated from these reactions acted as plasticizers and
caused slight decrease in glass transition temperature when aged at low temperature. At
higher ageing temperature or after longer ageing time, secondary polymerization
dominated while strengthening and increase in glass transition temperature were
observed. A significant difference between ageing below and above glass transition
temperature is that the fiber-matrix interface was damage at high temperature and the
decrease in related mechanical strength was observed, while interface was well
maintained at the studied low ageing temperature. In general, for pDCPD polymer, the
mechanical properties were stable throughout the studied ageing period, indicating a
superior thermal ageing resistance of this new resin.
In Chapter 4, the microscopic transverse compression behavior of general
unidirectional fiber composites was studied using both experimental and finite element
simulation methods. This work is not directly related to above ageing studies of pDCPD
polymer but is important to understand the failure mechanisms of fiber reinforced
polymers. The main parameter studied is the failure plane angle of unidirectional fiber
composites under transverse compressive loading. A model of hexagonally packed fibers
was developed to predict the failure plane angle, and consistency was found from both
experiments and finite element simulations. For strong fiber-matrix interface case, the
failure plane angle increases with fiber volume fraction, with lower and upper bounds.
For weak fiber-matrix interface case, the failure plane angle is not sensitive to fiber
107

volume fraction but is dependent on the failure plane angle of neat resin. Other findings
include the compression strength of unidirectional composites depends on both matrix
shear strength and fiber volume fraction. Significant stress and strain concentration was
observed for high fiber volume fraction condition and the composite can behave brittle
even if the matrix is ductile.
Future work includes several aspects. First, chemical analysis and characterization of
the polymer during ageing process will provide more information about the detailed
ageing process. For example, Fourier transform infrared spectroscopy (FTIR) analysis of
the volatile gases diffused out from pDCPD polymer during thermal ageing can
distinguish the types of gas and further provide information about the chemical reactions
during ageing. Second, the fiber-matrix interface property is of significant importance to
the mechanical performance of composites, and thus needs further thorough study.
Fracture surface images from SEM are only qualitative and other method is needed for
quantitative measurement of the interface strength. Single-fiber push-out method is
considered to be the method which provides interface strength value closest to real value.
This part of work is currently on-going and the progress is attached as appendix of this
dissertation. From the fiber push-out test, the fiber-matrix interface shear strength can be
quantitatively measured and the evolution of interface strength with ageing time and
ageing condition can be obtained. These results will provide valuable information of the
interface status, which helps understanding the detailed ageing mechanisms. Finally,
ageing mechanisms other than hygrothermal ageing and thermal oxidation ageing, and
combined ageing conditions should also be studied to obtain an overall concept of the
ageing resistance for this pDCPD polymer. These ageing mechanisms include freeze-
108

thaw cycling, ultra-violet radiation, creep and etc. Because in real application like a wind
turbine blade, several ageing factors (fatigue load, sunshine, moisture, salt, day-night
temperature cycling) coexist and their effects can couple with each other. The coupling
effect can be complex and nonlinear, so by understanding all these ageing effects we can
finally achieve the goal of predicting the life of a polymer composite component.

109

APPENDIX SINGLE-FIBER PUSH-OUT TESTER
A.1 Overview
Single-fiber push-out [1-9], single-fiber pull-out [10-13], fiber fragmentation [2] and
micro-droplet tensile test are the four commonly used methods to measure the strength of
fiber-matrix interface in long fiber composites. Among these methods, fiber push-out
method provides the most realistic value because it does not require specific sample
preparation. The tested sample is just cut and polished a thin piece from real composite
sample. The basic idea of this test is to cut a thin piece (thickness ~ 10x fiber diameter) of
the unidirectional fiber composite in the 90° direction, and push a single fiber out using a
flat-end tip. Fig. A1 shows an image of this test.

Fig. A1 Fiber push-out test of glass fiber composite

110

The tip used to push the fiber is made of diamond because metals are too soft on
that scale. The diameter of the tip should just be slightly smaller than the diameter of
the fiber, because if the diameter of the tip is too small, it will break the fiber rather
than push it out. Fig. A2 shows an example using a small tip to push out relatively
large fiber.

