University of Southern California Dissertations and Theses

Polymer flow for manufacturing fiber reinforced polymer composites

(USC Thesis Other)

Polymer flow for manufacturing fiber reinforced polymer composites

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2024 Patricio Martinez Martinez
POLYMER FLOW FOR MANUFACTURING FIBER REINFORCED POLYMER
COMPOSITES
by
Patricio Martinez Martinez
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
(MECHANICAL ENGINEERING)
December 2024

ii
Acknowledgements
First, I would like to thank my advisor, Prof. Steve Nutt. When I first took MASC 310 with
you in my sophomore year, I could not have foreseen how far I would come, thanks to you. You
have provided for me an opportunity to learn and become a part of the composites community.
With your help I’ve gathered skills, knowledge and connections that will support me for the rest
of my professional career.
Beyond Prof. Nutt, I would also like to thank my fellow members of the M.C. Gill Composite
Center. For being there, helping each other go through the PhD process. I would especially like to
thank our lab manager, Yunpeng Zhang, for keeping the lab equipment running, and fixing it
promptly whenever it went down, relieving us students from maintenance duties. Special thanks
to Tim Ceneta, Bill Edwards, Bo Jin and Mark Anders, who served as a mentors, peers and friends.
I want to thank the talented undergraduate researchers who worked under me during this time:
Andrius Stankus, Ryan Kraemer and Ashwini Balaganesh.
I would like to thank the following people for technical assistance and support: Arnaud
Dereims of ESI Group for support with the Visual-RTM/PAM-RTM software; and Philip Tayton
and Jack Burns from Mallinda, Inc for their support with the vitrimer resins used in the latter half
of this dissertation.
To my closest friends, Pablo Lamothe, Daniel Sada and Arturo Guerra, thank you for helping
me take my mind of work, and keeping our friendship strong despite me being so far away.
To Vivian, Haley and the rest of my D&D group; thank you for keeping me sane by forcing
me out of the lab with our scheduled D&D sessions. It gave me something to look forward to after
a busy week, and an opportunity to flex my creative muscles in a way my research did not.

iii
I would not be where I am now without my family. My mom and dad, Cristina and Enrique
Martinez, thank you for hearing me out, letting me vent and chat whenever I was walking home
from the lab. My siblings, Erika, Alejandro and Santiago; I know how lucky I am to be so close
with all of you, thank you for being there for me, whether it be discussing serious life decisions,
meeting together on Zoom, or just sharing funny pictures and videos; it really kept me going.

iv
Table of Contents
Acknowledgements ....................................................................................................................... ii
List of Tables................................................................................................................................ vii
List of Figures............................................................................................................................. viii
Abstract....................................................................................................................................... xiii
Chapter 1. Introduction................................................................................................................ 1
1.1 Motivation............................................................................................................................. 1
1.2 VBO Prepreg and Semi-preg ................................................................................................ 2
1.3 Liquid Molding Processes .................................................................................................... 3
1.3.1 Resin Transfer Molding ................................................................................................. 4
1.3.2 Vacuum Infusion ............................................................................................................ 5
1.4 Composite Sustainability ...................................................................................................... 6
1.5 Composite Joining ................................................................................................................ 8
1.6 Vitrimers ............................................................................................................................. 10
1.7 Scope of Dissertation .......................................................................................................... 12
Chapter 2. Droplet Spreading on Unidirectional Fiber Beds.................................................. 15
2.1 Introduction......................................................................................................................... 15
2.2 Materials and Methods........................................................................................................ 18
2.2.1 Resin characterization and fluid selection ................................................................... 18
2.2.2 Contact Angle Comparisons ........................................................................................ 19
2.2.3 Fiber Beds.................................................................................................................... 20
2.2.4 Droplet Deposition....................................................................................................... 22
2.2.5 Droplet Grid Tests........................................................................................................ 23
2.3 Results................................................................................................................................. 24
2.3.1 Contact Angle Comparisons ........................................................................................ 24
2.3.2 Single Droplet Test....................................................................................................... 25
2.3.3 Aggregate Tests Results............................................................................................... 26
2.3.4 Quilted Fiber Bed......................................................................................................... 28
2.3.5 Droplet Grid Optimization........................................................................................... 29
2.4 Discussion........................................................................................................................... 32
2.4.1 Absorption Kinetics ..................................................................................................... 32
2.4.2 Spread kinetics............................................................................................................. 33
2.4.3 Effects of Macrofeatures.............................................................................................. 35
2.4.4 Droplet Arrays.............................................................................................................. 35
2.4.5 Facsimile Fluid............................................................................................................. 36
2.5 Conclusions......................................................................................................................... 37

v
Chapter 3. Thickness Variation in Contoured Composite Parts by Vacuum Infusion......... 39
3.1 Introduction......................................................................................................................... 39
3.2 Experimental Methods........................................................................................................ 41
3.2.1 Fabric Characterization................................................................................................ 41
3.2.2 Laminate Production.................................................................................................... 43
3.2.3 In-situ Observation....................................................................................................... 44
3.2.4 Initial FEA Simulation ................................................................................................. 45
3.2.5 Revised Compaction and Geometry ............................................................................ 46
3.3 Results and Discussion ....................................................................................................... 47
3.3.1 In-situ Observation....................................................................................................... 47
3.3.2 Analysis of Polished Sections...................................................................................... 49
3.3.3 Simulation.................................................................................................................... 50
3.3.4 Thickness Measurements............................................................................................. 51
3.3.5 Modified Compaction Response.................................................................................. 53
3.3.6 Modified Simulations................................................................................................... 55
3.4 Conclusions......................................................................................................................... 59
Chapter 4. Flax-Reinforced Vitrimer Epoxy Composites Produced via RTM...................... 61
4.1 Introduction......................................................................................................................... 61
4.2 Materials and Methods........................................................................................................ 64
4.2.1 Resins........................................................................................................................... 64
4.2.2 Fabrics.......................................................................................................................... 65
4.2.3 RTM Part Production ................................................................................................... 67
4.2.4 Re-Forming.................................................................................................................. 68
4.2.5 Mechanical Testing ...................................................................................................... 69
4.3 Results and Discussion ....................................................................................................... 70
4.3.1 Cured Laminate General Properties............................................................................. 70
4.3.2 Mechanical Properties.................................................................................................. 72
4.3.3 Re-Forming.................................................................................................................. 74
4.4 Conclusions......................................................................................................................... 78
Chapter 5. Resistance Welding of Carbon Fiber Reinforced Vitrimer Composites............. 80
5.1 Introduction......................................................................................................................... 80
5.2 Materials and Methods........................................................................................................ 82
5.2.1 Materials ...................................................................................................................... 82
5.2.2 Resistance Welding Experimental Setup ..................................................................... 83
5.2.3 Laminate Preparation................................................................................................... 84
5.2.4 Heating Element Preparation ....................................................................................... 86
5.2.5 Welding and Lap Shear Testing ................................................................................... 87
5.2.6 Reversing the Welded Joint.......................................................................................... 89
5.3 Results and Discussion ....................................................................................................... 89
5.3.1 Laminate and Heating Element Manufacturing........................................................... 89
5.3.2 Welding Temperature Tests.......................................................................................... 90

vi
5.3.3 Repeated Welding ........................................................................................................ 93
5.3.4 Reversing the Welded Joint.......................................................................................... 99
5.4 Conclusions....................................................................................................................... 100
Chapter 6. Conclusions and Suggested Future Work............................................................ 103
6.1 Droplet Deposition and Semipreg..................................................................................... 103
6.1.1 Conclusions................................................................................................................ 103
6.1.2 Suggestions for Future Work ..................................................................................... 104
6.2 Vacuum Infusion Thickness Deviation and Prediction..................................................... 105
6.2.1 Conclusions................................................................................................................ 105
6.2.2 Suggestions for Future Work ..................................................................................... 106
6.3 Flax-reinforced Vitrimer Composites............................................................................... 107
6.3.1 Conclusions................................................................................................................ 107
6.3.2 Suggestions for Future Work ..................................................................................... 109
6.4 Vitrimer Welding................................................................................................................110
6.4.1 Conclusions.................................................................................................................110
6.4.2 Suggestions for Future Work ......................................................................................112
6.5 Broader Implications and Final Thoughts..........................................................................113
References...................................................................................................................................115
Appendix A : Droplet Spreading on Non-Unidirectional Fabrics ........................................ 126
A.1 Fabrics.............................................................................................................................. 126
A.2 Results.............................................................................................................................. 127
A.2.1 Time to Full Sorption ................................................................................................ 127
A.2.2 Surface Coverage over Time..................................................................................... 127
A.3 Conclusions...................................................................................................................... 128
Appendix B : Effects of Repeated Reprocessing on Neat Vitrimer Parts............................ 130
B.1 Neat Vitrimer Part Production and Reprocessing ............................................................ 130
B.2 Mechanical Testing .......................................................................................................... 132
B.3 Results.............................................................................................................................. 133
B.4 Conclusions...................................................................................................................... 135

vii
List of Tables
Table 2.1: Comparison of properties for the different fiber beds. NCF, non-crimp carbon
fabric. ......................................................................................................................... 21
Table 3.1: Properties of the triaxial fabric.................................................................................... 41
Table 3.2: Tool configurations used for experimental trials......................................................... 43
Table 3.3: Tool Geometries used in FEA. .................................................................................... 46
Table 4.1: Comparison of the general properties and cost of flax, glass [19], and carbon
fibers [20]................................................................................................................... 63
Table 4.2: General properties for fabrics used in this study......................................................... 66
Table 4.3: Test matrix detailing the specifics of all parts produced for this study. ...................... 68
Table 4.4: General properties of the cured laminates................................................................... 70
Table 4.5: Coupons cut from FV-2 and used for the reshaping trials........................................... 74
Table 5.1: Welding parameters used for the first set of lap-jointed specimens, determined
from temperature measurements................................................................................ 88
Table 5.2: Welding parameters used for the second set of lap-jointed specimens. ...................... 89

viii
List of Figures
Figure 1.1: Schematics for liquid molding processes. (a) Resin transfer molding, (b)
vacuum infusion........................................................................................................... 4
Figure 1.2: Diagrams explaining the three composite joining methods. ....................................... 9
Figure 1.3: Cross-section micrographs of carbon reinforced vitrimer composites produced
via (above) standard VBO prepreg method; (below) cured ply consolidation
method. .......................................................................................................................11
Figure 2.1: Rheological profiles for PMT-F4A, (a) Following the filming process thermal
cycle (b) With three separate thermal histories ......................................................... 19
Figure 2.2: Top and bottom views of the two distinct types of fabrics. Fiber direction is
indicated with arrows................................................................................................. 21
Figure 2.3: Diagram showing the droplet deposition test set up. ................................................ 22
Figure 2.4: Diagram showing the measurements made from side and top-views of the
droplet deposition test (left) and an example of the used for such measurements
(right), with the droplet edges outlined for clarity in the top view. .......................... 23
Figure 2.5: (a) Apparent contact angle of droplets of silicone oil and epoxy resin at 30 and
60 Pa·s viscosity. (b) Difference between the apparent contact angle of the 30
and 60 Pa·s droplets for both resin and facsimile fluid. ............................................ 25
Figure 2.6: Results from depositing 30 Pa·s viscosity fluid on fabric C. The left y-axis
shows results derived from the side view, while the right y-axis shows results
from top-down view. (a) Real time. (b) Time scale normalized to 𝑡ℎ0. ................... 26
Figure 2.7: Time lapse of side view for all combinations of fluid viscosity and fabric types.
Red line corresponds to the baseline from which height is measured, equal to the
height of the droplet at 𝑡 = 𝑡ℎ0 ℎ = 0. ..................................................................... 27
Figure 2.8: Relationship between time to full sorption and fabric type....................................... 27
Figure 2.9: Relationship between fiber bed type and spread, in directions across and along
the fibers. ................................................................................................................... 28
Figure 2.10: Images showing the contours of the area infiltrated by fluid over time, overlaid
onto a topo-graphical representation of the fiber bed for the 30 Pa·s viscosity
fluid tests on (above) fabric C and (below) fabric D. ............................................... 29
Figure 2.11: Diagrams of droplet grid guides used. From left to right: 1. square grid, 2.
staggered grid, 3. tight-staggered grid. ...................................................................... 30

ix
Figure 2.12: Images showing the extent of fluid coverage for all three grids, including 5 min
and 150 min after deposition. .................................................................................... 31
Figure 2.13: Graph showing the area covered by the droplets using all three grids.................... 32
Figure 3.1: Compression stress versus strain response for a single ply of fabric. ....................... 42
Figure 3.2: Permeability vs. Fiber Volume Fraction in x- and y-directions for the triaxial
fabric. ......................................................................................................................... 43
Figure 3.3: Experimental configuration for laminate infusion. (left) Schematic of the laser
displacement system. (right) Experimental setup. .................................................... 44
Figure 3.4: Position and thickness data during infusion with Tool B. (a) Laser sensor
acquired position data. (b) Thickness data, derived from position data. ................... 48
Figure 3.5: Micrograph of the cross-sections for the corner region for all five tool
geometries. For Tool A, the center of the laminate is displayed. The tool side is
facing downwards for all samples. ............................................................................ 49
Figure 3.6: Thickness across the centerline of the laminates from FEA simulation using the
laminate mesh. (a) Tool geometry set 1, varying corner angle; (b) tool geometry
set 2, varying corner radius........................................................................................ 51
Figure 3.7: Thickness measurements of laminates centerline for all experimental trials,
compared with FEA. The ‘inlet’ and ‘outlet’ points are indicated with enlarged
markers. A zoomed-in look at the corner region of the FEA results is also shown. .. 52
Figure 3.8: Process for obtaining the modified compaction results for Tool B. (a) Region
positions labeled on Tool B; (b) average thickness of the different regions and
pressure versus time; (c) thickness of the different regions versus pressure. ............ 54
Figure 3.9: Modified compaction stress versus strain responses for the (a) corner, (b) edge,
and (c) flange regions of each tool geometry. Also included is the single-ply
stress versus strain response used in LM simulations. .............................................. 55
Figure 3.10: Measured, UDG and LM uncompressed surface profiles for the corner region
of all tool geometries. ................................................................................................ 56
Figure 3.11: Thickness of UDG simulation vs. experimental results. (a) Plot showing the
entire part length, with ‘inlet’ and ‘outlet’ points labeled, and (b) enlargement of
the corner region. ....................................................................................................... 57
Figure 3.12: Thickness deviation between (a) the outlet and the inlet and (b) the outlet and
the corner. .................................................................................................................. 58
Figure 4.1: Rheology graphs for Vitrimax T130 (left) and RTM-6 (right) following the
specific injection-and-cure temperature cycles used for each resin. ......................... 65

x
Figure 4.2: Surface micrographs of the virgin flax (a), glass (b), and recycled flax (c)
fabrics ........................................................................................................................ 66
Figure 4.3: Test cell used to produce laminates in this study [93]............................................... 67
Figure 4.4: Tool used for bending of vitrimer parts. (Top) A picture of the tool. (Bottom)
Schematic................................................................................................................... 69
Figure 4.5: Cross-section micrographs for select laminates. (a) FV-1, (b) FR, (c) GV, (d)
rFV. Some visible voids are circled in yellow. .......................................................... 71
Figure 4.6: High resolution cross-sections of (left) sample FV-1 and (right) sample GV. ......... 72
Figure 4.7: Tensile test results for all samples. (left) Entire curves displayed. (right) Closeup of the flax laminates.............................................................................................. 72
Figure 4.8: Mechanical properties for all coupons. (a) Standard bar graph. (b) Properties
divided by density...................................................................................................... 73
Figure 4.9: Thickness values of the coupons cut from FV-2 after different degrees of
reshaping.................................................................................................................... 74
Figure 4.10: Micrographs of the center region of reshaped coupons. (a) Coupon 4 after
initial bending. (b) Coupon 5 after subsequent straightening.................................... 75
Figure 4.11: Results of the tensile testing of sample FV-2, with coupon 1 being as is and
coupons 2 and 3 having undergone bending and straightening. (Left) Stress vs.
strain curve for each coupon, with the fitted curve used to determine the tensile
modulus. (Right) Bar graph showing the aggregate tensile properties for each
coupon........................................................................................................................ 76
Figure 4.12: Micrographs of select FV-2 coupons after tensile testing. ...................................... 77
Figure 4.13: Cross-sections of bent coupons. (a) Standard pressing method with major
defects highlighted; (b) hand roller method. ............................................................. 78
Figure 5.1: Graphs showing the rheological behavior of the vitrimer resin. (a) Rheology of
the resin filming and prepreg pressing process; (b) Rheology during cure............... 83
Figure 5.2: Resistance welding setup. (a) Isometric view, (b) Labeled front view crosssection diagram, (c) Side view cross-section, (d) Circuit diagram............................ 84
Figure 5.3: Diagram showcasing the method used to produce heating elements. ....................... 87
Figure 5.4: Micrographs of select cross-sections (a) Vitrimer composite laminate, (b) Center
region of a virgin heating element ............................................................................. 90

xi
Figure 5.5: Graph showing the time required to reach 100, 130 and 180°C at specified
welding power values, and the welding time and welding power used for the two
sets of lap joints. ........................................................................................................ 91
Figure 5.6: Graph showing the shear strength for the lap joints with welding conditions
derived from the temperature tests. ........................................................................... 92
Figure 5.7: Micrograph cross-sections across the centerline of resistance welded lap-joints
for the second set of welding conditions. Bond lines are shown with dashed
yellow lines................................................................................................................ 93
Figure 5.8: Bar graph showing the shear strength for the third set of lap joints after repeated
welding. The × designates a joint that failed at the heating element, preventing
further testing. The joint 2-3 was mistakenly welded with 40W 135s for the sixth
weld, marked with *. ................................................................................................. 94
Figure 5.9: Micrographs showing the surface of the bottom adherend after lap-shear testing
for the third set of joints, after variable number of welds. ........................................ 96
Figure 5.10: Micrograph cross-sections across the centerline of resistance welded lap-joints
for iterative welding at 30 W for 450 s. Bond lines are shown with dashed
yellow lines................................................................................................................ 97
Figure 5.11: Graphs detailing the iterative welding tests. (a) Lap shear strength vs welding
iteration; (b) Joint stack thickness vs welding iteration............................................. 98
Figure 5.12: Photographs showing the surfaces of the adherends after reversing the weld. ..... 100
Figure A.1: Surface images of the two additional fabrics. (Left) plain weave; (right) twill
weave ....................................................................................................................... 126
Figure A.2: Graph showing the time to full sorption for all six fabrics for the 30 Pa·s
facsimile fluid .......................................................................................................... 127
Figure A.3: Images showing the surface covered by the droplet over the topographical map
of the fiber bed. Times of each contour are at 10, 100, 500, 1000, 2000 seconds
after droplet deposition, as well as at the last moment at which the droplet can be
observed. (above) PW Fabric; (below) TW Fabric.................................................. 128
Figure B.1: Neat resin thermoforming process. (a) As-cured neat resin sample. (b) Pressed
sample removes bubbles. (c) Reconsolidated, void-free sample............................ 131
Figure B.2: Set of pictures showing neat resin samples after certain number of reshaping
cycles, just before cutting and testing...................................................................... 132
Figure B.3: Results from the tensile test. (Left) maximum tensile strength, (Middle) Young's
modulus, (Right) maximum elongation................................................................... 133

xii
Figure B.4: Results from the three-point bending test. (Left) maximum flexural strength,
(Middle) flexural modulus, (Right) maximum elongation ...................................... 134
Figure B.5: Results from the short-beam shear tests. ................................................................ 135

xiii
Abstract
This thesis is focused on the importance of the polymer matrix in fiber reinforced polymer
composites. The aim is to improve knowledge of process parameters for the advancement of
composite manufacturing methods and processes. This work was separated into four projects that
explored different degrees of polymer flow behavior. The first project was on single-droplet
deposition on dry fabric, second on the relationship between infusion parameters and thickness
deviation for vacuum infusion, the third on the compatibility of a novel vitrimer resin formulation
and the RTM process, and the fourth on the flow of a cured thermoset vitrimer matrix via resistance
heating for fusion bonding of composites.
The first project investigated the flow behavior of a single droplet of resin onto dry fabric to
lay the groundwork for a robust prepreg format manufacturing method. Two facsimile fluids of
different viscosities, 30 and 60 Pa·s, were deposited onto the surface of four different
unidirectional fiber beds, and the droplet absorption time and spread distances were measured.
Time to full sorption was dominated by viscosity while spread behavior depended on the macrolevel topology of the fabric. Using the results of the single-droplet tests, a series of droplet arrays
was developed, showcasing how the droplet spread behavior can be leveraged to achieve different
discontinuous resin distributions.
The second project produced a workflow to predict the thickness deviation in V-shaped
laminates produced by vacuum infusion. This required modifications to the workflow of a
commercial infusion simulation software (PAM-RTM), including the development of a user
defined geometry and custom material properties. The relationship between fabric compaction
behavior and tool geometry was characterized and included in the simulation. The results of the

xiv
modified simulation workflow were compared with dynamic thickness measurements of vacuum
infused parts, showing significant accuracy improvements over the unmodified simulation.
The third project tested compatibility between a novel vitrimer polymer resin and the RTM
manufacturing process. After rheological characterization of the resin, processing conditions that
would allow for infusion were found and selected. Parts were produced using both the novel
vitrimer epoxy and conventional RTM epoxy as well as both flax and glass fabrics. The mechanical
behavior of these two sets of reinforcement and matrix were measured and compared. Flax fabric
was recovered from a recycled flax-vitrimer composite and used to produce a second-life laminate,
which showed little degradation in mechanical properties compared to the fresh flax laminate.
Finally, the reshaping behavior of vitrimer composites was put to the test, showcasing their ability
to reprocess while revealing the potential damage that can occur if reshaping was done carelessly.
The final project further investigated the unique properties of vitrimer epoxies. Due to the
vitrimers’ capacity for post-cure reprocessing and bonding, a resistance welding method developed
for thermoplastic composites was tested with vitrimer composites. Adherends were joined in a
single lap-joint geometry, and joint strength was tested. The bond strength of resistance welded
samples was observed to match or exceed the bond strength of those obtained through conventional
vitrimer manufacturing methods. Leveraging the reprocessability of vitrimers, adherends were rewelded repeatedly and measured to see the change in joint strength as a function of number of
weld cycles. Finally, as a proof-of-concept, the same welding parameters used for joining were
used to separate a resistance welded joint, showing the potential for temporary bonding via
resistance welding of vitrimers.
This dissertation offers numerous advancements and insights into composites manufacturing
and processing, setting the stage for a novel prepreg manufacturing method, developing a

xv
simulation workflow to detect potential defects in vacuum infused parts before manufacturing,
furthering the understanding of the mechanical properties of natural flax fibers and how they
compare to glass, characterizing the properties of vitrimers and potential for recycling, and
confirming the compatibility of and developing processes for vitrimer polymer manufacturing via
both RTM and resistance welding

1
Chapter 1. Introduction
1.1 Motivation
Composite materials, combinations of support and reinforcement, have been used in
everything from weapons, vehicles and buildings through the ages, appearing as early as 1500
B.C.E when a combination of mud and straw was made to provide structure to buildings [1]. The
materials known as composites in modern-day parlance, however, did not come into existence until
the mid-19th century [2]. Fiber-reinforced polymers (FRPs) – as the name suggests – are a polymer
matrix, thermoset or thermoplastic, reinforced by fibers that can vary both in material and length.
Continuous fibers possess high strength and elastic modulus; however, without a supporting matrix
they would easily bend and buckle. As an engineering material, carbon fiber reinforced polymers
(CFRPs) provide mechanical properties, stiffness and strength, that can match or surpass metals
such as aluminum and steel while providing significant weight reductions and resistance to
corrosion and fatigue [1–4]. Therefore, in the aerospace industry, where weight reductions lead to
reduced fuel needs, extended range, and overall lower costs, composite materials have been rapidly
growing in the past few decades, with projections to continue that expansion; modern airplanes
like the Boeing 787 and the Airbus A350 are more than 50% composite [4]. Much like there is
customizability in the material make-up of FRPs, there too exists customizability in the
manufacturing process – from infusion processes like wet-layup, vacuum infusion (VI), and resin
transfer molding (RTM), to the use of pre-impregnated fabric (prepreg), laid up either by hand or
through automated processes and cured [5]. The latter is the preferred method of manufacturing
for the aerospace industry, typically relying on autoclave cure for consistent high-quality parts with
complex geometries.

2
In this dissertation, four projects on composites are presented. The first two focused on the
improvement of robustness for out-of-autoclave (OoA) processes – vacuum bag-only (VBO)
prepreg and liquid molding in the first and second projects, respectively – that have a lower barrier
of entry. The last two focused on vitrimer composites, with the third project exploring their
potential to be a sustainable alternative to traditional epoxy composites, and the fourth covering
the joining of composite structures and their compatibility with preexisting manufacturing methods
1.2 VBO Prepreg and Semi-preg
Autoclave processing is a very forgiving method, yielding consistently high-quality parts with
low void content, high fiber volume fractions (𝑉𝑓) and excellent mechanical properties [6]. Outof-autoclave processes on the other hand, can match the quality of the autoclave; however,
consistency and robustness are not as high.
One method commonly used to produce VBO prepregs is the hot-melt process [7]. Resin is
spread onto release paper and then pressed on either side of carbon fabric [8], resulting in partial
impregnation that allows for in-plane air evacuation during the initial stages of the cure cycle. This
air evacuation method can lead to air entrapment for parts that deviate from standard conditions
and geometries [9,10]. Prior work to alleviate these issues led to the development of a custom
prepreg format relying on discontinuous resin distribution, dubbed semi-preg [11].
Semi-preg is an out-of-autoclave prepreg format that relies on through-thickness air
evacuation to yield high quality, robust parts, even in suboptimal processing conditions [12] and
for complex geometries [13]. A wide variety of manufacturing methods has been developed to
produce semi-preg, including direct resin deposition relying on fabric architecture [14], partial
deposition of masked resin films [12], and deposition of discontinuous resin films [15].

3
By allowing through-thickness permeability, air evacuation distances are significantly
reduced. Thus, air entrapped between individual plies, evolving from dissolved moisture, or
captured due to complex part geometry can more easily escape, reducing void content and
improving quality. Compared to aerospace industry standard VBO prepreg, semi-pregs
consistently yielded equivalent or higher quality parts.
One prepreg manufacturing method previously discussed but not yet implemented is the direct
deposition of individual resin droplets onto a dry fiber bed [11]. This method has the potential for
on-demand customization of resin content and distribution as well as reduction in both prepreg
waste and the need for freezer storage. However, the conditions for this method would rely on
careful understanding of the relationship between individual resin droplets and the fiber bed
architecture, which is explored in this work.
1.3 Liquid Molding Processes
Liquid molding is an umbrella term for composite manufacturing processes that rely on direct
infusion of polymer resin into dry fabric. Unlike prepreg, which requires freezer storage because
its resin is mixed and applied to the fabric several weeks or even months prior to layup and curing,
liquid molding processes typically mix the resin immediately prior to infusion. This reduces waste,
shipping and storage needs, and overall cost. The two liquid molding processes relevant to this
dissertation are vacuum infusion and resin transfer molding, shown in Figure 1.1.

