Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Efficient manufacturing and repair of out-of-autoclave prepreg composites
(USC Thesis Other)
Efficient manufacturing and repair of out-of-autoclave prepreg composites
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EFFICIENT MANUFACTURING AND REPAIR OF OUT-OF-AUTOCLAVE PREPREG
COMPOSITES
David B. Bender
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Chemical Engineering)
August 2021
ii
1 Acknowledgements
I would like to begin by thanking my advisor, Prof. Steve Nutt. Professor Nutt, you have given
me the knowledge, skills, and guidance to launch my career. You have challenged me with new
perspectives and provided invaluable insights on the research work I was doing. I have developed
and grown immensely during the last five years in your lab and appreciate the opportunity to have
been a part of it. I feel blessed to have started my scientific careers under your guidance.
I am also grateful to have had the opportunity to work with Dr. Timotei Centea. Tim
functioned as an unofficial second advisor and was pivotal to my career. Whenever I needed to
discuss a new research opportunity or ran into an obstacle, Tim was there to help me push ahead.
I will always remember your kind and genuine nature.
Thank you to my fellow PhD students at the M. C. Gill Composites Center for inspiring
me and encouraging me to bring my best self to work every. Special thanks to my teammates Mark
and Bill; it was a pleasure working with both of you. I want to thank Miranoush Chigani, the
brilliant and talented master’s researcher that I worked with. I also want to thank Vihan Krishnan
for his help in the design of tools for the lab. Finally, I want to acknowledge our lab manager
Yunpeng Zhang, for supporting my work.
For materials support I would like to acknowledge Guy Riddle at Tipton Goss for providing
materials that contributed to my work. It was a privilege to be able to come visit your facility and
learn from your experience in manufacturing prepregs.
I would also like to acknowledge my defense committee members for being supportive
during my studies.
And of course, a heartfelt thank you to my family. Mom and dad, you have supported my
intellectual endeavors my entire life, and I would not have made it this far without your
iii
unconditional love and support. Billy and John, thank you for believing in me and pushing me to
achieve my dreams. You are the best brothers anyone could ask for. Last, but certainly not least,
Esme, my wife, thank you for being by my side throughout this journey.
iv
2 Table of Contents
Acknowledgements ............................................................................................................................... ii
Table of Figures .................................................................................................................................... vi
Abstract ................................................................................................................................................ xi
1 CHAPTER 1: Introduction ...................................................................................................................... 1
1.1 Motivation ..................................................................................................................................... 1
1.2 Scope ............................................................................................................................................. 6
2 CHAPTER 2: Background ....................................................................................................................... 9
2.1 Experimental Methods.................................................................................................................. 9
2.1.1 Thermogravimetric Analysis (TGA): ...................................................................................... 9
2.1.2 Differential Scanning Calorimetry (DSC) ............................................................................... 9
2.1.3 Rheology ............................................................................................................................. 10
2.1.4 Dewetting ............................................................................................................................ 11
2.1.5 Double Vacuum Debulking (DVD): ...................................................................................... 11
2.1.6 Ultrasonic C-scans: .............................................................................................................. 13
2.1.7 Dynamic Mechanical Analysis (DMA).................................................................................. 14
2.1.8 Polished Section Microscopy .............................................................................................. 14
3 CHAPTER 3: Kinetic Modeling of Novel Thermoset Resins ................................................................. 16
3.1 Introduction ................................................................................................................................ 16
3.2 Methods ...................................................................................................................................... 20
3.3 Results ......................................................................................................................................... 25
3.4 Conclusion ................................................................................................................................... 33
4 CHAPTER 4: Rheology and Flow number analysis of novel thermoset resins .................................... 35
4.1 Introduction ................................................................................................................................ 35
4.2 Experimental ............................................................................................................................... 39
v
4.3 Results ......................................................................................................................................... 42
4.4 Conclusion ................................................................................................................................... 50
5 CHAPTER 5: Cure Cycle Optimization with Semi-pregs ....................................................................... 52
5.1 Introduction ................................................................................................................................ 52
5.2 Methods ...................................................................................................................................... 55
5.3 Results ......................................................................................................................................... 57
5.4 Conclusions ................................................................................................................................. 71
6 CHAPTER 6: Defect Reduction for OoA Prepregs in Repair Environments ......................................... 74
6.1 Introduction ................................................................................................................................ 74
6.2 Methods ...................................................................................................................................... 80
6.3 Results ......................................................................................................................................... 87
6.4 Conclusions ................................................................................................................................. 96
7 CHAPTER 7: In situ Analysis of Prepreg Processing in Repair Environments ...................................... 98
7.1 Introduction ................................................................................................................................ 98
7.2 Methods .................................................................................................................................... 102
7.3 Results ....................................................................................................................................... 103
7.4 Conclusions ............................................................................................................................... 112
8 CHAPTER 8: Conclusions and Future Work ....................................................................................... 114
8.1 Conclusions ............................................................................................................................... 114
8.2 Broader Implications ................................................................................................................. 117
8.3 Future Work .............................................................................................................................. 119
References ................................................................................................................................................ 122
vi
3 Table of Figures
Figure 1-1: Schematic of semi-preg formatting on a laminate surface. The gaps in the resin impart
through thickness air permeability that reduces porosity in processed laminates. .................................... 5
Figure 2-1: Schematic of a double vacuum debulking (DVD) setup over a laminate. The setup for this
procedure is complex and requires skilled technicians to operate the equipment. .................................. 12
Figure 3-1: TGA analysis of vinyl hybrid showing the mass percent remaining in the material with
increasing temperature. The resin retains most of its mass prior to cure at 105°C. .................................. 20
Figure 3-2: DSC raw data for isothermal hold showing the negative peak. The presence of the negative
peak required additional steps to be taken to gather useable data. ......................................................... 22
Figure 3-3: Example of a DSC isothermal run after accounting for the “negative peak” phenomena. Note
the starting degree of cure which is above zero. ........................................................................................ 23
Figure 3-4: DSC raw data used to perform calculations in Visual Basic. This data will be analyzed by the
code to gather to heat of reaction and degree of cure. ............................................................................. 24
Figure 3-5: Converted DSC data from the code analysis. A baseline was created and used to integrate the
heat flow, giving the heat of reaction of the resin. .................................................................................... 25
Figure 3-6: Modeling an isothermal test at 116°C with the conventional model. Major discrepancies
existed between the experimental and conventional model results. ........................................................ 26
Figure 3-7: Modeling a ramp test at 2°C/min with the conventional model. Minor systemic differences
occurred between the conventional model and experimental results. ..................................................... 27
Figure 3-8: Isotherm run with new model. The new model more closely matches data including at the
starting point at degree of cure 0.37 and final degree of cure 0.9. ............................................................ 30
Figure 3-9: Ramp run with new model. The new model does not exhibit the systemic differences like the
conventional model did. ............................................................................................................................. 31
vii
Figure 3-10: α max as a function of maximum temperature. The rapid change in α max begins at 105 °C and
quickly rises to above 0.95 by 120 °C. ......................................................................................................... 32
Figure 4-1: Semi-preg format achieved by applying dewetted film to dry fabric. The dry areas allow for
increased air evacuation in the through-thickness direction. .................................................................... 40
Figure 4-2: Viscosity comparison of epoxy and vinyl hybrid. The vinyl hybrid has a lower minimum
viscosity and sharper viscosity increase at curing temperatures. .............................................................. 42
Figure 4-3: Viscosity test with conventional model - ramp at 1°C/min. The model does not match the
initial decrease in viscosity nor the sharp increase in viscosity at the curing temperature of 105 °C. ...... 45
Figure 4-4: Viscosity test with new model – ramp at 1°C/min. The new model more closely matches the
initial viscosity decrease and sharp increase at 105 °C............................................................................... 47
Figure 4-5: Porosity of fabricated panels with high and low flow numbers. The low flow number panels
had significant flow related porosity while the high flow number panels exhibited minimal flow related
porosity. ...................................................................................................................................................... 49
Figure 5-1: Example of a typical OoA cure cycle. Note that the entire cycle takes over six hours until
laminate completion. .................................................................................................................................. 53
Figure 5-2: MATLAB simulation of cure cycle. In this plot, the maximum temperature was compared with
total cycle time to generate final degree of cure. ...................................................................................... 61
Figure 5-3: MATLAB plot of cure cycles. Normalized effective flow numbers were calculated after
varying total cycle times and mid stage hold temperatures. ....................................................................... 63
Figure 5-4: Sample test matrix. Panels 1-4 each have one or more variables removed from the cure cycle
to assess their impact on laminate quality. Panel 5 represents a fully optimized sample from a kinetic
and rheological perspective. ....................................................................................................................... 65
Figure 5-5: Panel 5 cure cycle. Note that unlike conventional cure cycles, the laminate is fabricated in
under an hour. ............................................................................................................................................ 66
viii
Figure 5-6: Porosity of fabricated panels percent porosity. Panels 2 and 5 had the lowest porosity other
than the control panel in line with them having both a mid-stage hold and debulking applied. .............. 67
Figure 5-7: Porosity of fabricated panel micrographs. Despite relatively small differences in the panel
cure cycles, the resulting laminate qualities were drastically different. .................................................... 68
Figure 5-8: Glass transition temperatures of fabricated. Panel two was the single panel that was
predicted to reach full cure and had a lower glass transition temperature as a result. ............................ 69
Figure 5-9: Interlaminar shear strength of panels. The strength of the panels was inversely proportional
to the amount of porosity in the laminates. ............................................................................................... 70
Figure 6-1 - Schematic of scarf and stepped repairs with over-plies. Stepped repairs allow for repair plies
to be inserted into the parent material but can be difficult to align. Scarfed repairs are more difficult to
machine but can offer additional layup flexibility and increase strength retention in the composite part.
.................................................................................................................................................................... 75
Figure 6-2 - Schematic of the impairment of gas edge breathing caused by impermeable bondlines in a
laminate repair patch. ................................................................................................................................. 78
Figure 6-3: a) Semi-preg formatting on prepreg surface with islands of resin b) sealed edges of test
panels that simulate the sealed edges of patch repairs ............................................................................. 82
Figure 6-4: Ultrasound images of Panels A and C (semi-preg and DVD). Panel A shows spatially consistent
void free areas while Panel C has an inconsistent interior. ........................................................................ 87
Figure 6-5 - Micrograph of DVD (Panel C) and semi-preg (Panel A) polished sections. The DVD panels
exhibited porosity while the semi-preg panels were void-free. ................................................................. 88
Figure 6-6: Microscopy of scarf panels. The DVD panel (Panel D) exhibited porosity while the semi-preg
panel (Panel B) was void-free. .................................................................................................................... 90
ix
Figure 6-7: Glass transition temperatures for Panels A (semi-preg) and C (DVD). The epoxy had a higher
glass transition temperature than the vinyl hybrid resin making it more viable for higher temperature
applications. ................................................................................................................................................ 91
Figure 6-8: Interlaminar shear strength for Panels A (semi-preg) and C (DVD). As expected, the epoxy had
a higher interlaminar shear strength than the vinyl hybrid resin. .............................................................. 92
Figure 6-9: Micrographs of fabricated panels. The panels fabricated with semi-preg formatting exhibited
fewer macro- and microvoids than those with conventional formatting. ................................................. 93
Figure 6-10: Quality of fabricated panels shown by percent porosity. The semi-preg formatted panels
had significantly lower porosity than the conventionally formatted panels. ............................................. 94
Figure 6-11: Short beam strength of Panels 1-4. The panels demonstrate the inverse correlation between
the void content and strength of laminates. .............................................................................................. 95
Figure 7-1: Schematic of repair tool plate with an in-situ observation window. A 3° scarf angle was
incorporated to follow best practices in repair procedures. .................................................................... 102
Figure 7-2: Surface images of DVD and semi-preg panels. The semi-preg Panel A do not exhibit the same
level of surface porosity as the DVD panel C. ........................................................................................... 104
Figure 7-3: In-situ images of DVD processing. Void formation and migration on the surface of the
laminate is seen in these images. ............................................................................................................. 105
Figure 7-4: In-situ images of semi-preg processing. Fewer voids remain after resin consolidation and are
able to escape through the laminate. ....................................................................................................... 107
Figure 7-5: Images acquired in-situ during processing (Panel 1). Any bubbles remaining after
consolidation were evacuated through the thickness of the laminate. ................................................... 109
Figure 7-6: Images acquired in-situ during processing (Panel 3). Bubbles that remain after consolidation
are not able to evacuate through the thickness of the laminate and remain as surface porosity. ......... 110
x
Figure 7-7: Images acquired in-situ during processing (Panel 2). Many bubbles remain after laminate
consolidation and remain as surface porosity despite the lower viscosity of the vinyl hybrid resin. ...... 111
xi
4 Abstract
This thesis is focused on the development of materials and models for the next generation of
out-of-autoclave (OoA) prepregs that reduce cycle times, ensure robust laminate quality, and can
be employed in the field for repairs.
The kinetic behavior of a novel resin archetype was defined with a new autocatalytic model
that accounts for features such as cure inhibition, fast-cure, and incomplete cure. A maximum
degree of cure term that depends on the curing temperature of the inhibited resin was incorporated
into the model. The model was shown to match experimental results closer than a conventional
autocatalytic cure equation.
A revised gel viscosity equation with a resin curing temperature term was created to describe
the rheology of this material. Resin elements such as rapid gelation and reduced viscosity below
the curing temperature were described by the model. The revised model was shown to predict
viscosity behavior more accurately than the conventional gel model. The rheology model was
employed to assess the predictive capability of effective flow number in composite laminates. It
was shown that different prepreg materials with the same effective flow number reach similar
levels of resin saturation and flow-based porosity.
The revised gel model and novel autocatalytic kinetic model were employed in a cure cycle
process optimization method to reduce cycle times without reducing laminate quality. Using the
optimization method, cure cycles of less than an hour were created. Panels that were fabricated
with the optimized cure cycles reached full cure and saturation. It was shown that air evacuation
has the greatest effect on the number of void defects in a laminate, with mid-stage hold having a
lesser impact.
xii
An in-field repair solution was presented featuring semi-preg formatting and a room
temperature stable resin. The performance and quality of the room temperature stable semi-preg
was compared to conventionally formatted epoxy prepregs and wet laid panels processed with
double vacuum debulking (DVD). The semi-preg exhibited significantly lower porosity than both
the conventional prepregs and the wet laid materials when processed in an in-field repair
environment.
An in-situ observation tool was created to generate insights on the cure processes of scarf
repair patches. The tool featured an observation window that allowed for video recordings of
panels undergoing cure. The in-field repair solution, conventional epoxy prepreg, and DVD wet
layup materials were examined using the scarf repair tool. Videos taken showed how gas bubbles
formed and remained in the wet laid DVD materials and conventional prepregs as void defects but
were able to be evacuated with the semi-preg formatting.
Overall, the work presented here offers insights on the processing of prepreg materials. The
models, optimization methods, and experimental results can provide an understanding of OoA
processing that can lead to reduced cycle times, improved laminate quality, and increased
flexibility for in-field repairs.
1 CHAPTER 1: Introduction
1.1 Motivation
Advanced composite materials for high performance applications are composed of a
continuous fiber reinforcement with a polymeric resin matrix [1]. Composites exhibit high
strength and stiffness compared to equivalent materials, but with a lower weight than equivalent
materials in other classes, such as metals [2]. They also have greater resistance to corrosion than
metals by virtue of resin matrices and can add flexibility to the design space during
manufacturing [3], [4]. Their unique combination of strength/stiffness to weight ratio have made
composites useful for decreasing the weight of moving vehicles [5]. Military applications of
composites began in the 1960s and quickly became the standard material for military aircraft.
Conversely, it was not until the release of the Boeing Dreamliner that a majority-composite
aircraft was employed in the commercial space. Composites became the material of choice for
race cars after McLaren revealed the MP4/1 in the 1980s. Composites are currently being used in
other racing and luxury cars. High performance applications in both automotive and aerospace
industries represented the first use cases for composites [6]. Carbon fiber thermoset composites
represent the largest use case of composites in high performance applications [7]. The limited
introduction of composites into widespread use can be explained by their cost [8]. High
performance composites are expensive to produce. Carbon fiber and performance thermoset
resins both have high prices per pound, and processing composite laminates in autoclaves further
increases the price of the final product. For high performance applications like race cars and
fighter jets, those costs are easily justified compared to more common applications of structural
2
materials. To increase the usage of high-performance materials, the cost of producing and
maintaining these composite structures needs to decrease.
As the leading mature use case for composite materials, the aerospace industry has been
responsible for standards used across other industries [9]. The military was the earliest adopter of
composite usage in aircraft and have developed procedures towards the maintenance and repair
of its structures [10]. Many of those procedures were transferred over into the commercial sector
as composites became commonplace in structural components. Different operating conditions
exist for commercial aircraft versus military aircraft. Mission-ready aircraft generally fly 500
hours per year while commercial aircraft can fly upwards of 3000 hours per year [11]. As a
result, military aircraft spend more time on the ground, allowing more opportunities for
inspection and repair. Therefore, the military can create procedures that work for military planes,
under assumption they spend more time on the ground than in the air. These procedures optimize
for the performance of the aircraft during the limited usage. While downtime is surely a concern
for military sectors, it is not seen as a critical parameter to minimize when compared to
performance [12]. Commercial aircraft face different pressures during their operation. Cost
models that incorporate downtime have been created to optimize profits for planes [13].
Reducing downtime as a means of increasing revenues brings up two aspects of composite repair
and maintenance to be considered. First, the pursuit of downtime reduction emphasizes the need
for organizations like the FAA to maintain adequate regulations for the ongoing safety of air
travel, so that maintenance is not improperly overlooked in the pursuit of profits [42]. Boeing’s
ongoing MAX saga can serve as a reminder to operators that fail to maintain adequate safety
controls [16]. Secondly, research into technology to reduce this downtime can be incentivized by
these cost models. In summary, process optimizations for both manufacturing and maintenance
3
can bring down the cost of employing composite materials in aerospace and promote their use
across additional industries.
When composites were first introduced, methods of repair did not exist. Parts damaged were
simply replaced, creating significant amounts of waste and increasing downtime for damaged
equipment [17]. As a result, repair procedures were developed to attempt to restore strength to
the laminates. However, many of these procedures were based off of experience with other
materials such as metals [15]. Metals are isotropic while composite materials are anisotropic,
leading to different behaviors under the same loading conditions [18]. Repair techniques like
doublers were attempted on composite structures with varying levels of success [19]. Bonded
repair methods were developed to restore strength in laminates without disrupting aerodynamics
of aircraft [20], [21]. Although it took years to be accepted, bonded repairs are now being
incorporated into repair guidelines in composite structures. Thus, research into the design and
processing of repairs has intensified. The procedures required for repairing composite parts differ
than that of traditional composite manufacturing but have many similar challenges for quality
control.
The expense of manufacturing and maintaining composites structures traditionally through
autoclaves has limited their use. Out-of-autoclave (OoA) prepregs offer an opportunity to reduce
the cost and increase throughput of manufacturing composite structures. OoA composites have
defect issues due to difficult air removal, especially with large or complex structures. Conventional
OoA composites rely on engineering evacuation channels (EVaCs) in the prepreg for gas egress.
While under vacuum, air flows in the plane of the laminate out to the edge of the panel, where it
egresses through edge breathing dams. Thus, laminates are held under vacuum at room temperature
for long periods, in a process known as debulking. In large structures, the distance air needs to
4
travel to evacuate causes processing challenges. Furthermore, complex structures with features
such as ply drops introduce barriers to the EVaCs which inhibit gas egress. The time it can take to
attempt to fully evacuate prepregs through debulking can be prohibitive. Reducing the time it takes
to fully evacuate air can lead to a great cost reduction for OoA prepregs and lead to increased
usage. Defects are especially prevalent when processing OoA prepregs in repair environments,
where methods such as double vacuum debulking (DVD) have been employed to attempt to reduce
porosity in the final laminates. However, due to repair conditions, DVD processing has not been
able to fully alleviate porosity concerns in repairs. In addition, processing standard epoxies
requires freezer storage, which is often not present in the field. While techniques such as wet layup
do not require freezer storage, defects are even more prevalent with this method despite DVD
processing. Using resins that can be stored at room temperature can obviate the need for freezer
storage or wet layup methods. Furthermore, using materials that can be processed without a DVD
apparatus can improve reliability and reduce the cost of repairing materials.
A method of improving air evacuation by virtue of through thickness air evacuation has been
introduced in a resin formatting scheme, referred to as semi-preg [22]. Semi-pregs have partially
impregnated discontinuous resin patterns on the fiber beds, expanding the degrees of freedom for
the EVaCs (Figure 1-1).
5
Figure 1-1: Schematic of semi-preg formatting on a laminate surface. The gaps in the resin impart through thickness
air permeability that reduces porosity in processed laminates.
Unlike conventional prepregs, semi-pregs do not rely on in plane air evacuation for gas
egress. They have been demonstrated to reduce porosity in a variety of imperfect processing
conditions [22]. Work has been done to optimize the resin format architecture based on air
permeability [23]. These advances can enable processing prepregs in repair environments to
become much less complex [24]. With fewer complexities in processing, the need for skilled
technicians becomes less pressing. With current operating methods (wet layup, DVD) for repairs,
the difference between a high-quality repair and a low-quality one can depend on human error
and experience, placing emphasis on the skill of the technicians. However, as material
6
developments continue in the direction of robust, high-quality, and simple processing, the need
for depot level facilities will decrease. In addition, the advent of semi-preg formatting can allow
for reduced debulking times. In many OoA processes, debulking can be the longest step in a cure
cycle, especially for large or complex parts. By introducing another degree of freedom to air
egress, the effective distance air needs to travel to evacuate is significantly lower. It also allows
for complex features such as ply drops to be bypassed in the air escape pathways. When paired
with fast-curing and/or low viscosity resins, the time required for many of the processing steps
needed for high quality laminate can be reduced.
1.2 Scope
This thesis covers experimental and modeling work to study innovative materials and
processing methods. Models were created or revised to predict the behavior of a new class of
resins. A resin with inhibited cure at lower temperatures, with fast-cure at higher temperatures,
and with low viscosity prior to gelation was modeled and employed in experiments. A prepreg
with this resin and semi-preg formatting was created to present an in-field repair solution. These
models were employed to create optimized cure cycles which informed the experimental
endeavors. Finally, a new tool to provide in-situ observation on the curing process for
composites was created and used to generate insights on the effectiveness of semi-preg
formatting.
Chapter 2 covers the experimental methods and procedures used in the analysis of composite
laminates fabricated in other chapters. Procedures covered include thermogravimetric analysis
(TGA), differential scanning calorimetry (DSC), rheology, resin dewetting, double vacuum
7
debulking (DVD), ultrasound analysis, dynamic mechanical analysis (DMA), and polished
section microscopy.
Chapter 3 covers kinetic modeling of thermoset resins with unusual cure features. The
features accounted for in the novel model include inhibited cure, incomplete cure, and fast-cure.
An equation for a maximum degree of cure term based on the curing temperature of the resin
was created. The predictive capability of the new fast-cure (FC) model was compared to a
conventional diffusion limited thermoset model. In addition, methods for recovering useable
DSC data from fast-cure resins and automating kinetic modeling were introduced.
Chapter 4 covers rheology modeling of thermoset resins with unique properties. A gel model
to describe viscosity of thermosets was revised to account for cure inhibition, low viscosity, and
fast gelation. The predictive capability of the revised model was compared to the conventional
gel model in effective flow number analysis. The efficacy of using effective flow number to
determine flow based void defects was analyzed. Laminates with different materials were used to
assess the predictive capability of flow number analysis.
Chapter 5 covers process optimization of thermoset materials. In particular, the material
modeled in chapters 3 and 4 was used due to its unique properties that enabled significant cure
cycle time reductions. A method of efficiently optimizing cure cycles based on kinetic and
rheology models was created. It allowed for a dynamic range of cure cycle parameters to be
altered, and the effects of those alterations on degree of cure and flow to be quickly assessed.
Using the models from chapter 3 and 4 with the optimization procedures created, laminates were
processed with cure cycles of less than an hour. The quality and performance of those laminates
were assessed.
8
Chapter 6 covers defect reduction for out-of-autoclave (OoA) composites in repair
environments. A material solution based on a room temperature stable resin and semi-preg
formatting was presented. The performance and quality of the new material was assessed against
conventional materials and processing methods. Prepregs with conventional formatting and
epoxy resins were compared to the novel semi-preg. In field repair techniques of DVD and wet
layup with two-part epoxies were also contrasted against the new material system.
Chapter 7 covers the use of a novel in-situ observation tool for composite repairs. The tool
allowed for videos to be taken of the cure process on a repair relevant scarfed surface. The
comparisons in chapter 6 were able to be analyzed with the tool. Insights on void formation and
reduction through semi-preg formatting were gained by using the tool.
Chapter 8 highlights the contributions of this work and key takeaways from the results. The
broader impact of these results is discussed. Finally, recommendations for future work related to
the content of this thesis are presented.
