2.

29
Continuous Molding of
Thermoplastic Composites
A. G. GIBSON
University of Newcastle upon Tyne, UK
2.29.1 INTRODUCTION

2.29.2 THE IMPREGNATION PROBLEM

2.29.3 PROCESS MODELING

2.29.4 IMPREGNATION PROCESSES

2.29.4.1 Melt Impregnation

2.29.4.1.1 Extrusion compounding
2.29.4.1.2 Strand impregnation processes
2.29.4.1.3 Belt press processes
2.29.4.2 Powder Processes
2.29.4.2.1 Aligned fiber techniques
2.29.4.2.2 The Radlite process
2.29.4.3 Fiber Co-mingling
2.29.4.4 Solvent-based Processes
2.29.4.5 Reactive Techniques

7
7
7
9
10
10
12
12
13
14

2.29.5 CONTINUOUS PROCESSES FOR THERMOPLASTIC COMPOSITES

2.29.5.1 Thermoplastic Pultrusion
2.29.5.1.1 Reinforced extrusions and moldings
2.29.5.1.2 Roll forming
2.29.5.2 Thermoplastic Tape Laying and Filament Winding
2.29.5.3 Reinforced Thermoplastic Pipe (RTP)
2.29.6 REFERENCES

2.29.1

14
14
16
16
16
17
18

INTRODUCTION

These are the factors driving the development

of thermoplastic composite technology, which
is now one of the most rapidly expanding sectors of the composite market. Surprisingly,
until the 1980s, the difficulty of achieving full
impregnation and wetting of the reinforcement
with thermoplastic resin prevented the costeffective manufacture of continuous fiber reinforced thermoplastic composites, and confined
commercial activity mainly to short fiber injection molding materials. Once the full potential
of thermoplastic matrix composites was realized, however, numerous promising processing

Thermoplastic matrix composites offer the

following advantages over their thermosetting
counterparts:
(i) a rapid fabrication cycle which avoids the
need for a curing reaction,
(ii) cleaner processing,
(iii) precursor materials with infinite shelf
life,
(iv) a broad selection of tough matrix materials, and
(v) recyclability.
1

Continuous Molding of Thermoplastic Composites

Figure 1 Alternative fabrication routes for the processing of thermoplastic matrix composites.

methods were developed and refined, with the

result that several manufacturing routes,
described in Figure 1, are now available
(Leach, 1989; Gibson and Manson, 1992; Cogswell, 1992).
Precursors for thermoplastic composites can
be broadly divided into two types: consolidated
and nonconsolidated (requiring further processing to produce wetting of the fibers by the
matrix) (see Chapters 2.16, 2.26 and 2.27, this
volume).
Each product form has particular advantages
and drawbacks. Fully consolidated material,
e.g., has optimum mechanical properties but is
springy, hard, and elastic, which renders it difficult to handle, shape, or lay up into complex
shapes. The nonconsolidated materials, by contrast, are much easier to handle and can, for
instance, be woven into fabrics that have good
drape properties. Their disadvantage is that
they need to be compressed or worked in some
way to consolidate them, and this results in
significant reduction in bulk, which is difficult
to accommodate in some molding processes.
The processes that have shown the most impressive growth to date are direct melt impregnation and fiber co-mingling. The former is
used for long fiber injection molding materials,
unidirectional thermoplastic prepregs, and
glass mat thermoplastics. Co-mingled products
are frequently converted into woven fabrics,
processed by vacuum bag consolidation or
press molding.
Although thermoplastic composites have
been available since the 1980s, their penetration
into the wider composites markets was initially
rather slow, and some of the anticipated advantages were more difficult to realize than had
been envisaged. The fast, chemistry-free processing cycle, for instance, was not free of problems. The concept of rapidly heating, forming,
and cooling is attractive, but some difficulties

needed to be overcome to put it into place. The

temperatures to which thermoplastics must be
heated for processing, for instance, are higher
than with thermosets, which presents problems
with equipment design and material compatibility. Moreover, if a part is to be formed in
matched steel tooling, it cannot be demolded
while the resin is still liquid, so how should it be
cooled? One answer is to preheat the composite,
then combine the forming and cooling operations. Hot stamping of glass mat thermoplastics
and other materials is now one of the most
active areas of thermoplastic composite technology.The underlying principle with all thermoplastic composites is to consider the complete
process cycle, rather than simply the preparation of pre-impregnated materials.
Early research concentrated on high performance applications and led to the development
of aromatic polymer composites which, in
many ways, outperformed their thermosetting
counterparts (Cogswell, 1992). It was soon realized, however, that thermoplastic technologies,
based on medium cost polymers, such as polypropylene, polyesters, and polyamides, was
capable of penetrating a much broader range
of markets and applications.
Problems with processing technologies have
now been overcome and the green requirements of clean processing and recyclability are
of increasing importance, so the thermoplastic
route, based on some of the technologies reviewed here, is assured of an increasingly important role in the future.

2.29.2

THE IMPREGNATION PROBLEM

The principles of manufacturing with thermoplastic composites have been discussed

widely by, e.g., Gibson and Manson (1992),

The Impregnation Problem

Figure 2

The problem of impregnating a roving, containing a large number of small diameter fibers, with a
viscous thermoplastic resin.

Chang and Lees (1988), Cogswell (1992), and

Mantell and Springer (1992). The key problem,
illustrated schematically in Figure 2, lies in the
effective impregnation (see Chapter 2.17, this
volume) of fiber tows (or a fiber network) with
a thermoplastic resin. The viscosity of conventional thermoplastics under melt processing
conditions is usually in the range
502000 Pa.s, which contrasts with thermosets
in their noncross-linked state where the viscosity seldom exceeds 50 Pa.s. In the case of the
2400 tex roving shown in Figure 2, continuous
impregnation would involve wetting the surface
of all the glass fibers. This surface, if
unwrapped and laid flat, would be equivalent
in area to a continuous strip 300 mm wide. The
viscous resin would have to be spread out to a
thickness roughly equivalent to the interfiber
spacing, i.e., approximately 9 microns thick.
The impregnation of a bed of fibers, Figure 3,
can be described in terms of Darcy's law which,
for 1-D flow, gives
u

dx K dp

dt Z dx

where u is the flow velocity of the resin, the

pressure gradient is dp/dx, K is the permeability
of the fiber array, and D is the polymer melt
viscosity. Observing that the pressure gradient
in this case is simply P/x and integrating Equation (1) for the boundary conditions in Figure 3
gives the impregnation time
timp

ZX2
2KP

This relationship, although approximate, serves

to demonstrate the problems involved in impregnating fibers arrays with thermoplastic

melt. The importance of the impregnation

length can be seen from the squared term in X.
The principle on which all effective impregnation processes operate is to manipulate the variables in Equation (2) to attain an economically
short impregnation time. For processes with
polymer melts the most straightforward method
of achieving this is clearly to minimize X.
The pressure, which drives the flow of polymer, contains components due to the external
mechanically applied pressure, Pm, and the capillary pressure, Pc, so that
P Pm Pc

The capillary pressure, discussed in greater detail elsewhere (Connor, 1995), results from the
change in surface energy when the polymer wets
the surface of the reinforcement. Unfortunately, capillary pressure values for polymer/
fiber systems are generally relatively small, in
the range 440 kPa, as observed by Ahn et al.
(1991), so the impregnation time would be very
long if one relied on capillary effects alone to
achieve impregnation (Lee and Springer, 1987).
In most impregnation processes the capillary
pressure is small enough to be neglected in
comparison to the applied pressure needed to
drive the process at an economic rate.
Permeability is a key parameter in impregnation processes. Assuming the fiber bed to consist of an array of parallel fibers, the
permeability can be estimated from the modified CarmanKozeny equation
K

r2f 1 vf 3
4c
v2f

Values of the Kozeny constant, c, are different

for flow parallel and perpendicular to the fibers

Continuous Molding of Thermoplastic Composites

Figure 3 Impregnation of a fiber array by resin under an externally applied pressure.