4
Figure 1.1: Schematics for liquid molding processes. (a) Resin transfer molding, (b) vacuum infusion.
The general steps for resin infusion are as follow: layers of dry fabric are placed onto a onesided tool, typically relying on some form of tackifier, tape, or adhesive to hold the fabric in place,
forming a preform; the preform is then sealed, evacuated, and low viscosity resin is infused into
the fabric, until fully saturated; and the part is cured, then demolded and trimmed as necessary.
1.3.1 Resin Transfer Molding
Resin transfer molding (Figure 1.1a) is a closed-mold process, where the dry fabric preform
is sealed between two matched tools and the resin is injected at elevated pressure. The two-mold
process allows for complex geometry, careful matching of geometric tolerances, and high fiber
volume content, approaching what is possible with prepreg manufacturing. Much like in autoclave
curing, the elevated pressure of RTM can suppress void evolution. Furthermore, using highpressure resin injection can allow for rapid manufacturing, with infusion times measured in the
order of minutes. Heated tools are used for curing, and tool-embedded pressure and temperature
sensors are common.

=

5
Due to the complexity of the infusion tool, fabric preforming is typically done on a separate
tool without heating, resin ports or sensors. Preforming may be done in stages, gradually shaping
the fabric to the final geometry. RTM parts are commonly net shape, with only minor resin
trimmings being removed.
Resin viscosity needs to be approximately < 10 Pa·s for infusion, with high pressure RTM
requiring even lower viscosity. Heating of the resin is common to reduce the viscosity, with RTM
resins typically stable at sub-cure temperatures. Resin systems are selected for long pot-life and
rapid cure.
RTM is limited by requiring high-complexity leak-proof tools that must withstand repeated
cycles of elevated temperature and pressures. As such, equipment costs for RTM are high, and
therefore it is a process preferred for small-to-medium scale parts, particularly with high-volume
production where modifications to the parts is not expected.
1.3.2 Vacuum Infusion
Vacuum infusion (Figure 1.1b), also known as vacuum assisted resin transfer molding
(VARTM), is a variation on RTM. VI uses only a one-sided tool, with the other tool replaced by a
vacuum bag. This significantly reduces cost compared to RTM; however it also reduces part
complexity, dimensional tolerance and achievable fiber volume fractions. VI is commonly used
for large scale parts, such as boat hulls and wind turbine blades, where a two-part mold would be
prohibitively expensive.
Resin selection for vacuum infusion also requires extended pot-life, but the need for low
viscosity is more important than in RTM, typically requiring sub-1 Pa·s. Room temperature cure
is also often a preference for VI resins, removing the need for ovens that can accommodate large
parts.

6
Unlike in RTM, the resin is held open to the air, with the driving pressure for infusion being
the difference between atmospheric pressure and vacuum in the fabric. Consequently, infusion
times tend to be longer for VI, and multiple-inlet systems are more common. Flow media are
commonly employed to ensure full saturation, significantly increasing the permeability at the part
surface and reducing the distance resin must flow through the fabric.
This same pressure difference driving infusion (1 atm) is also responsible for compaction of
the preform, which results in lower fiber volume fractions. Separate preforming is less common
with vacuum infusion, with fabric being laid up directly on the tool. Due to the flexible nature of
the vacuum bag, part dimensions are expected to change between layup, initial vacuum, and
infusion. As the preform saturates with resin, the pressure within the bag increases, reducing the
total compaction pressure, resulting in an increase in part thickness. Deviations in expected
geometries can lead to scrapped parts or require post-processing that can drive up the cost – as
such being able to predict these deviations before infusion is ideal.
The relationship between tool geometry and thickness variation in VI is explored in this work,
and a method for predicting final part geometry through simulation is demonstrated.
1.4 Composite Sustainability
Composites are often touted as sustainable materials for air travel due to significant lightweighting: lighter aircraft require less fuel and thus produce lower emissions. However, there are
several areas where composites could see significant sustainability improvements, three of which
are relevant to this work. The first is composite reinforcement, the second is end of life (EoL)
processing, and the third is repair of composite structures.
High performance composites use a wide variety of reinforcements; however, the two that
make up the largest fraction are carbon fiber and glass fiber. Carbon fibers are derived from

7
petroleum, and the production of carbon and glass fibers has a significant carbon footprint [16].
While there is some work trying to find a more sustainable alternative to carbon fiber precursors
[17,18], these fibers are generally non-renewable, and thus there is significant motivation towards
finding a sustainable alternative.
Natural fibers are seen as a potential alternative reinforcement. Sourced from plants, they
typically have lower emissions and energy consumption. However, the mechanical properties of
natural fibers tend to be lower than those of both carbon and glass [16,19,20]. Additionally, there
is significant variation in those same properties, arising from variations in the plants from which
they are sourced from. The potential to replace carbon is low due to the extreme difference in
strength and modulus, but some natural fibers can approach the properties of glass. Given the large
fraction of FRPs produced with glass fibers, particularly in non-aerospace industries, there is
potential for adoption of natural reinforcements to increase sustainability. A portion of this
dissertation focuses on flax fibers as a sustainable substitute for glass, specifically in combination
with a highly sustainable polymer matrix.
A significant portion of EoL processing of composites results in landfill disposal. Wind energy
alone is estimated to produce more than 2 Mt of waste by 2050, with end-of-life waste
corresponding to 55-84% of that [21]. The aerospace industry is also expected to produce
significant EoL waste, estimated to be over 20,000 tons of carbon fiber waste alone by 2035 [22].
As such there is significant drive towards improving recycling technologies. Current recycling
methods often lead to decreases in mechanical performance and reduction in fiber length [23],
reducing the potential uses for recycled materials – a case of down-cycling as opposed to true
recycling. Vitrimers, a novel class of polymer matrix that is the focus of the second half of this

8
dissertation, have demonstrated full recyclability through a simple method, and the effects of this
recycling process on the properties of the recovered fibers are investigated.
Damage to composite structures exacerbates the EoL processing issue, reducing part lifetime
and efficiency. Thus, repair of composites is important to restore lifetime and reduce the impact of
the damage. Repair is particularly important to the wind energy industry, where manufacturing
defects in wind blades often require repair, and costs must be kept at a minimum. Repairs are
regularly done in-field with limited processing capabilities due to a desire to minimize downtime
and cost [24]. Scarf repairs are often used, where the damaged section is removed and replaced
with new material. This material can be pre-cured to serve as a hard patch that requires bonding to
the parent structure, or it can be cured in-place as a soft patch [24,25]. Each of these has its own
advantages and disadvantages. Soft patch repairs can lead to material mismatch due to differences
in the materials used, as the processing capabilities available during repair are significantly
different than those available during manufacturing. Hard patch repairs require expensive tooling,
surface preparation for bonding and there may be difficulties in matching the part geometry [26].
1.5 Composite Joining
Joining of composite structures is typically done in one of three ways: mechanical fasteners,
adhesive bonding, or welding [27]. Mechanical fasteners are the lowest immediate cost option and
allow for temporary joining while also being the most flexible in terms of material. Fasteners
require some form of piercing through the composite as shown in Figure 1.2a, whether that be
drilling or riveting; this can significantly reduce the mechanical performance of the part, and as
such it is common to increase the thickness at the point of bonding through embedded doublers.
The doublers and fasteners increase both part count and part weight and can affect the surface
profile of the part, reducing fuel efficiency and life-time savings.

9
Figure 1.2: Diagrams explaining the three composite joining methods.
Adhesive bonding, as the name implies, is a process where two parts, composite or otherwise,
are joined together by an adhesive. The process is done in five stages: material selection, surface
preparation, adhesive application, pressing, and curing [27] – the last four stages depicted in
Figure 1.2b. Compatibility between the substrate and adhesive must be considered, requiring
surface treatment and favorable curing conditions. Surface preparation is crucial for a strong bond
and can be an expensive and time-consuming process. Unlike fasteners, adhesives provide a
permanent bond but don’t affect part count or weight. Adhesive bonding is the most materialagnostic process, capable of bonding metals, wood, and both thermoset and thermoplastic
composites. Finally, curing of the adhesive can take significant time, as the parts may not always
be suitable for oven curing after bonding. Co-cure is an alternative to adhesive bonding, where the
part and the adhesive are cured simultaneously to reduce processing time and eliminate the need
for surface pre-treatment. However, co-cure requires compatible resin formulations, careful preselection of parts and materials and can affect the available part geometry. Co-cure is common in
the manufacturing of sandwich structures, with the adhesive between the core and the face sheet
being cured in the same cycle.

10
Fusion welding relies on melting and subsequent freezing of the surface between the
adherends to form a continuous matrix, as depicted in Figure 1.2c. Of the three bonding methods,
it is the most restrictive because it requires compatible thermoplastics. Fusion welding uses the
simultaneous application of pressure and temperature. Temperature application is done through a
wide range of methods: frictional heating methods, like spin welding and ultrasonic welding; direct
heating methods, like infrared and laser welding; and electromagnetic welding methods, like
induction and resistance welding [28]. Each of these methods has some drawbacks. For example,
laser welding requires at least one of the substrates to be transparent, while resistance and induction
welding require a heat conductor at the welding surface. Some methods are only suitable for
welding thin parts. Compared to fasteners and adhesives, fusion welding is the most expensive
joining method, requiring specialized tools and highly skilled technicians; however, it does not add
parasitic weight, and welding times are short.
1.6 Vitrimers
Vitrimers are a type of thermoset polymer with a unique crosslink network. At service
temperatures, vitrimers behave like a typical thermoset. After elevating beyond a specific
temperature, however, bond exchange begins to occur, allowing the vitrimer matrix to ‘flow’ like
a viscoelastic fluid. This allows vitrimers to be reshaped, reprocessed and bonded after they have
been cured [29].
Within this study two different vitrimer formulations are used – Vitrimax T130 and Vitrimax
T100 – both produced by Mallinda, Inc. These vitrimers have imine bonds that allow for stress
relaxation and malleability at elevated temperatures [30,31]. Prior work with these polymers has
shown they can be reprocessed with little impact to mechanical properties. Furthermore, this

11
formulation is easily chemically recycled, allowing for recovery of long-scale reinforcement from
fiber reinforced vitrimer composites [32].
The unique bonding properties of vitrimers can also be leveraged for modifications to the
manufacturing process. The vitrimer formulations used in this study exhibit limited resin flow
during the cure stage, and as such, conventional VBO prepreg methods yield parts with high
porosity [33], as shown in Figure 1.3. Unlike conventional prepreg, parts produced via vitrimer
prepreg rely on full resin saturation in the pre-impregnation stage, which is achieved through
elevated temperature and pressure during the resin filming. Individual plies of vitrimer prepreg are
then cured into 1-ply laminates. The cured vitrimer composite plies are cut and stacked in the
desired layup sequence, generating cleaner waste than standard prepreg – cured composite waste
instead of prepreg scrap. The layup stack is consolidated into a final part via hot-pressing, bond
exchange allowing for the post-cure bonding of the individual plies into a single multi-ply
laminate.
Figure 1.3: Cross-section micrographs of carbon reinforced vitrimer composites produced via (above) standard
VBO prepreg method; (below) cured ply consolidation method.
Vitrimer polymers possess properties that allow for unique composite manufacturing methods,
with capabilities bridging the gap between thermoset and thermoplastics.

12
1.7 Scope of Dissertation
Chapter 2: This chapter reports a method to analyze parametric effects on the spread flow
kinetics of fluid droplets on unidirectional fiber beds. The purpose of this method is to guide the
design of droplet arrays for production of semi-preg. Volume-controlled droplets of a resin
facsimile fluid were deposited on various unidirectional carbon fiber beds and the flow behavior
was recorded. The time to full sorption (after deposition) and the maximum droplet spread
distances were measured. Experiments revealed that fluid viscosity dominated time to full
sorption—doubling the viscosity resulted in an 8- to 20-fold increase in sorption time, whereas
doubling fabric areal weight increased the time only by a factor of three. Droplet spread distance
was nearly invariant with fiber bed architecture and fluid viscosity. Overall, the study highlighted
the importance of fluid viscosity and the relationship between droplet size and fabric surface
features to predict single-droplet spreading. A series of droplet arrays were designed,
demonstrating how the results can be leveraged to achieve different resin distributions to produce
semi-preg optimized for OoA cure.
Chapter 3: This chapter reports an investigation into the thickness evolution of V-shaped
laminates produced by vacuum infusion. Thickness variations were monitored dynamically as a
function of VI process parameters. Process simulations were performed using finite element
software (PAM-RTM), and predictions of thickness along the length of the laminate were
compared with dynamic measurements. Initial simulations approximated the effects of corners on
preform deformation and fabric draping behavior. Subsequent modifications to the simulation
geometry and material properties were implemented to increase accuracy and more closely match
experimental measurements. User-defined geometry (UDG) simulations were used to predict both
the maximum corner deviation and the area of corner deviation with greater accuracy. A workflow

13
for use of analytical tools to design and control vacuum infusion processes was demonstrated. The
workflow leveraged process monitoring and modified process simulation tools to provide insight
into parametric effects and to guide process modifications to reduce product variability.
Chapter 4: This chapter reports a study where composite laminates were produced by RTM
using glass and flax fabrics and both vitrimer epoxy and aerospace-grade epoxy, both formulated
for liquid molding. Tensile and flexural properties were measured and compared, revealing that
the vitrimer composites exhibited equivalent performance in flexural strength and tensile modulus,
but slightly lower performance in tensile strength relative to reference epoxy composites. In
general, glass fiber composites outperformed flax fiber composites in tension. However, both glass
and flax fiber composites yielded roughly equivalent flexural strength and tensile modulus-toweight ratios. Flax fabrics were recovered from vitrimer composites by matrix dissolution, and a
second-life laminate showed full retention of the mechanical properties relative to those produced
from fresh flax. Finally, a demonstration of re-forming was undertaken, showing that simple pressforming can be used to modify the composite shape. However, re-forming to a flat configuration
resulted in local fiber damage and a decrease in mechanical properties. An alternative forming
method was demonstrated that resulted in less fiber damage, indicating that further refinements
might lead to a viable forming and re-forming process.
Chapter 5: This chapter reports an investigation into the welding behavior of prototype
vitrimer composites with respect to adjustable parameters and protocols. Resistance welding of
vitrimer composites using a method directly adapted from the welding of thermoplastics is
described. Adherend laminates were joined to a matrix-saturated carbon fiber heating element
through resistive heating, forming a single lap joint. Weld strengths matched or exceeded the
strength of composite parts produced using the manufacturer-recommended consolidation method.

14
Furthermore, repeating the welding process yielded greater shear strength, each sample easily
withstanding up to five weld-and-break cycles with no loss in strength. The findings highlighted
the suitability for repair of vitrimer matrix composites. Finally, a process for reversing a welded
joint was shown, demonstrating the potential for vitrimers for temporary joining.
Chapter 6: The final chapter provides a summary of the conclusions and outcomes of each
project, and the broader implications of the work presented in this dissertation. Suggestions are
made for future work to advance the work presented and further support the field of composites
research.

15
Chapter 2. Droplet Spreading on Unidirectional Fiber Beds1
2.1 Introduction
We investigate the effects of fiber bed architecture on the anisotropic flow behavior of fluid
droplets on and into unidirectional (UD) fiber beds. In particular, we determine the effects of fiber
bed areal weight and fluid viscosity on sorption time and spread distance. The work is motivated
by a need to support the design of prepreg formats with discontinuous resin distributions (semipregs).
Conventional out-of-autoclave (OoA) prepregs typically feature continuous resin films
partially impregnated into the fiber bed and thus rely on “edge breathing” for air removal [34–36].
Compared to autoclave prepregs, however, processing of OoA prepregs lacks robustness,
particularly in challenging conditions, such as poor vacuum, ply ramps, embedded doublers and
large parts [9,10]. In contrast, semi-pregs [37] feature discontinuous resin distributions that impart
high through-thickness permeability and increase process robustness compared to conventional
OoA prepregs [10,12,38,39]. Previous methods of fabricating semi-preg include hot-rolling resin
onto the tow overlaps of woven fiber beds [40], using a release film mask to press a discontinuous
film onto dry fibers [8] and dewetting a continuous resin film, then pressing onto dry fibers [38].
Recent work on semi-pregs has employed a polymer film dewetting approach to fabrication
[15,38,39,41]. The inherent versatility of the method permits deposition of a variety of patterns on
the fabric surface. However, the method has limitations. For example, the initial degree of
impregnation (DoI) is negligible, leading to a higher bulk factor than conventional vacuum bag1 This study was published in the Journal of Composites Science in January 2021 with Steve Nutt and Bo Jin as coauthors [128].

16
only (VBO) prepreg [41]. In addition, the method as described requires filming of a continuous
resin film prior to dewetting and pressing onto the fiber bed, adding cost to the manufacture.
Finally, for thin resin films, difficulties can arise in generating uniform discontinuous resin
distributions [41]. Thus, alternative methods of production are being evaluated, including gravure
printing and droplet deposition onto fiber beds [11]. Out of these methods, droplet deposition
allows for a resin distribution that is controlled based on the relationship between droplet position
and fiber bed architecture and does not require a prior filming step as the dewetting method does.
The present work constitutes a first step towards semi-preg production by droplet deposition.
Experiments were undertaken to understand the flow of a single droplet placed on a fiber bed
surface, with further scaling-up used to inform the future design of semi-preg with robust process
characteristics.
Fluid flow on porous surfaces and through fiber beds has broad relevance for composites
manufacturing [42,43] and has thus been studied extensively. However, most prior studies of
surface flow have assumed material isotropy, treating pores as tubes oriented normal to the surface
[44]. Fiber beds in composites, however, are anisotropic and permeability varies by orders of
magnitude in directions normal and parallel to the fibers [45]. Thus, studies of fluid flow through
fiber beds generally must consider flow through an anisotropic porous medium. Most often, such
studies consider flow through the entire fiber bed, leading to full saturation (e.g., modeling the
impregnation of an individual fiber tow [46]). These studies focus on infiltration, in which one
fluid (air) is fully displaced by another (resin) in a porous medium [47], and as such ignore flow
above and near the surface of the porous substrate.
Studies of fluid flow on a porous surface generally assume low-viscosity fluids (under 1 Pa·s)
[48]. Although similar low viscosity values are achieved by typical prepreg resins during cure,

17
these studies have limited relevance with regard to surface flow during droplet deposition. The
reason for this is that droplet deposition is performed at lower temperatures than those used during
cure to prevent advancing cure of the resin, leading to higher viscosities. The present study
considers higher viscosity fluids and focuses on local (near-surface) flow beneath individual
droplets during initial wet-out. The droplets used in semi-preg production must only wet-out
partially during the deposition stage in order to preserve connectivity of the dry spaces for air
evacuation during the de-bulking stage of VBO processing. Full flow and subsequent saturation of
the fiber bed only occurs after de-bulking, during cure of the laminate.
A combination of forces governs the fluid flow of a droplet on a solid surface, including
gravitational forces, viscous forces, and surface tension [48]. However, the effects of gravity can
be ignored for small droplets, leaving only viscous forces, surface tension and capillary forces to
govern the flow [49]. Similar forces control the flow of droplets dispersed on the surface of a
porous medium saturated with the same fluid [50]. However, for dry fiber beds, capillary effects
play a major role [42,43]. Specifically, standard wet-out phenomena lead to droplet spread on the
surface, increasing the coverage, while capillary effects promote absorption into the substrate,
reducing surface coverage and increasing impregnation [51].
In this study, we measured the surface flow of individual droplets on unidirectional fiber beds.
Facsimile fluids with moderate viscosity values were selected (30–60 Pa·s) to resemble polymer
resins used during hot-melt production of prepreg. Using the measured response of a single droplet,
we also produce droplet arrays to maximize surface coverage or to minimize interactions between
neighboring droplets. The results inform the design and production of semi-pregs, particularly the
spacings and patterns of resin droplets on fiber beds. Note that despite the match in fluid
viscosities, the facsimile fluid did not match the apparent contact angle nor the advancing droplet

18
edge slope. Thus, while the results presented here are useful for determining droplet spread
parameters and foreseeing how such droplets will behave in manufacturing conditions, further tests
must be performed with actual resin to determine parameters for prepregging.
The experiments reveal the effects of droplet parameters on spread rates. The droplet
absorption time depended strongly on viscosity: doubling the viscosity resulted in an 8- to 20-fold
increase in absorption time. However, droplet spread along the surface showed little variation, at
least within the fluid viscosity range tested. Similar relationships were noted with fiber bed areal
weight, which had little effect on spread distances, but caused marked changes in sorption time.
Finally, surface spread depended strongly on fiber bed architecture, particularly tow gaps caused
by stitching. Using these results, we demonstrated how droplets can be positioned on fiber beds to
ensure uniform impregnation, informing future methods for semi-preg production.
2.2 Materials and Methods
2.2.1 Resin characterization and fluid selection
An epoxy resin designed for aerospace applications was selected (PMT-F4A, Patz Materials
& Technologies, Benicia, USA). The resin viscosity was typical of B-staged resins used in the
production of conventional OoA prepreg. During prepreg production, the resin is pre-melted at
65–68 °C (150–155 °F), filmed at 68–72 °C (155–162 °F), then transferred to the fiber bed. The
process from pre-melting to cooling takes less than 90 min. Thus, in the present work, rheological
measurements were performed after each step in the thermal cycle, shown in Figure 2.1a. During
the initial pre-melting stage, resin viscosity averaged 49.7 Pa·s, while during the filming stage,
resin viscosity averaged 32.4 Pa·s, with a minimum of 27 Pa·s and a maximum of 52.5 Pa·s during
the entire cycle.

19
To ensure that the process was consistent with the manufacturer’s practices, the cure cycle
rheology was compared with a sample provided by the manufacturer that had not undergone the
melting process used for filming (A stage). The pre-melting process showed nearly identical cure
cycle rheology (Figure 2.1b).
(a) (b)
Figure 2.1: Rheological profiles for PMT-F4A, (a) Following the filming process thermal cycle (b) With three
separate thermal histories
Facsimile fluids were chosen to match the viscosity values of the resin during the filming
process (Figure 2.1a). Silicone viscosity standard fluids (General Purpose Silicone, Brookfield
Ametek, Middleborough, USA), with viscosities of 30 and 60 Pa·s (±1%), were selected. The use
of facsimile fluids enables droplet flow testing at room temperature, minimizing difficulties in
maintaining a uniform high temperature on the fiber bed while simultaneously recording data and
ensuring the viscosity does not change due to advancing cure.
2.2.2 Contact Angle Comparisons
Surface tension can influence fluid flow along the surface of a substrate and facsimile fluids,
despite matching resin viscosity, might exhibit different surface flow behavior. Consequently,
measurements of the apparent contact angle were performed, comparing the facsimile fluids to the
B-staged resin at the same viscosity. These measurements were performed using a goniometer
(Ramé-Hart Model 500, Plymouth, USA) from droplet deposition until apparent stabilization of

20
the contact angle. By determining the difference in contact angle between the facsimile and the
resin, a determination can be made of the validity of using the flow behavior observed for a
facsimile fluid to predict that of an actual resin.
2.2.3 Fiber Beds
Four unidirectional non-crimp carbon fabrics (UD NCFs) with different areal weights were
selected for use as dry fiber beds, including (A) 146 g/m2
(4.3 oz/yd2
), (B) 136 g/m2
(4.0 oz/yd2
),
(C) 305 g/m2
(9.0 oz/yd2
) and (D) 756 g/m2
(22.3 oz/yd2
) (FibreGlast Products #2596, #2585,
#2583 and #2595, Brookville, USA). UD NCFs were chosen because of previous work with similar
fabrics [6–8], their simplified geometry compared to woven fabrics, the similarity to tapes used in
automated tape layups and their growing use in vacuum infusion processes both in aerospace and
wind blades. Non-crimp fabrics rely on stitching to hold the fibers together, preventing the crimp
present on woven fabrics, increasing performance. Carbon fiber fabrics were selected due to their
common use in prepreg for the aerospace industry. Using a different fiber material would result in
different intra-tow capillary sizes based on fiber diameter, as well as differences in wettability
based on fabric-fluid surface parameters, leading to differences in flow. The fabrics contained
polyester binding to stitch layers together and impart ease of handling. The first three NCFs were
bound using polyester stitching perpendicular to the fibers on one side, spaced ~10 mm apart. The
heaviest weight fabric, however, featured binding in a diamond pattern and fiber tows with distinct
edges, as shown in Figure 2.2.

21
Figure 2.2: Top and bottom views of the two distinct types of fabrics. Fiber direction is indicated with arrows.
Prior to droplet deposition, a 19 mm strip of fabric was cut and the edges were secured with
tape to prevent fraying, leaving a 19 × 19 mm square of exposed fabric. The tape was positioned
perpendicular to the fiber direction, preventing free fiber edges from lifting and distorting the
surface. The fiber bed squares were examined using a digital stereo microscope (Keyence VHX5000, Osaka, Japan) to generate images and 3D contour maps of the surface. Fabric properties are
summarized in Table 2.1.
Table 2.1: Comparison of properties for the different fiber beds. NCF, non-crimp carbon fabric.
Fabric Areal Weight
[g/m2
]
Tow Count Fabric Style
A 136 12K Standard NCF
B 146 12K Standard NCF
C 305 24K Standard NCF
D 768 24K Quilted NCF

22
2.2.4 Droplet Deposition
Two devices were used to monitor the droplet deposition experiments: a goniometer and a
camera (LUMIX GH4, Matsushita Electric Co., Osaka, Japan). The sample was positioned on the
goniometer stage, with fibers aligned perpendicular to the goniometer light source and camera.
Samples were positioned with stitching facing downward. The second camera was positioned
directly above the sample (Figure 2.3). A syringe was used to deposit a single droplet of the
silicone oil facsimile and both cameras started recording as the droplet contacted the fiber. After
droplet deposition, the syringe was removed from the field of view.
Figure 2.3: Diagram showing the droplet deposition test set up.
The goniometer camera was used to record images at two-second intervals, while the topdown camera recorded images at ten-second intervals. The images were compiled and analyzed
separately. Software (MATLAB R2019a) was used to analyze the goniometer images, while the
top-down images were analyzed manually using image editing software. Droplet perimeter was
approximated using a brightness threshold, then further modified manually, as the contrast between
wet and dry was not perfect. In both cases, the dimensions of the fluid spread were measured from
images. Using the side view, droplet width and height were measured at each frame. Using the top

23
view, the droplet spread was measured in directions parallel and transverse to the fibers. Figure
2.4 shows diagrams illustrating the measurements recorded, as well as an example of one such
frame used for a single measurement. The goniometer camera detected fluid above the fiber
substrate only, while the top-down camera allowed for the observation of fluid imbibed by the
fiber bed.
Figure 2.4: Diagram showing the measurements made from side and top-views of the droplet deposition test (left)
and an example of the used for such measurements (right), with the droplet edges outlined for clarity in the top
view.
2.2.5 Droplet Grid Tests
Droplet arrays were deposited to demonstrate how the results from previous sections can
inform designs of resin patterns on semi-pregs. Droplet arrays of the facsimile fluid (30 Pa·s) were
deposited on a 20 × 20 mm area of fabric A. The arrays were evaluated with respect to two
parameters—(a) area covered by the fluid and (b) neighboring droplet interaction. Three distinct
droplet arrays were deposited. The first pattern consisted of droplets uniformly distributed in a 3
× 3 square grid. The second pattern was based on data obtained from the single droplet tests,
allowing the droplet positions to be arranged such that droplet overlap was minimized. The final

24
grid pattern used the same data with minor changes to the positioning to ensure droplet-to-droplet
interactions, showing how small deviations can result in differences in the final distribution.
Grid accuracy was maintained by producing a guide for the same syringe used for the single
droplet deposition tests. Droplets were aligned using the grid guides and dispensed one at a time
in raster fashion on the fiber bed. Droplet spreading was recorded using a top-down camera in the
same manner as for the single droplet tests. Since the positioning guide shielded the camera view,
data recording commenced 5 min after the first droplet was deposited. Images were captured at 10
sec intervals for up to 150 min. Using these images, the area covered by facsimile fluid was
recorded over time. Furthermore, a time lapse was generated using the captured images for each
droplet array.
2.3 Results
2.3.1 Contact Angle Comparisons
Using all three fluids—the resin and the two different viscosity silicone oils—differences in
surface flow phenomena were observed and recorded. Experiments were conducted to measure the
apparent contact angle as the droplet was deposited and absorbed into the fabric. As shown in
Figure 2.5a, the facsimile fluid did not match the apparent contact angle of the resin. Given the
non-static nature of the measured angle, it is more appropriate to refer to the metric not as the
contact angle, but as the edge slope of the droplet. However, as shown in Figure 2.5b, the
difference in edge slope between the lower and higher viscosity droplets was equivalent for both
the epoxy resin and the facsimile fluid.

