Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Fabrication and analysis of prepregs with discontinuous resin patterning for robust out-of-autoclave manufacturing
(USC Thesis Other)
Fabrication and analysis of prepregs with discontinuous resin patterning for robust out-of-autoclave manufacturing
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
FABRICATION AND ANALYSIS OF
PREPREGS WITH DISCONTINUOUS RESIN
PATTERNING FOR ROBUST OUT-OF-
AUTOCLAVE MANUFACTURING
by
Sarah G.K. Schechter
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)
May 2020
DEDICATION
I dedicate this manuscript to my loving husband, Aaron,
and our furry best friend, Shmuley.
iii
ACKNOWLEDGEMENTS
None of this work would be possible without the funding and donations
that supported my work. I was generously funded by the M.C. Gill
Composites Center and received donations of materials from Airtech
International, Patz Materials & Technologies, and Textreme. I am also
extremely grateful to the Achievement Rewards for College Scientists (ARCS)
Foundation for their annual scholarship.
I would like to extend my sincere thanks to my advisor, Dr. Steven
Nutt, for his constant guidance throughout my Ph.D. work. Dr. Nutt has been
supportive and has given me the freedom to pursue various projects without
objection. I am also very grateful for his scientific advice and knowledge and
many insightful discussions and suggestions.
There are several colleagues at USC who I would like to acknowledge.
I am deeply indebted to the late Dr. Timotei Centea. He provided guidance to
me that helped lay the groundwork for my career as a researcher. He
continues to inspire by his example and dedication to the students he served
over the course of his career. I am also extremely grateful to Dr. Lessa
Grunenfelder. She was instrumental in helping me define the path of my
research and I greatly appreciate the kindness she bestowed upon me. In
addition, I am greatly appreciative to our lab manager, Yunpeng Zhang,
whose calm presence was a source of comfort. He kept our labs running
smoothly and effectively, which allowed me to focus on my research rather
iv
than fixing equipment. Finally, I would like to thank two excellent
undergraduate students, Claire Carlton and Angela Yang, for their helpful
research assistance.
In addition, I would like to extend my deepest gratitude to my family. I
want to thank my husband and love of my life, Aaron, who has been a
constant source of support and encouragement during the challenges of
graduate school and life. I am particularly appreciative of his tireless work
editing my manuscripts and critiquing my presentations. I am truly thankful
for having him in my life. And, of course, I would not have been able to go
through my Ph.D. without our furry best friend, Shmuley, who provided me
with steadfast love, companionship and comfort. I also want to thank my
parents, Frank and Carol Katz, who raised me to be hard working, to be kind,
and to enjoy life along the way. I would not be who am I today without both of
my parents. I want to thank my smart older sister, Nicole Katz, for being a
role model in my life. Finally, I want to thank my father-in-law and mother-
in-law, Alan Schechter and Barbara Castler-Schechter, for always rooting for
me along the way and treating me like their own daughter.
v
TABLE OF CONTENTS
DEDICATION .................................................................................................... ii
ACKNOWLEDGEMENTS ................................................................................ iii
TABLE OF FIGURES ..................................................................................... viii
TABLE OF TABLES ....................................................................................... xiii
ABSTRACT ....................................................................................................... xiv
INTRODUCTION ................................................................................................ 1
CHAPTER I: Polymer Film Dewetting for Fabrication of Out-Of-
Autoclave Prepreg with High Through-Thickness Permeability ...................... 4
1. Background .................................................................................... 4
2. Materials ...................................................................................... 10
3. Film dewetting ............................................................................. 12
3.1 Surface opening dimensions ................................................ 12
3.2 Rheological behavior ............................................................ 18
4. Prepreg fabrication and characterization ................................... 20
4.1 Ammonia curing .................................................................. 20
4.1 X-ray micro-CT .................................................................... 23
4.1 Permeability ......................................................................... 25
5. Laminate characterization .......................................................... 29
5.1 In-situ visualization ............................................................. 29
5.2 Surface defects and bulk porosity ....................................... 32
5.3 Laminate structure .............................................................. 37
5.4 Bulk factor ............................................................................ 38
6. Conclusion .................................................................................... 39
vi
CHAPTER II: Effects of Resin Distribution Patterns on Through-
Thickness Air Removal in Vacuum-Bag-Only Prepregs .................................. 42
1. Background .................................................................................. 42
1.1 Background .......................................................................... 43
1.2 Objective ............................................................................... 45
2. Modeling ....................................................................................... 46
2.1 Model development .............................................................. 46
2.2 Parametric study ................................................................. 48
2.3 Statistical analysis .............................................................. 60
3. Experiments ................................................................................. 66
3.1 Materials .............................................................................. 66
3.2 Permeability ......................................................................... 68
3.3 Porosity ................................................................................ 71
4. Conclusion .................................................................................... 73
CHAPTER III: Air Evacuation and Resin Impregnation in Semi-pregs:
Effects of Feature Dimensions .......................................................................... 76
1. Background .................................................................................. 76
2. Materials and experimental methods ......................................... 82
3. Resin characterization ................................................................. 88
4. Resin flow model .......................................................................... 93
4.1 Model development .............................................................. 93
4.2 In situ observation ............................................................... 95
5. Room temperature debulk ......................................................... 102
6. Experimental validation ............................................................ 106
7. Conclusion .................................................................................. 109
vii
CHAPTER IV: Design and Application of Discontinuous Resin
Distribution Patterns for Vacuum-Bag-Only Prepregs ................................. 112
1. Background ................................................................................ 112
2. Materials and methods .............................................................. 116
3. Results ........................................................................................ 121
3.1 Quality analysis of semi-preg formats .............................. 121
3.2 Bulk factor .......................................................................... 126
3.3 Complex shapes ................................................................. 127
3.4 Design considerations and limitations for semi-preg
fabrication .................................................................................. 129
4. Conclusion .................................................................................. 140
CONCLUDING REMARKS ............................................................................ 142
RECOMMENDATIONS FOR FUTURE WORK ............................................ 145
REFERENCES ................................................................................................ 147
viii
TABLE OF FIGURES
Figure 1 Top view images of resin dewetted at 104 °C either still on the
substrate (silicone-coated backing paper) or pressed onto unidirectional
carbon fiber tape - (a) dewetted for 15 s on the substrate (b) dewetted for
4 min on the substrate (c) dewetted for 30 s pressed onto fiber bed
(d) dewetted for 4 min pressed onto fiber bed. .................................................. 13
Figure 2 Dimensions of surface openings created by the dewetting
process – (a) surface area exposed (b) average diameter of surface
openings (where half this amount, or the radius, is the longest distance
to wet-out during cure). ..................................................................................... 15
Figure 3 Diagram of 4-ply stacked prepreg demonstrating the
interconnectedness of z-direction air evacuation networks –
(a) continuous resin film [with in-plane air evacuation only], (b)
dewetted resin film with small openings, (c) dewetted resin film with
large openings. ................................................................................................... 17
Figure 4 Viscosity profiles of resin without any dewetting treatment
compared to resin dewetted for 30 s and 2 min at 104 °C. Times t1, t2,
and t3 correspond to the 1 h room temperature hold, the cure
temperature ramp to 121 °C, and the 2 h cure dwell, respectively. .................. 19
Figure 5 Laminates cured using the ammonia bath method to
visualize the fiber and resin distributions before oven curing –
(a) Control Prepreg, (b) Prepreg 104-30, (c) Prepreg 104-120. ......................... 22
Figure 6 2D and 3D images of X-ray micro-CT scans of the ammonia-
cured Control Prepreg and Prepreg 104-120. The images represent the
structure of the prepreg prior to cure and the resultant interconnection
of void space. Both prepregs exhibited intra-tow void space, whereas
only the Control Prepreg exhibited inter-ply (enclosed) void space. ................. 24
Figure 7 Through-thickness permeability values for 1, 2, 4, and 8 plies
of prepreg. The value of 0% surface area exposed represents dry fabric,
100% represents continuous film, and the values in between represent
the two levels of dewetting focused on in this work. Error bars were
omitted for viewing clarity (statistical information can be found in
Table 1). ............................................................................................................. 27
Figure 8 Screenshots of window vacuum bag curing – (a) Prepreg
104-30 (b) Prepreg 104-120. The times t1, t2, and t3 correspond to the
point in the cure cycle that the laminate is being exposed to (as
indicated on Fig. 4). ........................................................................................... 30
ix
Figure 9 (a) Surface porosity values of laminates created with
continuous resin film and dewetted resin film under the conditions of
standard cure and sealed edges. (b, c, d) Surface images of laminates
cured under the sealed edges condition with accompanying close-up of
porosity or other defects – (b) Control Prepreg (c) Prepreg 104-30
(d) Prepreg 104-120. .......................................................................................... 33
Figure 10 (a) Internal porosity values of laminates created with
continuous resin film and dewetted resin film under the conditions of
standard cure and sealed edges. (b, c, d) Cross-sectional images of
laminates cured under the sealed edges condition for internal porosity
measurements – (b) Control Prepreg, (c) Prepreg 104-30, (d) Prepreg
104-120. .............................................................................................................. 35
Figure 11 Bulk factor of laminates produced with continuous resin
film and discontinuous (dewetted) resin film. .................................................. 38
Figure 12 (a) The traditional manufacturing of prepreg and laminates
with continuous resin illustrating air entrapment at the resin-to-resin
interface and the mechanism of air removal in the mid-plane. (b) The
fabrication of prepreg and laminates with a discontinuous resin
distribution illustrating air removal in the through-thickness direction
and the potential for seal-off between resin layers (due to large resin
feature dimensions and random placement during stacking). ........................ 48
Figure 13 (a) Examples of the discontinuous resin patterns created via
the polymer film dewetting technique. (b) Examples of the discontinuous
resin patterns coded in MATLAB. ..................................................................... 49
Figure 14 A flow chart of the developed algorithms to create the resin
layers, resin interfaces, and 3D objects to compute the surface area
exposed (%) of the “stacked” images, sealed interfaces within the
specified ply count, and tortuosity, respectively. ............................................... 53
Figure 15 Examples of the 3D objects coded in MATLAB to represent a
laminate created with discontinuous resin patterns. Each object was
8 plies, quasi-isotropic, and 30% single layer surface area exposed. ............... 54
Figure 16 Surface area exposed (%) for a single resin layer for each of
the studied patterns [(a) stripes, (b) islands, (c) grid] with the specified
range of resin and dry space distances of 1 to 20 pixels. Diagrams of the
definition of the variables “Resin Distance” and “Dry Space Distance”
are below each graph. ........................................................................................ 56
x
Figure 17 Selected results of the projected surface area exposed (%)
[(a) Quasi-Isotropic (b) Stripes] as ply count increases from the code
developed. ........................................................................................................... 57
Figure 18 Selected results of sealed interfaces (%) [(a) [0/0]n
(b) Stripes] and tortuosity [(c) [0/90]n (d) Stripes] computations from
the code developed for 16 ply laminates. ........................................................... 58
Figure 19 (a) Summary of the sealed interfaces (%) and tortuosity
values for each of the 8 ply prepreg samples studied for permeability. (b)
Through-thickness permeability values (blue dots) of the tested prepregs
and a planar fit (multi-colored plane) via a linear regression model
with error bars indicated with red lines. .......................................................... 68
Figure 20 (a) The measured internal porosity of each of the 8 ply
prototype prepregs. (b-d) Cross-sections of the laminates made from
each of the 8 ply prototype prepregs. ................................................................. 73
Figure 21 These images show the semi-preg in the initial state (i.e., at
room temperature before cure). Both prepregs were dewetted for 1 min
30 s at 104 °C and subsequently pressed onto the unidirectional fiber
bed using a hydraulic press. (a) Image of the surface of the semi-preg
that was fabricated using a spike roller. (b) Image of the surface of the
semi-preg that was fabricated using a box cutter. ............................................ 83
Figure 22 Example images of the semi-preg surface on the transparent
tool plate during the cure cycle at (a) 75 °C, (b) 95 °C, and (c) 121 °C.
The last image shows that the surface opening was still not fully
infiltrated by the time the resin underwent gelation. ....................................... 85
Figure 23 The experimental data measured by DSC for the four
dynamic temperatures and the predictions obtained with the cure
kinetics model. ................................................................................................... 89
Figure 24 The predicted and measured viscosity evolution with time
and temperature. ............................................................................................... 92
Figure 25 (a) Diagram of the hexagonal arrangement of fibers and
labels for the “through-thickness” and “in-plane” directions.
(b) Diagram of the hexagonal arrangement of fibers and a description of
the average fiber radius, rf. (c) Micrograph of the hexagonal
arrangement of fibers......................................................................................... 95
Figure 26 The resulting resin flow distances, both experimentally and
from the model predictions, for favorable cure conditions (uniform ramp
xi
rate, accurate dwell temperature, full vacuum, and no aging of the
resin) at the three different ramp rates: 0.5, 1.0, and 2.0 °C/min. .................. 96
Figure 27 The resulting resin flow distances, both experimentally and
from the model predictions, for: (a) non-uniform ramp rate at 0.75, 1.5,
and 3.0 °C/min, (b) inaccurate dwell temperature at 111, 121, and
131 °C, (c) poor vacuum at 80%, 90% and 100% vacuum, and (d) resin
that has undergone aging where the degree of cure was 0, 0.15, or 0.30. ........ 98
Figure 28 Images of semi-preg made from each of the chosen
diameters: (a) 0.5 mm, (b) 2.0 mm, and (c) 4.0 mm. ...................................... 103
Figure 29 The time to evacuate 99.99% of air in 4-, 8-, and 16-ply
laminates for surface opening diameters between 0.5 – 4.0 mm. ................... 105
Figure 30 Images of the surface defects and internal void content of the
prototype semi-pregs fabricated using the following cases: (a) Favorable
Cure Conditions, (b) Aged Resin, and (c) Adjusted Aged Resin. .................... 108
Figure 31 Micrographs prior to cure of the discontinuous resin
distribution (fabricated via polymer film dewetting) on each of the fiber
bed types evaluated. ......................................................................................... 117
Figure 32 (a) A custom test fixture, which allowed a 60° concave and a
60° convex corner laminate (with a rounded radius of 9.5 mm) to be
manufactured simultaneously. (b) Locations of measurement along the
flanges and corners to calculate the coefficient of variation (CoV). ............... 120
Figure 33 Micrographs of the surface and the cross-sections of each of
the semi-preg formats evaluated. The fiber bed types are indicated as:
PW (plain weave), GF (glass fiber plain weave), 3KT (3K twill), ST
(spread tow), 5HS (5-harness satin), 6KT (6K twill), and 12KT (12K
twill). ................................................................................................................ 122
Figure 34 (a) Surface porosity and (b) internal porosity of each of the
prepreg and semi-preg formats evaluated. The fiber bed types are
indicated as in Fig 33. ..................................................................................... 125
Figure 35 The measured bulk factor of each of the prepreg and semi-
preg formats evaluated. ................................................................................... 127
Figure 36 Images of the part quality obtained from processing 5-
harness satin prepregs and semi-pregs at a concave and convex corner. ...... 128
xii
Figure 37 Micrographs of resin with various areal weights that were
cut at different distances to demonstrate the limitations of feature
dimensions for uniform patterning. ................................................................ 130
Figure 38 (a) Image of the surface of semi-preg with a DOI of zero.
(b) Image of the cross-section of a semi-preg with a DOI of zero.
(c) Image of surface of semi-preg that underwent elevated temperature
and pressure to increase the DOI. (d) Image of cross-section of semi-preg
with an increased DOI. .................................................................................... 133
Figure 39 Images of the flow of resin for various pattern types and
pattern placement with respect to the fiber bed during the cure cycle. .......... 136
Figure 40 Image of bismaleimide (BMI), cyanate ester, and a
commercial epoxy (CYCOM-5320-1) films that were created into
discontinuous distributions by initiating nucleation sites by a spike
roller and subsequently heating the resins at 104 °C. .................................... 139
xiii
TABLE OF TABLES
Table 1 Statistical data (average, maximum, and minimum) for the
permeability tests run for the Control Prepreg, Prepreg 104-30, Prepreg
104-120, and dry fibers at 1, 2, 4, and 8 plies. ................................................. 28
Table 2 List of the stacking sequences at 4, 8, and 16 plies implemented
into the code. ...................................................................................................... 50
Table 3 List of input and output variables of the developed code. .................. 52
Table 4 List of input and output variables of the developed code. .................. 61
Table 5 Summary of the n-way ANOVA results for (a) the Projected
Surface Area Exposed (%), (b) Sealed Interfaces (%), and (c) Tortuosity
calculations. ....................................................................................................... 63
Table 6 The constants for the resin cure kinetics model that
characterizes the resin system PMT-F4A. ......................................................... 90
Table 7 The constants for the rheological behavior model that
characterizes the resin system PMT F4A. ......................................................... 92
Table 8 The average and standard deviation of the effective
permeability, K, of laminates made with 4, 8, and 16 plies of semi-preg
with surface opening diameters of 0.5, 2.0, and 4.0 mm. ............................... 104
xiv
ABSTRACT
For aerospace composites manufacturing, prepregs processed via out-
of-autoclave (OoA) vacuum-bag-only (VBO) methods offer a viable alternative
to traditional autoclave manufacturing methods, which is costly and limiting.
However, OoA/VBO processing is limited to 0.1 MPa (1 atm) of consolidation
pressure, which is often insufficient to collapse porosity to acceptable levels.
Past work has shown that discontinuous resin films increase the capacity for
air evacuation in the z-direction (transverse) by creating additional egress
pathways and can virtually eliminate porosity caused by entrapped gases and
low-pressure VBO consolidation. The main objective of this work was to
determine a method to create high through-thickness permeability prepreg
for all fiber bed architectures, to develop methods to evaluate optimal
discontinuous resin patterns (i.e., maximize gas transport while ensuring
complete infiltration), and to understand the limitations of its applications in
various situations.
In Chapter I, polymer film dewetting on a substrate (independent of
fiber bed architecture) was explored, developed, and demonstrated as a
method to produce out-of-autoclave, vacuum bag-only (OoA/VBO) prepregs
with high transverse permeability and process robustness. The dimensions of
the surface openings created by dewetting were measured, and the percent
surface area exposed was calculated. Prepregs were fabricated with
continuous and dewetted (discontinuous) films to produce trial laminates.
xv
The laminates were cured under both standard and sub-optimal conditions,
and were characterized before, during, and after cure. Laminates fabricated
with dewetted resin consistently achieved near-zero porosity. In contrast,
laminates with continuous film displayed high levels of porosity, particularly
during sub-optimal cure. The findings demonstrate that dewetting can be
used effectively to produce OoA prepregs with high through-thickness
permeability, which can yield porosity-free laminates via VBO processing.
Furthermore, these results elucidate aspects of resin dewetting that are
critical in the creation of robust OoA prepregs.
In Chapter II, a geometric model was developed to guide the
fabrication of prepregs with various discontinuous patterns and laminates
with different orientations and ply counts. The model was used to evaluate
metrics related to gas transport: projected surface area exposed, sealed
interfaces, and tortuosity. Statistical analysis revealed that single layer
surface area exposed and ply count had the greatest effect on projected
surface area exposed; orientation had the greatest effect on sealed interfaces
and tortuosity. From these insights, prototype prepregs were fabricated to
measure through-thickness permeability. Prepregs with a large percentage of
sealed interfaces and high tortuosity exhibited lower permeability. The study
demonstrated a methodology to differentiate/screen patterns for gas
transport efficiency.
xvi
In Chapter III, a flow front model was developed based on resin cure
kinetics and rheological behavior, and then determined maximum dry space
dimensions for semi-pregs under a range of realistic manufacturing
conditions. Model predictions were validated in situ. Under controlled
laboratory cure conditions, small surface openings (≤3.7 mm) resulted in full
resin infiltration. Under adverse conditions (resin with accrued out-time), the
maximum opening size dropped 40% (to ≤2.2 mm). Using a mathematical
model, air evacuation time was calculated for various feature sizes using
permeability measurements. Model predictions were tested and verified via
fabrication of laminates. This methodology can be applied to other resin
systems to guide vacuum-bag-only prepreg design and support robust
production of composites.
Finally, in Chapter IV, semi-pregs were fabricated using fiber beds of
various weaves, fiber types, and areal weights (200-670 GSM). Flat and
curved laminates were produced and characterized, confirming that porosity-
free parts can be manufactured from a range of constituent materials using
VBO semi-pregs. Contoured laminates were produced with negligible
porosity, although a slight increase in bulk factor of prepreg plies was
observed (Δ~0.1). In addition, design considerations and limitations for the
fabrication of semi-pregs were presented. The findings demonstrate that
polymer film dewetting can be used effectively to produce semi-pregs that
xvii
yield porosity-free laminates via VBO processing, imparting robustness to
out-of-autoclave cure of prepreg laminates.
1
INTRODUCTION
A composite material is a material made from two or more constituent
materials with significantly different physical or chemical properties that,
when combined, produce a material with characteristics different from the
individual components. Fiber-reinforced polymer (FRP) is a composite
material made of a polymer matrix reinforced with fibers. Carbon fiber
reinforced polymer (CRFP) is an extremely strong and light fiber-reinforced
polymer, which contains carbon fibers. CFRPs are commonly used wherever
high strength-to-weight ratio and stiffness are required, such as in the
aerospace industry.
The manufacture of high-performance components from advanced
composites often requires autoclave processing. Industrial autoclaves are
pressure vessels used to process parts and materials, which require exposure
to elevated pressure and temperature. Typical aerospace-grade composites
produce a void content of about 1%. If void content is high, then mechanical
properties are poor. For example, an increase in void content from 1% to 3%
can result in a 20% loss in laminate mechanical properties.
Although the autoclave is the primary method for aerospace
component manufacturing, this production method has several drawbacks.
Autoclaves are highly capital-intensive, expensive to operate (i.e. energy
costs), and size limiting. Labor intensity and slow cycle times severely limit
production volume. In addition, the projected growth of aerospace composites
2
could outpace the available autoclave capacity and, at the same time, fail to
provide economic justification to ramp up the number of large autoclaves in
operation.
Out of autoclave (OoA) composite manufacturing is an alternative to
the traditional high-pressure autoclave curing process. OoA aims to achieve
the same quality as an autoclave but through a different method, specifically
vacuum bag only (VBO) method. Prepreg is a fibrous material pre-
impregnated with a particular synthetic resin. VBO prepreg processing
entails consolidating prepreg laminates within a standard industrial oven.
The use of VBO for secondary structures, such as flaps and fairings, is well
established. However, material with less than 1% void content and autoclave-
quality mechanical properties for primary parts, such as wings and fuselages,
has not been robustly delivered. Because the applied consolidation pressure
is limited to 0.1 MPa (1 atm), the prepreg resin pressure can be insufficient to
collapse bubbles that arise from entrapped air and/or evolved gases, which
thus can remain in place during gelation and leads to porosity.
A potential solution is to produce OoA prepregs with discontinuous
resin distributions. Previous works [1–3] demonstrated that in-plane
discontinuity of resin (and not a specific prepreg embodiment) is the prepreg
attribute that alone can nearly eliminate porosity in VBO-cured parts. This
topic is scantily explored in literature. In Chapter I, a potentially scalable
technique for producing OoA prepregs with discontinuous resin distribution
3
from any fiber bed is introduced. The technique can be readily adjusted to
produce a wide variety of complex patterns, a potentially fertile area of
exploration. In Chapter II, a methodology that allows for the screening (i.e.
evaluation and differentiation) of discontinuous resin patterns for OoA
prepregs based on efficient air evacuation was outlined. In Chapter III, a
strategy was outlined to determine dimensional guidelines for discontinuous
resin patterns required to impart robustness to Out-of-Autoclave/Vacuum-
Bag-Only processing of composite prepregs. Both of the methodologies
developed in Chapters II and III can be used to guide semi-preg development.
Finally, in Chapter IV, the application of discontinuous resin on different
fibers beds was evaluated for use in commonly encountered manufacturing
conditions for out-of-autoclave (OoA)/vacuum-bag-only (VBO) processing of
composite prepregs, and design considerations were determined for the
fabrication of such prepregs. This work combined can be used to guide
vacuum-bag-only prepreg design and support robust production of
composites.
4
CHAPTER I:
Polymer Film Dewetting for Fabrication of
Out-Of-Autoclave Prepreg with High
Through-Thickness Permeability
1. Background
Aerospace manufacturers seek to reduce costs of traditional composites
manufacturing methods by producing autoclave-quality composite structures
using out-of-autoclave/vacuum bag-only (OoA/VBO) methods [4,5]. The desire
to shift away from autoclave cure is motivated by high acquisition and
operation costs of autoclaves, resource-intensiveness, and throughput
limitations, which can constitute a production bottleneck. Currently, VBO
prepreg processing can match the part quality of autoclave cure, but only in
optimal situations, while sub-optimal manufacturing conditions lead to
greater defect levels, particularly porosity, and potentially degrade
mechanical performance [6]. The major sources of porosity during prepreg
cure are low process pressures (intrinsic to VBO cure) and incomplete
evacuation of entrapped gases during processing.
VBO prepreg processing entails consolidating prepreg laminates
within a standard industrial oven. Because the applied consolidation
pressure is limited to 0.1 MPa (1 atm), the prepreg resin pressure can be
insufficient to collapse bubbles that arise from entrapped air and/or evolved
gases, which thus can remain in place during gelation and lead to porosity.
5
The presence of entrapped gases in cured parts can lead to porosity levels as
high as 3–5%, resulting in unacceptable mechanical properties and part
rejection [7].
Technological advancements have been implemented to reduce
porosity during VBO prepreg cure. Conventional VBO prepregs are pre-
impregnated using a hot-melt approach, during which resin is formed into a
thin film on backing paper (without the use of solvents) and pressed onto the
fiber bed using rollers [7,8]. Hot-melt prepregging enables partial through-
thickness impregnation of the fiber bed, which can be used to create dry,
high-permeability gas evacuation pathways within the plane of the prepreg
ply. These pathways allow rapid in-plane gas transport and, when combined
with appropriate consumables (such as permeable edge-breathing dams),
reduce void content within cured parts [9].
Despite these technological advancements, laminates produced by
VBO prepreg processing cannot match the quality of autoclave processing in
all manufacturing conditions [6]. Autoclave pressure effectively suppresses
porosity caused by entrapped air, insufficient resin flow, and evolved gases,
while VBO processing is inherently more susceptible to these problems.