9
2 CHAPTER 2: Background
2.1 Experimental Methods
The key experimental techniques used to employ repairs and analyze composites will be
described in this section in sequential order. First, methods available for the characterization of a
composite material using kinetic analysis will be described - DSC and TGA. Second, methods to
investigate material viscoelastic properties will be examined in the Rheology section. Third,
dewetting and double vacuum debulking methods to process samples will be explored. Lastly,
methods to analyze material defects will be elaborated.
2.1.1 Thermogravimetric Analysis (TGA)
Thermogravimetry analysis is a method used to measure volatile content and investigate
the degradation behavior of materials. TGA determines the change in weight of a substance as a
function of time and temperature. In a TGA experiment, a sample is heated in a controlled time-
temperature environment. As the temperature rises, the sample’s weight loss is recorded as a
function of time and temperature. TGA is used to measure the release of volatiles from a resin
during cure. Given volatiles can create voids in composites laminates, knowing when they evolve
is critical for robust processing procedures.
2.1.2 Differential Scanning Calorimetry (DSC)
DSC measurements provide data as a function of time and temperature regarding
transitions in materials that involve endothermic or exothermic processes. In a DSC experiment, a
small sample is put into an isolated chamber and subjected to variable heat treatments. The heat
10
absorbed and/or released from the sample is measured and recorded. DSC results can be used to
analyze the kinetic behavior of resins. For polymeric resins, heat gets released as the resin
undergoes crosslinking. To assess the heat released, integration under the DSC data curves is done.
The total amount of heat released from a sample is measured to identify the amount of exotherm
per unit mass a particular resin has at full cure. Thus, in other DSC runs on the same material, it is
possible to assess the degree of cure by integrating the heat released at any given point and dividing
it by the total heat released for the same material. The degree of cure can then be modeled and
used to create optimal cure cycles based on the kinetics of the material.
2.1.3 Rheology
Rheology is the science of the deformation and flow of material. Under deformation, most
materials display viscoelastic behavior, acting partially as a solid and partially as a liquid. The
levels of viscous and elastic response vary based on stress, time, and temperature. A purely elastic
material (ideal solid) has a phase angle of 0°, while a purely viscous material (ideal liquid) has a
phase angle of 90°. Most thermoset materials display both viscous and elastic behavior as
viscoelastic materials. The complex modulus, which is separated into storage (elastic) modulus
and loss (viscous) modulus, and damping behavior are all calculated from stress, strain, and phase
angle. Complex viscosity is determined by incorporating the oscillation frequency, ω in rad/sec. A
rheometer is used to measure the viscoelastic behavior of a resin at different temperature
environments. A sample of resin is placed in between two parallel plates and subjected to variable
stresses and strains from the plates. The type of test preferred for these materials is the dynamic
test, which allows for the measurement of the storage and loss modulus. From that test, the storage
and loss moduli can be analyzed and used to determine viscoelastic behavior of the materials.
11
2.1.4 Dewetting
To create through thickness gas permeability on composite laminates, methods of creating
gaps in the resin layer need to be employed. Dewetting is one such approach, where the tendency
for resins to dewet when subjected to heat are used to create resin patterns on the surface of the
prepreg. This is needed to create semi-pregs. There are three main steps for creating dewetted
semi-preg materials. The first step involves creating nucleation sites on a thin resin film on backing
paper. The location and density of nucleation sites is important to control as they determine how
the resin will dewet when subjected to heat. A spike roller is a common way to introduce regular
patterns of nucleation sites to a resin film. The second step involves heating the resin film. The
temperature and amount of time inside an oven are critical for getting the correct pattern of resin
for any given material. Furthermore, creating the patterns with minimal crosslinking of the resin
is desired. For most resins, leaving the nucleated resin film longer in an oven or at higher
temperatures leads to more dewetting, and thus a drier surface area in the resulting semi-preg.
Finally, the third step is to apply the newly dewetted resin film to the fiber reinforcement. A heated
press can be used to put the two components together. Care must be taken such that the patterns
are not lost during the press process.
2.1.5 Double Vacuum Debulking (DVD)
One technique commonly employed for composite repairs is double vacuum debulking
(DVD), although additional complexity is associated with the setup. With double vacuum
debulking, a vacuum bag only (VBO) cure setup (release layer, laminate plies, breather, and
12
bagging with a vacuum outlet) is initially set up. A rigid closed structure is placed around the
VBO setup and bagged with a separate vacuum outlet (Figure 2-1).
Figure 2-1: Schematic of a double vacuum debulking (DVD) setup over a laminate. The setup for this procedure is
complex and requires skilled technicians to operate the equipment.
A vacuum is then applied to the outer structure, while leaving the inner vacuum off.
Thus, the compaction pressure normally associated with being under vacuum is no longer
applied to the laminate, but instead to the outer structure. The laminate consequently avoids
consolidation while under vacuum, allowing the gas in the laminate to escape more easily. After
debulking under the structure, the remainder of the cure is performed with a heat blanket
according to VBO guidelines. The DVD approach reduces void defects in laminates when
compared to equivalent VBO processed prepregs [25], [26]. While effective, DVD does not
always result in void free parts, especially in repair conditions [27]. The attempt to use
evacuation channels (EVaCs) to provide gas egress through an impermeable bondline remains a
hurdle to overcome in reducing porosity, despite the additional measure in DVD. Alternative
methods of robust repair processing for composites are desired.
13
2.1.6 Ultrasonic C-scans
Ultrasonic inspection is the most common contact NDE method employed for detecting
damages in composite laminates [78]–[80]. This method involves transmitting high frequency
sound (1 to 5 Mhz) into the part and monitoring received energy [32]. Transducers require a
medium in between the transducer and the part that will not cause much noise. Immersion
transducers are not in direct contact with the laminate and are submerged in water. Contact
transducers, as the name implies, do contact the sample. A thin layer of oil or of other couplant is
used for noise free transmission. The need for a couplant, water, or oil, is a downside of this
process, as it introduces the tested laminate to potential contamination or moisture. Ultrasound can
be conducted relatively quickly and can provide high resolution feedback on the presence of
damage in panels [33]. Pulse echo and through transmission are two types of transducer based
ultrasonic techniques [34]. The difference between the two is the location of the receiver. In pulse
echo transmission, the receiver is located on the same side of the part as the transducer and reads
the echo. In through transmission, the receiver is on the other side of the laminate. Recent efforts
into asymmetric frequency ultrasonic NDE has shown to be effective at locating defects [35]. An
alternative form of ultrasonic inspection is using lasers to generate ultrasonic waves at the test
surface [36]. Short pulses are used to generate the waves, while long pulses are used to detect
them. The laser form has the benefits of being non-contact, only requiring access to one side, and
not requiring a couplant. In this work, ultrasonic inspection is used to identify internal sources of
porosity in fabricated panels.
14
2.1.7 Dynamic Mechanical Analysis (DMA)
Dynamic mechanical analysis (DMA) is generally a more sensitive technique for detecting
transitions than the DSC method. In DMA, a sample of cured material is placed on a single or dual
cantilevered sample holder and subjected to forces at varying temperatures. The sensitivity
increase in DMA over DSC is because the properties measured are the dynamic modulus and
damping coefficient, which both change significantly when crystalline structures transition into
amorphous structures. The operating principle is that in these transitions, a proportionally larger
change takes place in the mechanical properties of a polymer than in its specific heat. Therefore,
dynamic mechanical analysis is the preferred method of measurement for glass transition
temperature and other minor phase/structure changes of polymers. The tan delta method was used
in this work to identify the glass transition temperature of the materials employed in the studies.
Tan delta is the ratio between the in-phase component of storage modulus to the out-of-phase
component of loss modulus. This ratio has a peak at a midpoint between the glassy and rubbery
states of the material. Thus, the temperature at which the material peaks is the temperature used as
a glass transition temperature.
2.1.8 Polished Section Microscopy
To determine the porosity content in the interior of composite laminates, polished section
microscopy was employed. It is a destructive technique which makes it unable to be employed
on structures in active use. However, for research, it gives insights on the location, distribution,
size, and amount of porosity in a composites laminate. First, the laminate is cut using a diamond
saw to create a smooth exposed sample interior. The sample is then mounted into a form that
makes it possible for polishing to occur. This can be done using mounting resin or with a light
15
wood such as balsa. The sample is then subjected to abrasive grinding to begin polishing the
surface. Silicon carbide paper at grits ranging from 240 to 4000 are used at increasing intervals
to polish the surface. Checking the sample for polishing artifacts is vital for a clean surface for
imaging. After abrasive polishing, the sample then polished with an alumina slurry on a cloth
surface. When polishing is finished, the sample is imaged on a digital microscope. The resulting
images can be used to determine porosity content. To get the percent void content, dark/light
thresholding is used to separate void content from laminate components. The percent porosity is
calculated by taking the void content area over the entire area of the laminate imaged.
16
3 CHAPTER 3: Kinetic Modeling of Novel Thermoset Resins
3.1 Introduction
The objective of the present work was to modify standard cure models to accurately predict
cure kinetics of a fast cure resin for prepregs that allow for fast cure. The new model was developed
to describe the kinetic behavior of the fast-curing resin, because conventional models do not
accurately predict the cure behavior.
Epoxies are widely used resins in the aerospace industry, and the cure kinetics are well-
characterized [37]. Cure cycles for VBO epoxy prepregs often last several hours. After
debulking, these cure cycles typically include an intermediate temperature dwell to allow resin to
flow and fully impregnate the composite. This stage is followed by a high-temperature dwell to
fully cure the laminate. The cure kinetics of the epoxy employed dictate the duration of the cure
step. Epoxy cure releases heat (exotherm), and thus ramp rates are limited to mitigate the risk of
damage from overheating [31]. Overheating can become uncontrolled and damage the processing
setup ranging from the bagging materials to the reinforcement material. Reducing the risk of
overheating, requiring lengthy intermediate dwell times, and the relatively slow crosslinking
process of epoxies contribute to longer cure cycles. Although the aerospace industry employs
long cycle times, the automotive industry requires much faster cure cycles to achieve high
production rates [38]. High performance applications such as military uses prioritize quality and
performance over production speed. While performance requirements for automotive structures
are less demanding than aerospace materials, composites capable of being employed in these and
other industries often require cure cycles of minutes as opposed to hours [39]. Deployment of
prepregs in industries with high production volumes may require resins formulated for fast cure
17
cycles, as well as prepregs suited to VBO curing. Fast curing requires an alternate cure chemistry
distinct from conventional resins, such as epoxies. The alternate resin chemistries need to
address the three factors responsible for long required cure cycles. The resin needs to have a low
exotherm to reduce the risk of overheating. It needs to have a rapid crosslinking process when
desired curing temperatures are met. Finally, the resin needs to potentially have lower viscosity
to reduce saturation times.
As thermoset resins employed in composites continue to become more complex, modeling
their behavior has required additional terms to be included [40]. The first model used was a
single nth order reaction expression shown below.
𝑑𝛼 𝑑𝑡 = 𝐴 ∗ exp (
− 𝐸 𝑅𝑇
) ( 1 − 𝛼 )
𝑛
The assumption behind this model is that at any given point, the cure rate of a material can
be accurately given by knowing the current degree of cure and temperature of the system. This
assumption is carried forward in future models as well. The expression has an Arrhenius
temperature dependence an nth degree reaction term to model thermoset behavior. This
expression falls short of accurately describing autocatalytic material behavior. In the expression,
the fastest rate of cure occurs when alpha is at 0, which is not the case for autocatalytic material.
Thus a model was described by Kamal et al that employed autocatalytic terms to model a single
epoxy cure reaction [41]. The expression below is the Kamal model.
𝑑 𝛼 𝑑𝑡 = ( 𝑘 1
+ 𝑘 2
𝛼 𝑚 ) ( 1 − 𝛼 )
𝑛
Where ki is given by the Arrhenius term for i = 1,2
18
𝑘 i
= 𝐴 𝑖 exp (
− 𝐸 𝑖 𝑅𝑇
)
The Kamal model introduces a second Arrhenius term with an 𝛼 𝑚 alpha term instead of
just a ( 1 − 𝛼 )
𝑛 term. This ensures that the model does not predict the fastest rate of cure when
alpha is at 0, making it more reflective of autocatalytic reactions such as those in epoxies. The
model however, does not account for physical effects such as vitrification or diffusion limitations
that alter the rate of cross linking. A second Arrhenius term is not always effective in
representing that phenomena. A term was added by Chern and Poehlein that accounts for
diffusion limitations on degree of cure. The rate constant is altered as shown below.
𝑘 d
= 𝑘 𝑐 exp ( − C ( 𝛼 − 𝛼 𝑐 ) )
In the expression C is a constant and 𝑘 𝑐 is the rate constant for the reaction. The value of
𝛼 𝑐 represents a value where the diffusion becomes the dominant factor controlling the rate of
reaction. However, there are periods where both the kinetic and diffusion factors control the
reaction, leading to an effective term defined by both 𝑘 d
and 𝑘 𝑐 to represent that region.
1
𝑘 𝑒 =
1
𝑘 𝑐 +
1
𝑘 𝑑
A term developed by Khanna and Chanda applies the expression to get a diffusion factor
𝑓 ( 𝛼 ) [42].
𝑓 ( 𝛼 ) =
1
1 + exp ( 𝐶 ( 𝛼 − 𝛼 𝑐 ) )
19
The term 𝑓 ( 𝛼 ) is multiplied to the autocatalytic expressions to account for diffusion
limitations during the cure of resins. The term 𝛼 𝑐 is determined empirically from the system.
However, Hubert observed that there is a linear relationship between the glass transition
temperature and ultimate degree of cure for a material [43]. The terms from that relationship are
𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇 𝑔 which can thus be substituted instead of 𝛼 𝑐 into the kinetic models.
𝛼 𝑐 = 𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇
The kinetic models that were developed to include these diffusion terms have been used to
successfully model many thermoset resins employed today. However, as resin development
continues, features that are unaccounted for in standard models are being introduced. In
particular, room temperature stability is a desired resin trait. Room temperature stability
obviates the need for freezer storage for prepregs. In-field conditions, such as in repair
processing environments, often do not have the ability for freezer storage. Thus, resins that have
inhibited cure kinetics are becoming increasingly common. Modeling these materials with the
conventional models is unfeasible. The conventional models describe materials with
autocatalytic curing that starts at relatively low temperatures. Inhibition that keeps the degree of
cure close to zero until a threshold temperature is reach is not compatible with the conventional
models. Furthermore, resins that cure quickly exacerbate the problems the conventional models
have in describing these resins.
This study addresses the limitations of conventional kinetic models to describe novel fast
curing thermoset resin behavior with room temperature stability. Current thermoset models are
unable to account for room temperature stability and its effect on cure behavior. New terms are
introduced to a conventional autocatalytic model to describe the unusual behavior of the resin. A
20
methodology to work with fast curing resins using DSC and reduce variation in baselining error
was created.
3.2 Methods
A fast-cure vinyl hybrid resin with characteristics suitable for the present study was
selected [44]. The resin, originally designed for resin infusion, featured a vinyl ester base but
differed from conventional vinyl esters in that styrene was not required to form crosslinks. The
absence of styrene reduced volatiles normally released during cure. The reduced release of
volatiles was confirmed by thermogravimetric analysis (TGA, TA Instruments Q5000).
Figure 3-1: TGA analysis of vinyl hybrid showing the mass percent remaining in the material with increasing
temperature. The resin retains most of its mass prior to cure at 105°C.
21
Prior to the cure of the vinyl hybrid resin at 105°C, <1% of mass loss occurred. This is in
contrast to styrene-based resins which have a greater mass loss prior to curing temperatures.
Evolved gases during cure can remain in the laminate as void defects. Thus, reducing mass loss by
avoiding a styrene-based chemistry is critical for ensuring robust quality control.
The vinyl hybrid resin relies on an unusual cure chemistry that features reaction inhibition
before the cure temperature is reached. The inhibition allows the resin to remain at low viscosity
prior to gelation and imparts room temperature stability, obviating the need for cold storage. Room
temperature stability thus imparts cost savings and convenience, particularly for applications such
as in-field repair. The rapid cure kinetics of the resin requires modification to conventional cure
kinetics models, as described in the modeling section.
The vinyl hybrid reaction kinetics were analyzed using dynamic scanning calorimetry
(DSC, TA Instruments Q2000), to measure heat flow data as a function of time and temperature.
Neat resin samples (10 mg) were employed. Test conditions included five dynamic ramps (1 – 10
°C/min) up to 200 °C and twelve isothermal holds (93 °C – 163 °C). Because of the rapid cure
reaction, the instrument did not capture the entirety of the exotherm. The result was a negative
peak that resulted from supplying heat to the pan while the resin underwent the exotherm. The
negative peak from instrument heating was thus convoluted with the peak from the exotherm. An
example of the phenomenon is shown in Figure 3-2.
22
Figure 3-2: DSC raw data for isothermal hold showing the negative peak. The presence of the negative peak
required additional steps to be taken to gather useable data.
To address the “negative peak” phenomenon, additional steps were taken to compensate
and to retain the information needed to generate an accurate cure model. First, the total heat of
reaction was measured for all the ramp runs, and the average was then used as the baseline. At
the end of every isothermal run, a ramp test was immediately performed on the isothermal
sample to measure the remaining heat yet to be released. The excess heat released from the ramp
run was subtracted from the total heat of reaction, yielding the heat released from the isothermal
run. The results from this procedure allowed for analysis to begin immediately after the
convolution from the “negative peak” phenomena was resolved. This resulted in data as seen
below.
23
Figure 3-3: Example of a DSC isothermal run after accounting for the “negative peak” phenomena. Note the
starting degree of cure which is above zero.
Code was written in Visual Basic and used to analyze the raw data for both the isothermal
and ramp runs. With a resin that cures as quickly as vinyl hybrid, keeping the process for
calculating the heat released by the material as consistent as possible is imperative to develop an
accurate model. The code began by starting with the dataset given from the DSC after being
converted into Excel form.
24
Figure 3-4: DSC raw data used to perform calculations in Visual Basic. This data will be analyzed by the code to
gather to heat of reaction and degree of cure.
A baseline was then selected by the code to be used to integrate the heat flow curve.
Although baselining is often done manually, automating this step with this material ensures
consistency across different trials and prevents human variation from causing inaccuracies in the
model. When the resin cures as quickly as vinyl hybrid does, this was necessary to reconciliate
the data from both isothermal and ramp DSC runs. Once the baseline is created the needed
integrations are calculated. This is done by first converting the heat flow into a format able to be
integrated using the baseline. For the ramp runs, this is just the single curve as seen in Figure
3-4. For the isothermal runs this is both the initial isothermal run as well as the secondary ramp
done to acquire the remaining heat released from the sample. A graph showing the baseline and
integrated form of the heat flow is generated to allow for a manual check on the results.
25
Figure 3-5: Converted DSC data from the code analysis. A baseline was created and used to integrate the heat flow,
giving the heat of reaction of the resin.
The heat released (determined from the code) was used to determine the resin degree-of-
cure after the negative peak phenomenon for isothermal samples. Thus, experiments frequently
yielded plots of degree-of-cure versus time that were offset from zero degree-of-cure (see for
example, Figure 3-6). Once the isothermal tests could be accurately adapted and included, the
results from both isothermal and dynamic experiments were incorporated into the modeling
analysis.
3.3 Results
The data from the DSC was first inserted into the widely-used kinetic model from Khoun
et al [3].
26
𝑑𝛼 𝑑𝑡 = 𝐴 ∗ exp (
− 𝐸 𝑎 𝑅𝑇
)
𝛼 𝑚 ( 1 − 𝛼 )
𝑛 1 + exp ( 𝐶 ( 𝛼 − ( 𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇 ) ) )
(1)
The expression describes autocatalytic cure with a rate-limiting diffusion factor. Using this
equation, multivariable minimization of error was applied to determine values for the terms. The
error was minimized across 17 isothermal and ramp DSC runs. However, systematic
discrepancies appeared between the cure data and values predicted from the model. The
discrepancies were especially apparent in isothermal DSC runs (Figure 3-6a).
Figure 3-6: Modeling an isothermal test at 116°C with the conventional model. Major discrepancies existed
between the experimental and conventional model results.
27
Figure 3-7: Modeling a ramp test at 2°C/min with the conventional model. Minor systemic differences occurred
between the conventional model and experimental results.
Despite efforts to determine terms that yielded a closer match, significant differences
remained between measured and model predictions. Also, in some isothermal runs, the resin did
not reach full cure. The conventional model had no way of accounting for this phenomenon to
this extent, and adjustments were required. For dynamic (ramp) tests, data sets correlated more
closely, yet still exhibited systematic discrepancies (Figure 3-7b).
For dynamic heating, the model predictions consistently lagged behind the measured data,
indicating delayed cure in early stages (up to =0.7). The delayed cure kinetics was consistent
with the room temperature stability of the vinyl hybrid. The measured cure rate also decelerated
more quickly than the model cure rate as full cure was approached. The observed rapid
deceleration in reaction rate indicated that the rate-limiting diffusion terms were more influential
28
than the model was predicting. However, the diffusion terms were not accurately represented in
the conventional model, because an important aspect of the vinyl hybrid resin was not
considered. The room temperature stability of vinyl hybrid required a modified cure cycle that
took this characteristic into account. Therefore, a phenomenological model (fast-cure, or FC
model) was formulated to account for the inhibition. The new expression includes a term to
account for inhibition, as well as a term to account for incomplete cure at lower temperatures.
𝑑𝛼 𝑑𝑡 = 𝐴 ∗ exp (
− 𝐸 𝑎 𝑅𝑇
)
𝛼 𝑚 ( 𝛼 𝑚𝑎𝑥
− 𝛼 )
𝑛 1 + exp ( 𝐶 ( 𝛼 − ( 𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇 ) ) ) + 𝑑 ( 1 − 𝛼 )
ℎ
(2)
In this expression, the cure rate is represented as a function of α and temperature. In the
numerator, the term (αmax – α) is used instead of (1 – α) to account for the resin not fully curing
at lower temperatures. A term like this has been seen in styrene-based cure reactions that are not
always able to reach full conversion. The vinyl hybrid resin does not contain any styrene as part
of its chemical structure; however, its behavior is reflective of styrene-based resins in this regard.
Additionally, the term has been seen to account for vitrification in resins. Although αmax in most
cases will be 1, the distinction is necessary for accurate modeling of the rapid cure kinetics, a
feature manifest in the isothermal measurements.
In the denominator, the term with d and h accounts for cure inhibition at lower
temperatures. This new term allows the degree of cure to remain near zero until a threshold α is
reached, after which the autocatalytic terms dominate. Inclusion of the new term altered the
values for other terms in the expression. For example, the activation energy Ea, and autocatalytic
terms n and m, were decreased in the new expression relative to the conventional (Khoun)
29
model. The model with the following constant values, A = 4.86*10
^7
s
^-1
, Ea = 5.02*10
^4
J/mol, m
= 0.39, n = 0.21, C = 1.08, αCo = -55.0, αCT = 0.12 K
^-1
, d = 1.83*10
^3
, and h = -1.43, more
accurately represent fast cure kinetics than the Khoun model. In the denominator, the effect of
the diffusion term became more pronounced, and the C term decreased, while α Co and αCT both
increased relative to the Khoun model.
When the FC model predictions were plotted alongside measured results, the data sets
were more closely aligned, especially for the isotherm tests. The percent deviation between
predictions of the revised model and measured data was <5%, with negligible variance (Figure
3-8). In contrast, the percent deviation for the conventional (Khoun) model degree of cure on
isothermal tests was often >30% and exhibited variability.
30
Figure 3-8: Isotherm run with new model. The new model more closely matches data including at the starting point
at degree of cure 0.37 and final degree of cure 0.9.
When dynamic heating runs were inserted into the FC model, the results did not display
the systematic discrepancies described above.
31
Figure 3-9: Ramp run with new model. The new model does not exhibit the systemic differences like the
conventional model did.
The FC model accounted for cure inhibition, and thus more accurately simulated early
stages of cure. The altered diffusion terms eliminated the lag predicted by the conventional
(Khoun) model and more closely matched the data.
The final degree of cure αmax does not always reach αmax =1 (shown in isothermal data
Figure 3-8). Thus, an expression for αmax as a function of the maximum temperature is required for
the FC model, and this expression must also include the threshold curing temperature. One such
model for αmax is:
𝛼 𝑚𝑎𝑥
= 1 −
1
1 + 𝑘 ∗ e x p ( 𝑞 ∗ ( 𝑇 𝑚𝑎𝑥
− 𝑇 𝑐 ) )
32
where the constants k and q were determined using minimization of error. The term TC
represents the temperature at which the cure reaction commences (105 °C, from DSC results).