25
(a) (b)
Figure 2.5: (a) Apparent contact angle of droplets of silicone oil and epoxy resin at 30 and 60 Pa·s viscosity. (b)
Difference between the apparent contact angle of the 30 and 60 Pa·s droplets for both resin and facsimile fluid.
2.3.2 Single Droplet Test
Experiments were performed for each substrate–facsimile fluid combination, while recording
droplet height (ℎ), droplet width above the surface (𝑤), fluid spread across fibers (𝑤𝑦) and fluid
spread along fibers (𝑤𝑥
). Note that w and h were recorded from the goniometer, while 𝑤𝑥 and 𝑤𝑦
were recorded from the top-down view. While both 𝑤 and 𝑤𝑥 represent droplet dimensions in the
direction parallel to the fibers, 𝑤 represents the width of the droplet seen above the surface, while
𝑤𝑥 includes fluid flow visible at the surface from the top-down view. The time to full sorption
(𝑡ℎ0
) was taken as the time required for the droplet height to reach a constant value. This time was
used to normalize the remainder of the time for the figure, as follows:
𝑡̂ =
𝑡
𝑡ℎ0
Figure 2.6 shows (a) droplet height and width versus time and (b) test results normalized by
time to full sorption. When plotted against the logarithm of normalized time, the height of the
droplet decreased approximately linearly. Similarly, the spread distance along the fibers also
increased linearly, which follows from Tanner’s Law (2.1) for a two-dimensional droplet, where
𝑅(𝑡) is the radius of the droplet, 𝛾 is the surface tension, 𝐵 is a constant, 𝜂 is fluid viscosity and 𝑉
is droplet volume [52,53]. In summary, the spread distance along the fibers, 𝑤𝑥 in this case, follows

26
a power law with time. In contrast, the spreading behavior across the fibers was distinctly nonlinear
(albeit noisy), exhibiting spread at the start followed by a quasi-stable plateau. Droplet spreading
generally followed patterns similar to those depicted in Figure 2.6, with variations in the rate of
growth and decay.
𝑅(𝑡) ≈ [
10𝛾
9𝐵𝜂 (
4𝑉
𝜋
)
3
]
1
10
∝ 𝑡
𝑛
(2.1)
(a) (b)
Figure 2.6: Results from depositing 30 Pa·s viscosity fluid on fabric C. The left y-axis shows results derived from
the side view, while the right y-axis shows results from top-down view. (a) Real time. (b) Time scale normalized to
𝑡ℎ0
.
2.3.3 Aggregate Tests Results
To visualize the evolution of droplet height with time, select images were assembled in an
array, as shown in Figure 2.7. For these images, the height of the droplet was mapped against
fractions of time to full sorption, 𝑡ℎ0. These images show that as fabric areal weight increased, inplane spreading of droplets generally occurred more rapidly, the only exception being the fabric
with the heaviest areal weight. In the case of the standard NCFs with the highest viscosity droplets,
most of the spreading occurred early; that is, the fluid spread out rapidly before it started being
absorbed into the fabric. The effects of gravity in assisting flow were strongest in early stages,
when droplet mass was most centralized.

27
Figure 2.7: Time lapse of side view for all combinations of fluid viscosity and fabric types. Red line corresponds to
the baseline from which height is measured, equal to the height of the droplet at 𝑡 = 𝑡ℎ0
(ℎ = 0).
Figure 2.8 was generated using the time to full sorption, 𝑡ℎ0, for each test. For the first three
fabric samples, lower areal weight correlated with increased time to full sorption for the lowviscosity oil. When fluid viscosity increased (from 30 to 60 Pa·s), 𝑡ℎ0 markedly increased (between
8- and 20-fold). In contrast, no similar correlation appeared for the heaviest fabric: fluid viscosity
did not affect time to full sorption for the 756 g/m2
fabric.
Figure 2.8: Relationship between time to full sorption and fabric type.
Figure 2.9 compares the maximum fluid flow distance in both directions of interest for each
test. As expected, the fluids spread longer distances along the fibers than across the fibers by a
factor of 2–3×. For the first three fabrics, spread distances along the fibers fell within a 1.9 mm
range and spread distances across the fibers were within a narrower 1.1 mm range. Spread

28
distances along the fibers did not fall in this range for fabric D, at least for the lower viscosity
fluid.
Figure 2.9: Relationship between fiber bed type and spread, in directions across and along the fibers.
2.3.4 Quilted Fiber Bed
Fabric D behaved differently from other fabrics due to its distinct architecture. This fabric
featured individual tows secured by a grid pattern of stitches, as opposed to simple unidirectional
stitches (shown previously in Figure 2.2). The gaps between individual tows afforded pathways
for fluid flow into the fiber bed, acting effectively as macrochannels between fiber bundles. Topdown images of fabric D revealed that most of the fluid flowed into the gaps (shown overlaid on
a topographical image of the fiber bed in Figure 2.10). The results in the previous section showed
that fabric D yielded the largest fluid flow distance along the fibers, accompanied by minimal flow
across the fibers. These observations support the assertions that irregularities in flow along the
fibers were caused by greater fluid penetration into the surface, and that penetration occurred
through and along the larger inter-tow gaps created by stitching. Comparing this result to a similar
test on one of the non-quilted fabrics, the bottom image in Figure 2.10 shows that fluid also flowed
into gaps/irregularities within the fiber bed, but that the flow was spread more uniformly along the
fibers.

29
Figure 2.10: Images showing the contours of the area infiltrated by fluid over time, overlaid onto a topo-graphical
representation of the fiber bed for the 30 Pa·s viscosity fluid tests on (above) fabric C and (below) fabric D.
The observations in Figure 2.10 highlight challenges when attempting to correlate the results
reported here to other fabric types. Fine-scale variations in the fiber bed, such as inter-tow gaps,
pinholes and other such irregularities common in woven fabrics, are intrinsic and affect fluid flow.
However, the impact of such variations on the flow of an individual droplet is likely to be more
pronounced. Due to the similarity in length scales of droplet size and surface features, droplet flow
is dominated by the positioning of the droplet on the surface. For example, in the case of the quilted
unidirectional fabric above, a droplet at the center of a tow bundle is likely to behave differently
from one deposited directly atop a tow gap, as shown in the top image of Figure 2.10. Since the
non-quilted UD NCFs, fabrics A–C, do not exhibit large-scale surface irregularities, the results
from these fabrics can be analyzed with fiber areal weight as the only variable.
2.3.5 Droplet Grid Optimization
Using the results from Figure 2.9, three droplet arrays were designed, including a control
pattern, with droplets arranged in a square grid, a staggered array to maximize coverage and

30
minimize droplet overlap and a tight-staggered array that ensured droplet interaction while
maximizing spread distance (grids 1–3, Figure 2.11). Coverage refers to the area fraction of the
surface that was covered by the facsimile resin upon full sorption and overlap refers to neighboring
droplets impinging on each other and coalescing into a single fluid pool.
Figure 2.11: Diagrams of droplet grid guides used. From left to right: 1. square grid, 2. staggered grid, 3. tightstaggered grid.
The dimensions of the grid arrays can be used to estimate the fiber loading and resin content
for a fully impregnated prepreg. Using the controlled volume of the droplet, the areal weight of
fabric A and a fiber density of ~1.7 × 106
g/m3
, eighteen droplets on a 20 × 20 mm area correspond
to a fiber volume fraction of 60–70%, assuming full impregnation and no voids. This range of fiber
loadings lies within the range used for commercial unidirectional prepregs. Since commercial
prepregs are produced by applying resin to both sides of a fiber bed, half the droplets needed to
achieve a proper volume fraction were used and nine droplets were placed within a 20 × 20 mm
area for the droplet grids.
Diagrams of the three grids are shown in Figure 2.11. Grid 1 featured droplets separated by
6.67 mm in a square array that extended across and along the fibers. The design for grid 2 was
informed by the observations that droplets spread 9.6 mm along the fibers and 4.1 mm across them.
The droplets were spaced 10 mm apart along the fibers and 3 mm apart across the fibers, with an

31
offset of 5 mm along the fibers between rows. Grid 3 featured a modification of grid 2, with
droplets spaced 9.5 mm apart along the fibers and 2.75 mm apart across the fibers.
Figure 2.12 shows the final images captured, as well as contours for the area covered by the
fluid for the first and last images. The images show that all three arrangements resulted in droplet
overlap. However, by design, grid 2 exhibited the least overlap, while grid 3 yielded the most. The
square grid, grid 1, showed overlap along the fibers and dry gaps between droplets (across the
fibers). Green and red outlines in Figure 2.12 show initial and final perimeters of the nine initial
droplets, respectively, as seen from the surface. At the end of each test, the red outlines show that
grid 1 resulted in three distinct fluid-covered regions, grid 2 with six and grid 3 with two.
Furthermore, upon completion of each test, the portion of the fluid area that extended beyond the
prescribed region was 5.3%, 0.6%, and 3.2% for grids 1, 2 and 3, respectively. These fluid regions
were determined only above the surface and do not include subsurface flow.
Figure 2.12: Images showing the extent of fluid coverage for all three grids, including 5 min and 150 min after
deposition.
The area covered by the droplets was recorded and plotted versus time (Figure 2.13). This
graph shows that grids 2 and 3 covered greater areas than grid 1, but the difference was small (3–
4%). Upon completion of the tests, grids 1–3 covered 192, 198 and 200mm2
, respectively.

32
Figure 2.13: Graph showing the area covered by the droplets using all three grids.
2.4 Discussion
2.4.1 Absorption Kinetics
The time to full sorption, 𝑡ℎ0, was inversely proportional to fiber bed areal weight. As shown
in Figure 2.8, for 30 Pa·s droplets, fabrics A, B and C exhibited 𝑡ℎ0 values of 33.8, 25.3, and 10.5
min, respectively. Furthermore, only fabric A exhibited full through-thickness penetration of the
fluid, as evidenced by the presence of facsimile fluid residue on the underside of the fabric. From
these results, we assert that heavier fabrics allowed for deeper through-thickness flow of the
droplet, reducing the time-to-saturation for droplets of the same size.
In contrast, increasing fluid viscosity generally increased sorption time, although the
relationship between viscosity and 𝑡ℎ0 was not proportional (based on the two silicone oils tested).
Figure 2.8 shows that, except for fabric D, increasing fluid viscosity by a factor of two led to an
8- to 20-fold increase in sorption time. In fabric D, high-viscosity resin readily penetrated the
macrolevel features (features that were absent in fabrics A through C).
The mechanism by which droplet height changed depended on fluid viscosity. For example,
the height change of the 60 Pa·s droplets was driven primarily by spreading on the fabric surface,
while the height change of 30 Pa·s droplets was dominated by more rapid penetration into the

33
fabric, shown previously in Figure 2.7. Higher viscosity fluid droplets spread longer distances
along the fiber bed surface, in some cases extending beyond the field of view. In contrast, droplets
of the lower viscosity fluid remained within the field of view until full sorption was reached. The
60 Pa·s droplets showed reduced absorption and spread to a large area, resulting in extended
absorption times.
The most important metric for production of semi-pregs via droplet deposition is the time to
full sorption, referring to the time until the droplet is fully within the fiber bed. As such, by ensuring
that a deposited droplet can achieve maximum spread within the time to full sorption, it should be
possible to produce prepreg with a degree of impregnation high enough to achieve bulk factors
similar to standard OoA prepregs (~10%). As shown in Figure 2.6, droplet flow continued after
𝑡ℎ0, allowing the droplet to flow further within the fiber bed and ensuring full saturation of the
fiber bed during cure. However, in practice, a balance must be achieved between the DoI, bulk
factor and droplet separation. In particular, the spreading time must be close enough to 𝑡ℎ0 to
achieve both a high DoI and low bulk factor, while still maintaining droplet separation and
allowing for full ply saturation during cure. Furthermore, given the differences in 𝑡ℎ0 achieved by
the minor increase in viscosity used in this study, droplet spreading can be virtually arrested by
simply cooling the prepreg.
2.4.2 Spread kinetics
Fabrics A–C were all uniform, unidirectional fabrics with similar smooth surfaces, as shown
in Figure 2.2, and as such similar droplet spread behavior was expected for these fabrics. Figure
2.9 shows that droplet spread distances fell within a narrow range, regardless of fiber areal weight.
Furthermore, there was no correlation between spread distance and areal weight. Droplet spread
distance was governed solely by surface topography, while areal weight had a negligible effect.

34
The droplet spread was characterized by two stages as well as two different directions. The
two spreading stages were (a) rapid spreading while the droplet remained atop the fabric surface,
followed by (b) slower spreading once the droplet was fully imbibed. In the test shown in Figure
2.6, 80% of spreading along the fibers occurred before full sorption, and 90% of spreading across
the fibers occurred before full sorption. Spreading above the surface was driven primarily by
gravity and surface tension, while spreading within the fiber bed was driven by capillary effects.
The spreading was also different along and across the fiber direction and droplets spread much
longer distances in the fiber direction than across fibers. Furthermore, most of the spreading
occurred before sorption in the through-thickness direction. The early-stage spreading was aided
by capillarity between aligned fibers, facilitating fluid flow along fibers and impairing flow across
the fibers after full absorption.
For fabrics A–C, the capillary radius was expected to remain the same, and when using the
same viscosity fluid, the remaining parameters were similarly unchanged. Therefore, the distance
traveled within the capillary, and as such the final distance spread, remained the same, as expected.
The horizontal capillary flow of a droplet is described by Washburn’s equation (2.2), where 𝐿 is
the distance traveled within the capillary, 𝛾 is the surface tension, 𝑟 is the capillary pore radius, 𝑡
is time, 𝜃 refers to the contact angle and 𝜂 is the viscosity [54].
𝐿 = √
𝛾𝑟𝑡 cos(𝜃)
2𝜂
(2.2)
An increase in viscosity was shown in Figure 2.5 to result in an increase in contact angle,
meaning the change in silicone oil resulted in two terms changing within Washburn’s equation.
Furthermore, referring to Equation (2.2), it can be seen that the increase in viscosity and increase
in angle effectively counteracted one another, resulting in only minor deviations in spread distance.

35
2.4.3 Effects of Macrofeatures
Macroscopic features of fabrics, particularly inter-tow gaps, strongly influenced droplet
spread. For example, Figure 2.10 showed that fluid flowed quickly into inter-tow regions of fabric
D. These regions served as channels for macroflow, as opposed to micro-flow within tows. Similar
results can be expected for woven fabrics, where the placement of the droplet, particularly with
regard to the proximity to pinholes in the weave, affects droplet spread more strongly than fabric
areal weight or intra-tow capillary sizes.
With regard to developing a method for predicting droplet spread, an analytical solution or a
numerical simulation would be useful for unidirectional fabrics. However, for woven fabrics,
which feature complex surface topography, the parameters of the fabric may have only minor
effects on surface spread compared to the position of the droplet on the fabric terrain. Generating
an analytical solution or simulation for such fabrics would require an added level of complexity
and rely on the specific surface features of the given weave.
2.4.4 Droplet Arrays
Droplet positioning and array designs can be guided by the results presented here, as shown
by the observations of spread behavior in the three different droplet grids. The findings support the
hypothesis that fluid flow occurs more rapidly along the fibers than across and provide
approximate indications of how a droplet will spread along the surface. The spread distances can
be used to arrange droplets in arrays that prevent/minimize fluid overlap and that ensure maximum,
controlled flow distances. However, these same results can also be used to ensure full surface
coverage, maximizing overlap and flow distances, as shown in Figure 2.12 and Figure 2.13. For
example, grid 3 exhibits overlapping fluid covered regions while maintaining similar coverage as

36
the other arrays. Due to the strong degree of anisotropy of unidirectional fiber beds, droplet
positioning must be staggered to ensure maximum surface coverage.
The droplet grid tests also revealed effects of droplet impingement. Neighboring droplets
influenced the direction of spread and in some cases altered the spreading behavior that would be
expected for a single droplet. As shown in Figure 2.12, once contact between droplets occurred,
droplets quickly spread to fill regions between neighboring droplets. Flow occurred even across
the fiber bed when driven by the surface tension of the fluid. Transverse flow was responsible for
the increase in surface area covered by grids 1 and 2, which included droplet interaction across the
fiber bed.
Using these insights into single-droplet behavior, droplets can be positioned in arrays that
minimize droplet interaction and maximize through-thickness air evacuation pathways.
Alternatively, droplets can be positioned to intentionally create a degree of interaction, achieving
a higher surface coverage. These capabilities can be leveraged for use in semi-preg design, which
requires discontinuous resin distribution, and more careful control of the resin distribution is
achievable with this method than with currently implemented methods. Furthermore, should other
novel composite designs emerge that rely on discreet resin droplets, this study may serve as an
initial guide.
2.4.5 Facsimile Fluid
Given the impact of droplet–substrate surface parameters on the driving forces behind droplet
spreading, matching viscosity is not sufficient to ensure accurate simulation of resin droplets.
Experiments showed that the apparent contact angles, or the evolving edge slopes, did not coincide
between the facsimile fluid and the resin, and as such the observed results of spreading distance
and absorption times for the facsimile fluid cannot be assigned one-to-one to the resin. However,

37
as Figure 2.5b shows, the relationship between the 30Pa·s droplet and the 60Pa·s droplet was
maintained both for resin and the silicone oil facsimile. Based on this result, we assert that the
relative effects of increasing the viscosity are equivalent when using actual resin droplets.
2.5 Conclusions
We have demonstrated the effects of unidirectional fiber beds on droplet flow on the surface.
The fabric feature that most strongly influenced the surface flow was the fiber bed architecture.
Macrolevel features of the quilted UD fabric (fabric D) dominated the surface flow of droplets.
Decreasing fiber areal weight and increasing viscosity both led to an increase in time to full
absorption. However, neither of these factors appreciably influenced the droplet surface coverage.
The findings presented entail multiple implications. First, predicting droplet surface flow is
most reliable for unidirectional fiber beds and the anisotropy of UD can be exploited to design
patterns that prevent or minimize droplet impingement. By heating or cooling the fluid, viscosity
values can be adjusted to allow longer working times for droplet deposition. Within the narrow
viscosity range used in this study, the actual droplet spread distances would see little variation.
These findings are potentially useful, yet challenges remain. For droplet deposition onto woven
fabrics, the specific interactions between droplets and macrolevel details of the fabric must be
tracked. In addition, an actual resin must eventually be used, as opposed to a facsimile fluid.
Finally, single droplet size was monitored throughout the study. However, the volume of resin
needed to achieve a proper volume fraction for prepreg differs with the fabric weight and smaller
droplets are likely more reliable for thinner fabrics. Attending to these challenges will inevitably
fall to those designing a droplet deposition system.
The findings highlight the groundwork required to develop practical methods to produce semipregs and indicate a pathway to achieving more efficient and robust OoA production. The spread

38
of resin after deposition can be manipulated by determining the time allowed for flow into the fiber
bed, as well as droplet position in relation to other droplets and to the fiber bed, allowing for
intelligent design of resin distribution in semi-preg. Such designs will facilitate the scaling-up for
manufacture of semi-pregs via droplet deposition, as opposed to current methods that rely on
prepreg manufacture one ply at a time. Semi-preg is a robust intermediate material for OoA
processing and can potentially restore robustness to levels comparable to conventional autoclave
manufacturing. In addition, OoA processes reduce costs and increase part throughput, allowing for
increased part complexity and accessibility.

39
Chapter 3. Thickness Variation in Contoured Composite Parts by Vacuum
Infusion2
3.1 Introduction
Thickness variations commonly arise in composite parts produced by VI, yet the evolution of
such defects and the effects of basic process parameters are not fully understood. The present study
addresses this gap in understanding and demonstrates a practical workflow that leverages process
monitoring, material characterization, and process simulations. Laminate thickness variations are
monitored along the length of V-shaped parts produced via VI, then compared with results of
process simulations conducted using commercial software (PAM-RTM). The effects of variations
to geometry and material properties were explored in the context of process simulations in an effort
to accurately account for fiber bridging at corners. The investigation demonstrates how process
monitoring can be used effectively in coordination with process simulation to understand and
control a common defect in VI parts.
VI is a cost-effective method to produce large composite parts such as wind blades and boat
hulls, although there is also potential to use the method to produce aerospace parts comprised of
carbon fiber reinforced polymer (CFRP) composites [55,56]. VI relies on a relatively small
pressure difference to draw resin into a fiber bed that lies between a rigid tool on one side and a
flexible vacuum bag on the other. The pressure difference between the resin reservoir at the inlet
and the vacuum at the outlet generally induces a gradient in thickness and fiber volume fraction
along the part length [57–59].
2 This study was published in Advanced Manufacturing: Polymer and Composites Science in October 2023 with
Steve Nutt and Bo Jin as co-authors [129].

40
The presence of corner contours in parts generally causes variations in local thickness in most
CFRP production processes, including VI [60–62]. Depending on the curvature of the tool,
thickness can increase or decrease at the corner due to reduction or augmentation of the compaction
force over the arc. As permeability depends on fiber volume fraction [63,64], local thickness
changes result in variations in flow behavior during infusion, as well as dimensional differences
once a part is cured [65].
To predict thickness variations before producing a part, process simulation tools for resin
infusion processes can be employed [66–69]. This practice has led to development of software
dedicated to finite element simulations of flow processes in porous media. Such software can be
used to determine appropriate inlet and outlet positions and timing sequences to ensure full
saturation, minimize fill times [70], and determine if a reinforcement-resin coupling is compatible
with infusion [71]. The process simulation software used in this study (PAM-RTM, ESI, Bagneux,
France) is widely used in industry, and has been used to accurately predict the thickness gradients,
mostly in flat laminates [58].
In this study, we demonstrate a method for process simulation to predict thickness deviations
in contoured parts produced by VI. In conjunction with this method, we deploy process monitoring
to monitor thickness deviations during infusion. These coordinated activities can be used to prevent
defects, particularly those associated with thickness variation. Application of process simulations
coordinated with process monitoring can be used to maintain geometric specifications, increase
geometric tolerances, and reduce part scrap. Flow simulations are widely used in industry for
various purposes, but rarely coordinated with process monitoring. The method demonstrated in
this work highlights the importance of fabric-and-tool compatibility and the deviations in laminate
geometry that typically arise from fabric bridging and inter-ply slippage.

41
3.2 Experimental Methods
Thickness variations were measured in VI parts produced on tools with concave and convex
corners. A laser displacement system was used to measure laminate thickness during infusion.
Following cure, the laminates were sectioned to measure the thickness of the centerline crosssection.
Software (PAM-RTM) was used to simulate infusion for a wide range of corner geometries,
including those matching the experimental parts. Analysis revealed that the standard procedure did
not adequately account for all corner effects and required modification. The compression response
and surface profile of the fabric were measured on specific tool geometries and used to generate a
second set of User Defined Geometry (UDG) simulations. Using these UDG’s, simulations
predicted thickness deviations at the corner much more accurately, both in terms of magnitude and
the affected area.
3.2.1 Fabric Characterization
Triaxial carbon fabric with a high areal weight (A&P Technologies, Cincinnati, USA) was
selected based on compatibility with VI and the selected resin [71]. General properties are listed
in Table 3.1. The areal weight, braid angle, and thickness were provided by the manufacturer and
independently verified. The volumetric density was measured using a gas pycnometer (Accupyc
1330, Micromeritics, Norcross, USA).
Table 3.1: Properties of the triaxial fabric
Property Value Units
Areal Weight 536 g/m2
Tow Angles 0/+60/-60 °
Thickness @ 55% Vf 0.5334 mm
Density 1.6742 g/m2
A single ply of the fabric was compressed at 1 mm per second using a rheometer (AR2000ex,
Texas Instruments, Dallas, USA), chosen for its high accuracy in position and force data to obtain

42
stress versus strain compression data for the fabric, for use in the initial simulations. The results of
this test is shown in Figure 3.1. Given the multiple compressions done on the fabric during vacuum
leak testing, the second compression was selected as the stress versus strain response for the fabric.
Figure 3.1: Compression stress versus strain response for a single ply of fabric.
Permeability values of the fabric were measured using a radial infusion setup, following the
same procedure of a previous study [71]. A [0]2 stack of 203.2 mm square plies was sandwiched
between aluminum and acrylic tool plates. Plastic shims positioned around the plies were used as
spacers to control thickness, with the entire assembly being sealed in a vacuum bag. The test was
repeated at least four times for each thickness spacing.
𝑉𝑓 =
𝑁 ∙ 𝜌𝐴
ℎ ∙ 𝜌𝑉
,
(3.1)
Permeability tests were performed using different spacer thickness values. Employing the
number of plies 𝑁, the areal weight 𝜌𝐴, the volumetric density 𝜌𝑉, and the thickness ℎ, the volume
fraction was obtained for each test via Equation (3.1). Following the procedure of Chan and Huang
[72] the x- and y-directional (0° and 90°) radii of the flow front captured in each image were used
to obtain in-plane permeability values for the fabric at each volume fraction. An exponential decay
curve was fitted to the data, as shown in Figure 3.2, as this relation is required for the process

43
simulation. The error bars were significant, as expected, due to the high variability and noise
intrinsic to fabric and fabric permeability measurements [73,74].
Figure 3.2: Permeability vs. Fiber Volume Fraction in x- and y-directions for the triaxial fabric.
3.2.2 Laminate Production
Laminates were produced using 305 × 203 mm plies, following an [0]8 layup sequence,
selected due to the quasi-isotropic nature of the fabric. Standard VI layup procedure was followed,
using conventional consumables. A 25.4 mm gap was cut into the centerline of the flow media to
permit direct observation of the laminate surface. Five distinct tool configurations were used
(Table 3.2). For contoured tool configurations (B-E), the fabric was first stacked, then placed at
the corner and allowed to drape onto the tool. The fabric was infused with epoxy resin and hardener
(FibreGlast #4500 and #4570, Brookville, USA). Resin was allowed to infuse until steady flow
was reached in the outlet line, at which point the inlet and outlet lines were sealed. The parts were
subsequently cured at room temperature according to supplier guidelines.
Table 3.2: Tool configurations used for experimental trials.
Tool ID Tool Curvature Corner Angle
[°]
Corner Radius
[mm]
A Flat N/A N/A

44
B Concave 90 10
C Concave 60 10
D Convex 90 15
3.2.3 In-situ Observation
A laser profile scanner (Micro-Epsilon, scanCONTROL 2600-100) was used to measure the
displacement of the top surface of the laminate during infusion (Figure 3.3). The frame ensured
that the centerline of the laminate was directly under the sensor, to measure the thickness
perpendicular to the ideal flow front. Given the sensor range, only the region at the corner was
measured, approximately 80 mm in either direction. The resin reservoir was attached to the side
of the frame, to ensure no significant height difference between reservoir and inlet point. After
curing, the part was demolded without moving the tool, and the laser profile scanner measured the
corresponding position of the tool surface.
After demolding, laminates were sectioned along the centerline, polished, and imaged using
a digital light microscope (Keyence VHX-5000, Osaka, Japan). Pixel measurements from the
micrographs were used to measure the thickness along the centerline.
Figure 3.3: Experimental configuration for laminate infusion. (left) Schematic of the laser displacement system.
(right) Experimental setup.