Fig. A2 Using 4 μm tip to push glass fiber

Another important parameter is the length-to-diameter ratio of the fiber in the
sample. This ratio should not be too small to cause unrealistic result because of the
end effect and the abrasive polish influence, nor too large that the fiber cannot be
pushed out. A reasonable ratio given by previous researchers is 10. Therefore, the
final thickness of the polished sample should be strictly controlled. Also, the
observed surface should be fine polished to give a clear view of the fibers. And good
surface finish will yield accurate strength value of the interface.
111

A.2 Apparatus setup
The overall setup of the tester consists of three parts: the pusher, the sample
support and the computer and interface (National Instruments) controlling the
pusher. The setup is based on scanning electron microscope (SEM) and Fig. A3
shows the components inside the SEM chamber. The pusher consists of the diamond
tip, the load sensor, the pushing rod and the driving motor behind the rod which is
outside of the chamber.

Fig. A3 Setup inside SEM chamber

The motor is a stepper motor (McLellan) with a resolution of 0.25 μm. The motor
is powered by a two-channel power supplier (MASTECH HY3005F-3). Load sensor
is a piezoelectric force sensor (Honeywell FSG020) with a resolution of 0.4 mN.
Motor control and load data acquisition are from a computer program (LabView).
112

We prepared two diamond tips (Hysitron): a 4 μm diameter tip for carbon fiber and
10 μm diameter tip for glass fiber.
Special sample support is required because the fiber cannot be blocked during
push-out. Most commonly used support types are ceramic foam and TEM grid.
Ceramic foam is porous and the fiber locations that can be pushed out are random.
TEM grid is small but the push-out locations are regularly arranged.

A.3 Sample preparation
A thin piece (~1.5 mm) of sample is first cut off from unidirectional composite
in the 90° direction using diamond saw. Then both sides of the sample are abrasively
polished using a disk grinder on SiC sand paper through 600, 1200, 2400, 4000 grit
level. When polishing one side, the other side is glued to the grinder using hot wax.
Thickness of the sample is monitored under optical microscope during the polishing
process. After abrasive polishing to the desired thickness, one side of the sample is
polished using polishing cloth and silica solution carefully to reach fine surface
finish. Since the diameter of common carbon fiber is ~8 μm, the final thickness of
the sample is ~80 μm. Sample at such thickness is as flexible as a piece of paper,
thus extreme care should be taken when handling it. The typical diameter of glass
fiber is ~15 μm and thus the thickness of the sample should be ~150 μm.

113

Appendix References
1. Kharrat M, Chateauminois A, On the interfacial behaviour of a glass/epoxy
composite during a micro-indentation test: assessment of interfacial shear
strength using reduced indentation curves, Composites A 1997;28:39-46.
2. Zhou XF, Wagner HD, Nutt SR, Interfacial properties of polymer composites
measured by push-out and fragmentation test, Composites A 2001;32:1543-1551.
3. Battisti A, Shetty DK, Single fiber push-out characterization of interfacial
properties of hierarchical CNT-carbon fiber composites prepared by
electrophoretic deposition, Composites Science and Technology 2014;95:121-
127.
4. Belnap JD, Shetty DK, Interfacial properties of sapphire/epoxy composites:
Comparison of fluorescence spectroscopy and fiber push-in techniques, Composites
Science and Technology 2005;65:1851-1860.
5. Tandon GP, Pagano NJ, Micromechanical analysis of the fiber push-out and re-
push test, Composites Science and Technology 1998;58:1709-1725.
6. Greisel M, Jä ger J, Moosburger-Will J, Sause MGR, Mueller WM, Horn S,
Influence of residual thermal stress in carbon fiber-reinforced thermoplastic
composites on interfacial fracture toughness evaluated by cyclic single-fiber
push-out tests, Composites A 2014;66:117-127.
7. Mueller WM, Moosburger-Will J, Sause MGR, Horn S, Microscopic analysis of
single-fiber push-out tests on ceramic matrix composites performed with
Berkovich and flat-end indenter and evaluation of interfacial fracture toughness,
Journal of the European Ceramic Society 2013;33:441-451.
114