The value determined from DSC measurements was compared with a separate minimization of
error calculation treating it as unknown. The value of TC determined from that procedure (104.7
°C) was nearly identical to the value measured by DSC (105 °C), demonstrating the accuracy
and the physical relevance to the cure onset temperature of vinyl hybrid. The results of the αmax
model were compared to measured data, showing a difference of ~8% (Figure 3-10).
Figure 3-10: α max as a function of maximum temperature. The rapid change in α max begins at 105 °C and quickly
rises to above 0.95 by 120 °C.
33
When combining the equations for αmax and
𝑑𝛼
𝑑𝑡
, a model that accurately predicts the
behavior of fast curing, inhibited resins is created.
3.4 Conclusion
Fast-curing semi-pregs were modeled and evaluated using a vinyl hybrid resin. The
conventional autocatalytic model was modified to account for peculiar features of the fast-cure
resin (FC), particularly the inhibition of cure. As new and complex materials are created, the
models to describe their behavior must become more intricate in response. The model created in
this work was a logical next step to model resins with the new behavior feature of cure inhibition.
Although the resin curing quickly made analyzing DSC data difficult, it allowed for the
creation of a reproducible method of analyzing fast curing resins. In addition, the code that was
created to account for fast curing DSC data can be used with any thermoset material, including
with conventional models. The automation of these procedures can shorten the time to generate
the models for other novel materials by reducing manual calculation time. Automating tedious
steps during kinetic calculations not only reduces variability in data analysis, but also makes it
easier for researchers to focus on the exploring the phenomena present in a material.
The form of the model created can be used with similar types of resins developed in the future,
including room temperature stable resins that are not fast curing. Other resins may be able to avoid
the inclusion of αmax as a separate equation instead of as a constant of 1, which was a unique artifact
of the fast-curing nature of the vinyl hybrid used in this study. Conversely, the equation for αmax
can be different for other materials, depending on its crosslinking behavior during cure.
Independent of the form of αmax, the model can be used to predict the behavior of room temperature
34
stable resins. With accurate models, optimization of cure cycles can be done to minimize
processing time without resulting in parts that are not fully cured. With fast curing resins, the
accuracy becomes more important, as the shorter cure cycles have less latitude for errors. Thus,
the resulting model created in this section will be used to optimize cure cycles in Chapter 5.
35
4 CHAPTER 4: Rheology and Flow number analysis of novel thermoset resins
4.1 Introduction
The objectives of the present work were to modify standard rheology models to accurately
predict viscosity behavior of a fast-cure, room temperature stable resin and to assess the impact
resin viscosity has on the quality of vacuum bag only (VBO) conventional and semi-preg prepregs
in repair environments. A new model was developed to describe the rheological behavior of the
fast-cure, room temperature stable resin. Flow number analysis was used to assess the accuracy of
the new model and panels were fabricated to pair modeling results with physical evidence.
Rheology is the study of flow in materials and understanding the rheological behavior of
materials used in composite materials is vital for their processing [45]. Unlike Newtonian liquids
which have a linear relationship between shear rate and shear stresses, thermoset resins are
viscoelastic with both viscous and elastic components [46]. Intrinsic properties of thermoset resins
that affect the viscoelastic properties include molecular weight and molecular weight distribution,
inactive filler content, degree of branching, and crosslinking activity [47]. When these materials
are subjected to cure cycles with changing temperatures, their viscosity behavior changes. For
most thermoset materials there are three stages in their viscosity profile [48]. The first stage is at
lower temperatures prior to heating or cure of the material. For infusion resins, the viscosity can
be low enough at these temperatures to allow for flow. However, for out-of-autoclave prepreg
resins, the resin is tailored to have a higher viscosity at temperatures around room temperature.
This is done to prevent saturation prior to processing and preserve the evacuation channels
(EVaCs) in the prepreg [49]. Although the preservation of EVaCs increases the cycle time and
36
difficulty to achieve full saturation, they are necessary to prevent excessive porosity in the panels.
Autoclave prepregs do not have the same porosity problems as out-of-autoclave prepregs yet
maintain the higher viscosities at lower temperatures [50], [51]. The second stage in thermoset
resin viscosity profiles occurs when the temperature rises but the resin has not begun to heavily
crosslink. At this stage, viscosity will reach its minimum since the viscosity of thermoset resins
decreases with increasing temperature prior to cure. It is during this stage that most laminate
saturation is achieved. The final stage occurs when the material begins crosslinking, which raises
the viscosity [52]. Crosslinking in thermoset resins converts the material that was able to flow,
into solid structures to act as the binding matrix in reinforced materials. Thus, the viscosity
increases until the material ceases flow after a point commonly referred to as the gel point [53].
Any reinforcement that has not been impregnated by the time the gel point is reached will not be
able to bond with the matrix, resulting in weak points akin to the porosity derived from evolved
gases. In composite laminates, the porosity can stem from incomplete saturation [54]. Resin has
less resistance to flow in between fiber tows and along the fiber direction compared to flowing
into the fiber tows. Thus, incomplete saturation results in voids, or microporosity, within the
interior of fiber tows [8]. Understanding the flow behavior of resins is critical to prevent defects
from occurring composite laminates.
Epoxies are the most common thermoset resin used in the aerospace industry, and their
rheological behavior is well-characterized [55]. As thermoset resins continue to become more
complex, modeling their behavior has required additional terms in the equations describing their
behavior. Macosko described thermoset viscosity starting with Arrhenius terms [45], [52].
1
𝑡 1
= 𝐶 ex p (
− 𝐸 𝑎 𝑅𝑇
)
37
It was established that the activation energy and temperature is an important variable to
describe rheological behavior. Thus, future forms of viscosity equations had the form of
𝜂 = 𝑓 ( 𝑇 , 𝛼 )
where 𝜂 is the viscosity of the resin material, although many forms use a variable 𝜂 0 that represents
the initial viscosity of the material prior to processing. The early forms of a gel model contained
the 𝜂 0 term instead of using an Arrhenius temperature dependence [37]. The gel model uses a
constant α gel to represent the degree of cure gel point of the material.
𝜂 𝜂 0
= (
α
𝑔𝑒 𝑙 α
𝑔𝑒 𝑙 − α
)
𝐴 + 𝐵 α
As α approaches α
𝑔𝑒 𝑙 , 𝜂 approaches infinity which translates physically into flow ceasing as
the degree of cure reaches the gel point. The terms A and B are calculated from rheometer data on
the samples. To include temperature as a variable instead of using 𝜂 0, the gel model was modified
to include the Arrhenius term.
𝜂 = 𝐴 ex p (
− 𝐸 𝑎 𝑅𝑇
) (
α
𝑔𝑒 𝑙 α
𝑔𝑒 𝑙 − α
)
𝐴 + 𝐵 α
As epoxies and other thermoset resins diversified themselves, models describing viscosity
behavior did as well. Additional Arrhenius terms to account for multiple reactions were added.
Additionally, the exponential for α
𝑔𝑒 𝑙 was expanded to include a quadratic term on α. For many
modern conventional epoxies, the modified gel model proposed by Khoun has been employed
[56].
𝜂 = 𝐴 1
ex p (
− 𝐸 𝑎 1
𝑅𝑇
) + 𝐴 2
ex p (
− 𝐸 𝑎 2
𝑅𝑇
) (
α
𝑔𝑒 𝑙 α
𝑔𝑒 𝑙 − α
)
𝐴 + 𝐵 α + C α
2
The Khoun model features two Arrhenius terms and the α
𝑔𝑒 𝑙 term with a quadratic exponent.
For conventional epoxies with uninhibited reactants, this model describes viscosity behavior well.
38
With the advent of thermosets with novel properties, new models are needed to describe these
unique behaviors.
Predictive viscosity models can be employed to create and analyze various cure cycles.
Knowing how viscous a material is at any given temperature and degree of cure allows for the
calculation of a number across a cure cycle that relates to total flow occurring during that cure
cycle. With an assumption of Newtonian flow prior to cure, a characteristic flow number can be
derived from resin characteristics and the viscosity profile.
𝑁 Fl
= (
𝜌 p
𝜌 𝑜 ) [ 1 − ( [
16 𝐹 ℎ
𝑜 2
3 𝜋 𝑅 4
∫ 𝜂 ( 𝑡 )
− 1
𝑑𝑡 𝑡 𝑔𝑒 𝑙 0
] + 1 )
− 0 . 5
] x 100
In this expression, 𝜌 p
is the resin density, 𝜌 p
is the prepreg density, F is the lamination force, h0 is
the initial thickness, and R is the effective radius of the resin [57]. In vacuum-bag-only processing
F comes from atmospheric pressure and the other constants are derived from the material itself.
The critical value for viscosity comes from the integrated expression which can be separated from
the entire flow number expression. The effective flow number can thus be defined as:
𝑁 𝐹𝑙 . 𝑒 𝑓𝑓 = ∫ 𝜂 ( 𝑡 )
− 1
𝑑𝑡
𝑡 𝑔 𝑒𝑙 0
The integral comes from the inverse relationship between viscosity and flow of the material. As
viscosity decreases, the amount a material can flow increases, thus the inverse of 𝜂 ( 𝑡 ) is integrated
in the effective flow number expression. Studies on the effective flow number have revealed two
controlling variables that affect the evolution of viscosity, and therefore the effective flow number.
With conventional thermosets that have the three phases of viscosity of evolution, there are
opposing factors that affect the viscosity of a resin during cure [58]. As the temperature increases,
the viscosity decreases as the activation energy for flow is met in the resin. Consequently, as the
temperature increases, crosslinking occurs which decreases the viscosity until the material is
39
solidified at the gel point. It stands to reason that the temperature profile determines the rheology
behavior. In consideration of the two opposing phenomena, the two factors that dictate the
viscosity profile evolution in a conventional thermoset are the heating rate and dwell temperatures.
It has been shown that increasing the heating rate increases the overall flow number in an epoxy
system [59]. Furthermore, higher dwell temperatures can also increase the effective flow number
in a cure cycle. Novel thermoset materials can change how cure cycles affect flow numbers.
Additives that inhibit cure can alter the viscosity profile and therefore the effective flow numbers
of the resins.
In this study, a fast-cure resin and a typical epoxy resin were used to produce conventional
prepreg and prepreg with high through-thickness permeability (semi-preg). New terms were
introduced to a conventional rheology model to describe the unusual viscosity behavior of the
resin. The model was employed in flow number analysis to design efficient cure cycles for the
prepregs that were processed with a single vacuum bag in a repair environment, delineating how
cure cycle features (intermediate hold time, intermediate temperature) can be specified to control
the quality and performance of cured laminates. The effectiveness of using effective flow
number to predict flow in a prepreg processing was assessed with panels fabricated from both
materials.
4.2 Experimental
Rheology behavior was measured using a rheometer (TA Instruments) with neat resin tested
between two parallel plates. Test conditions included five dynamic ramps (1 – 10 °C/min) up to
150°C. The results from the rheometer were incorporated into the modeling analysis. Panel quality
was assessed by polished section microscopy. Polished sections (25×13 mm) were prepared from
40
the four panels to measure porosity. Void content was estimated by measuring the ratio of void
area to total area of each sample.
Panels with vinyl hybrid were fabricated using resin films produced commercially (Tipton
Goss Advanced Materials Company), while panels with epoxy resin were also fabricated using
resin films produced commercially (PMT-F4, Patz Materials & Technology). To produce semi-
preg formatting, the conventional films were dewetted after nucleation sites were introduced
with a hand-held spike roller [60]. The dewetting was conducted on silicone paper in a
convection oven (Blue M Oven). The resin was heated in the oven for 2 minutes at 104°C until
dewetting occurred and established the semi-preg format [61]. Prepreg was then fabricated by
transferring either the formatted resin or continuous films to dry fabric (Figure 4-1).
Figure 4-1: Semi-preg format achieved by applying dewetted film to dry fabric. The dry areas allow for increased
air evacuation in the through-thickness direction.
2 mm
Resin Dry Fiber Tows
41
The prepregs were prepared using a 2x2 twill fabric of carbon fibers with 6k tows
(DowAksa A-38) with resin contents of 35%. Using the same fabric allows for a more accurate
comparison between effective flow numbers by eliminating the fabric as a potential source of
flow differential. A stacking sequence of [0°/45°]2s for 8 plies was used for all panels. A
transparent release agent (Frekote® 700-NC) was applied to the tool plate surface before
stacking the plies. Perforated release film (Airtech A4000 P8) and a Teflon-coated fiberglass
peel ply were placed on top of the laminate, and a nylon breather cloth (Airtech Airweave N10)
covered the laminate. Vacuum bagging was then overlaid on the surface and peripherally sealed
with sealant tape (Airtech GS213-3). A heat blanket (Briskheat SR512018X18C) was used to
heat the laminates after applying vacuum to the bag. A custom control system for temperature
control was built using a controller (Watlow PM6R1CA-AAAAAAA), solid state relay (SSR-
240-10A-DC1), and K-type thermocouple input. The temperature inside the sample was
determined using calibrations of test runs on the blanket system. For each material, two cycles
were conducted, one to give a high flow number, and one for a low flow number. For the epoxy
panels, the high flow number cycle began with a room temperature debulk for 30 minutes,
followed by a 5°C/min ramp to 93°C and a 30-minute dwell. A second ramp at 5°C/min to
121°C was applied and held for 120 minutes. Cured laminates were cooled to 20°C at 1.1°C/min.
The same cure cycle was used for the low flow number epoxy panels, except that the dwell at
93°C was eliminated. The flexibility of inhibited resins like the vinyl hybrid made it possible to
match the flow number of the conventional epoxy without using different temperatures. To
adjust the high flow number cycle to be the same as the epoxy, the vinyl hybrid was held at 93°C
for 13 minutes. The lower minimum viscosity made the shorter dwell necessary to match
effective flow numbers with the epoxy panel. For the low flow number cycle, the rate was
42
increased to 1.3°C/min. Unlike with conventional materials, the faster heating rate reduced the
effective flow number for vinyl hybrid relative to the 1.1°C/min. The lower viscosity and
inhibited cure for vinyl hybrid translates to a higher flow number than the epoxy with the same
cure cycle. Without a mid-stage dwell time to change, the amount of time until the vinyl hybrid
reaches 105°C was altered. The time was changed by increasing the heating rate.
4.3 Results
To highlight the differences between the rheology behavior between vinyl hybrid and
epoxy, a chart was made showing the viscosity profile of both materials (Figure 4-2).
Figure 4-2: Viscosity comparison of epoxy and vinyl hybrid. The vinyl hybrid has a lower minimum viscosity and
sharper viscosity increase at curing temperatures.
43
The first feature difference between the two materials is the sharpness of the velocity
increase. For the vinyl hybrid the viscosity increases sharply at 105°C while the epoxy has a
more gradual increase. The drastic increase in viscosity for the vinyl hybrid is consistent with its
fast-cure nature. Because the resin cures rapidly, the viscosity will similarly have a rapid increase
in response to the crosslinking. The epoxy is not a fast-cure resin and thus has a slower increase
in viscosity with rising temperature. The second feature difference is the lower minimum
viscosity of the vinyl hybrid. The vinyl hybrid reaches viscosity of below 1 Pa s while the epoxy
resin has a minimum viscosity of 10 Pa s. Directly, this means that vinyl hybrid has the greater
potential for flow at the mid stage dwells of 93°C than the epoxy does. However, the final
feature makes the vinyl hybrid flexible for matching effective flow numbers with other materials.
Although it is not apparent on the figure, the vinyl hybrid can stay at low viscosities during mid
stage dwells without curing, due to inhibited curatives in the resin. Being able to ensure a
constant viscosity at a dwell temperature for an arbitrary amount of time makes it simple to test
similar effective flow numbers. To match the flow number for a given cure cycle for the epoxy,
the only variable that needs to be adjusted is the amount of dwell time at 93°C. If no dwell time
is desired (such as the ensure a low effective flow number), the temperature rate can be changed
to increase or reduce the amount of time until 105°C. This feature was used to ensure
comparable effective flow numbers in the panels made from epoxy and vinyl hybrid.
To model the rheological properties of vinyl hybrid, the equations previously adapted to
accurately model the kinetics of fast-cure room temperature stable resins were used [62].
𝑑𝛼 𝑑𝑡 = 𝐴 ∗ exp (
− 𝐸 𝑎 𝑅𝑇
)
𝛼 𝑚 ( 𝛼 𝑚𝑎𝑥
− 𝛼 )
𝑛 1 + exp ( 𝐶 ( 𝛼 − ( 𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇 ) ) ) + 𝑑 ( 1 − 𝛼 )
ℎ
44
The 𝛼 𝑚𝑎𝑥
term instead of using a value of 1, was needed for the vinyl as the resin often
has incomplete cure for isothermal cycle tests. The equation for 𝛼 𝑚𝑎𝑥
that matches the vinyl
hybrid 𝛼 𝑚𝑎𝑥
profile uses the maximum temperature of a cure cycle to return the final degree of
cure.
𝛼 𝑚𝑎𝑥
= 1 −
1
1 + 𝑘 ∗ e x p ( 𝑞 ∗ ( 𝑇 𝑚𝑎𝑥
− 𝑇 𝑐 ) )
Introduced in the fast cure kinetic model was a cure temperature, denoted as 𝑇 𝑐 in the
model. This 𝑇 𝑐 was found to be at 105°C which matched kinetic data for vinyl hybrid. This term
will become important for modeling viscosity for vinyl hybrid as well. The degree of cure acquired
from these equations was used in the conventional modified gel model thermoset rheology model
[3].
𝜇 = 𝜇 1
( 𝑇 ) + 𝜇 2
( 𝑇 ) (
𝛼 𝑔 𝑒 𝑙 𝛼 𝑔𝑒 𝑙 − 𝛼 )
( 𝐴 ′
+ 𝐵 ′
𝛼 + 𝐶 ′
𝛼 2
)
𝜇 𝑖 ( 𝑇 ) = 𝐴 𝜇𝑖
e x p (
𝐸 𝜇𝑖
𝑅𝑇
) 𝑖 = 1 𝑜𝑟 2
Using these equations, multivariable minimization of error was applied to determine values
for the constant terms. Systematic discrepancies appeared between the viscosity data and values
predicted from the model (Figure 4-3).
45
Figure 4-3: Viscosity test with conventional model - ramp at 1°C/min. The model does not match the initial
decrease in viscosity nor the sharp increase in viscosity at the curing temperature of 105 °C.
Despite efforts to determine terms that yielded a closer match, differences between
measured data and model predictions persisted. The unusual characteristics of the vinyl hybrid
resin – room temperature stability and fast cure – were not accounted for adequately in the
conventional model, preventing accurate prediction of viscosity behavior, particularly at early
and late stages of heating. Viscosity predictions using the conventional model showed a gradual
increase beginning at 90°C, yet measurements showed that the vinyl hybrid did not begin
solidifying until 105°C. Furthermore, the rate at which vinyl hybrid reached its lowest viscosity
and its gelation rate were not reflected, leading to a lag on both ends of the viscosity profile. To
remedy these discrepancies and account for both the fast-cure transition (at 105°C) and the
46
inhibited cure prior to transition, slight modifications to the model are required. A term Tc is
introduced that reflects the transition temperature of 105°C, as shown below.
𝜇 = 𝜇 1
( 𝑇 ) + 𝜇 2
( 𝑇 ) ( 𝑦 𝑒 𝑧 ( 𝑇 𝑐 − 𝑇 )
) (
𝛼 𝑔 𝑒 𝑙 𝛼 𝑔 𝑒 𝑙 − 𝛼 )
( A′ + 𝐵 ′
𝛼 + 𝐶 ′
𝛼 2
)
𝜇 𝑖 ( 𝑇 ) = 𝐴 𝜇𝑖
e x p (
𝐸 𝜇𝑖
𝑅𝑇
) 𝑖 = 1 𝑜𝑟 2
The transition temperature Tc determined from rheometer measurements was compared
with a separate calculation of Tc (a multivariable minimization of error, where Tc was treated as
an unknown variable). The value determined from that calculation (105.3°C) was nearly identical
to the value measured using the rheometer. Using the constant values determined from error
minimization ( 𝐴 𝜇 1
= 8.72e-18 Pa s, 𝐴 𝜇 2
= 4.23e-3 Pa s, 𝐸 𝜇 1
= 84672 J/mol, 𝐸 𝜇 2
= 9462 J/mol, 𝐴 ′
= .46, 𝐵 ′
= 0.87, 𝐶 ′
= 3.8, 𝛼 𝑔𝑒 𝑙 = 1, y = 2.3, and z = -1.32), yields a phenomenological model
that more accurately simulates the viscosity profile of fast-cure resins. Activation energies were
reduced, more accurately reflecting the viscosity behavior during initial heating.
The predictions from the modified model were plotted alongside the measured viscosity,
yielding data sets that were closely aligned, with deviation of <5% and negligible variance
(Figure 4-4). In contrast, predictions using the conventional model exhibited deviations >25%.
47
Figure 4-4: Viscosity test with new model – ramp at 1°C/min. The new model more closely matches the initial
viscosity decrease and sharp increase at 105 °C.
Using the modified model for viscosity evolution, comparative analysis using flow
number was performed. Assuming Newtonian flow prior to gelation, the flow number 𝑁 Fl
is
defined as:
𝑁 Fl
= (
𝜌 p
𝜌 𝑜 ) [ 1 − ( [
16 𝐹 ℎ
𝑜 2
3 𝜋 𝑅 4
∫ 𝜂 ( 𝑡 )
− 1
𝑑𝑡 𝑡 𝑔𝑒 𝑙 0
] + 1 )
− 0 . 5
] x 100
48
where 𝜌 p
is the resin density, 𝜌 p
is the prepreg density, F is the force, h0 is the initial thickness,
and R is the effective radius of the resin [57]. The critical part of that expression for this study,
the effective flow number 𝑁 𝐹𝑙 . 𝑒 𝑓𝑓 is the integrated expression:
𝑁 𝐹𝑙 . 𝑒 𝑓𝑓 = ∫ 𝜂 ( 𝑡 )
− 1
𝑑𝑡
𝑡 𝑔 𝑒𝑙 0
where t and 𝜂 depend on the cure cycle, and a higher 𝑁 𝐹𝑙 . 𝑒 𝑓𝑓 corresponds to increased resin flow
[63].
The effective flow number was used to guide the design of cure cycles in this study. For
conventional epoxies, the effective flow number encompasses the balance between the decrease
in viscosity as temperature increases and the increase in viscosity that accompanies onset of
gelation at higher temperatures. These counteracting phenomena severely limit the range of
possible effective flow numbers for epoxies. However, vinyl hybrid resin exhibits cure inhibition
at temperatures below 105°C, suppressing the onset of gelation below that temperature. Thus, the
resin affords flexibility in design of cure cycles, allowing a wider range of effective flow
number. This flexibility was used to match the vinyl hybrid effective flow numbers to epoxy
flow numbers for comparative analysis.
To evaluate the utility of employing effective flow number to tailor cure cycles, panels
were fabricated with the vinyl hybrid and epoxy resins, and cure cycles were specified such that
the same effective flow number was achieved for equivalent epoxy and vinyl hybrid panels.
Ramp rates and hold times were adjusted to yield the same effective flow number for all panels.
Polished sections from four panels were analyzed to measure porosity levels (Figure 4-5). The
first two panels shown correspond to a low flow number regime, one for each material. The
remaining two were from a high flow number regime, one for each resin.
49
Figure 4-5: Porosity of fabricated panels with high and low flow numbers. The low flow number panels had
significant flow related porosity while the high flow number panels exhibited minimal flow related porosity.
The panels show that different materials with the same effective flow number achieve
similar levels of resin infiltration during the cure cycle. For the same low effective flow
numbers, similar levels of both micro- and macro-porosity were exhibited for the epoxy and
vinyl hybrid panels. Microporosity in these panels indicated that the resin did not fully saturate
the fiber tows, reflecting the low effective flow number. In addition, macroporosity appeared in
the high effective flow number panels because of the limited pathways for air egress. Cure cycles
1.0 mm
Low flow number epoxy
Low flow number vinyl hybrid
1.0 mm
1.0 mm
High flow number epoxy
High flow number vinyl hybrid
1.0 mm
50
with low effective flow number provided less time for air evacuation prior to gelation.
Conversely, panels with high effective flow number exhibited minimal porosity for both the
epoxy and vinyl hybrid resins. The primary difference observed between the vinyl hybrid and
epoxy panels was the presence of macroporosity in the epoxy panel stemming from insufficient
bubble migration. This issue was explored using in-situ observation in Chapter 7.
4.4 Conclusion
Cure profiles of resins that feature cure inhibition, low-viscosity, and fast-cure/gelation can
be accurately predicted by modifying a conventional model as described in this work. The
conventional model was well-matched to the viscosity behavior of vinyl hybrid for much of the
cure cycle, but was unable to account for the initial and final stages of cure. Cure processes for
aerospace repair often employ cure cycles with long mid-stage holds to ensure full saturation.