Process deficiencies (such as inadequate vacuum, insufficient air evacuation,
and/or high humidity) generate voids in laminates fabricated with VBO
prepregs. For example, VBO prepreg laminates are susceptible to micro-void
formation if the material and thermal conditions shift the resin viscosity from
6
the designed range [10]. These shifts can occur because of excessive out-time,
which advances the initial degree of cure and increases moisture uptake, both
of which alter the viscosity profile of the resin. Moisture absorbed from
ambient humidity also will increase porosity in VBO-processed laminates,
but not during autoclave cure [11]. Kardos developed a diffusion-based void
growth model to explain this trend [12]. The gas pressure within the
moisture-induced void can exceed the maximum resin pressure attainable
with VBO cure at 120 °C. Such phenomena also can arise during autoclave
processing, although autoclave pressures are typically sufficient to ensure
adequate resin flow and to suppress moisture-based void formation.
Grunenfelder et al. showed that increasing air evacuation in the
z-direction (transverse) virtually eliminated porosity caused by entrapped air
or moisture in OoA/VBO manufacturing [1,13]. To increase the through-
thickness permeability, the resin was distributed onto the fiber bed in
discontinuous strips, creating gaps that connect to the internal evacuation
channels to form an interconnected, three-dimensional network. The prepreg
(USCpreg [14]) was produced using a custom prepregging line to transfer the
resin pattern onto the fabric using heated rollers. Using the USCpreg format,
cured parts were produced with near-zero internal porosity and no surface
defects even at non-ideal manufacturing conditions. The major limitation of
the direct coating method was that the resin distribution depended entirely
7
on the surface topology of the fabric, with resin transferring to the raised tow
overlap sections but leaving recessed regions dry.
Tavares et al. [3] studied selectively impregnated prepregs called
“semipregs” to assess the effectiveness of high-permeability prepregs for co-
cure of honeycomb sandwich structures. Through-thickness air permeability
was measured at room temperature and during the cure cycle for both a
commercial semipreg (“Zpreg”) and an equivalent unidirectional prepreg
(constructed with continuous resin). Results showed that the permeability of
the semipreg was three orders of magnitude greater than the continuous film
prepreg before and during the cure cycle, owing to a network of dry
interconnected pore spaces. However, the large characteristic size of the dry
areas within the semipreg necessitated use of a low-viscosity resin, resulting
in unwanted resin bleed and high defect levels within the facesheets.
Furthermore, the resin distribution within the semipreg consisted of linear
strips, and the work did not describe a technique to create discontinuous
patterns of arbitrary shape and size.
Roman reported a method to fabricate high through-thickness
permeability prepreg [2]. In this method, resin was applied to the fiber bed
using hot-melt coating. Next, the backing paper covering the resin layer was
replaced with a polyester film. Covered with the film, the prepreg underwent
a heat cycle, during which the resin flowed away from fabric warps. As a
result, openings were created in the film, allowing gas removal in the
8
through-thickness direction. The resultant cured composite parts using this
method exhibited near-zero internal porosity.
The methods of Roman rely on the phenomenon of fluid “dewetting,”
the rupture of a thin liquid film on a solid substrate and the subsequent
formation of islands or droplets. The reverse phenomenon is “wetting,” in
which a liquid spreads and increases contact area with a solid surface.
Wetting/dewetting is a result of intermolecular forces between the liquid film,
solid surface, and the surrounding gas. The force balance between the
adhesive and cohesive forces between these surfaces dictates the degree of
wettability [15]. A useful parameter for gauging the wetting of a system is
the spreading coefficient, S, which relates the interfacial tensions between a
solid s, liquid l, and gas g denoted by γij with i, j = s, l, g:
𝑆 = 𝛾 𝑠𝑔
− ( 𝛾 𝑠𝑙
− 𝛾 𝑙𝑔
) (1)
𝑆 > 0, 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 𝑤𝑒𝑡𝑡𝑖𝑛𝑔 𝑜𝑐𝑐𝑢𝑟𝑠 (2)
𝑆 < 0, 𝑝𝑎𝑟𝑡𝑖𝑎𝑙 𝑤𝑒𝑡𝑡𝑖𝑛𝑔 𝑜𝑐𝑐𝑢𝑟𝑠 (3)
The surface energies are calculated from the Young Equation, a force balance
along the solid-liquid interface [16]:
𝛾 𝑠𝑔
= 𝛾 𝑠𝑙
+ 𝛾 𝑙𝑔
cos 𝜃 (4)
where θ is the contact angle of the liquid droplet. However, polymer films may
not dewet even if S < 0, because the film may be in a metastable state, i.e.,
below the glass transition temperature [17].
9
Kheghsi identified three stages of dewetting when forming new dry
patches and the locations on which dry patches nucleate [18]. First, the liquid
film must form on the solid substrate. Second, some disturbance must reduce
the liquid film thickness to near zero. Approximate solutions to predict
thickness prior to rupture were solved using the Navier-Stokes system
augmented with conjoining force [18]. Finally, the liquid film ruptures and
dry patches nucleate. As the dry patches grow, fluid material accumulates in
a rim surrounding the growing hole. Dry patch formation will occur at a pre-
existing patch or edge, or at a film-thinning disturbance caused by
evaporation, drainage due to gravity (i.e., sharp/rough surface), and/or
surface tension gradients.
The methods for producing prepregs with high through-thickness
permeability described above are not suitable for unidirectional fiber beds,
which are flat. The direct coating method described by Grunenfelder et al.
requires the warps of twill fabric to pool resin. The second method (Roman)
relies on the space created by overlapping tows as dewetting sites.
Unidirectional (UD) fiber beds comprise the majority of the aerospace
composites market. Therefore, a method is required to create high through-
thickness permeability prepreg for all fiber bed architectures.
In this work, we describe a simple technique to create discontinuous
resin patterns via polymer film dewetting on a substrate (versus being
dependent on the fiber bed architecture). This technique can be used to create
10
finely tuned patterns of various shapes and sizes. In addition, the patterned
resin films created with this technique can be applied to any fiber bed,
regardless of architecture. The technique can be used for lab-scale studies to
explore and determine the limitations of prepregs with discontinuous resin
patterns, and is potentially compatible with commercial processes used to
produce prepregs. Here, we use the technique to produce UD prepreg with
high through-thickness permeability (UdSCpreg). In this process, resin film is
first perforated on backing paper (the substrate) to create an array of
nucleation sites, after which the film is heated to cause the film to recede
from the nucleation sites. Finally, the dewetted resin is transferred onto UD
tape by briefly pressing. In comparison to conventional OoA prepreg formats
that feature continuous resin films, prepreg produced using this technique
exhibited nearly void-free cured laminates, even under challenging cure
conditions.
2. Materials
A UD carbon fiber bed (Fibre Glast Development Corporation, Ohio,
USA) and a toughened epoxy resin (PMT-F4, Patz Materials & Technology,
California, USA) were selected for the experiments. The epoxy resin was
designed for vacuum bag curing and featured medium-to-dry tack. The fabric
weight was 305 g per square meter (gsm) and the thickness was 0.36 mm. A
binder of polyester fill threads stitched in one direction held the UD fibers in
place. The tape exhibited negligible crimp, except around the binder. The
11
resin was 76 gsm, yielding prepreg with a resin content of 33% by weight.
The standard cure cycle included a ramp of 1.5 °C per min followed by a dwell
at 121 °C for two hours. The resin storage life was two years at −10 °C, and
the out-life was 120 days at room temperature.
Because of the resin thickness, the polymer dewetting process was not
spontaneous when heated. To facilitate dewetting, nucleation sites were
introduced using a hand-held spike roller (HR-2, Robert A. Main & Sons, Inc.,
New Jersey, USA). The spike roller pins were spaced at 6.35 mm, and the
roller was passed over the entire film in straight passes. The dewetting
process was carried out on silicone-coated release paper, which provided a
low-energy surface [19], which was required for dewetting. A standard oven
(Blue M Oven, Thermal Product Solutions, Pennsylvania, USA) was used to
heat the films for dewetting.
A toughened 76 gsm epoxy resin (PMT-F4, Patz Materials &
Technology, California, USA) was selected for the experiments. The epoxy
resin was designed for vacuum bag curing and featured medium-to-dry tack.
The standard cure cycle included a ramp of 1.5°C per min followed by a dwell
at 121°C for two hours. The resin storage life was two years at -10°C and the
out-life was 120 days at room temperature.
Because of the resin thickness, the polymer dewetting process was not
spontaneous when heated. To facilitate dewetting, nucleation sites were
introduced using a hand-held spike roller (HR-2, Robert A. Main & Sons, Inc.,
12
New Jersey, USA). The spike roller pins were spaced at 6.35 mm, and the
roller was passed over the entire film in straight passes. The dewetting
process was carried out on silicone-coated release paper, which provided a
low energy surface [19] which was required for dewetting. A standard oven
(Blue M Oven, Thermal Product Solutions, Pennsylvania, USA) was used to
heat the films for dewetting.
3. Film dewetting
3.1 Surface opening dimensions
The dewetting process was performed at three temperatures – 89 °C,
104 °C, and 119 °C – that were selected based on prior work and resin
kinetics. For example, Roman created surface openings in a similar epoxy
system at 104 °C [2], midway between the temperatures of 119 °C (+15 °C)
and 89 °C (−15 °C) chosen here. All three temperatures were below the
specified curing temperature of the resin (121 °C), reducing the risk of
changes in resin rheological properties due to dewetting.
The times used for dewetting spanned 15 s to 8 min, based on
preliminary trials. This time range allowed for substantial variations of
surface opening sizes at each sampling time and allowed surface openings to
reach maximum dimensions at each temperature.
The resin was thawed and cut to samples 75 × 150 mm on silicone-
coated release paper. Nucleation sites were introduced into the films by spike
rolling. The spike roller was passed over the entire surface without
13
overlapping passes to achieve uniform spacing of nucleation sites.
Subsequently, the sample was placed in an oven pre-heated to one of the
specified nominal temperatures, and held for times of 15 s to 8 min. The
sample was then removed from the oven and allowed to cool to room
temperature.
Figure 1 Top view images of resin dewetted at 104 °C either still on the substrate (silicone-
coated backing paper) or pressed onto unidirectional carbon fiber tape - (a) dewetted for 15 s
on the substrate (b) dewetted for 4 min on the substrate (c) dewetted for 30 s pressed onto fiber
bed (d) dewetted for 4 min pressed onto fiber bed.
Edwards et al. [20] reported the dewetting behavior of a
dielectrophoresis-induced film as it formed into a single equilibrium droplet.
Visual observations revealed that a rim formed, which receded at a constant
speed and constant dynamic contact angle. This event was followed by
relaxation into a spherical cap shape. The dewetting system presented here
matches to the description of these two regimes. Fig. 1a and 1b show the
14
dewetted resin on the substrate (silicone-coated backing paper) after 15 s and
4 min. At 15 s, a rim had developed with a width of about 0.3 mm. After
4 min, the rims of the adjacent holes had converged. The dewetted resin was
then pressed onto UD carbon fiber tape using an unheated hydraulic press.
Each sample was examined using a microscope (VHX-5000, Keyence
Corporation of America) over an area of 40 × 40 mm. Images of resin dewetted
at 104 °C for 30 s and 2 min are shown in Fig. 1c and 1d. The images are
representative of the surface opening sizes and shapes produced from the
dewetting process. The images were converted to black pixels for fibers and
white pixels for resin using image analysis software (ImageJ). The surface
area exposed by dewetting was calculated from the numbers of black and
white pixels:
𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝐴𝑟𝑒𝑎 𝐸𝑥𝑝𝑜𝑠𝑒𝑑 ( %) =
𝑝 𝑏𝑙𝑎𝑐𝑘 𝑝 𝑤 ℎ𝑖𝑡𝑒 + 𝑝 𝑏𝑙𝑎𝑐𝑘 𝑥 100 %
(5)
where p is the number of pixels. A greater percentage of exposed surface area
indicates that a larger portion of the prepreg ply surface will consist of
exposed dry fibers, rendering the prepreg far more permeable to gas than
continuous resin films during consolidation.
The geometric features of the patterns were analyzed to quantify
characteristics relevant to through-thickness permeability and impregnation.
For example, Fig. 2a shows the percent surface area exposed after 15 s to
8 min of dewetting at 89 °C, 104 °C, and 119 °C. At all three temperatures,
the percent of surface area exposed sharply increased in the first 1–2 min,
15
and then slowly increased with increasing time. The maximum exposure
achieved across all three temperatures was 65% at 119 °C for 8 min. Fig. 2b
shows the average diameter of the openings in each sample, where half this
amount, or the radius, represents the distance resin must flow in-plane to
fully wet the fiber bed. This flow length must be compatible with the resin
cure kinetics, viscosity, and gel time to prevent incomplete saturation and
flow-induced porosity. Second, the dimensions of the dry fiber regions will
govern the prepreg impregnation time, with larger-diameter openings leading
to longer times for air evacuation. Finally, opening size is also related to the
likelihood of creating a continuous, interconnected, three-dimensional
network of dry fibers within laminates formed of stacked prepreg plies. For
example, openings larger than half the distance between the initial spiked
nucleation sites (3.2 mm, in this case) are less likely to be sealed off during
lay-up than openings with smaller dimensions.
Figure 2 Dimensions of surface openings created by the dewetting process – (a) surface area
exposed (b) average diameter of surface openings (where half this amount, or the radius, is the
longest distance to wet-out during cure).
16
The average diameter of the surface openings follows the same trend
as the values of exposed surface area - the diameter increases sharply within
the first 1–2 min, then slowly increases with time. The maximum diameter
achieved across all three temperatures was 7.6 mm at 119 °C for 8 min. This
distance is larger than the space between the initial nucleation sites created
by the spike roller because many of the resin strands at the edges of the
growing opening were breaking at this point. Edwards [20] developed a
hydrodynamic model describing the rate of dewetting of a droplet. However,
the model was developed for a different geometry (spherical cap) than the
studies presented here (holes). Using the empirical data from this section (on
surface opening sizes), the rate of dewetting, dR/dt, was measured to be
1.7 mm/min (or 2.9 × 10
−5
m/s) at 104 °C during the first regime. This rate can
be used to calculate the Capillary Number, Ca, a description of the
interaction between the resin and the substrate:
𝐶𝑎 =
𝜇 γ
|
𝑑𝑅 𝑑𝑡 |
(6)
where μ is the dynamic viscosity of the epoxy at 104 °C (6.088 Pa * s), γ is the
interfacial tension of the epoxy and the substrate (on the order of 10
−2
N/m)
[21–23], and dR/dt is the rate of change of the radius of the surface openings
(on the order of 10
−5
m/s). Here, the Capillary Number was determined to be
on the order of 10
−3
, indicating that the dewetting process is dominated by
the effects of surface tension rather than by viscous forces. Thus, surface
17
tension forces will tend to minimize the surface area between the resin and
the substrate.
Figure 3 Diagram of 4-ply stacked prepreg demonstrating the interconnectedness of
z-direction air evacuation networks – (a) continuous resin film [with in-plane air evacuation
only], (b) dewetted resin film with small openings, (c) dewetted resin film with large openings.
The dewetted resin pattern can potentially affect the interconnectivity
and tortuosity of the dry pore network. Prepregs fabricated with continuous
resin film do not feature breathe-out pathways in the through-thickness
direction (Fig. 3a). However, the partial impregnation of the resin into the
fibers allows for air evacuation at the edges of the prepreg. Applying a
discontinuous resin pattern, on the other hand, creates additional pathways
in the through-thickness direction for gases to evacuate. Due to random
placement of plies during lay-up, smaller openings may not overlap with
openings in adjacent plies (Fig. 3b). However, larger openings are more likely
to overlap with another opening (Fig. 3c), resulting in an interconnected dry
18
fiber network of pathways for gas egress. The formation of a highly connected
network (i.e., breathing pathways to the surface) will promote rapid air
removal.
Based on the results of dewetting experiments, dewetting conditions of
30 s and 2 min at 104 °C were selected for subsequent tests. At 30 s of
dewetting, the surface area exposed was 13%, and the average diameter of
the openings was 1.7 mm. This condition represents a case in which full
saturation of the fiber bed is likely, but openings may be sealed off during
lay-up and/or prematurely during cure. After dewetting for 2 min, the
exposed surface area was 50%, and the average diameter of the openings was
5.6 mm. This condition may result in incomplete wet-out of the fiber bed
because of the large openings. However, a 3D network of interconnected dry
areas is likely to form, facilitating air evacuation during the cure cycle.
3.2 Rheological behavior
Dewetting involves briefly heating resin film below the prescribed cure
temperature, and this heat exposure introduces a risk of advancing the
degree of cure. To assess the extent of partial curing, the viscosity profiles of
three resin films – untreated, dewetted for 30 s at 104 °C, and 2 min at 104 °C
– were measured over the recommended cure cycle using a rheometer (TA
Instruments, Delaware, USA). A comparison of these three viscosity profiles
against the cure cycle is presented in Fig. 4. All of the viscosity profiles
decrease to a minimum after 57 min of the temperature ramp (1.5 °C/min to
19
121 °C), which are then followed by sharp increases in less than 8 min to over
104 Pa * s. The curves before and after dewetting show negligible differences.
Figure 4 Viscosity profiles of resin without any dewetting treatment compared to resin
dewetted for 30 s and 2 min at 104 °C. Times t1, t2, and t3 correspond to the 1 h room
temperature hold, the cure temperature ramp to 121 °C, and the 2 h cure dwell, respectively.
The minimum viscosity and flow number values were obtained for the
two dewetted resin samples and the untreated resin. These values were used
to evaluate the effect of dewetting on rheological behavior. The flow number
is defined as the integral sum of the inverse viscosity curve prior to gelation
[24], and provides a measure of resin fluidity:
𝐹𝑙𝑜𝑤 𝑁𝑢𝑚𝑏𝑒𝑟 = ∫
1
𝜇 ( 𝑡 )
𝑡 𝑔𝑒𝑙 𝑡 0
𝑑𝑡
(7)
The minimum viscosity of the untreated resin was 5.6 Pa * s, while for
both dewetted samples the value was slightly less − 5.3 Pa * s. The flow
number for the untreated resin was 210.4 Pa
−1
, while the flow number for
both dewetted samples was 212.4 Pa
−1
. These trends are counterintuitive
20
because high-temperature exposure during dewetting should cause
polymerization/cross-linking, which increases viscosity. However, the
difference (0.3 Pa * s in viscosity, 2 Pa
−1
in flow number) is negligible, and can
be attributed to measurement variability.
4. Prepreg fabrication and characterization
Two categories of prepreg with identical resin content (33 wt%) were
fabricated. “Control Prepreg” was produced by sandwiching UD carbon fiber
tape between two 76 gsm continuous resin films, which replicated the
architecture of conventional VBO prepregs. Conversely, prepreg with
discontinuous resin distribution was fabricated by combining UD tape with
identically dewetted resin sheets. The resin in this category was dewetted at
104 °C for either 30 s or 2 min and, for the remainder of the paper, is
designated as “Prepreg 104-30” and “Prepreg 104-120,” respectively. The
names designate the temperature and the time (in s) at which the resin film
was dewetted. For both categories, resin was applied to the fiber bed by
aligning and pressing constituents in an unheated hydraulic press
(G30H-18-BCX, Wabash MPI).
4.1 Ammonia curing
Polished sections of uncured prepreg were prepared by exposure to
ammonia vapor prior to polishing. Ammonia slowly cures epoxy resin at room
temperature, fixing the resin in place with negligible flow [25]. This
“ammonia curing” enables preparation of polished sections for evaluation of
21
the structure of the prepreg stack before cure. The resin distribution in as-
fabricated prepregs was examined to assess (1) the accumulation of resin due
to the dewetting process and (2) the breathe-out pathways between layers for
gas evacuation. Accumulated resin may lead to fiber bed deformation,
primarily manifested as waviness due to the fibers conforming around the
resin. Excessive waviness can lead to greater air entrapment and can also
induce non-uniformity in laminate thickness, fiber volume fraction, or shape
after cure, all of which adversely affect mechanical properties of the laminate.
To prepare samples for ammonia curing, resin and fiber tape pieces
were cut to 75 × 75 mm squares. The resin - either a continuous film or a
discontinuous pattern - was pressed on both sides of each fiber tape sample
using a hydraulic press. Plies were then stacked symmetrically along the
midplane in a [0/90]4s sequence, and each prepreg stack consisted of 16 plies.
Next, the prepreg stacks were suspended for ten days above an ammonia
bath inside a sealed glass vessel to slowly cure the epoxy [25].
Next, sections were cut from the ammonia-cured prepreg and polished
for imaging. Images of polished sections are shown in Fig. 5. Fibers parallel
to the surface [labeled “Fibers (0°)” in Fig. 5a] were unsupported and thus
susceptible to tear-out during polishing, causing affected regions to appear
black in the images. In addition, the fibers normal to the surface also tore out
in places because of partial impregnation and insufficient support from the
matrix. These areas appeared as dark regions within the fiber tows.
22
Figure 5 Laminates cured using the ammonia bath method to visualize the fiber and resin
distributions before oven curing – (a) Control Prepreg, (b) Prepreg 104-30, (c) Prepreg 104-120.
The prepreg fabrication protocols yielded distinctly different resin
distributions. For example, the Control Prepreg exhibited evenly distributed
resin ∼130 μm thick (Fig. 5a). Prepreg 104-30 showed a resin structure
similar to that of the continuous film (Fig. 5b). The resin was evenly
distributed with a thickness of ∼130 μm. In this case, there was also no clear
evidence of accumulated resin due to the dewetting process. However, for
Prepreg 104-120, resin accumulation was obvious (Fig. 5c). Some areas were
resin-free, while others showed a resin thickness up to 280 μm. Resin
accumulations in the prepreg may result in resin-rich regions in the cured
laminate. Additionally, all three samples yielded minimal resin
impregnation.
No fiber deformation was visible in laminates produced using Control
Prepreg or Prepreg 104-30 (Fig. 5a and 5b). However, Prepreg 104-120
exhibited waviness in the fiber bed due to resin accumulation (Fig. 5c), where
the average angle of fiber bed misalignment was 7.3° ± 2.3°. The average
amplitude of the agglomerates was 0.22 ± 0.05 mm and the average
23
wavelength was 2.45 ± 0.84 mm. This waviness can persist and propagate to
cured parts, compromising mechanical properties.
4.1 X-ray micro-CT
X-ray microtomography (micro-CT) yields 3D images and data that can
be used to visualize prepreg microstructures, including entrapped gas
bubbles and air evacuation pathways [26]. The ammonia-cured Control
Prepreg and Prepreg 104-120 were imaged by micro-CT. The samples were
cut using a diamond saw into 25 × 50 mm rectangles. Scans were performed
using an industrial CT system (GE Phoenix Nanotom M) with voxel size
14 µm, source voltage 70 kV, source current 180 mA, and exposure 500 ms.
The resulting series of parallel tomographic slices was used to visualize
microstructural features and perform quantitative measurements.
The 2D and 3D images generated are presented in Fig. 6. The darker
regions (black) represent void space, while the lighter regions (white-gray)
represent solid space (fibers and resin). The 2D images at left show
orthogonal cross-sections, with fiducial markers indicating the relative
position of each slice. The upper 3D image at right is a selected sub-volume
comprising 2-3 plies (of the 16 total plies), while the lower 3D image is a view
of the in-plane direction of the fibers of all 16 plies.
24
Figure 6 2D and 3D images of X-ray micro-CT scans of the ammonia-cured Control Prepreg
and Prepreg 104-120. The images represent the structure of the prepreg prior to cure and the
resultant interconnection of void space. Both prepregs exhibited intra-tow void space, whereas
only the Control Prepreg exhibited inter-ply (enclosed) void space.
Both the Control Prepreg and Prepreg 104-120 exhibit rod-like void
spaces parallel to the fiber direction (e.g., areas enclosed in blue boxes). The
Control Prepreg also exhibits large irregular void spaces (e.g., red boxes),
consistent with prior observations of air entrapped between resin-rich ply
interfaces [27]. Conversely, inter-ply air entrapment was negligible in
Prepreg 104-120. The percent volumetric void space of the Control Prepreg
and Prepreg 104-120 was 11.6% and 13.1%, respectively. The differences in
void content and morphology are attributable to the discontinuous resin
distribution in Prepreg 104-20 achieved by dewetting. This prepreg features
25
less exposed resin surface area, reduces the likelihood of resin-on-resin
mating, decreases risk of bubble entrapment, and increases the effective size
of the dry pore network within multi-ply laminates. Altogether, the CT data
underscores the differences in internal microstructure between prepregs
fabricated with continuous film versus discontinuous films.
4.1 Permeability
The effective (slip-corrected) transverse permeability was measured
and compared for the Control Prepreg, Prepreg 104-30, Prepreg 104-120, and
dry fibers. The transverse permeability of the experimental prepregs is
expected to differ markedly from conventional VBO prepregs. A custom test
fixture was used for the experiments [28], following the falling pressure
method described by Tavares [29]. Plies of dry fabric or prepreg were laid
over a cavity of known dimensions supported by stacks of honeycomb core.
Measurements were recorded for 1-, 2-, and 4-ply laminates. The edges of the
plies were sealed with vacuum tape to prevent edge-breathing and allow air
evacuation only in the through-thickness direction. The laminates were
covered with perforated release film and breather cloth, and vacuum-bagged.
Vacuum was drawn in the bag to compact the laminate and create a pressure
difference between the core cavity and the bag. The evolution of pressure in
the cavity was monitored over time using a pressure transducer and data
acquisition software (LabVIEW, National Instruments), and the
26
measurements were used to estimate an effective permeability coefficient. All
tests were performed at room temperature.
To obtain an average effective permeability value, two samples
(replicates) were tested for each experimental configuration, with a minimum
of five pressure decay trials per sample. Each trial was continued to the time
at which the cavity pressure stabilized (indicating flow had ceased), and the
configuration was then re-pressurized to begin the next trial. The data from
the first trial was omitted because air evacuates more quickly when the
consumables and plies have not been previously compressed.
Utilizing Darcy’s Law, the one-dimensional laminar flow of
compressible air at isothermal and adiabatic conditions through a porous
medium [27] can be described by
−
𝐾𝐴 𝑃 𝐵𝑎𝑔 𝐿𝜇 𝑉 𝐶𝑜𝑟𝑒 𝑡 = ln [
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)+ 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )− 𝑃 𝐵𝑎𝑔 )
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)− 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )+ 𝑃 𝐵𝑎𝑔 )
]
(8)
where K is the permeability scalar in the flow direction in m
2
, A is the cross-
sectional area (1.46 × 10
−2
m
2
), PBag is the pressure at the bag side (5 × 10
3
Pa),
PCore is the pressure at the honeycomb core side in Pa, L is the lateral
dimension in m, μ is the viscosity of air at room temperature
(1.85 × 10
−5
Pa * s), t is time in s, and VCore is the volume of the core
(7.87 × 10
−4
m
3
). Here, the vacuum level is assumed to be 95% (corresponding
to an absolute vacuum bag pressure of 5 kPa). Plotting the left-hand side
versus time yields a straight-line plot, the slope of which can be used to
determine the effective air permeability of the prepreg, K.