8. Kalinka G, Leistner A, Hampe A, characterisation of the fibre/matrix interface in
reinforced polymers by the push-in technique, Composites Science and
Technology 1997;51:845-851.
9. Netravali AN, Stone D, Ruoff S, Topoleski LTT, Continuous Micro-indenter
Push-through Technique for Measuring Interfacial Shear Strength of Fiber
Composites, Composites Science and Technology 1989;34:289-303.
10. Pitkethly MJ, Doble JB, Characterizing the fibre/matrix interface of carbon
fibre-reinforced composites using a single fibre pull-out test, Composites
1990;21:389-395.
11. Piggott MR, Xiong Y, Visualization of debonding of fully and partially
embedded glass fibres in epoxy resins, Composites Science and Technology
1994;52:535-540.
12. Graupner N, Fischer H, Ziegmann G, Mü ssig J, Improvement and analysis of
fibre/matrix adhesion of regenerated cellulose fibre reinforced PP-, MAPP- and
PLA-composites by the use of Eucalyptus globulus lignin, Composites B
2014;66:117-125.

115

COMPREHENSIVE REFERENCES (ALPHABETICAL)
Abaqus user manual, v6.11, Dassault Systemes (2011).

Abu-Sharkh BF, Hamid H, Degradation study of date palm fibre/polypropylene
composites in natural and artificial weathering: mechanical and thermal analysis,
Polymer Degradation and Stability 2004;85:967-973.

Agastra P, Mandell JF. Testing and Simulation of Damage Growth at Ply Drops in Wind
Turbine Blade Laminates. In: SAMPE, 2010, Paper 398.

Alessi S, Conduruta D, Pitarresi G, Dispenza C, Spadaro G, Accelerated ageing due to
moisture absorption of thermally cured epoxy resin/polyethersulphone blends. Thermal,
mechanical and morphological behavior, Polymer Degradation and Stability 2011;96:642-
648.

Arif MF, Meraghni F, Chemisky Y, Despringre N, Robert G. In situ damage mechanisms
investigation of pa66/gf30 composite: effect of relative humidity, Composites B
2014;58:487-495.

Arun KV, Basavarajappa S, Sherigara BS, Damage characterisation of glass/textile fabric
polymer hybrid composites in sea water environment, Materials and Design 2010;31:930-
939.

116

Assarar M, Scida D, Mahi AE, Poilane C, Ayad R, Influence of water ageing on mechanical
properties and damage events of two reinforced composite materials: Flax-fibres and glass-
fibres, Materials and Design 2011;32:788-795.

Awaja F, Nguyen MT, Zhang S, Arhatari B, The investigation of inner structural damage
of UV and heat degraded polymer composites using X-ray micro CT, Composites A
2011;42:408-418.

Barnes SE, Brown EC, Corrigan N, Coates PD, Harkin-Jones E, Edwards HGM, Raman
spectroscopic studies of the cure of dicyclopentadiene (DCPD), Spectrochim Acta A
2005;61:2946-2952.

Battisti A, Shetty DK, Single fiber push-out characterization of interfacial properties of
hierarchical CNT-carbon fiber composites prepared by electrophoretic deposition,
Composites Science and Technology 2014;95:121-127.

Belnap JD, Shetty DK, Interfacial properties of sapphire/epoxy composites: Comparison
of fluorescence spectroscopy and fiber push-in techniques, Composites Science and
Technology 2005;65:1851-1860.

Belnap JD, Shetty DK, Interfacial properties of sapphire/epoxy composites: Comparison
of fluorescence spectroscopy and fiber push-in techniques, Composites Science and
Technology 2005;65:1851-1860.
117

Canal LP, Segurado J, LLorca J, Failure surface of epoxy-modified fiber-reinforced
composites under transverse tension and out-of-plane shear, International Journal of
Solids and Structures 2009;46:2265-2274.

Celina MC, Review of polymer oxidation and its relationship with materials performance
and lifetime prediction, Polymer Degradation and Stability 2013;98:2419-2429.