and they are also dependent to some extent on

the particular resin/fiber system and on the
experimental method used, as observed by Williams et al. (1974). The applicability of this
approach to thermoplastic melts, which display
non-Newtonian characteristics, has been examined by Astrom (1992) and Astrom et al. (1992).
Improved models for the permeability of
composite fiber arrays have been proposed by
Gebart (1992) and Gutowski et al. (1987),
amongst others, but in all cases the permeability is predicted to be proportional to the square
of the fiber radius. This is exploited in the
production of special grades of glass fiber for
thermoplastic composites, which employ significantly larger fiber diameters than the 10 microns that is widespread for thermosets. Glass
fibers for continuous fiber reinforced thermoplastics are often supplied with 18 mm diameter,
sometimes up to 25 mm, for which Equation (4)
predicts permeability improvements of more
than a factor of six. It should be borne in
mind, however, that the strength of glass fibers
depends on diameter, and larger fibers often
have strengths as low as 1400 MPa, compared
to about 1800 MPa for 10 mm fiber.
Unfortunately carbon fibers, which are
79 mm in diameter, are not similarly amenable
to manipulation, so impregnation processes
generally run more slowly than those for glass.
Applied transverse stress significantly affects
the permeability by forcing the fibers closer
together. This deformation has been likened
to that of a nonlinear spring, and has been

studied by several workers, including Kim

et al. (1991) and Toll and Manson (1995).
Strictly speaking the K value in Equations
(1)(3) should be treated as pressure-dependent. One simple functional form of pressure
dependence that can conveniently be assumed
for modeling purposes is


Pm
KP K0 exp 0
P

where K0 is the permeability at low or negligible

pressure (from Equation (4) or similar) and P' is
a constant with the units of pressure.
The detrimental effect of compression on
permeability was not understood in the early
days, and it took some time before it was
realized that high levels of transverse pressure
are counterproductive to impregnation. Despite the presence of pressure in the denominator of Equation (2) the most successful
impregnation processes employ only modest
pressure. In practice it is often found that the
pressure which gives the highest value of the
product, KP is of the order of 12 MPa.
Viscosity, of course, appears in Equation (3)
and benefits can be obtained by keeping this to
a minimum, as discussed later. Viscosity can be
lowered by several means, which include the use
of local high temperatures, low molecular
weight thermoplastics, and, in certain processes, solvents, or plasticizers (Cogswell et al.,
1981a, 1981b; Cogswell, 1992; Goodman and
Loos, 1990).

Process Modeling

Figure 4 Model prediction for impregnation time (log scale) vs. applied pressure for various thermoplastic
composite systems of interest, using model parameters given in Table 1.
Table 1 Parameter values used in the process model.

Glass/polypropylene (low MW)

Glass/thermoplastic
Carbon/thermoplastic
Carbon/PEEK

Viscosity, Z (Pa.s)

vf

rf (mm)

10
100
100
500

0.3
0.3
0.6
0.6

9
9
3.5
3.5

Capillary pressure, Pc = 0.04 MPa; Kozeny constant, c = 9; P' = 2 bar; impregnation length,
X = 100300 mm.

The pronounced effect of the squared term in

flow length, already noted, provides the most
powerful means of reducing the impregnation
time. Clearly it is desirable that the polymer and
fiber components of a thermoplastic composite
should be intimately mingled together prior to
impregnation in a manner which minimizes X.
Most techniques for the preparation of wellwetted composites with thermoplastic matrices,
therefore, involve two stages:
(i) Stage 1: The intimate mechanical mingling of the fiber and matrix, and
(ii) Stage 2: Melting of the thermoplastic
polymer with the application of a modest level
of pressure to promote impregnation.
The three most important methods of reducing X are pin impregnation, powder impregnation, and commingling. These, and others,
will be discussed in the next sections.

2.29.3

PROCESS MODELING

Equations (1)(4) can be used, along with

appropriate values of the impregnation length,
X, as the basis of simple process models, as

reported by Chang and Lees (1988) and Gibson

and Manson (1992). In the present case, the
effect of pressure will be included through the
incorporation of Equation (5). Figure 4 shows
the predicted relationships between impregnation time and pressure for a range of thermoplastic composite systems of interest, calculated
from the data given in Table 1. The four fiber/
resin combinations chosen all relate to commercially interesting systems. X values of practical
interest for flow processes transverse to the
fibers are generally in the region 100300 mm.
Each material combination therefore produces
a band of typical impregnation times as a
function of increasing applied pressure. At
zero applied pressure the impregnation process
is driven solely by wetting effects, so the
impregnation times are long. As the applied
pressure increases, the impregnation time
decreases rapidly until the effect of the fiber
tow compressibility on permeability begins to
show. Eventually, at high values of applied
pressure, the impregnation time begins to
increase again, because of the pressure effect.
It can be seen therefore that high, applied pressures are generally counterproductive in the
impregnation process.

Continuous Molding of Thermoplastic Composites

Figure 5 Impregnation time (log scale) vs. pressure, as in Figure 4, with process windows for different
impregnation techniques superposed.

In the case of glass fiber with low molecular

weight polypropylene, melt viscosity values as
low as 10 Pa.s can be achieved, leading to very
low impregnation times. With other polymers,
however, it is often difficult to obtain materials
with viscosities lower than about 100 Pa.s,
which gives impregnation times an order of
magnitude larger. It can be seen that changing
from glass to carbon fiber results in more than
an order of magnitude increase in impregnation
time. This is due partly to the smaller fiber
diameter and partly to the fact that higher
fiber volume fractions are of interest in the
carbon fiber case. Finally, the most difficult
range of systems to melt impregnate can be
seen to be the ones based on carbon fiber with
high viscosity polymers, such as PEEK, where
the impregnation times extend to more than one
hour. In the latter case, special measures, beyond conventional melt impregnation, are often
required to achieve well-impregnated product.
Individual categories of impregnation process can each be characterized by a process
window of achievable pressure and impregnation time. Figure 5 shows the same material
impregnation data as Figure 4, with process
windows for different techniques superposed.
In each case the upper edge of the process
window indicates the practical limit that can
be achieved in terms of impregnation time. It
can be seen that the most rapid impregnation
times are achievable with pin impregnation,
which will be described in the next section. This
technique is not applicable, however, to the
materials of highest viscosity. The other continuous processes shown are the belt press,

which extends the applicable pressure and compounding extrusion, traditionally used for
injection molding materials, which is applicable
only to short fiber products.

2.29.4

IMPREGNATION PROCESSES

In this review of materials systems and processes, the need for compatibility and a strong
interface between the matrix resin and the fibers
is, in most cases, an important requirement that
significantly influences properties and applications. The advent of fibers specifically tailored
for use with engineering thermoplastics is a
significant advance and improvements are continuing in this area.
The glass/polypropylene system is worthy of
special note, as this is one of the most commercially significant material combinations (Gibson, 1994). Although pure polypropylene is
nonpolar, and adheres poorly to glass, the addition of an internal coupling agent in the form
of a functionalized polyolefin with maleic or
acrylic acid groups grafted on to the main chain
can confer a considerable improvement in interlaminar properties (Adur et al., 1989; Constable et al., 1989; Constable and Adur, 1991;
Rijsdijk et al., 1993). Functionalized additives
of the Polybond2 type are most effective when
used in conjunction with glass fibers treated
with a suitably compatible size. Another benefit, specific to polypropylene, is the ability to
clip the high molecular weight tail from the
polymer by melt processing with an organic

Impregnation Processes

Figure 6 The extrusion compounding process for the manufacture of short fiber reinforced thermoplastic
injection molding granules.

peroxide, producing resins of very low viscosity

that perform extremely well in the melt impregnation process, as seen in Figures 4 and 5.
With all reinforcements to be used with thermoplastics it should be remembered that the
processing temperatures are much higher than
those employed for thermosets, so the size coating on the glass should be appropriately temperature-resistant, in addition to being
compatible with the polymer.
2.29.4.1
2.29.4.1.1