45
3.2.4 Initial FEA Simulation
The simulation software relies primarily on Darcy’s Law [75], given in Equation (3.2), where
𝑣 is flow velocity, [𝐾] is the permeability tensor, 𝜇 is dynamic viscosity, and ∇𝑃 is the pressure
gradient. By including mechanical coupling, the permeability tensor is recalculated and updated
between calculation steps, thus accounting for the changes in fiber volume fraction [76]. While the
software can also simulate heating and curing conditions, the present study concerns only the
filling simulation.
𝑣 =
[𝐾]
𝜇
∇P, (3.2)
The initial stage of this investigation encompassed a fluid-mechanical coupled finite element
analysis (FEA, PAM-RTM). The fabric properties, listed in Table 3.1, served as input parameters
for initial simulations. A constant resin viscosity of 0.3 Pa·s, measured via rheology, was applied.
Inputs into the software included tool geometry, the number of plies, and the uncompressed
thickness of the fabric, all of which were used in the development of a Laminate Mesh (LM)
geometry. The laminate geometry for the LM approach involves setting the element size such that
each ply is one element thick. Exploiting the flexibility inherent in the simulations, a broad range
of tool geometries were selected, encompassing three diverse sets, detailed in Table 3.3.
The outcomes from these simulations were used to extract thickness values through the node
positions at the centerline of the top and tool surface.
All simulations maintained uniform processing conditions. Application of vacuum via the
outlet edge was constant. Simultaneously, a uniform pressure was exerted on the top surface,
ramping up to atmospheric pressure in a span of 90 s. Thereafter, this applied pressure was
maintained. After 120 s, the inlet was opened at atmospheric pressure. Subsequently, the nodes on

46
the tool surface were rigidly fixed, whereas the remaining nodes were constrained in-plane. The
simulation culminated with complete saturation of the laminate.
Table 3.3: Tool Geometries used in FEA.
Curvature Corner Angle
[°]
Corner Radius
[mm]
Set 1: Varying Corner Angle
Flat N/A N/A
Concave 15/30/45/60/75/90 10
Convex 15/30/45/60/75/90 10
Set 2: Varying Corner Radius
Flat N/A N/A
Concave 90 10/15/20/25
Convex 90 10/15/20/25
Set 3: Comparison to Experimental Trials
Flat N/A N/A
Concave 90 10
Concave 60 10
Convex 90 15
Convex 60 15
3.2.5 Revised Compaction and Geometry
Results from the first round of simulations revealed the need for a refined methodology for
predicting the final cured geometry. The simple assignment of tool geometry and generation of a
laminate mesh was insufficient. There was deviation from the expected compaction behavior when
the fabric was laid upon the tool: fabric bridging for the concave tools and fabric buckling for the
convex tools. This realization prompted a series of tests to determine more accurately the
compaction response and uncompacted geometry of the fabric on each tool.
Fabric was laid up on the tool and subsequently vacuum-sealed within a bag. A pair of pressure
sensors (XE-0050-008, Composite Integration, Saltash, United Kingdom) were connected to the
inlet and outlet lines. By controlling the vacuum level inside the bag, the top surface of the preform
was measured using the laser displacement sensor. The ply stack was placed under vacuum for a
minimum of one hour. Then, the pressure level inside the bag was increased by ~3400 Pa (1 inHg)
in fixed intervals, allowing the fabric to stabilize before the next increment. When the pressure

47
within the bag equaled atmospheric pressure, the process was reversed, increasing the vacuum
level in fixed intervals and allowing the fabric to rest between intervals until maximum vacuum
was reestablished.
Positional data from the laser displacement sensor was transformed to thickness data.
Combined with pressure, a thickness versus pressure response curve was generated for each tool
geometry for the intervals of steady pressure. Because the simulation software required uniform
material properties, each geometry was segmented into Flange, Corner, and several Edge regions.
From the pressure versus thickness data, a stress versus strain response was obtained for each
region, with the multiple Eedge regions being averaged into one. The data was subsequently fit to
derive a stress-strain curve for the Corner, Edge, and Flange of each tool geometry. Based on
previous experience with the software, exponential curves were used in the fitting.
The tool surface profile and the preform surface at full decompaction were measured and used
to construct a set of User Defined Geometries (UDG). Curve fitting was performed to ensure
preservation of symmetrical and smooth surfaces for simulations. The full curve was subsequently
divided into appropriate regions for each tool geometry, with each assigned the Corner, Edge, or
Flange elastic properties and uncompressed volume fraction. Instead of generating the laminate
mesh from individual ply properties, the UDG was used to generate a 3D solid, and then split into
elements such that each ply was one element thick. The second set of simulations used the same
processing conditions as the LM simulations, as described in the previous section.
3.3 Results and Discussion
3.3.1 In-situ Observation
The laser displacement system was used to monitor the top-surface position of the laminate
during infusion, transforming the positional data into thickness data. Figure 3.4 shows a typical

48
data set, for Tool B. In Figure 3.4a, the primary difference arises for the ‘After Vacuum Hold’
curve, which reveals preform decompression during/after resin infusion. Inspecting thickness,
Figure 3.4b more clearly shows the difference between the various points during infusion,
revealing marked reduction in thickness, both at the corner and along flanks, after vacuum hold.
Application of vacuum caused a decrease in thickness across the entire measurement domain, with
the average thickness dropping 1.33 mm (±0.17) in the flange region and 0.84 mm (±0.05) in the
corner region. Upon completion of infusion, the average thickness reverted nearly to the value
prior to vacuum pull, increasing by 1.42 mm (±0.28) and 0.88 mm (±0.08) in the flange and corner
regions, respectively. As the resin cured, the thickness changed due to cure shrinkage, dropping
0.40 mm (±0.19) in the flange and increasing by 0.07 mm (±0.07) in the corner.
Figure 3.4: Position and thickness data during infusion with Tool B. (a) Laser sensor acquired position data. (b)
Thickness data, derived from position data.
The anticipated difference in thickness between corner and flange regions was noted even
before initiating infusion. At all stages – prior, during, and post-infusion – the thickness at the
corner exceeded the flange thickness by approximately 5 mm. Once infusion concluded, the corner
thickness returned to the value prior to vacuum pull and remained constant post-cure without
significant cure shrinkage. This finding indicates that infusion introduces less variability in the
corner region thickness compared to the flange region.
Flat laminates produced with Tool A showed modest bag inflation, but no major pleating was
necessary to achieve bag integrity. Parts produced on Tool A geometry exhibited the most

49
compaction during vacuum pull, and least thickness increase during filling. Given the small
measuring area and the surface structure of the fabric, no significant thickness gradient was
observed during infusion, aside from the expected decompression with the infusion of resin.
3.3.2 Analysis of Polished Sections
Cross-sections of the laminates at centerline are shown in Figure 3.5, with the tool-side facing
down. Negligible porosity was observed in all parts, except for parts produced with Tool C, which
exhibited voids in corner regions. The concave laminates produced with Tools B and C exhibited
the morphology expected of concave corner parts – that is, fiber bridging, and increased thickness
at the corner. Note that the fabric in Tool B did not completely fill the laminate and showed resinrich regions on both the interior and exterior curve at the corner. Convex laminates revealed no
significant defects, and laminates produced with Tool E maintained constant thickness throughout.
Tow waviness was observed at the corner region in laminates produced with Tool D. Based on dry
fabric compaction observations, this phenomenon can be attributed to buckling and bunching of
the fabric at the corner during layup.
Figure 3.5: Micrograph of the cross-sections for the corner region for all five tool geometries. For Tool A, the center
of the laminate is displayed. The tool side is facing downwards for all samples.

50
3.3.3 Simulation
Preliminary FEA simulation outcomes from the Laminate Mesh approach are presented in
Figure 3.6. Two trends emerged from inspection, both of which were expected. First, sample
thickness decreased from the inlet to the outlet side for all tool geometries (Figure 3.5, Tools A,
B). The thickness gradient arises from the pressure difference between inlet and outlet. Second,
parts with concave corners exhibited corner thickening, whereas those with convex corners
exhibited corner thinning. As shown in Figure 3.6a, increasing the corner angle resulted in a larger
maximum thickness deviation and a wider corner region. The corner thickness deviation increased
from 0.8% to 1.5% for concave parts and 0.5% to 0.8% for convex, while the corner width
increases from 12.4 to 28.4mm for concave parts and 11.7 to 25.6mm for convex, respectively
Conversely, Figure 3.6b shows that an increase in the corner radius decreased the thickness
deviation at the corner, while expanding the corner region. Concave parts had the thickness
deviation decrease from 1.5% to 0.5%, while convex parts decreased from 0.8% to 0.3%; corner
width increased with corner radius, from 25.6 to 46.3 mm for both curvatures. These outcomes are
consistent with expectations, with larger radius and corner angle both result in a larger tool arc,
and a less sharp corner reduces pressure effects.

51
Figure 3.6: Thickness across the centerline of the laminates from FEA simulation using the laminate mesh. (a) Tool
geometry set 1, varying corner angle; (b) tool geometry set 2, varying corner radius.
The thickness deviation at the corner was minor (0.3–1.5%) compared to the thickness
deviation observed across the laminate length (16.8%). Even the sharpest corner geometry tested,
a 10 mm 90° concave corner, exhibited a maximum deviation at the corner that was a small fraction
(40%) of the thickness difference between inlet and outlet. Thickness measurements from
laminates produced with Tools B and C (shown previously in Figure 3.4) showed a more
pronounced thickness increase than predicted in simulations. In the following section, simulations
of these conditions are compared to experimental results.
3.3.4 Thickness Measurements
Cross-sectional measurements of laminate thickness revealed distinctive characteristics, and
these findings were compared with the FEA results shown in Figure 3.7. Primarily, variations in
thickness observed in the non-corner regions of the experimental parts were detected (both sides
of central peak), and these were attributed to differences in fabric alignment and intrinsic
reproducibility challenges associated with VI. Additionally, edge tapering was observed due to

52
misaligned ply edges, arising from the contour radius changing as plies were laid down and shifting
of the fabric at edges – an effect that was disregarded for this study. Hence, measurements 35 mm
from either edge, defined as the ‘inlet’ and ‘outlet’ locations for subsequent analysis, are indicated
in the graphs by the dashed circles. Minor waviness in the thickness curve was observed in all
laminates, attributed to the fabric weave.
Figure 3.7: Thickness measurements of laminates centerline for all experimental trials, compared with FEA. The
‘inlet’ and ‘outlet’ points are indicated with enlarged markers. A zoomed-in look at the corner region of the FEA
results is also shown.
Simulation results predicted a thickness difference of 0.71 mm between points 35 mm from
the outlet and the inlet. The part produced using Tool A exhibited the most pronounced gradient
and demonstrated a thickness difference of 0.48 mm, about 1/3 less than predicted. An inlet-outlet
thickness gradient was not detected in experimental trials due to a combination of effects from
thickness, waviness, and edges.
The trends in simulation results for corner regions generally aligned with cross-sectional
observations. Concave laminates exhibited a thickness increase at the corner. In contrast, convex

53
laminates exhibited only small variations in thickness. Tool D showed a minor increase in
thickness, while Tool E presented negligible deviation in thickness at the corner. The sample
showing maximum deviation, produced with Tool B, showed a corner-outlet difference of 5.1 mm,
while the simulation predicted a much smaller thickness difference (0.27 mm, 18× less). The
findings revealed a major limitation of simulations performed using the laminate mesh method.
These simulations did not accurately predict the final thickness of contoured laminates produced
via vacuum infusion.
3.3.5 Modified Compaction Response
The thickness of the corner region of the preform was measured during compaction and recompaction. The case for Tool B is shown in Figure 3.8. All regions followed a similar trend, with
the rate of thickness increasing in a semi-exponential fashion during decompaction. During recompaction, the thickness decreased more gradually. In general, the Flange region exhibited the
lowest initial thickness and smallest thickness change, while the Corner region exhibited the
greatest initial thickness and largest thickness change. The thickness of Edge regions fell between
those of the Corner and Flange, and regions nearest the corner exhibited the largest thickness
deviation. All these results fall in line with expectations, with the fabric bridging, bending and
bunching at the corner increasing the thickness while allowing for more space for compression,
whereas the Flange region exhibits behavior much like the flat laminate, with the Edge regions
forming a smooth transition between the two responses.

54
Figure 3.8: Process for obtaining the modified compaction results for Tool B. (a) Region positions labeled on Tool
B; (b) average thickness of the different regions and pressure versus time; (c) thickness of the different regions
versus pressure.
The compiled stress-strain curves for all tool geometries are illustrated in Figure 3.9. As
expected, the Corner and Edge regions exhibited stiffer responses than the Flange region for all
tools, and the Single Ply test and Tool A were more similar to the Flange region than the others,
while still being less stiff. The contoured tools all exhibited similar responses in the Edge region,
and likewise in the Corner region, except for Tool C, which showed a much stiffer response. The
difference in Tool C was attributed to bridging of fibers, imparting increased stiffness. One might
expect Tool B, which also exhibited bridging, to show a similar response; however, as shown in
Figure 3.5, the fabric did not fully conform to the corner, resulting in greater relaxation in corner
compression during the pre-infusion stage. The stress-strain curves were then used as material
properties for subsequent modified simulations.

55
Figure 3.9: Modified compaction stress versus strain responses for the (a) corner, (b) edge, and (c) flange regions of
each tool geometry. Also included is the single-ply stress versus strain response used in LM simulations.
3.3.6 Modified Simulations
To impart greater accuracy to simulations, modified compaction response tests were deployed
to develop a user defined geometry (UDG) for each tool. The tool-specific UDGs were used in
modified simulations to generate predictions and compare with those obtained with conventional
LM FEA. This UDG surface corresponded to the top surface profile of the preform, measured
when the bag was at atmospheric pressure post-decompaction (denoted by an asterisk in Figure
3.8b). Curve-fitting and symmetry was assigned to the measured profiles to develop the UDG
surface profiles, for compatibility with the software.
Figure 3.10 shows the measured surface profiles, together with the fitted curves employed for
the UDG FEA simulations and the geometry developed using the LM method. The measured
contours differ significantly from those generated by the LM method. For concave tools (Figure
3.10b, c), the LM simulation did not account for fabric bridging at the corner, behavior that was
exhibited in the measured geometry and used in the UDG surface. While less pronounced, similar
differences were observed in the convex tools (Figure 3.10d, e): the measured geometry showed
a modest increase in thickness at the corner for Tool D, while Tool E showed a major deviation

56
from the expected radial arc. In both cases, the deviations were attributed to fabric buckling and
draping issues at the corner observed during layup. When laying down fabric on the convex tools,
the innermost plies (closest to the tool), experienced compression, which is then transferred
through the thickness. Tool E experienced the most compression, leading to minor fabric buckling,
visible from the profile.
Figure 3.10: Measured, UDG and LM uncompressed surface profiles for the corner region of all tool geometries.
Thickness values generated by the UDG FEA simulations were obtained using the nodes at
the centerline of the tool surface and the top surface, following the procedure used for the initial
set of simulations. These results, plotted alongside the experimental measurements, are presented
in Figure 3.11. A thickness gradient from inlet to outlet was again observed. However, unlike the
LM simulations, the UDG simulations consistently demonstrated the expected corner thickening,
which was attributed to the geometry of the uncompressed laminate, already displaying thickening
due to the fiber effects during preform layup.

57
Figure 3.11: Thickness of UDG simulation vs. experimental results. (a) Plot showing the entire part length, with
‘inlet’ and ‘outlet’ points labeled, and (b) enlargement of the corner region.
The thickness curve for Tool E did not conform to a smooth arc, unlike samples produced with
other tools. This anomaly was attributed to deviation from a circular arc that was measured in the
uncompressed fabric, illustrated in Figure 3.10e. Although this deviation was not observed in the
laminate produced with Tool E, the thickness variation of Tool D showed a similar departure from
the smooth arc.
Values of thickness deviations for relevant positions of the laminates are presented in Figure
3.12, shown as the deviation from the thickness at the outlet. On average, LM simulations predicted
a thickness deviation that was 0.28× experimental values at the corner, and 1.63× at the inlet,
while the UDG simulations predicted a thickness deviation that on average was much closer to
measured values (0.91×) at both corner and inlet. In general, the LM simulations overpredicted
the general thickness deviation at the flange regions, and significantly under-predicted the corner
thickness deviation. The UDG simulations predicted a thickness deviation that was much closer to
measured values throughout the laminate length, within 30% in non-corner regions, and 20%
elsewhere.

58
Figure 3.12: Thickness deviation between (a) the outlet and the inlet and (b) the outlet and the corner.
Comparing the experimental measurements to the UDG FEA simulations, greater accuracy
was generally achieved in the region of laminate experiencing thickness deviation due to the
corner, referred to as the corner width. For concave tools, UDG FEA simulations predicted a corner
width 7% less (±4) than the measured value, while LM simulations were less accurate, predicting
a width 43% less (±3) than measured. The width of the corner area of convex laminates was
characterized by non-uniformity of the thickness distribution, and results of simulations contained
apparent discrepancies. Notably, simulations consistently underpredicted the thickness of flange
regions and of the flat laminate, Tool A. At the corner, the simulation predicted lower thickness for
concave parts, and higher thickness for convex parts. These discrepancies arose from the difference
in compaction effect once the fabric was saturated with resin. Resin saturation decreased the degree
of compression of the laminate relative to dry compaction. The decrease was attributed to sharing
of the compaction force between fabric and resin, affecting both overall thickness and thickness in
concave regions. In addition, the presence of resin is expected to lubricate fibers, leading to
nesting, allowing fabrics to conform to convex tools and reducing thickness, an effect not
accounted for in FEA simulations.
In summary, a marked thickness deviation arose from corners in laminates produced via
vacuum infusion, clearly outweighing thickness variations caused by the inherent pressure

59
gradient. Material bridging at corners resulted in significant thickening and largely dictated the
geometry of the contoured laminates. Minor thickness deviations were observed between the initial
vacuum pull post-infusion and the fully cured stage. Despite a temporary thickness gradient
emerging upon closure of the inlet and outlet, subsequent resin redistribution within the part
yielded only a slight gradient across the contoured components. To account for the dominance of
fiber bridging on thickness deviation, a solution was developed by measuring the post-layup
geometry and using it as the starting geometry for the UDG FEA simulations. These modified
simulations more closely matched the measured thickness values.
3.4 Conclusions
Process simulations were conducted to predict laminate thickness in parts with controlled
corner geometry. To achieve accurate predictions, the method of assigning material properties was
modified to account for fiber bridging at corners, resulting in deviation of < 10% from
experimental results. The results demonstrated a method for using commercial FEA simulation
software to more accurate predict thickness deviations at corners. Refinements of this method can
be integrated into commercial simulation codes and increase accuracy and utility. Most VI
simulation codes today focus exclusively on resin flow, on achieving full saturation, and
determining positions of inlet and outlet. The addition of mechanical effects in VI simulations can
guide modification of pre-processing parameters and reduce thickness variations in cured
laminates, thus achieving tighter tolerances. On the other hand, achieving greater accuracy in
simulations also increases preproduction analysis time. Nevertheless, computation time is far
cheaper than materials and labor, particularly for production of large parts.
While greater accuracy was achieved with the User Defined Geometry method than with the
simple Laminate Mesh method, refinements to the method are required to further increase

60
simulation fidelity. Alternative methods of segregating the corner region into distinct regions with
different compaction response or increasing the number of regions to achieve a more continuous
effect within the simulations may also present advantages. Integration of draping simulations with
finite element codes for VI process simulations can be leveraged to increase predictive accuracy
of geometry features. For example, simulation of fabric draping during layup and initial vacuum
pull will more accurately predict initial laminate geometry, which can be used as the UDG to
simulate final part thickness, eliminating the need for post-layup fabric measurements. The present
investigation focused only on the filling stage, neglecting potential geometric changes occurring
during post-filling and curing. Such deviations were observed in experimental trials, and while
minor, efforts to simulate such changes [77] are expected to be useful. FEA simulations may
benefit from future efforts to gauge the wet compression response of fabrics. Finally, examination
of specific responses for non-woven, unidirectional, or non-crimp fabrics is expected to yield
valuable insights and broaden utility of process simulations.
High areal weight fabrics are often used for vacuum infusion, but do not readily conform to
sharp corners. Given that the pre-infused preform thickness distribution was shown to dominate
the post-cured part thickness for concave parts, efforts to leverage techniques to increase
uniformity during layup may be beneficial. As reported in previous studies, convex corners exhibit
greater thickness uniformity than concave corners, and as such are preferred when possible.
Methods to increase fabric-tool parity at corners, such as preform binders and pressure intensifiers,
are almost mandatory, particularly in cases where large concave curvatures cannot be avoided.

61
Chapter 4. Flax-Reinforced Vitrimer Epoxy Composites Produced via RTM3
4.1 Introduction
Composite materials are instrumental in efforts to increase the sustainability of systems,
enabling increased fuel efficiency, extended vehicle range, and renewable energy structures like
wind blades. However, composite materials themselves rank poorly with respect to sustainability,
possessing large carbon footprints for production, large production scrap, and no complete
recycling pathway at end of life. In this study, we demonstrate the feasibility of producing natural
fiber composites comprising a fully recyclable vitrimer epoxy matrix using standard resin transfer
molding (RTM). The aim is to demonstrate feasibility and to highlight the merits and drawbacks
of the materials and processes, with the expectation that continued efforts to refine and optimize
will be demonstrated in the future.
The use of composite materials spans the aerospace, wind energy, marine, automotive, and
sporting goods industries. However, production scrap rates are upwards of 30% for carbon fiber
composites [78], and there is no viable recycling pathway for matrix materials. The wind turbine
industry alone will generate nearly half a million tons of carbon fiber-reinforced polymer (CFRP)
waste by 2050 [79]. Therefore, there is a strong incentive to find recycling methods to reduce such
waste. Currently, two of the main methods used for recycling are mechanical and thermal recycling
[22]. These methods have downsides: mechanical recycling is achieved by shredding the material,
reducing the length and alignment of recovered fibers, and can damage fibers as well [80], serving
more as downcycling. Thermal recycling involves pyrolyzing the resin, which can result in closeto-virgin fibers, but requires extreme temperatures, can produce harmful gasses, and eliminates
3 This study was published in the Journal of Composites Science in July 2024 with Steve Nutt as co-author [96].

62
any possibility of recycling the matrix [22]. Chemical recycling can yield close-to-virgin fibers
while allowing for reuse of the matrix, although the methods have not yet been scaled up [81,82].
Most carbon fiber recycling focuses on recovering fibers while eliminating and discarding the
resin. While carbon fiber constitutes the majority of raw material costs, aerospace-quality resin
prices are not insignificant [83], and discarding entails waste. In glass fiber-reinforced polymers
(GFRPs), the fibers are much cheaper than carbon [84], and as such, the cost fraction associated
with the matrix is significantly larger. Thus, alternative polymers that can be recycled, such as
vitrimers, are appealing.
Vitrimers contain thermally reversible covalent bonds that impart processing characteristics
normally associated with thermoplastics [85]. Even after curing, vitrimers can be reshaped at
elevated temperatures [30]. Furthermore, through a closed-loop process, the vitrimer polymer
matrix can be dissolved and reused, leaving the fabric intact [31]. Such a material may be suited
to applications in the automotive and sports industries, where minor repairs are more expected than
in the aerospace industry and component lifecycles are shorter.
Carbon fiber production is energy-intensive [86], which drives the need to reuse the fibers
[87]. Carbon fibers are produced from PAN or pitch precursor fibers, both of which are derived
from fossil fuels [88]. One alternative to synthetic fibers is natural fibers derived from plants,
which are, thus, renewable. Mechanical properties of natural fibers can approach those of lowerend glass fibers [89]. For applications where the remarkable, exceptional performance of carbon
fibers is unnecessary, but weight savings are still important (e.g., automotive), natural fibers may
find utility. For this study, flax fibers were selected. Flax fibers have been used in textiles dating
back to 5000 BC [90]. Flax fabric has mechanical properties that approach those of glass fibers,
but have a much lower density and cost.

63
Table 4.1 shows a comparison of select properties of flax, E-glass, and carbon fibers. Flax
fibers and fabrics, in principle, offer lower cost and greater sustainability relative to carbon and
glass fibers.
Table 4.1: Comparison of the general properties and cost of flax, glass [19], and carbon fibers [20].
Fiber Diameter
[µm]
Density
[103 kg/m3
]
Tensile Strength
[MPa]
Elastic Modulus
[GPa]
Cost
[USD/kg]
Flax 12 – 600 1.4 – 1.5 343 – 2000 27.6 – 103 0.30 – 1.55
E-Glass <17 2.50 – 2.59 2000 – 3500 70 – 76 1.65 – 3.25
Carbon 5 – 10 1.75 – 2.18 2000 – 7000 200 – 900 10 – 20
Most FRP manufacturing methods, such as autoclave cure, vacuum-bag-only prepreg (VBO),
and vacuum infusion (VI), are considered too slow for automotive processes, while high-pressure
RTM has been widely practiced. In the present work, resin transfer molding (RTM) was selected
to accomplish the project goals. RTM relies on pressurized infusion of low-viscosity resin into a
rigid closed mold containing a dry fabric preform [91]. Once fully infused, the mold is heated, and
the part is cured and subsequently demolded. By increasing the injection pressure, it is possible to
reduce the cycle time to under 10 min [92], and the process produces a net-shape part requiring
little finishing.
The objective of this study is to demonstrate the feasibility of producing recyclable flax–
vitrimer composites via RTM. Two reinforcements were tested: flax and glass fabrics and two
matrices—vitrimer epoxy and an industry-standard epoxy. The mechanical properties of the
product composites highlight the relative merits and limitations of the less common constituents—
flax and vitrimer epoxy—relative to conventional fibers and matrices. A complete recycling
pathway of the vitrimer composite is demonstrated and evaluated, along with its capacity for
forming and re-forming. The study demonstrates the use of lab-scale methods to evaluate how new
materials perform in manufacturing processes, an effort that helps to identify potential challenges
for scale-up and insertion into practice.

64
4.2 Materials and Methods
4.2.1 Resins
Two resins were selected: a vitrimer epoxy (Vitrimax T130, Mallinda Inc., Denver, CO, USA),
and an epoxy resin (RTM-6, Hexcel, Stamford, CT, USA), which was used as a baseline for
comparison.
The commercial epoxy was cured following the manufacturer’s recommended cure profile.
For the vitrimer epoxy resin, isothermal rheology tests were performed to determine a suitable
infusion temperature, maximizing the time during which the viscosity was less than 1 Pa·s, −55°C.
The rheology plots for standard infusion-and-cure cycles for both resins are shown in Figure 4.1.
Unlike the vitrimer epoxy, the viscosity of the commercial epoxy resin remained low (below 1
Pa·s) for the entire infusion cycle. In contrast, the vitrimer resin sustained low viscosity (<1 Pa·s)
for ~20min, with the viscosity steadily increasing from the moment of mixing. This behavior
effectively limited the resin pot-life and required infusion immediately after mixing. Vitrimer
gelation occurred during the temperature ramp to the first dwell for the vitrimer, while the
commercial epoxy gelled approximately 20 min into the 2 h dwell (cure).