Chateauminois A, Chabert B, Soulier JP, Vincent L, Hygrothermal ageing effects on the
static fatigue of glass/epoxy composites, Composites 1993;24:547-555.

Christensen RM, DeTeresa SJ, Failure plane orientations for transverse loading of a
unidirectional fiber composite, International Journal of Solids and Structures
2003;40:7055-7062.

Colin X, Verdu J, Strategy for studying thermal oxidation of organic matrix composites,
Composites Science and Technology 2005;65:411-419.

Collings TA, Transverse compressive behaviour of unidirectional carbon fibre reinforced
plastics, Composites 1974;5:108-116.

Correa E, Mantic V, Paris F, A micromechanical view of inter-fibre failure of composite
materials under compression transverse to the fibres, Composites Science and
118

Technology 2008;68:2010-2021.

Correa E, Mantic V, Paris F, Numerical characterisation of the fibre-matrix interface
crack growth in composites under transverse compression, Engineering Fracture
Mechanics 2008;75:4085-4103.

Daniel IM, Luo J, Schubel PM, Werner BT, Interfiber/interlaminar failure of composites
under multi-axial states of stress, Composites Science and Technology 2009;69:764-771.

DeNeve B, Shanahan MER. Water absorption by an epoxy resin and its effect on the
mechanical properties and infra-red spectra. Polymer 1993;34:5099-5105.

Dimonie D, Dimonie M, Munteanu V, Iovu H, Couve J, Abadie MJ. Nature of the first
exothermic peak on the DSC curve of linear polydicyclopentadiene obtained with a special
catalytic system, Polymer Degradation and Stability 2000;70:319-324.

Favre JP, Auvray MH, Cheneau-Henr P, Galiotis C, Vlatias C, Paipetis A, Pegoraro M,
Severini F, Di Landro L, Yuan L, Fiber/matrix mechanical interaction in carbon
fiber/bismaleimide model composites, Polymer Composites 1996;17:937-947.

Fiedler B, Hojo M, Ochiai S, Schulte K, Ando M, Failure behavior of an epoxy matrix
under different kinds of static loading, Composites Science and Technology
2001;61:1615-1624.
119

Foulc MP, Bergeret A, Ferry L, Ienny P, Crespy A, Study of hygrothermal ageing of glass
fibre reinforced PET composites, Polymer Degradation and Stability 2005;89:461-470.

Gellert EP, Turley DM. Seawater immersion ageing of glass-fibre reinforced polymer
laminates for marine applications. Composites A 1999;30:1259-1265.

Ghassemieh E, Nassehi V, Prediction of Failure and Fracture Mechanisms of Polymeric
Composites Using Finite Element Analysis, Polymer Composites 2001;22:528-554.

Gigliotti M, Grandidier JC, Lafarie-Frenot MC, Assessment of chemo-mechanical
couplings in polymer matrix materials exposed to thermo-oxidative environments at high
temperatures and under tensile loadings, Mechanics of Materials 2011;43:431-443.

Gigliotti M, Olivier L, Vu DQ, Grandidier JC, Lafarie-Frenot MC, Local shrinkage and
stress induced by thermo-oxidation in composite materials at high temperatures, Journal
of the Mechanics and Physics of Solids 2011;59:696-712.

Gigliotti M, Pannier Y, Minervino M, Lafarie-Frenot MC, Corigliano P, The effect of a
thermo-oxidative environment on the behaviour of multistable [0/90] unsymmetric
composite plates, Composite Structures 2013;106:863-872.

120

Gigliotti M, Grandidier JC, Lafarie-Frenot MC, The employment of 0/90 unsymmetric
samples for the characterization of the thermo-oxidation behaviour of composite
materials at high temperatures, Composite Structures 2011;93:2109-2119.

Gonzalez C, LLorca J, Mechanical behavior of unidirectional fiber-reinforced polymers
under transverse compression: Microscopic mechanisms and modeling, Composites
Science and Technology 2007;67:2795-2806.