Melt Impregnation
Extrusion compounding

Until about 1980 the main examples of thermoplastic matrix composites were short fiber
reinforced injection molding materials (see
Chapter 2.30, this volume), manufactured by
the process of extrusion compounding, shown
in Figure 6. This is still one of the most effective
methods of producing molding pellets and
accounts for more than half the current tonnage
of glass fiber reinforced thermoplastics.
Significant improvements have taken place
over the years in the design of compounding
extruders. These include the facility for downstream addition of the reinforcement, which
reduces work by allowing the polymer to be
fully melted before it is brought into contact
with the fibers. Nevertheless, significant levels
of fiber breakage are virtually unavoidable if
the fibers are to be fully wetted by the melt
compounding route.
The residual fiber length in injection molding
compounds is usually 100600 mm, with a
weight average fiber length of about 250 mm.
This compares to critical fiber length values of
about 200 mm for nylon and 400 mm for polypropylene. While the addition of short glass
fibers to polymers such as polypropylene or
nylon undoubtedly results in useful increases
of stiffness and strength the reinforcing efficiency is only of the order of 40%, as observed

by Bader and Bowyer (1973). Several studies

have been carried out on the breakage of fibers
during compounding, including those of Franzen et al. (1989) and von Turkovich and Erwin
(1983).
One interesting argument in relation to fiber
length preservation is that, in processes which,
by mixing, impose a fully random fiber orientation distribution, the fiber aspect ratio is controlled ultimately by the ability of randomly
oriented fibers to pack in space. Evans and
Gibson (1986) showed that for random fiber
orientation there is a simple inverse relationship
between the maximum achievable fiber aspect
ratio (l/d)max and fiber volume fraction, vf so
that
 
l
k

d max vf

This is consistent with the results of the experimental studies of Milewski (1973, 1978) on
random fiber packing (for an experimentally
determined value of k = 5.3), and also describes
the residual fiber length after melt compounding in many cases. Equation (2), therefore,
places a fundamental restriction on the fiber
length that can be retained in processes that
randomize the orientation of the fibers.

2.29.4.1.2

Strand impregnation processes

As early as the 1960s, attempts were being

made to overcome the limitations of the compounding extruder. Several workers realized
that it was undesirable to process the reinforcement through the extruder along with the polymer, as this caused problems with machine wear
as well as with fiber breakage. The most promising early attempts to overcome the restrictions of melt compounding involved the
passage of fiber tows through cross-head dies
similar to those used for wire-coating extrusion,
as shown in Figure 7. The process patented by

Continuous Molding of Thermoplastic Composites

Figure 7 Cross-head die, or wire coating extrusion process for the continuous manufacture of partially
impregnated injection molding materials.

Figure 8 The pin impregnation process for continuous pultrusion of well-impregnated unidirectional
tows: (a) impregnation bath and (b) mechanism of
impregnation.

Bradt (1959, 1962) employed a technique similar to this, as did those of Moyer (1976) and
Hattori et al. (1977), in which a feedstock based
on a polymer melt or aqueous latex was combined with glass fibers to form fiber concentrates. Experiments on cross-head extrusion
were also reported by Bader and Bowyer
(1973).
Cross-head extrusion, followed by chopping
of the strand into pellets for molding, seemed to
offer a number of advantages over the compounding extruder, including reduced machine
wear and lower power consumption. However,
in the early continuous strand process impregnation was incomplete, with the melt rarely
penetrating beyond the outer layers of the pellet. In hindsight, it can be seen that the reason
for this is the pressure dependence of the permeability mentioned earlier: the external pres-

sure exerted on the tow by the polymer melt

served to lock the individual fibers together,
severely hindering the penetration process.
With only partial impregnation, the task of
wetting the fibers at the core of each pellet was
transferred from the extruder to the injection
molding machine, and this caused some machine wear and often produced unwetted or
undispersed fibers in moldings. Nevertheless,
some longer fibers were often retained intact,
which produced parts with improved impact
properties, so early processes of the strand impregnation type achieved some success in niche
markets.
The problem of achieving a substantially
completely wetted product by the melt route
was solved in the 1980s with an improved melt
pultrusion technique for long fiber pellets
(Cogswell et al., 1981a, 1981b). The Verton
process, which pointed the way towards many
current thermoplastic impregnation processes,
addressed the problem of reducing the flow
length for impregnation by opening and flattening the fiber tow.
Patents for the new process taught the importance of spreading the fiber tows and of
ensuring that the melt viscosity of the polymer
was as low as possible. Impregnation devices,
Figure 8, are quite simple and resemble superficially the baths used in the processing of
thermosetting composites. Pin impregnation
works, first by using cylindrical spreader surfaces to open up and spread out the roving.
When this is accomplished the thickness of the
fiber tow can be reduced to just a few fiber
diameters, resulting in a flow length of the
order of 100200 mm. Following this, a film of
polymer melt is entrained between the moving
fibers and the pin. This generates a modest
pressure that drives the melt into the fibers, as
shown. The process usually requires several
pins to achieve good impregnation.
The flow mechanisms operating in pin
impregnation have been studied by a number
of workers, including Chandler et al. (1992),

Impregnation Processes

Figure 9 Arrangement of impregnation cells on a production unit for long fiber injection molding
materials.

Peltonen et al. (1992), Bijsterbosch and Gaymans (1993a, 1993b), and Bates and Charrier
(1999). As might be expected, there have been
several variants of the process, some of which
have been patented. Benefits have been variously claimed, for instance, for rotating the
pins, varying the pin surface profile, injecting
the resin through slots in the pin, or through a
porous pin material. Commercial versions of
the process generally employ many impregnation cells, side by side, fed from a common
extruder, as in Figure 9. In an alternative configuration a single broad impregnation cavity
may be employed, through which many separate rovings are passed and the pins may be
replaced by curved, wavelike, or angled spreader surfaces (Hawley, 1982).
Although many thermoplastics have been
processed by melt impregnation, nylon 66 and
polypropylene have been found to work very
effectively. The molecular weight needed for
good mechanical properties in thermoplastic
matrix composites is substantially lower than
that required in conventional thermoplastic
products, permitting the use of thermoplastic
melts of much lower viscosity than would normally be encountered: 30 Pa.s and even as low
as 10 Pa.s have been advocated (Cogswell et al.
1981a, 1981b). Low viscosity grades of polypropylene, which are particularly suitable for
this process, may be prepared by compounding
the resin with an organic peroxide. It has also
been suggested that viscosity can be lowered in
the vicinity of the impregnation pins by heating
them to a higher temperature than the rest of
the bath.
In most processing operations a compromise
has to be achieved between line speed, degree of

wetting, and polymer viscosity. Nevertheless,

line speeds in excess of 30 m min71 are often
encountered in processes for nylon or polypropylene injection molding pellets. As might be
expected, higher fiber volume fractions, or the
use of carbon fibers instead of glass, necessitate
the employment of lower line speeds, because of
the lower permeability.
Pin impregnation is used most commonly for
the manufacture of long fiber injection molding
pellets. Variants of the process, however, are
readily adaptable for making continuous thermoplastic prepregs, such as Plytron2 (glass/
polypropylene) and APC (carbon/PEEK).

2.29.4.1.3

Belt press processes

A belt press enables the processes of pressing

(with consolidation and elimination of air) and
cooling to be carried out in a continuous operation. These processes are mainly used in the
manufacture of glass mat thermoplastic
(GMT) prepregs, for subsequent processing
by hot stamping or pressing (see Chapter 2.27,
this volume). In the Azdel-Laminate process,
Figure 10, the reinforcement can take the form
either of chopped strands or continuous fiber
swirl mat. A film of matrix polymer, most
frequently polypropylene, is inserted or
extruded between the layers of mat prior to
entering the belt press. In one version, recycled
in-house scrap is incorporated into the central
layer of the GMT.
The moving belt is a stainless steel band
and, within the press, there are stations for both
heating and cooling the product, while exerting

10

Continuous Molding of Thermoplastic Composites

Figure 10 The melt process for the manufacture of stampable glass mat thermoplastics (GMTs).

form of strands or bundles, e.g., the flow of

polymer may often be sufficient only to surround the bundles, without fully impregnating
individual strands. The key to the success of this
type of process is the recognition that, in many
circumstances, including GMT manufacture,
complete impregnation is not necessary.
Belt press processes are also beginning to be
investigated for processing woven continuous
fiber materials, but again it is difficult to
achieve complete impregnation by this route.