65
Figure 4.1: Rheology graphs for Vitrimax T130 (left) and RTM-6 (right) following the specific injection-and-cure
temperature cycles used for each resin.
4.2.2 Fabrics
The three fabrics used included a 2 × 2 twill flax fabric with areal weight 200 g/m2
, a 2 × 2
twill glass fabric with similar areal weight (GF-22-200-100, Easy Composites, Ltd., Stoke-onTrent, UK), and a recycled flax fabric recovered from a flax–vitrimer composite. The recycled
fabric was extracted by first placing the flax–vitrimer laminate in a solution of resin precursors.
Due to the specific chemistry of the vitrimer, the matrix underwent imine exchange within the
solution, resulting in depolymerization. After 24 h, the fabric was removed, rinsed with ethanol,
and dried [11]. Surface micrographs of the three fabrics are shown in Figure 4.2 (VHX-5000,
Keyence, Itaska, IL, USA). Note the smaller tow-size of the glass fabric relative to the flax. The
only apparent distinction between virgin and recycled flax is a slight discoloration of the recycled
fabric.

66
Figure 4.2: Surface micrographs of the virgin flax (a), glass (b), and recycled flax (c) fabrics
The general characteristics of the three fabrics are tabulated in Table 4.2. Areal weight was
measured by weighing individual plies of each fabric, while density was measured using a gas
pycnometer (Micromeritics, Accupyc 1330, Norcross, GA, USA) and thickness was measured
with digital calipers. Tow width was measured from the micrographs in Figure 4.2. The three
fabrics had similar areal weights, but the uncompressed thickness of the glass fabric was ~1/3 of
the flax fabric. Similarly, the density of the glass fabric was roughly 50% greater than the flax. The
tow widths were also narrower in the glass fabric, allowing for a tighter weave and a more flexible
fabric. There was a negligible difference in the physical properties of the virgin and recycled flax
fabrics.
Table 4.2: General properties for fabrics used in this study.
Fabric Weave Areal Weight
[g/m2
]
Density
[103 kg/m3
]
Thickness
[mm]
Tow Width
[mm]
Flax 2 × 2 Twill 202 ± 6 1.46 ± 0.03 0.43 ± 0.02 2.58 ± 0.37
Glass 2 × 2 Twill 200 ± 7 2.57 ± 0.01 0.14 ± 0.01 0.71 ± 0.04
Recycled
Flax 2 × 2 Twill 207 ± 7 1.47 ± 0.01 0.43 ± 0.02 2.69 ± 0.32
Preforms for RTM were produced from 160 × 90mm plies—12 for the flax fabrics and 23 for
the glass fabrics, to achieve comparable thickness. A thin layer of binder (Airtac Mega, Airtech,
Huntington Beach, CA, USA) was sprayed onto the top surface of the first ply, and a second ply
was aligned on top. Using a custom mold, the stack was then compressed under ~500 kPa for 5
min at room temperature using a hot press (Genesis, Wabash MPI, Wabash, IN, USA). After

67
pressing, the preform was weighed and measured before repeating the process for the next ply until
a preform thickness of approximately 3.1 mm was achieved. Once fully consolidated, preforms
were trimmed to 150 × 80 mm (3 × 5 in).
4.2.3 RTM Part Production
A custom-built test cell (dubbed ‘mini-RTM’) was used to produce composite plates, as shown
in Figure 4.3. The test cell allowed for temperature and pressure measurements within the mold
while featuring a glass window for direct observation during infusion [93].
Figure 4.3: Test cell used to produce laminates in this study [93].
Leak testing was undertaken to ensure a satisfactory seal, and resin was infused following
procedures specific to the resin used. The vitrimer resin was mixed, then immediately vacuumdegassed for 5 min. Resin was poured into a pressure pot and heated to 55°C. The mold was also
heated to 55°C. The pressure pot was pressurized to 350 kPa (50 psi), and resin was allowed to
infuse. After saturation, the outlet was sealed and the mold was heated to 150°C. The laminate was
cured at 150°C for 60min, then post-cured at 180 °C for 60min. The procedure for the commercial
epoxy was similar, although the pressure pot was held at 70°C and the mold at 120°C during

68
infusion. Finally, the cure lasted 2 h at 180°C. The cure cycles are shown in the rheology curves
of Figure 4.1. After curing, select cross sections were examined to investigate the microstructural
features of the different laminates. Two sets of cross sections were produced: a full set of abrasive
polished sections for general inspection and a flax and glass cross-section produced using an ion
polisher (JEOL IB-09010CP, Peabody, MA, USA) for higher resolution microscopy. The test
matrix of the parts produced in this study is shown in Table 4.3.
Table 4.3: Test matrix detailing the specifics of all parts produced for this study.
Sample ID Fabric Resin Num Plies
FV-1 Flax Vitrimer 13
FV-2 Flax Vitrimer 12
FR Flax RTM6 12
GV Glass Vitrimer 23
rFV Recycled Flax Vitrimer 12
Following demolding, all laminates were characterized. Thickness and density were
measured, and the volume fraction was calculated based on the density of the laminate (𝜌𝑐
), the
matrix (𝜌𝑚), and the fabric (𝜌𝑓) through the rule of mixtures given in Equation (4.1).
𝑉𝑓 =
𝜌𝑐 − 𝜌𝑚
𝜌𝑓 − 𝜌𝑚
(4.1)
4.2.4 Re-Forming
A tool was designed for reshaping flax–vitrimer composites (Figure 4.4). The matched metal
tool featured a 20° 25.4 mm radius convex and concave corner. Threaded holes were included in
the lower tool, with matching unthreaded holes in the upper tool, to allow for positioning of
threaded rods for alignment
The forming trials were conducted with Sample FV-2. The laminate was cut into five coupons,
the first three of which were 19 × 127 mm, and these were used for mechanical tests, while the
remaining two were 19 × 63.5 mm and were used to prepare polished sections. Two of the three

69
full sized coupons and both half-length coupons were subjected to bending, and all half-length
coupons but one were subsequently straightened. Coupons were prepared by cutting the cured
laminate using a waterjet cutter (ProtoMAX, Omax Corporation, Kent, WA, USA). After cutting,
coupons were allowed to air dry for at least 24 h.
Figure 4.4: Tool used for bending of vitrimer parts. (Top) A picture of the tool. (Bottom) Schematic.
The bending process began by pre-heating both sides of the tool to 140°C. The composite
coupon was then carefully positioned over the center of the corner, and the top tool was replaced.
The entire tool was then re-heated for 15 min before cooling on an aluminum plate for 45 min and
demolding. Straightening followed the same procedure, replacing the matched tool with two
aluminum plates while using the top tool as a weight on top of the upper plate. In both cases, no
external pressure was applied, and the compaction force on the laminate was derived from the
weight of the top tool (1.9 kg).
4.2.5 Mechanical Testing
Two types of mechanical tests were conducted. Tensile tests in accordance with ASTM D3039
[94] provided the tensile strength and Young’s modulus, both fabric-dominated properties. Shortbeam-shear tests (ASTM D2344 [95]) provided the flexural strength, a matrix-dominated property.
All tests were performed on a load frame (Universal Testing Machine 5567, Instron, Norwood, IL,

70
USA), and displacements were measured using a 3D DIC system (ARAMIS, Trilion, King of
Prussia, PA, USA).
4.3 Results and Discussion
4.3.1 Cured Laminate General Properties
The general properties of the laminates are shown in Table 4.4, with Sample ID corresponding
to the material combination, F revers to flax fabric, G to glass, rF to recycled flax, while V and R
refer to vitrimer and RTM-6 resin respectively. The initial flax–vitrimer panel (FV-1) was produced
using 13 plies, although subsequent plates were produced with 12 plies. Note that the plates
produced with flax had higher volume fraction and lower thickness than the plate produced with
glass. Also, all flax laminates had densities roughly 75% of the densities of the glass composite
plates (despite a higher fiber volume fraction) due to the lower density of flax fabric.
Table 4.4: General properties of the cured laminates.
Sample ID Num Plies Thickness
[mm]
𝑽𝒇
[%]
Density
[103 kg/m3
]
FV-1 13 3.35 ± 0.05 59 ± 4 1.302 ± 0.003
FV-2 12 3.24 ± 0.02 54 ± 4 1.280 ± 0.005
FR 12 3.18 ± 0.04 48 ± 4 1.291 ± 0.001
GV 23 3.58 ± 0.04 45 ± 1 1.751 ± 0.015
rFV 12 3.19 ± 0.04 53 ± 2 1.284 ± 0.001
Cross-sections are shown in Figure 4.5. The distinct color difference for the FR part (Figure
4.5b), which allows for observation of the 0° fibers, is attributed to the different matrix
formulation. The significant void content in the glass laminate (GV) (Figure 4.5c) is attributed to
the tighter weave leading to greater air entrapment: The vitrimer resin exhibited off gassing during
infusion, entrapping gas. Resin-rich pockets were observed in the glass laminate, although further
from the inlet and outlet side (the top of the cross section), these same pockets contained large

71
voids. Finally, both FV-1 and rFV plates showed negligible differences (Figure 4.5a, d), showing
that virgin and recycled flax fabrics yielded similar microstructures.
Figure 4.5: Cross-section micrographs for select laminates. (a) FV-1, (b) FR, (c) GV, (d) rFV. Some visible voids
are circled in yellow.
The ion-polished cross-sections of FV-1 and GV, allowing for observation of the individual
fiber bundles, are shown in Figure 4.6, highlighting important differences between the two fibers.
First, glass fibers were uniformly round, with an average diameter of 12 ± 2 µm, while flax fibers
varied widely and were irregular, roughly rectangular or elliptical. Flax fibers showed a width-toheight ratio of 1.7 ± 0.5, the shorter dimension being 15 ± 4 µm and the larger being 25 ± 5 µm.
Furthermore, the flax fibers were often aggregated, with the edges of individual fibers conforming
to the shape of neighboring fibers. Some aggregates were so tight that the edges of individual fibers
could not be easily discerned. Finally, the lumen in each fiber appeared as a thin channel at the
fiber center; often resembling a void, but sometimes filled with matrix.

72
Figure 4.6: High resolution cross-sections of (left) sample FV-1 and (right) sample GV.
4.3.2 Mechanical Properties
Figure 4.7 shows the results of select tensile tests for each laminate. Note that the GV plate
surpassed the other laminates in tensile strength and modulus, showing a distinct performance gap
between the two types of fibers. All three flax laminates were grouped close together, with the
stress–strain curves nearly overlapping.
Figure 4.7: Tensile test results for all samples. (left) Entire curves displayed. (right) Close-up of the flax laminates
The mechanical properties of all laminates are summarized in Figure 4.8a, highlighting the
differences in the tensile properties of flax and glass laminates. Comparing FV-1 and GV, the glass
composite tensile strength was nearly 200% greater than the flax laminate, while the modulus was
~33% greater, despite the significant void content in the glass laminate. Comparing the different

73
matrix materials, the tensile strength of FR was 27 ± 9% greater than FV-1, while the tensile
modulus for the FR plate was 15 ± 4% less than that of FV-1. Comparing the flexural strength
obtained via short-beam shear tests, the flex strength of FR was 30 ± 18% greater than FV-1, while
the strength of GV was within the same bounds, with the flex strength being 7 ± 11% greater than
FV-1.
Figure 4.8: Mechanical properties for all coupons. (a) Standard bar graph. (b) Properties divided by density.
Despite having undergone the recycling process, the part with recycled fabric exhibited
mechanical properties similar to the fresh flax part. The laminate produced with recycled flax
fabric (rFV) showed tensile and flex strength approximately equivalent to the fresh flax fabric
laminate (FV-1), within 3 ± 12% and 16 ± 15%, with only a slight decrease in the elastic modulus
(14 ± 4%).
The strength-to-weight and modulus-to-weight ratios of the laminates are shown in Figure 8b.
The glass–fiber plate exhibited a greater tensile strength-to-weight ratio (122 ± 26%) than the flax
composite. However, the modulus-to-weight ratio was approximately equal (within 4%),
presumably because the greater weight of the glass fabric compensated for the difference in the
modulus. Furthermore, the flexural strength-to-weight ratio was lower for GV (21 ± 8% less) than
for FV-1, which was not unexpected. Flexural strength tends to be matrix-dominated, and increased
fiber stiffness typically causes only a modest increase in flexural strength. This phenomenon was

74
offset by the lower fiber volume fraction and increased void content in the glass laminate. Thus,
the two types of laminates yielded similar strengths, and thus, the GF laminate yielded a notably
lower strength-to-weight ratio.
4.3.3 Re-Forming
Laminate FV-2 was cut into five coupons (Table 4.5). Changes to the coupon thickness were
recorded during the initial forming (bending) and following re-forming (straightening), as shown
in Figure 4.9. After initial forming, the average thickness increased by 4.1 ± 0.5% across coupons
2–5. Following re-forming to flat coupons, the thickness decreased on average by 0.8 ± 0.5%, and
a total average thickness increase of 3.9 ± 0.3% occurred after the two-step forming process.
Table 4.5: Coupons cut from FV-2 and used for the reshaping trials.
Coupon # Processing Testing
1 None Tensile
2 Bending and Straightening Tensile
3 Bending and Straightening Tensile
4 Bending Micrograph
5 Bending and Straightening Micrograph
Figure 4.9: Thickness values of the coupons cut from FV-2 after different degrees of reshaping.
Bend-forming consistently caused a distinct defect, visible as a protrusion on the coupon
facing the convex tool. A similar defect was observed on the opposite side of the coupons after re-

75
forming. The protrusion appeared on the side of the coupon experiencing compression each time,
likely due to the fiber buckling under compression.
Polished sections of the corner region at different junctures during forming/re-forming are
shown in Figure 4.10. The surface defect was associated with fiber buckling, as shown in Figure
4.10a, albeit to a lesser degree. Wrinkles in the fabric were confined to the upper half of the plate,
corresponding to regions under compression. After re-forming (flattening), fiber waviness was
reduced but not eliminated (Figure 4.10b).
Figure 4.10: Micrographs of the center region of reshaped coupons. (a) Coupon 4 after initial bending. (b) Coupon
5 after subsequent straightening.
Tensile testing was performed on the first three coupons after re-forming (Figure 4.11).
Coupons 2 and 3 failed along the trace of the former corner bend. All three coupons followed a
similar initial linear curve before deflecting to a less steep secondary curve for the re-formed
coupons 2 and 3. While the modulus was negligibly greater for the re-formed coupons (5 ± 2%),
the tensile strength decreased (by nearly 50%). Re-forming caused negligible changes to the elastic
properties but weakened the fibers such that they failed at half the stress.

76
Figure 4.11: Results of the tensile testing of sample FV-2, with coupon 1 being as is and coupons 2 and 3 having
undergone bending and straightening. (Left) Stress vs. strain curve for each coupon, with the fitted curve used to
determine the tensile modulus. (Right) Bar graph showing the aggregate tensile properties for each coupon
Further insight into the observed strength reduction was provided by sections of the tested
coupons (Figure 4.12). At sites remote from the bend, the buckling of the fabric fibers was
negligible. Comparing Coupon 1 to Coupons 2 and 3 at the edges of the cross-section revealed
negligible differences. However, at the point of fracture, the fibers at the mid-plane of the beam
showed distinct differences. In Coupon 1, the fibers remained unbent, while in Coupons 2 and 3,
the fibers were bent away from the centerline. The shaping processes damaged and displaced fibers
such that they were no longer aligned with the loading direction. Fiber misalignment led to
premature failure and decreased tensile strength.

77
Figure 4.12: Micrographs of select FV-2 coupons after tensile testing.
Using a hand roller, an attempt was made to reduce the fabric distortion from bend-forming.
Gradual bending should allow fibers to conform to the corner more easily and prevent or minimize
the observed buckling and wrinkling. This process followed a similar procedure to the forming
process: the tool was heated to 140°C, the coupon was aligned with the corner of the tool, and the
roller was used to force the coupon to bend. Once the coupon was fully bent, the top tool was
positioned, and the process was resumed as before. The coupons were not long enough for tensile
testing after re-forming. However, a similar-sized coupon was bent using the standard method, and
the cross-sections of both were compared (Figure 4.13). The laminate formed using the roller
showed no defects, while the laminate formed without roller application showed distinct wrinkling
on the concave side (yellow ellipses, Figure 4.13a).

78
Figure 4.13: Cross-sections of bent coupons. (a) Standard pressing method with major defects highlighted; (b) hand
roller method.
4.4 Conclusions
Woven flax fibers were well-suited to RTM and yielded porosity-free composite plates, while
glass–fiber composites showed porosity because of the finer weave fabric. Flax fabric composites
also yielded superior flex strength-to-weight ratios and comparable tensile modulus-to-weight
ratios relative to the glass–fiber composites. Despite these advantages, glass–fiber composites
exhibited superior tensile strength values (~200% greater), which was not unexpected in light of
the reference values. However, because composite applications are often compression-critical and
subject to multi-axial loads, there may be potential to substitute flax fabrics for glass fabrics in
specific instances. One drawback to the future use of flax–fiber composites stems from the
observed variability in fiber size and dispersion within a given batch of fabric. Compounding this
issue are reports of batch-to-batch variability that will manifest in variations in mechanical
properties. Such variability, along with the shortage of property data and experience, must be
addressed before the use of flax fibers can be widespread.

79
Natural fibers such as flax are a “greener” alternative to synthetic fibers, and as such, they
may be useful substitutes for synthetic fibers, particularly in applications where increased
sustainability is sought and lower performance levels can be tolerated. These opportunities may
emerge in large-volume applications (wind blades, electric vehicles) because of the potential to
significantly reduce carbon footprint. However, even with natural fiber reinforcements, composites
will still be sent to landfills at the end of their life, highlighting the need for solutions to composite
recycling. Recycling solutions, if developed, may well change the calculus of fiber sustainability.
The vitrimer epoxy used here offers the prospects of increased sustainability as well as
exceptional formability. As was shown, the strength of the vitrimer laminates was approximately
30% less than the laminates produced with commercial epoxy in both tension and flex, but the
tensile modulus was matched. Furthermore, recovered flax fibers were virtually indistinguishable
from fresh flax fibers and yielded composites with nearly identical mechanical performances.
However, the processing characteristics of the commercial epoxy (pot life, fusion time, minimum
viscosity, ease of cleanup) were superior to those of the vitrimer epoxy. Nevertheless, this
difference is expected to narrow with future refinements.
Natural fibers undoubtedly afford one avenue towards increased sustainability, and the use of
vitrimer epoxy matrices complements that approach. First and foremost, the vitrimer epoxy matrix
allows for convenient separation of fabric and cured matrix at room temperature, affording the
possibility of repeated re-use, as opposed to downcycling at end of life via pyrolysis. Furthermore,
vitrimers greatly expand the processing space of thermoset composites, allowing for joining by
welding, thermo-forming, and repairs. These attributes are expected to compensate for the
aforementioned drawbacks to natural fibers and to offer a viable pathway for insertion into
practice.

80
Chapter 5. Resistance Welding of Carbon Fiber Reinforced Vitrimer
Composites4
5.1 Introduction
The attributes of carbon fiber-reinforced polymer (CFRP) composites have led to widespread
use. However, manufacturing of CFRP structures poses challenges, foremost of which are joining,
repairs, and recycling at end-of-life (EOL). These challenges can be mitigated by use of vitrimer
matrices. For example, a vitrimer epoxy has been shown to be both recyclable and reformable,
with properties not unlike a thermoplastic polymer [30]. In this study, we demonstrate a
thermoplastic-like welding process via resistance heating applied to vitrimers. The intent is to
elucidate the joining characteristics of the vitrimer, and to highlight how the unusual processing
characteristics can be leveraged to advance composites manufacturing.
Some vitrimers have been shown to be recyclable by EOL processing [31]. In such cases, the
vitrimer matrix can be dissolved in a solution of its own monomer. The recovered monomer can
be reused to form new vitrimer with negligible loss in performance characteristics [31]. The
recycling process can be applied to vitrimer composites, also. The fibers and matrix can be fully
separated, leaving aligned, continuous fibers, and fibers can be reused, with only a minor
decrement in mechanical performance [96].
Vitrimers exhibit other distinctive process characteristics, particularly with regard to shaping
and bonding. By applying heat or pressure, imine bonds within the matrix can be temporarily
broken and reformed, imparting formability [30,31]. Bond exchange can be performed for vitrimer
composites, allowing for reconsolidation, repair, and joining [33,97–100]. Much prior work on
4 This study has been submitted for publication with co-author Steve Nutt and is currently under review as of
October 2024.

81
vitrimer-vitrimer bonding involves hotpressing [99], although there are few reports on bonding
methods similar to thermoplastic welding [101]. Because of similarities in behavior of vitrimers
and thermoplastics [102], it is logical to attempt to leverage the manufacturing methods developed
for thermoplastic composites.
A distinctive advantage of thermoplastic composites is the ability to join parts via welding. In
contrast, thermoset composites are commonly joined with mechanical fasteners [103], adhesives
[104–107], co-curing [105], or even chemical reaction [102]. Joining with fasteners allows
disassembly for maintenance, although fasteners increase part count, system weight, and typically
require additional thickness (doublers) at the fasteners. Joining with thermoset adhesives and cocuring is nonreversible, and requires modifications to either the cure cycle, part mold, or post-cure
processing. In contrast, thermoplastic bonding is process flexible, and fusion welding is common
used for both part production and joining of individual parts. Multiple methods for fusion welding
have been developed, including hot-pressing, ultrasonic welding, induction welding, and the focus
of this study, resistance welding [26,28,108,109].
Fusion welding of thermoplastics involves heating the polymer to melting, applying pressure,
and allowing bonds to form across the adherends, then cooling to freeze the polymer, producing a
join that features a continuous thermoplastic polymer matrix [108–110]. Resistance welding
generates heat by passing electric current through a heating element, typically a metal mesh or
carbon fiber array placed between two thermoplastic adherends [111]. Consideration of the heating
element properties [112], as well as the power and time [113] required for welding is required to
ensure a strong weld. Resistance welding can be fast and scaled to larger parts [28,110,111].
The goal of the present study is to analyze and demonstrate resistance welding of vitrimer
composites, and thus leverage the ability of vitrimers to re-form bonds when heated. We developed

82
a resistance welding rig inspired by thermoplastic welding work and used it to weld vitrimer
composites. Adherends were welded in a single lap joint configuration, with initial welding
parameters determined by measuring temperatures. After exploring parametric effects, we were
able to form bonds that surpassed the strength of bonds formed through hot press welding.
Furthermore, after lap shear testing, the joints were rewelded, resulting in an unexpected increase
in weld strength. Repeated rewelding after lap-shear testing maintained the high shear strength,
even after seven weld-and-break cycles. The weld experiments in this work, while not fully
optimized, demonstrate potential for a reversible welding process for vitrimer composite joints.
Resistance welding was shown to be a suitable method for bonding vitrimer composites,
although the vitrimer composites required longer weld times than typical thermoplastics. The
ability to repeatedly weld and re-weld highlights the potential for vitrimer composite repairs that
can retain and possibly increase joint strength. Practical, reproducible, and scalable joining
methods can reduce part count and process complexity, as well as manufacturing costs.
5.2 Materials and Methods
5.2.1 Materials
A vitrimer epoxy resin suitable for composite manufacturing was selected for this study
(Vitrimax T100, Mallinda Inc, Denver, USA). The resin was developed for prepreg manufacturing,
with rheological properties suitable for filming (Figure 5.1). The cured density of the vitrimer was
1.07 ± 0.02 × 106
g/m3
.
A 2 × 2 twill fabric was selected (FibreGlast-1069, FibreGlast Development Corp, Brookville,
USA), and the same fabric was used for both heating elements and laminate manufacturing. Woven
fabric has been shown to be effective for use as a heating element, yielding more uniform
temperatures than unidirectional fibers [112]. The measured volumetric density of the fabric was

83
1.77 ± 0.01 × 106
g/m3
, the areal weight was 197 ± 3 g/m2
, and the dry thickness was 0.21 ± 0.03
mm.
Figure 5.1: Graphs showing the rheological behavior of the vitrimer resin. (a) Rheology of the resin filming and
prepreg pressing process; (b) Rheology during cure.
5.2.2 Resistance Welding Experimental Setup
A resistance welding rig was designed and assembled (Figure 5.2), following an earlier design
for resistance welding of thermoplastic composites [112,114,115]. The rig consisted of two sets of
two 76.2 × 25.4 × 12.7 mm brass bar clamps, positioned 28.4 mm apart. The clamps were secured
with brass threaded bolts and nuts. The lower set of nuts supplied compaction force, ensuring
electrical contact between the clamps and the heating element, and were tightened to a torque of 9
Nm. A custom-designed base for ensuring consistent spacing of the rig was 3D printed in-house
(Funmat HT, Intamsys, Eden Prairie, MN, USA) using PEEK filament. A combination of silicone
strips and garrolite bars were used as pressure-transferring material, support and thermal and
electrical insulation, placed above and below the adherends during welding. A 1-32 V DC, 30 A
programmable power supply (1901B, BK Precision) was used to control voltage and current. A
LabView virtual instrument was developed for both control and datalogging, using a DAQ (NI-

84
9219, National Instruments Corp., Austin, TX, USA). The compaction force on the adherend stack
during welding was supplied by a universal test machine (Instron 5567, Instron, Norwood, MA,
USA), with a 25.4 mm cube of garollite serving as a pressure applicator.
Data measured during welding included compaction force, crosshead displacement, current
and voltage. Power and resistance across the welding rig were calculated using the measured
voltage and current.
Figure 5.2: Resistance welding setup. (a) Isometric view, (b) Labeled front view cross-section diagram, (c) Side
view cross-section, (d) Circuit diagram
5.2.3 Laminate Preparation
Vitrimer-carbon laminates were prepared following the instructions provided by the resin
supplier. The procedure consisted of filming resin, pressing the film on either side of dry fabric,
curing the individual prepreg plies, and then consolidating the 1-ply laminates into the desired
final thickness. This process, while unorthodox, has been used to produce high quality parts, and
standard out-of-autoclave prepreg processing was not suitable for this resin system [33].