27
Figure 7 Through-thickness permeability values for 1, 2, 4, and 8 plies of prepreg. The value
of 0% surface area exposed represents dry fabric, 100% represents continuous film, and the
values in between represent the two levels of dewetting focused on in this work. Error bars
were omitted for viewing clarity (statistical information can be found in Table 1).
A summary of the permeability values of the 1, 2, and 4 ply laminates
versus fiber surface area exposed is presented in Fig. 7. The Control Prepreg
is represented by 0% surface exposure, while dry fibers represent 100%
surface exposure. The percent surface area exposed was 13% and 50% for
Prepreg 104-30 and Prepreg 104-120, respectively.
28
Table 1 Statistical data (average, maximum, and minimum) for the permeability tests run
for the Control Prepreg, Prepreg 104-30, Prepreg 104-120, and dry fibers at 1, 2, 4, and 8
plies.
The transverse permeability of the experimental prepregs is expected
to differ markedly from conventional VBO prepregs, and experiments were
undertaken to measure this property. Resin film dewetting substantially
increases the room-temperature permeability of each laminate configuration.
For the Control Prepreg, little airflow was detectable for all ply counts,
resulting in permeability values less than 0.1 × 10
−16
m
2
. Kratz [28] reported
similar results, and lack of airflow was attributed to the inherent
topographical features of the UD tape (i.e., absence of gaps in the resin film).
29
Prepreg 104-30 yielded four- to- six-fold increases in transverse permeability
(to values of 0.5–0.7 × 10
−16
m
2
) as compared to the Control Prepreg. Prepreg
104-120 demonstrated a further increase in transverse permeability, with
values 36–52 times greater than the values obtained for the Control Prepreg
(3.7–5.3 × 10
−16
m
2
, and 30–50% of those of UD dry fibers. Tavares et al. [3]
likewise reported that the air permeability of two plies of the semipreg
format was three orders of magnitude greater than two plies of an equivalent
continuous film format. The transverse permeability decreased slightly as ply
count increased, owing to an increase in the tortuosity of the dry pore
network, or possibly to differences in compaction and dimpling between thin
(compliant) and thick (stiff) laminates (similar to Grunenfelder et al. [1]).
However, single-ply and 8-ply laminates exhibited similar trends with
respect to surface area exposed. As shown by Grunenfelder et al. [1,13],
prepreg with greater transverse permeability generally results in more rapid
air evacuation, and nearly eliminates porosity in cured laminates.
5. Laminate characterization
5.1 In-situ visualization
Insights into air removal mechanisms during VBO cure of prepregs can
be provided by use of a transparent tool plate, as described by Hu et al. In
their work, they described a technique for in situ visualization of prepreg
during cure [30,31]. Using similar methods here, 8-ply prepreg stacks were
fabricated using Prepreg 104-30 and Prepreg 104-120. The stacks were laid
30
against a glass window of an oven and vacuum-bagged with standard
consumables, including edge-breathing dams. The standard cure cycle was
used, and prepregs were observed during cure using a digital microscope
(Dino-Lite US, Dunwell Tech, Torrance, CA).
Figure 8 Screenshots of window vacuum bag curing – (a) Prepreg 104-30 (b) Prepreg
104-120. The times t1, t2, and t3 correspond to the point in the cure cycle that the laminate is
being exposed to (as indicated on Fig. 4).
Fig. 8 shows the progression of resin flow at the tool/part interface at
different times during the cure cycle. Images in the first column show the
surface openings during the room temperature vacuum hold. The second
column shows the same openings during the heating ramp. At this point, the
resin began to flow, primarily along the fiber direction. The third column
shows surface openings during the cure dwell, at which point the openings
were fully saturated with resin. However, vestigial evidence of the surface
openings remains, manifesting as regions with a different visual appearance.
31
These regions are fully saturated, but the resin film covering the fiber bed is
marginally thinner than in adjacent areas. This indicates that aspects of the
laminate microstructure were affected by the initial prepreg format (e.g.,
resulting in a different resin film thickness at the surface, or changes in fiber
arrangement).
The sequences in Fig. 8 show that in-plane resin flow occurs primarily
along the fiber direction due to the anisotropic in-plane permeability of the
aligned fiber bed. Therefore, the radius of the openings represents the longest
flow distance of the resin to complete saturation of the fiber bed. For this
fiber bed, resin system, and cure cycle, full saturation of the visible area
required 16 min more for Prepreg 104-120, which fully saturated after 41 min
after the onset of the temperature ramp, than for Prepreg 104-30, which fully
saturated after 25 min.
The images in Fig. 8 also show that in Prepreg 104-120, greater
volumes of bubbles are trapped between the surface of the glass than in
Prepreg 104-30. The greater volume of trapped air in Prepreg 104-120 can be
attributed to the larger surface area of resin in direct contact with the glass.
The tackiness of the resin combined with the natural surface texture
invariably entraps air at the tool face, while dry fibers do not. The
observation provides insight into how such prepregs will behave when laid
onto tool molds. In particular, air entrapped at the surface between the tool
mold and the prepreg can give rise to surface pits [32].
32
5.2 Surface defects and bulk porosity
Laminates were fabricated to assess levels of surface and internal
porosity. Sixteen plies were stacked symmetrically along the midplane in a
[0/90]4s sequence. Initially, each ply was cut to 150 × 150 mm, and after
stacking; the edges of the stack were trimmed, resulting in dimensions of
∼140 × 140 mm. Two cure conditions were employed. The first condition
consisted of a standard “ramp-hold” cure cycle, and standard vacuum bag
consumables were used, including edge-breathing dams created from vacuum
sealant tape wrapped in fiberglass and placed at the laminate perimeter. The
second condition was identical to the first condition except for one key
difference - the edge-breathing dams were replaced with sealant tape to
prevent in-plane air removal at the ply boundaries and allow air evacuation
only in the through-thickness direction. This case was intended to
approximate commonly encountered process conditions, which prevent or
limit in-plane air evacuation in OoA prepregs, such as large or complex parts,
parts with corners, etc.
Surface porosity and defects. Laminates produced with conventional
VBO prepreg typically show surface porosity as a result of air entrapment at
the tool face [32]. While surface porosity normally does not affect the
mechanical performance of cured parts, it is a manufacturing defect that
requires costly rework. Surface void contents were measured on the laminate
surface facing the glass tool plate during the cure cycle. Images 38 × 38 mm
33
were recorded using a digital microscope at 5 regions across the surface.
Images were analyzed using software (ImageJ), and a series of manual steps
to produce a binary porosity image. The surface porosity was calculated in
the same manner as the calculation of exposed surface area described in
Sec. I.3.1. Finally, an average surface porosity was determined from the five
images.
Figure 9 (a) Surface porosity values of laminates created with continuous resin film and
dewetted resin film under the conditions of standard cure and sealed edges. (b, c, d) Surface
images of laminates cured under the sealed edges condition with accompanying close-up of
porosity or other defects – (b) Control Prepreg (c) Prepreg 104-30 (d) Prepreg 104-120.
The surface porosity levels in laminates produced with and without
edge-breathing were measured and compared, as shown in Fig. 9a. Edge-
breathing yielded surface porosity near 1% when Control Prepregs were used,
but the porosity levels dropped more than 50% when Prepregs 104-30 and
104-120 were used. When edge breathing during cure was eliminated, the
difference was far more pronounced. Surface porosity was 8% in Control
laminates, less than 1% in 104-30 laminates, and negligible in 104-120
laminates (Fig. 9a).
34
Laminates produced with Control Prepreg showed narrow surface pits
3–15 mm in length and 0.5–2 mm in width (Fig. 9b). The distinctive elongated
shape of the surface pits reflects the fact that air entrapped at the tool/part
interface (and resin) flows primarily along the fiber direction. Laminates
produced with Prepreg 104-30 were nearly free of surface porosity, showing
occasional small pits (Fig. 9c). In contrast, laminates produced from Prepreg
104-120 showed a periodic array of equiaxed surface pits with spacing
identical to the film perforations (Fig. 9d). Furthermore, the saturated film
openings appeared as faint patches on the laminate surface, an effect
attributed to fiber waviness and a thinner resin film in these regions. Such
an artifact was not visible for laminates produced with Prepreg 104-30 (see
Fig. 9c).
Bulk porosity. For measurements of bulk void content, mutually
orthogonal sections were prepared from each laminate center. Cross-sections
were polished to 4000 grit on a polishing wheel (LaboPol-2, Struers), and
regions 20 × 5 mm were examined with a microscope. The average bulk void
content of each laminate was measured in the same manner as the surface
porosity.
For the Control Prepreg, the bulk void content nearly tripled, from
1.2% for standard cure conditions, to 3.2% with sealed edges (Fig. 10a). In
contrast, for laminates created with Prepreg 104-30, the bulk void content
was only 0.2–0.3% for both curing conditions, due to both fewer and smaller
35
voids. For laminates created with Prepreg 104-120, the void content
decreased further to 0.1% for both curing conditions. The insensitivity of
dewetted prepregs to restricted in-plane air evacuation demonstrates that air
evacuation occurred almost exclusively by breathe-out in the z-direction.
Similarly, the results indicate that continuous film laminates, which
represent equivalence to conventional VBO prepregs, rely exclusively on in-
plane breathe-out pathways, which are far less robust and less effective than
z-direction pathways, when present.
Figure 10 (a) Internal porosity values of laminates created with continuous resin film and
dewetted resin film under the conditions of standard cure and sealed edges. (b, c, d) Cross-
sectional images of laminates cured under the sealed edges condition for internal porosity
measurements – (b) Control Prepreg, (c) Prepreg 104-30, (d) Prepreg 104-120.
Martinez et al. [33] conducted studies on a thick laminate (64 plies) of
unidirectional prepreg produced with dewetted resin films and compared the
results to a control prepreg (128 plies of Cycom 5320-1, Solvay) of the same
thickness. The control prepreg exhibited 8.5% internal porosity, while the
laminate produced with discontinuous resin (with optimized distribution)
showed 2.4% internal porosity. The study demonstrated that an increase in
the flow distance in the through-thickness direction increases the amount of
porosity in both the control prepreg and the prepreg produced with dewetted
resin. However, the porosity value for the prepreg with dewetted resin is
36
approximately 25% of the porosity of the control prepreg. These results
indicate that the benefits of a discontinuous resin pattern, as described here,
persist even as laminate thickness increases.
Two types of bulk porosity are observed when standard VBO prepregs
are cured [34]. Inter-ply porosity consists of entrapped gas bubbles within
resin-rich regions, typically between plies. Intra-tow porosity manifests
within fiber tow bundles, and is generally attributed to incomplete resin
infiltration due to inadequate resin rheological properties or process
conditions. These pores are often classified as gas-induced and flow-induced,
respectively. The Control Prepreg exhibited both inter-ply and intra-tow
porosity (Fig. 10b), indicating both gas-induced and flow-induced porosity.
The porosity type in laminates created with either Prepreg 104-30 or Prepreg
104-120 was primarily intra-tow porosity (Fig. 10c and 10d). These results
indicate that the type of porosity produced using the discontinuous patterns
was flow-induced and not from entrapped air.
Statistical significance. The independent (unpaired) t-test is commonly
used to determine if the means of two independent data sets differ
significantly. A t-test was conducted for the data sets from the Control
Prepreg and either Prepreg 104-30 or Prepreg 104-120 for each cure condition
(standard and sealed edges), and for both surface and bulk porosity. Due to
limited sample sizes, each specimen was assumed to be an individual sample.
The difference in the means of two sample sets is considered statistically
37
significant if the p-value determined from the t-test is less than the chosen
threshold value (usually 0.05) [35]. All tests were deemed statistically
significant (p < 0.05) except for the surface porosity values of the Control
Prepreg and Prepreg 104-120 under standard cure conditions, where the p-
value was 0.071. Thus, the porosity values for prepreg produced with
discontinuous resin patterns was significantly less than porosity values
associated with the Control Prepreg.
5.3 Laminate structure
Earlier, in Sec. I.4.1, uncured laminates created using Prepreg 104-120
were shown to exhibit resin-rich regions and fiber waviness, raising questions
about the extent to which such non-uniformity would persist post-cure. The
cross-sections used for internal porosity measurements provide the
opportunity to assess the structure of the laminates, particularly with regard
to this question of resin distribution and fiber integrity. The Control Prepreg
exhibited a standard VBO prepreg microstructure, with resin evenly
distributed within the fiber bed, uniform thickness between plies, and
straight fibers (Fig. 10a). For both Prepreg 104-30 and 104-120, the
laminates exhibited a similar structure post-cure, demonstrating that the
consolidation process can achieve microstructural uniformity (Fig. 10b
and 10c).
38
5.4 Bulk factor
The bulk factor is particularly important for curved parts, where a
large bulk factor can result in wrinkling and/or bridging of plies. The bulk
factor is the ratio of the initial thickness of the prepreg stack to the final
thickness of the laminate:
𝐵𝑢𝑙𝑘 𝐹𝑎𝑐𝑡𝑜𝑟 =
𝑡 𝑖 𝑡 𝑓 , 𝑡 = 𝑡 ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠
(9)
In principle, the ideal bulk factor value is 1.0, a value that represents
no change in thickness. However, prepregs that incorporate dry areas
intrinsically possess bulk factors >1.
Figure 11 Bulk factor of laminates produced with continuous resin film and discontinuous
(dewetted) resin film.
Here, the bulk factor for the laminates created with Control Prepreg
ranged from 1.17 to 1.26 depending on cure conditions (Fig. 11). As the extent
of dewetting increased, the bulk factor increased to 1.30–1.40, representing
39
increases of 5–13%. The increase in bulk factor is not unexpected, because
the dewetting process accumulates resin locally, creating regions thicker (or
thinner) than the original film thickness.
6. Conclusion
A method to create high through-thickness permeability VBO-OoA
prepreg using UD carbon fiber tape was demonstrated and characterized.
The approach was distinguished by the use of polymer film dewetting (prior
to combining with fibers) to create periodic openings in the resin, through
which gas evacuation during cure occurred during laminate consolidation.
Laminates fabricated using prepregs produced by this method yielded near-
zero void contents even when in-plane gas evacuation (edge breathing) was
eliminated. In contrast, conventional OoA prepreg produced with continuous
films resulted in laminates that exhibited much greater porosity (up to 8%)
for both standard cure conditions and sealed edges.
The experiments revealed three important aspects for creating a
prepreg format for high transverse gas permeability. First, a high level of
interconnectedness of (dry) evacuation channels between plies must be
achieved to create accessible pathways for gas evacuation and prevent
premature occlusion. Measurements from permeability testing revealed that
the transverse permeability increased as feature size dimensions of the
discontinuous resin films increased. Therefore, the pathways are more
accessible for through-thickness gas evacuation as feature sizes increase.
40
Second, the openings must be sufficiently large to prevent closure of channels
in laminates with large ply counts both during lay-up and/or during cure, but
also small enough to prevent formation of surface defects and achieve full
saturation during cure. The surface porosity results indicate that
discontinuous patterns do exist that result in little to no porosity and without
surface defects, although large feature sizes (large distances between resin
features) will result in surface defects. Finally, the shape of the resin
accumulations during dewetting can affect part quality by influencing the
bulk factor, with greater dewetting levels potentially conflicting with lay-up
and consolidation of contoured parts.
This work builds on previous studies [1,2], introducing a potentially
scalable technique for producing OoA prepregs with discontinuous resin
distributions from any fiber bed. Furthermore, the work demonstrated that
in-plane discontinuity of resin (and not a specific prepreg embodiment) is the
prepreg attribute that alone can nearly eliminate porosity in VBO-cured
parts. Previous studies relied on a method for producing prepregs that relied
on fabric topography to impart resin discontinuity. Furthermore, our work
has not attempted to optimize the dewetting technique or resin distribution,
and thus further refinements and advances in efficiency may well be possible.
The most effective distribution of resin for a given fiber bed and application is
currently unexplored, but the dewetting technique presented here can be
41
readily adjusted to produce a wide variety of complex patterns, a potentially
fertile area for exploration.
OoA/VBO prepreg processing presently suffers from a lack of
robustness in the manufacturing process, and unacceptable defect levels
often-observed in non-ideal manufacturing conditions or part geometries. The
method described here [14] enables the production of UD fiber prepregs with
discontinuous resin distributions, and such prepregs will enable manufacture
of composite parts with low defect levels even in sub-optimal processing
conditions. In addition, dewetting is potentially backwards compatible with
hot-melt prepregging, since in principle, imprint/de-wet steps can be
incorporated into existing prepreg production lines. The inherent
manufacturing robustness imparted by the methods presented can, in turn,
expand the applicable uses of VBO prepregs within aerospace manufacturing
and into other non-aerospace applications.
* This study was published in Composites: Part A in November 2018
[36]. Also, this work was presented at the SAMPE Conference 2018 [37],
where it won First Place in the Outstanding Technical Paper Awards. The
work was also featured in an article published in the magazine, Composites
World [38].
42
CHAPTER II:
Effects of Resin Distribution Patterns on
Through-Thickness Air Removal in
Vacuum-Bag-Only Prepregs
1. Background
Prepregs processed via out-of-autoclave (OoA) vacuum-bag-only (VBO)
methods offer an increasingly viable alternative to traditional autoclave
manufacturing methods [39]. VBO prepregs alleviate problems associated
with the autoclave, including high costs from acquiring and operating
pressure vessels, limited throughput (i.e., production bottlenecks), and high
resource use (e.g., energy, nitrogen gas). However, OoA/VBO processing is
limited to 0.1 MPa (1 atm) of consolidation pressure, which is often
insufficient to collapse porosity to acceptable levels. Consequently, if
entrapped gases are not completely evacuated during early stages of
processing, VBO-cured parts will exhibit porosity and diminished mechanical
properties [6].
Conventional VBO prepregs are fabricated by pre-impregnating the
fiber bed using a hot-melt process: pre-catalyzed resin is formed into a thin
film on backing paper and pressed onto the fiber bed using rollers [7,8]. By
design, the resulting prepreg includes dry, highly permeable gas evacuation
pathways at the mid-plane of the ply. Combined with appropriate
consumables, such as edge-breathing dams, these pathways can be sufficient
43
to achieve low defect levels (i.e., porosity) within cured parts [9]. However, in
laminates fabricated from conventional VBO prepregs, sub-optimal
conditions such as inadequate vacuum, high levels of absorbed moisture in
the resin, or part geometries that prevent in-plane air removal can lead to
unacceptable void levels [10,11].
1.1 Background
Past work has shown that discontinuous resin films increase the
capacity for air evacuation in the z-direction (transverse) by creating
additional egress pathways, and can virtually eliminate porosity caused by
entrapped gases and low-pressure VBO consolidation [1–3]. For example,
Roman showed that the resultant cured composite parts using prepreg with
discontinuous resin film exhibited near-zero internal porosity [2]. In related
work, Grunenfelder et al. showed that, even under non-ideal manufacturing
conditions, a discontinuous prepreg format (denoted USCpreg) produced
near-zero internal porosity and no surface defects [1]. In contrast, under
identical manufacturing conditions, prepreg with continuous resin resulted in
high porosity levels. Grunenfelder’s work demonstrated that the USCpreg
format was more robust than conventional VBO prepregs. Finally, Tavares et
al. measured the through-thickness air permeability at room temperature
and processing conditions of a commercial semipreg (“Zpreg”) and an
equivalent unidirectional prepreg constructed with continuous resin [3].
Results showed that the permeability of the semipreg was three orders of
44
magnitude greater than the continuous film prepreg before and during the
cure cycle, owing to a network of dry interconnected pore spaces. Their work
showed that the initial overall dry space in prepreg prior to cure increased
when the resin distribution was discontinuous. Nevertheless, these works
together demonstrated the increased robustness of prepreg with
discontinuous resin. This beneficial effect can be attributed to the much
shorter distance for air evacuation in the through-thickness direction as
compared to the in-plane direction. However, methods of production allowing
for the creation of arbitrary patterns on a chosen fiber bed were not identified
or disclosed.
Previously, we demonstrated a simple technique to create
discontinuous resin patterns via polymer film dewetting on a substrate [36].
Resin film was perforated on silicone-coated backing paper (i.e., the
substrate) to create an array of nucleation sites. The film was then heated,
which caused resin to recede from the perforated sites, nucleating dewetted
openings. The resin recession was driven by the surface tension between the
low surface energy substrate (the silicone-coated backing paper) and the resin
[16–19]. Finally, the dewetted resin was transferred onto the fiber bed by
pressing briefly. This technique can be used to create a variety of resin
patterns, including stripes, grids, and islands. In addition, the patterned
resin films created with this technique can be applied to any dry fiber bed.
45
1.2 Objective
Polymer film dewetting allows for the creation of resin distribution
patterns with varying shapes and sizes. However, this capability
substantially enlarges the design space for the prepeg format. The optimal
pattern for efficient air evacuation has not been identified yet, nor is it
obvious. In addition, the relationships between pattern characteristics,
laminate design (i.e., stacking orientation, ply count), and prepreg properties
with regards to z-direction air evacuation are unclear. Because of the
extensive design space possible for VBO prepregs featuring these new
formats, experimental screening based on prepreg production, laminate
layup, and permeability testing will be arduous, time-consuming, and
ineffective in highlighting governing trends. Thus, the present work focused
on developing a method to (1) screen (evaluate and differentiate) between
discontinuous resin patterns without excessive experimentation or set-up
time, and (2) identify prepreg characteristics that likely favor rapid air
evacuation in the transverse direction in the initial state (i.e., at room
temperature, before significant flow-compaction has occurred).
These objectives were accomplished by developing a program (in
MATLAB) to generate geometric models of the prepreg (in 2D) and of an
arbitrary laminate (in 3D). Methods were implemented for measuring several
attributes from these models: projected surface area exposed of stacked
prepreg, sealed interfaces, and tortuosity of the air evacuation pathway.
46
Subsequently, a factorial design study was conducted to determine the effect
of pattern factors on each of these attributes. The factorial design was
analyzed using n-way ANOVA and multi-comparison tests. The results from
the multi-comparison tests were used to select laminates for experimental
validation. Finally, the effective air permeability at room temperature of
these laminates was tested. The study effectively outlines a methodology that
can readily discriminate between prepreg designs and allow for the rapid
selection of prepregs that efficiently evacuate air.
2. Modeling
2.1 Model development
The primary goal of modeling was to create a method for rapidly
screening prepreg configurations (and resulting laminates) with respect to
initial capacity for transverse air evacuation at room temperature. Thus,
explicitly modeling resin flow and fiber bed compaction for an arbitrary
discontinuous prepreg laminate was excluded from the scope of this first
effort. The capacity for gas flow within complex-shaped porous media can be
estimated using different approaches, including geometric analysis (employed
here), analytical models of flow in porous media, and computational
simulations. Analytical models capture and express the underlying physics,
and such models have been used to generate process maps to guide material
or process development [40–43]. Computational models allow detailed, often
multi-physics analysis of physical phenomena within complex domains [44].
47
However, both methods must be configured and solved for a specific domain
geometry. While parameterization of scales is possible, neither option is
inherently well suited to broad, rapid screening of arbitrary prepreg and
laminate patterns. For these reasons, this study focused on investigating the
viability of a geometric model that is simple to implement (i.e., parametric)
and execute, and that can reduce a large design space to a small subset of
potential solutions, which can be evaluated experimentally.
The model developed here relied on simplifying assumptions. Prior
work has shown that gas transport occurs much faster through dry regions
(i.e., unsaturated fiber reinforcements) than through saturated regions
[1,3,45]. Hence, the model was based on the premise that gas evacuation
capacity is primarily governed by the characteristics (i.e., size and
connectivity) of the dry pore network, rather than by bubble migration
through resin. Hence, void growth and transport within the fluid domain
were not modeled. Previous studies have also indicated that, following initial
compaction, the microstructure of partially-impregnated prepregs does not
evolve substantially under room-temperature vacuum compaction [46,47].
For this reason, the model utilized a rigid geometric framework, without
simulating compaction. In addition, the thickness of a resin layer and a fiber
layer were arbitrarily defined and remained constant across prepreg designs.
However, in actuality, the thickness of the resin layer would increase with a
decrease in resin feature dimensions.
48
2.2 Parametric study
2.2.1 Methods
Figure 12 (a) The traditional manufacturing of prepreg and laminates with continuous resin
illustrating air entrapment at the resin-to-resin interface and the mechanism of air removal in
the mid-plane. (b) The fabrication of prepreg and laminates with a discontinuous resin
distribution illustrating air removal in the through-thickness direction and the potential for
seal-off between resin layers (due to large resin feature dimensions and random placement
during stacking).
Prepregs are generally produced by placing a continuous layer of resin
on either or both side(s) of a fiber bed [Fig. 12a]. Usually, in OoA prepregs,
the fiber bed is partially impregnated by the resin, where the dry space
allows for in-plane evacuation of air via the ply mid-plane. In OoA prepreg
formats that feature discontinuous resin, surface openings allow for
additional air evacuation in the transverse direction [Fig. 12b]. A laminate is
constructed by stacking plies of such prepreg. The placement of the resin on
the fiber bed and the plies of prepreg for a laminate are random (i.e., no
intentional alignment of features). The space between the resin layers of the
stacked prepregs is an interface in which air can be readily entrapped. In
49
addition, with discontinuous resin, this interface can either be sealed, if
surface openings are too small, or not sealed, allowing for an interconnected
evacuation pathway in the transverse direction.
Figure 13 (a) Examples of the discontinuous resin patterns created via the polymer film
dewetting technique. (b) Examples of the discontinuous resin patterns coded in MATLAB.
The code was written to create patterns that replicate those that have
been created by the polymer film dewetting technique [36]. These patterns
included stripes, islands, and grids [Fig. 13a]. To create these patterns,
rectangles in 2D coordinates were created using a four-element vector of the
form [x y w h]. The x and y elements determine the location, and the w and h
elements determine the size. Curvature was also specified, where a value of 0
results in no curvature, and a value of 1 results in maximum curvature. For
stripes, the curvature of the rectangle was set to 0, whereas for islands and
50
grids, the curvature was set to 1. The islands and grid patterns were
programmed to be mutually reciprocal by computing the complement of the
image. In addition, the images were cropped to the same size to eliminate any
automatically created borders. Fig. 13b shows examples of a single resin ply
created by this method for each of the three pattern types studied. Black
regions represent void space (defined as a pixel intensity of 0) and white
regions represent resin (a value of 1).
Table 2 List of the stacking sequences at 4, 8, and 16 plies implemented into the code.
The code was designed so that the laminate was symmetrical about the
midpoint [Table 2]. Thus, the resin layers that make up the interface at the
midpoint were oriented in the same direction. Each resin layer was randomly
translated in the x- and y-directions by generating a single uniformly
distributed random number in the interval (0,1). This random number was
multiplied by the phase, which was the sum of the resin distance and the dry
51
space distance, then added to the location elements in the four-element
vector.
Three key output attributes were defined to generate a broad
description of the air evacuation capacity of a laminate constructed with
discontinuous resin patterns: (a) the two-dimensional projected exposed
surface area (%) of a laminate, (b) the percentage of sealed interfaces in a
laminate (relative to the total number of inter-ply regions), and (c) the
tortuosity of the dry pore network pathways in the laminate. The 2D
projection of the exposed surface area (%) described the presence and amount
of direct (unoccluded) transverse-oriented pathways. However, even with zero
projected surface area exposed, gases can move in the through-thickness
direction by migrating laterally through the dry fibers. Hence, the percentage
of sealed interfaces described the number of occlusions in the modeled pore
network for flow in the transverse direction. A sealed interface is created
when the resin films on adjacent prepreg plies are stacked such that film
openings are sealed off, effectively generating a continuous film. In such
cases, continuous through-thickness evacuation pathways will not exist from
laminate bottom to top. Finally, tortuosity described the distance gas would
have to travel within in the laminate to escape (assuming no sealed
interfaces). If the laminate contained a sealed interface, then the laminate
would not have a tortuous path available for air evacuation. Large tortuosity
values were expected to reduce the momentum of gas flow, thereby reducing
52
permeability, even in the absence of sealed interfaces. The input and output
variables are outlined in Table 3.
Table 3 List of input and output variables of the developed code.
Each parameter computation was iterated 25 times to achieve an
average value. When prepreg and laminates are manufactured, the resin
layers are randomly placed on the fiber bed, and thus the iterations would
compensate for the randomness in making prepregs and laminates. The
parameters were calculated for 4, 8, and 16 plies (i.e., 8, 16, and 32 resin
layers or 3, 7, and 15 interfaces). Three orientations were programmed for
computation: [0/0]n, [0/90]n, and [0/90/±45]n (i.e., quasi-isotropic).
53
Figure 14 A flow chart of the developed algorithms to create the resin layers, resin interfaces,
and 3D objects to compute the surface area exposed (%) of the “stacked” images, sealed
interfaces within the specified ply count, and tortuosity, respectively.
The code for the three parameters is illustrated in a flowchart in
Fig. 14. For calculations of projected surface area exposed (%), the laminates
(or “stacks”) were created by matrix addition of the appropriate number of
resin layers according to the target ply count. The surface area exposed (%),
ASE, then was calculated using the following equation:
𝐴 𝑆𝐸
=
𝑁𝑜 . 𝑜𝑓 𝑍𝑒𝑟𝑜 𝐸𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑇𝑜𝑡𝑎𝑙 𝑁𝑜 . 𝑜𝑓 𝐸𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑥 100%
(10)
54
For sealed interfaces (%) calculations, an interface was created by the matrix
addition of two resin layers. The number of interfaces created for each
laminate was 3 for 4 plies, 7 for 8 plies, and 15 for 16 plies. The exposed
surface area (%) of the interface was calculated from Eqn. 10. A threshold
value of 0.1% was employed, where, if the surface area exposed (%) was less
than 0.1%, then the interface was considered sealed. Otherwise, the interface
was not sealed. The percentage of sealed interfaces was calculated by:
𝑆𝑒𝑎𝑙𝑒𝑑 𝐼𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒𝑠 ( %) =
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑁𝑜 . 𝑜𝑓 𝑆𝑒𝑎𝑙𝑒𝑑 𝐼𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒𝑠 𝑇𝑜𝑡𝑎𝑙 𝑁𝑜 . 𝑜𝑓 𝐼𝑛𝑡𝑒𝑟𝑓𝑎𝑐 𝑒 𝑠 𝑥 100 %
(11)
For tortuosity calculations, 3D objects were created to represent the
laminates fabricated with discontinuous resin. The dry fiber layers between
the resin layers were modeled as dry space by creating matrices of all zeroes.
The “fiber layers” were not rotated nor translated. The “resin layers” were
constructed just as before, in that the layers were rotated and randomly
translated (in plane). Each fiber layer was four pixels thick, and each resin
layer was two pixels thick.
Figure 15 Examples of the 3D objects coded in MATLAB to represent a laminate created with
discontinuous resin patterns. Each object was 8 plies, quasi-isotropic, and 30% single layer
surface area exposed.
55
Examples of how the 3D objects were rendered are shown in Fig. 15. Next,
the tortuosity of the transverse air evacuation pathways was calculated using
the TORT3D.m program developed by Al-Raoush et al. [48]. The code read a
segmented image and determined all possible tortuous pathways required to
compute tortuosity. The algorithm conducted a guided search for the
connected path in the void space of the image, utilizing the medial surface of
the void space. The average of all connected pathways in that direction was
used to compute tortuosity. Geometric tortuosity, τg, was defined as:
𝜏 𝑔 =
〈𝐿 𝑔 〉
𝐿 𝑠
(12)
where 〈𝐿 𝑔 〉 is the average length of true paths through the porous media, and
Ls is the length of the straight-line path across the porous media in the
direction of flow. The tortuosity value, τ, acquired from the code was defined
as:
𝜏 =
∑ 𝑙 𝑝 𝑖 𝑛 𝑖 =1
𝑛 𝑙 𝑠
(13)
where lp is the given path through the void space that connects the boundary
of the images in the direction of flow, ls is the corresponding straight line, and
n is the number of paths. If the 3D object contained a sealed interface, the
program returned a tortuosity value of “NaN,” meaning the program was
unable to compute a value.
56
2.2.2 Results
Figure 16 Surface area exposed (%) for a single resin layer for each of the studied patterns
[(a) stripes, (b) islands, (c) grid] with the specified range of resin and dry space distances of 1
to 20 pixels. Diagrams of the definition of the variables “Resin Distance” and “Dry Space
Distance” are below each graph.
Single layer ASE. The exposed surface area (%) [Eqn. 10] of a single
discontinuous resin film was determined for all three pattern types for resin
widths or diameters specified between 1 and 20 pixels, and dry space distance
specified between 1 and 20 pixels [Fig. 16]. The graphs in Figs. 16a-c show
the limitations of the amount of initial (or single layer) surface area exposed.
The stripes pattern [Fig. 16a] yielded the largest range of exposed surface
area (%), with values from 10% to 90%. However, an upper limit of 90%
appeared to exist for the stripes pattern. For the islands pattern [Fig. 16b],
no upper limit appeared to exist, but a lower limit of 15% did manifest. As for
the grid pattern [Fig. 16c], no lower limit existed, but an upper limit of 65%
57
did exist. With all three constraints considered, the range of surface area
exposed (%) that could be used for comparison between the three patterns
was 15% – 65%.
Figure 17 Selected results of the projected surface area exposed (%) [(a) Quasi-Isotropic
(b) Stripes] as ply count increases from the code developed.
Projected ASE. The projected surface area exposed (%) was calculated as
each resin layer was added to the stack. These computations were performed
for single-layer surface area exposed values of 30% and 60%. Fig. 17a
compares the projected surface area exposed (%) of the three patterns
(stripes, islands, and grid). Provided the single layer surface area exposed (%)
value was the same, the decrease in projected surface area exposed (%) was
the same across the pattern types. Fig. 17b compares the projected exposed
surface area (%) for the stripes pattern oriented in [0/0]n, [0/90]n, and
[0/90/±45]n. Orientation had a negligible effect on the behavior of how the
projected surface area exposed (%) decreased. These graphs also show the
point where the projected surface area exposed (%) approached zero. For 30%
58
single layer surface area exposed, the projected surface area exposed (%) was
nearly zero after just 2 plies (or 4 resin layers). For 60% single layer surface
area exposed, the projected surface area exposed (%) was nearly zero after
5 plies are laid down (or 10 resin layers).
Figure 18 Selected results of sealed interfaces (%) [(a) [0/0]n (b) Stripes] and tortuosity
[(c) [0/90]n (d) Stripes] computations from the code developed for 16 ply laminates.
Sealed interfaces. The percentage of sealed interfaces (%) was
calculated for 16-ply laminates with values of single resin layer surface area
exposed between 7.5% and 60%. Fig. 18a compares the sealed interfaces
across the three patterns (stripes, islands, and grid) with an orientation of
[0/0]n. The striped pattern resulted in substantial amounts of sealed
59
interfaces (10.1% - 87.2%) at values of single layer surface area exposed (%)
less than 60%. The grid pattern also produced substantial amounts of sealed
interfaces (9.9% – 61.1%) when the single layer surface area exposed (%) was
less than 45%. On the other hand, islands only yielded sealed interfaces when
the single layer surface area exposed was less than 30%. However, an islands
design with a single layer surface area exposed of 7.5% could not be obtained,
so the sealed interfaces could not be determined at this value.
Fig. 18b compares the number of sealed interfaces for 16-ply laminates
laid up with stripes in each of the three orientations ([0/0]n, [0/90]n, and
[0/90/±45]n). Intuitively, stripes would appear to be the most sensitive of the
three patterns to orientation. Here, the data shows that [0/90]n and
[0/90/±45]n laminates behaved identically, where the proportion of sealed
interfaces was markedly less than for [0/0]n. Intuitively, laminates produced
with resin stripes oriented in either [0/90]n and [0/90/45]n should not exhibit
any sealed interfaces because the stripes were orthogonal. However, at the
midpoint, the stacking was symmetric (the stripes were parallel), which could
cause a sealed interface in certain cases of the randomly oriented set
[Table 2].
Tortuosity. Tortuosity (a unitless quantity) was calculated for 16-ply
laminates. Fig. 18c compares these laminates stacked in the orientation of
[0/90]n for stripes, islands, and grid. On average, the stripes had greater
tortuosity than either the island or grid patterns. The island and grid
60
patterns behaved similarly with high single layer surface areas exposed.
Fig. 18d compares the 16 ply stripes laminates oriented in [0/0]n, [0/90]n, and
[0/90/±45]n. The [0/90]n laminate on average had a smaller tortuosity.
However, the [0/90/±45]n laminate resulted in unusually high tortuosity. The
high tortuosity can be attributed to the fact that each ply of a laminate
produced with a quasi-isotropic stacking orientation was oriented in a
different direction. These different orientations would result in more indirect
and convoluted pathways than a simpler orientation like [0/0]n and [0/90]n.
2.3 Statistical analysis
Full factorial experiment. A full factorial experiment consists of two or
more factors such as, in this case, pattern type and ply count. Each factor has
discrete values or levels, such as, here, 4, 8 and 16 plies for ply count. The
resulting values from the full factorial experiment take on all possible
combinations of these levels across all such factors [49,50]. Such an
experiment allows evaluation of the effects of each factor on the dependent
variables, as well as the importance of interactions between factors. Here,
each experiment (for each dependent variable) was denoted as a 34 factorial,
which identifies the number of factors (4), the number of levels each factor
has (3), and the number of experimental conditions in the design (34 = 81).
Also, three dependent variables (projected surface area exposed, sealed
interfaces, and tortuosity) were being calculated. Therefore, the total number
of calculations was 243 (3 × 81). The number of physical experiments that
61
would be required to perform such a screening instead (with 25 iterations of
each experiment) demonstrates the utility of the geometric model, which
would be much faster to execute. As listed in Table 4, the four factors (with
the three levels) studied were pattern type (stripes, islands, grid), single layer
surface area exposed (30%, 45%, 60%), ply count (4, 8, 16), and stacking
orientation ([0/0]n, [0/90]n, [0/90/±45]n).
Table 4 List of input and output variables of the developed code.
Analysis of variance (ANOVA). The factorial experiment responses
were analyzed using Analysis of Variance (ANOVA). ANOVA is a collection of
statistical models, and their associated estimation procedure (such as the
variation among and between groups) is used to analyze the differences
among group means in a sample [51,52]. In its simplest form, ANOVA
provides a statistical test of whether the population means of several groups
were equal, and therefore generalizes the t-test to more than two groups.
A test result (calculated from the null hypothesis and the sample) was
treated as statistically significant if it was deemed unlikely to have occurred
by chance, assuming the truth of the null hypothesis. A statistically
62
significant result, i.e., when a probability (p-value) was less than a pre-
specified threshold (here, p ≤ 0.05), justified the rejection of the null
hypothesis. The null hypothesis was that all groups were random samples
from the same population. Rejecting the null hypothesis signified that the
difference in observed effects between groups was unlikely to be because of
random chance.
Initially, each experiment was performed using a 4-way ANOVA.
However, terms often would have missing p-values (represented as “NaN”),
indicating missing factor combinations and/or that the model had higher-
order terms. Thus, a 3-way ANOVA was performed instead, which allowed
for the identification of statistically insignificant terms. These terms were
then omitted, where their effects were pooled into the error term. The n-way
ANOVA was then performed again without the statistically insignificant
terms. The ANOVA results for each response (dependent) variable are
presented in Table 5.
63
Table 5 Summary of the n-way ANOVA results for (a) the Projected Surface Area
Exposed (%), (b) Sealed Interfaces (%), and (c) Tortuosity calculations.
For surface area exposed, the p-value was less than 0.05 (i.e.,
statistically significant) for only three of the factors – single layer surface
64
area exposed, ply count, and orientation. This result signified that the
projected surface area exposed (%) was insensitive to the pattern type.
Comparing the F-values, the largest F-values for the projected surface area
exposed (%) response variable was for ply count and single ply surface area
exposed (%), which indicated that these two variables were the most
influential on projected surface area exposed. The F-value is defined as
𝐹 =
𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑠𝑎𝑚𝑝𝑙𝑒 𝑚𝑒𝑎𝑛𝑠 𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 𝑤𝑖𝑡 ℎ𝑖𝑛 𝑡 ℎ𝑒 𝑠𝑎𝑚𝑝𝑙𝑒𝑠
(14)
where the null hypothesis was that these two variance values were roughly
equal (F-value = 1). Therefore, the larger the F-value, the greater the relative
variance among the group, and thus, the greater the dependence of the
dependent variable. The p-value indicates the probability of obtaining an
F-value as extreme or more extreme than the one observed under the
assumption that the null hypothesis was true.
For sealed interfaces (%), the ANOVA results indicated that all factors
except ply count were significant, whereas for tortuosity, all factors were
significant. Therefore, the number of plies of prepreg in a laminate had no
effect on the percentage of sealed interfaces (i.e., not affected by scale), but
ply count did affect tortuosity. The largest F-value for both the sealed
interfaces and for tortuosity was the stacking orientation. The significance of
the orientation may have been heavily skewed by its effect on the stripes
pattern, which was the most sensitive to orientation.
65
Multiple comparison tests. Multiple comparison tests were performed
to determine which groups of factors yielded statistically significant
differences. To preserve clarity, the resulting interactive graphs of the
estimates and comparison intervals were omitted. On the interactive graphs,
each group mean was represented by a symbol, where a line extending out
from the symbol represents the interval. Two group means were significantly
different if their intervals were disjointed; they were not significantly
different if their intervals overlapped. Any group can be selected and the
graph will highlight all other groups that were significantly different.
For the 2D projected surface area exposed (%), the pattern type was
determined to be insignificant, so the results were pooled together without
regard to pattern type. The group with the largest mean (and being
significantly different from the rest) was the 4-ply laminates made with 60%
single layer surface area exposed and oriented [0/0]n. The next two largest
(significantly different) means, in order, were the same 4 ply laminates with
60% single resin surface area exposed, but oriented [0/90]n and [0/90/±45]n.
These conditions identified prepreg designs with the highest quantity of
uninhibited air evacuation pathways in comparison to the other conditions
tested (i.e. 4 plies, 60% ASE, [0/0]n). However, all three conditions were for
4-ply laminates, indicating that uninhibited air evacuation pathways did not
exist for larger ply counts.
66
For sealed interfaces and tortuosity, ply count was ignored, because it
was insignificant from the n-way ANOVA results. For sealed interfaces, three
laminates exhibited the largest (significantly different) means, where all
three were oriented [0/0]n. Two of the laminates were stripes with 30% or
45% surface areas exposed, while the third laminate was a grid with 30%
surface area exposed. These results indicated prepreg designs that would
result in large amounts of sealed interfaces (i.e. stripes, 30% or 45% ASE,
[0/0]n and grid, 30% ASE, [0/0]n). For tortuosity, the largest mean was
associated with the striped pattern with 30% surface area exposed and
stacked in a [0/90/±45]n orientation. These results indicated a prepreg design
that would result in high tortuosity (i.e., stripes, 30% ASE, quasi-isotropic
stacking). These multi-comparison results will be demonstrated in
experimental verification testing, described next.
3. Experiments
Experiments were performed to (1) verify if geometric modeling can
discriminate between prepregs with high and low room-temperature through-
thickness gas permeability, and (2) determine which output attribute(s) most
closely correlated to effective permeability and porosity in cured laminates.
3.1 Materials
A unidirectional (UD) non-crimp carbon fiber bed (Fibre Glast
Development Corporation, Ohio, USA) and a toughened epoxy resin (PMT-F4,
Patz Materials & Technology, California, USA) were selected for testing. The
67
epoxy resin was designed for vacuum bag curing and featured medium-to-dry
tack. The fabric weight was 305 g/m2 (gsm), and the thickness was 0.36 mm.
A binder of polyester fill threads stitched in one direction held the UD fibers
in place. The tape exhibited negligible crimp, except around the binder
threads. For the permeability tests, the resin film weighed 152 gsm, yielding
prepreg with a resin content of 50% by weight, which was much larger than
standard UD prepreg (i.e., 33%). Thick resin was used to ensure that the
dewetting process yielded features that were close to the model image. The
resin, at thinner thicknesses, occasionally dewetted at surface imperfections
such as tears or depressions. For porosity measurements, prepregs were
fabricated from two thicknesses of resin film – either 152 gsm or 76 gsm. The
152 gsm film was also used in permeability measurements, while the 76 gsm
film was selected to match typical commercial prepregs. The 76 gsm film
yielded prepreg with a resin content of 33%. The standard cure cycle included
a ramp of 1.5 °C per min, followed by a two-hour dwell at 121 °C.
Dewetting was performed as described previously [36]. To facilitate
dewetting, nucleation sites were introduced using a box cutter. The dewetting
process was carried out on silicone-coated release paper. A standard oven
(Blue M Oven, Thermal Product Solutions, Pennsylvania, USA) was used to
heat the films for dewetting. After dewetting, the resin was applied to the
fiber bed by aligning and pressing constituents in an unheated hydraulic
press (G30H-18-BCX, Wabash MPI).
68
3.2 Permeability
Figure 19 (a) Summary of the sealed interfaces (%) and tortuosity values for each of the 8 ply
prepreg samples studied for permeability. (b) Through-thickness permeability values (blue
dots) of the tested prepregs and a planar fit (multi-colored plane) via a linear regression model
with error bars indicated with red lines.
The effective transverse permeability was measured and compared for
8 plies of prepreg produced from the striped pattern, with 30% surface area
exposed for each of the orientations, [0/0]n, [0/90]n, and [0/90/±45]n. These
conditions were chosen to distinguish how sealed interfaces and tortuosity
affected through-thickness air evacuation. The orientation [0/0]n represented
the case of a large amount of sealed interfaces and a small tortuosity. On the
other hand, [0/90/±45]n represented the opposite case, where the amount of
sealed interfaces was small and the tortuosity was large. Finally, [0/90]n
represented the case where the proportions of sealed interfaces and the
tortuosity were small. The values for sealed interfaces and tortuosity for
these prepregs are summarized in Fig. 19a. Projected surface area exposed
was not tested, since the results indicated that only ply count and single
layer surface area exposed affected this parameter. Ply count was a
69
specification that cannot be manipulated in a manufacturing setting, and a
change in feature dimensions would invariably change values for the number
of sealed interfaces and tortuosity.
In addition to samples created specifically for this study, samples from
a previous report [36] were also included, thereby expanding the permeability
data set to encompass other pattern types and feature dimensions. These
samples included both a grid pattern and stacked in a [0/90]n orientation. The
samples featured a single layer surface area exposed (%) of either 13% or
50%. The prepreg with an exposed surface area of 13% was likely to have a
large number of sealed interfaces (~ 11%), as well as tortuous air evacuation
pathways. The prepreg with an exposed surface area of 50% was likely to
have no sealed interfaces (0%), and less tortuous air evacuation pathways.
A custom test fixture was used for the experiments [28], following the
falling pressure method described by Tavares et al. [29] and Kratz et al. [28].
Plies of prepreg were laid over a cavity of known dimensions supported by
stacks of honeycomb core. The edges of the plies were sealed with vacuum
tape to prevent edge breathing, thus permitting air evacuation only in the
through-thickness direction. The laminates were covered with perforated
release film and breather cloth, and were vacuum-bagged. Vacuum was
drawn in the bag to compact the laminate and create a pressure difference
between the core cavity and the bag. The evolution of pressure in the cavity
was monitored over time using a pressure transducer and data acquisition
70
software (LabVIEW, National Instruments), and the measurements were
used to estimate an effective (slip-enhanced) permeability coefficient. Gas
flow in porous media generally enhances permeability through slippage of air
molecules along the boundaries of the air-filled pores [53,54]. All tests were
performed at room temperature (~20°C).
To obtain an average effective permeability value, two samples
(replicates) were tested for each experimental configuration, with a minimum
of three pressure decay trials per sample. Each trial was conducted to the
time at which the cavity pressure stabilized (indicating flow had ceased), and
the configuration was then re-pressurized to begin the next trial. The data
from the first trial was omitted, because air evacuated more quickly when the
consumables and plies had not been previously compacted.
Using Darcy’s Law, the one-dimensional laminar flow of compressible
air at isothermal and adiabatic conditions through a porous medium [28] can
be described by
−
𝐾𝐴 𝑃 𝐵𝑎𝑔 𝐿𝜇 𝑉 𝐶𝑜𝑟𝑒 𝑡 = 𝑙𝑛 [
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)+ 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )− 𝑃 𝐵𝑎𝑔 )
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)− 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )+ 𝑃 𝐵𝑎𝑔 )
]
(15)
where K is the permeability scalar in the flow direction in m
2
, A is the cross-
sectional area (1.46 × 10
−2
m
2
), PBag is the pressure at the bag side (5 × 10
3
Pa),
PCore is the pressure at the honeycomb core side in Pa, L is the lateral
dimension in m, μ is the viscosity of air at room temperature
(1.85 × 10
−5
Pa*s), t is time in s, and VCore is the volume of the core
(7.87 × 10
−4
m
3
). Here, the vacuum level was assumed to be 95%
71
(corresponding to an absolute vacuum bag pressure of 5 kPa). Plotting the
left-hand side versus time yields a straight-line plot, the slope of which can
be used to determine the effective air permeability of the prepreg, K.
A summary of the permeability values of the laminates and a plane
created by a linear regression fit of the data are presented in Fig. 19b. The
equation for the linear regression model and the coefficient of determination
(R2) are also displayed in the graph. Sealed interfaces and tortuosity were
defined by the symbols nsealed and τ, respectively. The value for R2 (0.966)
indicated that the model fit the data. The planar fit indicated that an
increase in the percentage of sealed interfaces and in the tortuosity does
indeed decrease the through-thickness permeability of the prepregs (a less
desirable outcome for robust manufacturing). The equation indicated that the
percentage of sealed interfaces had a larger effect on the decrease in
permeability than did tortuosity.
3.3 Porosity
Laminates were fabricated to determine parametric effects on internal
porosity for each of the prototype prepregs evaluated in permeability tests.
Laminates consisted of 8 plies of prepreg featuring a striped pattern of resin
and 30% surface area exposed for each of the ply sequences, [0/0]n, [0/90]n,
and [0/90/±45]n. Initially, each ply was cut to 150 × 150 mm, and trimmed to
140 × 140 mm after stacking. Standard vacuum bag procedures were used,
except the edges were sealed using sealant tape to prevent in-plane air
72
removal at the ply boundaries, thus allowing air evacuation only in the
through-thickness direction. Sealed edges approximate commonly
encountered process conditions that often restrict or prevent in-plane air
evacuation – features such as large or complex parts and parts with ply drops
or corners. A standard “ramp-hold” cure cycle was used.
Measurements of bulk void contents were performed using image
analysis of polished sections prepared from each laminate center. Void
contents were calculated from binary images processed to yield black and
white pixels.
𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦 ( %) =
𝑝 𝑏𝑙𝑎𝑐𝑘 𝑝 𝑤 ℎ𝑖𝑡𝑒 + 𝑝 𝑏𝑙𝑎𝑐𝑘 × 100 %
(16)
where p is the number of pixels. Finally, an average internal porosity was
determined from four images.
Laminates were fabricated using both 76 gsm and 152 gsm resin, and
measured porosity levels are shown in Fig. 20a. The laminate with [0/90]n
sequence featured no sealed interfaces and no tortuous air evacuation
pathways, and yielded a void content of only ~0.1% (Fig. 20b). In contrast, the
porosity levels in the [0/0]n and [0/90/±45]n laminates were 1.0% (Fig. 20c)
and 0.8% (Fig. 20d), respectively. The [0/0]n laminate represented the case of
a large number of sealed interfaces, but relatively non-tortuous pathways for
air evacuation, while the [0/90/±45]n laminate represented the case of
tortuous evacuation pathways and unsealed interfaces.
73
Figure 20 (a) The measured internal porosity of each of the 8 ply prototype prepregs. (b-d)
Cross-sections of the laminates made from each of the 8 ply prototype prepregs.
The experiments demonstrated that the prepreg and laminate features
included in the study strongly influence the efficiency of gas evacuation
during the cure cycle. The results also indicated specific aspects of prepreg
design that most strongly affect this efficiency. In particular, sealed
interfaces had a much greater effect on air permeability and void content
after cure than tortuosity. The results and the methodology described here
can be used to guide the design of discontinuous resin patterns for OoA
prepregs.
4. Conclusion
This work outlined a methodology that allows for the rapid screening
(i.e., evaluation and differentiation) of discontinuous resin patterns for VBO
74
prepregs, which can be used to guide prepreg development. The methodology
stems from previous work [36] that reported a technique to create
discontinuous resin patterns of arbitrary shapes and sizes via a polymer film
dewetting technique, which created a large design space for choosing a resin
distribution. Due to the overwhelming number of choices in designing a
prepreg with discontinuous resin format, selection of an appropriate resin
distribution was not obvious. Thus, the methodology presented here allows
one to differentiate between pattern types, feature dimensions, stacking
orientations, and ply counts (with as many iterations as specified) without
any physical experimentation by the use of simple geometric models.
Experiments validate that the characteristics identified using the geometric
model do indeed affect through-thickness air permeability and void content in
cured laminates. In practice, the design space can be greatly reduced by
eliminating patterns that do not allow for rapid air evacuation.
Various aspects of optimal design for prepregs with discontinuous
resin distribution were outside the scope of this work, but would be important
to pursue for future research. For example, the scope of this study did not
include the changes in bulk factor due to feature dimensions and the
subsequent effect on part quality. Smaller resin features with large surface
openings will generally result in more efficient air evacuation, but the bulk
factor will be much larger. A large bulk factor is associated with wrinkling in
curved surfaces, which will decrease the mechanical properties of the
75
composite part. Secondly, the flow of resin during cure in relation to
maximum feature dimensions was not evaluated. Full infiltration of a large
surface opening requires longer flow distances and would be more
challenging than small surface openings. Understanding the size restrictions
on gap sizes with regard to resin flow would be useful to guide future prepreg
designs for optimal resin distributions.
OoA/VBO prepreg processing presently suffers from a lack of
robustness during manufacturing, often yielding unacceptable defect levels
when manufacturing conditions are not fully controlled, are non-ideal, or
complex part geometries are involved. The method described here provides a
pathway to determine favorable (and, eventually, optimal) designs for
prepregs with discontinuous resin distributions. Such prepregs can enable
the manufacturing of composite parts with low defect levels without
autoclaves, even in sub-optimal processing conditions. The robustness
imparted by the methods presented can, in turn, expand the applicable uses
of VBO prepregs within aerospace manufacturing and into other non-
aerospace applications.
* This study was published in the journal Composites: Part A [55].
76
CHAPTER III:
Air Evacuation and Resin Impregnation in
Semi-pregs: Effects of Feature Dimensions
1. Background
In this work, we outline a strategy to determine dimensional
guidelines for discontinuous resin patterns required to impart robustness to
Out-of-Autoclave (OoA)/Vacuum-Bag-Only (VBO) processing of composite
prepregs. Currently, aerospace-quality composite materials are cured in
autoclaves at pressures of ~5-8 atmospheres. Autoclave pressure suppresses
porosity caused by entrapped air, insufficient resin flow, and evolved gases,
and consistently yield parts with low defect content, a quality often referred
to as robustness. However, the use of an autoclave has drawbacks, including
high capital and operating costs, limited throughput (production bottlenecks),
part size restrictions, and high resource use (energy, nitrogen). An appealing
alternative is OoA/VBO processing, in which composites are cured in
conventional industrial ovens. VBO processing, however, is inherently
susceptible to defects from adverse process conditions, such as poor vacuum,
incomplete air evacuation, and/or high humidity. Because at most 0.1 MPa
(1 atm) of pressure is exerted on the laminate during VBO processing, air and
other gases can remain entrapped during layup and cure. Once the resin gels
during the cure cycle, trapped bubbles cannot migrate, resulting in voids and
leading to unacceptable porosity levels [1,2,7,36,37]. High defect levels (>1%)
77
diminish mechanical properties and require part rejection. For VBO
processing to gain broad acceptance in aerospace, the manufacturing process
must yield high quality parts (low porosity) consistently.
Background. Conventional VBO prepregs are fabricated by partially
impregnating the fiber bed using a hot-melt process, in which pre-catalyzed
resin is formed into a thin film on backing paper and pressed into the fiber
bed via rollers [7,8]. By design, the resulting prepreg includes dry, highly
permeable gas evacuation pathways at the mid-plane of the ply. During
manufacturing, using edge-breathing dams and appropriate consumables,
these pathways can be sufficient to achieve low defect levels (i.e., porosity)
within cured parts [9]. However, in laminates fabricated from conventional
VBO prepregs, adverse conditions, such as inadequate vacuum, high levels of
absorbed moisture in the resin, or part geometries that prevent in-plane air
removal (including large part size), can lead to unacceptable void levels that
degrade the mechanical properties of the part [10,11].
VBO prepregs featuring discontinuous resin distributions (i.e, semi-
pregs) can potentially limit process-induced defects and impart robustness to
the process of manufacturing of aerospace-quality composite materials via
OoA methods [1–3,13,14,36,37]. Grunenfelder et al. [1,13,14] showed that
discontinuous resin in prepreg provided air evacuation pathways in the
transverse direction, virtually eliminating porosity caused by entrapped air
or moisture in OoA/VBO processing. Using this format, cured parts were
78
produced with near-zero internal porosity and no surface defects, even under
adverse manufacturing conditions. In contrast, conventional VBO prepreg
with continuous resin resulted in unacceptable levels of porosity (3-8%),
despite identical process conditions.
The benefit of semi-pregs derives from though-thickness pathways for
gas egress, which impart capacity for air evacuation in the z-direction
(transverse). These pathways can potentially eliminate porosity caused by
entrapped gases and low-pressure VBO consolidation. Tavares [3] measured
the through-thickness air permeability of a commercially available semi-preg
and an equivalent unidirectional prepreg constructed with continuous resin.
Results showed that the permeability of the semi-preg was three orders of
magnitude greater than the continuous film prepreg before and during the
cure cycle, owing to a network of dry interconnected pore spaces.
Discontinuous resin inherently increases the efficiency of evacuating air and,
in addition, traps less air at inter-ply interfaces.
While semi-preg formats are commercial products, design and
fabrication of such materials has not been thoroughly reported. Previous
work by the authors [36,37] focused on developing a technique to create finely
tuned resin distribution patterns with varying shapes and sizes that could be
applied to a range of reinforcement architectures. The method relied on
polymer film dewetting on a low-surface-energy substrate. In this process,
resin film was first perforated on backing paper (the substrate) to create an
79
array of nucleation sites. The film was then heated, causing resin to recede
(dewet) from the nucleation sites. The resin recession was driven by the
surface tension between the low surface energy substrate (the silicone-coated
backing paper) and the resin [17–19]. The dewetted resin was transferred
onto a fiber bed by pressing the constituents in a hydraulic press. Compared
to conventional OoA prepreg formats that feature continuous resin films,
prepreg produced using this technique exhibited nearly void-free cured
laminates, even when curing under adverse conditions.
Composite part production via out-of-autoclave processing requires
both efficient air evacuation and full fiber bed infiltration (saturation). While
the use of semi-pregs results in efficient air evacuation, discontinuous resin
introduces dry fiber regions and increases resin flow distances. Increased
resin flow distances may lead to incomplete fiber bed infiltration and
ultimately, porosity. Efficient air evacuation of semi-pregs has been
addressed in previous work, but an analysis of both efficient air evacuation
and subsequent full fiber bed infiltration has not been reported. For example,
Tavares [3] developed a 1D resin flow front model to fit results of through-
thickness permeability experiments and predict the air permeability of semi-
pregs throughout the cure cycle. The commercial semi-preg analyzed,
however, did not achieve full fiber bed infiltration during cure. In a separate
study, a geometric model was used to guide semi-preg design for efficient air
evacuation [55]. In the present work, we systematically address the design
80
and evaluation of semi-pregs for both efficient air evacuation and full fiber
bed infiltration, considering favorable cure conditions as well as realistic
adverse process conditions (i.e., poor vacuum, resin with accrued out-time).
Judicious selection of discontinuous resin distribution can mitigate macro-
porosity, promote efficient air evacuation from the laminate, and eliminate
micro-porosity arising from incomplete fiber wet-out. Additionally, the
method presented here can inform future efforts to develop process models
required to further expand applications of VBO prepregs.
Objectives. A key challenge to producing composite parts with semi-
pregs is ensuring full air evacuation while saturating the fiber bed during the
cure cycle. Whereas previous work [55] focused on the design of semi-preg
designs that resulted in efficient air evacuation using a geometric model, the
current work focuses on model-driven design and evaluation of semi-pregs
that saturate the fiber bed. While large surface openings maximize air
transport, infiltration of large openings requires long lateral flow distances
[36,37]. Conversely, smaller surface openings may ensure full fiber bed wet-
out because of shorter flow distances. However, small surface openings may
be sealed by adjacent prepreg plies if the surface openings do not overlap.
Thus, understanding and determining the range of surface opening sizes
required for full resin infiltration and full air evacuation is valuable to inform
semi-preg design. Feature size restrictions, moreover, will depend on process
81
parameters, such as ramp rate, dwell temperature, vacuum level, and resin
out-time.
The objectives of this work were (1) to define the range of feature
dimensions for a given resin system to ensure complete infiltration while
maximizing gas transport, (2) to provide the experimental basis for
evaluating this range, and (3) to demonstrate the efficacy of the semi-preg
design method in lab-scale laminates. These objectives were motivated by the
needs (a) to develop design guidelines for semi-pregs and (b) to increase
process robustness in OoA/VBO composite manufacturing.
In the present work, we employ in situ observations of prototype semi-
pregs to validate a developed model for resin flow fronts, then use the results
to inform the design of resin patterns for VBO cure of laminates. A model for
resin flow front progression was developed to predict the maximum flow
distance under a range of cure conditions. Both favorable and adverse process
conditions were analyzed to guide design of appropriate resin pattern
dimensions based on representative manufacturing parameters. The adverse
cure conditions studied were non-uniform ramp rate, inaccurate dwell
temperature, partial vacuum, and resin with accrued out-time. These
conditions limit the minimum resin viscosity and flow time achieved during
the cure cycle. Semi-preg was produced with dimensions informed by the
resin flow model, prototype laminates were fabricated, and the permeability
of the laminates was measured. The results from the in situ observations and
82
the prototype laminates corroborate the resin flow front model in the cure
conditions tested in this study. These results indicate that the mathematical
models used in this study can be used to guide the design of semi-pregs with
dry space dimensions required for full tow saturation and efficient air
evacuation.
2. Materials and experimental methods
Prepreg fabrication. Following procedures described in previous work
[36,37], prepreg samples were fabricated using a unidirectional (UD) carbon
fiber reinforcement (Fibre Glast Development Corporation, Ohio, USA) and a
toughened epoxy resin (PMT-F4, Patz Materials & Technology, California,
USA), which featured medium-to-dry tack. The UD fabric featured an areal
weight of 305 grams per square meter (gsm) and a thickness of 0.36 mm.
Polyester fill threads were stitched in one direction to hold the UD fibers in
place. The tape exhibited negligible crimp, except for the region adjacent to
this stitching. The areal weight of resin for all prepreg plies was 76 gsm,
yielding prepreg with a resin content of 33% by weight. The manufacturer’s
recommended cure cycle for the resin was a ramp of 1.5 – 3.0 °C per min to
121 °C followed by a two-hour dwell. Prior to this study, the resin was stored
in a freezer for approximately 6 months and had accumulated about 7 days of
out-life. The material was within specifications, with a recommended storage
life of two years at −10 °C, and an out-life of 120 days at room temperature.
83
Figure 21 These images show the semi-preg in the initial state (i.e., at room temperature
before cure). Both prepregs were dewetted for 1 min 30 s at 104 °C and subsequently pressed
onto the unidirectional fiber bed using a hydraulic press. (a) Image of the surface of the semi-
preg that was fabricated using a spike roller. (b) Image of the surface of the semi-preg that
was fabricated using a box cutter.
Semi-pregs were produced via a method of selective dewetting of resin
film. With the neat resin film on the silicone-coated backing paper (a low-
surface-energy substrate [19]), nucleation sites were introduced using a spike
roller (HR-2, Robert A. Main & Sons, Inc., New Jersey, USA) or a box cutter
knife. The spike roller pins were spaced at 6.35 mm, and the roller was
passed over the entire film in straight passes. Alternatively, a box cutter was
used to create evenly spaced resin strips by first scoring parallel lines in the
film. The resin film was then placed in an air-circulating oven (Blue M Oven,
Thermal Product Solutions, Pennsylvania, USA), to enlarge openings at the
nucleation sites. Subsequently, the resin film was attached to the UD fiber
bed by pressing the constituents briefly in an unheated hydraulic press
(G30H-18-BCX, Wabash MPI). Images of the resultant discontinuous resin
84
pattern on the UD fiber bed using the spike roller or the box cutter are
presented in Figs. 21a-b.
Flow front tracking. To monitor the resin flow front during processing,
the surfaces of the semi-pregs were tracked in situ using the technique
described by Hu et al. [30,31]. Four-ply semi-preg stacks (75 75 mm) were
fabricated as described above. The laminates were vacuum-bagged using
standard consumables, with the exception of the edge-breathing dams
typically required for VBO cure. Edge breathing was eliminated by sealing
the perimeter of each laminate with vacuum tape, thus permitting air
evacuation only in the through-thickness direction. Each laminate was
subjected to a 1-hr room-temperature vacuum hold prior to the cure cycle.
Resin flow was monitored and recorded during cure using a digital
microscope and time-lapse video acquisition (Dino-Lite US, Dunwell Tech,
Torrance, CA, USA). Temperature was monitored with a USB thermocouple
data logger (Lascar Electronics EasyLog EL-USB-TC-LCD), which was
attached adjacent to the laminate and directly on the tool plate. Pressure was
tracked using a voltage transducer and data acquisition software (LabVIEW,
National Instruments). The maximum pressure of the system was 96.5 kPa
(28.5 in Hg). To assess the influence of vacuum level on resin flow during the
cure cycle, a regulator was inserted between the vacuum pump and the
vacuum bag-assembled laminate, which was adjusted according to the test
conditions. To quantify the progression of resin infiltration, images from the
85
recorded video were analyzed. These images were selected from the initial
temperature ramp of the cure cycle, at temperatures between 26 and 121 °C
with an interval of 5 °C [example images presented in Figs. 22a-c]. An
average diameter of three to six surface openings at each temperature was
measured using image-processing software (ImageJ).
Figure 22 Example images of the semi-preg surface on the transparent tool plate during the
cure cycle at (a) 75 °C, (b) 95 °C, and (c) 121 °C. The last image shows that the surface
opening was still not fully infiltrated by the time the resin underwent gelation.
Experiments were conducted to evaluate the progression of resin flow
front for aged resin during processing. To replicate the increase in degree-of-
cure caused by ambient temperature aging (out-time), resin was artificially
aged by placing the material in an oven at 85 °C for either 75 min or 109 min.
This aging increased the degree of cure, α, to 0.15 and 0.30, respectively.
These values were chosen to replicate the values obtained by Kim et al. [56]
using a standard industrial epoxy resin system (CYCOM
®
5320-1), where α0 =
0.15 corresponds to 28 days of out-time and α0 = 0.30 corresponds to 49 days.
Permeability measurements. Using results from in situ observations,
the effective (slip-corrected) transverse permeability was measured and
compared for the semi-pregs with various dimensions of the surface openings.
86
A custom test fixture was used for the experiments [28], following the falling
pressure method described by Tavares et al. [29]. Plies of prepreg were laid
over a cavity of known dimensions supported by stacks of honeycomb core.
Measurements were recorded for 4-, 8-, and 16-ply laminates. The edges of
the plies were sealed with vacuum tape to allow air evacuation only in the
through-thickness direction (no transverse air evacuation). The laminates
were covered with consumables (perforated release film and breather cloth),
and then vacuum bagged. Vacuum was drawn in the bag to compact the
laminate and create a pressure differential between the core cavity and the
bag. The evolution of pressure in the cavity was monitored over time using a
pressure transducer and data acquisition software, and the measurements
were used to estimate an effective permeability coefficient. All tests were
performed at room temperature.
To obtain an average effective permeability value, two samples
(replicates) were tested for each experimental configuration, with a minimum
of four pressure decay trials per sample. Each trial was performed until the
cavity reached a pressure of 35 kPa, ensuring a uniform level of compaction.
Subsequently, the configuration was re-pressurized to begin the next trial.
The data from the first trial was omitted because air evacuates more quickly
when the consumables and plies have not been previously compressed.
87
One-dimensional laminar flow of compressible air at isothermal and
adiabatic conditions through a porous medium [28] can be described by
Darcy’s Law:
−
𝐾𝐴 𝑃 𝐵𝑎𝑔 𝐿𝜇 𝑉 𝐶𝑜𝑟𝑒 𝑡 = ln [
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)+ 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )− 𝑃 𝐵𝑎𝑔 )
( 𝑃 𝐶𝑜𝑟𝑒 ( 0)− 𝑃 𝐵𝑎𝑔 ) ( 𝑃 𝐶𝑜𝑟𝑒 ( 𝑡 )+ 𝑃 𝐵𝑎𝑔 )
]
(17)
where K is the permeability scalar in the flow direction in m
2
, A is the cross-
sectional area (1.46 × 10
−2
m
2
), PBag is the pressure at the bag side (5 × 10
3
Pa),
PCore is the pressure at the honeycomb core side in Pa, L is the lateral
dimension in m, μ is dynamic viscosity of air (1.85 × 10
-5
Pa·s at room
temperature), t is time in s, and VCore is the volume of the core
(7.87 × 10
−4
m
3
). The vacuum level was assumed to be 95% (corresponding to
an absolute vacuum bag pressure of 5 kPa). Plotting the left-hand side versus
time yields a straight-line plot, the slope of which can be used to determine
the effective air permeability of the prepreg, K.
Porosity. Finally, using the results from in situ observation and
permeability testing, prototype prepregs were fabricated and cured for
analysis of part quality. Each prepreg ply was cut to 150 × 150 mm. After
stacking, the edges of the stack were trimmed, resulting in dimensions of
140 × 140 mm. The oven cure cycle was programmed to reach a dwell
temperature of 121 °C with a ramp rate of 2 °C/min. For measurements of
bulk void content, mutually orthogonal sections were prepared from each
laminate center. Cross-sections were polished to 4000 grit on a polishing
88
wheel (LaboPol-2, Struers), and regions 20 × 5 mm were examined with a
microscope. Images were analyzed using software (ImageJ). A series of
manual steps produced a binary porosity image. The void content was
calculated from the number of black and white pixels:
𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦 =
𝑝 𝑏𝑙𝑎𝑐𝑘 𝑝 𝑏𝑙𝑎𝑐𝑘 + 𝑝 𝑤 ℎ𝑖𝑡𝑒 × 100 %
(18)
where p is the number of pixels.
3. Resin characterization
Resin cure kinetics model. Following the methodology of Khoun [57]
and Kratz [58], modulated differential scanning calorimetry (TA Instruments
DSC Q2000) was used to measure the heat flow of resin samples in dynamic
and isothermal conditions. Tests were performed under a nitrogen purge at a
flow rate of 50 mL/min. For each measurement, 10 ± 2 mg of neat resin was
sealed in aluminum hermitic DSC pans (Tzero, TA Instruments). The testing
conditions were chosen based on typical process conditions recommended by
the manufacturer. To obtain the total heat of reaction of the resin, four ramps
were performed at 1, 2, 3, and 5 °C/min from -60 °C to 250 °C, with a
temperature modulation of ± 0.5 °C/min. Four isothermal dwells were
performed at temperatures between 100 °C and 130 °C. Isothermal testing
was performed by heating the DSC cell rapidly to the dwell temperature then
holding until negligible heat flow was measured. The DSC cell was then
cooled to 20 °C, followed by a secondary heating to 250 °C at a constant
89
heating rate of 1.5 °C/min. This final step was performed to measure residual
heat of reaction.
Figure 23 The experimental data measured by DSC for the four dynamic temperatures and
the predictions obtained with the cure kinetics model.
Measured heat flow was converted into cure rate using the technique
described by Khoun [57]. The DSC heat flow signal was plotted as a function
of time, and a linear integration of the area under the heat flow curve was
used to calculate the total heat of reaction: 408.5 ± 8.8 J/g for the resin
system analyzed here. The degree-of-cure of the resin was obtained by
integrating the area under the curve of cure rate versus time. A cure kinetics
model was fit to the experimental data, with cure rate described by an
autocatalytic model with diffusion factor [59]:
𝑑𝛼 𝑑𝑡 = 𝐾 𝛼 𝑚 ( 1 − 𝛼 )
𝑛 1 + 𝑒𝑥𝑝 (𝐶 ( 𝛼 − ( 𝛼 𝐶 0
+ 𝛼 𝐶𝑇
𝑇 ) ) )
(19)
90
where α is the degree-of-cure (0 ≤ α ≤ 1), t is time in s, m and n are reaction
exponents, C is a constant, T is temperature in °C, and αC0 and αCT (1/K) are
fitting parameters. The initial degree-of-cure of the resin was assumed to be
0.005. The rate constant, K, accounts for an Arrhenius temperature
dependence:
𝐾 = 𝐴 𝑒𝑥𝑝 (
−𝐸 𝑎 𝑅𝑇
)
(20)
where A is the frequency factor of the cure reaction (1/s), Ea is activation
energy (J/mol), and R is the universal gas constant. Fig. 23 compares the
dynamic DSC data for the four experimental test conditions with the
predictions obtained using the kinetics model presented in Eqn. 19. The
numerical values of the model parameters are listed in Table 6. These values
were used as a starting point to develop a model for rheological behavior.
Table 6 The constants for the resin cure kinetics model that characterizes the resin system
PMT-F4A.
Rheological behavior model. Rheology experiments (TA Instruments
AR2000) performed on neat resin were used to fit a semi-empirical model for
91
evolution of viscosity [60]. Dynamic scans at heating rates of 2 – 5 °C/min
were performed using a 25 mm parallel plate geometry in oscillatory mode at
0.1% strain and 1 Hz. Resin sample thickness was maintained between 1 –
2 mm. To characterize the rheological behavior of the resin, a model
considering the influence of both temperature and the degree-of-cure was
implemented. The following equation presents the modified gel model used to
describe the evolution of resin viscosity [57,61]:
𝜇 = 𝜇 1
( 𝑇 )+ 𝜇 2
( 𝑇 )(
𝛼 𝑔𝑒𝑙 𝛼 𝑔𝑒𝑙 − 𝛼 )
( 𝐴 ′
+𝐵 ′
𝛼 +𝐶 ′
𝛼 2
)
(21)
𝜇 𝑖 ( 𝑇 ) = 𝐴 𝜇𝑖
𝑒𝑥𝑝 (
𝐸 𝜇𝑖
𝑅𝑇
)
(22)
where αgel is the degree-of-cure at the gel point and Aμi, Eμi, A′ , B′ , and C′ are
constants. From the temperature-time history measured in each experiment,
the degree-of-cure of the resin was calculated using the cure kinetics model
and the gel point was determined as αgel = 0.85. Equality of the storage and
loss shear moduli, G′ and G′′ , was used as a criterion for gelation [62].
After simple manipulation, Eqn. 19 can be expressed as a linear
relationship between viscosity and inverse temperature. Using this approach
and dynamic data collected prior to the gel point, a linear regression was
used to calculate the constants, Aμ1, Eμ1, Aμ2, and Eμ2. The remaining model
parameters, A′ , B′ , and C′ , were calculated using a least squares nonlinear
regression between viscosity and temperature data. Fig. 24 compares the
92
predicted and measured viscosity evolution as a function of time and
temperature. The model constants are summarized in Table 7.
Figure 24 The predicted and measured viscosity evolution with time and temperature.
Table 7 The constants for the rheological behavior model that characterizes the resin system
PMT F4A.
93
4. Resin flow model
4.1 Model development
When a stack of prepreg plies is heated, the decrease in resin viscosity
accelerates resin flow and impregnation of dry fiber regions. Impregnation
continues until either the flow fronts on either side of a dry fiber opening
meet, or the degree of cure (and, consequently, viscosity) increases to the
point that flow is impeded or arrested. Here, the evolution of the resin flow
front was modeled by considering saturated flow of resin within a rigid fiber
bed [3]. The model development relied on two assumptions: (i) unidirectional
flow of the resin (along the fiber direction) only, and (ii) infinite resin supply
(flow stops when both fronts meet and not when the initial quantity of resin
has been depleted). Combining the mass conservation equation and Darcy’s
law, the advancement of a saturated UD flow front, L, into a rigid fiber bed,
under constant applied pressure can be determined as follows [63]:
( 1 − 𝑉 𝑓 )
𝑑𝐿 𝑑𝑡 = −
𝐾 𝑟 𝜇 𝑟 Δ𝑃 𝐿
(23)
where μr is the resin viscosity (Pa·s) and ΔP is the applied pressure (Pa). For
the semi-pregs studied here, the applied pressure was assumed to be equal to
the vacuum pressure. The initial degree of impregnation of these semi-pregs
was negligible, and thus during initial stages of impregnation, the resin
pressure was also assumed equal to the vacuum pressure. This assumption
does not hold at high degrees of impregnation, as the fiber bed carries some of
the applied pressure. In that case, the applied pressure carried by the resin is
94
less than the vacuum pressure, reducing the pressure that drives the resin
into the fiber bed. The fiber volume fraction, Vf, of the prepreg was
determined using [64]:
𝑉 𝑓 =
𝑛 ∙ 𝑚 𝑓 𝑍 ∙ 𝜌 𝑓
(24)
where n is the number of plies, mf is the fiber areal weight (305 g/m
2
), and ρf
is the fiber density (1.75 g/cm
3
). Here, the value for Vf was calculated to be
0.6189 (fiber volume fraction of 61.89%). The constant, Kr, is a scalar and
represents the permeability of the fabric to the resin in the direction of flow.
In the UD fiber bed used in this work, the flow advances along the direction
of the fibers and between fiber bundles [3]. Considering a hexagonal
arrangement of the fibers (Fig. 25), the permeability for the flow parallel to
the fiber direction can be calculated using Gebart’s formula [65]:
𝐾 𝑟 =
8𝑟 𝑓 2
53
( 1 − 𝑉 𝑓 )
3
𝑉 𝑓 2
(25)
where rf is the average fiber radius, 5 μm. Here, the calculated value of Kr
was 5.45 × 10
-13
m
2
.
95
Figure 25 (a) Diagram of the hexagonal arrangement of fibers and labels for the “through-
thickness” and “in-plane” directions. (b) Diagram of the hexagonal arrangement of fibers and
a description of the average fiber radius, rf. (c) Micrograph of the hexagonal arrangement of
fibers.
The viscosity evolution of the epoxy resin was considered by
introducing a function μr(t). In this way, the advancement of the resin flow
front, L(t), was modeled considering the cure cycle. Then Eqn. 23 becomes:
1
2
𝐿 2
( 𝑡 ) = −
𝐾 𝑟 ( 1 − 𝑉 𝑓 )
∆𝑃 ∫
𝑑𝑡 𝜇 𝑟 ( 𝑡 )
𝑡 0
(26)
The integral of dt/μr(t) was determined from the rheological model
characterization of the resin, as discussed in Sec. III.3.
4.2 In situ observation
Conditions can arise during composite cure cycles that alter the
viscosity profile of the resin and prevent full resin infiltration, particularly
for semi-pregs with large surface openings, including (1) non-uniform ramp
rate, (2) inaccurate dwell temperature, (3) partial vacuum, and (4) excessive
out-time of resin. Below, achievable resin flow distances were evaluated for
each of these adverse conditions during cure. Flow during the one-hour room
temperature hold was assumed to be negligible and was not evaluated here.
96
Favorable cure conditions. Favorable cure conditions consist of a
uniform ramp rate, a suitable dwell temperature and time sufficient for resin
flow, full vacuum, and resin with minimal out-time. To achieve a uniform
ramp rate in this study, the oven was programmed to a temperature above
the target temperature. Here, the programmed temperature was 200 °C for a
target temperature of 121 °C. Once the laminate reached the target
temperature of 121 °C, the oven was quickly reprogrammed to maintain this
temperature. The rationale for using a higher set point temperature is
described in detail in the next section.
Figure 26 The resulting resin flow distances, both experimentally and from the model
predictions, for favorable cure conditions (uniform ramp rate, accurate dwell temperature, full
vacuum, and no aging of the resin) at the three different ramp rates: 0.5, 1.0, and 2.0 °C/min.
Fig. 26 shows measured and predicted resin flow distances for ramp
rates of 0.5, 1.0, and 2.0 °C/min. The fastest ramp rate (2.0 °C/min) resulted
in the longest resin flow distance– ~1.85 mm, corresponding to a total surface
97
opening diameter of 3.70 mm. In contrast, the slowest ramp rate (0.5 °C/min)
yielded the shortest flow distance – ~1.50 mm, or a diameter of 3.00 mm (a
difference of 0.70 mm when compared to the value from the fastest ramp
rate). The resin flow rates also decreased with decreasing ramp rate (1.11 ×
10
-2
mm/min for 0.50 °C/min versus 3.99 × 10
-2
mm/min for 3.0 °C/min).
Comparing measured and predicted values [Fig. 26], the model
accurately captured variations in the resin flow front caused by the process
conditions selected. Under favorable conditions, the maximum feature size for
the resin system studied was 3.8 mm. Additionally, the data showed that a
slow ramp rate can result in insufficient resin infiltration, as flow distance
decreased with ramp rate.
Non-uniform ramp rate. Normally in composite processing, an oven is
programmed to reach a set dwell temperature with a specific ramp rate.
However, there is typically a lag, such that the temperature read by the
thermocouple in the oven reaches the target dwell temperature before the
laminate plies. In practice, the ramp rate measured by a thermocouple within
the laminate will decrease as it approaches the dwell temperature.
98
Figure 27 The resulting resin flow distances, both experimentally and from the model
predictions, for: (a) non-uniform ramp rate at 0.75, 1.5, and 3.0 °C/min, (b) inaccurate dwell
temperature at 111, 121, and 131 °C, (c) poor vacuum at 80%, 90% and 100% vacuum, and
(d) resin that has undergone aging where the degree of cure was 0, 0.15, or 0.30.
To explore the influence of this temperature variation on resin flow,
the oven cure cycle was programmed to reach a dwell temperature of 121 °C
with ramp rates of 0.75, 1.50, or 3.00 °C/min. The resulting flow fronts, both
experimental and model predictions, are illustrated in Fig. 27a. At the fastest
ramp rate (3.00 °C/min), the measured data matches the model predictions.
At this higher rate, the ramp rate of the laminate was maintained until the
dwell temperature was reached. However, at the slower ramp rates (0.75 and
1.50 °C/min), the measured data did not reach the flow distances predicted by
the model due to differences in oven and laminate temperatures. During the
99
1.50 °C/min ramp, the measured ramp rate shifted to 0.50 °C/min at 105 °C
and remained at this rate until reaching a dwell temperature of 121 °C. The
adjusted model prediction in Fig. 27a includes the ramp rate shift at 105 °C.
This decrease in ramp rate also decreased the flow distance, from 1.85 mm to
1.55 mm (or a surface opening diameter of 3.70 mm to 3.10 mm). During the
0.75 °C/min ramp, the rate shifted to ~0.25 °C/min, starting at 80 °C until
121 °C (see the adjusted model prediction in Fig. 27a). This decrease in ramp
rate decreased the flow distance from 1.70 mm to 1.35 mm (or a surface
opening diameter of 3.4 mm to 2.7 mm), a difference of 0.35 mm (or 0.7 mm in
diameter).
A slower ramp rate led to a larger gap between the desired and actual
(non-uniform) ramp rate. This ultimately reduced the resin flow distance.
Values for the predicted flow distances for the 1.5 °C/min and 3.0 °C/min
ramp rate, interestingly, were both the same as the 1.85 mm (or diameter of
3.70 mm) distance for the 2.0 °C/min ramp rate described in the section
above. This result indicates that flow distance did not increase with
increasing ramp rate beyond ~1.50 °C/min. Thus, for resin system used here,
3.70 mm was the maximum allowable diameter of a surface opening required
to achieve full infiltration.
Inaccurate dwell temperature. In practice, the laminate temperature
can differ from the programmed dwell temperature due to inevitable thermal
gradients, faulty equipment, or location in the oven (i.e., near a vent hole or a
100
heating element). To examine the effect of departures from dwell
temperature, a series of experiments were carried out in which the dwell
temperature was programmed to be 111 °C, 121 °C, or 131 °C, all achieved
via a ramp rate of 1 °C/min. The resulting flow distances are presented in
Fig. 27b. The maximum flow distance did not differ between 121 °C and
131 °C, at 1.80 mm (or 3.6 mm diameter). A slight decrease in flow distance
was observed with a dwell at 111 °C, 1.70 mm (or a surface opening diameter
of 3.4 mm). Therefore, the effect of a 10 °C decrease in dwell temperature
(reducing flow distance by 0.10 mm) was less than the effects of changes in
ramp rate (reducing flow distance up to 0.70 mm), whether under a uniform
or non-uniform condition. As with ramp rate, an increase in dwell
temperature did not produce additional changes in flow distance above a
threshold value (here, 121 °C).
Partial vacuum. In practice, the condition of partial vacuum frequently
arises, most often because of bag leaks, and this condition reduces the driving
force for resin flow, the difference between atmospheric pressure and bag
pressure, or the compaction pressure. In this section, the bag pressure was
adjusted to achieve specific levels in relation to the maximum pressure
achieved by the pump: 80% (77.2 kPa), 90% (86.9 kPa), and 100% (96.5 kPa).
The ramp rate was 1 °C/min, and the dwell temperature was programmed as
described above under “favorable cure conditions.”
101
The resulting flow distances are presented in Fig. 27c. With each
subsequent drop in the vacuum level, a decrease in maximum flow distance
and flow rate occurred. At 80% vacuum, the maximum flow distance
decreased to 1.60 mm (or 3.4 mm in surface opening diameter), compared to
1.80 mm (or 3.6 mm diameter) at full vacuum. However, the maximum flow
distances both occurred at ~84 min, which was due to the decrease in resin
flow rate at reduced vacuum. The flow rate for the 80% vacuum condition
decreased to 1.89 × 10
-2
mm/min, compared to 2.11 × 10
-2
mm/min at 100%
vacuum. The reduction in vacuum pressure decreased compaction, which
ultimately decreased the resin flow rate. Note that reduced vacuum pressure
can also retard or reduce air evacuation from dry regions of prepreg plies, and
residual air can prevent resin from infiltrating the fiber bed. Air entrapped in
this way, while not accounted for in the flow model, may contribute to the
reduced flow distances observed.
Out-time. When resin has accrued minimal out-time, minimum
viscosity and maximum flow time generally can be achieved for a given cure
cycle, and flow dynamics allow for full wet-out. However, large composite
parts can accrue weeks of out-time during lay-up, advancing the degree of
cure of the pre-catalyzed resin [56,66–68]. As a consequence, the minimum
viscosity increases with out-time, and the amount of time the resin will flow
decreases during the heated portion of cure cycle, particularly the first dwell.
102
The flow distances for different aging periods are shown in Fig. 27d.
The rheological behavior model was manipulated by adjusting the initial
degree of cure. Here, the laminates underwent a cure cycle of 1 °C/min with a
dwell temperature of 121 °C (programmed as described above under
“favorable cure conditions”). For fresh resin (no out-time), the maximum flow
distance was 1.75 mm (surface opening diameter of 3.5 mm). At α0 = 0.15, the
maximum flow distance decreased to 1.25 mm (surface opening of 2.5 mm), a
reduction of 0.50 mm compared to fresh resin. At α0 = 0.30, the maximum
flow distance decreased to 1.10 mm (surface opening diameter of 2.20 mm), a
reduction of 0.65 mm compared to fresh resin. The reduction in maximum
flow distance from fresh resin to α0 = 0.30 was the largest amongst all the
adverse process conditions evaluated. Note that the flow rate did not change
with an increase in out-time.
5. Room temperature debulk
Design of semi-pregs requires assessment of the efficiency for air
evacuation from the laminates. Larger surface openings yield more rapid air
evacuation but require longer resin flow distances. In principle, prepreg
design must therefore balance the objectives of rapid air evacuation and
complete saturation of the fiber bed.
In Sec. III.4, maximum surface opening dimensions were identified for
a range of manufacturing conditions. The largest opening diameter (3.7 mm)
was associated with a uniform ramp rate of 2 °C/min and represents a
103
judicious balance of the two objectives above. The smallest opening (2.2 mm)
was associated with resin aged to o = 0.30 (equivalent to an out-time of 49
days for CYCOM
®
5320-1 [56]), and represents a worst-case scenario in which
the two objectives are not balanced.
Figure 28 Images of semi-preg made from each of the chosen diameters: (a) 0.5 mm,
(b) 2.0 mm, and (c) 4.0 mm.
Permeability measurements of air evacuation in the through-thickness
direction were conducted for semi-pregs with the diameters of the two cases
described above. In addition, a third diameter, 0.5 mm, was included to
understand air evacuation where resin infiltration was not likely to be
inadequate. Images of prepreg with each of the chosen diameters are shown
in Figs. 28a-c. The permeability values obtained from these experiments were
used to determine the amount of time required to evacuate 99.99% of air in
the laminate, as described below.
104
Table 8 The average and standard deviation of the effective permeability, K, of laminates
made with 4, 8, and 16 plies of semi-preg with surface opening diameters of 0.5, 2.0, and
4.0 mm.
Measured permeability. A summary of the permeability values of the
4-, 8-, and 16-ply laminates versus surface opening diameter is presented in
Table 8. The effective permeability, K, value decreased with decreasing
diameter, as expected. For prepregs with fixed surface opening sizes,
permeability values were independent of the number of plies. This finding
has implications for fabrication of thick parts, where typically longer room
temperature vacuum holds (debulks) are required for prepreg stacks with
reduced permeability (K).
Debulk time. A method typically used to evacuate the air entrapped
during the lay-up process of the composite structure is to use debulking,
which typically exerts vacuum pressure on the pre-cured lay-up at room
temperature. A model for one-dimensional gas transport in composite
laminates based on Darcy flow and the ideal gas law is presented below
105
[40,41,69]. A simple expression for the mass fraction of gas removed as a
function of time and general scaling laws for gas evacuation in laminates was
developed. Time, t, required to reach a mass fraction of gas remaining in the
laminate, m/m0, as a function of the ratio of the width of the laminate
squared, W
2
, to the permeability, K, is described as:
𝑡 =
𝜇 𝑝 0
𝑊 2
𝐾 [−
1
0.9
ln (
𝑚 𝑚 0
)]
1
0.6
(27)
where μ is dynamic viscosity of air (1.85 × 10
-5
Pa·s at room temperature) and
p0 is the initial pressure (here, assumed 95% vacuum or 96 × 10
3
Pa).
Figure 29 The time to evacuate 99.99% of air in 4-, 8-, and 16-ply laminates for surface
opening diameters between 0.5 – 4.0 mm.
The time to evacuate 99.99% of air in 4-, 8-, and 16-ply laminates for
surface opening diameters between 0.5 – 4.0 mm is presented in Fig. 29. The
largest surface openings (diameter = 4.0 mm) required only 26 s for
106
evacuation of air from 4 plies of prepreg and 2.5 min for 16 plies. For a
diameter of 2.0 mm, the debulk time increased to 5.5 min for 4 plies, and
20.5 min for 16 plies. Finally, for the smallest surface openings (diameter =
0.5 mm), the evacuation time was 18.5 min for 4 plies and 180 min for 16
plies. However, all these times were far less than the time required to debulk
prepreg with continuous resin film. Using data from a previous study [36,37],
the time required to evacuate 99.99% of air from 4 plies or 8 plies of prepreg
made with continuous film was 24.5 and 32 hr, respectively.
These results indicate that air evacuation was more efficient in semi-
pregs with larger resin feature dimensions. The debulk time prior to the cure
cycle was therefore greatly reduced when using larger surface openings. The
trade-off, however, was that overly large feature dimensions can limit resin
infiltration and prevent fiber wet-out during the cure cycle.
6. Experimental validation
Laminates were fabricated from prototype semi-pregs to determine the
debulk times required and the internal porosity after cure. The Favorable
Cure Conditions (FCC) case consisted of an 8-ply laminate made from semi-
preg featuring a striped pattern of resin with resin strips of 3.7 mm
separated by the same distance. The feature dimensions were determined
from Sec. III.4.2 (“favorable cure conditions”), where the maximum dry space
distance for full resin infiltration was 3.7 mm for the FCC (i.e., fast ramp
rate, vacuum at highest efficiency, fresh resin, etc.). Permeability
107
measurements for this laminate design resulted in a minimum debulk time of
slightly less than 2 min to evacuate 99.99% of the air. The Aged Resin (AR)
case consisted of the same laminate design as the FCC case, except that the
resin was aged to o = 0.3. The minimum debulk time was the same for this
case. The Adjusted AR case consisted of resin strips separated by 2.2 mm.
Feature dimensions were determined using the resin flow front model in
Sec. III.4.2 (“out-time”), where the maximum dry space distance for full resin
infiltration of resin aged to α0 = 0.30 was 2.2 mm. Permeability tests for this
laminate design resulted in a minimum debulk time of 2 min 27 s to evacuate
99.99% of the air.
The laminate surface for all three cases resulted in zero porosity (see
Figs. 30a-c). However, for the AR case, evidence of the surface openings was
visible as vestiges of the pattern. This pattern indicated local and periodic
variations in resin content at the surface, from resin-rich regions (lighter
areas) to resin-lean area (darker regions). However, dry exposed fibers (i.e.,
surface porosity) were not present. Evidence of the surface openings for the
FCC case was also present, but the features were much less distinct. For the
Adjusted AR case, no surface defects were observed.
The three panels showed marked differences in internal porosity,
however. The FCC case displayed an average void content of 0.1% (Fig. 30a).
On the other hand, the average internal porosity in laminates made from
semi-pregs featuring the AR case was 1.6% (Fig. 30b). The porosity in this
108
case was expected to be much greater than in the FCC case, as the results
from this study indicated that aged resin required resin feature spacings that
were no more than 2.2 mm. The internal porosity of the Adjusted AR case,
designed to account for this prediction, dropped to 0.3%, a decrease of 1.3%
from the AR case.
Figure 30 Images of the surface defects and internal void content of the prototype semi-pregs
fabricated using the following cases: (a) Favorable Cure Conditions, (b) Aged Resin, and
(c) Adjusted Aged Resin.
Experimental results described above validated the model predictions
using the methodology presented. The debulk time calculated for both the
FCC and the Adjusted AR cases were sufficient to fully evacuate air from the
laminates. In addition, the feature dimensions fabricated for both the FCC
case and the Adjusted AR case resulted in full fiber bed infiltration, whereas
the AR case resulted in insufficient fiber bed wet-out. The results and the
methodology described here can be used to guide the design of discontinuous
109
resin patterns for OoA prepregs to ensure both rapid air evacuation and
complete resin saturation of fiber beds.
7. Conclusion
A methodology is described for determining the maximum dry space
dimensions for semi-pregs produced using a given resin system. While the
work focused on a single resin system, the method itself can be applied to any
thermoset epoxy. The methodology stems from earlier work [36,37], in which
a technique was described for creating discontinuous resin patterns of
arbitrary shapes and sizes via polymer film dewetting. That technique
opened a large design space for prepreg format (semi-preg resin distribution).
While earlier work [55] focused on the design and evaluation of semi-pregs
based solely on efficient air evacuation, the present work addresses the
design of semi-pregs that takes into account the objective of fiber bed
saturation.
For the resin studied, a cure cycle was identified that yielded a defect-
free microstructure, and surface openings as large as 3.7 mm ensured full
fiber bed wet-out. The effects of adverse cure conditions, including inadequate
temperature profile, reduced vacuum quality, and resin with accrued out-
time, were explored to determine semi-pregs designs to accommodate reduced
resin flow distances. In contrast with the favorable cure condition, the
maximum surface opening required to ensure full fiber bed wet-out dropped
to 2.2 mm for the worst-case scenario (of those considered). The influence of
110
these dry space dimensions on the effective air permeability of the laminates,
and ultimately, the time required for evacuation, was determined. The
results show that in practice, the design space can be greatly reduced and
defined by eliminating patterns that do not allow for (a) full resin infiltration
of the fiber bed or (b) sufficient air evacuation.
The model presented provides guidance to specify the dimensions of
resin distribution patterns sufficient for both efficient air evacuation and full
resin infiltration. However, some aspects of prepreg design were not
considered here. First, the evaluation of the resin flow fronts under adverse
conditions focused on singular contributions. The combined effects of multiple
adverse conditions (such as aged resin, non-uniform ramp rate, and poor
vacuum quality) will exacerbate the issue, and impede full resin infiltration.
Second, only one-dimensional resin infiltration was considered. However, the
mechanism for resin infiltration of more complex fiber beds (particularly
woven fabrics) is expected to differ and involve 2D or 3D multi-scale flow.
Finally, changing surface opening dimensions ultimately changes the
thickness of the resin because of resin recession from nucleation sites during
dewetting. Larger surface openings require thicker resin and, ultimately,
yield laminates with a larger bulk factor. Prepreg with a large bulk factor can
cause laminate wrinkling at part corners and degrade mechanical properties.
Understanding how resin feature dimensions affect prepreg bulk factor and
cured part quality is useful to part fabricators and requires further study.
111
The bulk factor of these resins may be reduced by increasing the degree of
impregnation. However, it remains unclear how to incorporate impregnation
while retaining the desired pattern. Nonetheless, the method described here
is versatile and affords opportunity to design and create discontinuous resin
patterns and thus tailor resin distributions to fiber bed features and even
part geometry.
The current study can be used to inform the design of VBO semi-preg
for virtually any resin system. OoA/VBO prepreg processing presently suffers
from a lack of manufacturing robustness, often yielding unacceptable defect
levels when conditions are adverse or not adequately controlled. The method
described here outlines a pathway to determine favorable (and, eventually,
optimal) designs for semi-pregs. Such prepregs can potentially restore
robustness to composite manufacturing via VBO processing, yielding low
defect levels without autoclave pressures. In addition, dewetting is
potentially backwards-compatible with hot-melt prepregging, since in
principle, imprint/de-wet steps can be incorporated into existing prepreg
production lines. The robustness imparted by the methods presented can, in
turn, expand applicable uses of VBO prepregs both in aerospace and non-
aerospace applications.
* This study has been submitted for publication.
112
CHAPTER IV:
Design and Application of
Discontinuous Resin Distribution Patterns
for Vacuum-Bag-Only Prepregs
1. Background
The application of discontinuous resin on different fibers beds was
evaluated for use in commonly encountered manufacturing conditions for out-
of-autoclave (OoA)/vacuum-bag-only (VBO) processing of composite prepregs,
and design considerations were determined for the fabrication of such
prepregs. The study was motivated by limitations of current methods for
producing composite parts for aerospace, which are commonly cured in high-
pressure autoclaves (heated pressure vessels). These methods consistently
yield defect-free parts, but autoclave processing has drawbacks, including
high capital and operational cost, size restrictions, limited throughput
(production bottlenecks), and high resource use (e.g., energy, nitrogen gas).
Alternative manufacturing methods are sought to reduce costs, overcome
limitations, and accelerate production rates, particularly in aerospace.
VBO processing of composites offers a viable alternative to
conventional autoclave cure methods [4,70]. VBO processing is an OoA
method in which prepreg is vacuum bagged and cured in an oven. VBO
processed parts can match the quality of parts manufactured in an autoclave
under favorable conditions. However, adverse process conditions, including
113
poor vacuum, incomplete air evacuation, and/or high humidity, often yield
defects, particularly porosity, which degrade mechanical performance [6].
Because VBO processing is limited to a consolidation pressure of 0.1 MPa
(1 atm), air and other gases can be trapped during layup. Once the resin gels,
trapped bubbles remain, resulting in potentially unacceptable porosity levels
(>1%). For VBO processing to gain wider acceptance in high-performance
applications, the process must consistently yield low porosity levels.
Background. Previous work has shown that use of OoA/VBO prepregs
with a discontinuous resin pattern (semi-pregs) virtually eliminates porosity
caused by entrapped air or moisture, without an autoclave cure [2]. For
example, Grunenfelder [1,14,71] showed that even under adverse process
conditions, parts cured using semi-preg contained near-zero internal porosity
and no surface defects. In contrast, conventional VBO prepreg (with
continuous resin films) yielded unacceptable levels of porosity (3-8%).
Tavares [3] measured the through-thickness air permeability of a commercial
semi-preg (“Zpreg”) and an equivalent unidirectional (UD) prepreg
constructed with continuous resin. The permeability of the semi-preg
material was three orders of magnitude greater than that of the continuous
film prepreg before and during the cure cycle, a result attributed to a network
of dry interconnected pore spaces in the semi-preg design. As these studies
demonstrate, prepregs featuring discontinuous resin films increase the
capacity for air evacuation in the z-direction (transverse) by creating efficient
114
egress pathways. This effect is attributed to the much shorter air evacuation
distances in the through-thickness direction (on the order of millimeters) as
compared to the in-plane direction (on the order of meters). Discontinuous
resin not only increases the efficiency of air evacuation but also entraps less
air between plies. VBO prepregs with discontinuous resin have the potential
to mitigate process-induced defects and impart robustness to the
manufacture of aerospace composites via OoA methods.
Semi-pregs can be produced by different methods. We have previously
reported one such method, which yields finely tuned resin distribution
patterns and high through-thickness permeable prepregs [36,37]. The method
relies on polymer film dewetting on a low surface energy substrate [19].
Using this method, resin film is first perforated on a silicone-coated backing
paper (substrate) to create an array of nucleation sites. The film is then
heated, causing resin to recede from the nucleation sites. The resin recession
is driven by the surface tension between the low surface energy substrate
(the silicone-coated backing paper) and the resin [16–19]. The dewetted resin
is transferred onto a fiber bed by briefly pressing the constituents in a
hydraulic press. This technique can be used to create a variety of resin
patterns, including stripes, islands, and grids. In addition, the resin patterns
created with this technique can in principle be applied to any dry fiber bed.
Objectives. The goal of this study is to assess the design and
application of the semi-preg format in various common but challenging
115
manufacturing conditions. Previous work has demonstrated the benefits of
discontinuous resin application in prepreg processing. What has yet to be
developed is a broad range of product forms for semi-pregs. The key
developments described in this work are: (1) the fabrication of semi-preg
produced from fiber beds with various fiber types, weave types, and areal
weights, and (2) the production of complex parts using a woven semi-preg
format. In this context, design considerations and limitations of semi-preg
fabrication are also discussed.
Results reveal that low porosity laminates can be fabricated using
semi-pregs with various woven fiber bed architectures. The use of
discontinuous resin increases overall resin thickness, which slightly increases
the bulk factor. To address concerns regarding the increase in bulk factor,
curved laminates with concave and convex corners were produced.
Ultimately, part quality did not decrease with the use of semi-pregs in the
production of complex shapes. Finally, initial design guidelines are offered
that outline the dimensional limitations of resin patterns, degree-of-
impregnation (DOI) of semi-preg, placement of constituents considering the
general direction of the fiber bed and the resin pattern, and the use of the
polymer film dewetting technique on some of the most widely used resin
systems. This work provides the first guidelines for the design of semi-pregs –
an important modification of OoA prepreg format that has the potential to
enhance the robustness of VBO processing of prepreg laminates.
116
2. Materials and methods
Prepreg fabrication. A total of seven fiber beds were evaluated. Three
fiber areal weights (200, 370, and 670 GSM) were selected. Unless otherwise
noted, fabrics were carbon fiber. At 200 GSM, four fabric types were
evaluated: plain weave, glass fiber plain weave (Fibre Glast Development
Corporation, Ohio, USA), 3K twill (Fibre Glast), and spread tow (Textreme,
Sweden). The production of spread tow fiber beds involves the spreading of a
tow into a thin and flat UD tape and then weaving into a fabric. For
370 GSM, a 5-harness satin (Fibre Glast) and a 6K twill (Fibre Glast) fiber
bed were evaluated. Finally, at 670 GSM, a 12K twill fiber bed (Fibre Glast)
was evaluated. A twill weave was selected for each areal weight to enable
direct comparisons. The designations 3K, 6K and 12K refer to the numbers of
fibers within each tow (i.e., 3,000, 6,000 or 12,000 fibers per tow).
Semi-preg and conventional prepreg plies were fabricated with epoxy
resin (PMT-F4, Patz Materials & Technology, California, USA) at a resin
content of 35-36%. Semi-pregs were produced via selective dewetting of resin
film [36,37]. With the neat resin film on the silicone-coated backing paper (a
low surface energy substrate), nucleation sites were introduced using either a
hand-held spike roller (HR-2, Robert A. Main & Sons, Inc., New Jersey, USA)
or a box cutter. The spike roller pins were spaced at 6.35 mm, and the roller
was passed over the entire film in straight passes. Using a box cutter, the
resin on the silicone-coated backing paper was scored in straight passes. The
117
resin film was then placed in an air-circulating oven (Blue M Oven, Thermal
Product Solutions, Pennsylvania, USA), to dewet and grow the openings at
the nucleation sites. The resin film was heated for 2 min at 104 °C.
Subsequently, the resin film was attached to the fiber beds by pressing the
constituents briefly in an unheated hydraulic press (G30H-18-BCX, Wabash
MPI, Indiana, USA). Images of discontinuous resin patterns produced using a
spike roller, after application to various fiber beds, are presented in Fig. 31.
Figure 31 Micrographs prior to cure of the discontinuous resin distribution (fabricated via
polymer film dewetting) on each of the fiber bed types evaluated.
To produce flat laminates, 16 plies of prepreg were stacked in a [0/90]4s
sequence, where applicable. Initially, each ply was cut to 150 mm × 150 mm,
and after stacking, the edges of the stack were trimmed, resulting in
dimensions of 140 mm × 140 mm. The laminates were vacuum bagged using
standard consumables. Rather than utilizing edge breathing (common
118
practice with commercial VBO prepregs), the perimeter of each laminate was
sealed with vacuum tape to restrict air evacuation solely to the through-
thickness direction. Sealed edges were used to approximate process
conditions that limit or prevent in-plane air evacuation (e.g., large or complex
parts, parts with ply drops or corners). Laminates were cured according to
the recommended cure cycle: a ramp of 1.5 °C per min to 121 °C followed by a
two-hour dwell.
Commercial prepregs are produced from various resin systems,
including cyanate ester and bismaleimide (BMI). To assess the compatibility
of semi-preg fabrication via resin dewetting with different resin systems, the
technique was employed using a cyanate ester resin film (54 GSM, PMT-F27,
Patz Materials & Technology, California, USA), a BMI resin film (71.5 GSM,
RS-8HT, Toray, California, USA), and a toughened epoxy resin film (71.5
GSM, CYCOM 5320-1, Solvay, USA).
Porosity measurements. To evaluate the surface porosity of cured
laminates, images of regions 38 mm × 38 mm were recorded using a digital
microscope (VHX-5000, Keyence Corporation of America, California, USA).
These images were recorded at three locations across the laminate surface.
For bulk void content, mutually orthogonal sections were prepared from the
center of each laminate. Cross-sections were polished, and regions 5 mm × 20
mm were imaged.
119
Images were analyzed using software (ImageJ) to determine void
contents. A series of steps was performed manually to produce binary
porosity images. Using the software, images were converted to black pixels
for voids and white pixels for the rest of the laminate. The percent porosity
was calculated from the number of black and white pixels:
𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦 ( %) =
𝑝 𝑏𝑙𝑎𝑐𝑘 𝑝 𝑤 ℎ𝑖𝑡𝑒 + 𝑝 𝑏𝑙𝑎𝑐𝑘 𝑥 100 %
(28)
where p is the number of pixels.
Bulk factor. Bulk factor is defined as the ratio of the initial thickness,
ti, of the prepreg stack (prior to cure) to the final thickness, tf, of the laminate
(after cure):
𝐵𝑢𝑙𝑘 𝐹𝑎𝑐𝑡𝑜𝑟 =
𝑡 𝑖 𝑡 𝑓
(29)
To calculate the bulk factor, the thickness of a laminate was measured prior
to cure using a caliper at four locations around the perimeter. After cure, the
laminate was sectioned, and a caliper was used to measure the thickness of
the laminate at four locations throughout the cross-section.
Complex shapes. To fabricate curved laminates, a custom test fixture
[72] was utilized, featuring a 60° concave and a 60° convex corner (with a
rounded radius of 9.5 mm), yielding parts with common geometric
complexities [Fig. 32a]. Tooling was machined from a single billet of
aluminum to avoid leaks, and tool surfaces were fine polished. 8-ply
120
laminates with dimensions of 75 mm × 130 mm were produced using a
370 GSM 5-harness satin fiber bed.
Figure 32 (a) A custom test fixture, which allowed a 60° concave and a 60° convex corner
laminate (with a rounded radius of 9.5 mm) to be manufactured simultaneously. (b) Locations
of measurement along the flanges and corners to calculate the coefficient of variation (CoV).
An accepted metric for corner quality is thickness variability between
the flanges and the curved portion. Parts produced at a concave corner are
expected to have corner thickening from pooling of resin. On the other hand,
parts made at a convex corner are expected to have corner thinning. To
determine the thickness variability, nine locations were measured at the
flanges and corner, as illustrated in Fig. 32b. The average thickness, x̄, was
calculated using the following equation:
𝑥 ̅ =
∑ 𝑥 𝑖 𝑛
(30)
where xi is the thickness at each individual location and n is the number of
measurement locations. The coefficient of variation (CoV) was then calculated
using the following equation:
𝐶𝑜𝑒𝑓𝑓𝑖𝑐𝑒𝑛𝑡 𝑜𝑓 𝑉𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 ( 𝐶𝑜𝑉 ) =
1
𝑥 ̅ √
∑( 𝑥 − 𝑥 ̅)
2
( 𝑛 − 1)
(31)
121
where the variables are defined identically as the previous equation.
In situ monitoring. To monitor the resin flow front during processing,
the surfaces of the semi-pregs were tracked in situ using the technique
described by Hu et al. [30,31]. Four-ply semi-preg stacks using a 5-harness
satin fiber bed were fabricated using either a grid or a striped pattern. The
stacks were laid against a glass window of an oven and vacuum-bagged with
standard consumables, sealing edges to prevent gas egress. Resin flow was
monitored and recorded during a standard cure cycle using a digital
microscope and time-lapse video acquisition (Dino-Lite US, Dunwell Tech,
Torrance, CA, USA). Temperature was monitored with a USB thermocouple
data logger (Lascar Electronics EasyLog EL-USB-TC-LCD), which was
attached adjacent to the laminate and directly on the glass tool plate.
3. Results
3.1 Quality analysis of semi-preg formats
Key differences exist between UD and woven fiber beds. Woven fabric
exhibits crimp, or fiber waviness, whereas crimp in UD fiber beds is
negligible. Crimp varies based on weave type and can range from minimal
(i.e., spread tow) to large (i.e., plain weave). High crimp generally reduces
laminate strength [73]. Secondly, woven fiber beds contain pinhole openings
between crossing tows, which enhance transverse air evacuation, while UD
fiber beds do not. Finally, flow of resin in a woven fiber bed is
multidimensional. Resin flows more rapidly through pinholes and in the
122
depressions at tow intersections. Resin therefore flows through pinholes and
depressions (macro-flow) before saturating tows (micro-flow). This dual-scale
flow does not occur in UD fiber beds. These differences are expected to affect
both gas and resin flow in semi-preg materials.
Figure 33 Micrographs of the surface and the cross-sections of each of the semi-preg formats
evaluated. The fiber bed types are indicated as: PW (plain weave), GF (glass fiber plain
weave), 3KT (3K twill), ST (spread tow), 5HS (5-harness satin), 6KT (6K twill), and 12KT
(12K twill).
123
Both conventional prepregs and semi-pregs were fabricated with a
variety of fiber types, weave types, and fiber bed areal weights, after which
cured laminate quality was assessed. As described in Sec. IV.2, laminates
were fabricated from three areal weights (200, 370, and 670 GSM), four
weave types (plain weave, twill, satin weave, and spread tow), and two fiber
types (carbon and glass fibers). Micrographs of surface and internal porosity
of all samples are presented in Fig. 33, with the seven fiber bed types
indicated as PW (plain weave), GF (glass fiber plain weave), 3KT (3K twill),
ST (spread tow), 5HS (5-harness satin), 6KT (6K twill), and 12KT (12K twill).
The first and second columns of Fig. 33 depict the surface and internal
porosity, respectively, of laminates produced with conventional OoA prepreg
formats (continuous film) using the seven different fiber beds. All seven
conventional laminates exhibited high levels of surface and internal porosity,
both of which were apparent via visual inspection. In all seven conventional
laminates, porosity was concentrated at locations corresponding to fabric
pinholes and the depressions at tow intersections. During cure, these features
create points of low pressure towards which resin and air migrate.
Specifically, in the conventional spread tow laminate (fourth row, first and
second columns), in addition to porosity at the pinholes and depressions,
intra-tow porosity was also observed due to the flatness of this fabric. In
these prepregs produced with spread tow fabric, the surface porosity observed
124
resembled defects observed with UD fiber beds, as shown in previous work
[36,37].
The third and fourth columns of Fig. 33 depict the surface and internal
porosity, respectively, of laminates produced with semi-preg using the seven
different fiber beds. Each of the four semi-preg laminates fabricated with 200
GSM fiber beds displayed near-zero surface and internal porosity, as depicted
in the first through fourth rows of the third and fourth columns. Notably, in
the laminate made from spread tow semi-preg (fourth row, third column),
surface porosity was not detected, but vestiges of the applied discontinuous
resin pattern were observed on the cured part surface. Similar surface
features have been reported in previous studies of semi-preg materials made
from UD fiber beds [36,37]. Semi-preg laminates produced with 370 and 670
GSM fiber beds exhibited both surface and internal porosity after curing, as
depicted in the fifth through seventh rows of the third through fourth
columns. The surface and internal porosity was attributed to incomplete
saturation of the fiber bed during cure, indicating that resin flow distances
were too long to fully infiltrate. This intra-tow (flow-induced) porosity was
greater for 670 GSM than for 370 GSM fabrics, indicating a potential upper
limit on fabric thickness for the present format of semi-preg processing. No
inter-ply porosity (gas induced) was observed in any semi-preg samples.
125
Figure 34 (a) Surface porosity and (b) internal porosity of each of the prepreg and semi-preg
formats evaluated. The fiber bed types are indicated as in Fig 33.
A graphic summary of the surface and internal porosity for all samples
is shown in Fig. 34, with the fiber bed types indicated as in Fig. 33. The
surface porosity (Fig. 34a) of all conventional prepreg samples was greater
than 1%, except for the 3K twill sample at 0.4%. All semi-preg samples, in
contrast, exhibited less than 1% surface porosity, with most samples
displaying no surface voids. While surface porosity was visible for the fiber
beds with larger areal weights (≥370 GSM), the measured surface porosity
was ~0.1%. Internal porosity (Fig. 34b) for all samples produced using
prepreg with continuous film was 1% or more, with porosity levels increasing
with fiber areal weight. Except for the panel produced with 12K twill, the
semi-preg samples showed less than 1% porosity. Interestingly, the 12K twill
laminate produced from semi-preg resulted in more porosity (12.0%) than the
sample fabricated from prepreg with continuous resin (7.1%). These findings
indicate that an upper limit on fabric weight may exist for successful semi-
126
preg processing, or that the cure process or semi-preg format might require
modification.
The potential benefits of semi-preg materials have been previously
reported, but a versatile range of product forms has not been developed.
Here, we have demonstrated that semi-preg can be fabricated from a range of
fiber types, weave types, and fiber bed areal weights. The laminates produced
from semi-preg with low and intermediate areal weight fabrics showed lower
porosity (< 1%) than comparable laminates produced with conventional
format prepreg. However, in laminates produced from high areal weight
fabrics, porosity increased markedly (12.0%), because fiber bundles were too
thick to fully impregnate during cure. To resolve this issue, further work
must be undertaken, such as employing a higher degree of impregnation to
reduce the flow distance (which will be discussed in Sec. IV.3.4) or utilizing a
tailored resin system with longer flow times at low viscosity.
3.2 Bulk factor
Bulk factor also was calculated for each laminate described in
Sec. IV.3.1. Bulk factor is relevant for fabrication of contoured parts because
a large bulk factor can cause wrinkling and/or bridging of plies. In principle,
a bulk factor of 1.0 represents no change in thickness during cure, a
characteristic generally preferred. However, prepregs that incorporate dry
regions for air evacuation intrinsically possess bulk factors > 1. Bulk factor is
further increased when discontinuous resin distributions are used [36,37].
127
For the semi-preg and prepreg formats evaluated (Fig. 35), bulk factors
varied with weave type because of the inherent differences in fiber bed
characteristics (i.e., crimp). In addition, the bulk factor generally increased as
areal weight increased. The average difference in bulk factor between
conventional prepreg and semi-preg with the same fiber type was ~0.1. The
fiber bed architecture that produced the largest overall bulk factor was
5-harness satin (5HS). This fiber bed was selected for further study via
fabrication of panels with concave and convex corner geometries (discussed in
Sec. IV.3.3).
Figure 35 The measured bulk factor of each of the prepreg and semi-preg formats evaluated.
3.3 Complex shapes
Laminates with complex geometries were evaluated to determine if
and how the bulk factor increase associated with a semi-preg design affected
part quality. Laminates with concave and convex corners were cured using
both semi-pregs and conventional format prepregs, for a total of four curved
128
laminates. Prepregs with a higher bulk factor undergo greater compaction
during cure and often will not conform readily to a curved surface. When the
bulk factor is high, prepreg and consumable materials can bridge over
concave molds or wrinkle over convex tooling [74], introducing in-plane
stresses and reducing the compaction pressure at corners.
Figure 36 Images of the part quality obtained from processing 5-harness satin prepregs and
semi-pregs at a concave and convex corner.
Images of the four sample cross-sections are presented in Fig. 36. As
the four images show, neither the semi-preg nor the conventional format
prepreg produced wrinkling or extensive bridging at laminate corners.
However, a key distinction between the two prepreg formats was the levels of
porosity in the cured laminates. The conventional prepreg sample exhibited
both intra-tow and inter-ply porosity, as shown in the first column. In
contrast, the semi-preg sample exhibited negligible intra-tow porosity, as
129
shown in the second column. These results were consistent with those
discussed in Sec. IV.3.1.
The Coefficient of Variation (CoV) (Eqn. 31) is a metric of corner part
quality which describes the variability in thickness between the flanges and
the curved surface of a complex part. The calculated CoV of both convex and
concave corners made with conventional prepreg was 0.08. For the corners
made with semi-preg, the CoV was 0.09, indicating that discontinuous film
may slightly increase the variability in corner thickness. This increase can be
attributed to either inherent experimental variation or to the higher bulk
factor of the semi-preg.
3.4 Design considerations and limitations for semi-preg fabrication
Design considerations and limitations for the fabrication of semi-pregs
were explored in the context of processing. Certain aspects of design and
fabrication of semi-pregs are critical to achieving high quality parts via
OoA/VBO processing. Here, we examine: (1) dimensional limitations of the
discontinuous pattern due to resin thickness and to uniformity; (2) degree of
impregnation of the resin in the fiber bed; (3) placement of discontinuous
patterns and fiber beds that are either isotropic or anisotropic; and (4)
polymer film dewetting of different resin systems.
Feature dimensions. Polymer film dewetting typically occurs at the
edge of a dry region (i.e., perforations, tears, or deep depressions). When
producing resin patterns for semi-preg by dewetting, nucleation sites are
130
introduced by scoring or piercing the resin where openings are desired.
However, a regular pattern is more difficult to achieve with resin that is non-
uniform (high discontinuity), because unintentional nucleation sites will
exist. Non-uniformity in resin films typically consists of uneven thickness,
gaps, and dry space, which often arises during filming. These defects are
more prevalent in thinner resin films and have a greater effect.
Figure 37 Micrographs of resin with various areal weights that were cut at different
distances to demonstrate the limitations of feature dimensions for uniform patterning.
Examples of striped resin patterns with different spacings of imposed
nucleation sites are presented in Fig. 37. The first row depicts the resulting
discontinuous resin patterns at 26 GSM when scored at 5-mm and 1-mm
intervals. As shown in the left micrograph, the resin did not dewet uniformly
131
when scored at 5 mm, and dewetting nucleated spontaneously between the
scorings. As depicted in the right micrograph, the resulting pattern became
more uniform when the scoring was reduced to 1-mm intervals, although
some dewetting still occurred along the resin stripes.
The second row depicts the resulting discontinuous resin patterns at
54 GSM when scored at 10-mm and 5-mm intervals. As shown in the right
micrograph, by doubling the resin areal weight from 26 GSM to 54 GSM,
uniform stripes were achieved with dewetting after scoring at 5-mm
intervals. However, as depicted in the left micrograph, when scoring spacing
was increased to 10 mm, feature precision was lost, and resin stripes were
irregular in shape. The third row depicts a resulting discontinuous resin
pattern at 188 GSM. Here, the patterning was uniform, even with scoring at
20-mm intervals.
Uniform patterns can be achieved with thin resin films if nucleation
sites are closely spaced. However, uniform patterns are difficult to achieve for
thin and non-uniform resin films when nucleation sites are spaced further
apart, as depicted in the left micrographs of the first and second rows of
Fig. 37. On the other hand, uniform patterns are readily achieved with thick
resin films, as shown in the third row. To achieve uniform patterns across all
resin thicknesses, the resin filming process must be carefully controlled to
restrict random dewetting.
132
Degree of impregnation. The method described for fabricating semi-
pregs via polymer film dewetting results in a degree of impregnation (DOI)
near zero. Low DOI values require transverse micro-flow to saturate fiber
tows prior to gelation. Results in Sec. IV.3.1 revealed that low porosity
laminates were not achieved when semi-preg was fabricated with the thickest
fiber bed (12K, 670 GSM). Presumably, the porosity resulted from the longer
flow distances required, which prevented full saturation of fiber tows. One
approach to this problem is to increase the DOI to reduce the flow distances
required to achieve saturation during cure. However, attempts to increase
DOI can potentially eliminate resin discontinuities and compromise the semi-
preg format, as described below.
The original intention of creating conventional OoA prepregs with
partial DOI (versus full saturation of the fiber bed) was to retain dry space at
the center of fiber tows, which was a technological advancement in prepreg
manufacturing [9]. In conventional OoA prepregs, these in-plane dry regions
allow air to evacuate the prepreg via edge breathing. DOI is defined as:
𝐷𝑒𝑔𝑟𝑒𝑒 𝑜𝑓 𝐼𝑚𝑝𝑟𝑒𝑔𝑛𝑎𝑡𝑖𝑜𝑛 ( 𝐷𝑂𝐼 ) = 1 −
𝐴 𝑑 𝑟𝑦
𝐴 𝑡𝑜𝑤
(4-5)
where Adry is the dry tow area and Atow is the entire tow area (dry and
saturated) [75]. Commercial OoA prepregs exhibit levels of DOI which range
from 0.05 to 0.5. However, the prepreg fabrication method presented here
involves briefly pressing fibers and resin films (continuous or discontinuous)
133
using an unheated hydraulic press. This method results in a negligible DOI
into the fiber bed, and the resin adheres only to the outer fibers of tows.
Figure 38 (a) Image of the surface of semi-preg with a DOI of zero. (b) Image of the cross-
section of a semi-preg with a DOI of zero. (c) Image of surface of semi-preg that underwent
elevated temperature and pressure to increase the DOI. (d) Image of cross-section of semi-preg
with an increased DOI.
Images of the DOI of the resin into a 5-harness satin fiber bed
produced in this manner is presented in Figs. 38a-b. The micrograph of the
surface is depicted in Fig. 38a, and the micrograph of the cross-section is
depicted in Fig. 38b. The cross-sectional image (Fig. 38b) shows, in addition
to exhibiting near-zero DOI, extensive macro-flow between tows, as well as
entrapped air bubbles. A larger DOI would reduce the resin flow distances
required, as well as the bulk factor. Increasing the DOI was attempted by
applying a pressure of 0.09 MPa to the semi-preg at 50 °C for 15 min using a
134
hydraulic press. Images of the surface and cross-section of the semi-preg after
this treatment are displayed in Figs. 38c-d. This treatment altered and
disrupted the discontinuous resin pattern, as illustrated by comparing Fig.
38a and Fig. 38c. When pressure and temperature were applied, resin flowed
preferentially along the length of the fibers before infiltrating the fiber tows.
In Fig. 38d, an example of an entire tow bundle is outlined in purple, and the
remaining dry fiber region is outlined in red. Using Eqn. 32, the volumes of
these two outlined regions were used to calculate the DOI. Here, the DOI
increased to 0.25 +/- 0.10.
A negligible DOI in semi-preg materials with thick fiber beds
inevitably will result in long flow distances. However, conventional means of
increasing the DOI (i.e., high temperature and applied pressure) alter
discontinuous resin patterns. Further development in prepreg production
methods will be required to achieve semi-pregs with controlled resin
patterning and increased DOI (and reduced bulk factors).
Isotropy & anisotropy. Some fiber beds (i.e., UD) and some
discontinuous resin patterns (i.e., stripes) are anisotropic. Other fiber beds
(i.e., satin weaves) are not anisotropic, but have adjacent, unidirectionally
aligned segments of tows (AUDAST). An example of an AUDAST fiber bed is
5-harness satin fabric. On one side of this fabric, segments of adjacent tows
are aligned globally in one direction. On the obverse side, segments of
adjacent tows are also aligned globally in one direction, but perpendicular to
135
the segments of the first side. The segments on each side of the fabric are
woven together, which results in each side of the fabric containing
perpendicular tow warps. Thus, the fabric surface is not entirely isotropic.
This locally anisotropic structure of the AUDAST fabric surface affects resin
flow in semi-pregs, particularly during initial stages, and can potentially
disrupt, distort, and even eliminate patterns of discontinuous resin.
When imposing resin patterns on such fiber beds (anisotropic and
AUDAST), the placement of the resin pattern in relation to the fiber
orientation at the fabric surface can affect the ability to saturate the fiber
bed. During cure, resin flow occurs preferentially along fiber directions,
rather than transversely [36,37,76]. Thus, if an anisotropic resin pattern is
placed parallel to an anisotropic or AUDAST fiber bed (e.g., striped resin
pattern on 5-harness satin fiber bed), fiber bed saturation may not be
achieved. Such care need not be required with isotropic fabrics (i.e., plain,
twill weaves) and/or an isotropic resin patterns (i.e., uniform grid, islands).
To determine the influence of resin pattern anisotropy and fiber
orientation on fiber bed impregnation, in situ observations were performed
using a transparent tool plate [30,31]. A typical cure cycle was employed
using vacuum bagged semi-pregs with select resin distributions applied to a
5-harness satin fiber bed. Fig. 39 shows the progression of resin flow at the
tool/part interface at different temperatures during the cure of each of the
semi-preg formats analyzed.
136
The first row of Fig. 39 depicts the flow of resin during the cure cycle of
a 5-harness satin semi-preg with an isotropic grid pattern. Resin quickly
filled pinholes and depressions created by tow crimp of the 5-harness satin
fabric (macro-flow), as shown in the 50 C micrograph. Subsequently, micro-
pore spaces within the fiber tow bundles were slowly filled by micro-flow, as
shown in the 85 C micrograph. By the end of the cure cycle, the fiber bed
was saturated, and exhibited only minor surface porosity, as shown in the
“End of Cure” micrograph.
Figure 39 Images of the flow of resin for various pattern types and pattern placement with
respect to the fiber bed during the cure cycle.
The second row of Fig. 39 depicts the flow of resin during the cure cycle
of a 5-harness satin semi-preg with an anisotropic striped pattern applied
parallel to the primary fiber direction. Resin initially filled adjacent pinholes
and the depressions of adjacent tows, as shown in the 50 C micrograph.
Thereafter, resin flowed preferentially into fiber tows that were
137
perpendicular to the resin stripes, as shown in the 85 C micrograph. This
process was much slower than the infiltration times required for the grid
pattern and resulted in regions of dry fiber tows (~9.2% of the entire surface
area). The third row of Fig. 39 depicts the flow of resin during the cure cycle
of a 5-harness satin semi-preg with an anisotropic striped pattern applied
perpendicular to the primary fiber direction. Resin flow saturated this fiber
bed much like the grid pattern depicted in the first row.
The results described above show that anisotropic resin patterns (e.g.,
stripes), when aligned parallel to an anisotropic or AUDAST fiber bed, yield
only partial saturation. On the other hand, isotropic resin patterns, such as
grids, are more robust and immune to adverse effects of fiber orientation.
Thus, when designing semi-preg formats, isotropic patterns can be
advantageous, as they effectively mitigate non-uniform flow issues and
achieve full fiber bed saturation regardless of fiber orientation. Anisotropic
patterns such as stripes must be intelligently applied to ensure proper
placement with reference to the fiber orientation. However, stripes may be
simpler and more convenient to produce relative to other pattern designs,
such as islands or grid, depending on the production method of the
discontinuous resin distribution [1–3].
Alternative resin systems. Prepregs are produced with a wide variety of
thermoset resin types and formulations. To explore the versatility of semi-
preg fabrication via polymer film dewetting, the technique was applied to
138
select resin systems: cyanate ester, BMI, and a standard commercial OoA
epoxy (CYCOM 5320-1, Solvay, USA).
Images of the discontinuous resin patterns pressed onto a UD carbon
fiber bed and onto a plain weave glass fiber bed are shown in Fig. 40. A grid
pattern of cyanate ester was created and applied to glass fibers (cyanate ester
is often reinforced with fibers other than carbon), as shown in the left
micrograph. Similarly, grid patterns of BMI resin and a commercial epoxy
resin (CYCOM 5320-1) were created and applied to UD carbon fibers, as
shown in center and right micrographs. To achieve the same feature
dimensions as the epoxy resin patterns studied in previous sections, the
cyanate ester and BMI resin were heated to 104 °C for only 10 – 15 s. If
heated for longer times, the resin film dewetted extensively, forming small
droplets. However, achieving the same feature dimensions in the commercial
epoxy required heating to 104 °C for more than 12 min. While a dewetting
process can potentially be applied to any resin system, dewetting conditions
require adaptation to the specific resin system and substrate. The BMI and
cyanate ester films used here required less time (10 – 15 s) to achieve feature
dimensions equivalent to the original epoxy resin system (2 min), a result of
the higher interfacial tension with the silicone-coated backing paper. The
commercial epoxy film had a lower interfacial tension with the backing paper
and thus required more time ( ˃ 12 min) to achieve similar dimensions.
139
Figure 40 Image of bismaleimide (BMI), cyanate ester, and a commercial epoxy
(CYCOM-5320-1) films that were created into discontinuous distributions by initiating
nucleation sites by a spike roller and subsequently heating the resins at 104 °C.
The results shown in Fig. 40 and described above demonstrate that
polymer film dewetting can be used to produce discontinuous resin patterns
from a range of resin systems. However, a resin can only be utilized for the
fabrication of OoA parts if resin cure kinetics are amenable to low pressure
processing [6]. VBO prepreg systems are designed to remain relatively
viscous in the early stages of cure to limit infiltration, prevent resin bleed,
and retain sufficient dry areas for air evacuation [70]. The overall viscosity
profile, however, must also permit sufficient flow during the cure cycle to
fully saturate the fiber bed [4]. The rheological evolution of VBO resins must,
in essence, balance the need to prevent voids caused by entrapped gases (air
and cure induced volatiles) as well as voids caused by insufficient flow.