Graupner N, Fischer H, Ziegmann G, Mü ssig J, Improvement and analysis of fibre/matrix
adhesion of regenerated cellulose fibre reinforced PP-, MAPP- and PLA-composites by
the use of Eucalyptus globulus lignin, Composites B 2014;66:117-125.

Greisel M, Jä ger J, Moosburger-Will J, Sause MGR, Mueller WM, Horn S, Influence of
residual thermal stress in carbon fiber-reinforced thermoplastic composites on interfacial
fracture toughness evaluated by cyclic single-fiber push-out tests, Composites A
2014;66:117-127.

Guisandez J, Tiemblo P, Gó mez-Elvira JM, Change of thermal and dynamic-mechanical
behaviour of a metallocene isotactic polypropylene during low-temperature thermo-
oxidation, Polymer Degradation and Stability 2005;87:543-553.

121

Haque MH, Upadhyaya P, Roy S, Ware T, Voit W, Lu H, The changes in flexural
properties and microstructures of carbon fiber bismaleimide composite after exposure to
a high temperature, Composite Structures 2014;108:57-64.

Hashin Z, Failure Criteria for Unidirectional Fiber Composites, Journal of Applied
Mechanics 1980;47:329-334.

Hsu SY, Vogler TJ, Kyriakides S, Inelastic behavior of an AS4/PEEK composite under
combined transverse compression and shear, International Journal of Plasticity
1999;15:783-806.

Hu Y, Kar NK, Nutt SR, Transverse compression failure of unidirectional composites,
Polymer Composites 2014.

Huang YK, Frings PH, Hennes E, Mechanical properties of Zylon/epoxy composite,
Composites B 2002;33:109-115.

Kalinka G, Leistner A, Hampe A, characterisation of the fibre/matrix interface in
reinforced polymers by the push-in technique, Composites Science and Technology
1997;51:845-851.

Kar NK, Hu Y, Barjasteh E, Nutt SR. Tension-tension fatigue of hybrid composite rods.
Composites B 2012;43:2115-2124.
122

Kellas S, Morton J, Curtis PT, The effect of hygrothermal environments upon the tensile
and compressive strengths of notched CFRP laminates, Composites 1990;42:41-62.

Kempf M, Skrabala O, Altstadt V, Acoustic emission analysis for characterisation of
damage mechanisms in fibre reinforced thermosetting polyurethane and epoxy,
Composites B 2014;56:477-483.

Kharrat M, Chateauminois A, On the interfacial behaviour of a glass/epoxy composite
during a micro-indentation test: assessment of interfacial shear strength using reduced
indentation curves, Composites A 1997;28:39-46.

Koenig JL, Emadipour H, Mechanical characterization of the interfacial strength of glass-
reinforced composites, Polymer Composites 1985;6:142-150.

Kok JMM, Meijer HEH, Deformation, yield and fracture of unidirectional composites in
transverse loading 1. Influence of fibre volume fraction and test-temperature, Composites
A 1999;30:905-916.

Kotsikos G, Evans JT, Gibson AG, Hale JM, Environmentally enhanced fatigue damage
in glass fibre reinforced composites characterised by acoustic emission. Composites A
2000;31:969-977.

123

Kravchenko OG, Li C, Strachan A, Kravchenko SG, Pipes RB. Prediction of the chemical
and thermal shrinkage in a thermoset Polymer. Composites A 66 (2014) 35-43.

Kumar S, Murthy Reddy KVVS, Kumar A, Rohini Devi G, Development and
characterization of polymer-ceramic continuous fiber reinforced functionally graded
composites for aerospace application, Aerospace Science and Technology 2013;26:185-
191.

Kuptsov AH, Zhizhin GN, Handbook of Fourier transform Raman and infrared spectra of
polymers, Elsevier 1998, ISBN 0444826203, page 155.

Lafarie-Frenot MC, Grandidier JC, Gigliotti M, Olivier L, Colin X, Verdub J, Cinquin J,
Thermo-oxidation behaviour of composite materials at high temperatures: A review of
research activities carried out within the COMEDI program, Polymer Degradation and
Stability 2010;95:965-974.