85
The resin weight content was greater than 40%, per manufacturer recommendation, and the
equivalent target fiber volume fraction was ~47%. This value was used to calculate the required
resin film thickness, through Equation (5.1), where 𝑡𝑟𝑒𝑠𝑖𝑛 refers to the one-sided resin film
thickness, 𝜌𝑎,𝑓 is the areal density of the fabric, 𝑤𝑓 is the fiber weight fraction and 𝜌𝑚 is the matrix
volumetric density. To achieve 𝑉𝑓 = 47%, a film thickness of 0.064mm was required; the addition
of a 0.152 mm thick backing paper (Release Ease® 236 TFNP, Airtech International Inc,
Huntington Beach, CA, USA), required a minimum spacer of 0.216 mm, however an aluminum
spacer of 0.229 ± 0.01 mm was the closest available. This was expected to reduce the fiber volume
fraction to approximately 42%, corresponding to a resin weight of 54%.
𝑡𝑟𝑒𝑠𝑖𝑛 =
𝜌𝑎,𝑓(1 − 𝑤𝑓)
𝜌𝑓 − 𝜌𝑚
(5.1)
Resin filming was performed by clamping a 25.4 mm diameter steel rod, preheated to 80°C,
onto a flat granite surface using the spacers to control the gap. The backing paper was threaded
into this gap, and pre-mixed resin was spread across the width of the paper, behind the rod. By
slowly pulling on the backing paper, the rod allowed a set thickness of resin to pass, yielding a
~0.077 mm thick resin film. The filmed resin was pressed on either side of carbon fabric. A hotpress (Genesis, Wabash MPI, Wabash, IN, USA) was used to apply 0.4 MPa at 100°C for 1 minute,
to allow the resin to fully impregnate the fabric, yielding carbon-vitrimer prepreg.
These single-ply prepregs were then cured, hanging in the oven, at 160°C for 1 hour, yielding
fully cured 1-ply laminates. The cured plies were flattened at 130°C for 2-4 seconds using minimal
pressure to remove any warping developed during cure. To produce the final laminates, the cured
plies were cut and stacked, following a [0]6 layup, and pressed at 1.4 MPa and 130°C for 5 minutes.
After hot-pressing, the laminates were cooled between room-temperature aluminum plates.
Laminates were then sectioned using a waterjet cutter (ProtoMAX, Omax Corporation, Kent, WA,

86
USA), into 31.8 × 25.4 mm coupons for cross-sectioning, and 101.6 × 25.4 mm coupons for the
lap-shear bond tests.
Volume fractions of the cured laminates were obtained using Equation (4.1), where 𝜌𝑐
, 𝜌𝑚,
and 𝜌𝑓 refer to the volumetric densities of the composite laminate, cured matrix and fabric,
respectively. All densities were measured using a gas pycnometer (Micromeritics, Accupyc 1330,
Norcross, US).
5.2.4 Heating Element Preparation
Heating element design was governed by two constraints: first, the 25.4 mm square region
between the adherends had to consist of fully saturated carbon-vitrimer composite and second, the
region under the clamps had to be dry carbon fiber. The first requirement arose from the need to
prevent a large dry-fiber region between the two adherends. Such a feature could lead to a
delamination-like failure in the heating element itself, as opposed to a matrix failure at the bond
line. The second requirement arose from the need for electrical contact between the brass clamps
and the heating element. [111,112].
A method involving discontinuous resin film distribution was deployed to produce the
discontinuous matrix [11,12,14,41]. The relatively high viscosity during cure prevented significant
change in resin distribution, and so the goal was to obtain the desired resin geometry in the
preparation of the prepreg. A two-stage modification to the vitrimer prepregging method described
above was used to produce the heating elements (Figure 5.3).
Heating elements were trimmed to 101.6 × 25.4 mm strips, with a 25.4 mm square matrix
region at the center. Resistance for each heating element was measured immediately before
welding, with the entire welding stack already in place, and the appropriate welding pressure
already applied.

87
Figure 5.3: Diagram showcasing the method used to produce heating elements.
5.2.5 Welding and Lap Shear Testing
Welding followed the arrangement depicted in Figure 5.2. The available parameters were
welding time, welding power and welding pressure. For this study, welding pressure was
maintained at 1.38 MPa, the recommended pressure for consolidation of vitrimer prepregs.
The adherends were secured above and below the heating element with a 25.4 mm overlap
single lap joint geometry. Compression was applied at a rate of 2mm/min, until the desired pressure
was achieved, at which point the system was switched to load-controlled to maintain that pressure.
With the pressure held constant, resistance was measured by driving a low current (< 3 A) for 15
seconds. The system was allowed to return to room temperature before starting the welding
procedure. During welding current was controlled to maintain the desired welding power until the
welding time was reached, accounting for deviations in the resistance. At that point, the system
was powered off and allowed to cool down for a minimum of 5 minutes. Pressure was maintained
throughout the entire process.
Lap joint coupons were tested for shear strength following ASTM Standards D1002 and
D5868 [116,117], using a test speed of 1 mm/min.

88
To determine suitable welding parameters, an initial investigation into the temperatures
obtained in the system was undertaken. Four thermocouples were positioned at the center of the
lap-joint stack, one on each side of each adherend. Temperature was measured for Welding Power
ranging from 15 W to 82 W. Based on the properties given by the manufacturer, vitrimer reshaping
begins at 100°C, and vitrimer bonding is to be done between 130 and 180°C. Thus, the time
required for the bonding surfaces to reach these temperatures was recorded.
Using the 𝑡180𝐶, the time required to reach 180°C, lap-joints were made for welding powers
ranging from 30 W to 60 W; given in Table 5.1. After testing, a second set of joints were developed
by increasing the weld time by 50%. Finally, a control was developed using the hot-press. A lapjoint stack, including both adherends and a heating element, was placed in the hot press and
consolidated at 130°C with 1.4 MPa for 5 minutes.
Table 5.1: Welding parameters used for the first set of lap-jointed specimens, determined from temperature
measurements
Lap Joint # Pressure
[MPa]
Power
[W]
Time (t180C)
[s]
1-1 1.38 30 300
1-2 1.38 40 91
1-3 1.38 50 74
1-4 1.38 60 44
With the results of the previous section, welding parameters were set for the second stage of
testing, given in Table 5.2. Following lap-shear testing, adherends were welded again, using the
same heating element and welding conditions. This process was repeated six times, or until the
joint failed in a manner that prevented future welding. Smaller coupons were used to obtain crosssections of the lap-joints after the first weld for these conditions.
A final set of joints were prepared using the welding parameters that yielded the highest shear
strength. Joints were welded, tested and then re-welded and cross sectioned, with the process
repeated for cross sections after the first five welds.

89
Table 5.2: Welding parameters used for the second set of lap-jointed specimens.
Lap Joint # Pressure
[MPa]
Power
[W]
Time (t180C)
[s]
2-1 1.38 30 300
2-2 1.38 40 135
2-3 1.38 50 108
2-4 1.38 60 62
2-5 1.38 30 450
2-6 1.4 Hot-Press @ 130°C 300
5.2.6 Reversing the Welded Joint
Using the welding conditions developed in the previous section, a reverse welding procedure
was developed. The top adherend was shortened to 28.6 mm and bonded to a garrolite slab and a
threaded insert using epoxy adhesive (EA E-20NS, Loctite, Düsseldorf, Germany). This was used
to apply compressive and tensile forces without interfering with the welding rig. The bottom
adherend and the brass clamps were secured to the welding rig using c-clamps. Welding followed
the standard procedure (30 W for 450 s).
Reverse welding used the same welding power, holding at 30 W for 120 s, while maintaining
uniform force on the adherend. After 120 s, tension was applied at 5 mm/min until the top adherend
separated from the heating element, ending the test.
5.3 Results and Discussion
5.3.1 Laminate and Heating Element Manufacturing
The fiber volume fraction of the laminates produced ranged from 30% to 37%, corresponding
to a resin weight fraction of 50 - 58%, well above the 40% required for consolidation. Thickness
was measured at 2.23 ± 0.16 mm. Laminates were inspected via cross-section imaging (Figure
5.4a). Some porosity was observed in the laminates, previously seen with this manufacturing
method and fabric [33].
Virgin heating elements featured a Vf of 28 ± 2% and a thickness of 0.62 ± 0.05 mm in the
weld region. A cross-section of the heating element is shown in Figure 5.4b. The average

90
resistance for the virgin heating elements was measured to be 0.80 ± 0.08 Ω, which is within the
expected range carbon fiber heating elements [112]. Including the measured resistance for parts
undergoing subsequent welding increases the resistance to 0.84 ± 0.14 Ω. This increase can be
explained by the manipulation required of the lap joint structure for testing and realigning with the
welding rig, resulting in occasional shearing, fraying and otherwise interacting with the bare fibers
used for clamping. However, the increase was small enough to be considered negligible during
welding, and the current was consistently adjusted to maintain the selected welding power.
Figure 5.4: Micrographs of select cross-sections (a) Vitrimer composite laminate, (b) Center region of a virgin
heating element
5.3.2 Welding Temperature Tests
The results from the power measurements during welding are shown in Figure 5.5. The timepower relationship follows a power curve, as shown by the fitted curves the time to reach each
temperature.

91
Figure 5.5: Graph showing the time required to reach 100, 130 and 180°C at specified welding power values, and
the welding time and welding power used for the two sets of lap joints.
The results of the time-power measurements were used to guide selection of the welding
parameters for the first set, outlined in Table 5.1. Joints were also fabricated with a 50% increase
in weld time, based on observations in preliminary trials that indicated that longer weld times
produced stronger welds. The shear strength of these joints is shown in Figure 5.6, alongside the
shear strength of the control joint produced via hot pressing. The results showed that 30 W
achieved the greatest weld strength, matching the results from hot-press welding, both of which
required a weld time of 300 s. Increasing the weld time by 50% increased the shear strength by
27%, 197%, and 54% for the 30, 40 and 50 W tests respectively, while the 60 W test resulted in a
25% decrease, corresponding to a 0.6 MPa drop. The decrease for the 60 W test is expected to be
due to both cases not providing sufficient welding time. The test parameters yielding the greatest
strength were selected for repeated welds. (The 30 W 300 s test was also included because it
approached the strength of the hot-press welded control, and the 60 W 62 s test was selected over
the 44 s test due to difficulties in performing the shorter test).

92
Figure 5.6: Graph showing the shear strength for the lap joints with welding conditions derived from the
temperature tests.
Cross-sections for coupon-scale joints produced using the second set of weld parameters are
shown in Figure 5.7. Joint 2-4 (Figure 5.7d), was produced with highest power and lowest weld
time, and the boundary between heating element and adherends was visible, indicting a poor bond
and explaining the lower strength recorded. In comparison, for joints welded at 30 W, (Figure 5.7a
and e) the boundary was difficult to distinguish. While no clear boundary was evident, the effects
of heating on the matrix were manifest by differences in gray level. In general, the boundary was
more distinct between the bottom adherend and the heating element, which was attributed to the
fact that lap joints consistently failed in this region. The differences in local conditions between
the top and bottom adherend arose from the geometry of the welding stack: the flexibility of the
heating element and silicone insulation allowed for deformation. This resulted in intimate contact
with the top adherend, to which pressure was applied directly, whereas the bottom adherend was
supported under the heating element.

93
Figure 5.7: Micrograph cross-sections across the centerline of resistance welded lap-joints for the second set of
welding conditions. Bond lines are shown with dashed yellow lines.
5.3.3 Repeated Welding
The results of lap shear tests for repeated welds are shown in Figure 5.8. For all joints, the
shear strength increased after the first or second weld, including the hot-press weld. Joints 2-1, 2-
3 and 2-5 all consistently surpassed the control after 3 welds. The joints failed primarily in the
matrix between the heating element and the bottom adherend, with the heating element generally
remained attached to the top adherend, except for the tests marked with an ×, where the failure
occurred through the heating element. The relationship between welding conditions and shear
strength was further confirmed at the sixth welding of the 2-3 joint, which was incorrectly welded
at 40 W for 135 s; this resulted in lower shear strength, more closely matching the strength of 2-2,
which was welded with those parameters. The shear strength recovered, however, after the seventh
weld performed with the appropriate welding parameters.

94
Figure 5.8: Bar graph showing the shear strength for the third set of lap joints after repeated welding. The ×
designates a joint that failed at the heating element, preventing further testing. The joint 2-3 was mistakenly welded
with 40W 135s for the sixth weld, marked with *.
On average, the strength increased by 44% between the first and second welds, excluding 2-
1 and 2-3, with the first showing a steady increase of ~13% for first four welds while the latter
increased only 4% after the second weld but 90% after the third. Between the third and fourth
weld, joints 2-2 and 2-4 both showed an approximate 30% increase in strength, while 2-5 and 2-6
decreased by 2 and 7% respectively. By the fourth weld, all joints showed an increase in strength
of 4.8 ± 1.1 MPa compared to the first weld, or at least a 33% increase, with the smallest increase
for the hot press welded joint.
The reason the strength increases occurred at different weld numbers for different conditions
was attributed to the total dwell time above welding conditions (> 130°C). For 2-1, 2-5 and 2-6,
there was sufficient time to allow the weld line to fully form, matching or exceeding the 5 minutes
used in consolidation. The two welded joints exceeded 200s above the welding conditions during
each weld. Meanwhile, the 2-2 joint showed increase in strength after both second and third welds,
dwelling ~100 s above 130°C each time. Finally, the shorter weld times (tests 2-3 and 2-4) showed
the largest increment in shear strength after 3 and 4 welds, respectively, as these weld times

95
featured only ~60 s and ~40 s above 130°C respectively. For a high strength weld, adherends
required sufficient dwell time at temperatures, which allowed surfaces to become flat and smooth
before bonding.
Fracture surfaces shown in Figure 5.9 revealed the effects of increased weld strength. Matrix
transfer manifested on the outer edges of the bond region for the first weld, and the amount of
excess matrix debris decreased as the number of welds increased. The fracture surfaces also
became more uniform after the second or third weld, matching the observed stabilization of shear
strength. Finally, the effects of weld time were visible in samples 2-4 and 2-5; the short weld time
(2-4) caused negligible matrix transfer, consonant with the low strength observed, and the surface
nearly matched the virgin unwelded surface for the first few tests. The longest weld time (2-5)
yielded the most severe deformation. The extended time allowed the adherend to reach temperature
and reshape.
The final set of welded joints were produced using the parameters for 2-5: 30 W for 450 s,
four joints, each sectioned after a select number of welds. The results from these tests are shown
in Figure 5.10 and Figure 5.11. Shear strength increased as the number of welds increased.
Welding at 30 W for 450 s produced a shear strength of 12.7 ± 2.1 MPa after the first weld, 14.1 ±
0.9 MPa after 2 welds, 14.5 ± 0.9 MPa after 3 welds and 15.6 ± 1.9 MPa after the fourth weld.
These values matched or surpassed the largest shear strength obtained via hot-press welding of
11.9 ± 0.2 MPa, which was obtained after 5 welds.

96
Figure 5.9: Micrographs showing the surface of the bottom adherend after lap-shear testing for the third set of
joints, after variable number of welds.
The thickness of the lap joint decreased with increasing weld iterations - 5.2 ± 2.5% after the
first weld, 2.9 ± 0.5% after the second, 2.6 ± 0.9% after three welds, 2.0 ± 0.4% after four and 2.0
± 0.7% after five welds, compared with the thickness of the previous iteration. The reshaping
temperature was exceeded during welding, increasing compaction of all joint elements, as
expected. Comparing the shear strength changes with the thickness changes, the largest increase
in shear strength occurred in the joint with the largest change in thickness - the joint sectioned after
3 welds, between first and second welding. This finding, coupled with the matrix buildup evident

97
at the edges of the heating element, indicated that repeated welding consistently reduced the matrix
thickness at the bond line with each weld until there was just enough to achieve a bond. This
assertion is supported by comparing the thickness of the heating element before welding and after
cross-sectioning. After the first weld, there is a 36 ± 5% decrease in thickness, while after 2, 3 and
4 welds a decrease of 50 ± 8%, 50 ± 6% and 50 ± 8% was measured. For the joint after 5 welds,
the exact boundary of the heating element was more difficult to determine, but the thickness
decrease was approximately 44 ± 10%.
Figure 5.10: Micrograph cross-sections across the centerline of resistance welded lap-joints for iterative welding at
30 W for 450 s. Bond lines are shown with dashed yellow lines.

98
Figure 5.11: Graphs detailing the iterative welding tests. (a) Lap shear strength vs welding iteration; (b) Joint stack
thickness vs welding iteration.
The thickness of the heating element stabilized after the second welding, a trend that also held
for the shear strength when welding at 30 W for 450 s. The shear strength did not stabilize as
quickly for other welding conditions because the decreased welding time reduced the time for
excess matrix to extrude from the bond line, requiring more welding iterations to achieve a similar
effect.
The effects of repeated welding caused matrix discoloration at the bond line, shown in Figure
5.9d, e. This observation correlated to behavior observed for repeated processing of the vitrimer
[33], which caused degradation and loss of strength. However, unlike previous reports, only the
surface of the matrix was being reworked with each successive weld. Thus, degradation did not
strongly affect the mechanical properties of the matrix, at least through six welds. This specific
welding condition also featured 150 s beyond the maximum consolidation temperature
recommended by the manufacturer, which led to the observed discoloration.
Occasionally, welding caused deformation of the adherends, in which the top adherend tended
to bend a few degrees, and care had to be taken to ensure the garrolite supports were sufficient to
support the adherends, as any sections of overlapping adherend would have caused the garrolite to

99
impress into the material. These findings were attributed to the temperature required for reshaping
the matrix (100°C) which was less than temperatures used during welding (130 – 180<°C).
However, reshaping was not difficult, and deviations in the coupons were easily corrected through
application of heat and pressure using a heat gun and aluminum plates. This measure maintained
the parallel lap joint structure and minimized torque during testing.
5.3.4 Reversing the Welded Joint
Using the welding conditions that achieved the highest strength, 30 W for 450 s, a pair of
adherends were joined and subsequently separated. The reverse welding process occurred in five
stages. First, current was driven through the heating element for 120 s; second, tension was applied,
with a slight drop in the rate of load increase visible, likely due to the flexibility of the heating
element, and compliance in the clamps. The third stage followed a linear increase in tension up
until the bond breaks, with threads of vitrimer remaining on the adherends, giving the fourth stage.
Finally, the joint was fully disconnected, and the top adherend is fully separated from the heating
element.
The photographs in Figure 5.12 show that some damage occurred in both adherends after
reversing the weld. Matrix transfer between the heating element and the top adherend was evident,
and tendrils of vitrimer matrix were observed on both, extending outwards. Furthermore, the lower
adherend experienced deformation at the edges due to the clamps used to hold it. However, results
obtained from repeated welding indicated that it was possible to realign the adherends and produce
a new weld.
At this stage, the reverse welding was tested as a proof of concept, and thus the weld was
reversed only on one side of the lap joint. Fully separating the heating element from both adherends
would require a modified welding rig. Furthermore, there were variables that could be modified to

100
increase the quality of separation, including the heating time pre-tension and the speed of
separation.
Figure 5.12: Photographs showing the surfaces of the adherends after reversing the weld.
5.4 Conclusions
Resistance welding was shown to be a potentially suitable method for joining vitrimer
composites. The shear strength of lap joints produced by resistance welding matched or exceeded
the strength of joints produced via hot press consolidation, the method recommended for joining
by the matrix supplier. The average shear strength (12.7 MPa) is reasonable, comparable to some

101
commercial adhesives [118,119], although less than high-performance adhesives [120] and what
has been achieved with resistance welding of thermoplastic composites [113,121,122]. Comparing
welding of vitrimer composites to thermoplastic composites, the vitrimer requires less welding
power, due to the lower temperature required for joining. However, the short bonding time, one of
the benefits touted for resistance welding, was not observed with vitrimers. The strongest bonds
formed after multiple minutes, matching the weld time for hot press welding. Furthermore, the
lower temperatures for welding also translates to lower temperatures for shape change. The
combination of extended welding time and closer proximity to shape-change temperatures requires
that vitrimer welding be performed under specific conditions to prevent unwanted shape-change
of adherends.
Vitrimer resistance welding findings demonstrate potential implications for repair. Joints
broken via lap-shear testing were readily re-formed, with no need for surface preparation, and
iterative welding increased shear strengths. Coupled with the ability to reversibly weld a joint,
vitrimers also have potential as temporary adhesives that do not add parasitic weight or boost part
count. While other methods are suitable for vitrimer bonding, including hot-press consolidation,
reversing the bond requires localized heating at the bond line, something capable through resistive
heating as shown in this study, but also potentially through methods such as induction, microwave,
laser, and other non-contact heating methods used for thermoplastic welding.
Sustainability is an increasingly important issue for composite materials, particularly EOL
processing. Vitrimer composites can potentially extend service life (via ease and effectiveness of
repair) and reduce waste through recyclability. The surprising results of iterative rewelding may
afford extended service lives for some applications. In addition, the cured-ply consolidation
processing could be suited to automotive applications, affording short process times and

102
eliminating freezer storage of prepregs. The ease of repair and recyclability make vitrimers a
suitable candidate for the wind industry, where repairs are performed in the field, typically done
using lower performance wet-layup resins and with limited equipment and infrastructure; and
where EOL processing has been criticized as yielding large amounts of waste.

103
Chapter 6. Conclusions and Suggested Future Work
6.1 Droplet Deposition and Semipreg
6.1.1 Conclusions
1. Direct droplet deposition as a method for producing semipreg: The viability of direct
droplet deposition as a method for producing a discontinuous resin distribution on carbon
fiber fabrics was shown. Prior to work done in 0 UD semipreg relied on first producing a
discontinuous resin film that was then applied to the fabric. It was demonstrated that by
properly characterizing the fabric, the depositing fluid, and the relationship between them,
it is possible to create a droplet array that will produce the desired surface resin distribution.
It was necessary to run a battery of tests to determine the appropriate droplet array,
however, highlighting the need for characterization and prediction. A droplet deposition
method would allow for on-demand prepreg production, which lowers costs by eliminating
the need for freezer storage and reducing waste from cutting scrap and expired prepreg and
providing easier prepreg customization.
2. Control over resin viscosity allows for control over droplet spreading: Doubling the resin
viscosity resulted in an 8- to 20-fold increase in sorption time for the simple fabrics.
However, the change in viscosity did not significantly affect spreading distance, suggesting
that increasing viscosity served only to slow down the spreading and not affect spread
geometry. Other work on semipreg has shown that the percentage of surface coverage can
affect the efficiency of air evacuation and final part quality [15], so relying on full sorption
may not always be ideal. Further increasing the viscosity, such as by freezing the material,
could arrest flow entirely, allowing for control over the final resin distribution and the
degree of impregnation.

104
3. Effect of surface topology on droplet spreading: Fabric surface topology was the most
significant effect on droplet spread behavior. Capillary effects were the primary driving
force for droplet flow; as such, the spacing between fiber bundles in Fabric D led to the
widest spread distance. Additional work detailing droplet flow on woven fabrics, included
in 0, further highlighted this, with regions of tow over and underlaps dominating the
direction of droplet flow [123]. For fabrics with significant surface topology, the
relationship between surface feature and droplet sizes had to be considered and could limit
the positions for droplet deposition, requiring large enough droplets to disregard the fabric
topology, or a monitoring system coupled with the deposition system, to fully control
droplet position. Furthermore, with Fabric D, increasing the droplet viscosity did not
significantly affect spreading or absorption time. On the other hand, low weight
unidirectional fabrics like Fabrics A-C, commonly used for aerospace prepreg, lacked
macro-level topology. This was shown to result in similar droplet spreading for all fabrics,
allowing droplet deposition to be position-agnostic.
6.1.2 Suggestions for Future Work
An expansion of this study with more droplet viscosities, including an investigation into the
effects of actively changing viscosity could improve the accuracy predicted droplet arrays. Work
into a simulated method of predicting droplet spreading was suggested, but never advanced.
Droplets in this study were delivered slowly, relying mainly on gravity; however, in a
manufacturing environment a faster delivery method will be necessary, so understanding the effect
of droplet impact speed on spread and absorption will be required. Questions remain about the
subsurface absorption of droplets, given that the current work focused entirely on surface effects.

105
The most impactful future work will be implementing droplet deposition as a manufacturing
method for semipreg. This work is currently underway at the M.C. Gill Composite Center, and a
droplet deposition system capable of rapid delivery of resin droplets onto fabric surfaces has been
developed. This system will enable more advanced studies that compare the quality of semipreg
produced via droplet deposition with previously used semipreg methods and conventional prepreg.
6.2 Vacuum Infusion Thickness Deviation and Prediction
6.2.1 Conclusions
1. Vacuum infusion process monitoring tool: An experimental setup was designed and built
that enabled in-situ visualization of vacuum infusion processing. This tool allowed for realtime monitoring of inlet and outlet pressure, part thickness and flow front position during
vacuum infusion. The setup was also used for characterization of the preform and fabric
response to the tool geometry, which was crucial for accurate simulation. These capabilities
offer increased potential for future VI studies at the M.C. Gill Composites Center.
2. Workflow for predicting thickness deviation at a corner for VI parts: A method was
developed that utilized commercial simulation software PAM-RTM/Visual-RTM to more
accurately predict the effect that corners have on thickness deviation. The process required
precise characterization of the relationship between tool geometry and fabric drapability,
with appropriate modifications to the simulation inputs. The modified simulations had <
10% deviation from experimental results, accurately predicting part geometry. By
improving pre-manufacture predictions, dimensional tolerances could be reduced, postmanufacture repairs or modifications minimized, and fewer parts scrapped.
3. Relationship between preform and cured part geometry: The motivation behind the study
presented in Chapter 3 was understanding the compounding effects of corners with the

106
inherent thickness deviation present in parts produced via linear vacuum infusion. Based
on the results, it is clear that the corner effect dominated. Furthermore, the draping of the
fabric in the corner, the geometry of the prefom prior to infusion and the response obtained
during debulking had the biggest effect on thickness deviation, and understanding these
was crucial in generating the User Defined Geometries that yielded more accurate
simulations. Based on these results, it is clear that concave corners should be avoided
whenever possible because convex corners reduce thickness deviation at the corner, and
are less likely to result in resin buildup and consumable bridging.
6.2.2 Suggestions for Future Work
As presented, the modified workflow requires characterization of the specific fabric-and-tool
geometry combination immediately prior to infusion, which reduces the impact as a premanufacturing process. The addition of fabric drapability simulation to predict that relationship
would remove this requirement and allow the thickness prediction simulation to depend solely on
the independent fabric characteristics. Furthermore, the current process only monitors thickness
effects during infusion, but the addition of a curing simulation could be implemented to predict
final part dimensions and account for the spring-back that is common in contoured parts [124].
The study focused solely on linear infusion of a single corner with one degree of curvature.
While such a geometry is common, more complex geometries must be considered. A 3D system
accounting for the out-of-plane effects of the corner on the fabric will be required for both
monitoring and simulation. Applying the simulation workflow to a demonstrator part will serve as
validation of the process and showcase its value as a manufacturing tool.
Finally, the vacuum infusion in-situ visualization experimental setup can assist future projects.
Intended for VI, the ability to simultaneously measure vacuum bag pressure and part thickness can

107
be leveraged for studies on debulking. The current sensor serves a limitation, preventing in-oven
monitoring, but a heated tool can also allow the experimental setup to be used to monitor thickness
changes for out-of-autoclave prepreg parts.
6.3 Flax-reinforced Vitrimer Composites
6.3.1 Conclusions
1. Refurbishing of the Mini-RTM visualization tool: While the in-situ visualization tool used
for Chapter 4 had been made by a previous composites group member, it was refurbished
and improved for this project. The addition of pressure sensors at the inlet and outlet allowed
for more accurate leak detection and improved monitoring of processing conditions. Any
users trained in the device had left the M.C. Gill Composites Center by the time it was needed
for this study, requiring troubleshooting and learning its operation. Thus, effort has and will
be undertaken to ensure the device can be easily used in future studies, by developing an indepth instruction manual.
2. Vitrimer resin compatibility with RTM: The Vitrimax T130 resin was not originally
developed with intent to use with infusion methods, with a custom formulation developed
specifically for infusion. It was necessary to characterize the rheological behavior of the
resin and determine appropriate processing conditions for liquid molding. For proper
manufacturing the resin had to maintain a low viscosity for a long enough time to allow for
resin degassing and infusing. Holding the resin at 55°C gave the resin ~1 hr under 100 Pa·s,
which was satisfactory for research purposes. However, these characteristics may not be
suitable for industry-scale infusion, where near indefinite pot life and sub 10 Pa·s viscosities
are preferred. Further modifications to the resin chemistry will be required if the resin is to
be used in industry.

