advanced high speed programmable preforming - wiki.ornl.gov ...

ADVANCED HIGH SPEED PROGRAMMABLE PREFORMING
Robert E. Norris, Jr 1 , Ronny D. Lomax 1 , Fue Xiong 1 , Jeffrey S. Dahl 2 , Patrick J. Blanchard 2
1 Oak Ridge National Laboratory
Bethel Valley Road
Oak Ridge, TN 37831
2 Ford Motor Company
Research and Innovation Center
Dearborn, MI 48121
ABSTRACT
Polymer-matrix composites offer greater stiffness and strength per unit weight than conventional
materials resulting in new opportunities for lightweighting of automotive and heavy vehicles.
Other benefits include design flexibility, less corrosion susceptibility, and the ability to tailor
properties to specific load requirements. However, widespread implementation of structural
composites requires lower-cost manufacturing processes than those that are currently available.
Advanced, directed-fiber preforming processes have demonstrated exceptional value for rapid
preforming of large, glass-reinforced, automotive composite structures. This is due to process
flexibility and inherently low material scrap rate. Hence directed fiber performing processes
offer a low cost manufacturing methodology for producing preforms for a variety of structural
automotive components.
This paper describes work conducted at the Oak Ridge National Laboratory (ORNL), focused on
the development and demonstration of a high speed chopper gun to enhance throughput
capabilities. ORNL and the Automotive Composites Consortium (ACC) revised the design of a
standard chopper gun to expand the operational envelope, enabling delivery of up to 20kg/min.
A prototype unit was fabricated and used to demonstrate continuous chopping of multiple roving
at high output over extended periods. In addition fiber handling system modifications were
completed to sustain the high output the modified chopper affords. These hardware upgrades are
documented along with results of process characterization and capabilities assessment.
1. INTRODUCTION
Polymer-matrix composite materials offer a number of benefits in lightweighting of automotive
and heavy vehicles, including greater stiffness and strength per unit weight than conventional
materials, easier formability, less corrosion susceptibility, the ability to tailor properties to
specific load requirements, and enhanced noise and vibration damping. However, widespread
implementation of carbon fiber (CF) composites, which are among the materials with the greatest
weight-saving potential, will require lower-cost materials and manufacturing processes than are
currently available. Advanced preforming processes offer opportunities to facilitate the
widespread use of carbon composites.
The process utilized in this work is based on equipment and techniques originally developed and
demonstrated by Owens Corning in Battice, Belgium and Aplicator System AB in Molnlycke,

Sweden and termed P4, (Programmable Powdered Preforming Process). Aplicator was the
hardware developer and manufactured the research machine for ORNL. In the traditional
embodiment of the process, reinforcing fiber is chopped and sprayed onto a preforming screen
along with a powdered binder. The fiber and binder are held in place on the screen by air flow
drawn through the screen by a large suction fan on the opposite side of the screen from the fiber
deposition. In addition to being able to place the chopped fiber and binder where they are of
most benefit structurally via the robot program, one unique aspect of the Aplicator P4 is the
ability to also change the fiber cut length “on-the-fly” by programming variable fiber lengths as
well. Ductwork is then rotated or translated into place over the screen so that air from a
continuously circulating hot air loop can then be directed through the fiber and binder to set the
binder. After exposing the preform to an adequate time at the higher temperature, the preform is
cooled, the ductwork removed, and the part extracted. The preform is then moved to a molding
station where Structural Reaction Injection Molding (SRIM) resin is rapidly injected and the
mold closed under pressure to distribute the resin throughout the preform and for consolidation
during final curing. The binder holds the reinforcing fiber in place to maintain the fiber
architecture in handling and during the thermosetting resin injection. This process can achieve
relatively high fiber volume fractions with uniform fiber distribution while working with low
cost product forms. Preforming and liquid molding using these processes have been
demonstrated by the Automotive Composites Consortium in its Focal Project 2 pickup truck
box [1] and implemented in limited production for the GMC Silverado and for several parts of
Aston Martin vehicles.
Analyses have indicated a potential for greater than 60% weight savings for a CF-intensive bodyin-white
under the assumption of a thickness design constraint of 1.5 mm. The analyses also
indicated the potential for saving an additional 15% if the thickness constraint is reduced to
1 mm. Unfortunately, evidence suggests that 1.5 mm may be a practical limit for liquid molding
due to resin infiltration difficulties. However, thermoplastic preforms, in which the reinforcing
fiber and matrix (in the form of thermoplastic fiber during deposition) are both deposited in the
preforming step, offer a potential path to obtaining thinner sections and, consequently, additional
weight savings as well as greater potential for recyclability. Among other drivers for the use of
thermoplastic materials include reduced cycle times, enhanced damage tolerance, potential
rework with less waste, and fewer resin safety and health issues associated with storage,
handling, and processing of these systems.
One approach to manufacturing thermoplastic composites is to chop and spray the thermoplastic
material in fiber form along with the reinforcing fiber. A product form with commingled glass
and thermoplastic fibers that facilitates this type of processing is a product (Twintex) produced
by Owens Corning Vetrotex. Although the form utilized for this work consists of commingled
glass and polypropylene fibers, the manufacturer has listed other variants and the basic
evaluation process can be extended to a variety of materials and processes based on this concept.
The ACC Materials and Processing working groups has conducted work with Ecole
Polytechnique Federal de Lausanne (EPFL) Laboratory in Switzerland to study processing
effects and material properties leading to cost- model comparisons of various thermoplasticcomposite
processes. With support of EPFL and the ACC, ORNL fabricated approximately 200
sample panels using the polypropylene form of Twintex in various fiber lengths (25 - 75 mm)
and preform thickness (2 mm - 4 mm) along with variations in preforming compaction station

time (10-60s) and air temperature (180-240°C). Material properties from the ORNL panels
processed by EPFL are competitive with similar materials and, in some cases, significantly
improved [2-4]. Overall economics for this process were shown by EPFL analysis to approach
targets as chopping speeds improve, leading to additional near-term focus on demonstrating the
technical and economic feasibility of higher-speed chopping processes. This paper discusses
how significantly higher chopping speeds were achieved and demonstrated [5].
2. CHOPPER EVALUATION AND DEVELOPMENT
2.1 Baseline and Alternative Chopper Hardware
The EPFL analysis indicated economics for thermoplastic preforming of a generic liftgate
application (5 kg) in a deposition cycle time of ~30 seconds with this process can be attractive
versus competitive processes such as Glass Mat Thermoplastics (GMT) and Direct Long Fiber
Thermoplastics (D-LFT). This would require fiber deposition rates in the 12 to 15 kg/minute
range using a single deposition head system, leading to our focus on demonstrating the technical
and economic feasibility of higher-speed chopping processes. For comparison, the baseline P4
chopper as shown in Figure 1 is designed to process 2 tows at up to 625 m/min as supplied by
Aplicator. For the baseline 2400 tex glass fiber for which the machine was designed to process,
this translates to an output of 3,000 g/min and is the machine-rated output. In this work, we are
utilizing the Twintex roving TR PP 60 N 1870, which is a natural color polypropylene with a
60% glass content by weight at 1870 TEX. For this particular roving, processed at the maximum
fiber feed velocity, the material output translates to ~2.3 kg/min. Processing trials determined
the feasibility of running two tows through each tube of the P4 chopper, which effectively
doubled the throughput. However, this rate was difficult to sustain without material jamming
and additional tows could not be added due to physical and motor torque limitations. The P4
chopper allows selection of an infinitely-variable fiber length within a relatively wide range, but
utilization of more than two cut lengths within a single production operation has been very rare.
In order to potentially improve preforming throughput, a prototype chopper unit from Wolfangel
Composite Technology in Heimerdingen, Germany was procured. This chopper was designed to
process up to 5 tows at a maximum fiber velocity of ~900 m/min, which for the 1870 TEX
Twintex material translates to a material output of ~8.4 kg/min. The Wolfangel chopper as
shown in Figure 2 affords two distinct fiber lengths determined by blade spacing in either the
inside or outside half of the cutter roll. The positioning of the cutter roll is controlled by
pneumatic actuation; thereby switching the location of the cutter roller with respect to the fibers
and changing the fiber length. Economic goals were to be able to achieve 12 to 15 kg/min,
which would require increasing the number of tows being processed, speeding up the peripheral
speed, or combinations of both as described in the following sections.

Figure 1. P4 chopper as supplied by Aplicator. Housings with Aplicator identification
enclose servos for feeding and chopping fiber tows with variable cut lengths.
Figure 2. Wolfangel Chopper showing cutter wheel extended. Cutting takes place
on inner portion of the wheel in contact with the rubber roller.

2.2 Alternative Chopper Evaluation
Trials were conducted to evaluate the chopper using the supplied manual chopper controls and to
capture representative chopper-control signals from the P4 machine to approximate simulation of
the fabrication of a preform suitable for fabrication of an ~5kg liftgate structure similar in shape
to that shown in Figure 3. In typical processing, the chopper starts and stops very abruptly in
tight coordination with the robot movement as the robot traverses across each pass and then
indexes though multiple passes in uniformly spraying up the part profile. Obviously, the
economics of the process are improved as speed of the preforming process is increased up to the
repeatability and mechanical-loading capabilities of the preforming system. A program was
written in LabView to capture representative control signals from the robot that could then be
used to control the chopper for evaluation independent of operation of the entire preforming
system. A separate LabView program was written to provide necessary control signals to the
chopper using the data previously captured and manipulated into the necessary format for
LabView control, as shown in Figure 4. Also, hardware and program capability was provided to
automatically switch fiber length during processing.
Figure 3. Approximate liftgate dimensions utilized in simulation program.
Figure 4. Chopper control signals in simulation program.

A larger set of trials was conducted at ORNL with ACC support to evaluate the Wolfangel
chopper simulating the liftgate fabrication. During these trials, a number of enhancements were
made to processing conditions, including fiber ejection at the chopper as shown in Figure 5, and
other fiber handling equipment (as described in more detail in a later section). It was determined
that the chopper could easily handle up to 3 tows at close to the upper speed limit (900 m/min) to
deliver about 5.0 kg/min, but that the chopper motor stalled and did not have adequate torque to
handle more tows due to the pressure required to fully sever the tows. The system was supplied
with capability to meter 4 bar of pneumatic actuation pressure to the rubber roller/cutter wheel
interface and we found it necessary to operate at or slightly above this level to achieve adequate
severing at the upper levels of output. However, it appeared that achieving both higher speeds
and higher torque with a larger motor would be feasible with relatively minor modifications.
Figure 5. Photo during ORNL/ACC trials showing enhanced fiber ejection.
The existing Servowatt motor was rated at 0.75 Newton(N)-m maximum torque and 4,000
revolutions per minute (rpm). An Allen Bradley MPL-B430 motor and drive system was located
that was rated at 20 N-m maximum torque and 5,000 rpm. With this motor operating at its speed
rating, it should be able to achieve 12 kg/min utilizing about 6 tows with a maximum output of
over 18 kg/min utilizing 9 tows. The system was reconfigured with the new motor while
utilizing as much of the existing chopper hardware as possible as shown in Figure 6, although a
number of components had to be modified in order to utilize this physically larger and higherpowered
motor system.
The chopper system was upgraded and then the system was characterized and conditions
tweaked in extensive trials throughout a wide rage of potential operating speed ranges while
mounted on the test stand. Initial trials approached output targets, but the chopper motor/head
interface was inadequate for the higher torque and additional side loading added to increase the
cutter/roller interface pressures beyond the original allowance for 4 bar actuating pressure,

esulting in hardware failure. The insufficient hardware was upgraded and subsequent detailed
experiments were conducted at operating speeds from 30-100% of maximum cutter velocity
utilizing several tow number variations from 3 to 9, a range of cutter actuation pressures from 4-
10 bar, and fiber cut lengths of 25 mm and 75mm while mounted on the test stand. These trials
demonstrated the chopper to be capable of achieving more than 18 kg/min output when using 9
tows and the representative programming conditions described previously. As shown in Table 1,
good severing quality was achieved at these conditions when using 10 bar air pressure on the
cutter/rubber roller interface for the 25mm cut length, while the 75mm length at these conditions
was largely, but not completely, cut. This trend toward better cutting at the shorter length was
essentially replicated for all conditions at which any difference was noted. As expected,
improved severing was achieved at higher air pressures for essentially all speeds (not shown)
regardless of the number of tows being processed. Also expected was the finding (also not
shown) that cutting quality decreased with increasing the number of tows being simultaneously
cut. However, we somewhat unexpectedly found that cutting quality actually improved with
increased speed regardless of the number of tows being cut or interface pressure. When we could
not achieve complete tow severing, typically the glass fibers were completely or largely cut
while the polypropylene was typically the material not fully cut. This is not considered a
significant issue as the polypropylene is actually melted and flows to become the composite
matrix in subsequent operations. Since cutting quality was definitely improved at higher speeds
while somewhat decreased with larger numbers of tows, it is clear that for a given output target
that one would attempt to run at speeds approaching the upper end of the chopper capabilities
and run with the fewest numbers of tows necessary to achieve that output. It is worth noting that
this was achieved with the chopper mounted on a stationary test stand with the distance from
fiber spools to chopper significantly shorter than would be the case for operation on the robot.
As discussed in more detail later, higher velocities coupled with rapid stops and starts required to
maximize overall deposition rate create issues with managing tow inertia that are exacerbated as
the distance from spool to chopper is increased. Table 2 shows chopper output as a function of
the number of tows of the Twintex product and percentage of motor capacity.
Figure 6. Wolfangel Chopper with upgraded Allen Bradley Motor.

Table 1. Tow cut quality as a function of number of tows being cut, cut length, and chopper
output. To be considered good, essentially all reinforcing fiber is severed and only small,
isolated amounts of polypropylene are not completely severed.
Output # of Tows →
3 5 7 9
% Cut Length → 25mm 75mm 25mm 75mm 25mm 75mm 25mm 75mm
Poor x x
Needs Improvement x x
30 Marginal x
Acceptable x
Good x x
Poor x
Needs Improvement x x
50 Marginal
Acceptable x x
Good
Poor
x x x
Needs Improvement x x
70 Marginal x
Acceptable x
Good
Poor
x x x x
Needs Improvement x
90 Marginal
Acceptable x
Good
Poor
x x x x x x
Needs Improvement x
100 Marginal
Acceptable x
Good x x x x x x
Tests run at 10 bar roller/cutter interface actuation pressure
Table 2. Chopper output and velocity as a function of motor speed and number of tows.
Output Speed Velocity 1 Tow 2 Tows 3 Tows 4 Tows 5 Tows 6 Tows 7 Tows 8 Tows 9 Tows
% RPM m/min kg/min kg/min kg/min kg/min kg/min kg/min kg/min kg/min kg/min
5% 250 56
10% 500 112
15% 750 167
20% 1000 223
25% 1250 279
30% 1500 335
35% 1750 390
40% 2000 446
45% 2250 502
50% 2500 558
55% 2750 613
60% 3000 669
65% 3250 725
70% 3500 781
75% 3750 837
80% 4000 892
85% 4250 948
90% 4500 1,004
95% 4750 1,060
100% 5000 1,115
Chopper out with 1870 tex tows
0.10
0.21
0.31
0.42
0.52
0.63
0.73
0.83
0.94
1.04
1.15
1.25
1.36
1.46
1.56
1.67
1.77
1.88
1.98
2.09
0.21
0.42
0.63
0.83
1.04
1.25
1.46
1.67
1.88
2.09
2.29
2.50
2.71
2.92
3.13
3.34
3.55
3.75
3.96
4.17
0.31
0.63
0.94
1.25
1.56
1.88
2.19
2.50
2.82
3.13
3.44
3.75
4.07
4.38
4.69
5.01
5.32
5.63
5.94
6.26
0.42
0.83
1.25
1.67
2.09
2.50
2.92
3.34
3.75
4.17
4.59
5.01
5.42
5.84
6.26
6.67
7.09
7.51
7.93
8.34
0.52
1.04
1.56
2.09
2.61
3.13
3.65
4.17
4.69
5.21
5.74
6.26
6.78
7.30
7.82
8.34
8.86
9.39
9.91
10.43
0.63
1.25
1.88
2.50
3.13
3.75
4.38
5.01
5.63
6.26
6.88
7.51
8.13
8.76
9.39
10.01
10.64
11.26
11.89
12.51
0.73
1.46
2.19
2.92
3.65
4.38
5.11
5.84
6.57
7.30
8.03
8.76
9.49
10.22
10.95
11.68
12.41
13.14
13.87
14.60
0.83
1.67
2.50
3.34
4.17
5.01
5.84
6.67
7.51
8.34
9.18
10.01
10.85
11.68
12.51
13.35
14.18
15.02
15.85
16.69
0.94
1.88
2.82
3.75
4.69
5.63
6.57
7.51
8.45
9.39
10.32
11.26
12.20
13.14
14.08
15.02
15.96
16.90
17.83
18.77

2.3 Processing Experience Extended to Robot Mounting
This characterization experience was utilized to identify likely fiber handling impediments to
running the Wolfangel chopper on the preforming robot at these high output speeds, and system
improvements have been developed and implemented during the course of this activity. The
high speeds and large numbers of tows traversing through the system are taxing; more difficult
problems are encountered with the fiber inertia involved with the rapid starting and stopping of
the fiber associated with the robot starting and stopping at the end of each pass. A variety of
changes such as employing a number of hardened fiber guides in the system, utilizing an air
ejection system at the chopper output as previously shown, and the use of funnels to manage the
fiber “whipping” out of the spool were successfully implemented to mitigate speed and inertia
problems. With hardware changes and operational experience, capability to run at target outputs
on this stand for extended time periods was demonstrated.
Taking advantage of this background, system hardware and controls necessary to mount the
upgraded Wolfangel chopper on the robot (Figure 7 and 8) have been completed and the chopper
has now been characterized in making representative liftgate preforms and other test sections.
As previously mentioned, fiber inertia causes problems in working with the chopper mounted on
a stationary stand. These problems are exacerbated with abrupt movements of the robot and
further increased with the number and severity of fiber pathway turns required to reach the
chopper in various positions and also as the length of unsupported spans in the distance between
the fiber spools and the chopper increases. A number of additional changes have been made and
others continue to be implemented to improve the fiber delivery to the robot. Among the
changes are implementation of an “air braking” mechanism developed internally from related
hardware (Figure 9) to maintain tension close to the spools and use of fiber spreading and
ceramic eyelets to guide fibers (Figure 10).
Figure 7. Wolfangel chopper mounted on robot.

Figure 8. Upgraded chopper with air ejection system at the bottom.
Figure 9. Air braking mechanism to mitigate fiber inertia problems.
Figure 10. Fiber guides utilized to provide spacing and to mitigate tangling with 9 tow
deposition.

2.4 Chopper Deposition Uniformity
To understand and demonstrate the capabilities for uniform fiber deposition at high output rates,
the chopper has also been evaluated for uniform fiber distribution at various speeds taking into
account the rapid stopping and starting required for high speed deposition. A series of tests were
run assuming that the robot would be depositing a 6-pass pattern (3 passes down and back each
without stopping) over lengths of 0.5m, 1m, and 1.5m. Tests were run at 3 different robot speeds
and 3 different chopper output levels corresponding to 50, 75, and 100% of rated capability each.
The amount of material deposited was weighed for each test to judge deposition uniformity.
Actual results from this testing are shown in the Table 3. Table 4 shows the results that would be
expected for all other points normalized to the central data point for the 1.0m section at 75%
chopper speed and 75% robot speed with the other data calculated by speed ratios. The data
differences at each point are shown in Table 5 as a % difference from the expected value. As
might be somewhat expected, there tends to be slightly more fiber deposited than predicted at the
lower chopper speeds and slightly less than predicted at the higher chopper speeds. However,
the trends appear to be mixed versus the robot speed as slightly more fiber is deposited at both
higher robot and lower chopper speeds versus the reverse at both higher robot speeds and higher
chopper speeds. No effort has been expended to adjust robot dispenseware programming gains
or other settings to this point.
Table 3. Measured chopper output as a function of chopper speed and robot speed for 6 passes
over specified section lengths.
Robot Speed 100% 75% 50%
Chopper Speed Section Length Mass Mass Mass
m kg kg kg
100% 1.5 2.656 3.531 5.464
1 1.734 2.325 3.619
0.5 0.860 1.145 1.792
75% 1.5 2.057 2.771 4.193
1 1.334 1.820 2.800
0.5 0.669 0.896 1.389
50% 1.5 1.404 1.886 2.737
1 0.925 1.251 1.829
0.5 0.476 0.617 0.921

Table 4. Predicted chopper output based on scaling chopper and robot speeds and normalizing to
the measured output at 75% robot and chopper speeds and 1.0 m length.
Robot Speed 100% 75% 50%
Chopper Speed Section Length Mass Mass Mass
m kg kg kg
100% 1.5 2.730 3.640 5.460
1 1.820 2.427 3.640
0.5 0.910 1.213 1.820
75% 1.5 2.048 2.730 4.095
1 1.365 1.820 2.730
0.5 0.683 0.910 1.365
50% 1.5 1.365 1.820 2.730
1 0.910 1.213 1.820
0.5 0.455 0.607 0.910
Table 5. Percentage difference of measured output from Table 3 and predicted output in Table 4.
Robot Speed 100% 75% 50%
Chopper Speed Section Length Mass Mass Mass
m kg kg kg
100% 1.5 -2.7% 3.0% 0.1%
1 -4.7% -4.2% -0.6%
0.5 -5.5% -5.6% -1.5%
75% 1.5 0.5% 1.5% 2.4%
1 -2.3% 0.0% 2.6%
0.5 -2.0% -1.6% 1.8%
50% 1.5 2.9% 3.6% 0.3%
1 1.6% 3.1% 0.5%
0.5 4.6% 1.7% 1.2%

2.5 Liftgate Demonstration
With the combination of the system upgrades and process tweaking as previously described, we
were able to meet the objective of 12-15 kg/min deposition rate and demonstrate capability to
deposit a complete liftgate preform (Figure 11) in the 30 seconds assumed in the earlier process
comparison and economic study. Our capabilities exceed the deposition rate assumed for the
upper limit of a single robot/chopper system and the lower end of a dual system as assumed in
this study. Since the objective of this effort was to demonstrate preforming system output level
capability to verify economic model assumptions and at this point not to determine the upper
level output practical, we did not optimize the robot program for the time cycle and have not
attempted to run with more tows to increase this output. Also, since tooling does not exist to do
detailed evaluations of a final part, panels have not been molded from which properties can be
obtained. It is anticipated that further processing speed and product property relationships can be
established in follow-on work where tooling would be available. Plans are to demonstrate and
evaluate these relationships in looking at potential applications such as the ACC Focal Project 4
Composite Seat.
Figure 11. Representative unconsolidated liftgate preform.

3. CONCLUSIONS
Capability to uniformly sever and deposit thermoplastic tow preforms has been demonstrated at
output rates exceeding 15 kg/min using a modified chopper prototype mounted on the robotic
preforming system at ORNL. This effectively more than triples output previously achievable
with baseline equipment. With the combination of the system upgrades and process tweaking as
previously described, we were able to meet the objective of 12-15 kg/min deposition rate and
demonstrate capability to deposit a complete liftgate preform in the 30 seconds assumed in the
earlier process comparison and economic study. Our capabilities exceed the deposition rate
assumed for the upper limit of a single robot/chopper system as assumed in this study. Without
significant focus to this point on the programming and controls part of the system, this deposition
has also shown to be relatively uniform, varying approximately 5% above and below predictions
in this demonstration effort.
4. PLANNED FUTURE WORK
The ACC Focal Project 4 Composite Seat baseline design utilizes glass fiber with a
polypropylene matrix. Tooling is being procured for molding and testing potential materials
system approaches including the preforming approach as described here. ORNL will provide
preforms for evaluation in the seat project manufactured from the current Twintex baseline
material and utilizing the existing preforming screens. The first goal is to demonstrate technical
feasibility for this product form for this application before working towards optimizing robot and
chopping speeds and preform quantities per cycle specifically related to this application. ORNL
will collaborate with the ACC in evaluating this product form as well as the economics involved
in this process.
Hybrid-fiber preforms that can effectively take advantage of utilizing carbon and glass fibers
together in the same structure offer potential benefit in terms of economics and property
enhancement versus using only glass or carbon fiber independently and may be a good route for
actually introducing more carbon fiber at somewhat lower quantities than necessary for allcarbon
components in automotive applications. Since this process uses reinforcing fibers in
relatively simple and low cost product forms and allows placement of materials in specific
locations as desired for structural optimization, we are evaluating a variety of approaches to
introducing and utilizing hybrid forms in applications that could have significant impact to DOE
goals in lightweighting of vehicles.
5. ACKNOWLEDGEMENTS
This Research sponsored by the U.S. Department of Energy, Assistant Secretary for Energy
Efficiency and Renewable Energy, Office of Vehicle Technologies, as part of the Lightweight
Materials Program. LabView programming was provided by Donald L. Erdman. The authors
would also like to express appreciation for related contributions of others at ORNL, the ACC,
and EPFL.

6. REFERENCES
1. N. G. Chavka and J. S. Dahl, P4: Glass Fiber Preforming Technology for Automotive
Applications”, 44 th International SAMPE Symposium, May 1999.
2. S. T. Jespersen , F. Baudry, D. Schmäh, M. D. Wakeman, V. Michaud, P. Blanchard, R. E.
Norris, J.-A. E. Månson, “Rapid Processing of Net-Shape Thermoplastic Planar-Random
Composite Preforms”, Applied Composite Materials (2009) 16:55–71.
3. S. T. Jespersen , F. Baudry, M. D. Wakeman, V. Michaud, P. Blanchard, R. Norris, J.-A. E.
Månson, “Consolidation of Net-shape Random Fiber Thermoplastic Composite Preforms”,
Polymer Composites, published online March 2009.
4. S. T. Jespersen , F. Baudry, M. D. Wakeman, V. Michaud, P. Blanchard, R. Norris, J.-A. E.
Månson, “Flow Properties of Tailored Net-Shape Thermoplastic Composite Preforms”,
Applied Composite Materials, published online September 2009.
5. M. D. Wakeman, G. Francois, J-A. E Manson, “Cost Modeling of Thermoplastic P4 Process
Technology”, Progress Report1 for the Automotive Composites Consortium, Ecole
Polytechnique Fédérale de Lausanne, April 2006.