With vinyl hybrid in particular, the final normalized flow number depends on mid-stage hold
time at its lowest viscosity of below 1 Pa s. Vinyl hybrid and other fast-cure resins can be
employed in faster cure cycles that do not feature mid-stage holds in applications such as
infusion. Thus, model predictions at the initial and final stages of cure become essential for the
accurate calculation of flow number and design of effective cure cycles. With accurate modeling
and flow number analysis, cure cycles that utilize the unique properties of fast-cure resins can be
created for any process requirement. A process for cure cycle optimization with fast-cure
inhibited resins using the model created here is discussed in Chapter 7.
In this study, a link between flow number and laminate quality was found. High effective
flow number panels were shown to have minimal flow related porosity while low effective flow
51
number panels had significant porosity. In addition, those results were consistent across two
materials with different viscosity profiles. The resins had a similar initial thickness and density
which allowed for the use of effective flow number instead the entire flow number expression.
Conditions between the panels featuring different resins were kept as close as possible to enable
the effective flow number to be compared. However, in practical applications it is not always
possible to do comparative analyses in a way that allows the effective flow number to be used as
it is currently defined. The generalized flow number expression will be needed when exploring
varying cure conditions with alternative materials. However, a factor that the current flow
number does not account for is reinforcement fabric geometry. The permeability of fiber
reinforcement to resin flow is affected by the direction and shape of the weaves, which in turn
affects the saturation by the resin. As equation for flow number currently stands, the saturation
into a 3k tow plain weave is equivalent to a 12k unidirectional fabric. A term that accounts for
the differences between reinforcement types can help the flow number become a greater
predictive tool for designing cure cycles. The term needs to account for the differences in flow
between tow sizing as well as the difference in geometry. For the tow size, a function of
permeability based on the tow sizing for the materials can be used. For the geometry, empirical
results from tests where everything about the reinforcement and resin material are kept the same
except the reinforcement geometry, need to be gathered. With both factors included into flow
number, optimization of cure cycles with novel materials and processing environments becomes
possible.
52
5 CHAPTER 5: Cure Cycle Optimization with Semi-pregs
5.1 Introduction
The objectives of the present work were to employ kinetic and rheology models to create
optimized cure cycles while maintaining robust quality controls. A process that enables efficient
cure cycle optimization was created. In the process, multiple cure cycle variables can be quickly
compared to determine their impact on laminate performance. Panels were made to test the
optimization procedures and the models created for a novel fast-cure, low-velocity resin.
The production rate of high-performance composites is presently limited by lengthy cycle
times and costly, labor-intensive manufacturing processes that require curing in autoclaves. VBO
processing of prepregs can mitigate these limitations, particularly the need for autoclaves, but
VBO processing sacrifices process robustness, requires long cycle times (hours), and frequently
requires protracted debulking (vacuum holds).
53
Figure 5-1: Example of a typical OoA cure cycle. Note that the entire cycle takes over six hours until laminate
completion.
These limitations stem largely from both the resin chemistry and the conventional design
of VBO prepregs, which rely exclusively on edge-breathing for air removal. Nevertheless, the
use of out-of-autoclave (OoA) processes continues to grow, driven by the need to reduce both the
process cycle time and the manufacturing cost of composite laminates. Doing both will reduce
barriers to entry and expand the use of these high-performance materials.
In the aerospace industry, production of composite parts from prepregs generally involves
curing in autoclaves, primarily to limit porosity and to yield consistency in part quality. Voids
within composite materials reduce mechanical performance, generating stress concentrations that
lead to premature failure. Consequently, parts with porosity levels >1% are generally rejected
[64]. Autoclaves apply pressure (and temperature) on the laminates during cure that suppress
volatile evolution from the resin [51]. Thus, autoclave processing consistently yields void-free
parts. However, capital and operating costs are high relative to ovens, and throughput is severely
restricted, both by the autoclave and by the kinetics of cure in conventional thermoset prepregs
54
[65]. The lengthy startup time and intensive resource consumption of autoclaves make it difficult
to increase production efficiency, and prevent in-field processing often required for repairs and
modifications [66]. These issues provide the impetus for development of out-of-autoclave
(OoA)/vacuum bag only (VBO) prepregs, which are designed to be cured in conventional ovens
[6].
The key to VBO processing of prepregs lies in control of porosity [67]. Without super-
ambient pressure to suppress void development, extra care must be taken to ensure complete
removal of air and/or evolved gases from the prepreg while it is being cured. The design of
conventional VBO prepregs features partially impregnated resin on both sides of the fiber bed,
creating a dry channel in the ply midplane that allows gases to egress via edge-breathing dams
[68]. Air trapped between and within plies, as well as evolved volatiles, must be evacuated
before the resin fully saturates the fiber bed and gels [69]. However, gases that are not
completely removed remain as voids after gelation. Thus, imperfect or adverse conditions during
cure of edge-breathing VBO prepreg can result in porosity levels >1%. Because of this, VBO
prepreg plies must undergo “debulking” (extended vacuum hold) to remove air prior to cure [70].
Although the setup time for ovens or heat blankets is less than that of autoclaves, debulking can
often render VBO processing more time consuming than autoclave cure, particularly for large or
complex parts. Thus, VBO processing does not always reduce cycling times.
Compounding the challenges above, scenarios arise where gas evacuation via edge-
breathing is not feasible. One such example is the co-cure of composite repair patches [71].
Because patches are inserted into fully cured parent material, the pathways for gas egress via
edge-breathing are generally blocked. The situation is further complicated by adhesive layers,
which ensure strong bonding but prevent air egress. Although methods are available to mitigate
55
these problems (such as double vacuum debulking (DVD)), they are difficult to implement and
do not ensure void-free parts [27]. Because of the difficulty of these methods, hard patches cured
in autoclaves are often employed instead of co-cured patches, despite often being weaker and
less versatile [72]. With VBO prepregs, whether in a manufacturing environment or an infield
repair, these issues are exacerbated in larger and thicker parts. When manufacturers consider
substituting composites for conventional materials, they are deterred by the large capital
investments required, or by lack of robustness of VBO processing. In both autoclave and VBO
methods, making a single part requires hours of processing in a cycle akin to Figure 5-1.
With the advent of semi-preg formatting, debulking may no longer be the rate limiting step
for OoA prepreg processing. In addition, as seen in Chapters 3 and 4, resins with unique kinetics
and viscosities are being developed and accurately modeled. Deploying these models to assess
their impact on laminate quality and performance during fast curing processes requires a
framework to analyze multiple inputs. A MATLAB modeling framework was developed and
employed to allow for comparative analyses for cure cycles when altering multiple input
variables at the same time. This allows for tests to be performed to compare the effects of
particular inputs as well as optimize cure cycles for faster cure. Prepregs with vinyl hybrid and
semi-preg formatting were fabricated and tested using this framework and the impact on
performance and quality of input variables was assessed.
5.2 Methods
56
The vinyl hybrid resin that was employed in Chapters 3 and 4 was employed. To make the
semi-preg material, the resin was dewetted. The first step was creating dewetting nucleation sites
through spike rolling the resin film on the backing paper. The resin was then put into an oven
(LR Environmental Blue M) to dewet. The samples were heated at 104 °C for 2 minutes until the
resin formed the semi-preg pattern. The resin film was then pressed onto the reinforcement. The
fabric was a 2x2-twill carbon fabric with 6K tows (Dow Aksa A-38).
Identical layup procedures were used for all panels. Non-perforated fluorinated ethylene
propylene (FEP) release film (Airtech A4000) was taped to an aluminum tool plate, 450 450
mm. A thin tool plate was chosen to minimize the heat sink effects of a thicker plate. The
prepreg measuring 150 150 mm was then laid up in a quasi-isotropic stacking sequence of
[0°/45°]2s with 8 plies for each panel.
The panel edges were sealed using sealant tape. The sealant tape dams prevented edge
breathing that would normally reduce debulking time, but more importantly, sealed edges
demonstrated the efficacy of the semi-preg format for air evacuation. Sealing edges also
eliminated concerns of resin bleed. To complete the vacuum bagging, the laminate was covered
with a layer of perforated FEP release film. The perforations were spaced every 50 mm to limit
resin bleed from the low-viscosity resin.
A layer of nylon breather cloth (Airtech Airweave N10) was placed atop the perforated
FEP layer. The vacuum bag was then sealed and placed in an insulated chamber. The bag was
then covered with a heat blanket. The insulation reduced heat losses to the environment, allowing
rapid and more accurate temperature control. Heat blanket temperature was controlled using K-
type thermocouples and a controller (Watlow PM6R1CA-AAAAAAA). The laminate
57
temperature was measured using K-type thermocouples centrally located on the bagging film.
Vacuum was applied with a dedicated vacuum pump (Busch R5). Bag pressure was monitored
using a gauge on the bag, and a vacuum level of >28.5” Hg was maintained throughout the cure
cycle.
Glass transition temperature was determined using dynamic mechanical analysis (DMA,
TA Instruments Q800). Samples measuring 60 12 mm were cut from the panels and tested with
dual cantilever mode. The samples were heated at 3 °C/min beyond the glass transition
temperature. The Tg was identified as the peak of the tangent delta curve.
Porosity was measured from polished sections, 25 mm. Cross sections were ground and
polished to 4000 grit. Images were acquired using a digital stereo microscope (Keyence VHX
600) at 200 magnification. Void content was calculated by measuring the ratio of void area to
the total prescribed area for each polished sample. Void areas were indicated using a
combination of binary thresholding and manual input.
Interlaminar shear strength was measured according to ASTM 2344D. Samples, having a
width of two times the thickness and a span greater than six times the thickness, were prepared.
Samples were loaded at a rate of 1mm/min until failure. The maximum load was used to
calculate short beam shear strength.
5.3 Results
58
To optimize cure cycles, predictive models for the kinetic and rheology behavior of the
resin are required. In the previous chapters, these models were created for the unique properties
of the vinyl hybrid resin. The first model was a kinetic model able to match the inhibited cure
behavior and fast-cure nature of the resin.
𝑑𝛼 𝑑𝑡 = 𝐴 ∗ exp (
− 𝐸 𝑎 𝑅𝑇
)
𝛼 𝑚 ( 𝛼 𝑚𝑎𝑥
− 𝛼 )
𝑛 1 + exp ( 𝐶 ( 𝛼 − ( 𝛼 𝐶𝑂
+ 𝛼 𝐶𝑇
𝑇 ) ) ) + 𝑑 ( 1 − 𝛼 )
ℎ
Unlike many resins, vinyl hybrid does not always reach full cure. This effect is more
pronounced during an isothermal cure cycle. Thus, 𝛼 𝑚𝑎𝑥
needs to be calculated for this material.
A form of 𝛼 𝑚𝑎𝑥
that matches the vinyl hybrid behavior uses the properties of the resin.
𝛼 𝑚𝑎𝑥
= 1 −
1
1 + 𝑘 ∗ e x p ( 𝑞 ∗ ( 𝑇 𝑚𝑎𝑥
− 𝑇 𝑐 ) )
The term 𝑇 𝑐 refers to the curing temperature of the material at 105 °C. This expression is
substituted into the degree of cure equation to describe the kinetic behavior of the vinyl hybrid
resin.
The second model is a modified gel model that is able to account for the inhibited cure
and low viscosity of the vinyl hybrid.
𝜇 = 𝜇 1
( 𝑇 ) + 𝜇 2
( 𝑇 ) ( 𝑦 𝑒 𝑧 ( 𝑇 𝑐 − 𝑇 )
) (
𝛼 𝑔 𝑒 𝑙 𝛼 𝑔 𝑒 𝑙 − 𝛼 )
( A′ + 𝐵 ′
𝛼 + 𝐶 ′
𝛼 2
)
The Arrhenius terms, 𝜇 𝑖 ( 𝑇 ) , are defined as.
𝜇 𝑖 ( 𝑇 ) = 𝐴 𝜇𝑖
e x p (
𝐸 𝜇𝑖
𝑅𝑇
) 𝑖 = 1 𝑜𝑟 2
59
The 𝑇 𝑐 term also appears in the rheology model. In both models 𝑇 𝑐 has a value of 105 °C
and refers to the temperature at which the inhibition of the vinyl hybrid ends, and rapid curing
begins. Unlike degree of cure, viscosity alone does not yield any insight into the processing of a
composite laminate. The predictive model of viscosity first needs to be converted into an
effective flow number 𝑁 𝐹𝑙 . 𝑒 𝑓 𝑓 that correlates to the flow of the resin into the reinforcement.
𝑁 𝐹𝑙 . 𝑒 𝑓 𝑓 = ∫ 𝜂 ( 𝑡 )
− 1
𝑑𝑡 𝑡 𝑔𝑒 𝑙 0
It was shown in Chapter 4 that the effective flow number can be a useful tool in predicting
how the viscosity can translate into panel quality. The optimizations and panels fabricated in this
study further confirm its use as a predictive tool.
Both the kinetic and rheology models were incorporated into the optimization analysis. To
predict the cure behavior of the vinyl hybrid resin, a cure cycle map was constructed to guide
efforts to minimize cycle time. A similar procedure was used for both the rheology and kinetic
models, and their results were analyzed in conjunction. This ensured that an optimized cycle
would be optimized for both kinetic and rheology, rather than just one or the other.
For the kinetic model optimization, a code was written (MATLAB) to simulate different
cure cycles and compare final degrees of cure. First, a final degree of cure, αmax, was determined
using the expression for αmax, then input into the equation for the kinetic model. This equation
was then employed to generate a final degree of cure for any specified cure cycle. The aspects of
a cure cycle that were varied included temperature ramp rate, mid-stage hold time, mid-stage
temperature, maximum temperature, and hold time at maximum temperature. Each of these
aspects influence the final degree of cure from a cure cycle, and their magnitudes were
60
considered accordingly. In each optimization run, one of the temperature-based aspects was
tabulated against one of the time-based aspects. For example, the maximum temperature at the
final dwell could be tabulated against the mid-stage hold time. For the time-based aspects, they
were incorporated into the total cure cycle time. In the previous temperature, maximum
temperature would be compared against total cycle time. The results were used to generate a 3D
plot of final degree of cure as a function of the two varied aspects (Figure 5-2). In the plot, the
red plane represents an value of 0.99, and the purple plane includes possible combinations of
time and temperature that meet or exceed that value.
61
Figure 5-2: MATLAB simulation of cure cycle. In this plot, the maximum temperature was compared with total
cycle time to generate final degree of cure.
Possible cure cycle
parameters
0.99 degree of cure
plane
Degree of
cure surface
plot
62
The space visible above the red plane represents possible cure parameterization. Through
iterations of this process, one can design cure cycles of minimum duration that also achieve a
final degree of cure of at least 0.99 that minimizes processing time.
A similar procedure was used to optimize cure cycles based on rheology. Using the
modified model for rheology for the vinyl hybrid resin, a cure cycle map was constructed to
guide cure cycle design. Code was written (MATLAB) to generate normalized flow numbers for
simulated cure cycles with a range of parameters. First, the final degree of cure, 𝛼 𝑚𝑎𝑥
, was
determined and input into the kinetic equation to generate degree of cure, 𝛼 , at any point in a
given cure cycle. Next, the 𝛼 term and cure cycle parameters were used to generate the viscosity
profile for the cure cycle. An integration of the inverse viscosity profile was taken to calculate
the effective flow number, 𝑁 𝐹𝑙 . 𝑒 𝑓 𝑓 , for the cure cycle. The effective flow number was normalized
to a value of effective flow number that consistently yielded full saturation from the high flow
number panels in Chapter 4. The normalization was done to allow for a similar analysis akin to
degree of cure for the kinetic parameters. Parameters of the cure cycle were varied, including
temperature ramp rate, mid-stage hold time, and mid-stage temperature. The dependence of
normalized effective flow number on intermediate dwell temperature and total cycle time were
used to create a 3D plot (Figure 5-3). In the plot, the red plane represents a flow number of 0.99
and the purple plane indicates possible cure cycles that meet or exceed that value. The
normalized flow number values above 1 were displayed as 1 on the figure for clarity.
63
Figure 5-3: MATLAB plot of cure cycles. Normalized effective flow numbers were calculated after varying total
cycle times and mid stage hold temperatures.
Through iterations of this process, the cure cycles employed on the panels were chosen.
The visual graphs that represent either full cure, or full saturation made it simple to select cure
cycle parameters that minimized time. Because the graphs were generated through an automated
procedure, iterating cure cycles until a cycle that met testing parameters was done quickly.
Having the predictive models from Chapters 3 and 4 made the cure cycle optimization for this
Possible cure cycle
parameters
0.99 normalized
flow number
Degree of normalized
flow number surface
plot
64
novel material to happen much more quickly than having to fabricate and analyze a series of test
laminates.
To validate the model predictions and test the effects of varying aspects of the cure cycle,
panels were fabricated using vinyl hybrid semi-preg. A control panel was cured using a five-hour
cure cycle to produce a void-free laminate and provide a benchmark for mechanical
performance. This cycle was the following: 30-minute debulk at room temperature, 60-minute
ramp to 90 °C, 30-minute hold at 90 °C, 30-minute ramp to 150 °C, 60-minute hold at 150 °C,
and finally a 90-minute cool down to room temperature. In addition, panels were fabricated using
five fast-cure cycles and compared against the control panel. The cure cycle parameters that were
varied included maximum temperature, mid-stage hold time, and room temperature debulking
time. These parameters were chosen because they had significant impacts on the cure cycle
parameters during optimization, or they are important for air evacuation. The maximum
temperature was identified as the variable that made the most impact on maximizing degree of
cure without increasing cure cycle times. The mid-stage hold time was identified as the variable
that had the largest impact on the normalized effective flow number. Finally, the debulking time
was chosen to examine the effect semi-preg formatting had on air evacuation. Each sample was
held at the maximum temperature for 10 minutes. Maximum temperatures were 150 °C or 120
°C to test the predictive accuracy of the kinetic model and subsequent optimization. The models
predicted full cure ( =1) for 150 °C, and =0.17 for 120 °C. A duration of 10 minutes was
assigned to both the mid-stage hold at 90 °C and to room temperature debulking, and the total
ramp time to reach the final temperature was also 10 minutes. For cure cycles with a mid-stage
hold, 5 minutes was assigned to each temperature ramp before and after the hold. The test matrix
of fabricated panels with variables chosen for each is shown in Figure 5-4.
65
Figure 5-4: Sample test matrix. Panels 1-4 each have one or more variables removed from the cure cycle to assess
their impact on laminate quality. Panel 5 represents a fully optimized sample from a kinetic and rheological
perspective.
As an example, the cure cycle for Panel 5 consisted of the following steps: 10-minute
debulk at room temperature, 5-minute ramp to 90 °C, 10-minute hold at 90 °C, 5-minute ramp to
150 °C, 10-minute hold at 150 °C, and finally a 10-minute cool down to room temperature.
66
Figure 5-5: Panel 5 cure cycle. Note that unlike conventional cure cycles, the laminate is fabricated in under an
hour.
Panel 5 represented the optimization for both rheology and kinetics. The final degree of
cure and normalized effective flow number for that cycle were both above 0.99, representing full
cure and saturation respectively. After fabrication, the panels were then evaluated for quality and
performance metrics. The percent porosity of the panels (measured from polished sections) is
shown in Figure 5-6.
67
Figure 5-6: Porosity of fabricated panels percent porosity. Panels 2 and 5 had the lowest porosity other than the
control panel in line with them having both a mid-stage hold and debulking applied.
The porosity of the panels (Figure 5-6) depended strongly on the room temperature debulk
and mid-stage hold times. The control panel showed <0.1% porosity. Panels 2 and 5 included
both steps and exhibited the lowest porosity among fast-cured samples (3.0% and 3.3%
respectively). This finding is consistent with assertion that porosity depends strongly on the
debulk and mid-stage hold, since both steps were included in the cure cycles for each panel.
Sample 4 exhibited lower porosity than Sample 3, indicating that debulking had a greater effect
on porosity reduction than a mid-stage hold when the steps were mutually independent. Sample 1
included neither step, nor exhibited the highest porosity level.
6.5
3.0
5.8
4.9
3.3
0.1
68
Figure 5-7: Porosity of fabricated panel micrographs. Despite relatively small differences in the panel cure cycles,
the resulting laminate qualities were drastically different.
The micrographs in Figure 5-7 highlight the differences in void characteristics between the
different samples. The top sample is the control panel, and showed negligible porosity, while the
middle image is Panel 2 (3% porosity), and the bottom image is Panel 1 (6.5% porosity).
The porosity data indicate that both a mid-stage hold and a room-temperature debulk
effectively reduce porosity, with debulking having a stronger effect. Panel 1 was fabricated
without debulking or mid-stage hold, and exhibited both micro- and macro-voids. In Panel 2,
interply voids stemming from air entrapment were prevalent, despite fabrication with both
debulking and mid-stage hold steps. Similar voids were present in Panel 5, which was fabricated
with the same cure cycle except for a different Tmax. Unlike Panels 2 and 5, Panel 4 exhibited
higher levels of interply porosity, despite fabrication with the same debulking time. However,
Panel 4 spent 5 minutes less time below 105 °C than Panel 5, which led to the higher macrovoid
Control
Panel 2
Panel 1
69
content. Panels 1 and 3, both of which did not have a debulking step, exhibited the most porosity.
Although all panels exhibited porosity (except the control panel), increasing debulking time
reduced porosity proportionately.
The measured values of Tg confirm model predictions for the degree of cure in Figure
5-8.
Figure 5-8: Glass transition temperatures of fabricated. Panel two was the single panel that was predicted to reach
full cure and had a lower glass transition temperature as a result.
70
Panel 2, which underwent the lower maximum temperature cycle, exhibited a lower Tg
than other panels. The FC model predicted full cure versus incomplete cure when comparing two
different cure temperatures, 120 °C and 150 °C, and a hold time of 10 minutes. Furthermore,
other panels exhibited Tg values similar to the control panel, demonstrating that despite a cure
time of 10 minutes, the panels reached full cure using a cycle designed using model predictions.
Results for the short beam strength of the panels are shown in Figure 5-9.
Figure 5-9: Interlaminar shear strength of panels. The strength of the panels was inversely proportional to the
amount of porosity in the laminates.
Figure 5-9 shows interlaminar shear strength (ILSS) values for the 5 panels. The value
for Panel 2 is not shown in Figure 5-9, because those panels were not fully cured, and thus
exhibited lower strength values than other panels to the extent that an accurate measure was
unable to be gathered. The trend in panel strength shows increasing ILS as panel number
71
increases (and porosity decreases). Porosity reportedly decreases ILS, and the panel ILS values
exhibit the same dependence [28], [73]. The ILS for Panel 5, with the lowest porosity of the
panels outside of the control, was >95% of the ILS of the control panel.
The ILS data demonstrate that with appropriate cure cycles and use of semi-preg formats to
enhance air removal, fast-cure resins can be used with vacuum-bag cure to achieve mechanical
performance similar autoclave cured prepregs. In addition, the revised kinetic and rheology
models can be used to design an effective fast-cure cycle that achieves full cure and saturation.
Although the ILS values for fast-cured panels were less than that of the control panel, the
difference was modest (< 5%), and was attributed to porosity within the panels. By reducing
porosity and designing a suitable fast-cure cycle, panels with properties equivalent to a control
panel can be achieved.
5.4 Conclusions
Fast-curing semi-pregs were modeled and evaluated using a vinyl hybrid resin. Models that
account for the unique features of vinyl hybrid were employed to design efficient cure cycles.
Both kinetic and rheology models were employed to generate predictive degrees of cure and
normalized effective flow numbers respectively. Cure cycles the time scale of minutes were
created through cure cycle optimizations. Panels were produced with short cure cycles (< 1 hour)
based on model calculations. Panels exhibited full cure and porosity levels of 1-7%. In principle,
these porosity levels can be reduced by extending the debulk time and/or the mid-stage dwell, or
using slower ramps to promote air evacuation during cure. A third model that can assess the
72
amount of air evacuation in the same way the kinetic and rheology models were able to would be
helpful. It was seen that air evacuation had the greatest effect on total porosity in the panels, thus
having a way to optimize air evacuation would be a significant piece in designing efficient cure
cycles. The model that describes air evacuation behavior would be closer in form to the viscosity
modeling than the kinetic modeling. Variables such as three-dimensional air permeability, bag
pressure, and other relevant gas flow variables could be analyzed. Assumptions such as no
evolved gases from the resin would need to be made, or additional variables would need to be
included to account for gas evolution with rising temperatures. An integration of the model
created for air evacuation could be done. Unlike rheology, the inverse of the gas model would
likely not be needed to be taken to generate a gas flow number. This third model could address
the third major factor affecting porosity. Optimizations with all three models could be done using
the same procedure outlines in this chapter.
Although there may always be tradeoffs between shorter cure cycles and part quality,
inherent advantages of semi-preg will generally allow shorter cycle times when compared to
conventional prepregs. Despite aerospace applications generally requiring the lowest porosity
levels, shorter cure cycles without autoclaves will have clear benefits. Industries that produce in
larger volumes (such as automotive) generally require shorter cycle times, but can often tolerate
higher porosity levels. Further refinements to semi-preg format and resin formulation may be
required to adapt to the needs of different industrial sectors.