2.29.4.2
2.29.4.2.1

Figure 11 Powder impregnation: (a) powder impregnated roving and (b) schematic of a powder
impregnation process.

a through-thickness pressure to ensure that a

certain level of impregnation takes place as the
polymer is melted, followed by consolidation as
it cools. The outstanding advantage of the belt
press is that, despite its relatively high capital
cost, it makes possible the continuous manufacture of thermoplastic prepregs. Because the
individual rovings are not spread out to minimize the flow length, belt processes generally do
not achieve full wet-out of the prepreg by the
polymer melt. When the reinforcement is in the

Powder Processes
Aligned fiber techniques

Processes of this type rely on incorporating

polymer, in the form of powder, into the interstices between the fibers, as shown in
Figure 11(a), in order to reduce considerably
the flow length required for impregnation. This
method is particularly attractive when the polymer is available in the form of powder. One of
the earliest references to powder impregnation
is the patent of Price (1973). The hot tube
process, as it was known, was used successfully
for some years as an alternative to melt impregnation for long fiber injection molding pellets.
Several others, including Chabrier et al. (1986),
Muzzy et al. (1989, 1990), de Jager (1987),
Hartness (1988), Iyer and Drzal (1990), Ogden
et al. (1992), Ye et al. (1992), Hugh et al. (1992),
Baucom and Marchello (1994), and Wagner
and Colton (1994), subsequently discussed or
patented versions of the process. Ramani et al.
(1992), Texier et al. (1993), and Vodermeyer
et al. (1993) described similar processes based
on aqueous suspensions of powder. These various processes have been reviewed in greater
detail by Miller and Gibson (1996).

Impregnation Processes

11

Figure 12 The melt bridge model for powder impregnation (Connor et al., 1993) shows bridges between
fibers which flow axially under applied pressure, to achieve impregnation.

Fiber tows can be impregnated continuously

with powder by various means, the simplest of
which is through the action of pins or rollers in
a bed or suspension of the powder. The mechanism of initial dispersion of the powder within
the strand is rather similar to that described
previously for melt impregnation. The strand is
opened up and spread by the action of pins,
then the powder is mechanically forced between
the fibers by the trapping of particles between
the moving fibers and the surface of the pin.
There are several versions of the powder
impregnation process, one of which is shown
in Figure 11(b). It is possible, e.g., to employ
gas fluidization or mechanical vibration to agitate the powder. Gas fluidization presents difficulties in terms of the finer polymer particles
being blown out of the bed. It is usually necessary to trap these particles and return them to
the bed. There can also be an explosion hazard
if air is used as the fluidizing medium; the use of
inert gas as a substitute significantly increases
the cost of the process. Some processes employ
electrostatic charge to attract the polymer to
the fibers, while others rely simply on mechanical forces. It is possible to employ a venturi or
an air knife to open up the rovings prior to
insertion of the polymer particles.
The polymer powder should ideally be in the
form of near-round particles. There are economic advantages in using polymers which are
available as powder direct from the polymerization process and these, in addition, tend to be
more regular in shape than those produced by
grinding processes such as cryogrinding. In
principle it might be seen as desirable for the
powder particle size to be small, and similar to
the fiber diameter, but this is often difficult to

achieve, as particle size tends to be limited by

economic considerations. In practice, powders
with diameters up to 150 mm have been used
with success.
One key problem is control of the resin to
fiber ratio. In some versions of the process an
excess of powder is incorporated at the impregnation stage. The surplus is then removed by
means such as scraping, mechanical vibration,
or the use of a wipe-off die (following melting
and consolidation). One convenient method of
controlling resin/fiber ratio is to use a powder
bed in which the impregnation pins are on an
inclined ladder, as shown in Figure 11. The pins
above the powder level have the effect of removing powder, so polymer pick-up can be
maintained accurately by controlling the level
of powder in the bed.
Processing the impregnated strand into a consolidated laminate involves two further steps.
First, the polymer particles are melted using a
hot air or infrared oven. Following that, it is
then necessary to consolidate the material. Various models have been proposed to describe the
consolidation process by, among others, Lee
and Springer (1987), Connor et al. (1995), and
Haupert and Friedrich (1995). Generally, it is
agreed that the main mechanism of consolidation in powder impregnated tows involves flow
of resin along the fibers, since the axial permeability is about an order of magnitude greater
than that in the transverse direction. According
to Connor et al. (1995) the melted particles form
bridges between adjacent fibers, as shown in Figure 12. Lateral pressure must then be applied to
promote consolidation, so that each bridge
spreads along the fiber, displacing air, until it
meets a neighboring one coming the other way.

12

Continuous Molding of Thermoplastic Composites

Figure 13 The Radlite process for the manufacture of GMT sheet from thermoplastic powder and fibers.

In continuous processing, consolidation may

be achieved by drawing the moving tow over a
number of pins or cylindrical surfaces in much
the same way as discussed above for strand
impregnation. This process has been described
by Miller et al. (1998). Alternatively the tow
may be pulled into a convergent die. In cyclic
molding processes the consolidation is achieved
by direct pressing in the case of matched die
molding or through the use of combinations of
vacuum and external pressure in the case of
autoclave or vacuum bag molding.
The consolidation step can be carried out inline with powder impregnation, but more frequently it is desirable to process the tows into a
convenient form, such as woven fabric, prior to
final fabrication. Two methods have been proposed to ensure that the powder remains within
the roving during subsequent handling. Ganga
(1986) recommended that the impregnated tow
be coated with a thin-walled tube of polymer
film, whereas Muzzy and Colton (1995) describe the fusion of the powder to keep it in
place. In the latter case the flexibility of the tow
is retained.

2.29.4.2.2

The Radlite process

The Radlite process for GMT, Figure 13,

employs paper-making technology to disperse
polymer powder and glass fibers in an aqueous
foam. This is then deposited on a continuous
moving filter and the aqueous component
removed by vacuum. After drying, the composite paper is passed through a belt press,
where the polymer is melted and the composite
consolidated by the application of moderate
pressure. By virtue of the foam dispersion
step, GMTs made by the Radlite process gen-

erally contain more completely dispersed fibers,

compared with those made by the melt route,
where the fibers tend to remain in bundles. The
paper-making deposition process also results in
some preferential fiber orientation in the
machine direction.

2.29.4.3

Fiber Co-mingling

A further method of achieving initial intimacy between fires and matrix is by the random
intermingling of fiber and matrix materials in
the form of parallel fibers (Coldicott et al., 1989;
Hamada et al., 1993; Miao et al., 1994; Svensson et al., 1998). Co-mingling, Figure 14(a), can
be achieved with either continuous or discontinuous fibers. In the processes employed by
Vetrotex (St. John, 1995) and Trevira Neckelmann (1999) glass and thermoplastic fibers are
mingled continuously to form a single tow. It
should be remembered that, even when the two
different fiber types are dispersed in a completely random manner, fiber-rich and resin-rich
areas will naturally be present, so the flow
length needed to achieve complete impregnation may be several times the interfiber spacing,
as shown in Figure 14(b). The co-mingled systems used most widely are glass/polypropylene,
glass/PET (and the copolymer, PETG), and
glass polyethylene.
The largest commercial tonnage of comingled material processed is glass/polypropylene. It is recognized, however, that the interlaminar properties achievable with glass/
polypropylene may not be sufficient for some
structural applications. Although the technology exists to achieve good coupling between
polypropylene and glass (Constable et al.,
1989), the relatively low glass transition tem-

Impregnation Processes

13

Figure 14 Fiber co-mingling: (a) co-mingled fibers tow and (b) resin and fiber-rich regions resulting from
random dispersal.

perature limits the interlaminar shear strength

achievable at room temperature. Since both
compressive strength and flexural strength are
determined to a large extent by the interlaminar
properties, the limitations on these properties
restrict the load-bearing applications in which
polypropylene-based composites may be used.
The limitations of polypropylene have stimulated interest in cost-effective matrix materials
with higher glass transition temperatures, notably polyesters and polyamides. Recently, comingled materials have been developed based
on PET and its amorphous copolymer PETG
(Trevira Neckelmann, 1999), the latter being
favored in certain types of process because of
the high melting point of PET (285 8C).
The co-mingling processes employed by
EMS-Chemie (1999) and Schappe Techniques
(France) to process carbon fibers and nylon 12
(McDonnell et al., 1999), and by others to
process various materials (Ives and Williams,
1988; Coldicott et al., 1989) employs stretchbreaking and twisting to combine two different
types of fiber. The technique is well known in
the textile industry as a method for combining
different fiber types. Stretch-breaking is performed by passing the fiber tow over two sets
of rollers, the second rotating at a higher speed
than the first. Although it might be expected
that stretch-broken reinforcement, being discontinuous, would give products of inferior
strength it should be borne in mind that the
distance between breaks is very large compared
with the critical fiber length. It is also argued
that the stretch-breaking process acts mainly on
the statistical weak points in the fiber, which
would in any case have broken on loading of
the composite, and that the slight lack of alignment of the twisted structure gives improved
interlaminar properties. Often it has been found
that well-consolidated composites from the
stretch-breaking route have properties similar
to or only slightly inferior to those of continuous fiber materials.

Co-mingled materials may be processed in

the form of continuous rovings, by pultrusion,
filament winding, or braiding. Alternatively
they may be woven into fabrics for subsequent
fabrication operations. As in the case of powder
impregnated prepregs, the fabrication step for
co-mingled composites requires both melting of
the thermoplastic fibers and consolidation
under pressure, to cause the reinforcing fibers
to be impregnated and wetted by the matrix.
The mechanisms of impregnation have been
investigated and models for the process have
been proposed by, amongst others, Bernet et al.
(1999), Van West et al. (1991), Klinkmuller et al.
(1995), Cain et al. (1997), Miller et al. (1998),
and Wilks et al. (1999).
An important factor in developing successful
co-mingled systems is control of the diameter,
properties and structure of the polymer fibers,
which are different from those in textile fibers.
For co-mingling it is desirable that the polymeric and reinforcing fibers should have similar
diameter, which requires the polymer fibers to
be of lower diameter than they would be in
conventional textiles. It is also important that
the polymer fibers should not contract excessively or move around on melting so, unlike
textile fibers, generally they are processed under
conditions that minimize molecular orientation. Finally, fibers for co-mingling generally
contain polymer of lower molecular weight, in
order to promote flow and impregnation.

2.29.4.4

Solvent-based Processes

The purpose of solvent processing is to reduce

the viscosity of the thermoplastic in order to
promote flow. Solvent processes have been described by Turton and McAinsh (1974), Goodman and Loos (1990), and Savadori and Cutolo
(1993). Solvent-based technologies, such as
those employed by Ten Cate (1998), are gener-

14

Continuous Molding of Thermoplastic Composites

ally applicable to amorphous, rather than crystalline polymers, as the latter are not usually
easy to dissolve under normal conditions. Polymers that can be conveniently processed by
solution techniques include thermoplastic polyetherimide (PEI), polysulfone, polyether sulfone, and polyphenyl sulfone.
Solutions of thermoplastic polymer can be
used to impregnate either unidirectional tows
or woven fabrics. The most significant technical
problem is in handling the complete removal of
the solvent after wet-out of the prepreg, since
residual solvent can be deleterious to the properties of the composite.
When choosing composites based on amorphous resins, their potential susceptibility to
specific solvents needs to be borne in mind.
Nevertheless, such materials have been used in
a range of engineering applications (Ten Cate,
1998). The most successful are those based on
PEI, which has excellent fire smoke and toxicity
characteristics. Applications of the Ten Cate
Cetex range of PEI-based composites include
aircraft interiors and external fuselage ice impact protection.
An alternative to the solvent technique is the
use of a plasticizer, which may either be allowed
to remain in the structure after processing, with
some detriment to properties, or be removed.
One such technique, appropriate to PEEK or
PPS composites has been described (Cogswell,
1992; Cogswell and Staniland, 1985; Cogswell
and Measuria, 1988) where it was claimed to be
beneficial for the impregnation of carbon fiber
fibers with minimum damage to the fibers.

2.29.4.5

Reactive Techniques

An alternative method of avoiding the difficulties associated with high viscosity is to process the matrix resin in a condition where the
molecular weight is low, and to follow the
impregnation step with a polymerization or
chain extension reaction. In situ polymerization
has been considered in the case of monomers
such as methyl methacrylate, lactams, and systems leading to polyketones (Cogswell, 1992).
Recently, an improved system has been developed for the anionic ring-opening polymerization of caprolactam and laurolactam, the
precursors for nylons 6 and 12 respectively
(EMS-Chemie, 1999). This has led to interest
in in situ polymerized polyamide 12 composites
(O'Mairtin et al., 1999; Luisier et al., 2000). In
addition to the use of chemical initiators and/or
heat to promote polymerization, light and
UV-initiated polymerization have also been
considered.

In certain cases it has been found beneficial

to carry out the impregnation with a low molecular weight thermoplastic, then increase the
molecular weight by chain extension. Some
thermoplastics, e.g., PPS and PEEK, increase
their molecular weight at high temperatures
and this may be used to advantage, although
chain extension may sometimes be accompanied by less desirable cross-linking reactions. In
the case of PEEK (Ward et al., 1987; Cogswell,
1992), it has been found possible to impregnate
specially-sized carbon fiber with a low viscosity
version of the resin using a process similar to
the pin impregnation process described in
Figure 8. The molecular weight is then substantially increased by the action of a reagent previously applied to the fibers. This reagent is
claimed to promote chain extension initially in
the important region close to the resinfiber
interface. This sequence of melt impregnation
and chain extension results in a thermoplastic
composite with unique properties, due to the
combination of excellent impregnation, high
molecular weight resin and a good fibermatrix
interface. The successful development and exploitation by ICI, and later Fiberite, of the
technology for the manufacture of aromatic
polymer composite (APC), as the carbon/
PEEK system is referred to, is one of the most
notable achievements to date in the area of
thermoplastic matrix composites.

2.29.5

CONTINUOUS PROCESSES FOR

THERMOPLASTIC COMPOSITES

The processing routes available for thermoplastic matrix composites have been summarized in Figure 1. This section will deal
specifically with those techniques that involve
continuous or near continuous movement of
the reinforcement through the process, namely:
(i) thermoplastic pultrusion,
(ii) thermoplastic filament winding (including tape laying), and
(iii) helical tape winding.
The other cyclic processes are dealt with in
Chapter 2.28, this volume.

2.29.5.1

Thermoplastic Pultrusion

Thermoplastic matrix composites offer some

potential advantages over their thermosetting
counterparts for the production of sectional
products by pultrusion. Perhaps the greatest
benefit is the elimination of the cure reaction,
which appears to offer the prospect of much
faster speeds. A number of studies have taken

Continuous Processes for Thermoplastic Composites

15

Figure 15 Schematic of the thermoplastic pultrusion process.

place recently with the aim of making thermoplastic pultrusion a viable commercial technique (Hawley, 1982; Larock et al., 1989;
Astrom and Pipes, 1991; Devlin et al., 1992;
Michaeli and Jurss, 1996; Miller et al., 1998).
One option is to conduct he process in-line with
one of the impregnation processes described
previously (melt impregnation, powder impregnation or fiber co-mingling). Alternatively, it
may be operated with a pre-impregnated precursor material.
The pultrusion process, shown in Figure 15,
has the following stages:
(i) in-feed,
(ii) melting of the resin in the precursor
material,
(iii) consolidation,
(iv) forming, and
(v) cooling.
Careful control of the spatial arrangement
and tension of the in-feed is essential for a
good quality product with regular spatial distribution of the reinforcement. The melting
stage should make allowance for the low thermal diffusivity of the in-feed materials. To
achieve melting throughout the section of the
in-coming product it is necessary to separate all
the tows or tapes of material prior to heating, to
maximize the area available for heat transfer.
For line speeds above about 1 m min71, infrared (IR) heating, or a mixture of this and hot air
is desirable. IR allows a high rate of energy
transfer to be achieved, and the heating effect
penetrates some distance into the material. A
frequently used design of IR oven for pultrusion preheating consists of upper and lower
shells which can be opened up in the case of a
line stoppage, to allow the energy input to the
product to be halted. This avoids the risk of
igniting the polymer.
In the case of powder impregnated or comingled precursors it is necessary to carry out

a consolidation operation to achieve wet-out.

This is best accomplished by the use of pins,
similar to those described for melt impregnation. Pulling the pretensioned tows over the
pins results in the necessary lateral pressure to
cause the powdered or co-mingled material to
consolidate and wet-out. This stage is not necessary if fully consolidated precursor is used.
Following wet-out, the strands must be
brought together carefully, usually in a gathering die, to form the shape of the product. Two
different process configurations have been advocated for the forming/cooling stages, as
shown in Figure 16. In the two-die process,
a heated consolidation die is followed by a
cooled shaping die. The convergence in the
hot die generates a substantial pressure, which
assists in consolidation of the product. The die
geometry may also result in a certain amount of
wipe-off of resin. While the two-die system has
been found to operate successfully at low
speeds its drawback is the very high pull-forces
that are developed due to the resin viscosity and
the relatively large surface area of the die. The
pull-force is almost proportional to velocity, so
the line speeds are often significantly lower than
those in conventional thermoset pultrusion
(about 1 m min71), and the high interfacial
stresses in the dies can lead to problems with
surface appearance.
In the second configuration in Figure 16(b),
the contact time between the product and the
forming devices is substantially reduced, resulting in much lower pulling forces and higher
speeds, up to 20 m min71. A short gathering
and wipe-off die is followed by a cooling bath,
containing a series of product consolidation
devices. The down-stream configuration resembles the forming dies sometimes used in the
extrusion of thermoplastic sections. The consolidation devices used may either be short,
water-cooled forming dies or, when the product

16

Continuous Molding of Thermoplastic Composites

Figure 16 Alternative die configuration for thermoplastic pultrusion (a) two die process and (b) tooling
configuration for rapid pultrusion (Miller et al., 1998).

Figure 17 Schematic of the tape-laying and consolidation process for thermoplastic matrix composites.

section permits it, rollers. With dies, it is desirable that they are split and spring-loaded, so
only moderate pressure is applied to the moving
product. The best surface finish is achieved with
low-die temperatures.

proposed for forming profiles such as top hat

sections, from thermoplastic prepreg sheet. The
process changes the shape, but not the thickness, of the section and requires a stand of
several rolls.

2.29.5.1.1

2.29.5.2

Reinforced extrusions and moldings

Coating thermoplastic pultrusions by means

of a cross-head die has been advocated to improve surface appearance. Likewise, the use of
pultrusions to strengthen thermoplastic injection moldings (Hawley, 1982) and thermoplastic profiles has been considered, but practical
difficulties are encountered in achieving sufficient fiber content, within the shape envelope
of the component, and in ensuring that the
prepreg is fully bonded to the thermoplastic
matrix.
2.29.5.1.2

Roll forming

A semicontinuous process (Cogswell, 1992),

similar to that used in metal working, was

Thermoplastic Tape Laying and

Filament Winding

Processes where the final shape of the part is

built up by the accurate placing of plies, as in
Figure 17, are known collectively as fiber placement techniques (see Chapter 2.31, this
volume). Techniques such as tape laying, by
numerically controlled multi-axis machines
are well-established in thermoset prepreg processing. Thermoplastic prepregs do not adhere
in the cold state, but fiber placement processes
can be very effective provided the plies are
bonded to one another as they are placed.
Processes which achieve consolidated parts via
this route, without the use of autoclaves or
further heat treatment, have considerable
potential, as they utilize the capability of the
material to be rapidly formed and solidified.

Continuous Processes for Thermoplastic Composites

With all consolidation processes of this type
(as with conventional welding of thermoplastics) the requirement is that the surfaces to be
joined should be heated to a temperature above
the melting point of the thermoplastic and held
together for sufficient time to enable a bond to
be established through molecular diffusion. The
practical requirements can be illustrated by
considering the situation where two components, initially at temperatures T1 and T2 are
brought together with the aim of achieving a
weld. The temperature of the interface between
the components, Ti, will be given approximately by
Ti = (T1 + T2)/2

(7)

Bond formation requires that the interfacial

temperature should exceed the melting point
of the polymer. In many thermoplastic placement processes the component being laid down
is heated to a temperature just a little above the
melting point. With the substrate at ambient
temperature it can be seen that the thermal
condition for bonding is not fulfilled. There
are three ways of addressing this problem:
(i) to raise the temperature of the material
being placed to a level much higher than the
melting point,
(ii) to increase the substrate temperature,
and
(iii) to add extra heat to the bond region.
All three methods have been examined. Solution (i) alone is generally insufficient to solve
the problem, since heating the polymer much
above its melting point will cause degradation,
unless the it is blanketed with inert gas to
prevent oxidation or unless the period at the
high temperature is very short. Most processes
rely on a combination of (i) and (ii) to solve the
problem, or on adding extra energy (iii) near to
the bonding point. Sources of heating that have
been investigated include hot gas, including
sometimes a flame, focused IR energy and
laser energy. In addition to achieving the necessary thermal conditions it is usually necessary
to apply a degree of local consolidation pressure to eliminate air and fulfill the condition for
molecular diffusion across the bond line. When
the process is operated under the optimum
thermal conditions it is possible to produce
complex structures with good interlaminar
properties and low thermally induced stresses.
High precision multiaxis machines for fiber
placement are now available from a number
of manufacturers, including Automated
Dynamics (1999).
Large structures may be built up using tapelaying, but difficulties have been found sometimes in achieving full consolidation. Processing

17

factors have been discussed in a number of

publications, including Grove (1988), Grove
and Short (1988), Beyeler and Gucieri (1988),
and Beyeler et al. (1988).
The manufacturing problems for thermoplastic filament winding are similar to those
encountered in tape laying and a range of solutions have been proposed. Some employ a combination of mandrel heating and hot air to
achieve the necessary thermal conditions,
whereas others use an unheated mandrel, the
additional energy being supplied by focused
radiation or hot gas. Some processes rely solely
on the tension of the fiber tow to apply consolidation pressure to the bond region, while
others lay down the in-coming tow with a roller
or hot shoe to apply additional consolidation
pressure.
As with pultrusion, thermoplastic filament
winding can be operated with precursor material from any of the sources described above.
Processes have been reported where the precursor is melt impregnated (Moyer, 1976; Cogswell
et al. 1981a, 1981b; Fahrer, 1997; Fisher and
Gibson, 1998), powder impregnated (Astrom
and Pipes, 1990; Wagner and Colton, 1994;
Romagna et al., 1995; Haupert and Friedrich,
1995) and co-mingled (Fedro et al., 1989; Elliott
et al., 1999). Recently, considerable success has
been achieved in high speed winding, at linear
speeds of 50m/min or more, with co-mingled
glass/PETG melted and wetted out in-line with
the process, making use of controlled winding
tension to achieve the necessary consolidation
force. This version of the process offers possibilities for the rapid manufacture of gas bottles
for a range of applications (Elliott et al., 1999).
One of the advantages of thermoplastic filament winding is that, once consolidated and
cooled, the material has immediate mechanical
strength, unlike its thermosetting counterpart,
which requires to be cured. This offers the
possibility of following severely nongeodesic
paths, of making shapes with re-entrant or
concave curvature and of removing the mandrel
altogetherall of which would be difficult or
impossible with thermosets. Thermoplastic filament winding now has the possibility of being
used in large scale and mass production environments, which will result in considerably enhanced use of thermoplastic matrix composites
in this type of application.
2.29.5.3

Reinforced Thermoplastic Pipe (RTP)

Pressurized pipework is an excellent loadbearing application for thermoplastic matrix

composites. Because the main loads in the
pipe wall are all tensile, the significance of

18

Continuous Molding of Thermoplastic Composites

Figure 18 Schematic of the tape-winding process used for RTP manufacture, showing thermoplastic liner,
layers of tape reinforcement, and extruded outer cover.

impregnation and bonding between the layers

within the structure is greatly reduced. Indeed
some lack of bonding may even be regarded as
advantageous, as it increases the axial flexibility
of the pipe. Continuous fiber reinforced thermoplastic pipes (RTPs) offer considerable performance
enhancement,
compared
to
unreinforced thermoplastic pipe, while retaining the advantages of light weight, ease of installation and corrosion resistance. As shown in
Figure 18, an RTP comprises an inner liner and
outer cover of thermoplastic (usually medium
density polyethylene), with an even number of
load-bearing +558 reinforcing plies, usually of
aramid fiber.
Because RTPs are required in long lengths, a
continuous winding process is preferred over
conventional filament winding. The most favorable technique at present is continuous helical winding of reinforcing tape, as shown in
Figure 18. The tape comprises yarns of reinforcing fiber, encapsulated in a polymeric matrix.
The advantage of this configuration is that the
tape acts as a carrier, permitting many reinforcing yarns to be applied simultaneously, without the need for a large number of rotating
bobbins on the winding assembly. Depending
on the manufacturer and the application the
reinforcement may or may not be bonded to the
liner and covers.
At present the most attractive reinforcement
for RTP is aramid fiber, the reason being that,
unlike glass or carbon, aramid can perform
satisfactorily in tension in the unwetted state.
In contrast to the other materials described in
this chapter the yarns in the reinforcing tape
generally are not impregnated by the thermoplastic resin. Use of nonimpregnated fibers in
this case has certain advantages, including flexibility. Lengths of several hundred metres of
RTP may continuously manufactured and
coiled for transport.

On the basis of strength per unit cost, nonimpregnated aramid/PE tape is less expensive
than impregnated glass/PE. Glass fiber reinforcement would be competitive in this application, however, if impregnation could be carried
out at the same time as the winding operation,
and a process has been developed to do this
(Fisher and Gibson, 1998).
At present the main applications of RTP are
in high pressure transport of natural gas and
oilfield fluids (Frost, 1999). In the medium
term, however, RTPs have a wide range of
other potential applications and the future appears excellent for processes to manufacture
this type of product.

2.29.6

REFERENCES

A. M. Adur, R. C. Constable and J. A. Humenik, in `SPI/

Composites Institute 44th Annual Conference, Paper
11-A', Dallas, TX, February 69, 1989, Society of the
Plastics Industry Inc., NY, 1989.
K. J. Ahn, J. C. Seferis and J. C. Berg, Polymer Composites, 1991, 12, 146152.
B. T. Astrom, Ph.D. Thesis, Royal Institute of Technology, Stockholm, Sweden, 1992.
B. T. Astrom and R. B. Pipes, J. Thermoplastic Comp.
Mat., 1990, 3, 314324.
B. T. Astrom and R. B. Pipes, SAMPE Q., 1991, 22(4),
5565.
B. T. Astrom, R. B. Pipes and S. G. Advani, J. Comp.
Mat., 1992, 26(9), 13511373.
Automated Dynamics, Product Literature on Fiber Placement Technologies, Schenectady, NY, 1999.
M. G. Bader and W. H. Bowyer, Composites, 1973, 4,
150156.
P. J. Bates and J. M. Charrier, J. Thermoplastic Comp.
Mat., 1999, 12, 276296.
R. M. Baucom and J. M. Marchello, J. Advanced Mat.,
1994, 25(4), 3142.
N. Bernet, V. Michaud, P-E. Bourban and J.-A. E.
Manson, J. Comp. Mat., 1999, 33(8), 751772.
E. P. Beyeler and S. I. Gucieri, J. Heat Transfer, 1988,
110, 424430.
E. P. Beyeler, W. Phillips and S. I. Gucieri, J. Thermoplastic Composites, 1988, 1, 107121.

References
H. Bijsterbosch and R. J. Gaymans, Composites Manufacturing, 1993a, 4(2), 8591.
H. Bijsterbosch and R. J. Gaymans, Composites Manufacturing, 1993b, 4(3), 133137.
R. Bradt, US Pat. 2 877 501 (1959).
R. Bradt, US Pat. 3 042 570 (1962).
T. A. Cain, M. D. Wakeman, R. Brooks, A. C. Long and
C. D. Rudd, in `Proceedings of the ICCM-11', Gold
Coast, Queensland, Australia, Woodhead Publishing
Ltd., Cambridge, UK, 1997.
G. Chabrier, G. Moine, R. Maurion and R. Szabo, US
Pat. 4 626 306 (1986).
H. W. Chandler, B. J. Devlin and A. G. Gibson, Plastics,
Rubber and Composites Processing and Applications,
1992, 18(4), 215220.
I. Y. Chang and J. K. Lees, J. Thermoplastic Comp. Mat.,
1998, 1, 277.
B. J. Chapman, Ph.D. Thesis, University of Newcastle
upon Tyne, UK, 1999.
F. N. Cogswell, `Thermoplastic Aromatic Polymer Composites: A Study of Structure, Processing and Properties of Carbon Fibre Reinforced PEEK and Related
Materials', Butterworth-Heinemann, Oxford, UK,
1992.
F. N. Cogswell, D. J. Hezzell and P. J. Williams, US Pat.
4 549 920 (1981a).
F. N. Cogswell, D. J. Hezzell and P. J. Williams, US Pat.
4 559 262 (1981b).
F. N. Cogswell and U. Measuria, US Pat. 4 624 886
(1988).
F. N. Cogswell and P. A. Staniland, US Pat. 4 541 884
(1985).
R. J. Coldicott, T. Longdon, S. Green and P. J. Ives, in
`34th International SAMPE Symposium', eds. G. A.
Zakrzewski, D. Mazenko, S. T. Peters and C. D. Dean,
SAMPE, Covina, CA, 1989.
J. Colton and D. Leach, in `49th Annual Technical
Conference', ANTEC, Society of Plastics Engineers,
Brookfield Center, CT, 1991, pp. 20582061.
M. Connor, Ph.D. Thesis, EPFL Lausanne, Switzerland,
1995.
M. Connor, S. Toll, J.-A. E. Manson and A. G. Gibson,
J. Thermoplastic Composite Materials, 1995, 8, 138162.
R. C. Constable and A. M. Adur, ANTEC, 1991, 1892
1896.
R. C. Constable, J. A. Humenik and A. M. Adur,
ANTEC, 1989, 13891393.
B. J. Devlin, M. D. Williams, J. A. Quinn and A. G.
Gibson, Composites Manufacturing, 1992, 2(34), 203
209.
J. Elliott, G. Bishop and J. Francomb, in `ICAC 1999, 6th
Int. Conference on Automated Composites', Bristol, UK,
September 2324, Institute of Materials, London, 1999,
pp. 237244.
EMS-Chemie AG, CH-7013 Domat/Ems, Switzerland,
Product data, 1999.
K. E. Evans and A. G. Gibson, Comp. Sci. Tech., 1986,
25, 149162.
A. Fahrer, Ph.D. Thesis, University of Newcastle upon
Tyne, UK, 1997.
M. J. Fedro, D. W. Jensen and H. T. Hahn, in `Proceedings of Fabricating Composites 1989', Society of Manufacturing Engineers, 1989, pp. 118.
E. H. Fisher and A. G. Gibson, Plastics, Rubber and
Composites, 1998, 27(10), 447451.
B. Franzen, C. Klason, J. Kubat and T. Kitano, Composites, 1989, 20(1), 6576.
S. R. Frost, in `Composite Materials for Offshore Operations-2', eds. S. S. Wang, J. G. Williams and K. H. Lo,
American Bureau of Shipping, Houston, TX, 1999, pp.
341360.
R. A. Ganga, US Pat. 4 614 678 (1986).
B. R. Gebart, J. Comp. Mat., 1992, 26, 11001133.

19

A. G. Gibson, in `Polypropylene: Structure, Blends and

Composites', ed. J. Karger-Kocsis, Chapman and Hall,
London, 1994, chap., 24.
A. G. Gibson and J.-A. E. Manson, Composites Manufacturing, 1992, 3(4), 223233.
K. E. Goodman and A. C. Loos, J. Thermoplastic Comp.
Mat., 1990, 3, 3440.
S. M. Grove, Composites, 1988, 19, 367375.
S. M. Grove and D. Short, Plastics, Rubber and Composites Processing and Applications, 1988, 10(1), 3544.
T. G. Gutowski, Z. Cai, S. Bauer, D. Boucher, J. Kingery
and S. Wineman, J. Composite Materials, 1987, 21,
650669.
H. Hamada, Z. Maekawa, N. Ikegawa and T. Matsuo, J.
Polymer Composites, 1993, 14(4), 308.
T. Hartness, J. Thermoplastic Comp. Mat., 1988, 1, 210
220.
K. Hattori, E. L. Lowery, W. H. Chadbourne and W. M.
Boyer, US Pat. 4 037 011 (1977).
F. Haupert and K. Friedrich, Composites Manufacturing,
1995, 6(34), 201204.
R. C. Hawley, US Pat. 4 312 917 (1982).
M. K. Hugh, J. M. Marchello, R. M. Baucom and N. J.
Johnston, in `International SAMPE Symposium and
Exhibition', eds. G. C. Grimes, R. Turpin, G. Forsberg,
B. M. Rasmussen and J. Whithey, SAMPE, Covina, CA,
1992, vol. 37, pp. 10401051.
P. J. Ives and D. J. Williams, in `33rd International
SAMPE Symposium', eds. G. Carrillo, E. D. Newell,
W. D. Brown and P. Phelan, SAMPE, Covina, CA,
1988, pp. 858869
S. R. Iyer and L. T. Drzal, J. Thermoplastic Comp. Mat.,
1990, 3, 325355.
G. de Jager, Int. Pat. WO 87 00 563 (1987).
C. St. John, Proc. ICCM-10, 1995, 3, 757764.
Y. R. Kim, S. P. McCarthy and J. P. Fanucci, Polymer
Composites, 1991, 12(1), 1319.
V. Klinkmuller, M. K. Um, M. Steffens, K. Friedrich and
B. S. Kim, Applied Comp. Mat., 1995, 1, 351371.
J. A. Larock, H. T. Hahn and D. J. Evans, in `44th
Annual Conference', Society of the Plastics Industry,
New York, 1989
D. C. Leach, in `Advanced Composites', ed. I. K. Partridge, Elsevier Applied Science, London, 1989, chap. 2,
pp. 43111.
W. I. Lee and G. S. Springer, J. Comp. Mat., 1987, 21,
10171055.
A. Luisier, P.-E. Bourban and J.-A. E. Manson, J. Appl.
Polym. Sci., in press, 2000.
C. S. Mantell and G. S. Springer, J. Comp. Mat., 1992,
26(16), 23482360.
P. McDonnell, K. P. McGarvey, L. Rochford and C. M.
O'Bradaigh, in `ICAC 1999, 6th International Conference on Automated Composites', Bristol, UK, September 2324, Institute of Materials, London, 1999, pp. 63
71.
M. Miao, Y. L. How and K. P. S. Cheng, Textile Res. J.,
1994, 64(10), 563.
W. Michaeli and D. Jurss, Composites Part A, 1996,
27A(1), 37.
J. V. Milewski, Composites, 1973, Nov., 258265.
J. V. Milewski, `Handbook of Filler and Reinforcements
for Plastics', Van Nostrand Reinhold, London, 1978.
A. H. Miller, N. Dodds, J. M. Hale and A. G. Gibson,
Composites A, 1998, 29A, 773782.
A. H. Miller and A. G. Gibson, Polymers and Polymer
Composites, 1996, 4(7), 459481.
R. L. Moyer, US Pat. 3 933 726 (1976).
J. D. Muzzy and J. Colton, US Pat. 5 409 757 (1995).
J. D. Muzzy, B. Varughese, V. Thammongkol and W.
Tincher, SAMPE J., 1989, 25(5), 1521.
J. D. Muzzy, B. Varughese and P. H. Yang, SPE
ANTEC, 1990, 13851389.

20

Continuous Molding of Thermoplastic Composites

A. L. Ogden, M. W. Hyer, G. L. Wilkes and A. C. Loos,

J. Thermoplastic Comp. Mat., 1992, 5, 1424.
P. O'Mairtin, P. McDonnell, R. Eider and C. M. O'Bradaigh, in `ICAC 1999, 6th International Conference on
Automated Composites', Bristol, UK, September 23
24, Institute of Materials, London, 1999, pp. 4552.
P. Peltonen, K. Lahteenkorva, E. J. Paakonen, P. K.
Jarvela and P. Tormala, J. Thermoplastic Comp. Mat.,
1992, 5, 318328.
R. V. Price, US Pat. 3 742 106 (1973).
K. Ramani, M. Tryfondis and J. Gentry, in `International
SAMPE Technical Conference', eds. T. S. Reinhart, M.
Rosenow, R. A. Cull, E. Struckholt, F. H. Froes and
W. Wallace, SAMPE, Covina, CA, 1992, pp. 128142.
H. A. Rijsdijk, M. Contant and A. A. J. M. Peijs, in
`Proceedings of ICCM-9', Madrid, Spain, Woodhead
Publishing, Cambridge, UK, 1993, pp. 251258.
J. Romagna, G. Ziegmann and M. A. Flemming, Composites Manufacturing, 1995, 6(34), 205210.
A. Savadori and D. Cutolo, Macromol. Chem. Macromol.
Symp., 1993, 68, 109131.
N. Svensson, R. Shishoo and M. Gilchrist, J. Thermoplastic Comp. Mat., 1998, 11, 121133.
Ten Cate, Cetex Product Literature, Ten Cate Advanced
Composites, Fountain Valley, CA, 1999.
A. Texier, R. M. Davis, K. R. Lyon, A. Gugnor, J. E.
McGrath, H. Marand and J. S. Riffle, Polymer J.,

1993, 34(4), 896905.

S. Toll and J.-A. E. Manson, J. Applied Mechanics, 1995,
62, 223226.
Trevira Neckelmann, Product Literature, 1999, DK-8600,
Silkeborg, Denmark.
R. von Turkovich and L. Erwin, Polymer Engineering and
Science, 1983, 23(13), 743749.
N. Turton and J. McAinsh, US Pat. 3 785 916 (1974).
A. M. Vodermayer, J. C. Kaerger and G. Hinrichsen,
Composites Manufacturing, 1993, 4(3), 123.
P. Wagner and J. Colton, J. Polymer Comp., 1994, 15(6),
436441.
B. P. van West, R. B. Pipes and S. G. Advani, J. Polymer
Comp., 1991, 12(6), 417427.
B. P. van West, R. B. Pipes, M. Keefe and S. G. Advani,
Composites Manufacturing, 1991, 2(1), 1021.
M. V. Ward, E. Neild and P. A. Staniland, Eur. Pat.
0 125 816 B1 (1987).
C. E. Wilks, C. D. Rudd, A. C. Long and C. F. Johnson,
in `ICAC 1999, 6th International Conference on Automated Composites', Bristol, UK, September 2324,
Institute of Materials, London, 1999, pp. 179186.
J. G. Williams, C. E. M. Morris and B. C. Ennis, Polymer
Engineering Science, 1974, 14(6), 413419.
L. Ye, V. Klinkmuller and K. Friedrich, J. Thermoplastic
Comp. Mat., 1992, 5, 3248.

Copyright # 2000 Elsevier Science Ltd.

All rights reserved. No part of this publication may be reproduced, stored in any
retrieval system or transmitted in any form or by any means: electronic,
electrostatic, magnetic tape, mechanical, photocopying, recording or otherwise,
without permission in writing from the publishers.

Comprehensive Composite Materials

ISBN (set): 0-08 0429939
Volume 2; (ISBN: 0-080437206); pp. 979998