Le Gac PY, Choqueuse D, Paris M, Recher G, Zimmer C, Melot D, Durability of
polydicyclopentadiene under high temperature, high pressure and seawater (offshore oil
production conditions), Polymer Degradation and Stability 2013;98:809-817.

Liao K, Schultheisz CR, Hunston DL, Long-term environmental fatigue of pultruded
glass-fiber-reinforced composites under flexural loading, International Journal of Fatigue
1999;21:485-495.
124

Liao YT, A study of glass fiber-epoxy composite interfaces, Polymer Composites
1989;10:424-428.

McBagonluri F, Garcia K, Hayes M, Verghese KNE, Lesko JJ, Characterization of
fatigue and combined environment on durability performance of glass/vinyl ester
composite for infrastructure applications, International Journal of Fatigue 2000;22:53-64.

Mené trey Ph, Willam KJ, A triaxial failure criterion for concrete and its Generalization,
ACI Structural Journal 1995;92:311-318.

Meziere Y, Bunsell AR, Favry Y, Teissedre JC, Do AT, Large strain cyclic fatigue testing
of unidirectional carbon fibre reinforced epoxy resin, Composites A 2005;36:1627-1636.

Middleton J, The effect of ozone on polymer degradation in Polymer Core Composite
Conductors, Polymer Degradation and Stability 2013;98:436-445.

Mivehchi H, Varvani-Farahani A, The Effect of Temperature on Fatigue Strength and
Cumulative Fatigue Damage of FRP Composites, Procedia Engineering 2 2010;2011-
2020.

Mouzakis DE, Zoga H, Galiotis C, Accelerated environmental ageing study of
polyester/glass fiber reinforced composites (GFRPCs), Composites B 2008;39:467-475.
125

Mueller WM, Moosburger-Will J, Sause MGR, Horn S, Microscopic analysis of single-
fiber push-out tests on ceramic matrix composites performed with Berkovich and flat-end
indenter and evaluation of interfacial fracture toughness, Journal of the European
Ceramic Society 2013;33:441-451.

Muhlebach A, Schaaf PA, Hafner A, Setiabudi F, Thermal stability and degradation of
hydrocarbon metathesis polymers, Journal of Molecular Catalysis A: Chemical
1998;132:181-188.

Netravali AN, Stone D, Ruoff S, Topoleski LTT, Continuous Micro-indenter Push-
through Technique for Measuring Interfacial Shear Strength of Fiber Composites,
Composites Science and Technology 1989;34:289-303.

Olivier L, Ho NQ, Grandidier JC, Lafarie-Frenot MC, Characterization by ultra-micro
indentation of an oxidized epoxy polymer: Correlation with the predictions of a kinetic
model of oxidation, Polymer Degradation and Stability 2008;93:489-497.

Park SJ, Seo MK, Lee DR, Studies on the mechanical and mechanical interfacial
properties of carbon-carbon composites impregnated with an oxidation inhibitor, Carbon
2003;41:2991-3002.

Park Y, Kweon J, Choi J, Failure characteristics of carbon/BMI-Nomex sandwich joints in
126

various hygrothermal conditions, Composites B 2014;60:213-221.

Patel SR, Case SW, Durability of hygrothermally aged graphite/epoxy woven composite
under combined hygrothermal conditions, International Journal of Fatigue 2002;24:1295-
1301.

Pauchard V, Chateauminois A, Grosjean F, Odru P, In situ analysis of delayed fibre failure
within water-aged GFRP under static fatigue conditions, International Journal of Fatigue
2002;24:447-454.

Paul B, A modification of the Coulomb-Mohr theory of fracture, Journal of Applied
Mechanics 1961;06:259-268.

Piggott MR, Xiong Y, Visualization of debonding of fully and partially embedded glass
fibres in epoxy resins, Composites Science and Technology 1994;52:535-540.

Pitkethly MJ, Doble JB, Characterizing the fibre/matrix interface of carbon fibre-
reinforced composites using a single fibre pull-out test, Composites 1990;21:389-395.