The processability and room temperature stability of the vinyl hybrid resin are appealing
features for applications beyond manufacturing of primary structures. One such application is
repair of composite structures. Conventional repair materials and methods require specialized
depots to perform repairs because of the limitations of employing conventional OoA prepregs.
73
The room temperature stability of the vinyl hybrid obviates the need for cold storage of prepregs,
an important attribute for in-field repairs. In addition, the relatively low cure temperature
required and smaller exotherm of the vinyl hybrid resin (compared to epoxies) are features well-
suited to repair of sensitive structures. The semi-preg format could potentially simplify the
process for in-field repairs, reducing the need for expertise and specialized equipment.
In this work, two unconventional practices were combined and evaluated – fast-cure resin
and VBO cure of semi-preg. The semi-preg format reduces the times required for debulking,
hold and cure times, shrinking the process window for conventional materials from several hours
to less than an hour. While these practices are mutually compatible, they can be deployed
separately, also. When used with conventional thermosets (e.g., epoxies), the semi-preg format
affords distinct advantages with respect to air removal during consolidation. Similarly, vinyl
hybrid formulations can be used in conventional prepreg formats, offering both rapid cure
characteristics and eliminating cold storage requirements. Nevertheless, the greatest benefits will
accrue from deploying both technologies simultaneously. Doing so may ultimately enable
expansion of composite materials into new markets and applications.
74
6 CHAPTER 6: Defect Reduction for OoA Prepregs in Repair Environments
6.1 Introduction
Carbon fiber composites exhibit high specific strength and stiffness as well as resistance to
fatigue life and environmental attack by moisture and solvents [74]–[76]. The demand for such
materials has steadily increased and is projected to continue to grow [77]. However, these materials
are not immune to damage stemming from service loads or from manufacturing defects. Such
damage can occur at different length scales, ranging from matrix cracks to ply delaminations to
broad structural damage [78], [79]. Regardless of the scale of the damage, the damage adversely
affects part performance, increasing safety risks in service [80]. Historically, damaged parts have
been completely replaced to reduce risk to performance [81]. However, complex structures render
this practice impractical, especially in aerospace applications. Therefore, procedures for on-aircraft
repair have become necessary [82].
To execute repair to a composite part, damage is first assessed by nondestructive testing
[83]. The damaged area is then removed and prepared for bonding to a patch, usually through
abrasive grinding [84]. Attachment is achieved either through mechanical attachment or through
adhesive bonding (or both). Although commonly employed, mechanical fastening invariably
generates stress concentrations, weakening the resulting part. Adhesive bonding offers the
advantage allowing full strength, when performed correctly [85]. For bonding, a scarf joint is most
widely practiced, and is preferable to a lap or stepped bondline due to increased strength [86], [87].
75
Figure 6-1 - Schematic of scarf and stepped repairs with over-plies. Stepped repairs allow for repair plies to be
inserted into the parent material but can be difficult to align. Scarfed repairs are more difficult to machine but can
offer additional layup flexibility and increase strength retention in the composite part.
In stepped repairs, material is removed in a step pattern, often attempting to match the
parent material [88]. The steps allow for a logical layup of plies into the parent laminate.
However, it can be difficult to match the stepped geometry with the prepreg or wet layup lies
inserted into the patch. Mismatches can lead to areas of lower fiber volume fraction or porosity
buildup in the corners [89]. Scarfed material removal avoids the corners of stepped repairs by
creating a straight bondline at a specific angle in the parent laminate [86]. While it can simplify
processing by virtue of layup flexibility, machining scarfed repairs is more difficult than that of
stepped, especially with manual techniques. Scarfing presents more opportunities for error in the
machining process itself when compared to stepped patches. Inadequate surface preparation,
material selection, and handling of a repair will result in a patch with lesser strength than the
76
parent material [90]. Experimental and modeling work has guided efforts to optimize scarf
repairs, from the angle of the scarf to the procedure to maximize adhesion on the parent surface
[80]. However there has not been extensive investigation into tailoring the material systems for
the conditions present during repair.
Two types of patches - hard (cured) or soft (uncured) - can be deployed in repairs. A hard
patch is fabricated in a controlled production setting and benefits from the reliability of those
procedures. The hard patch is co-bonded to the damaged area using an adhesive, and thus does not
have the ability to conform to complex/irregular shapes. Hard patches generally are not deployed
for in-field repairs where manufacturing capabilities are absent. Soft patches are fabricated by
applying uncured plies of pre-impregnated fibers (prepregs) or wet laid fabric/resin, then co-curing
with the adhesive [85]. Soft patches are processed on the parent laminate at the repair site. This
offers several advantages compared to hard patches, mostly regarding flexibility of repair [91].
Because the repair patch materials begin in the uncured state, they can conform to the
specifications needed to perform the repair. Without the need for outside manufacturing, soft
patches allow for reduced logistical demand and downtime for the equipment being repaired [92].
However, due to quality restraints from in-field repair conditions, many aerospace and automotive
repairs are done in a depot. Requiring a depot to perform maintenance eliminates the potential
upside of reduced downtime for in-field repair. Depots provide facilities not present in the field
that aid in performing high quality repairs [93]. These facilities include the presence of skilled
technicians, cold storage, and processing equipment such as double vacuum debulking (DVD)
capabilities. Soft patches can be performed by wet layup or with prepreg materials. They are
becoming more popular due to their flexibility, and work has been done to improve aspects of the
procedure, from preparation to processing. Co-curing provides higher joint strength compared to
77
co-bonded joints [94]. Thus, soft patches are often preferred for repair, provided patch quality can
be comparable to the parent manufactured material. Soft patches are generally performed out-of-
autoclave (OOA) due to the prohibitively large sizes of structures being repaired [66].
Autoclave curing consistently yields high quality composite laminates and imparts process
robustness [68]. On the other hand, autoclave curing restricts part size and throughput, and requires
major capital and operational investments. In an attempt to avoid these, efforts have been devoted
to out-of-autoclave processing of prepregs [6]. Without autoclave pressures to suppress volatiles
(mostly moisture) and entrapped air, these process routes generally sacrifice robustness.
Consequently, careful control over the bagging and curing process is required to achieve laminate
quality comparable to autoclave cure. OoA prepregs offer precise control of fiber orientations and
volume fractions, simple layup procedures, and consistent properties [95]. However, without
autoclave pressures during cure, prepreg format is the key variable in the production of high quality
parts [96].
During VBO cure, voids can form within fiber tows (microvoids) or between tows
(macrovoids) as a result of faulty processing or environmental conditions [70], [97]. Void
mitigation on the interior and surface of composite parts has been an area of emphasis for OOA
prepregs [98]. Partially impregnated prepregs demonstrate lower porosity compared to fully
impregnated prepregs [64]. Employing hot-melt prepregging to partially impregnate both sides of
the fibers allows re-work and yields laminate quality comparable to autoclave-cured prepregs [99].
The main mechanism behind this porosity reduction stems from engineered vacuum channels
(EVaCs) that promote in-plane air evacuation [68]. The dry tows then get impregnated with the
resin during the cure cycle, driven by pressure differences from vacuum bagging [69]. A variety
of resin distributions have been adopted in commercial prepregs, yet all rely on similar EVaCs for
78
gas removal and consolidating laminate layers [100]. EVaCs achieve varying levels of success in
reducing void content, yet depend strongly on material layup procedures. During layup, edge
breathing is required to connect to the EVaCs and allow gas egress to the vacuum outlet. Breathing
dams are placed on the laminate periphery to ensure gas egress. However, when complex contours
or embedded plies are present, edge breathing is impaired.
Figure 6-2 - Schematic of the impairment of gas edge breathing caused by impermeable bondlines in a laminate
repair patch.
Repairs are an example where edge breathing cannot be achieved due to the impermeable
bondline (Figure 6-2). The absence of edge breathing would seemingly preclude the use of OoA
prepregs in repairs, since conventional OoA prepregs require edge breathing for gas egress.
Alternative measures include breathable adhesives, resin infusion and other methods to simulate
edge breathing in repair environments [71], [101]. One technique commonly employed for repairs
is double vacuum debulking (DVD), although additional complexity is associated with the setup
[66]. During DVD, a rigid closed structure is placed around the vacuum-bagged laminate. A
separate vacuum is applied to the enclosing structure, allowing for gas removal during debulk
without laminate consolidation (atmospheric pressure can hinder air evacuation). The DVD
79
approach reduces porosity in laminates compared to equivalent single-vacuum-bagged samples
[25], [26]. While effective, DVD is more difficult to implement than standard procedures, and does
not guarantee void-free parts [27]. The difficulties with producing high quality laminates are
magnified when performing repairs in the field.
When performing a repair in the field, the challenges associated with OOA processing are
magnified. The resources and infrastructure typically present in production environments are
generally unavailable in the field. However, the ability to perform high-quality repairs in the field
is needed to reduce downtime in depots and extend the service life of aircraft and similar systems.
Currently there is no satisfactory solution for structural repairs in the field, where conditions are
challenging [80]. For example, freezers are generally absent in the field, precluding the possibility
of storing prepregs for repairs. Low-temperature storage is required to retard the cure reaction that
progresses even at room temperature [102]. Thus, in-field repairs commonly rely on wet layup,
sacrificing the uniformity and consistency intrinsic to prepregged laminates [103]. Each ply is laid
up by hand using resin mixed on-site and spread across the fabric surface. Composite repairs
performed by wet lay-up invariably lack the consistency, quality, and performance levels typically
achieved with autoclave cure of prepregs [79]. DVD processing mitigates these challenges
partially, but is difficult to deploy in the field [104]. Requiring depot level facilities to perform
repairs obviates any flexibility. In-field repairs have the most flexibility yet are unable to be
consistently employed because current OOA prepregs are not designed for in-field conditions.
A new material system is presented that addresses the issues with in-field repairs describe
above. Although the material also can be used for production of original parts, it directly addresses
the process difficulties and inferior patch quality associated with in-field repair conditions. The
material is a semi-preg that features a resin format (distribution) that allows for through-thickness
80
gas permeability [105]. Because incomplete air evacuation is a major cause of surface and bulk
porosity, this format expedites gas egress and effectively eliminates porosity in composite parts,
especially in repair conditions. Furthermore, the semi-preg format allows for conventional vacuum
bagging, eliminating the need for DVD. An additional feature of the material system is a vinyl
hybrid resin which is stable at room temperature and thus does not require freezer storage. These
features – semi-preg that does not require freezer storage – are well-suited to in-field repairs of
composite structures.
This study describes the processing of a semi-preg system suited to repair-relevant VBO
conditions. The material system was contrasted with the conventional repair procedures of wet
layup and DVD. The material system was also contrasted with conventional formatted prepregs.
Composite repair patches with no apparent porosity were produced with the semi-preg system,
while DVD/wet layup and conventional prepreg panels had void defects in the laminates.
6.2 Methods
This study involves two separate comparisons against the vinyl hybrid semi-preg system.
The first is a comparison of the new material system versus repairs conducted with double
vacuum debulking (DVD) with wet layup. The panels created will be labeled using a letter
system (A-D). The second is a comparison of the new material system against a conventionally
formatted epoxy prepreg system. The panels created for that comparison will be labeled with a
numerical system (1-4).
81
For the DVD comparison, two matrix materials were employed for this study, a vinyl hybrid
resin and a two-part epoxy resin for wet layup. Using these resins, four types of panels were
produced and analyzed. Two panels (A and B) were produced using the prototype semi-preg vinyl
hybrid resin and VBO processing, heretofore referred to as semi-preg panels. Two more panels (C
and D) were produced using the epoxy resin and wet layup/DVD, which we will refer to as DVD
panels. A and C panels were processed on a tool plate with sealed edges, while B and D panels
were processed on a scarf repair tool. Thus, the four panels were A (semi-preg tool plate), B (semi-
preg repair), C (DVD tool plate), and D (DVD repair). The DVD panels (C and D) were fabricated
using current state-of-the-art procedures for in-field repair, while the semi-preg panels (A and B)
were cured using VBO processing. Panel quality was analyzed by porosity measurements of
polished sections for defect analysis and by ultrasonic C-scans for porosity distribution.
Mechanical performance of the panels was evaluated through measurements of interlaminar shear
strength (ILSS) and glass transition temperature (Tg).
All panels were prepared using a carbon fiber 2 ×2 twill fabric with 6k tows (DowAksa A-
38) and a resin content of 35%. An 8-ply quasi-isotropic stacking sequence of [0°/45°]2s was
employed for all the panels. The DVD panels employed an epoxy adhesive paste commonly used
for co-cured scarf repairs (Henkel Loctite EA 9390) [101]. The layup procedure was performed
using techniques commonly used for wet layup repairs, including the use of DVD [106]. Panels
fabricated from semi-preg featured a vinyl hybrid resin (VH-37, Polynt) with discontinuous resin
formatting, produced commercially (Tipton Goss Advanced Materials Company) [105]. The
distribution of the resin across the surface is influenced by the contours of the fabric. Thus, the
2x2 twill fabric yields a resin pattern comprised of islands of resin on the surface, as shown in
Figure 6-3a. The formatting allows for edges to be sealed as seen in Figure 6-3b.
82
Figure 6-3: a) Semi-preg formatting on prepreg surface with islands of resin b) sealed edges of test panels that
simulate the sealed edges of patch repairs
DVD panels were fabricated by wet layup and DVD procedures. An external advisor with
extensive wet layup/DVD experience was consulted with to ensure compliance with state-of-the-
art repair protocols [106].
For Panel C (DVD over tool plate), nonperforated fluorinated ethylene propylene (FEP)
release film (Airtech A4000) was taped to an aluminum tool plate and a layer of Teflon-coated
fiberglass peel ply was layered on top. Eight dry plies (420×420) mm were laid up in a quasi-
isotropic stacking sequence [0°/45°]2s with a layer of resin (EA 9390) between each ply. Perforated
release film (Airtech A4000 P8) and a peel ply was placed on top of the laminate. Edge-breathing
was allowed for Panel C, and fiberglass bleeder plies were employed, with nylon breather cloth
(Airtech Airweave N10) covering the laminate. Vacuum bagging was then overlaid on the surface
and sealed with sealant tape (Airtech GS213-3). A wooden box with an open end was placed over
this setup with a breather blanket on top. Another vacuum bag was placed over top of the box.
While under the DVD box, the three cycles were applied to the laminate - debulking, compaction,
2 mm
83
and cure. Vacuum was applied to both bags, and the debulk step was 50°C for 60 minutes. The
debulk box vacuum was then vented to allow for compaction for 30 minutes. Final cure was
achieved in an oven with 2°C/min ramp to 118°C and held for 150 minutes.
Panel A (semi-preg on tool plate) was fabricated using VBO layup and bagging. Non-
perforated release film was laid onto a flat aluminum tool. Prepreg plies measuring 380×380mm
were laid up in the same stacking sequence as the DVD panel. Unlike the DVD panel, edge
breathing was prevented by sealing the edges of the prepreg using sealant tape. The intent was to
simulate the conditions of a scarfed repair, with restricted air evacuation. Two layers of sealant
tape were used to fully seal the edges of the panel (Figure 6-3b). Teflon-coated fiberglass peel ply,
perforated FEP release film and a layer of breather cloth were put on top of the laminate. The
perforations in this release film were spaced every 50mm to limit resin bleed due to the low
viscosity of the resin. The sealed vacuum bag assembly was cured in a convection oven (LR
Environmental Blue M). There was first a room temperature debulk for 30 minutes followed by a
1.1°C ramp to 93°C and held for 30 minutes. A second ramp of 1.1°C until 121°C was done and
held for 60 minutes. A cool down of 1.1°C/min was done until room temperature was reached.
The temperature was measured using a K-type thermocouple on top of the sample. The same
sizing, release agent and heat blanket were employed as the DVD panel.
Similar procedures were employed for Panels B and D (semi-preg and DVD panels) with
the scarf repair tool with the following differences. Instead of 420×420 mm plies, the dimensions
of the scarf repair baseline (50×50mm) was the starting point for ply sizing. Each successive ply
was cut proportionally large to account for the 3° scarf incline chosen to replicate best practices
for repairs [107], [108]. Instead of an opaque FEP release film, a transparent release agent
(Frekote® 700-NC) was employed. Finally, a heat blanket (Briskheat SR512018X18C) was
84
employed to deliver the same temperature cycle instead of an oven. A custom control system for
temperature control was built using a controller (Watlow PM6R1CA-AAAAAAA), solid state
relay (SSR-240-10A-DC1), and K-type thermocouple input. The temperature inside the sample
was determined using calibrations of test runs on the blanket system.
To ensure consistent values for glass transition temperature and ILSS, steps were taken to
ensure all samples were at full cure. Samples from each panel were postcured, and results from
postcured samples were compared to that of the original panels. Postcure consisted of heating
samples to 121°C at 1.1°C/min then heating to 149°C at 0.4°C/min and holding for 60 minutes
(semi-preg) or for 120 minutes (DVD). Results showed that the semi-preg samples remained
similar, while DVD panel property values increased with postcure. Thus, DVD samples were
postcured prior to subsequent testing while semi-preg samples were not postcured.
For the comparison of the new material to a conventional prepreg system, two matrix
materials were selected, including a vinyl hybrid resin and a conventional prepreg epoxy. Using
these resins, four types of panels were produced and analyzed. Panels 1 and 2 were produced
using the prototype vinyl hybrid resin. Panel 1 featured semi-preg formatting, while Panel 2
featured conventional OoA prepreg formatting. Two more panels (Panels 3 and 4) were produced
using the epoxy resin. Panel 3 featured semi-preg formatting, while Panel 4 featured
conventional OoA prepreg formatting. Thus, the four panels consisted of Panel 1 (semi-preg
vinyl hybrid), Panel 2 (conventional vinyl hybrid), Panel 3 (semi-preg epoxy), and Panel 4
(conventional epoxy). All panels were cured using VBO processing on a scarf repair tool
Mechanical performance of the panels was evaluated through measurements of interlaminar
shear strength (ILSS). Panel quality was assessed by porosity measurements of polished sections.
85
Panels with vinyl hybrid were fabricated using resin films produced commercially
(Tipton Goss Advanced Materials Company) [105], while panels with epoxy resin were also
fabricated using resin films produced commercially (PMT-F4, Patz Materials & Technology). To
produce semi-preg formatting, the conventional films were dewetted after nucleation sites were
introduced with a hand-held spike roller [60]. The dewetting was conducted on silicone paper in
a convection oven (Blue M Oven). The resin was heated in the oven for 2 minutes at 104°C until
dewetting occurred and established the semi-preg format [61]. Prepreg was then fabricated by
transferring either the formatted resin or continuous films to dry fabric.
All prepregs (1-4) were prepared using a 2x2 twill fabric of carbon fibers with 6k tows
(DowAksa A-38) with resin contents of 35%. A stacking sequence of [0°/45°]2s for 8 plies was
used for all panels. The size baseline scarf tool plate (50x50mm) was the initial ply size.
Successive plies were cut proportionally larger to account for the 3° scarf incline to replicate best
practices for repairs [107], [108]. A transparent release agent (Frekote® 700-NC) was applied to
the tool surface before stacking the plies. Perforated release film (Airtech A4000 P8) and a
Teflon-coated fiberglass peel ply were placed on top of the laminate, and a nylon breather cloth
(Airtech Airweave N10) covered the laminate. Vacuum bagging was then overlaid on the surface
and peripherally sealed with sealant tape (Airtech GS213-3). A heat blanket (Briskheat
SR512018X18C) was used to heat the laminates after applying vacuum to the bag. A custom
control system for temperature control was built using a controller (Watlow PM6R1CA-
AAAAAAA), solid state relay (SSR-240-10A-DC1), and K-type thermocouple input. The
temperature inside the sample was determined using calibrations of test runs on the blanket
system. For the vinyl hybrid panels, the cycle began with a room temperature debulk for 30
86
minutes, followed by a 5°C/min ramp to 93°C and a 30-minute dwell. A second ramp at 5°C/min
to 121°C was applied and held for 30 minutes. Cured laminates were cooled to 20°C at
1.1°C/min. The same cure cycle was used for the epoxy panels, except that the final hold at
121°C was two hours to ensure full cure (unlike the vinyl hybrid, the epoxy was not fast-cure).
Panel quality distributions was assessed by ultrasound. Ultrasonic C-scans were performed
on the DVD and semi-preg tool plate panels to assess microstructural uniformity (NDT
Automation UPK-T36). A transducer with a 10 MHz frequency scanned the panel while being
gated by the echo from a glass reflector plate. The resulting scans were transcribed into the images
used in this study.
Void defect analysis was conducted using polished section microscopy. Polished sections
(25×13 mm) were prepared all panels in both the DVD/set layup comparison and conventionally
formatted prepreg comparison for microscopic inspection to assess porosity. Cross sections of the
panels were polished (Struers Labopol 2) using abrasive paper up to 4000 grit. Images were
acquired using a microscope (Keyence VHX 5000) at 200× magnification and collated using built-
in functions. Void content was estimated by measuring the ratio of void area to total area of each
sample. The void areas were selected by dark area filters on the microscope which were the same
for each sample and verified visually.
Panel properties were assessed by measuring Tg and shear strength. Dynamic mechanical
analysis (DMA, TA Instruments Q800) was used to measure glass transition temperature (Tg) of
cured panels. Samples (60×12 mm) were cut from the panels and tested in dual cantilever mode
while heating at 3°C/min to 200°C. The value of Tg were identified as the peak of the tangent delta
curve. Interlaminar shear strength was measured using the short-beam-shear (SBS) method
87
outlined in ASTM D2344. Samples cut from panels in both the DVD/wet layup comparison and
the conventionally formatted prepreg comparison were tested using a load frame (Instron 5585H).
6.3 Results
In the comparison between the new material system and DVD/wet layup, data to compare
the quality of fabricated panels was taken. Representative ultrasonic C-scans of the semi-preg
(Panel A) and DVD (Panel C) panels fabricated on the tool plate are shown in Figure 6-4.
Figure 6-4: Ultrasound images of Panels A and C (semi-preg and DVD). Panel A shows spatially consistent void
free areas while Panel C has an inconsistent interior.
C-scans from the DVD panel exhibit varying shades of blue with regions of red and yellow
across the panel, indicating non-uniform attenuation and the likelihood of porosity distributed in
these regions. The red/yellow areas appear in central regions of the DVD panel. This observation
is consistent with findings from composites produced from OoA prepregs, where air bubbles are
88
often trapped near the center of panels because it is more difficult to reach panel edges (edge
breathing is the primary air evacuation mechanism for such prepregs) [22]. The need for
technologies addressing this shortcoming of DVD processing becomes apparent upon discovery
that even the relatively small and thin panels produced here exhibited unacceptable levels of
porosity. In contrast, the panels produced from semi-preg were markedly more uniform than the
DVD panels. The findings demonstrate the benefit of prepregs with through-thickness gas
permeability to laminate quality when curing in the absence of autoclave pressure. The reliance on
through-thickness gas transport, as opposed to edge breathing, imparts a multitude of much shorter,
redundant pathways for gas egress compared to conventional OoA prepregs, as well as inherent
process robustness and compatibility with a range of constraints stemming from geometry and cure
conditions. Polished sections of cured laminates afford an opportunity for quantitative analysis of
porosity. Representative micrographs from polished sections of Panels A and C are shown in
Figure 6-5.
Figure 6-5 - Micrograph of DVD (Panel C) and semi-preg (Panel A) polished sections. The DVD panels exhibited
porosity while the semi-preg panels were void-free.
Panel C (DVD) exhibited average porosity values of 2.2% across different areas of the
panel. Both macro- and micro-porosity were evident. Macroporosity resulted from air trapped
1.0 mm
Semi-preg
DVD
1.0 mm
89
within the laminate that failed to reach an outlet, while microporosity appeared within the carbon
fiber tows and stemmed from insufficient resin flow into the tows, either because of insufficient
time or insufficiently low viscosity. The relatively low viscosity of the resin argues against the
latter possibility (steps were required to prevent excessive resin loss via bleeding during cure), and
thus entrapped air was the likely cause of both macro- and micro-porosity. Ultrasonic C-scans of
the DVD panel revealed that porosity was distributed throughout the sample, although the
concentration was higher in central regions.
Comparing the results of the DVD panel with those of the semi-preg panel reveals marked
contrasts. The latter panel exhibits a conspicuous absence of both macro- and micro-porosity (<
0.1%). The absence of macroporosity itself can be attributed to more effective and efficient air
evacuation during cure. In contrast, microporosity present in the DVD panel was attributed to the
opposite cause – insufficient gas evacuation. Remarkably, the absence of porosity in panels
fabricated with semi-preg occurred despite both a shorter cure cycle and a much simpler setup than
DVD. The contrast demonstrates the effectiveness of the semi-preg format in achieving gas
evacuation despite sealed edges. Furthermore, the low viscosity of the vinyl hybrid resin (relative
to conventional epoxies) achieved tow impregnation while preserving multiple pathways for gas
egress.
Panels B and D were made on a scarf repair tool. Thus, the edges were sealed by the
geometry of the processing environment. However, the shape of the fabricated laminate makes
ultrasound analysis unfeasible. Representative micrographs from polished sections of Panel D
(DVD) and Panel B (semi-preg) panels are shown in Figure 6-6.
90
Figure 6-6: Microscopy of scarf panels. The DVD panel (Panel D) exhibited porosity while the semi-preg panel
(Panel B) was void-free.
Panel D (DVD) exhibited porosity levels similar to the counterparts produced on the tool
plate. The DVD panel showed 1.9% porosity with a similar amount of macroporosity. However,
it did not exhibit the same microporosity characteristics as the tool plate panels. The repair tool
plate has distinguishing features, including the scarfing along edges and the central observation
window, which accommodates a much smaller repair patch. The shorter travel distance and
potential for additional air pathways along the scarf edges, in contrast to sealed edge dams explain
the lower porosity compared to the tool plate DVD samples. The semi-preg panels exhibited 0.1%
porosity on average, usually with no apparent voids. Air bubbles present in the semi-preg were
evacuated prior to full cure.
Porosity analysis provides one metric to compare the quality of repairs performed with
the semi-preg vinyl hybrid system versus the DVD wet layup configuration. However, the
performance of laminates produced with a new material system can be benchmarked to those
produced with conventional epoxy, such as the one employed in Panels C and D. Figure 6-7
1.0 mm
DVD
Semi-preg 1.0 mm
91
shows the glass transition temperature (Tg) for panels produced with semi-preg (Panel A) and
those produced via DVD (Panel C), with error bars of one standard deviation.
Figure 6-7: Glass transition temperatures for Panels A (semi-preg) and C (DVD). The epoxy had a higher glass
transition temperature than the vinyl hybrid resin making it more viable for higher temperature applications.
Glass transition temperature is often used as an indicator of mechanical performance of a
resin [109]. Note that the epoxy has a higher Tg than the vinyl hybrid (179.6°C vs 160.1°C),
beyond the 95% significant level. Because the vinyl hybrid is a prototype formulation and not
optimized, it is not surprising that the Tg for vinyl hybrid is slightly lower than the commercial
epoxy. To determine the strength difference interlaminar shear strength (ILSS) is shown in
92
Figure 6-8: Interlaminar shear strength for Panels A (semi-preg) and C (DVD). As expected, the epoxy had a higher
interlaminar shear strength than the vinyl hybrid resin.
Interlaminar shear strength (ILSS) is also an important metric for mechanical
performance. ILSS is a matrix-dominated metric, and is characterized by multi-axial loading, a
condition most components experience in service. Porosity has a strong effect on ILSS and other
matrix properties [67]. As expected after viewing the Tg results, the ILSS values for epoxy were
greater than those of the vinyl hybrid (44.7 MPa vs 40.1 MPa). However, the difference was not
statistically different at the 95% confidence level (unlike Tg values). Modifications to the resin
formulations, both vinyl hybrid and epoxy, could increase performance levels to more closely
approximate that of the wet layup epoxy.
In the comparison between the new material system and a conventionally formatted
prepreg, data to assess the quality differences between fabricated panels was gathered.
Micrographs (Panels 1-4) show the difference in void characteristics between the four panels
(Figure 6-9).
93
Figure 6-9: Micrographs of fabricated panels. The panels fabricated with semi-preg formatting exhibited fewer
macro- and microvoids than those with conventional formatting.
Panel 1 exhibited negligible porosity, stemming from its semi-preg formatting and low
viscosity resin. By contrast, Panel 2 with its conventional formatting had higher levels of void
defects. The bubbles seen Figure 7-7b-d, did not fully evacuate through the laminate and remained
as macroporosity. Panel 3 with semi-preg formatting and epoxy resin contained reduced levels of
defects when compared to Panel 2, indicating that while low viscosity resins can assist in reducing
defect levels, semi-preg formatting is more effective at lowering porosity. The number and location
of macroporosity in Panel 3, matched the bubbles in the video samples from Figure 7-6. Panel 4
Panel 1 1.0 mm
Panel 2
Panel 3
Panel 4
1.0 mm
1.0 mm
1.0 mm
94
with its higher viscosity resin and conventional formatting exhibited higher levels of both micro-
and macro-porosity.
The microscopy graph quantifies the porosity trends seen in the microscopy samples
(Figure 6-10).
Figure 6-10: Quality of fabricated panels shown by percent porosity. The semi-preg formatted panels had
significantly lower porosity than the conventionally formatted panels.
Panel 1 (vinyl hybrid semi-preg) exhibited 0.1% porosity with most samples showing no
void defects. Panel 2 had 2.2% void content, significantly greater than Panel 1. The sole difference
between the two panels, resin formatting, was responsible for the presence of higher levels of
macroporosity in Panel 2. Panel 3 had lower porosity than Panel 2 with 1.2%, despite having a
higher viscosity resin (epoxy) and a cure cycle with a lower flow number. The semi-preg
formatting of Panel 3 allowed most gas to escape prior to saturation akin to Panel 1, leading to a
reduction in void content. Finally, Panel 4 exhibited 8.4% porosity with the widest range between
samples. The combination of repair conditions, conventional formatting, and lower viscosity
95
epoxy resin, resulted in a level of porosity that would be rejected on an in-field repair despite lower
quality standards [106].
Figure 6-11: Short beam strength of Panels 1-4. The panels demonstrate the inverse correlation between the void
content and strength of laminates.
Interlaminar shear strength is an important measure for mechanical performance because
ILSS is matrix-dominated and represents most structures in service with multi-axial loading. Figure
6-11 highlights the differences between the strength of epoxy and vinyl hybrid. Panels 3 and 4
(epoxy), despite having higher void content than Panels 1 and 2 respectively, exhibited higher
strength than the vinyl hybrid panels. The drop-off in strength due to porosity between Panels 3
and 4 is greater than the strength knockdown in Panels 1 and 2, in line with the greater increase in
porosity [50]. Because vinyl hybrid is a prototype resin, lower strength is expected before
refinements to the formulation are made.
96
6.4 Conclusions
A constant concern for in-field soft repairs is the inconsistent quality of repaired panels, as
well as the complexity of current state-of-the-art repair methods. A new material system was
introduced to address these concerns, and was evaluated for suitability in soft patch scarf repairs.
The semi-preg formatting of the material system addressed the concern of panel quality.
Employing a semi-preg format effectively eliminates the need for DVD processing by providing
redundant pathways for through-thickness air evacuation. These pathways are so effective that
they eliminate porosity, even when edge breathing is not possible, an attribute well-suited to in-
field repairs. The material system also features a vinyl hybrid resin that eliminates the need for
refrigerated storage and is well-suited to wet layup. These two features allow for high quality
repairs with a simple bagging process. The semi-preg panels demonstrated consistent absence of
porosity, while wet laid DVD panels showed both micro- and macro-porosity.
Semi-preg formatting coupled with vinyl hybrid creates an effective solution for robust
patch repairs when compared to conventionally formatted prepregs. Analysis from fabricated
panels confirmed the complementary effects of semi-preg format and vinyl hybrid resin on
reducing porosity by simultaneously promoting gas evacuation and bubble migration. The sealed
edges of the scarf did not present an air evacuation problem for the semi-preg materials like it
does for conventional prepreg composites. Panels exhibited porosity levels of 0-9% depending
on the resin type and format used. The vinyl hybrid with semi-preg formatting had the lowest
void content while the epoxy with conventional formatting had the highest.
97
The material system presented in this work provides an opportunity to perform simple in
field repairs with minimal processing equipment. The resin is room temperature stable, obviating
the freezer storage requirement. The semi-preg format eliminates the need for double vacuum
bag (DVD) processing, often required for wet layup or other repair procedures [110]. Although
the resin has lower strength than epoxy counterparts, the simplicity of processing the semi-preg
and getting a void-free part represents a fundamental change in how repairs could be approached.
Improvements to the formulation to increase interlaminar shear strength would result in a high
performance prepreg able to be easily employed for in-field repair, something that does not
currently exist. While the semi-preg format and resin were designed for in field-repair, they can
be employed in manufacturing to broaden the composites application space.
98
7 CHAPTER 7: In situ Analysis of Prepreg Processing in Repair Environments
7.1 Introduction
The objective of the present work was to employ in-situ analysis to assess the effect of semi-
preg formatting on air evacuation and porosity for in-field scarf patch repair conditions. A custom-
built tool featuring a transparent window was designed and deployed to provide insight into the
curing process of a scarfed patch. Semi-preg featuring a discontinuous pattern of room-
temperature-stable resin on a woven fiber bed was produced and used to simulate in-field repairs.
The in-situ analysis was performed on the panels fabricated in Chapter 6.
Porosity in composites can be an initiation site for damage in composite structures. Voids
are the most common manufacturing defects in prepreg processing [111]. Void defects in
composites laminates act as stress concentrators, lowering the strength of the overall structure
[50], [67], [112]. The two types of strengths affected the most by porosity are the matrix
dominated properties of interlaminar shear strength and flexural strength [112]. For every 1% of
porosity, a knockdown in about 5% of strength can be found. Percent porosity can often dictate
whether a manufactured or repaired laminate is deemed acceptable for service [113]. In
recognition of difficult quality control for repairs, the acceptable levels of void defect
percentages are higher in repairs than in manufactured components. In applications such as wind
energy, porosity can be prevalent in areas of the turbine blades that do not carry structural loads
[114]. Thus, the tolerances for repair patches on blades are much looser than tolerances in
aerospace. In autoclave processed prepregs, voids are forced into solution by virtue of additional
pressure, leaving few voids after processing [65]. The tolerance for manufactured components
99
made out-of-autoclave are considered with respect to the autoclave standard. Values of 1-2%
porosity can be deemed acceptable in those composite laminates. With repairs on load-bearing
structures, tolerances can be as high as 4% [106]. Detecting the presence of voids or other
damage within the laminates has been a focus in composite maintenance.
The accurate assessment of damage and defects in composites is essential for successful
maintenance and repair of composite structures [115]. Given the difficulty to assess defects with
visual inspection of the cured laminate surface, many alternative non-destructive evaluation
(NDE) or non-destructive testing (NDT) methods have been explored. They can be broken into
two main categories: contact and non-contact, although some methods include elements of both.
Contact methods, as the name implies, require contact between the testing equipment and
the composite structure. They include some of the oldest ways of determining the existence of
damage in composite structures. The simplest test that is still performed in the field is the tap
test, a form of acoustic emission analysis [116]. It involves a technician tapping the structure and
listening to the acoustic feedback. This method is imprecise and relies on the experience and skill
of the technician. This method may be used as a preliminary test but yields little useable data on
defects in the laminates. Ultrasonic inspection is the most common contact NDE method
employed for detecting damages [78]–[80]. This method involves transmitting high frequency
sound (1 to 5 Mhz) into the part and monitoring received energy [32]. The need for a couplant,
water, or oil, is a downside of this process, as it introduces the tested laminate to potential
contamination or moisture. Ultrasound can be conducted relatively quickly and can provide high
resolution feedback on the presence of damage or voids in panels [33]. It has been used to show
the distribution of porosity in fabricated panels in Chapter 6.
100
Non-contact methods, as the name implies, do not rely on contact between the laminate
and testing equipment. Radiography, specifically x-ray radiography, is the most common non-
contact NDE method [117]. The test sends short wavelength electromagnetic radiation to
determine defect content. The wavelength of choice depends on the thickness of the laminate,
with shorter wavelengths (gamma) being used on thicker laminates. The laminate will absorb
different amounts of radiation depending on the uniformity. There are many varieties of the
method including, computed radiography, digital radioscopy, and computed tomography.
Related to radiography is micro CT scanning [118]. It provides 3D images with high resolution
of the interior of panels and is frequently used in research to get detailed images of damaged or
tested parts, which can help with insights into damage formation and defect analysis [119],
[120]. Thermography is a method that uses infrared radiation to detect damage in composite
parts [94]. The thermal decay in the composite laminate is measured and analyzed, relying on a
composite structures’ low thermal conductivity to provide contrast. This type of testing can be
used to assess large areas of laminates and has passive and active forms. Two main forms of
thermography exist: passive and active. Thermography has the ability to measure large parts, but
can lack the small scale precision of other tests [122]. Shearography is a technique that uses laser
illumination to identify internal damage [122]–[125]. A laser pattern on a composite sample is
recorded with and without stress to measure the surface differences [125]. Electromagnetic (ET)
testing methods have been employed to evaluate composite defects. ET methods involve
inducing electromagnetic fields in the composite panel and observing the electromagnetic
responses.
Uniting all the contact and non-contact methods of analyzing composite laminates for
repairs is that they all occur on fully cured panels. While the post-cured data on damage or
101
defects are useful to assess repair materials and procedures, in-situ can provide additional insight
on how those defects were formed in the laminate. However, composite repairs have been
difficult to analyze while the laminate is being cured. The bagging setup for curing composites
on parent materials is meant to isolate the patch from the outside environmental and makes
performing analysis during processing demanding. When procedures such as DVD are
employed, those challenges become magnified. Insight on what is happening in the composite
laminate while under vacuum and being cured is crucial for understanding where porosity comes
from and how to minimize void defects. Thus in-situ methods of analysis desired to analyze
composite repairs.
In many forms of composites manufacturing, in-situ analysis has been employed to
generate insights into the cure process for laminates. The methods of in-situ analysis have a
commonality in that a viewing window is introduced into a relevant process. A transparent
acrylic surface has been used to monitor dual scale resin flow with fabric fiber reinforcements
[68]. A similar setup has been used to investigate bubble motion in fabric prepregs while under
constant temperature and pressure conditions [64]. Glass plates have also been used as a window
into the curing process. The surface porosity of prepregs was analyzed using a glass plate and
cameras for insight on surface porosity formation in laminates [98]. Bubble migration was
analyzed in unidirectional (UD) prepregs using a glass plate as well [126]. Sandwich structures
have also been analyzed in autoclave conditions [127]. In each of those studies, the ability to
view the material while it is curing was vital to draw insights on the processes being analyzed.
Composite repair has many unique aspects that differentiate it from more standard composite
manufacturing. Additionally, semi-preg formatting introduces another dimension of air
evacuation which has yet to be analyzed in-situ.
102
In this study, a tool that enables in-situ of a scarf repair was created. The tool was
employed to analyze semi-preg formatted material. The semi-preg laminates were compared to
laminates fabricated with DVD/wet layup procedures and conventionally formatted prepregs.
Videos of the laminates during processing were taken to generate insight into the defect
formation process.
7.2 Methods
To allow in-situ observation of resin flow and void formation, a tool plate featuring an
observation window was custom-built. The aluminum base plates were machined to simulate a
scarf profile and to accommodate a glass observation window (Figure 7-1), which was clamped
and sealed to prevent air leakage.
Figure 7-1: Schematic of repair tool plate with an in-situ observation window. A 3° scarf angle was incorporated to
follow best practices in repair procedures.
103
The distinctive features of the tool are the central observation window and the 3° scarf
incline. The legs of the tool accommodate video camera and illumination during the cure cycles.
Images were recorded at 1 Hz during the early stage of cure, and every ten seconds thereafter until
resin flow ceased.
The samples in this study were fabricated the same way as samples in Chapter 6. As such,
the procedures used to fabricate the panels can be found in the methods section of that chapter.
The naming scheme behind the panels will also remain the same. Panels A-D were part of a
comparison between a vinyl hybrid semi-preg and DVD/wet layup processing. Panels B and D
were done on the scarf tool plate, while Panels A and C were done on a standard tool plate. Panels
1-4 were part of a comparison between the vinyl hybrid semi-preg and conventionally formatted
epoxy prepregs. All four of those panels were fabricated on the scarf tool.
7.3 Results
The scarf tool was employed to assess the variations between DVD/wet layup and semi-preg
VBO processing. However, prior to its deployment, the samples made on a standard tool plate
were analyzed (Panels A and C). The difference in the surface quality of the Panels A and C
fabricated on the standard tool plate is apparent (Figure 7-2).
104
Figure 7-2: Surface images of DVD and semi-preg panels. The semi-preg Panel A do not exhibit the same level of
surface porosity as the DVD panel C.
Panel C exhibited surface voids concentrated in regions between tows. In twill fabrics, the
spaces between tows offered pathways of least resistance for pockets of air, and local pressure
gradients drove gas bubbles to these sinks. Bubbles were often trapped at these sites and remained
as surface defects. Wet layup covered each ply with resin and thus sealed the plies against through-
thickness gas transport, resulting in gas bubbles coalescing in depressions associated with tow
cross-overs. Without through thickness pathways, gas egress can occur only at ply edges (generally
orders of magnitude farther away). Whether these voids were present throughout the cure process
or evolved during processing is unclear. For this reason, the scarf tool plate was designed to furnish
insight into gas transport during scarf repair processing. In the semi-preg panel, no such surface
defects remained. The biggest difference between the panels was the ability for air to evacuate in
the through thickness direction in the semi-preg panels. Thus, semi-preg formatting can explain
the difference in surface quality between the panels.
DVD Semi-preg
105
The images below were selected from video frames recorded during DVD cure on the scarf repair
tool plate in Panel D (Figure 7-3).
Figure 7-3: In-situ images of DVD processing. Void formation and migration on the surface of the laminate is seen
in these images.
The initial distribution of resin after wet layup is shown in Figure 7-3a. Although some
regions are not entirely covered in resin, the distribution does not provide the multitude of
redundant, through-thickness air evacuation pathways as in the semi-preg layups. When the DVD
apparatus is deployed and the cure cycle initiated, the resin begins to wet out the panel, and some
uncovered regions begin to shrink (Figure 7-3b). At this stage, the benefit of DVD versus VBO
2 mm
2 mm 2 mm
2 mm
a
c
b
d
106
processing for wet laid materials becomes apparent from the shrinkage of air pockets. In the DVD
process, steps are taken to prevent compaction that could potentially block air evacuation channels
in the laminate. However, air pockets remain on the sample, mostly in gaps between tows. The
limits of wet layup for fabricating panels with low porosity become apparent as the nearly
continuous layer of resin prevented air evacuation during debulking. A representative air pocket is
circled in red (Figure 7-3b-d) to highlight movements during cure. In Figure 7-3c, the air pocket
has moved down along inter-tow channels towards the panel edge nearest to the vacuum port.
Eventually the bubble settles into its final position (Figure 7-3d). The final site is separated from
its original position by multiple tows, revealing the absence of through-thickness permeability
since the bubble migrated only in-plane. Moreover, the bubble size did not change from Figure
7-3b-d, indicating that no gas escaped, despite the DVD process seeking to preserve in-plane
pathways for egress. The wet layup process renders each ply independent with respect to air
evacuation. Similar events occur in all plies, resulting in internal porosity in the cured laminates.
The images from Panel B below (Figure 7-4) help demonstrate how processing semi-pregs differs
from prepreg formats that rely on edge breathing and from DVD.
107
Figure 7-4: In-situ images of semi-preg processing. Fewer voids remain after resin consolidation and are able to
escape through the laminate.
Figure 7-4a shows the initial distribution of resin in the semi-preg, comprised of islands of
resin separated by gaps for through-thickness gas egress. During cure cycle heating, the resin
spread across the sample while compaction began. Although much of the air escapes quickly
(especially compared to the DVD panel), some bubbles remained in the resin prior to gelation
(highlighted with a red circle in Figure 7-4b-c. The bubble was initially situated squarely in the
center of a tow, and this factor prevented rapid removal through gaps and pinholes like all other
air pockets. As the resin continued to spread and infiltrate fiber tows, the bubble migrated to the
edge of a tow, elongating in the process. The bubble then migrated between fiber tows until
2 mm a 2 mm b
2 mm c 2 mm d
108
reaching a pinhole at a corner. Immediately upon reaching the corner, the bubble began to shrink
such that most of the air in the elongated bubble traveling along the edge escaped before it was
able to settle into a roughly hemispherical shape in the corner against the plate. Thus, the bubble
was barely perceptible Figure 7-4c in the corner before eventually disappearing.
The elimination of gas bubbles during cure of semi-preg panels demonstrated the
effectiveness of formatted prepregs. Figure 7-4d shows the defect-free surface that remained post-
gelation. Complete removal of gas bubbles was achieved much earlier in the cure cycle relative to
the DVD panel. Note that the bubbles disappeared through the panel, unlike the DVD panel (Figure
7-3). Polished sections of cured panels revealed zero bulk porosity, supporting the contention that
bubbles that disappeared from the surface also exited the panel. The resin distribution on the
surface ensured that spots at tow corners were among the last to be filled by resin, and therefore
acted as through-thickness air evacuation channels for bubbles. Because air removal occurred via
short and redundant through-thickness pathways (as opposed to edge breathing), porosity-free
laminates can be produced in conditions where edge breathing is impaired or prevented, such as
large parts and in-field repairs.
The scarf repair tool was instrumental in analyzing the difference in processing semi-preg
and conventional prepregs. To observe resin flow and air evacuation behavior, samples were cured
on the repair tool with the in-situ observation window. The images below were selected from video
frames taken on Panel 1, the vinyl hybrid semi-preg laminate (Figure 7-5).
109
Figure 7-5: Images acquired in-situ during processing (Panel 1). Any bubbles remaining after consolidation were
evacuated through the thickness of the laminate.
Figure 7-5a shows the initial formatting of resin in the semi-preg, comprised of a resin grid
with islands of dry gaps for through-thickness egress. The resin spread across the sample while
compaction commenced. While much of the air evacuated quickly, some bubbles remained in the
resin prior to gelation (circle in Figure 7-5c). Despite full saturation, the circled bubble was able
to shrink in the corner of the fiber tows before eventually disappearing (Figure 7-5d). The low
viscosity of vinyl hybrid allowed easier bubble migration along the fiber tows through the
thickness of the laminate. The reduction and eventual elimination of gas bubbles during cure of
Panel 1 demonstrated the efficacy of semi-preg formatting in repair environments, especially when
combined with a resin that allows for high flow number cure cycles.
2 mm
2 mm
2 mm
2 mm
a b
c d
Remaining
bubble
Evacuated
bubble
110
The images below were selected from video frames taken on Panel 3, the epoxy semi-preg
laminate (Figure 7-6).
Figure 7-6: Images acquired in-situ during processing (Panel 3). Bubbles that remain after consolidation are not able
to evacuate through the thickness of the laminate and remain as surface porosity.
Figure 7-6a shows a similar resin formatting to that of Figure 7-5a. The semi-preg format
initially allowed air to escape, reducing the number of bubbles on the laminate surface. When the
resin reached full saturation, more bubbles remained than that of the vinyl hybrid panel. One such
bubble is highlighted in the circle in Figure 7-6c. However, unlike the vinyl hybrid panel, this
111
bubble was unable to escape through the thickness of the laminate (Figure 7-6d). The bubble
remained, and became surface porosity. Bubbles were also trapped in the thickness of the laminate.
The images below were selected from video frames taken on Panel 2, the vinyl hybrid
conventional laminate (Figure 7-7).
Figure 7-7: Images acquired in-situ during processing (Panel 2). Many bubbles remain after laminate consolidation
and remain as surface porosity despite the lower viscosity of the vinyl hybrid resin.
Figure 7-7a shows the initial distribution of resin fully covering surface. As the panel began
curing, many elongated bubbles or channels formed in between plies (Figure 7-7b). As opposed to
the semi-preg, the conventional format did not prevent bubble formation. The low viscosity of the
112
vinyl hybrid allowed many of those bubbles to escape into the laminate, many of which reached
the edge of the panel to be evacuated (Figure 7-7c). However, many bubbles remained on the
surface as defects (Figure 7-7d). The conventional epoxy displayed the same resin/bubble
progression throughout the cure cycle, albeit with more bubbles at the end remaining as defects
due to lower viscosity.
7.4 Conclusions
A custom tool was used for in-situ observation and analysis of scarf repairs. Video footage
and microscopy showed that through-thickness air egress in semi-preg laminates effectively
eliminated porosity. Semi-preg formatting and vinyl hybrid worked well in conjunction to address
the challenges of in-field repair conditions. In some repairs, a different resin may be required (other
than vinyl hybrid). In such cases, the semi-preg format could be employed with other resins (e.g.,
epoxies) to limit/eliminate porosity. However, the same semi-preg format may not yield void-free
parts for any fabric type or prepreg layup. For example, thicker or more complex parts may require
optimized resin distribution patterns or different cure cycles to ensure complete gas evacuation.
The ability to conduct in-situ analysis through a scarfed repair tool will be valuable in this regard,
providing guidance for optimizing semi-pregs for specific repair conditions. Although the tools
and materials discussed here are well-suited to repairs, the semi-preg format is broadly applicable
to OoA manufacturing, potentially increasing the application space of composites. Ensuring high
quality laminates with simple processing methods will advance the use of high-performance
composite materials.
113
Semi-preg formatting enables low-viscosity resin usage through robust processing in a
sealed environment. In conventional prepreg manufacturing, resins with low viscosities such as
vinyl hybrid bleed into edge breathing dams and into breather layers through perforated release
films. Patch repairs limit the bleed possible from low-viscosity resins by nature of sealed off
edges from the parent material. Furthermore, using release films with fewer perforations also
limit bleed into the breather. With sealed edges and reduced breather bleed, low-viscosity resins
become a unique strength that increases saturation and reduces porosity through bubble
migration. The concept of using semi-preg formatting and low-viscosity resins to promote
process robustness can extend beyond repair. Reducing sources of bleed with low-viscosity
semi-pregs can result in robust procedures for manufacturing high quality laminates. Successful
demonstration of robust processing of low-viscosity resin semi-pregs can widen the scope of
resins able to be safely employed on prepregs.
114
8 CHAPTER 8: Conclusions and Future Work
8.1 Conclusions
The work presented here in this thesis lays the foundation for future work in prepreg
development. New materials with unique properties will continue to be created to meet the
demands of high-performance applications. Semi-preg formatting opens the design space for
using prepreg materials in applications normally reserved for specialized techniques due to
processing difficulties.
Features not present in traditional thermoset materials are being incorporated in modern
resins. Resins can now cure quickly, yet also be inhibited at room temperatures. The combination
of those two features introduces a phenomenon where a resin will not always reach the full
degree of cure, depending on the cure cycle used. Having models that accurately describe the
behavior of these resins is critical for the effective use of these materials. To describe the
behavior of such materials, a new kinetic model was created. To gather data on fast cure
materials using DSC, a method of extracting heats of reaction from negative peak effects was
created. This was combined with an automated process to turn raw data into useable results. The
fast-cure kinetic model incorporates terms to account for inhibition of cure at lower
temperatures. The new terms allowed for the autocatalytic terms to reflect the fast-cure nature of
the material more accurately when compared to a conventional thermoset kinetic model. An
equation was created to predict the maximum degree of cure based on both the curing
temperature of the material in question and maximum temperature of the cure cycle. The
maximum degree of term was incorporated into the fast-cure kinetic model. The new kinetic
model was shown to accurately predict the kinetic behavior of a fast-cure, inhibited resin. Thus,
115
the model can be used to predict the behavior of other room inhibited cure resins, regardless of
cure speed.
The kinetic features of the novel resin archetype also impact its rheology behavior. Fast-cure
translates to rapid gelation at curing temperatures. Inhibition of cure reduces the viscosity of the
material prior to the curing temperature, relative to conventional thermoset resins without
inhibition. The gel model used to describe the viscosity of thermosets was revised to account for
these features. Like in kinetic modeling, a term was added that included the curing temperature
of the material in question to account for inhibition. The revised model was shown to match the
behavior of the resin in question. Thus, the revised model can be used to predict the behavior of
other inhibited cure resins. Cure inhibition also provided an opportunity to assess the
effectiveness of using flow number as a predictor of saturation in composite laminates. This
would occur through cure cycle flow number matching with a conventional epoxy resin. Because
the novel resin was able to maintain a constant viscosity at a temperature below its curing
temperature, altering the flow number was able to be done by simply changing the mid-stage
hold time. It was shown that different materials with the same effective flow number had similar
levels of saturation. When keeping variables such as reinforcement type constant, effective flow
number can be used to predict the saturation from any given cure cycle.
The kinetic and rheology models created for the novel resin were employed in a cure cycle
process optimization. A MATLAB code was written to incorporate both kinetic and rheology
models and to generate plots from multiple cure cycle variables. Kinetic parameters, such as
maximum temperature and curing stage hold time, were altered, and their effect on the total cure
cycle time and final degree of cure was assessed. Rheology parameters such as mid-stage hold
time and mid-stage temperature were modified, and their effect on the total cure cycle time and
116
normalized flow number was evaluated. Cure cycles of less than an hour were created from both
kinetic and rheology analyses. Panels with varying cure cycle features, mid-stage hold time,
maximum temperature, and debulking time, were fabricated to assess the impact of each feature
on fast-cure semi-preg quality. It was shown that cure cycles of less than an hour are possible
with fast-cure prepregs with semi-preg formatting. Furthermore, the accuracy of the kinetic and
rheology models from chapters 3 and 4 were validated from comparative results.
Using the materials and modeling from chapters 3, 4, and 5, an in-field composite repair
solution was presented. The material featured a room temperature stable resin and semi-preg
formatting. The room temperature stability addressed the lack of cold storage available in the
field and obviated the need for wet layup. The semi-preg formatting addressed the need for high
quality repair patches without using processes such as DVD. The novel material was compared
to conventional prepreg laminates in a repair environment, as well as wet laid DVD panels. The
material system was shown to be processed with minimal human expertise driven variation.
While both the wet laid DVD panels and the conventional prepreg laminates exhibited porosity,
the novel material was void free when processed in repair conditions. While the strength and
glass transition for the resin was lower than that of the epoxies, additional development on the
resin can address those shortcomings. The material as designed represents an improvement to
current repair materials and methods. The innovations in semi-preg formatting and room
temperature stability can be employed separately but offer a uniquely useful solution for in-field
repair when combined.
An in-situ examination scarf repair tool was created to give insight into the cure procedure in
repair environments. It had a 3° scarf incline to replicate best practices for repairing composite
laminates on parent materials. The tool featured an observation window by which videos could
117
be taken of the cure process. Videos from the novel material system in chapter 6, conventionally
formatted epoxies, and wet laid DVD systems were analyzed. The void evolution in the materials
was compared to assess the impact of features like semi-preg formatting and low viscosity resins
on the quality of produced repair patches. It was shown that semi-preg formatting and low
viscosity resins combine to create a robust solution for void-free laminates in repairs. The
conventionally formatted prepregs and wet laid DVD materials exhibited porosity in the
produced panels. The results show the impact semi-preg formatting has on void reduction.
Incorporating semi-preg formatting for manufacturing and repair prepregs can reduce the cost of
producing and maintaining prepreg.
8.2 Broader Implications
The results from this thesis can increase the usage of composites through more cost-effective
manufacturing and maintenance. The models, optimizations, material solutions, and in-situ
analysis all contribute to a greater understanding of composite processing. The results can be
used to reduce cycle times, improve the ease of maintenance, and increase the reliability of OoA
manufacturing.
Materials with inhibited cure that imparts room temperature stability have been created.
Prepreg out-time is currently a concern in both manufacturing and repair of composites. In
manufacturing, out-time can determine whether a material is within specifications, and therefore
whether the material will be used or become waste. Cold storage is not always present in field
conditions, thus having a prepreg without needing cold storage allows prepregs to be used for
118
patch repairs. Understanding the kinetic behavior of resins is key to designing effective cure
cycles. The cure model presented in this thesis can be used on future materials with inhibited
cure. The rheology of such materials was also modeled using a revised gel model. The model can
be used to describe the behavior of future resins features cure inhibition. Effective flow number
was shown to be effective at predicting saturation in composite laminates. Cure cycle design
relies on predictive metrics that translate into laminate quality or performance. Having an extra
predictive metric in effective flow number increases the number of available tools for engineers
creating cure cycles.
A process optimization using kinetic and rheology models was created to reduce cure cycle
times without sacrificing laminate quality and performance. It was shown that laminates can be
fabricated with OoA cure cycles of less than an hour with full saturation and degree of cure.
Many current OoA materials require hours of processing, which adds to the manufacturing costs.
Using the materials and process optimizations in this thesis to reduce cycle times can greatly
reduce manufacturing costs. While the aerospace industry, especially the military, has been able
to bear the long manufacturing times for high performance applications, other industries that
stand to benefit from lower composite material usage are unable to do so. With cure cycles of
less than an hour, composite use can become more competitive in industries like automotive.
An in-field repair solution featuring a room temperature stable resin and semi-preg
formatting was presented. The prepreg was shown to have robust quality when processed with
repair conditions. Current repairs to aircraft are performed in maintenance depots. Technicians
with a high level of expertise are required at these depots to perform methods like wet layup and
double vacuum debulking. The material solution presented does not require the same level of
expertise to ensure high quality patches. It also can be deployed without requiring the logistical
119
challenges of cold storage in the field. Employing a material system with these features can
greatly reduce the maintenance expense of composite structures, thus reducing the overall cost of
using composite materials. Additionally, with reduced cycle times and additional flexibility, the
material system can be used to improve manufacturing efficiency, even though it was originally
designed for repair. With these innovations, the cost of using composites is reduced, and they can
be employed in broader applications.
8.3 Future Work
The kinetic and rheology models were created based off a material with cure inhibition and
fast-cure properties. As additional resins, especially epoxies, with inhibited cure are developed,
the model’s performance on those materials should be tested.
Flow number analysis was shown to be effective between prepregs where all other variables
besides resin selection were kept constant. In practical applications, it is not possible to isolate
variables in a laboratory setting. Thus, incorporating scaling variables into the flow number
analysis to account for factors that would impact saturation could be done. These variables could
include tow size, reinforcement materials, fabric type, consolidation pressure, and fiber volume
fraction.
Semi-preg formatting adds an additional variable in percent surface coverage that would
affect the saturation of a laminate. Assessing the effect of surface coverage, pattern type, and
resin gaps on the saturation of a laminate with effective flow number can be done to improve the
predictive capabilities of flow number analysis for semi-preg materials. The in-situ observation
120
tool can be employed to analyze how these three features affect resin flow and void evolution in
repair laminates. It may be possible to define a variable that correlates to the ability of semi-preg
formatted materials to evacuate air in the through thickness direction. A possible calculation
could be an integration of prepreg air permeability throughout the cure cycle used. Unlike
effective flow number, the inverse would not be taken because permeability is directly correlated
with air flow. The variable could be used alongside the degree of cure and the effective flow
number as a tool to guide material design. The three features mentioned above can be
incorporated alongside any other factors that would affect the gas permeability of the prepreg
over time. Other factors to consider include resin flow that erases resin gaps and bag pressure.
Effective flow number would be a good starting point to account for resin flow that erases resin
gaps in a semi-preg material and therefore decreases permeability.
It was shown with in-situ observation that combining semi-preg formatting with low
viscosity resins can allow for high quality laminate fabrication. The conjoined features allowed
for faster gas evacuation, resin saturation and bubble migration. In these studies, a sealed
environment to prevent resin bleed from the low viscosity resin was created by reducing the
perforations in the bag side release film and sealing the edges of the laminate with sealant tape.
Sealing the edges represents a departure from conventional prepreg processing, where edge
breathing is needed to evacuate air. However, with low viscosity resins, edge dams are an outlet
for significant resin bleed. Thus, sealed edges are necessary to work with low viscosity resin
prepregs. With semi-preg formatting, air can escape through the thickness of the part and
primarily through the release film, obviating the need for edge breathing. Reducing perforations
in the bag side release film was also done to reduce resin bleed into the breather cloth. While
sealing the edges and reducing film perforations was effective in creating a mostly sealed
121
environment, a semi-permeable release film can allow for a fully sealed off environment for the
laminate. Films are now available that are impermeable to resin but allow air to pass through.
Using these films in conjunction with low viscosity, semi-preg formatted resins could improve
the quality of produced laminates.
Finally, the materials used in this thesis were flat laminates. Sandwich panels are commonly
employed to increase the stiffness of composite panels. The performance of the repair material
solution presented in this thesis can be studied in a sandwich configuration. The in-situ tool can
be used to analyze the effectiveness of semi-preg formatting when used in sandwich panel
repairs. Further demonstrations of the effectiveness of semi-preg formatting in manufacturing
and repair can increase the serviceability of composite materials.
122
9 References
[1] F. Campbell, Manufacturing Technology for Aerospace Structural Materials. 2006.
[2] B. A. Newcomb, “Processing , structure , and properties of carbon fibers,” Compos. Part
A, vol. 91, pp. 262–282, 2016.
[3] L. Khoun, T. Centea, and P. Hubert, “Characterization Methodology of Thermoset Resins
for the Processing of Composite Materials — Case Study: CYCOM 890RTM Epoxy
Resin,” J. Compos. Mater., vol. 44, no. 11, pp. 1397–1415, Jun. 2010, doi:
10.1177/0021998309353960.
[4] N. Guermazi, A. Ben Tarjem, I. Ksouri, and H. F. Ayedi, “On the durability of FRP
composites for aircraft structures in hygrothermal conditioning,” Compos. Part B Eng.,
vol. 85, pp. 294–304, Feb. 2016, doi: 10.1016/j.compositesb.2015.09.035.
[5] S. A. Bello, J. O. Agunsoye, S. B. Hassan, M. G. Zebase Kana, and I. A. Raheem,
“Tribology in Industry Epoxy Resin Based Composites, Mechanical and Tribological
Properties: A Review,” 2015. Accessed: Mar. 11, 2021. [Online]. Available:
www.tribology.fink.rs.
[6] T. Centea, L. K. Grunenfelder, and S. R. Nutt, “A review of out-of-autoclave prepregs –
Material properties, process phenomena, and manufacturing considerations,” Compos.
Part A Appl. Sci. Manuf., vol. 70, pp. 132–154, 2015, doi:
10.1016/j.compositesa.2014.09.029.
[7] K. Yu, Q. Shi, M. L. Dunn, T. Wang, and H. J. Qi, “Carbon Fiber Reinforced Thermoset
Composite with Near 100% Recyclability,” Adv. Funct. Mater., vol. 26, no. 33, pp. 6098–
6106, Sep. 2016, doi: 10.1002/adfm.201602056.
123
[8] T. A. Cender, J. J. J. Gangloff, P. Simacek, and S. G. Advani, “Void reduction during out-
of-autoclave thermoset prepreg composite processing,” SAMPE Int. Symp., 2013.
[9] R. M’Saoubi et al., “High performance cutting of advanced aerospace alloys and
composite materials,” CIRP Ann. - Manuf. Technol., vol. 64, no. 2, pp. 557–580, 2015,
doi: 10.1016/j.cirp.2015.05.002.
[10] J. J. Mazza and K. M. Storage, “Bonded Repair in the United States Air Force and Work
to Expand Future Capability,” STO-MP-AVT-266, pp. 1–22, 2018, doi: 978-92-837-2172-
7.
[11] Improving the efficiency of engines for large nonfighter aircraft. National Academies
Press, 2007.
[12] “Aircraft mission-capable rates hit new low in Air Force, despite efforts to improve.”
https://www.airforcetimes.com/news/your-air-force/2019/07/26/aircraft-mission-capable-
rates-hit-new-low-in-air-force-despite-efforts-to-improve/ (accessed Mar. 13, 2021).
[13] R. Saltoʇlu, N. Humaira, and G. Inalhan, “Aircraft Scheduled Airframe Maintenance and
Downtime Integrated Cost Model,” Adv. Oper. Res., vol. 2016, 2016, doi:
10.1155/2016/2576825.
[14] “Guidelines for the Development of Process Specifications, Instructions, and Controls for
the Fabrication of Fiber-Reinforced Polymer Composites,” 2003.
[15] C. Seaton, “Guidelines for the Development of a Critical Composite Maintenance and
Repair Issues Awareness Course,” 2009.
[16] “Boeing 737 MAX Reading Room.”
124
https://www.faa.gov/foia/electronic_reading_room/boeing_reading_room/ (accessed Mar.
15, 2021).
[17] K. K€ Uchler, E. Staiger, R.-D. Hund, O. Diestel, M. Kirsten, and C. Cherif, “Local repair
procedure for carbon-fiber-reinforced plastics by refilling with a thermoset matrix,” J.
Appl. Polym. Sci, p. 42964, 2015, doi: 10.1002/app.42964.
[18] M. E. Kazemi, L. Shanmugam, D. Lu, X. Wang, B. Wang, and J. Yang, “Mechanical
properties and failure modes of hybrid fiber reinforced polymer composites with a novel
liquid thermoplastic resin, Elium®,” Compos. Part A Appl. Sci. Manuf., vol. 125, p.
105523, Oct. 2019, doi: 10.1016/j.compositesa.2019.105523.
[19] J. M. Zhang, “Design and analysis of mechanically fastened composite joints and repairs,”
Eng. Anal. Bound. Elem., vol. 25, no. 6, pp. 431–441, Jun. 2001, doi: 10.1016/S0955-
7997(01)00049-2.
[20] R. Jones, W. K. Chiu, and R. Smith, “Airworthiness of composite repairs: Failure
mechanisms,” Eng. Fail. Anal., vol. 2, no. 2, pp. 117–128, 1995, doi: 10.1016/1350-
6307(95)00011-E.
[21] J. E. Robson, F. L. Matthews, and A. J. Kinloch, “The bonded repair of fibre composites:
Effect com of composite moisture content,” Compos. Sci. Technol., vol. 52, no. 2, pp.
235–246, Jan. 1994, doi: 10.1016/0266-3538(94)90208-9.
[22] L. K. Grunenfelder, “Defect control in out- of-autoclave manufacturing of structural
elements,” 2012.
[23] S. G. K. Schechter, L. K. Grunenfelder, and S. R. Nutt, “Design and application of
125
discontinuous resin distribution patterns for semi-pregs,” Adv. Manuf. Polym. Compos.
Sci., vol. 6, no. 2, pp. 72–85, 2020, doi: 10.1080/20550340.2020.1736864.
[24] S. G. K. Schechter, L. K. Grunenfelder, and S. R. Nutt, “Air evacuation and resin
impregnation in semi-pregs: effects of feature dimensions,” Adv. Manuf. Polym. Compos.
Sci., vol. 6, no. 2, pp. 101–114, Apr. 2020, doi: 10.1080/20550340.2020.1768348.
[25] A. N. Rider, A. A. Baker, C. H. Wang, and G. Smith, “An enhanced vacuum cure
technique for on-aircraft repair of carbon-bismaleimide composites,” Appl. Compos.
Mater., vol. 18, no. 3, pp. 231–251, 2011, doi: 10.1007/s10443-010-9148-9.
[26] H. M. Chong et al., “Out-of-autoclave scarf repair of interlayer toughened carbon fibre
composites using double vacuum debulking of patch,” Compos. Part A Appl. Sci. Manuf.,
vol. 107, pp. 224–234, 2018, doi: 10.1016/j.compositesa.2018.01.001.
[27] T. H. Hou and B. J. Jensen, “Double-vacuum-bag technology for volatile management in
composite fabrication,” Polym. Compos., vol. 29, no. 8, pp. 906–914, 2008, doi:
10.1002/pc.20475.
[28] M. L. Costa, S. F. M. d. Almeida, and M. C. Rezende, “The influence of porosity on the
interlaminar shear strength of carbon/epoxy and carbon/bismaleimide fabric laminates,”
Compos. Sci. Technol., vol. 61, no. 14, pp. 2101–2108, Nov. 2001, doi: 10.1016/S0266-
3538(01)00157-9.
[29] F. Lambinet and Z. S. Khodaei, “Damage detection & localization on composite patch
repair under different environmental effects,” Eng. Res. Express, vol. 2, no. 4, p. 045032,
Dec. 2020, doi: 10.1088/2631-8695/abd0d3.
126
[30] D. N. Markatos, K. I. Tserpes, E. Rau, S. Markus, B. Ehrhart, and S. Pantelakis, “The
effects of manufacturing-induced and in-service related bonding quality reduction on the
mode-I fracture toughness of composite bonded joints for aeronautical use,” Compos. Part
B Eng., vol. 45, no. 1, pp. 556–564, Feb. 2013, doi: 10.1016/j.compositesb.2012.05.052.
[31] C. M. Warnock and T. M. Briggs, “Cure cycle development and qualification for thick-
section composites,” Int. SAMPE Tech. Conf., vol. 2016-Janua, 2016.
[32] G. P. M. Fierro, F. Ciampa, D. Ginzburg, E. Onder, and M. Meo, “Nonlinear ultrasound
modelling and validation of fatigue damage,” J. Sound Vib., vol. 343, pp. 121–130, May
2015, doi: 10.1016/j.jsv.2014.10.008.
[33] T. Zweschper, G. Riegert, A. Dillenz, and G. Busse, “Ultrasound excited thermography -
advances due to frequency modulated elastic waves,” Quant. Infrared Thermogr. J., vol.
2, no. 1, pp. 65–76, 2005, doi: 10.3166/qirt.2.65-76.
[34] Y. J. Lee, J. R. Lee, and J. B. Ihn, “Composite repair patch evaluation using pulse-echo
laser ultrasonic correlation mapping method,” Compos. Struct., vol. 204, pp. 395–401,
Nov. 2018, doi: 10.1016/j.compstruct.2018.07.124.
[35] F. Liu, Z. Zhou, S. Liu, Y. Yang, and L. Zhang, “Evaluation of carbon fiber composite
repairs using asymmetric-frequency ultrasound waves,” Compos. Part B Eng., vol. 181, p.
107534, Jan. 2020, doi: 10.1016/j.compositesb.2019.107534.
[36] M. Choquet, R. Heon, C. Padioleau, P. Bouchard, C. Neron, and J.-P. Monchalin,
“LASER-ULTRASONIC INSPECTION OF THE COMPOSITE STRUCTURE OF AN
AIRCRAFT IN A MAINTENANCE HANGAR.”
127
[37] P. J. Halley and M. E. Mackay, “Chemorheology of Thermosets - An Overview,” Polym.
Eng. Sci., vol. 36, no. 5, pp. 593–609, 1996, doi: 10.1002/pen.10447.
[38] D. H. J. A. Lukaszewicz, C. Ward, and K. D. Potter, “The engineering aspects of
automated prepreg layup: History, present and future,” Compos. Part B Eng., vol. 43, no.
3, pp. 997–1009, 2012, doi: 10.1016/j.compositesb.2011.12.003.
[39] P. Malnati and J. Sloan, “Fast and Faster: Rapid-cure resins drive down cycle times,”
2018. https://www.compositesworld.com/articles/fast-and-faster-rapid-cure-epoxies-drive-
down-cycle-times.
[40] Thermal Characterization of Polymeric Materials. Elsevier, 1981.
[41] M. R. Kamal and S. Sourour, “Kinetics and thermal characterization of thermoset cure,”
Polym. Eng. Sci., vol. 13, no. 1, pp. 59–64, 1973, doi: 10.1002/pen.760130110.
[42] U. Khanna and M. Chanda, “Kinetics of anhydride curing of isophthalic diglycidyl ester
using differential scanning calorimetry,” J. Appl. Polym. Sci., vol. 49, no. 2, pp. 319–329,
Jul. 1993, doi: 10.1002/app.1993.070490212.
[43] P. Hubert, “Cure kinetics and viscosity models for Hexcel 8552 epoxy resin | Request
PDF,” SAMPE Proceeding, 2001.
[44] P. Malnati, “Hybrid resin system: Epoxy benefits, without the epoxy,” Composites World,
2019. https://www.compositesworld.com/blog/post/hybrid-resin-system-epoxy-benefits-
without-the-epoxy.
[45] F. G. Mussatti and C. W. Macosko, “Rheology of network forming systems,” Polym. Eng.
Sci., vol. 13, no. 3, pp. 236–240, 1973, doi: 10.1002/pen.760130312.
128
[46] J. D. Ferry, Viscoelastic Properties of Polymers - John D. Ferry, Third Edit. Canada: John
Wiley & Sons, 1980.
[47] P. Nogueira et al., “Effect of water sorption on the structure and mechanical properties of
an epoxy resin system,” J. Appl. Polym. Sci., vol. 80, no. 1, pp. 71–80, Apr. 2001, doi:
10.1002/1097-4628(20010404)80:1<71::AID-APP1077>3.0.CO;2-H.
[48] A. Y. Malkin and S. G. Kulichikhin, “Rheokinetics of curing,” Adv. Polym. Sci., vol. 101,
pp. 216–257, 1991, doi: 10.1007/bfb0018003.
[49] R. Helmus, J. Kratz, K. Potter, P. Hubert, and R. Hinterhölzl, “An experimental technique
to characterize interply void formation in unidirectional prepregs,” J. Compos. Mater.,
vol. 51, no. 5, pp. 579–591, Mar. 2017, doi: 10.1177/0021998316650273.
[50] M. Mehdikhani, L. Gorbatikh, I. Verpoest, and S. V. Lomov, “Voids in fiber-reinforced
polymer composites: A review on their formation, characteristics, and effects on
mechanical performance,” J. Compos. Mater., vol. 53, no. 12, pp. 1579–1669, 2019, doi:
10.1177/0021998318772152.
[51] P. Hubert, G. Fernlund, and A. Poursartip, “Autoclave processing for composites,” in
Manufacturing Techniques for Polymer Matrix Composites (PMCs), Elsevier, 2012, pp.
414–434.
[52] C. W. Macosko, “Rheological changes during crosslinking,” Br. Polym. J., vol. 17, no. 2,
pp. 239–245, Jun. 1985, doi: 10.1002/pi.4980170228.
[53] T. Vidil, F. Tournilhac, S. Musso, A. Robisson, and L. Leibler, “Control of reactions and
network structures of epoxy thermosets,” Prog. Polym. Sci., vol. 62, pp. 126–179, 2016,
129
doi: 10.1016/j.progpolymsci.2016.06.003.
[54] S.-S. Chae, G.-E. Lee, H. Ahn, J.-H. Choi, and J.-H. Kweon, “Tensile Strength of
Composite Laminate Repaired Using Heat-blanket and a Novel Pressurization System,”
Compos. Res., vol. 31, no. 1, pp. 1–7, 2018, doi: 10.7234/composres.2018.31.1.001.
[55] J. J. Gangloff, C. Daniel, and S. G. Advani, “A model of two-phase resin and void flow
during composites processing,” Int. J. Multiph. Flow, 2014, doi:
10.1016/j.ijmultiphaseflow.2014.05.015.
[56] L. Khoun, T. Centea, and P. Hubert, “Characterization methodology of thermoset resins
for the processing of composite materials -Case study,” J. Compos. Mater., vol. 44, no.
11, pp. 1397–1415, Jun. 2010, doi: 10.1177/0021998309353960.
[57] I. K. Varma and P. V. Satya Bhama, “Epoxy Resins: Effect of Amines on Curing
Characteristics and Properties,” J. Compos. Mater., vol. 20, no. 5, pp. 410–422, 1986, doi:
10.1177/002199838602000501.
[58] L. Woo, “Modeling void formation and unsaturated flow in liquid composite molding
processes: a survey and review,” J. Reinf. Plast. Compos., vol. 30, no. 11, pp. 957–977,
Aug. 2011, doi: 10.1177/0731684411411338.
[59] D. Kim and S. R. Nutt, “Processability of DDS isomers-cured epoxy resin: Effects of
amine/epoxy ratio, humidity, and out-time,” Polym. Eng. Sci., 2017, doi:
10.1002/pen.24738.
[60] S. G. K. Schechter, L. K. Grunenfelder, and S. R. Nutt, “Air evacuation and resin
impregnation in semi-pregs: effects of feature dimensions,” Adv. Manuf. Polym. Compos.
130
Sci., vol. 6, no. 2, pp. 101–114, 2020, doi: 10.1080/20550340.2020.1768348.
[61] S. G. K. Schechter, T. Centea, and S. R. Nutt, “Polymer film dewetting for fabrication of
out-of-autoclave prepreg with high through-thickness permeability,” Compos. Part A
Appl. Sci. Manuf., vol. 114, no. August, pp. 86–96, 2018, doi:
10.1016/j.compositesa.2018.08.002.
[62] D. B. Bender, T. Centea, and S. Nutt, “Fast cure of stable semi-pregs via VBO cure,” Adv.
Manuf. Polym. Compos. Sci., vol. 0, no. 0, pp. 1–11, Jan. 2021, doi:
10.1080/20550340.2020.1869891.
[63] B. G. Min, Z. H. Stachurski, and J. H. Hodgkin, “Cure kinetics of elementary reactions of
a diglycidyl ether of bisphenol A/diaminodiphenylsulfone epoxy resin: 2. Conversion
versus time,” Polymer (Guildf)., vol. 34, no. 21, pp. 4488–4495, 1993, doi: 10.1016/0032-
3861(93)90155-4.
[64] J. J. Gangloff, T. A. Cender, V. Eskizeybek, P. Simacek, and S. G. Advani, “Entrapment
and venting of bubbles during vacuum bag prepreg processing,” J. Compos. Mater., vol.
51, no. 19, pp. 2757–2768, Aug. 2017, doi: 10.1177/0021998316676325.
[65] A. R. Upadhya, G. N. Dayananda, G. M. Kamalakannan, J. Ramaswamy Setty, and J.
Christopher Daniel, “Autoclaves for Aerospace Applications: Issues and Challenges,” Int.
J. Aerosp. Eng., vol. 2011, pp. 1–11, 2011, doi: 10.1155/2011/985871.
[66] G. R. Sherwin, “Non-autoclave processing of advanced composite repairs,” Int. J. Adhes.
Adhes., vol. 19, no. 2, pp. 155–159, 1999, doi: 10.1016/S0143-7496(98)00030-X.
[67] H. Koushyar, S. Alavi-Soltani, B. Minaie, and M. Violette, “Effects of variation in
131
autoclave pressure, temperature, and vacuum-application time on porosity and mechanical
properties of a carbon fiber/epoxy composite,” J. Compos. Mater., vol. 46, no. 16, pp.
1985–2004, 2012, doi: 10.1177/0021998311429618.
[68] T. A. Cender, P. Simacek, S. Davis, and S. G. Advani, “Gas Evacuation from Partially
Saturated Woven Fiber Laminates,” Transp. Porous Media, vol. 115, no. 3, pp. 541–562,
Dec. 2016, doi: 10.1007/s11242-016-0784-x.
[69] T. Kourkoutsaki, S. Comas-Cardona, C. Binetruy, R. K. Upadhyay, and R. Hinterhoelzl,
“The impact of air evacuation on the impregnation time of Out-of-Autoclave prepregs,”
Compos. Part A Appl. Sci. Manuf., vol. 79, pp. 30–42, 2015, doi:
10.1016/j.compositesa.2015.08.034.
[70] T. Centea and P. Hubert, “Out-of-autoclave prepreg consolidation under deficient pressure
conditions,” J. Compos. Mater., vol. 48, no. 16, pp. 2033–2045, Jul. 2014, doi:
10.1177/0021998313494101.
[71] M. Préau and P. Hubert, “Processing of co-bonded scarf repairs: Void reduction strategies
and influence on strength recovery,” Compos. Part A Appl. Sci. Manuf., vol. 84, pp. 236–
245, 2016, doi: 10.1016/j.compositesa.2016.01.016.
[72] E. Archer and A. McIlhagger, “Repair of damaged aerospace composite structures,” in
Polymer Composites in the Aerospace Industry, Elsevier, 2015, pp. 393–412.
[73] C. C. Chamis, L. M. Handler, and J. M. Manderscheid, “Comp2007–002,” pp. 1–8, 2019.
[74] S. Bickerton, R. J. Lin, A. A. S. I. Singh, and A. A. S. I. Singh, Processing and
Fabrication of Advanced Materials-XXV Editors Processing and Fabrication Of
132
Advanced Materials-XXV Editors. .
[75] B. K. Fink, T. A. Bogetti, M. A. Stone, and J. W. Gillespie, “Thermochemical Response
of Vinyl-Ester Resin,” no. January, 2002.
[76] F. Fischer, U. Beier, F. Wolff-Fabris, and V. Altstädt, “Toughened high performance
epoxy resin system for aerospace applications,” Sci. Eng. Compos. Mater., vol. 18, no. 4,
pp. 209–215, 2011, doi: 10.1515/SECM.2011.042.
[77] M. Holmes, “Global carbon fibre market remains on upward trend,” Reinf. Plast., vol. 58,
no. 6, pp. 38–45, 2014, doi: 10.1016/S0034-3617(14)70251-6.
[78] S. M. Spearing, P. A. Lagace, and H. L. N. McManus, “On the role of lengthscale in the
prediction of failure of composite structures: Assessment and needs,” Appl. Compos.
Mater., vol. 5, no. 3, pp. 139–149, 1998, doi: 10.1023/a:1008876701815.
[79] N. M. Barkoula, B. Alcock, N. O. Cabrera, and T. Peijs, “Impact Resistance and Damage
Tolerance of Scarf-Repaired Composite Structures: An Experimental Investigation,”
Polym. Polym. Compos., vol. 16, no. 2, pp. 101–113, 2008, doi: 10.1002/pc.
[80] K. B. Katnam, L. F. M. Da Silva, and T. M. Young, “Bonded repair of composite aircraft
structures: A review of scientific challenges and opportunities,” Prog. Aerosp. Sci., vol.
61, pp. 26–42, 2013, doi: 10.1016/j.paerosci.2013.03.003.
[81] K. Küchler, E. Staiger, R. D. Hund, O. Diestel, M. Kirsten, and C. Cherif, “Local repair
procedure for carbon-fiber-reinforced plastics by refilling with a thermoset matrix,” J.
Appl. Polym. Sci., vol. 133, no. 6, pp. 4–9, 2016, doi: 10.1002/app.42964.
[82] R. Donner, “Composite Repair at FedEx Express,” no. May, pp. 22–27, 2012.
133
[83] C. Garnier, M.-L. Pastor, F. Eyma, and B. Lorrain, “The detection of aeronautical defects
in situ on composite structures using Non Destructive Testing,” Compos. Struct., vol. 93,
no. 5, pp. 1328–1336, 2011, doi: 10.1016/j.compstruct.2010.10.017.
[84] B. Whittingham, A. A. Baker, A. Harman, and D. Bitton, “Micrographic studies on
adhesively bonded scarf repairs to thick composite aircraft structure,” Compos. Part A
Appl. Sci. Manuf., vol. 40, no. 9, pp. 1419–1432, 2009, doi:
10.1016/j.compositesa.2008.12.011.
[85] J.-H. Lu and J. P. Youngblood, “Adhesive bonding of carbon fiber reinforced composite
using UV-curing epoxy resin,” Compos. Part B Eng., vol. 82, pp. 221–225, Dec. 2015,
doi: 10.1016/j.compositesb.2015.08.022.
[86] C. H. Wang and A. Gunnion, “Design Methodology for Scarf Repairs to Composite
Structures.”
[87] J.-S. Yoo, V.-H. Truong, M.-Y. Park, J.-H. Choi, and J.-H. Kweon, “Parametric study on
static and fatigue strength recovery of scarf-patch-repaired composite laminates,”
Compos. Struct., vol. 140, pp. 417–432, 2016, doi:
http://dx.doi.org/10.1016/j.compstruct.2015.12.041.
[88] E. Ghazali, M. L. Dano, A. Gakwaya, and C. O. Amyot, “Experimental and numerical
studies of stepped-scarf circular repairs in composite sandwich panels,” Int. J. Adhes.
Adhes., vol. 82, pp. 41–49, Apr. 2018, doi: 10.1016/j.ijadhadh.2017.12.008.
[89] D. Zaremba et al., “Repair preparation of fiber-reinforced plastics by the machining of a
stepped peripheral zone,” Stroj. Vestnik/Journal Mech. Eng., vol. 58, no. 10, pp. 571–577,
2012, doi: 10.5545/sv-jme.2012.305.
134
[90] K. W. Jolly I, Schlogl S, Wolfahrt M, Pinter G, Fleischmann M, “Chemical
functionalization of composite surfaces for improved structural bonded repairs,” Compos.
Part B Eng., vol. 69, pp. 296–303, 2015, doi: 10.1016/j.compositesb.2014.10.020.
[91] H. M. Chong et al., “Out-of-autoclave scarf repair of interlayer toughened carbon fibre
composites using double vacuum debulking of patch,” Compos. Part A Appl. Sci. Manuf.,
vol. 107, pp. 224–234, Apr. 2018, doi: 10.1016/j.compositesa.2018.01.001.
[92] A. Benkheira, M. Belhouari, and S. Benbarek, “Comparison of Double- and Single-
Bonded Repairs to Symmetrical Composite Structures,” J. Fail. Anal. Prev., vol. 18, no. 6,
pp. 1601–1606, Dec. 2018, doi: 10.1007/s11668-018-0557-7.
[93] “DoD Maintenance Depot Capabilities and Services Public-Private Partnerships.”
[94] C. Dransfeld, “An analytical model for B-stage joining and co- curing of carbon fibre
epoxy composites Composites : Part A,” Compos. Part A, vol. 87, no. July, pp. 282–289,
2016, doi: 10.1016/j.compositesa.2016.05.009.
[95] M. Molyneux, P. Murray, and B. P. Murray, “Prepreg, tape and fabric technology for
advanced composites,” Composites, vol. 14, no. 2, pp. 87–91, 1983, doi: 10.1016/S0010-
4361(83)80003-2.
[96] B. Thorfinnson and T. Biermann, “Degree of Impregnation of Prepregs--Effects on
Porosity,” in Advanced Materials Technology, 1987, p. 9.
[97] S. L. Agius, K. J. C. Magniez, and B. L. Fox, “Cure behaviour and void development
within rapidly cured out-of-autoclave composites,” Compos. Part B Eng., 2013, doi:
10.1016/j.compositesb.2012.11.020.
135
[98] L. Hamill, T. Centea, and S. Nutt, “Surface porosity during vacuum bag-only prepreg
processing: Causes and mitigation strategies,” Compos. Part A Appl. Sci. Manuf., vol. 75,
pp. 1–10, 2015, doi: 10.1016/j.compositesa.2015.04.009.
[99] J. Heth, “From art to science: A prepreg Overview,” High Perform. Compos., pp. 32–36,
2000.
[100] R. Helmus, T. Centea, P. Hubert, and R. Hinterholzl, “Out-of-autoclave prepreg
consolidation: Coupled air evacuation and prepreg impregnation modeling,” J. Compos.
Mater., vol. 0, no. 0, pp. 1–11, 2015, doi: 10.1177/0021998315592005.
[101] P. H. Wang, “The Comparison of Composite Aircraft Field Repair Method ( CAFRM )
with Traditional Aircraft Repair Technologies,” 2013.
[102] Y. Gu, M. Li, Z. Zhan, and Y. Li, “Effects of resin storage aging on rheological property
and consolidation of composite laminates,” Polym. Compos., vol. 30, no. 8, pp. 1081–
1090, 2009, doi: 10.1002/pc.20659.
[103] K. Mason, “The craft of aircraft repair,” Composites World, 2005.
http://www.compositesworld.com/articles/the-craft-of-aircraft-repair (accessed Jan. 23,
2017).
[104] L. Dorworth, “Composite repair: Lessons learned, challenges and opportunities, Part I,”
CompositesWorld, 2016. http://www.compositesworld.com/columns/composite-repair-
lessons-learned-challenges-and-opportunities-part-i.
[105] L. K. Grunenfelder, A. Dills, T. Centea, and S. Nutt, “Effect of prepreg format on defect
control in out-of-autoclave processing,” Compos. Part A Appl. Sci. Manuf., vol. 93, pp.
136
88–99, 2017, doi: 10.1016/j.compositesa.2016.10.027.
[106] NAVAIR, “General Composite Repair,” 2005.
[107] H. Bendemra, P. Compston, and P. J. Crothers, “Optimisation study of tapered scarf and
stepped-lap joints in composite repair patches,” Compos. Struct., vol. 130, pp. 1–8, 2015,
doi: 10.1016/j.compstruct.2015.04.016.
[108] A. J. Gunnion and I. Herszberg, “Parametric study of scarf joints in composite structures,”
Compos. Struct., vol. 75, no. 1–4, pp. 364–376, 2006, doi:
10.1016/j.compstruct.2006.04.053.
[109] E. Ruiz, “Thermomechanical Properties during Cure of Glass-Polyester RTM Composites:
Elastic and Viscoelastic Modeling,” J. Compos. Mater., vol. 39, no. 10, pp. 881–916,
2005, doi: 10.1177/0021998305048732.
[110] N. M. Barkoula, B. Alcock, N. O. Cabrera, and T. Peijs, “Double-Vacuum-Bag
Technology for Volatile Management in Composite Fabrication,” Polym. Polym.
Compos., vol. 16, no. 2, pp. 101–113, 2008, doi: 10.1002/pc.
[111] L. K. Grunenfelder and S. R. Nutt, “Void formation in composite prepregs - Effect of
dissolved moisture,” Compos. Sci. Technol., vol. 70, no. 16, pp. 2304–2309, Dec. 2010,
doi: 10.1016/j.compscitech.2010.09.009.
[112] K. J. Bowles and S. Frimpong, “Void Effects on the Interlaminar Shear Strength of
Unidirectional Graphite-Fiber-Reinforced Composites,” J. Compos. Mater., vol. 26, no.
10, pp. 1487–1509, Oct. 1992, doi: 10.1177/002199839202601006.
[113] “Guidelines and Recommended Criteria for the Development of a Material Specification
137
for Carbon Fiber/Epoxy Fabric Prepregs,” 2007.
[114] M. Jureczko, M. Pawlak, and A. Mȩzyk, “Optimisation of wind turbine blades,” J. Mater.
Process. Technol., vol. 167, no. 2–3, pp. 463–471, Aug. 2005, doi:
10.1016/j.jmatprotec.2005.06.055.
[115] R. Talreja and J. Varna, Modeling damage, fatigue and failure of composite materials.
Elsevier Inc., 2015.
[116] J. C. S. Queiroz, Y. T. B. Santos, I. C. da Silva, and C. T. T. Farias, “Damage Detection in
Composite Materials Using Tap Test Technique and Neural Networks,” Journal of
Nondestructive Evaluation, vol. 40, no. 1. Springer, p. 27, Mar. 01, 2021, doi:
10.1007/s10921-021-00759-9.
[117] P. Reimers, W. B. Gilboy, and J. Goebbels, “Recent developments in the industrial
application of computerized tomography with ionizing radiation,” NDT Int., vol. 17, no. 4,
pp. 197–207, Aug. 1984, doi: 10.1016/0308-9126(84)90021-X.
[118] F. Awaja, M. T. Nguyen, S. Zhang, and B. Arhatari, “The investigation of inner structural
damage of UV and heat degraded polymer composites using X-ray micro CT,” Compos.
Part A Appl. Sci. Manuf., vol. 42, no. 4, pp. 408–418, Apr. 2011, doi:
10.1016/j.compositesa.2010.12.015.
[119] M. Barburski, I. Straumit, X. Zhang, M. Wevers, and S. V. Lomov, “Micro-CT analysis of
internal structure of sheared textile composite reinforcement,” Compos. Part A Appl. Sci.
Manuf., vol. 73, pp. 45–54, Jun. 2015, doi: 10.1016/j.compositesa.2015.03.008.
[120] A. T. Zehnder, V. Patel, and T. J. Rose, “Micro-CT Imaging of Fibers in Composite
138
Laminates under High Strain Bending,” Exp. Tech., vol. 44, no. 5, pp. 531–540, Oct.
2020, doi: 10.1007/s40799-020-00374-9.
[121] G. M. Carlomagno, F. Ricci, C. Meola, G. M. Carlomagno, V. Lopresto, and G. Caprino,
“Investigation of Impact Damage in Composites with Infrared Thermography,” 2011.
Accessed: Mar. 11, 2021. [Online]. Available:
https://www.researchgate.net/publication/266016352.
[122] Y. S. Chen, Y. Y. Hung, S. P. Ng, Y. H. Huang, and L. Liu, “Review and comparison of
shearography and active thermography for nondestructive testing and evaluation
(NDT&E),” in ICEM 2008: International Conference on Experimental Mechanics
2008, Nov. 2008, vol. 7375, p. 73754W, doi: 10.1117/12.839322.
[123] Y. Y. Hung, W. D. Luo, L. Lin, and H. M. Shang, “Evaluating the soundness of bonding
using shearography,” Compos. Struct., vol. 50, no. 4, pp. 353–362, Dec. 2000, doi:
10.1016/S0263-8223(00)00109-4.
[124] D. Francis, R. P. Tatam, and R. M. Groves, “Shearography technology and applications: A
review,” Measurement Science and Technology, vol. 21, no. 10. Institute of Physics
Publishing, p. 102001, Aug. 25, 2010, doi: 10.1088/0957-0233/21/10/102001.
[125] D. Hofmann, G. Pandarese, G. M. Revel, E. P. Tomasini, and R. Pezzoni, “Optimization
of the excitation and measurement procedures in nondestructive testing using
shearography,” Rev. Sci. Instrum., vol. 79, no. 11, p. 115105, Nov. 2008, doi:
10.1063/1.3002423.
[126] W. Hu, L. K. Grunenfelder, T. Centea, and S. Nutt, “In situ monitoring and analysis of
void evolution in unidirectional prepreg,” J. Compos. Mater., vol. 52, no. 21, pp. 2847–
139
2858, Sep. 2018, doi: 10.1177/0021998318759183.
[127] M. Anders, D. Zebrine, T. Centea, and S. R. Nutt, “Process diagnostics for co-cure of
sandwich structures using in situ visualization,” Compos. Part A Appl. Sci. Manuf., vol.
116, no. May 2018, pp. 24–35, 2019, doi: 10.1016/j.compositesa.2018.09.029.
Abstract (if available)
Abstract
This thesis is focused on the development of materials and models for the next generation of out-of-autoclave (OoA) prepregs that reduce cycle times, ensure robust laminate quality, and can be employed in the field for repairs. ❧ The kinetic behavior of a novel resin archetype was defined with a new autocatalytic model that accounts for features such as cure inhibition, fast-cure, and incomplete cure. A maximum degree of cure term that depends on the curing temperature of the inhibited resin was incorporated into the model. The model was shown to match experimental results closer than a conventional autocatalytic cure equation. ❧ A revised gel viscosity equation with a resin curing temperature term was created to describe the rheology of this material. Resin elements such as rapid gelation and reduced viscosity below the curing temperature were described by the model. The revised model was shown to predict viscosity behavior more accurately than the conventional gel model. The rheology model was employed to assess the predictive capability of effective flow number in composite laminates. It was shown that different prepreg materials with the same effective flow number reach similar levels of resin saturation and flow-based porosity. ❧ The revised gel model and novel autocatalytic kinetic model were employed in a cure cycle process optimization method to reduce cycle times without reducing laminate quality. Using the optimization method, cure cycles of less than an hour were created. Panels that were fabricated with the optimized cure cycles reached full cure and saturation. It was shown that air evacuation has the greatest effect on the number of void defects in a laminate, with mid-stage hold having a lesser impact. ❧ An in-field repair solution was presented featuring semi-preg formatting and a room temperature stable resin. The performance and quality of the room temperature stable semi-preg was compared to conventionally formatted epoxy prepregs and wet laid panels processed with double vacuum debulking (DVD). The semi-preg exhibited significantly lower porosity than both the conventional prepregs and the wet laid materials when processed in an in-field repair environment. ❧ An in-situ observation tool was created to generate insights on the cure processes of scarf repair patches. The tool featured an observation window that allowed for video recordings of panels undergoing cure. The in-field repair solution, conventional epoxy prepreg, and DVD wet layup materials were examined using the scarf repair tool. Videos taken showed how gas bubbles formed and remained in the wet laid DVD materials and conventional prepregs as void defects but were able to be evacuated with the semi-preg formatting. ❧ Overall, the work presented here offers insights on the processing of prepreg materials. The models, optimization methods, and experimental results can provide an understanding of OoA processing that can lead to reduced cycle times, improved laminate quality, and increased flexibility for in-field repairs.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Fabrication and analysis of prepregs with discontinuous resin patterning for robust out-of-autoclave manufacturing
PDF
Sustainable manufacturing of out-of-autoclave (OoA) prepreg: processing challenges
PDF
Defect control in vacuum bag only processing of composite prepregs
PDF
In situ process monitoring for modeling and defect mitigation in composites processing
PDF
Material and process development and optimization for efficient manufacturing of polymer composites
PDF
Void evolution in vacuum bag-only prepregs
PDF
Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites
PDF
Developing efficient methods for the manufacture and analysis of composites structures
PDF
Chemical recycling of amine/epoxy composites at atmospheric pressure
PDF
In situ process analysis for defect control during composites manufacturing
PDF
Vacuum-bag-only processing of composites
PDF
Novel processing of liquid substrates via initiated chemical vapor depostion
PDF
Polymer flow for manufacturing fiber reinforced polymer composites
PDF
Sound transmission through acoustic metamaterials and prepreg processing science
PDF
Characterization, process analysis, and recycling of a benzoxazine-epoxy resin for structural composites
PDF
Processing, thermal-oxidative stability and thermal cyclic fatigue of phenylethynyl-terminated polyimides
PDF
Slurry based stereolithography: a solid freeform fabrication method of ceramics and composites
PDF
Vapor phase deposition of dense and porous polymer coatings and membranes for increased sustainability and practical applications
PDF
High temperature creep behaviors of additively manufactured IN625
PDF
Development of composite oriented strand board and structures
Asset Metadata
Creator
Bender, David Boursier
(author)
Core Title
Efficient manufacturing and repair of out-of-autoclave prepreg composites
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Chemical Engineering
Degree Conferral Date
2021-08
Publication Date
07/29/2021
Defense Date
07/26/2021
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Aerospace,composites,fast-cure,Materials,OAI-PMH Harvest,out-of-autoclave,prepreg,repair
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Nutt, Steve (
committee chair
), Grunenfelder, Lessa (
committee member
), Gupta, Malancha (
committee member
), Ravichandran, Jayakanth (
committee member
), Williams, Travis (
committee member
)
Creator Email
dbender@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC15661060
Unique identifier
UC15661060
Legacy Identifier
etd-BenderDavi-9929
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Bender, David Boursier
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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
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
Repository Email
cisadmin@lib.usc.edu
Tags
composites
fast-cure
out-of-autoclave
prepreg
repair