Provided these conditions can be met, a semi-preg fabricated with dewetted
resin can be expected to consistently yield high quality laminates with
OoA/VBO cure, even under adverse process conditions.
140
4. Conclusion
In general, laminates fabricated using semi-preg with a variety of fiber
bed architectures exhibited near-zero porosity. In addition, semi-preg formats
increased the bulk factor of VBO-fabricated laminates by ~0.1. Nevertheless,
production of complex shapes using semi-preg revealed that part quality (i.e.,
wrinkling, bridging, and thickness variation) was similar to the quality
achieved with conventional prepreg. By incorporating multiple design
considerations, the dewetting technique was applied to other resin systems,
such as cyanate ester and BMI, to create uniform and repeatable
discontinuous resin patterns.
The use of semi-pregs to produce parts via VBO processing shows
promise and may indeed impart robustness. Nevertheless, some key
limitations of the approach were identified, and these may require further
study. Current fabrication methods to create semi-pregs result in larger bulk
factors than conventional OoA prepregs (Δ~0.1), due primarily to the near-
zero DOI, and secondarily to thicker resin from the dewetting process. In
addition to longer flow distances, the low DOI will limit the ability to rework
plies (lift and reposition) when required. Only practical experience will
determine how serious these drawbacks will be. However, the advantages
imparted by the short breathe-out distances in semi-pregs (enhancing
robustness to the VBO manufacturing process) are expected to outweigh
these drawbacks. In particular, the use of semi-pregs may help to restore the
141
process robustness sacrificed by abandoning autoclave curing for OoA
processing. Nonetheless, one crucial challenge that remains is to demonstrate
the scalability of the dewetting process. Current work has demonstrated only
a lab-scale batch process and a continuous process has not yet been shown.
However, the process may in fact be backwards-compatible with hot-melt
prepregging, since in principle, imprint/de-wet steps can be incorporated into
existing prepreg production lines.
The development of semi-preg materials is a potential route toward
robust OoA composite manufacturing. This work established that a large
range of semi-preg product forms can be fabricated and manufactured into
high quality laminates for both flat and complex parts. The inherent
manufacturing robustness imparted by the methods presented here can
potentially expand the applicable uses of VBO prepregs within aerospace
manufacturing and to other non-aerospace applications.
* This study has been submitted for publication.
142
CONCLUDING REMARKS
In the presented work, several research goals were accomplished. Key
results of each study are discussed along with broader implications. In
Chapter I, a method to create high through-thickness permeability VBO-OoA
prepreg using UD carbon fiber tape was demonstrated and characterized.
The approach was distinguished by the use of polymer film dewetting (prior
to combining with fibers) to create periodic openings in the resin, through
which gas evacuation during cure occurred during laminate consolidation.
Laminates fabricated using prepregs produced by this method yielded near-
zero void contents even when in-plane gas evacuation (edge breathing) was
eliminated. In contrast, conventional OoA prepreg produced with continuous
films resulted in laminates that exhibited much greater porosity (up to 8 %)
for both standard cure conditions and sealed edges. This work introduces a
potentially scalable technique (imprint/de-wet steps can be incorporated into
existing prepreg production lines) for producing OoA prepregs with
discontinuous resin distributions from any fiber bed.
In Chapter II, a methodology was outlined that allows for the rapid
screening (i.e., evaluation and differentiation) of discontinuous resin patterns
for VBO prepregs, which can be used to guide prepreg development. Due to
the overwhelming number of choices in designing a prepreg with
discontinuous resin format, selection of an appropriate resin distribution was
not obvious. Thus, the methodology presented here allows one to differentiate
143
between pattern types, feature dimensions, stacking orientations, and ply
counts (with as many iterations as specified) without any physical
experimentation by the use of simple geometric models. Experiments validate
that the characteristics identified using the geometric model do indeed affect
through-thickness air permeability and void content in cured laminates. In
practice, the design space can be greatly reduced by eliminating patterns that
do not allow for rapid air evacuation.
In Chapter III, a methodology is described for determining the
maximum resin feature dimensions for semi-pregs produced using a given
resin system. While the work focused on a single resin system, the method
itself can be applied to any thermoset epoxy. In addition, the influence of
these feature dimensions on the effective air permeability of the laminates,
and ultimately, the time required for evacuation, was determined. The model
presented provides guidance to specify the dimensions of resin distribution
patterns sufficient for both efficient air evacuation and full resin infiltration.
In Chapter IV, laminates fabricated using semi-preg with a variety of
fiber bed architectures exhibited near-zero porosity. In addition, semi-preg
formats increased the bulk factor of VBO-fabricated laminates by ~0.1.
Nevertheless, production of complex shapes using semi-preg revealed that
part quality (i.e., wrinkling, bridging, and thickness variation) was similar to
the quality achieved with conventional prepreg. By incorporating multiple
design considerations, the dewetting technique was applied to other resin
144
systems, such as cyanate ester and BMI, to create uniform and repeatable
discontinuous resin patterns.
OoA/VBO prepreg processing presently suffers from a lack of
robustness in the manufacturing process, and unacceptable defect levels often
observed in non-ideal manufacturing conditions or part geometries. The
methods described here enables the production of UD fiber prepregs with
discontinuous resin distributions, and such prepregs will enable manufacture
of composite parts with low defect levels even in sub-optimal processing
conditions. The inherent manufacturing robustness imparted by the methods
presented can, in turn, expand the applicable uses of VBO prepregs within
aerospace manufacturing and into other non-aerospace applications.
145
RECOMMENDATIONS FOR FUTURE WORK
A crucial challenge that remains is to demonstrate the scalability of
the dewetting process. The work presented here introduces a potentially
scalable technique for producing OoA prepregs with discontinuous resin
distributions from any fiber bed. Dewetting is potentially backwards-
compatible with hot-melt prepregging, since in principle, imprint/de-wet
steps can be incorporated into existing prepreg production lines. However,
this work has only been demonstrated in a lab-scale batch process and a
continuous process has not yet been shown. Therefore, future work should
include demonstrating this potential.
The resin flow front model presented in this work only considered one-
dimensional resin infiltration. However, the mechanism for resin infiltration
of more complex fiber beds (particularly woven fabrics) is expected to differ
and involve 2D or 3D multi-scale flow. Therefore, a multi-dimensional resin
flow front model should be a future area of study for prepregs with
discontinuous resin.
The current fabrication methods to create semi-pregs result in larger
bulk factors than conventional OoA prepregs (Δ~0.1), due primarily to the
near-zero DOI, and secondarily to thicker resin from the dewetting process.
In addition to longer flow distances, the low DOI will limit the ability to
rework plies (lift and reposition) when required. Further research will
determine how serious these drawbacks will be. However, the advantages
146
imparted by the short breathe-out distances in semi-pregs (enhancing
robustness to the VBO manufacturing process) are expected to outweigh
these drawbacks.
147
REFERENCES
[1] Grunenfelder LK, Dills A, Centea T, Nutt S. Effect of prepreg format on defect
control in out-of-autoclave processing. Compos Part A Appl Sci Manuf
2017;93:88–99. doi:10.1016/j.compositesa.2016.10.027.
[2] Roman M, Howard SJ, Boyd JD. Curable Prepregs With Surface Openings.
US 2014/0174641 A1, 2014.
[3] Tavares SS, Michaud V, Månson JAE. Assessment of semi-impregnated
fabrics in honeycomb sandwich structures. Compos Part A Appl Sci Manuf
2010;41:8–15. doi:10.1016/j.compositesa.2009.09.005.
[4] Steele M, Corden T, Gibbs A. The development of out-of-autoclave composite
prepreg technology for aerospace applications. SAMPE Tech Conf, Long
Beach, CA: 2001.
[5] Ridgard C. Out of autoclave composite technology for aerospace, defense and
space structures. SAMPE Tech Conf, 2009.
[6] Centea T, Grunenfelder LK, Nutt SR. A review of out-of-autoclave prepregs -
Material properties, process phenomena, and manufacturing considerations.
Compos Part A Appl Sci Manuf 2015;70:132–54.
doi:10.1016/j.compositesa.2014.09.029.
[7] Jackson K, Crabtree B. Autoclave quality composites tooling for composite
from vacuum bag only processing. SAMPE Tech Conf, Long Beach, CA: 2002.
[8] Heth J. From art to science: A prepreg overview. High-Performance Compos
2000;8:32–6.
[9] Repecka L, Boyd J. Vacuum-bag-only-curable prepregs that produce void-free
parts. SAMPE Tech Conf, Long Beach, CA: 2002.
[10] Centea T, Hubert P. Modelling the effect of material properties and process
parameters on tow impregnation in out-of-autoclave prepregs. Compos Part A
Appl Sci Manuf 2012;43:1505–13. doi:10.1016/j.compositesa.2012.03.028.
[11] Grunenfelder LK, Nutt SR. Void formation in composite prepregs - Effect of
dissolved moisture. Compos Sci Technol 2010;70:2304–9.
doi:10.1016/j.compscitech.2010.09.009.
[12] Kardos J, Dudukovic M, Dave R. Void growth and resin transport during
processing of thermosetting — Matrix composites. Adv Polym Sci
1986;80:102–23. doi:10.1007/3-540-16423-5.
[13] Grunenfelder LK, Katz S, Centea T, Nutt S. Through-Thickness Permeable
Prepreg For Robust Vacuum Bag Only Processing. SAMPE Tech Conf, Seattle:
2017.
[14] Nutt S, Grunenfelder L, Centea T. High Permeability Composite Prepreg
Constructions and Methods for Making the Same. PCT/US2018/012665, 2018.
148
[15] Eustathopoulos N, Nicholas M, Drevet B. Wettability at High Temperatures.
Pergamon; 1999.
[16] Coussy O. Surface Energy and Capillarity. Mech. Phys. Porous Solids, John
Wiley & Sons, Ltd; 2010, p. 107–47.
[17] Roy S, Bandyopadhyay D, Karim A, Mukherjee R. Interplay of substrate
surface energy and nanoparticle concentration in suppressing polymer thin
film dewetting. Macromolecules 2015;48:373–82. doi:10.1021/ma501262x.
[18] Kheshgi HS, Scriven LE. Dewetting: Nucleation and growth of dry regions.
Chem Eng Sci 1991;46:519–26. doi:10.1016/0009-2509(91)80012-N.
[19] Kohl JG, Singer IL. Pull-off behavior of epoxy bonded to silicone duplex
coatings. Prog Org Coatings 1999;36:15–20. doi:10.1016/S0300-9440(98)00074-
5.
[20] Edwards AMJ, Ledesma-aguilar R, Newton MI, Brown C V, Mchale G. Not
spreading in reverse : The dewetting of a liquid film into a single drop. Sci Adv
2016.
[21] Bhowmick AK, Konar J, Kole S, Narayanan S. Surface properties of EPDM,
silicone rubber, and their blend during aging. J Appl Polym Sci 1995;57:631–
7. doi:10.1002/app.1995.070570513.
[22] Page SA, Berg JC, Månson JAE. Characterization of epoxy resin surface
energetics. J Adhes Sci Technol 2001;15:153–70.
doi:10.1163/156856101743382.
[23] Abbott JR, Higgins BG. Surface tension of a curing epoxy. J Polym Sci Part A
Polym Chem 1988;26:1985–8. doi:10.1002/pola.1988.080260724.
[24] Centea T, Peters G, Hendrie K, Nutt SR. Effects of thermal gradients on
defect formation during the consolidation of partially impregnated prepregs. J
Compos Mater 2017;51:3987–4003. doi:10.1177/0021998317733317.
[25] Howard S. Evaluating Consolidation Dynamics - Study of Key Consolidation
Parameters in a Vacuum Bag Only (VBO) Cure. SAMPE 2016 Tech. Conf. Soc.
Adv. Mater. Process Eng., 2016.
[26] Zhang D, Heider D, Gillespie JW. Void reduction of high-performance
thermoplastic composites via oven vacuum bag processing. J Compos Mater
2017;51:4219–30. doi:10.1177/0021998317700700.
[27] Centea T, Hubert P. Measuring the impregnation of an out-of-autoclave
prepreg by micro-CT. Compos Sci Technol 2011;71:593–9.
doi:10.1016/j.compscitech.2010.12.009.
[28] Kratz J, Hubert P. Anisotropic air permeability in out-of-autoclave prepregs:
Effect on honeycomb panel evacuation prior to cure. Compos Part A Appl Sci
Manuf 2013;49:179–91. doi:10.1016/j.compositesa.2013.02.013.
149
[29] Tavares SS, Michaud V, Månson JAE. Through thickness air permeability of
prepregs during cure. Compos Part A Appl Sci Manuf 2009;40:1587–96.
doi:10.1016/j.compositesa.2009.07.004.
[30] Hu W, Grunenfelder LK, Centea T, Nutt S. In-situ monitoring of void
evolution in unidirectional prepreg. J Compos Mater 2017.
doi:10.1177/0021998318759183.
[31] Hu W, Grunenfelder LK, Nutt SR. In-Situ Observation Of Void Transport
During Vacuum Bag-Only Cure. SAMPE Tech Conf, Long Beach, CA: 2016.
[32] Hamill L, Centea T, Nutt S. Surface porosity during vacuum bag-only prepreg
processing: Causes and mitigation strategies. Compos Part A Appl Sci Manuf
2015;75. doi:10.1016/j.compositesa.2015.04.009.
[33] Martinez P, Nutt SR. Robust Manufacturing Of Complex-Shaped Parts Using
Out-Of-Autoclave Prepregs With Discontinuous Formats. CAMx Conf. Proc.,
Dallas, TX: 2018.
[34] Cender TA, Simacek P, Advani SG. Resin film impregnation in fabric prepregs
with dual length scale permeability. Compos Part A Appl Sci Manuf
2013;53:118–28. doi:10.1016/j.compositesa.2013.05.013.
[35] de Winter JCF. Using the Student’s t-test with extremely small sample sizes.
Pr Assessment, Res Evalutaion 2013;18. doi:Retrieved from
http://pareonline.net.
[36] Schechter SGK, Centea T, Nutt SR. Polymer film dewetting for fabrication of
out-of-autoclave prepreg with high through-thickness permeability. Compos
Part A Appl Sci Manuf 2018;114:86–96.
doi:10.1016/j.compositesa.2018.08.002.
[37] Schechter SGK, Centea T, Nutt S. Fabrication Of Out-Of-Autoclave Prepreg
With High Through- Thickness Permeability By Polymer Film Dewetting.
Sampe Tech Conf., Long Beach, CA: 2018, p. 2018.
[38] Sloan J. Developing Next Generation Composites Talent. Compos World Mag
2019:22–5.
[39] Hahn GL, Bond GG, Fogarty JH. Non-Autoclave (Prepreg) Manufacturing
Technology: Part Scale-Up with CYCOM 5320-1 Prepregs. Proc. SAMPE 2011
Conf., Long Beach, CA: 2011.
[40] Arafath ARA, Fernlund G, Poursartip A. Gas transport in prepregs: model
and permeability experiments. Proc 17th Int Conf Compos Mater 2009:1–9.
[41] Helmus R, Centea T, Hubert P, Hinterhölzl R. Out-of-autoclave prepreg
consolidation: Coupled air evacuation and prepreg impregnation modeling. J
Compos Mater 2016;50:1403–13. doi:10.1177/0021998315592005.
[42] Helmus R, Hinterhölzl R, Hubert P. A stochastic approach to model material
variation determining tow impregnation in out-of-autoclave prepreg
consolidation. Compos Part A Appl Sci Manuf 2015;77:293–300.
doi:10.1016/j.compositesa.2015.03.021.
150
[43] Kourkoutsaki T, Comas-Cardona S, Binetruy C, Upadhyay RK, Hinterhoelzl
R. The impact of air evacuation on the impregnation time of Out-of-Autoclave
prepregs. Compos Part A Appl Sci Manuf 2015;79:30–42.
doi:10.1016/j.compositesa.2015.08.034.
[44] Kompiš V, Qin QH, Fu ZJ, Chen CS, Droppa P, Kelemen M, et al. Parallel
computational models for composites reinforced by CNT-fibres. Eng Anal
Bound Elem 2012;36:47–52. doi:10.1016/j.enganabound.2011.04.009.
[45] Fahrang L, Fernlund G. Void evolution and gas transport during cure in out-
of-autoclave prepreg laminates. Proc. SAMPE 2011 Conf., 2011.
[46] Centea T, Hubert P. Measuring the impregnation of an out-of-autoclave
prepreg by micro-CT. Compos Sci Technol 2011.
doi:10.1016/j.compscitech.2010.12.009.
[47] Centea T, Hubert P. Out-of-autoclave prepreg consolidation under deficient
pressure conditions. J Compos Mater 2013:0021998313494101-.
doi:10.1177/0021998313494101.
[48] Al-Raoush RI, Madhoun IT. TORT3D: A MATLAB code to compute geometric
tortuosity from 3D images of unconsolidated porous media. Powder Technol
2017;320:99–107. doi:10.1016/j.powtec.2017.06.066.
[49] Allen TT. Introduction to engineering statistics and six sigma: Statistical
quality control and design of experiments and systems. 2006. doi:10.1007/1-
84628-200-4.
[50] Antony J. Design of Experiments for Engineers and Scientists. 2003.
doi:10.1016/B978-0-7506-4709-0.X5000-5.
[51] Saleh AKME, Arashi M, Tabatabaey SMM. Statistical Inference for Models
with Multivariate t-Distributed Errors. 2014. doi:10.1002/9781118853931.
[52] Girden ER. Two-factor study with repeated measures on one factor. ANOVA,
1992. doi:http://dx.doi.org/10.4135/9781412983419.
[53] Bear J. Dynamics of fluids in porous media. New York: American Elsevier
Pub. Co.; 1972.
[54] Scheidegger AE. The physics of flow through porous media. Toronto:
University of Toronto Press; 1958.
[55] Schechter SGK, Centea T, Nutt S. Effects of resin distribution patterns on
through-thickness air removal in vacuum-bag-only prepregs. Compos Part A
Appl Sci Manuf 2020;130. doi:10.1016/j.compositesa.2019.105723.
[56] Kim D, Centea T, Nutt SR. Modelling and monitoring of out-time and
moisture absorption effects on cure kinetics and viscosity for an out-of-
autoclave (OoA) prepreg. Compos Sci Technol 2017;138:201–8.
doi:10.1016/j.compscitech.2016.11.023.
151
[57] Khoun L, Centea T, Hubert P. Characterization methodology of thermoset
resins for the processing of composite materials -Case study. J Compos Mater
2010;44:1397–415. doi:10.1177/0021998309353960.
[58] Kratz J, Hsiao K, Fernlund G, Hubert P. Thermal models for MTM45-1 and
Cycom 5320 out-of-autoclave prepreg resins. J Compos Mater 2013;47:341–52.
doi:10.1177/0021998312440131.
[59] Hubert P, Johnston A, Poursartip A, Nelson K. Cure Kinetics and Viscosity
Models for Hexcel 8552 Epoxy Resin. Int. SAMPE Symp. Exhib., vol. 46, 2001,
p. 8552.
[60] Halley PJ, Mackay ME. Chemorheology of Thermosets - An Overview. Polym
Eng Sci n.d.;36.
[61] Castro JM, Macosko CW. Kinetics and rheology of typical polyurethane
reaction injection molding systems. Soc. Plast. Eng., 1980, p. 434–8.
[62] Mezger TG. The Rheology Handbook: For Users of Rotational and Oscillatory
Rheometers. 2nd ed. Hannover: Vincentz Networks; 2006.
[63] Michaud V, Mortensen A. Infiltration processing of fibrereinforced composites:
governing phenomena 2001;32.
[64] Gay D. Composite materials : design and applications. Third. Boca Raton:
CRC Press, Taylor & Francis Group; 2015.
[65] Gebart BR. Permeability of unidirectional reinforcements for RTM. J Compos
Mater 1992;26:1100–33.
[66] Grunenfelder LK, Nutt SR. Out Time Effects on VBO Prepreg and Laminate
Properties. Proc SAMPE 2011 Conf 2011.
[67] Centea T, Kratz J, Hubert P. The effect of out-time and cure cycle on the
consolidation of out-of-autoclave prepregs. Proc. 11th Int. Conf. Flow Process.
Compos. Mater., Auckland, New Zealand: 2012.
[68] Grunenfelder LK, Centea T, Hubert P, Nutt SR. Effect of room-temperature
out-time on tow impregnation in an out-of-autoclave prepreg. Compos Part A
Appl Sci Manuf 2013.
[69] Centea T, Nutt SR. Manufacturing cost relationships for vacuum bag-only
prepreg processing. J Compos Mater 2016;50:2305–21.
doi:10.1177/0021998315602949.
[70] Ridgard C. Out of Autoclave Composite Technology for Aerospace, Defense
and Space Structures. Proc SAMPE 2009 Conf 2009.
[71] Grunenfelder LK, Katz S, Centea T, Nutt SR. Through-Thickness Permeable
Prepreg for Robust Vacuum Bag Only Processing. SAMPE J 2018.
[72] Ma Y, Centea T, Nutt SR. Defect reduction strategies for the manufacture of
contoured laminates using vacuum BAG-only prepregs. Polym Compos 2017.
doi:10.1002/pc.23773.
152
[73] Hörrmann S, Adumitroaie A, Viechtbauer C, Schagerl M. The effect of fiber
waviness on the fatigue life of CFRP materials. Int J Fatigue 2016;90:139–47.
doi:10.1016/j.ijfatigue.2016.04.029.
[74] Albert C, Fernlund G. Spring-in and warpage of angled composite laminates.
Compos Sci Technol 2002;62:1895–912. doi:10.1016/S0266-3538(02)00105-7.
[75] Palardy-Sim M, Hubert P. Characterization of the degree of impregnation of
out-of-autoclave prepreg. ICCM Int Conf Compos Mater 2015;2015-July:19–
24.
[76] Martinez P, Jin B, Nutt S. Effect of Fiber Bed Achitecture on Single Resin
Droplet Spread for Prepreg Manufacturing. CAMx Conf. Proc., Anaheim, CA:
2019.
Abstract (if available)
Abstract
For aerospace composites manufacturing, prepregs processed via out-of-autoclave (OoA) vacuum-bag-only (VBO) methods offer a viable alternative to traditional autoclave manufacturing methods, which is costly and limiting. However, OoA/VBO processing is limited to 0.1 MPa (1 atm) of consolidation pressure, which is often insufficient to collapse porosity to acceptable levels. Past work has shown that discontinuous resin films increase the capacity for air evacuation in the z-direction (transverse) by creating additional egress pathways and can virtually eliminate porosity caused by entrapped gases and low-pressure VBO consolidation. The main objective of this work was to determine a method to create high through-thickness permeability prepreg for all fiber bed architectures, to develop methods to evaluate optimal discontinuous resin patterns (i.e., maximize gas transport while ensuring complete infiltration), and to understand the limitations of its applications in various situations. ❧ In Chapter I, polymer film dewetting on a substrate (independent of fiber bed architecture) was explored, developed, and demonstrated as a method to produce out-of-autoclave, vacuum bag-only (OoA/VBO) prepregs with high transverse permeability and process robustness. The dimensions of the surface openings created by dewetting were measured, and the percent surface area exposed was calculated. Prepregs were fabricated with continuous and dewetted (discontinuous) films to produce trial laminates. The laminates were cured under both standard and sub-optimal conditions, and were characterized before, during, and after cure. Laminates fabricated with dewetted resin consistently achieved near-zero porosity. In contrast, laminates with continuous film displayed high levels of porosity, particularly during sub-optimal cure. The findings demonstrate that dewetting can be used effectively to produce OoA prepregs with high through-thickness permeability, which can yield porosity-free laminates via VBO processing. Furthermore, these results elucidate aspects of resin dewetting that are critical in the creation of robust OoA prepregs. ❧ In Chapter II, a geometric model was developed to guide the fabrication of prepregs with various discontinuous patterns and laminates with different orientations and ply counts. The model was used to evaluate metrics related to gas transport: projected surface area exposed, sealed interfaces, and tortuosity. Statistical analysis revealed that single layer surface area exposed and ply count had the greatest effect on projected surface area exposed
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Efficient manufacturing and repair of out-of-autoclave prepreg composites
PDF
Void evolution in vacuum bag-only prepregs
PDF
In situ process monitoring for modeling and defect mitigation in composites processing
PDF
Defect control in vacuum bag only processing of composite prepregs
PDF
Sustainable manufacturing of out-of-autoclave (OoA) prepreg: processing challenges
PDF
Developing efficient methods for the manufacture and analysis of composites structures
PDF
Material and process development and optimization for efficient manufacturing of polymer composites
PDF
In situ process analysis for defect control during composites manufacturing
PDF
Chemical recycling of amine/epoxy composites at atmospheric pressure
PDF
Characterization, process analysis, and recycling of a benzoxazine-epoxy resin for structural composites
PDF
Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites
PDF
Slurry based stereolithography: a solid freeform fabrication method of ceramics and composites
PDF
Fabrication, deposition, and characterization of size-selected metal nanoclusters with a magnetron sputtering gas aggregation source
PDF
Electrodeposition of platinum-iridium coatings and nanowires for neurostimulating applications: fabrication, characterization and in-vivo retinal stimulation/recording EIS studies of hexavalent a...
PDF
Investigation of gas transport and sorption in shales
Asset Metadata
Creator
Schechter, Sarah Grace Katz
(author)
Core Title
Fabrication and analysis of prepregs with discontinuous resin patterning for robust out-of-autoclave manufacturing
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Chemical Engineering
Publication Date
02/14/2020
Defense Date
01/30/2020
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
carbon fiber,defects,epoxy,OAI-PMH Harvest,out of autoclave processing,permeability,polymer matrix composites,porosity,prepreg,vacuum bag only
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Nutt, Steven (
committee chair
), Grunenfelder, Lessa (
committee member
), Ravichandran, Jayakanth (
committee member
)
Creator Email
katz.sarahgrace@gmail.com,sarahgka@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-269923
Unique identifier
UC11674219
Identifier
etd-SchechterS-8168.pdf (filename),usctheses-c89-269923 (legacy record id)
Legacy Identifier
etd-SchechterS-8168.pdf
Dmrecord
269923
Document Type
Dissertation
Rights
Schechter, Sarah Grace Katz
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
carbon fiber
defects
epoxy
out of autoclave processing
permeability
polymer matrix composites
porosity
prepreg
vacuum bag only