Pinho ST, Davila CG, Camanho PP, Iannucci L, Robinson P, Failure models and criteria
for FRP under in-plane or three-dimensional stress states including shear non-linearity,
NASA/TM-2005-213530.

127

Poodts E, Minak G, Zucchelli A, Impact of sea-water on the quasi static and fatigue
flexural properties of GFRP, Composite Structures 2013;97:222-230.

Puck A, Schurmann H, Failure analysis of FRP laminates by means of physically based
phenomenological models, Composites Science and Technology 1998;58:1633-1662.

Quinson R, Perez J, Rink M, Pavan A, Yield criteria for amorphous glassy polymers,
Journal of Materials Science 1997;32:1371-1379.

Rao V, Drzal LT, The dependence of interfacial shear strength on matrix and lnterphase
properties, Polymer Composites 1991;12:48-56.

Richaud E, Le Gac PY, Jacques V, Thermooxidative ageing of polydicyclopentadiene in
glassy state, Polymer Degradation and Stability 2014;102:95-104.

Romanowicz M, A numerical approach for predicting the failure locus of fiber reinforced
composites under combined transverse compression and axial tension, Computational
Materials Science 2012;51:7-12.

Samborsky DD, Agastra P, Mandell JF, Fatigue Trends for Wind Blade Infusion Resins
and Fabrics, In: AIAA SDM, 2010, Wind Energy Session, Paper 2820.

128

Schieffer A, Mairea JF, Lé vê queb D, A coupled analysis of mechanical behaviour and
ageing for polymer-matrix composites, Composites Science and Technology
2002;62:543-549.

Schmitz GK, Metcalfe AG, Stress Corrosion of E-Glass Fibers, I&EC Product Research
and Development 1966;5:1-8.

Segurado J, LLorca J, A numerical approximation to the elastic properties of sphere-
reinforced composites, Journal of the Mechanics and Physics of Solids 2002;50:2107-
2121.

Tandon GP, Pagano NJ, Micromechanical analysis of the fiber push-out and re-push test,
Composites Science and Technology 1998;58:1709-1725.

Totry E, Gonzalez C, LLorca J, Failure locus of fiber-reinforced composites under
transverse compression and out-of-plane shear, Composites Science and Technology
2008;68:829-839.

Tsai SW, Wu EM, A general theory of strength for anisotropic materials, Journal of
Composite Materials 1971;5:58-80.

Vauthier E, Abry JC, Bailliez T, Chateauminois A, Interactions between hygrothermal
ageing and fatigue damage in unidirectional glass/epoxy composites, Composites Science
129

and Technology 1998;58:687-692.

Vu DQ, Gigliotti M, Lafarie-Frenot MC, Experimental characterization of thermo-
oxidation-induced shrinkage and damage in polymer–matrix composites, Composites:
Part A 2012;43:577-586.

Vu DQ, Gigliotti M, Lafarie-Frenot MC, The effect of thermo-oxidation on matrix
cracking of cross-ply [0/90] composite laminates, Composites A 2013;44:114-121.

Yee AF, Du J, Thouless MD, Toughening of epoxies, Polymer Blends, 2000.

Zenkiewicz M, Rauchfleisz M, Czupryńska J, Polański J, Karasiewicz T, Engelgard W,
Effects of electron-beam irradiation on surface oxidation of polymer composites, Applied
Surface Science 2007;253:8992-8999.

Zhou J, Lucas JP, Hygrothermal effects of epoxy resin, Polymer 1999;40:5505-5522.

Zhou XF, Wagner HD, Nutt SR, Interfacial properties of polymer composites measured
by push-out and fragmentation tests, Composites A 2001;32:1543-1551.