108
3. Flax fabric as a substitute for glass: This project further corroborated what prior literature
had shown about flax fibers as a reinforcement: the tensile properties are lower than those
of glass fibers. However, this work highlighted that, accounting for the difference in weight,
flax fabric provides a similar modulus-to-weight ratio, and glass only has a strength-toweight ratio that is ~2x that of flax. Furthermore, both fabrics showed similar flexural
strength. Thus, flax is a viable reinforcement for composites in weight-critical applications,
particularly in non-structural elements. While not discussed in the study directly, the
mechanical properties obtained and cost values from literature show that flax also provides
higher tensile strength-, flexural strength-, and modulus-to-cost values, making it the
economic option. As the more sustainable option, flax should be considered in industries
where lightweighting and cost are crucial, such as the automotive and wind energy
industries.
4. Vitrimer recyclability: Previous work had shows that the vitrimer polymer is fully
recyclable. The work in Chapter 4 highlighted the observation that fabric recovered from
recycled vitrimer composites exhibited no loss in tensile or flexural strength, and only a
minor decrease in tensile modulus. This recycling method has been shown to be upscalable
and enables recovery of fabric that maintains long-range order. Flax is also known to readily
take up moisture [125], so the submersion into solution for recycling and cleaning was
expected to have more significant impact than on synthetic fibers, suggesting recycling of
such fibers may be even more successful.
5. Vitrimer composite reshaping: Reshaping the vitrimer matrix is touted as one of the unique
benefits of the material. However, in this project, some potential limitations of vitrimer
reprocessing were observed. Attempting a straightforward hot-press reshaping damaged the

109
fibers, resulting in significant decrease in strength. A method for potentially reducing this
damage was demonstrated, but not tested beyond cross-sectional analysis. Given the novel
nature of vitrimers as a matrix material for composites, it is important to fully understand
the advantages and limitations, to properly leverage vitrimers for composite manufacturing.
6.3.2 Suggestions for Future Work
A follow-up study on the recycling of carbon-reinforced vitrimer composites was suggested.
Said study would focus on the effects of repeated recycling on the mechanical performance of the
fibers, producing third-, fourth- and fifth-life parts. The current work focused on the mechanical
properties of composites produced with recycled material, however for fiber recycling projects it
is common to measure the properties of the fibers themselves. Due to the variance expected from
flax fibers, this measurement was not done in this project, but future work on vitrimer recycling
should investigate the effects of recycling on individual fibers. Single-fiber testing and measuring
any surface effects via XRD and high magnification microscopy will serve to further characterize
the effect vitrimer recycling has on the reinforcement.
Further work on the reprocessing of vitrimer composites is a necessity. Reshaping a multi-ply
laminate was damages the composite, significantly reducing strength, so similar damage might
occur for single plies. Given the manufacturer-recommended method of using vitrimers for
composites is to hot press-form single ply laminates together, such an investigation could be
crucial. Furthermore, this work set the stage for potential methods to reduce the damage from
reshaping with an alternative reshaping method, which could also serve as an avenue for future
work.

110
6.4 Vitrimer Welding
6.4.1 Conclusions
1. Resistance welding as a viable method for bonding vitrimers: The goal of the work
presented in Chapter 5 was to prove vitrimer-vitrimer bonding was possible via resistance
welding. This was successfully proven, showing that with appropriate welding conditions,
resistance welding can match and outperform hot-press consolidation as a method for
joining vitrimer adherends. Resistance welding is a more versatile method than hot-press
consolidation, enabling control of the welding temperature and more adaptability to parts.
Both methods showed potential for repair of broken bonds, showing an increase in shear
strength after iterative bonding. Lap joints produced with resistance welding often showed
significant deformation, as the combination of compaction pressure and heat from the bond
line extended into the adherends. This effect is expected to be more prominent on thinner
adherends whereas hot-press consolidation is expected to be more successful on thinner
parts. The ease of adapting resistance welding for vitrimers highlighted the potential for
other thermoplastic processing methods to be adapted for the material.
2. Repair of vitrimers: The most meaningful result of this study was the fact that it is not
only possible to bond vitrimer composites, but also to repeatedly re-bond after breaking.
Furthermore, after multiple iterations of the weld-and-break cycle, the shear strength of the
bond increases. Originally intended as an investigation into joining methods, this result
highlighted the potential for vitrimers in repairs. The unique properties of vitrimer
composites places them as ideal candidates for hard patch repairs – post-cure reshaping
precludes the need for highly accurate tooling to match patch geometry to the parent

111
material, ease of bonding provides a continuous matrix across the bond and restoration (or
increase) of pre-break matrix strength minimizes the damage after repair.
3. Strength retention after vitrimer reprocessing: This study also showcased the ability for
vitrimers to retain or increase in strength following reprocessing. As preliminary work to
this study, an investigation had been done to determine the effects of mechanical
reprocessing of neat vitrimer parts on mechanical properties. Originally presented at a
technical conference (SAMPE 2024), part of this work is included in Appendix B. These
results all show that vitrimer matrices do not appear to lose strength when undergoing the
first few recycling processes. This capability is limited to the matrix, and reprocessing of
vitrimer composites must be done more carefully. However, reprocessability means
vitrimers could be a suitable material for use as feedstock, for processes such as 3D printing
and extrusion, removing the need for cold storage of the mixed matrix and for pre-mixing
entirely. Vitrimers have potential to serve as fully shelf-stable thermosets, with the capacity
to be processed like thermoplastics.
4. Separation of welded joints: Resistance welding is a process that requires leaving the
heating element embedded in the part after joining. As a consequence, this same heating
element could be leveraged to separate the welded parts. The work in Chapter 5 showed
that vitrimer composites that had been joined via resistance welding could be separated
using the same process. This capability could be utilized in part repairs, by allowing
separation and replacement of a damaged component from the parent structure. Similarly,
it could facilitate the dismantling of composite structures at end-of-life, reducing the need
for cutting, and allowing for more intact components to be used in part recycling.

112
Temporary joining could also see use in maintenance by separating a section for inspection
or readjustment.
6.4.2 Suggestions for Future Work
Vitrimers are novel materials, with much to be characterized and studied. One suggestion for
future studies is adapting other thermoplastic processes for vitrimers. For example, one can look
into other thermoplastic fusion welding methods, use of vitrimer as a 3D printing thermoplastic
filament, and more.
Furthermore, a study into the creep behavior of vitrimers is necessary. Thermoset composites
are often touted as having excellent creep resistance, and given the low temperatures used for
reprocessing, vitrimers likely perform poorly at elevated temperatures. Improvements into the
resin formulation, to raise the vitrification temperature, could improve creep behavior, but would
consequently increase the temperatures and power required for both vitrimer consolidation and
welding methods.
Chapter 5 presented a proof-of-concept investigation into reversing a welded joint. Future
analysis to improve the quality of the separation might entail measuring the impact of the
separation on adherend structure, evaluating strength of a joint that has been separated, and
reattaching the original adherends, or joining the separated adherend to a new one. Iterating on and
characterizing the effect of resistance heating parameters used for separation is also an avenue for
future work.
Finally, this work highlighted the potential for vitrimers for repairs. Thus, a comparison
between a traditional hard patch repair, a soft patch repair, and a vitrimer hard patch repair will
provide valuable insight into their performance. Soft patch repairs are often preferred for in-field
repairs of wind energy structures as they eliminate the need for complex and expensive tooling. A

113
vitrimer hard patch could provide similar benefits as it could be cut into the required dimensions
and reshaped on-site into a geometry to match the parent structure.
6.5 Broader Implications and Final Thoughts
Composite materials are often identified solely by the reinforcement component, with CFRP
and GFRP products being referred to simply as carbon fiber and glass fiber products, disregarding
the polymer. However, not only do the polymer matrices makes upwards of 30% of the composite
by volume, but they are also the primary factor in considering what composite manufacturing
process to use. This dissertation focuses on the polymer matrix, from a single droplet of resin
interacting with the fabric surface, to liquid molding processes at ambient and elevated pressures,
and finally the unique cured-polymer flow available to vitrimers.
This dissertation presents four separate projects, all centered around the analysis of the
polymer matrix for composite manufacturing processes. The development and implementation of
in-situ monitoring tools to enhance manufacturing process control have been a significant focus at
the M.C. Gill Composites Center. The introduction of the vacuum infusion monitoring system, the
refurbishment of the Mini-RTM test cell, and the creation of the resistance welding rig will support
ongoing efforts in this area. Insights gained from these lab-scale analysis tools could be translated
into industrial scale manufacturing, leading to reduced manufacturing errors and improved part
quality control.
Vitrimer composites possess unique processing characteristics that sets them apart from
thermosets and thermoplastics. These distinctive properties hold significant promise in both
plastics and composite manufacturing. However, to fully leverage their potential, it is essential to
understand the limitations of these properties. Furthermore, vitrimers face the challenge of
disrupting the status quo, replacing not just pre-existing polymer matrices, but allowing for

114
modifications into the processing and post-processing of composites. By highlighting the unique
properties of vitrimers, and finding a specific niche for them, significant advancements in
composite manufacturing can be achieved.

115
References
[1] R. Reddy Nagavally, Composite Materials - History, Types, Fabrication Techniques,
Advantages, and Applications, in: IRF International Conference, Bengaluru, India, 2016:
pp. 25–30.
[2] B. Harris, A perspective view of composite materials development, Mater Des 12 (1991)
259–272.
[3] P. S, S. KM, N. K, S. S, Fiber Reinforced Composites - A Review, Journal of Material
Science & Engineering 06 (2017). https://doi.org/10.4172/2169-0022.1000341.
[4] Shivi Kesarwani, Polymer Composites in Aviation Sector, International Journal of
Engineering Research And 6 (2017) 518–525.
https://doi.org/10.17577/IJERTV6IS060291.
[5] D.K. Rajak, P.H. Wagh, E. Linul, Manufacturing technologies of carbon/glass fiberreinforced polymer composites and their properties: A review, Polymers (Basel) 13
(2021). https://doi.org/10.3390/polym13213721.
[6] S.Y. Park, C.H. Choi, W.J. Choi, S.S. Hwang, A Comparison of the Properties of Carbon
Fiber Epoxy Composites Produced by Non-autoclave with Vacuum Bag Only Prepreg and
Autoclave Process, Applied Composite Materials 26 (2019) 187–204.
https://doi.org/10.1007/s10443-018-9688-y.
[7] P. Hubert, T. Centea, L. Grunefelder, S. Nutt, J. Kratz, A. Levy, Out-of-autoclave prepreg
processing, in: Comprehensive Composite Materials II, Elsevier, 2017: pp. 63–94.
https://doi.org/10.1016/B978-0-12-803581-8.09900-8.
[8] K.J. Ahn, J.C. Seferis, Prepreg processing science and engineering, Polym Eng Sci 33
(1993) 1177–1188. https://doi.org/10.1002/pen.760331805.
[9] T. Centea, P. Hubert, Out-of-autoclave prepreg consolidation under deficient pressure
conditions, J Compos Mater 48 (2014) 2033–2045.
https://doi.org/10.1177/0021998313494101.
[10] T.A. Cender, P. Simacek, S. Davis, S.G. Advani, Gas Evacuation from Partially Saturated
Woven Fiber Laminates, Transp Porous Media 115 (2016) 541–562.
https://doi.org/10.1007/s11242-016-0784-x.
[11] S. Nutt, L. Grunenfelder, T. Centea, High-Permeability Composite Prepreg Constructions
and Methods for Making the Same, WO 2018/129378 A1, 2018.
[12] W.T. Edwards, P. Martinez, S.R. Nutt, Process robustness and defect formation
mechanisms in unidirectional semipreg, Advanced Manufacturing: Polymer & Composites
Science (2020) 1–14. https://doi.org/10.1080/20550340.2020.1834789.

116
[13] P. Martinez, S.R. Nutt, Robust Manufacturing of Complex-Shaped Parts Using Out-ofAutoclave Prepregs with Discontinuous Formats, in: CAMX, Dallas, TX, 2018.
[14] L.K. Grunenfelder, A. Dills, T. Centea, S.R. Nutt, S.G.K. Schechter, T. Centea, S.R. Nutt,
L.K. Grunenfelder, S.R. Nutt, S.G.K. Schechter, L.K. Grunenfelder, S.R. Nutt, Effect of
prepreg format on defect control in out-of-autoclave processing, Compos Part A Appl Sci
Manuf 93 (2017) 88–99. https://doi.org/10.1016/j.compositesa.2016.10.027.
[15] S.G.K. Schechter, T. Centea, S. Nutt, Effects of resin distribution patterns on throughthickness air removal in vacuum-bag-only prepregs, Composites Part A 130 (2020)
105723. https://doi.org/10.1016/j.compositesa.2019.105723.
[16] S. Maiti, M.R. Islam, M.A. Uddin, S. Afroj, S.J. Eichhorn, N. Karim, Sustainable FiberReinforced Composites: A Review, Adv Sustain Syst 6 (2022).
https://doi.org/10.1002/adsu.202200258.
[17] M.O. Guerrero-Pérez, M.A. Bañares, New reaction: Conversion of glycerol into
acrylonitrile, ChemSusChem 1 (2008) 511–513. https://doi.org/10.1002/cssc.200800023.
[18] E.R. Braga, R.R. Braga, L.A. Magalhães Pontes, Technical and Economic Analysis for the
Production of Acrylonitrile from Crude Glycerol, Chem Eng Technol 44 (2021) 2228–
2235. https://doi.org/10.1002/ceat.202100299.
[19] D.B. Dittenber, H.V.S. Gangarao, Critical review of recent publications on use of natural
composites in infrastructure, Compos Part A Appl Sci Manuf 43 (2012) 1419–1429.
https://doi.org/10.1016/j.compositesa.2011.11.019.
[20] Y. Liu, S. Kumar, Recent progress in fabrication, structure, and properties of carbon fibers,
Polymer Reviews 52 (2012) 234–258. https://doi.org/10.1080/15583724.2012.705410.
[21] P. Liu, C.Y. Barlow, Wind turbine blade waste in 2050, Waste Management 62 (2017)
229–240. https://doi.org/10.1016/j.wasman.2017.02.007.
[22] A. Isa, N. Nosbi, M. Che Ismail, H. Md Akil, W.F.F. Wan Ali, M.F. Omar, A Review on
Recycling of Carbon Fibres: Methods to Reinforce and Expected Fibre Composite
Degradations, Materials 15 (2022). https://doi.org/10.3390/ma15144991.
[23] S. Pimenta, S.T. Pinho, Recycling carbon fibre reinforced polymers for structural
applications: Technology review and market outlook, Waste Management 31 (2011) 378–
392. https://doi.org/10.1016/j.wasman.2010.09.019.
[24] K.B. Katnam, A.J. Comer, D. Roy, L.F.M. Da Silva, T.M. Young, Composite repair in
wind turbine blades: An overview, Journal of Adhesion 91 (2015) 113–139.
https://doi.org/10.1080/00218464.2014.900449.
[25] C. Ghafafian, V. Trappe, Prolonging structural integrity—Fatigue of scarf repairs for wind
turbine blade shell applications, Compos Part A Appl Sci Manuf 167 (2023).
https://doi.org/10.1016/j.compositesa.2022.107419.

117
[26] J. Barroeta Robles, M. Dubé, P. Hubert, A. Yousefpour, Repair of thermoplastic
composites: an overview, Advanced Manufacturing: Polymer and Composites Science 8
(2022) 68–96. https://doi.org/10.1080/20550340.2022.2057137.
[27] F. Delzendehrooy, A. Akhavan-Safar, A.Q. Barbosa, R. Beygi, D. Cardoso, R.J.C. Carbas,
E.A.S. Marques, L.F.M. da Silva, A comprehensive review on structural joining
techniques in the marine industry, Compos Struct 289 (2022).
https://doi.org/10.1016/j.compstruct.2022.115490.
[28] A.P. da Costa, E.C. Botelho, M.L. Costa, N.E. Narita, J.R. Tarpani, A review of welding
technologies for thermoplastic composites in aerospace applications, Journal of Aerospace
Technology and Management 4 (2012) 255–265.
https://doi.org/10.5028/jatm.2012.04033912.
[29] Y. Yang, Y. Xu, Y. Ji, Y. Wei, Functional epoxy vitrimers and composites, Prog Mater Sci
120 (2021). https://doi.org/10.1016/j.pmatsci.2020.100710.
[30] P. Taynton, K. Yu, R.K. Shoemaker, Y. Jin, H.J. Qi, W. Zhang, Heat- or water-driven
malleability in a highly recyclable covalent network polymer, Advanced Materials 26
(2014) 3938–3942. https://doi.org/10.1002/adma.201400317.
[31] P. Taynton, H. Ni, C. Zhu, K. Yu, S. Loob, Y. Jin, H.J. Qi, W. Zhang, Repairable woven
carbon fiber composites with full recyclability enabled by malleable polyimine networks,
Advanced Materials 28 (2016) 2904–2909. https://doi.org/10.1002/adma.201505245.
[32] P. Taynton, Design for recyclability at the molecular level: Dynamic covalent chemistry in
polymers, in: AMI Chemical Recycling, 2023.
[33] P. Martinez, S. Nutt, Vitrimers: Thermoplastic-like Properties in a Thermoset Polymer for
Fiber Reinforced Composites, in: SAMPE Conference, Long Beach, 2024: pp. 813–825.
[34] M. Steele, T. Corden, A. Gibbs, The development of out-of-autoclave composite prepreg
technology for aerospace applications, in: Sampe Tech Conference, 2011.
[35] B. Thorfinnson, T. Biermann, Degree of Impregnation of Prepregs - Effects on Porosity,
in: 32nd International Sampe Symposium, 1987.
[36] L. Repecka, J. Boyd, Vacuum-bag-only-curable prepregs that produce void-free parts, in:
Sampe Tech Conference, Long Beach CA, 2002.
[37] S.S. Tavares, V. Michaud, J.A.E. Månson, Through thickness air permeability of prepregs
during cure, Compos Part A Appl Sci Manuf 40 (2009) 1587–1596.
https://doi.org/10.1016/j.compositesa.2009.07.004.
[38] S.G.K. Schechter, T. Centea, S.R. Nutt, Polymer film dewetting for fabrication of out-ofautoclave prepreg with high through-thickness permeability, Compos Part A Appl Sci
Manuf 114 (2018) 86–96. https://doi.org/10.1016/j.compositesa.2018.08.002.

118
[39] P. Martinez, S. Nutt, Robust manufacturing of complex-shaped parts using out-ofautoclave prepregs with discontinuous formats, in: CAMX 2018 - Composites and
Advanced Materials Expo, Dallas, Texas, 2018.
[40] L.K. Grunenfelder, C. Fisher, C. Cabble, S. Thomas, S.R. Nutt, Defect Control in Out-ofAutoclave Manufacturing of Structural Elements, in: Sampe Tech Conference, 2012.
[41] S.G.K. Schechter, L.K. Grunenfelder, S.R. Nutt, Design and application of discontinuous
resin distribution patterns for semi-pregs, Advanced Manufacturing: Polymer &
Composites Science 6 (2020) 72–85. https://doi.org/10.1080/20550340.2020.1736864.
[42] S.H. Davis, L.M. Hocking, Spreading and imbibition of viscous liquid on a porous base,
Physics of Fluids 11 (1999) 48–57. https://doi.org/10.1063/1.869901.
[43] S.H. Davis, L.M. Hocking, Spreading and imbibition of viscous liquid on a porous base.
II, Physics of Fluids 12 (2000) 1646–1655. https://doi.org/10.1063/1.870416.
[44] N. Alleborn, H. Raszillier, Spreading and sorption of a droplet on a porous substrate,
Chem Eng Sci 59 (2004) 2071–2088. https://doi.org/10.1016/j.ces.2004.02.006.
[45] S. Comas-Cardona, C. Binetruy, P. Krawczak, Unidirectional compression of fibre
reinforcements. Part 2: A continuous permeability tensor measurement, Compos Sci
Technol 67 (2007) 638–645. https://doi.org/10.1016/j.compscitech.2006.07.020.
[46] T. Centea, P. Hubert, Modelling the effect of material properties and process parameters
on tow impregnation in out-of-autoclave prepregs, Composites: Part A 43 (2012) 1505–
1513. https://doi.org/10.1016/j.compositesa.2012.03.028.
[47] V. Michaud, A. Mortensen, Infiltration processing of fibre reinforced composites :
governing phenomena, Composites: Part A 32 (2001) 981–996.
[48] P. Neogi, C.A. Miller, Spreading kinetics of a drop on a rough solid surface, J Colloid
Interface Sci 92 (1983) 338–349. https://doi.org/10.1016/0021-9797(83)90156-X.
[49] L.M. Hocking, The wetting of a plane surface by a fluid, Physics of Fluids 7 (1995) 1214–
1220. https://doi.org/10.1063/1.868579.
[50] V.M. Starov, S.R. Kosvintsev, V.D. Sobolev, M.G. Velarde, S.A. Zhdanov, Spreading of
liquid drops over saturated porous layers, J Colloid Interface Sci 246 (2002) 372–379.
https://doi.org/10.1006/jcis.2001.8077.
[51] V.M. Starov, S.A. Zhdanov, M.G. Velarde, Spreading of liquid drops over dry porous
layers: Complete wetting case, Langmuir 18 (2002) 9744–9750.
https://doi.org/10.1021/la025759y.
[52] L.H. Tanner, The spreading of silicone oil drops on horizontal surfaces, J Phys D Appl
Phys 12 (1979) 1473–1484. https://doi.org/10.1088/0022-3727/12/9/009.
[53] D. Bonn, J. Eggers, J. Indekeu, J. Meunier, Wetting and spreading, Rev Mod Phys 81
(2009) 739–805. https://doi.org/10.1103/RevModPhys.81.739.

119
[54] E.W. Washburn, The Dynamics of Capillary Flow, Physical Review 17 (1921) 273–283.
https://doi.org/10.1103/PhysRev.17.273.
[55] S. van Oosterom, T. Allen, M. Battley, S. Bickerton, An objective comparison of common
vacuum assisted resin infusion processes, Compos Part A Appl Sci Manuf 125 (2019)
105528. https://doi.org/10.1016/j.compositesa.2019.105528.
[56] T. Osborne, An introduction to resin infusion, Reinforced Plastics 58 (2014) 25–29.
https://doi.org/10.1016/S0034-3617(14)70038-4.
[57] K.D. Tackitt, S.M. Walsh, Experimental study of thickness gradient formation in the vartm
process, Materials and Manufacturing Processes 20 (2005) 607–627.
https://doi.org/10.1081/AMP-200041896.
[58] Z. Grofti, B. Croteau-Labouly, R. Boukhili, P. Causse, S. Belouettar, H. Perrin, F. Trochu,
Simultaneous characterization of preform expansion and permeability in vacuum assisted
resin infusion, Polym Compos 43 (2022) 3560–3573. https://doi.org/10.1002/pc.26637.
[59] B. Yenilmez, M. Senan, E. Murat Sozer, Variation of part thickness and compaction
pressure in vacuum infusion process, Compos Sci Technol 69 (2009) 1710–1719.
https://doi.org/10.1016/j.compscitech.2008.05.009.
[60] A. Levy, P. Hubert, Vacuum-bagged composite laminate forming processes: Predicting
thickness deviation in complex shapes, Compos Part A Appl Sci Manuf 126 (2019).
https://doi.org/10.1016/j.compositesa.2019.105568.
[61] Y. Ma, T. Centea, S.R. Nutt, Defect reduction strategies for the manufacture of contoured
laminates using vacuum BAG-only prepregs, Polym Compos 38 (2017) 2016–2025.
https://doi.org/10.1002/pc.23773.
[62] G. Fernlund, J. Grifith, R. Courdji, A. Poursartip, Experimental and numerical study of the
effect of caul-sheets on corner thinning of composite laminates, n.d.
www.elsevier.com/locate/compositesa.
[63] J. KOZENY, Uber Kapillare Leitung der Wasser in Boden, Royal Academy of Science,
Vienna, Proc. Class I 136 (1927) 271–306.
https://cir.nii.ac.jp/crid/1571417124968987776.bib?lang=en (accessed May 27, 2022).
[64] P.G. Carman, Fluid flow through granular beds, Chemical Engineering Research and
Design 75 (1997) S32–S48. https://doi.org/10.1016/S0263-8762(97)80003-2.
[65] S. Bickerton, E.M. Sozer, S.G. Advani, Fabric structure and mold curvature effects on
preform permeability and mold filling in the RTM process. Part II. Predictions and
comparisons with experiments, n.d. www.elsevier.com/locate/compositesa.
[66] F. Trochu, E. Ruiz, V. Achim, S. Soukane, Advanced numerical simulation of liquid
composite molding for process analysis and optimization, in: Compos Part A Appl Sci
Manuf, Elsevier Ltd, 2006: pp. 890–902.
https://doi.org/10.1016/j.compositesa.2005.06.003.

120
[67] D. Adhikari, S. Gururaja, S. Hemchandra, Vacuum infusion in porous preform with
different mould configurations: Flow simulation and experimental validation, Journal of
Reinforced Plastics and Composites 40 (2021) 321–338.
https://doi.org/10.1177/0731684420960209.
[68] B. Yang, T. Jin, F. Bi, J. Li, Modeling the resin flow and numerical simulation of the
filling stage for vacuum-assisted resin infusion process, Journal of Reinforced Plastics and
Composites 33 (2014) 1976–1992. https://doi.org/10.1177/0731684414551039.
[69] Q. Govignon, S. Bickerton, P.A. Kelly, Simulation of the reinforcement compaction and
resin flow during the complete resin infusion process, Compos Part A Appl Sci Manuf 41
(2010) 45–57. https://doi.org/10.1016/j.compositesa.2009.07.007.
[70] P.P. Maung, T.L. Htet, G. V. Malysheva, Simulation and optimization of vacuum assisted
resin infusion process for large-sized structures made of carbon fiber-reinforced plastics,
in: IOP Conf Ser Mater Sci Eng, Institute of Physics Publishing, 2020.
https://doi.org/10.1088/1757-899X/709/2/022041.
[71] P. Martinez, B. Jin, S. Nutt, Permeability of Multiaxial and Non-Crimp Fabrics for
Vacuum Infusion, in: CAMX 2022 - Composites and Advanced Materials Expo, Anaheim,
2022.
[72] A.W. Chan, S. Hwang, Anisotropic In-Plane Permeability of Fabric Media, Polym Eng Sci
31 (1991) 1233–1239.
[73] R. Arbter, J.M. Beraud, C. Binetruy, L. Bizet, J. Bréard, S. Comas-Cardona, C. Demaria,
A. Endruweit, P. Ermanni, F. Gommer, S. Hasanovic, P. Henrat, F. Klunker, B. Laine, S.
Lavanchy, S. V. Lomov, A. Long, V. Michaud, G. Morren, E. Ruiz, H. Sol, F. Trochu, B.
Verleye, M. Wietgrefe, W. Wu, G. Ziegmann, Experimental determination of the
permeability of textiles: A benchmark exercise, Compos Part A Appl Sci Manuf 42 (2011)
1157–1168. https://doi.org/10.1016/j.compositesa.2011.04.021.
[74] N. Vernet, E. Ruiz, S. Advani, J.B. Alms, M. Aubert, M. Barburski, B. Barari, J.M.
Beraud, D.C. Berg, N. Correia, M. Danzi, T. Delavière, M. Dickert, C. Di Fratta, A.
Endruweit, P. Ermanni, G. Francucci, J.A. Garcia, A. George, C. Hahn, F. Klunker, S. V.
Lomov, A. Long, B. Louis, J. Maldonado, R. Meier, V. Michaud, H. Perrin, K. Pillai, E.
Rodriguez, F. Trochu, S. Verheyden, M. Weitgrefe, W. Xiong, S. Zaremba, G. Ziegmann,
Experimental determination of the permeability of engineering textiles: Benchmark II,
Compos Part A Appl Sci Manuf 61 (2014) 172–184.
https://doi.org/10.1016/j.compositesa.2014.02.010.
[75] H. Darcy, Les fontaines publiques de la ville de Dijon, Dalmont, Paris, 1856.
[76] ESI Group, Private communication, (2019).
[77] B. Caglar, B. Yenilmez, E.M. Sozer, Modeling of post-filling stage in vacuum infusion
using compaction characterization, J Compos Mater 49 (2015) 1947–1960.
https://doi.org/10.1177/0021998314541305.

121
[78] M.F. Khurshid, M. Hengstermann, M.M.B. Hasan, A. Abdkader, C. Cherif, Recent
developments in the processing of waste carbon fibre for thermoplastic composites – A
review, J Compos Mater 54 (2020) 1925–1944.
https://doi.org/10.1177/0021998319886043.
[79] A. Lefeuvre, S. Garnier, L. Jacquemin, B. Pillain, G. Sonnemann, Anticipating in-use
stocks of carbon fibre reinforced polymers and related waste generated by the wind power
sector until 2050, Resour Conserv Recycl 141 (2019) 30–39.
https://doi.org/10.1016/j.resconrec.2018.10.008.
[80] Y. Yang, R. Boom, B. Irion, D.J. van Heerden, P. Kuiper, H. de Wit, Recycling of
composite materials, Chemical Engineering and Processing: Process Intensification 51
(2012) 53–68. https://doi.org/10.1016/j.cep.2011.09.007.
[81] Z. Yu, Y.J. Lim, T. Williams, S. Nutt, A rapid electrochemical method to recycle carbon
fiber composites using methyl radicals, Green Chemistry (2023).
https://doi.org/10.1039/d3gc01765f.
[82] C.A. Navarro, C.R. Giffin, B. Zhang, Z. Yu, S.R. Nutt, T.J. Williams, A structural
chemistry look at composites recycling, Mater Horiz 7 (2020) 2479–2486.
https://doi.org/10.1039/d0mh01085e.
[83] F. Meng, J. McKechnie, S.J. Pickering, An assessment of financial viability of recycled
carbon fibre in automotive applications, Compos Part A Appl Sci Manuf 109 (2018) 207–
220. https://doi.org/10.1016/j.compositesa.2018.03.011.
[84] M.G. Bader, Selection of composite materials and manufacturing routes for cost-effective
performance, Compos Part A Appl Sci Manuf 33 (2002) 913–934.
https://doi.org/https://doi.org/10.1016/S1359-835X(02)00044-1.
[85] W. Post, A. Susa, R. Blaauw, K. Molenveld, R.J.I. Knoop, A Review on the Potential and
Limitations of Recyclable Thermosets for Structural Applications, Polymer Reviews 60
(2020) 359–388. https://doi.org/10.1080/15583724.2019.1673406.
[86] E. Asmatulu, M. Overcash, J. Twomey, Recycling of Aircraft: State of the Art in 2011,
Journal of Industrial Engineering 2013 (2013) 1–8. https://doi.org/10.1155/2013/960581.
[87] F. Meng, J. McKechnie, T. Turner, K.H. Wong, S.J. Pickering, Environmental Aspects of
Use of Recycled Carbon Fiber Composites in Automotive Applications, Environ Sci
Technol 51 (2017) 12727–12736. https://doi.org/10.1021/acs.est.7b04069.
[88] J. McLaren, The technology roadmap for plant/crop-based renewable resources 2020,
National Renewable Energy Lab, Golden, CO, 1999. https://www.osti.gov/biblio/756319
(accessed July 17, 2023).
[89] D. Nabi Saheb, J.P. Jog, Natural fiber polymer composites: A review, Advances in
Polymer Technology 18 (1999) 351–363. https://doi.org/10.1002/(SICI)1098-
2329(199924)18:4<351::AID-ADV6>3.0.CO;2-X.

122
[90] L. Yan, N. Chouw, K. Jayaraman, Flax fibre and its composites - A review, Compos B Eng
56 (2014) 296–317. https://doi.org/10.1016/j.compositesb.2013.08.014.
[91] B. Rackers, Resin Transfer Moulding for Aerospace Structures, Springer Netherlands,
1998. https://doi.org/10.1007/978-94-011-4437-7.
[92] A. Vita, V. Castorani, M. Germani, M. Marconi, Comparative life cycle assessment of
low-pressure RTM, compression RTM and high-pressure RTM manufacturing processes
to produce CFRP car hoods, in: Procedia CIRP, Elsevier B.V., 2019: pp. 352–357.
https://doi.org/10.1016/j.procir.2019.01.109.
[93] M. Anders, J. Lo, T. Centea, S.R. Nutt, Eliminating volatile-induced surface porosity
during resin transfer molding of a benzoxazine/epoxy blend, Compos Part A Appl Sci
Manuf 84 (2016) 442–454. https://doi.org/10.1016/j.compositesa.2016.02.024.
[94] ASTM D3039, Standard Test Method for Tensile Properties of Polymer Matrix Composite
Materials, 2017th ed., ASTM International, West Conshohocken, PA, 2017.
https://doi.org/10.1520/D3039_D3039M-17.
[95] ASTM D2344, Standard Test Method for Short-Beam Strength of Polymer Matrix
Composite Materials and Their Laminates, ASTM International, West Conshohocken, PA,
2022. https://doi.org/10.1520/D2344_D2344M-22.
[96] P. Martinez, S. Nutt, Flax–Reinforced Vitrimer Epoxy Composites Produced via RTM,
Journal of Composites Science 8 (2024) 275.
https://doi.org/https://doi.org/10.3390/jcs8070275.
[97] Q. Shi, C. Jin, Z. Chen, L. An, T. Wang, On the Welding of Vitrimers: Chemistry,
Mechanics and Applications, Adv Funct Mater 33 (2023).
https://doi.org/10.1002/adfm.202300288.
[98] A. Kandemir, M.L. Longana, I. Hamerton, S.J. Eichhorn, Developing aligned
discontinuous flax fibre composites: Sustainable matrix selection and repair performance
of vitrimers, Compos B Eng 243 (2022).
https://doi.org/10.1016/j.compositesb.2022.110139.
[99] E. Chabert, J. Vial, J.P. Cauchois, M. Mihaluta, F. Tournilhac, Multiple welding of long
fiber epoxy vitrimer composites, Soft Matter 12 (2016) 4838–4845.
https://doi.org/10.1039/c6sm00257a.
[100] P. Yu, H. Wang, T. Li, G. Wang, Z. Jia, X. Dong, Y. Xu, Q. Ma, D. Zhang, H. Ding, B. Yu,
Mechanically Robust, Recyclable, and Self-Healing Polyimine Networks, Advanced
Science 10 (2023). https://doi.org/10.1002/advs.202300958.
[101] A.K. Biswal, A. Nandi, H. Wang, A. Vashisth, Ultrasonic welding of fiber reinforced
vitrimer composites, Compos Sci Technol 242 (2023).
https://doi.org/10.1016/j.compscitech.2023.110202.

123
[102] A. Vashchuk, Y. Kobzar, Chemical welding of polymer networks, Mater Today Chem 24
(2022). https://doi.org/10.1016/j.mtchem.2022.100803.
[103] S.D. Thoppul, J. Finegan, R.F. Gibson, Mechanics of mechanically fastened joints in
polymer-matrix composite structures - A review, Compos Sci Technol 69 (2009) 301–329.
https://doi.org/10.1016/j.compscitech.2008.09.037.
[104] J. Kupski, S. Teixeira de Freitas, Design of adhesively bonded lap joints with laminated
CFRP adherends: Review, challenges and new opportunities for aerospace structures,
Compos Struct 268 (2021) 113923. https://doi.org/10.1016/j.compstruct.2021.113923.
[105] M.G. Song, J.H. Kweon, J.H. Choi, J.H. Byun, M.H. Song, S.J. Shin, T.J. Lee, Effect of
manufacturing methods on the shear strength of composite single-lap bonded joints,
Compos Struct 92 (2010) 2194–2202. https://doi.org/10.1016/j.compstruct.2009.08.041.
[106] X. Wu, K. He, Z. Gong, Z. Liu, J. Jiang, The shear strength of composite secondary
bonded single-lap joints with different fabrication methods, J Adhes Sci Technol 34 (2020)
936–948. https://doi.org/10.1080/01694243.2019.1690775.
[107] R. Balkova, S. Holcnerova, V. Cech, Testing of adhesives for bonding of polymer
composites, 2002.
[108] A. Yousefpour, M. Hojjati, J.P. Immarigeon, Fusion bonding/welding of thermoplastic
composites, Journal of Thermoplastic Composite Materials 17 (2004) 303–341.
https://doi.org/10.1177/0892705704045187.
[109] A. Siddique, Z. Iqbal, Y. Nawab, K. Shaker, A review of joining techniques for
thermoplastic composite materials, Journal of Thermoplastic Composite Materials (2022).
https://doi.org/10.1177/08927057221096662.
[110] V.K. Stokes, Joining methods for plastics and plastic composites: An overview, Polym
Eng Sci 29 (1989) 1310–1324. https://doi.org/10.1002/pen.760291903.
[111] D. Stavrov, H.E.N. Bersee, Resistance welding of thermoplastic composites-an overview,
Compos Part A Appl Sci Manuf 36 (2005) 39–54.
https://doi.org/10.1016/j.compositesa.2004.06.030.
[112] C. Ageorges, L. Ye, M. Hou, Experimental investigation of the resistance welding for
thermoplastic-matrix composites. Part I: heating element and heat transfer, Compos Sci
Technol 60 (2000) 1027–1039.
[113] C. Ageorges, L. Ye, M. Hou, Experimental investigation of the resistance welding of
thermoplastic-matrix composites. Part II: optimum processing window and mechanical
performance, Compos Sci Technol 60 (2000) 1191–1202. https://doi.org/10.1016/S0266-
3538(00)00025-7.
[114] M. Dubé, P. Hubert, J.N.A.H. Gallet, D. Stavrov, H.E.N. Bersee, A. Yousefpour, Metal
mesh heating element size effect in resistance welding of thermoplastic composites, J
Compos Mater 46 (2012) 911–919. https://doi.org/10.1177/0021998311412986.

124
[115] M. Hou, L. Ye, Y.-W. Mai, An Experimental Study of Resistance Welding of Carbon Fibre
Fabric Reinforced Polyetherimide (CF Fabric/PEI) Composite Material, 1999.
[116] ASTM D1002-10, Standard Test Method for Apparent Shear Strength of Single-Lap-Joint
Adhesively Bonded Metal Specimens by Tension Loading (Metal-to-Metal) 1, 2019.
https://doi.org/10.1520/D1002-10R19.
[117] ASTM D5868-01, Standard Test Method for Lap Shear Adhesion for Fiber Reinforced
Plastic (FRP) Bonding, 2023.
[118] A.F. Ávila, P. de O. Bueno, An experimental and numerical study on adhesive joints for
composites, Compos Struct 64 (2004) 531–537.
https://doi.org/10.1016/j.compstruct.2003.09.052.
[119] J.A.B.P. Neto, R.D.S.G. Campilho, L.F.M. Da Silva, Parametric study of adhesive joints
with composites, in: Int J Adhes Adhes, 2012: pp. 96–101.
https://doi.org/10.1016/j.ijadhadh.2012.01.019.
[120] X. Wu, K. He, Z. Gong, Z. Liu, J. Jiang, The shear strength of composite secondary
bonded single-lap joints with different fabrication methods, J Adhes Sci Technol 34 (2020)
936–948. https://doi.org/10.1080/01694243.2019.1690775.
[121] R. Balkova, S. Holcnerova, V. Cech, Testing of adhesives for bonding of polymer
composites, 2002.
[122] R.E. Murray, J. Roadman, R. Beach, Fusion joining of thermoplastic composite wind
turbine blades: Lap-shear bond characterization, Renew Energy 140 (2019) 501–512.
https://doi.org/10.1016/j.renene.2019.03.085.
[123] P. Martinez, B. Jin, S. Nutt, Effect of fiber bed architecture on single resin droplet spread
for prepreg manufacturing, in: CAMX 2019 - Composites and Advanced Materials Expo,
2019.
[124] T. Gajjar, S.J. Joshi, K.M. Patel, Prediction of Spring-back Deformation for CFRP
Reflectors Manufactured using Various Processes, International Journal of Applied
Engineering Research 13 (2018) 144–148. http://www.ripublication.com.
[125] A. Moudood, A. Rahman, A. Öchsner, M. Islam, G. Francucci, Flax fiber and its
composites: An overview of water and moisture absorption impact on their performance,
Journal of Reinforced Plastics and Composites 38 (2019) 323–339.
https://doi.org/10.1177/0731684418818893.
[126] ASTM D638, Standard Test Method for Tensile Properties of Plastics, ASTM
International, West Conshohocken, PA, 2022. https://doi.org/10.1520/D0638-14.
[127] ASTM D790, Standard Test Methods for Flexural Properties of Unreinforced and
Reinforced Plastics and Electrical Insulating Materials 1, ASTM International, West
Conshohocken, PA, 2017. https://doi.org/10.1520/D0790-17.

125
[128] P. Martinez, B.C. Jin, S. Nutt, Droplet spreading on unidirectional fiber beds, Journal of
Composites Science 5 (2021). https://doi.org/10.3390/jcs5010013.
[129] P. Martinez, B. Jin, S. Nutt, Thickness variation in contoured composite parts by vacuum
infusion, Advanced Manufacturing: Polymer and Composites Science 9 (2023).
https://doi.org/10.1080/20550340.2023.2279604.
[130] P. Martinez, B. Jin, S.R. Nutt, Effect of Fiber Bed Architecture on Single Resin Droplet
Spreading for Prepreg Manufacturing, SAMPE Journal 56 (2020) 18–26.

126
Appendix A: Droplet Spreading on Non-Unidirectional Fabrics5
A.1 Fabrics
Two fabrics were selected for this study, in addition to the fabrics included in 0. The first was
a plain weave (PW) fabric (FibreGlast Product #2363) with an areal weight of 119 g/m2
, a tow
count of 1K and a tow width of 0.94mm. The second was a 2×2 twill weave (TW) fabric
(FibreGlast Product #1069) with an areal weight of 193 g/m2
, a 3K tow count, and a tow width of
1.79 mm. Surface photographs of the two fabrics are given in Figure A.1.
Testing on these fabrics was limited, with only the 30 Pa·s facsimile fluid droplet used in
testing, and the spread distance not being directly measured.
Figure A.1: Surface images of the two additional fabrics. (Left) plain weave; (right) twill weave
5 This work was presented at the Composites and Advanced Materials Expo (CAMX) in September 2019, in
Anaheim, CA, and subsequently featured in the SAMPE Journal in July/August 2020 with Bo Jin and Steve Nutt as
co-authors [123,130].

127
A.2 Results
A.2.1 Time to Full Sorption
The time to full sorption was 560 s for the PW fabric and 787 s for the TW fabric. These values
are presented alongside those from 0 in Figure A.2. Despite being closer in areal weight to fabrics
A and B, the sorption times for the additional fabrics were closer to those of fabrics C and D.
Figure A.2: Graph showing the time to full sorption for all six fabrics for the 30 Pa·s facsimile fluid
A.2.2 Surface Coverage over Time
The surface coverage of the fluid is shown in Figure A.3. Unlike the similar figures presented
in 0, these were not normalized by 𝑡ℎ0. As expected, with fibers aligned in two directions, the fluid
spread in both x- and y-directions.

128
Figure A.3: Images showing the surface covered by the droplet over the topographical map of the fiber bed. Times
of each contour are at 10, 100, 500, 1000, 2000 seconds after droplet deposition, as well as at the last moment at
which the droplet can be observed. (above) PW Fabric; (below) TW Fabric
A.3 Conclusions
The addition of pinholes in the fabric increased the rate of fluid absorption, reducing the
sorption time despite the lower areal weight. The effect of tow overlaps and underlaps on the
spreading can be seen in Figure A.3. For the PW fabric, these regions served as boundary lines,
preventing flow in some cases, causing the droplet to spread mostly to the south-east of the field
of view. The effect of the tow boundaries can be more clearly seen in the TW fabric, however, with
significant flow occurring along the tows the droplet was deposited on, and the edges of those tows
delineating the end points of the flow distances. Much like the effect of depositing a droplet in the
boundary between two bundles on Fabric D, the positioning of the droplet on the TW fabric entirely
dictated the spreading behavior. On the other hand, while the fabric topology did affect the flow
boundaries for the PW fabric, the spread was more even in both x- and y-directions.

129
The results of these tests, coupled with the results obtained for Fabric D, further emphasized
the effect that droplet size vs macro feature size has on the spread of the droplet. In cases where
the topology of the fabric is significantly smaller than the droplet, as in Fabrics A-C and the PW
fabric, the orientation of the fibers is what will dictate flow behavior. However, when the size of
the droplet and the macro-level features of the fabric are on a similar scale, then these features will
control the spread behavior.

130
Appendix B: Effects of Repeated Reprocessing on Neat Vitrimer Parts6
B.1 Neat Vitrimer Part Production and Reprocessing
Neat resin samples were produced by degassing the resin before and after mixing in a vacuum
oven (LR Environmental), then pouring into silicone tray. Both resin and hardener were degassed
for 15 minutes at 60°C, followed by mixing the resin for 2 minutes, and then degassing under halfvacuum for 10 minutes at 60°C. A heat gun was used to attempt to remove any bubbles. The resin
was then cured uncovered in an oven, following a cure cycle provided by the supplier: a 20-minute
hold at 80°C, followed by a 20-minute hold at 115°C and finally a 3-hour hold at 135°C, with 20-
minute ramp times in between.
Despite multiple degassing and careful removal of residual bubbles before curing, neat resin
samples consistently had bubbles in the cured product, arising from volatiles emitted during cure.
Leveraging the thermoforming properties of vitrimers, a process to remove bubbles from the cured
product was developed: The sample would be pressed to a film thickness, thin enough that no
bubbles would remain, and then cut and consolidated back to a thickness of ~3 mm.
The reshaping process was done using a hot-press (Wabash Genesis). With the press heated to
130°C, the resin was pressed for 5 minutes using 10 tons of force for the initial thickness reduction.
The following consolidation was performed with 3.175 mm (0.125 in) aluminum spacers and at
approximately 1.4 MPa (200 psi) compression. This resulted in near defect-free neat resin samples.
An overview of the process is shown in Figure B.1, note the high amount of bubbles present in
Figure B.1a, which are gone by Figure B.1b.
6 This work was presented at the Composites and SAMPE Conference in May 2024, in Long Beach, CA with Steve
Nutt as co-author [33].

131
Figure B.1: Neat resin thermoforming process. (a) As-cured neat resin sample. (b) Pressed sample removes bubbles.
(c) Reconsolidated, void-free sample
Following each set of tests, all part fragments were recovered, cleaned with ethanol, allowed
to air dry and then reshaped. Generally, each thermoforming cycle used 10 tons of force, apart
from the final step prior to testing, which again used 1.4 MPa and aluminum spacers.
Two neat resin parts were produced and studied, henceforth referred to as Sample 1 and
Sample 2. Sample 1 was reshaped 21 times, with tests being performed after cycles 3, 6, 9, 12 and
21. Sample 2 was reshaped 10 times, with tests performed after 2, 4, and 10 cycles. Samples before
being cut for testing are shown in Figure B.2. Darkening of the polymer was observed as the
number of cycles increases. Loss of material is also apparent in Sample 1, due to the smaller size
of each part; this loss comes from the material lost during cutting of the test coupons, as well as
pieces that could not be recovered after testing.
S1-C3 S1-C6 S1-C9
S1-C12 S1-C21

132
S2-C2 S2-C4 S2-C11
Figure B.2: Set of pictures showing neat resin samples after certain number of reshaping cycles, just before cutting
and testing.
B.2 Mechanical Testing
Three sets of tests were performed to determine the effect that the reshaping had on the
mechanical properties of the polymer. Tensile, using ASTM D638 [126]; short beam shear, using
ASTM D2344 [95]; and three-point bending, using ASTM D790 [127]. These tests were selected
due to the variety of properties measured, with the inclusion of short-beam shear, as it is typically
a test performed to measure matrix-dominated flexural strength. Furthermore, given the small size
of SBS samples, including the tests was simple. Test geometries were prepared using a waterjet
cutter, resulting in some material being lost.
Test coupons for tensile were prepared following the Type I specification of ASTM D638,
dog-bone samples with a gauge section of width 12.7 mm (0.5 in) and length 57 mm (2.25 in). For
short-beam shear, coupons were approximately 3 mm thick 18 mm long and 6mm wide, though
each coupon was cut to specific dimensions based on the individual coupon thickness, per the
ASTM 2344 standard. Finally, the three-point bending coupons were 12.6 mm (0.5 in) wide and
62 mm long, following the appropriate standard.

133
Testing speeds were 5mm/min, 1mm/min and 2mm/min for tensile, short beam shear and three
point bending, respectively. Displacement for the tensile testing was measured using a 3D DIC
system (ARAMIS, GOM), with a speckle pattern painted onto the surface of the testing coupons
using water soluble paint, so it could be removed before reforming. Displacement for the threepoint bending tests was measured using a strain gauge.
B.3 Results
The results of this test are shown in Figure B.3. Unfortunately, due to the loss of material,
there was a limitation on the number of coupons produced from Sample 1 after 6 cycles, resulting
in only one test coupon for S1-C9, S1-C12 and S1-C21. Within the range of 3-12 cycles, there
does not appear to be a significant loss in strength, modulus or elongation for Sample 1; however,
between 12 and 21 cycles, the strength and elongation dropped by 38% and 45%, respectively,
while the modulus increased 7%. Sample 2 also showed little change within the first 10 cycles.
Figure B.3: Results from the tensile test. (Left) maximum tensile strength, (Middle) Young's modulus, (Right)
maximum elongation.
The results from the three-point bending test were similar to the tensile tests, flexural strength,
flexural modulus, and flexural elongation, shown in Figure B.4. Sample 1 showed no significant
decrease in properties between cycles 6 and 21; three-point bending tests only started after 6 cycles,
so S1-C3 is omitted. The results from Sample 2 are more interesting, however. A steady decrease
in both flexural strength and flexural modulus were observed as the number of cycled increased,
while the elongation increased significantly. When testing S2-C10, samples were exhibiting much

134
more plastic behavior, with the samples visibly bending, and not snapping as was observed in
previous tests.
Figure B.4: Results from the three-point bending test. (Left) maximum flexural strength, (Middle) flexural modulus,
(Right) maximum elongation
The short-beam shear tests results are shown in Figure B.5. Despite only providing one
property, the results of these tests show a much clearer trendline than either tensile or three-point
bending tests: increasing the number of cycles results in a decrease in short-beam strength. From
Sample 1, the strength appears to be exponentially decreasing, or could be reaching an asymptote,
and of course more testing would be required to confirm this. By cycles 9 and 10, short beam
strength had dropped by 50 and 69% for samples 2 and 1 respectively. It was observed that samples
for S1-C21 were quite fragile, and several coupons delaminated simply during coupon preparation.
Much like during the three-point bending test, S2-C10 exhibited unique behavior, with the coupons
bending and deforming in the test fixture, with the tests being terminated due to the load-head
travel exceeding the specimen thickness, as opposed to the snapping failure observed in other tests.
Given the matrix tends to be load bearing in short-beam shear tests of composites, testing this
property for neat resin was important.

135
Figure B.5: Results from the short-beam shear tests.
Samples in cycles below 9 fractured as a bulk material, whereas delamination failure was
observed more commonly in S1-C9,12,21 and specially in S2-C10, which became fully
delaminated during three-point bending, which the authors attribute the resulting mechanical
properties to.
B.4 Conclusions
Vitrimers were shown to be able to be repeatedly reprocessed, through cycles of flattening,
cutting, stacking and flattening again. This did not significantly reduce mechanical properties of
the polymer for at least 6 cycles, indicating that the material can be mechanically recycled, without
resorting to the full chemical recycling process available to this vitrimer. While a decrease in
properties in the first set of cycles was observed, it remained under a 20% decrease, highlighting
that the process used to remove bubbles by reshaping a cured panel is satisfactory.
The mechanical properties of the vitrimer have been shown to be below typical aerospace
thermosets, in both Chapter 4 and Chapter 5, however as a novel material it is possible to improve
these. The capacity for reprocessing places vitrimers as a sustainable material, allowing for both
mechanical and chemical recycling, and suggests possible use in the field of repairs.

Abstract (if available)

Abstract This thesis is focused on the importance of the polymer matrix in fiber reinforced polymer composites manufacturing. The aim is to improve knowledge of process parameters for the advancement of composite manufacturing methods and processes. This work is separated into four projects, evolving in complexity and flow behavior. The first project was on single droplet deposition on dry fabric. The second investigation revolved around the relationship between infusion parameters and thickness deviation for vacuum infusion. The third investigation was into the compatibility of a novel resin formulation and the RTM process, as well as comparing a sustainable natural fiber to conventional glass fiber. Finally, the fourth study investigated flow of a cured thermoset vitrimer matrix via resistance heating for fusion bonding of composites.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

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

Characterization, process analysis, and recycling of a benzoxazine-epoxy resin for structural composites

PDF

Chemical depolymerization of amine-epoxy cured carbon fiber reinforced polymer composites and their re-use

PDF

Ageing and mechanical failure of fiber reinforced polymers

PDF

Chemical recycling of amine/epoxy composites at atmospheric pressure

PDF

Sound transmission through acoustic metamaterials and prepreg processing science

PDF

Fiber reinforced hybrid phenolic foam

PDF

Process planning for robotic additive manufacturing

PDF

Developing efficient methods for the manufacture and analysis of composites structures

PDF

Fabrication and analysis of prepregs with discontinuous resin patterning for robust out-of-autoclave manufacturing

PDF

Fabrication of polymer films on liquid substrates via initiated chemical vapor deposition: controlling morphology and composition

PDF

Efficient manufacturing and repair of out-of-autoclave prepreg composites

PDF

Understanding structural changes of polymer films during vapor phase deposition

PDF

Sustainable manufacturing of out-of-autoclave (OoA) prepreg: processing challenges

PDF

Motion-assisted vat photopolymerization: an approach to high-resolution additive manufacturing

PDF

In situ process analysis for defect control during composites manufacturing

PDF

Development of composite oriented strand board and structures

PDF

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

PDF

Comprehensive uncertainty quantification in composites manufacturing processes

Asset Metadata

Creator Martinez Martinez, Patricio (author)

Core Title Polymer flow for manufacturing fiber reinforced polymer composites

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Mechanical Engineering

Degree Conferral Date 2024-12

Publication Date 11/13/2024

Defense Date 10/24/2024

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

Tag composite manufacturing,composites,fiber reinforced polymers,flax,OAI-PMH Harvest,resistance welding,RTM,semipreg,thermoset resin,vacuum infusion,vitrimers

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nutt, Steve (committee chair), Gupta, Satyandra (committee member), Williams, Travis (committee member)

Creator Email mart136@usc.edu,pmm530@gmail.com

Unique identifier UC11399DIAC

Identifier etd-MartinezMa-13629.pdf (filename)

Legacy Identifier etd-MartinezMa-13629

Document Type Dissertation

Format theses (aat)

Rights Martinez Martinez, Patricio

Internet Media Type application/pdf

Type texts

Source 20241113-usctheses-batch-1222 (batch), 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.

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