Abstract (if available)

Abstract Fiber reinforced polymers (FRPs) have been widely used in the aerospace industry for decades. However, the use of FRPs in civilian infrastructure applications is increasing, particularly in applications such as high-voltage transmission lines, wind turbine blades, and off-shore oil platforms. In such applications, long-term environmental ageing is a major concern because these structures, unlike aircraft, are expected to provide decades of service with minimal inspection or maintenance. This presents a major challenge for FRPs because ageing has complex effects on mechanical performance, making it difficult to forecast property retention and predict component lifetime. The overriding focus of the present work is to determine the effects of different types of ageing on the mechanical behavior of FRPs. Two FRP systems are considered, each with distinct matrix chemistry and intended service applications. The first system is a polydicyclopentadiene (pDCPD) resin with intrinsic hydrophobicity and exceptionally low viscosity, attributes that are suitable for fabrication of large parts and humid service conditions, such as wind turbine blades and components for oil and gas production. The second system is a unidirectional (UD) composite rod intended as the load-bearing core for high voltage power transmission lines (overhead conductors). ❧ For the first system, pDCPD is an attractive candidate resin for potential use in corrosive environments because of its intrinsic hydrophobicity, which imparts resistance to ageing in humid environments. We first investigate the hygrothermal ageing behavior of pDCPD composites and effects on mechanical behavior, including the fatigue behavior. Next, we investigate the thermal oxidation ageing of pDCPD resins and composites to determine effects on mechanical behavior. By investigating the ageing behavior and resulting degradation in mechanical properties, the water diffusion mechanism, salt corrosion behavior, and oxidized layer evolution of pDCPD composites are better understood. These insights provide a basis for lifetime predictions based on ageing rules and the composite can be optimized to achieve improved performance. ❧ For the second system, UD FRPs have high strength in the longitudinal direction, but low transverse strength, often resulting in premature damage initiation. The study here investigates the transverse compression behavior of UD FRPs using both experimental and finite element simulation. The investigation provides a basis for optimizing the rod design and improving the characteristics of this new product.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Polymer flow for manufacturing fiber reinforced polymer composites

PDF

Environmental effects on the hybrid glass fiber/carbon fiber composites

PDF

Hydrothermo degradation of polydicyclopentadiene

PDF

Constitutive model of concrete confined by advanced fiber composite materials and applications in seismic retrofitting

$Fatigue and fracture of pultruded composite rods developed for overhead conductor cables$

PDF

Fatigue and fracture of pultruded composite rods developed for overhead conductor cables

$On the dynamic fracture behavior of polymeric materials subjected to extreme conditions$

PDF

On the dynamic fracture behavior of polymeric materials subjected to extreme conditions

PDF

Mechanical behavior and microstructure optimization of ultrafine-grained aluminum alloys and nanocomposites

PDF

In situ process monitoring for modeling and defect mitigation in composites processing

PDF

Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites

PDF

Fiber reinforced hybrid phenolic foam

PDF

Void evolution in vacuum bag-only prepregs

PDF

Synthesis and mechanical behavior of highly nanotwinned metals

PDF

Multiwall carbon nanotubes reinforced epoxy nanocomposites

PDF

Mechanical behavior and deformation of nanotwinned metallic alloys

PDF

In situ process analysis for defect control during composites manufacturing

PDF

Defect control in vacuum bag only processing of composite prepregs

PDF

Structure and behavior of nano metallic multilayers under thermal and mechanical loading

PDF

Material and process development and optimization for efficient manufacturing of polymer composites

PDF

Chemical recycling of amine/epoxy composites at atmospheric pressure

PDF

Development of composite oriented strand board and structures

Asset Metadata

Creator Hu, Yinghui (author)

Core Title Ageing and mechanical failure of fiber reinforced polymers

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Mechanical Engineering

Publication Date 06/10/2015

Defense Date 03/02/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Environmental degradation,fatigue,fiber-matrix interface,OAI-PMH Harvest,polymer-matrix composites (PMCs)

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nutt, Steven R. (committee chair), Eliasson, Veronica (committee member), Sammis, Charles G. (committee member)

Creator Email huyinghui2008@gmail.com,yinghuih@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-571305

Unique identifier UC11300647

Identifier etd-HuYinghui-3463.pdf (filename),usctheses-c3-571305 (legacy record id)

Legacy Identifier etd-HuYinghui-3463.pdf

Dmrecord 571305

Document Type Dissertation

Format application/pdf (imt)

Rights Hu, Yinghui

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA