CFD Modeling of the Closed Injection Wet-Out Process For Pultrusion

CFD Modeling of the Closed Injection Wet-Out
Process For Pultrusion
Authors:
Mark Brennan, Huntsman Polyurethanes, Everslaan 45, Everberg 3078, Belgium, and
Michael Connolly & Trent Shidaker, Huntsman Polyurethanes, 2190 Executive Hills
Boulevard, Auburn Hills, MI, USA 48326
Michael Connolly
Product Manager-Urethane Composites
Huntsman Polyurethanes
2190 Executive Hills Boulevard
Auburn Hills, Michigan USA 48326
Phone: +1-248-322-7300
Fax: +1-248-322-7303
Email: michael_connolly@huntsman.com
About the Authors:
Michael Connolly, Ph. D. is Product Manager-Urethane Composites for Huntsman
Polyurethanes in Auburn Hills, Michigan, USA. Michael has been employed in
polyurethane product development for automotive, consumer and industrial
applications for 12 years at Huntsman and has 19 years experience in a range of
polymer materials development roles. He received his doctoral degree in Polymer
Science and Engineering from the University of Massachusetts-Amherst in 1989.
Mark Brennan is PU Expert in Computational Physics for Huntsman Polyurethanes in
Everberg, Belgium. Mark has been employed in process and product research for 7
years at Huntsman and has 12 years experience in applying Computational Fluid
Dynamics and Finite Element Analysis to solve industrial applications. He received
his master s degree in Mathematics from University of Dublin, Trinity College in
1996.
Trent Shidaker is Development Manager for Polyurethane Composites and Rigid
Materials at Huntsman Polyurethanes in Auburn Hills, MI. Trent has been employed
as manager or project leader in polyurethane foam and composite development at
Huntsman for 12 years.
Abstract:
The practice of closed injection wet-out in pultrusion has become increasingly utilized
over the last five years. This paper presents a mathematical model for simple
geometries of the closed injection wet-out process used in pultrusion and represents
an initial step to understanding wetting phenomena in complex shapes with complex
reinforcement. The moving porous model used is both a generalization of the Navier-
Stokes equations and Darcy s law used for flows in porous media. The model can
predict pressure build-up, flow patterns and composite void content and evaluate the
influence of process conditions, injection box design, and material properties on these
parameters.

INTRODUCTION
Over the past decade, the global
pultrusion industry has increasingly
adopted closed injection wet-out for
profile processing. This trend has been
driven by awareness of worker safety
and mandates by regulatory agencies to
reduce volatile organic compounds
(VOC s) and by the availability of new
resin technology. Polyurethane (PU)
resins, in particular, have become an
accepted alternate technology which
contain no VOC s, especially for
applications requiring high strength
and durability combined with the
potential for high pultrusion line
speeds (Ref. 1-3 and references
therein). However, the relatively fast
reactivity of PU resins requires
processing via closed injection wetout.
The complex profile geometries and
reinforcement schedules used in stateof-the-art
profiles make efficient wetout
challenging in a constrained state
within an injection box. This
challenge is even greater for PU resins,
considering their inherently fast
reactivity. Hence, a more thorough
understanding of wet-out dynamics in
injection processing for pultrusion is
needed to expand acceptance of this
process for PU resins and accelerate its
implementation.
This report describes a mathematical
model for closed injection wet-out in
the pultrusion process. The model
predicts the impact of injection box
geometry, process conditions, resin
viscosity, reinforcement fraction and
permeability on flow patterns, pressure
profiles and final part density.
EXPERIMENTAL
CFD simulations were carried on an
injection box for a die profile
dimension of 5.21cm x 0.648 cm
(2.05 x 0.255 ). The line speeds were
varied between 0.6 and 1.5 m/min (24
to 60 inch/min) with injection box
lengths from 30.5 to 53.3 cm (12 to
21 ). Resin viscosity was varied
between 600 and 2400 cP.
Experimental pultrusion runs were
made to verify the CFD results.
RIMLine ® SK 97007 polyol and
Suprasec ®
9700 isocyanate and
experimental variants were combined
as the polyurethane matrix. Profiles
were pultruded with 120 rovings for a
total glass content of 77 wt% (60 vol
%). A two-component metering unit
(Electric Willie from GS
Manufacturing) with gear pump flow
control was used to ensure delivery of
the polyurethane resin at constant
pressure. Injection boxes were
fabricated using aluminum with HDPE
guide plates having small holes to
prevent resin leakage. Digital pressure
transducers or pressure gauges were
placed at ~2.5 cm (1 ) and ~10 cm (4 )
from the die face as well as the
injection port. The typical injection
box setup is shown in Figure 1.
MATHEMATICAL MODEL
Due to difficulty in modeling
macroscopic transport of resin and
reinforcement based on microscopic
variation of velocity between fibers,
the resin and reinforcement are better
treated as a moving porous fiber
medium. The moving porous model
used is both a generalization of the
Navier-Stokes equations and Darcy s
law used for flows in porous media.
The balance equations for mass and
momentum for steady incompressible
flow are
u
0
2
u u p u S
u, p, , , and S are the superficial
velocity, pressure, density, viscosity
and momentum sources, respectively.

The momentum sources, S, defining
the moving porous medium are
S u U
K
K, and U are the permeability tensor,
porosity and velocity of the fiber
bundle. The geometry of the injection
box is shown in Figure 2. This
approach is similar to the model
demonstrated by Jeswani et al. [Ref. 4]
The approach is a phenomenological
one with the main assumption being
that the surface tension effects are not
important or are captured in the
permeability tensor. The
reinforcements are initially dry and the
resin must penetrate the fiber bundles.
Capillary forces of attraction and
repulsion act at the flow front. These
forces, which depend on the surface
tension of the resin and on its ability to
adhere to the surface of the fibers, have
an effect of increasing or decreasing
the effective pressure at the resin front.
However, they are assumed to be
sufficiently small to be neglected by
the model. This assumption can be
checked with micro-scale modeling of
air-resin interface flow in fiber
bundles.
A further assumption in the model is
that degree of conversion and hence
build-up of viscosity is small and,
hence, can be ignored in the injection
box region. It is also assumed that
there is no shear rate dependence of
viscosity. These complexities can be
added to the model without great
difficulty.
Porosity and Fiber Volume Fraction
The porosity of the fiber bundle varies
with axial distance in the model
1 Vf ( z)
hence, the stream wise and transverse
permeability throughout the injection
is not constant. An example of porosity
variation throughout the injection box
is shown in Figure 3.
Permeability Model
The Gebart permeability model [Ref.
5] with a hexagonal fiber arrangement
was chosen to model the permeability
in both the axial and transverse fiber
directions. The model is particularly
suited to pultrusion processes as the
fibers are unidirectional. In the axial
direction, the permeability is defined
by
K
z
2D
2 3
f
2
c
(1 )
where Df is the effective fiber diameter
and c a model constant. The
permeability in the transverse direction
is
K K
x y
C D
V
1
2
f f ,max
1
4 (1 )
where Vf,max is the maximum volume
fraction of fibers and C1 a model
constant. Profiles of stream wise and
transverse permeability are shown in
Figure 4.
Boundary Conditions
Where the fiber reinforcement enters
the injection box, the pressure is set to
atmospheric pressure and a positive
pressure is set at the injection port. At
the exit of the computational domain, a
small distance into the die, the axial
velocity is set to the porosity times the
pull speed.
5
2

Flow Solver
The mathematical model was solved
by the flow solver ANSYS-CFX 11.0
with additional source terms added to
account for the moving porous
medium.
RESULTS AND DISCUSSION
Flow Patterns in the Injection Box
Streamlines, created by releasing massless
particles from the feed ports, show
how the resin impregnates the fiber
bundle in Figure 5. Resin injected
through the feed ports spreads over the
top of the bundle, then impregnates
into it and finally is pulled through the
injection box, achieving good wet-out.
The streamlines are colored to indicate
the pressure gradient.
Pressure Profiles in the Injection Box
Figure 6 shows the centerline pressure
profiles in the injection box for
different line speeds. As expected the
final pressure rise increases with line
speed. For example, at a line speed of
0.6 m/min, the pressure rise at the die
face is 23 bar, rising to 57 bar at a line
speed of 1.5 m/min. The experimental
results (blue squares in Fig. 6) compare
favorably with the CFD results at ~10
cm from the die face. However, the
injection box pressure measured at
~2.5 cm from the die face is much
higher than predicted. The model
assumes a uniform reinforcement
matrix and does not take into account
factors such as roving twist or
resin/fiber interactions which would
likely be magnified in the constrained
state near the die face. Further
experimental work is in progress to
verify and expand on these results.
Figure 7 shows the influence of the
viscosity of the resin, glass fraction,
box length and box height (i.e., the
effective taper angle of the box) on the
final pressure rise in the injection box.
As expected for a fully wet-out part,
this relationship is linear for viscosity
and box length. While the physics of
the pressure rise are not completely
understood, the model predicts nonlinear
relationships for box height and
glass fraction. The model can also look
at the influence of other geometrical
factors such as injection location(s)
and other operating parameters
including injection pressure on the
pressure profiles in the box.
Fiber Wet-Out in the Injection Box
Figure 8 shows the resin flow-front
development in the injection box. To
achieve good wet-out, both flow fronts
should reach the center of the fiber
bundle but also spread to all regions of
the injection box. In Figure 8, the resin
is colored to indicate residence time in
the box. It can be observed in this case
that the pultruded part is completely
wet-out by the time it enters the die.
The model can be used to illustrate
process conditions under which profile
wet-out becomes an issue. It can also
be used to evaluate the influence of
injection box geometry and operating
parameters such as injection pressure
on wetting efficiency.

CONCLUSIONS
A mathematical model of a pultrusion
injection box has been created. A
permeability model for unidirectional
reinforcements based on porosity and
direction has been used. The model
can predict injection box pressure
profiles and flow patterns giving an
estimation of reinforcement wet-out.
The model has proven to be versatile
and can evaluate a number of factors
influencing process performance
including injection box geometry, the
number and location of the injection
ports, injection pressure, line speed,
resin physical properties, fiber
permeability and fiber fraction on
pressure rise in the injection box.
The model predicts the appropriate
trends when compared with
experimental data and experience.
Further experimental work is planned
to verify additional geometric and
process variations. With the use of this
model, the typical empirical approach
taken with injection box development
can be reduced or eliminated. Hence,
the development time line can be
shortened and adoption of closed
injection wet-out and PU resin systems
may be accelerated in the pultrusion
industry.
All information contained herein
is provided "as is" without any
warranties, express or implied, and
under no circumstances shall the
authors or Huntsman be liable for any
damages of any nature whatsoever
resulting from the use or reliance upon
such information. Nothing contain in
this publication should be construed as
a license under any intellectual
property right of any entity, or as a
suggestion, recommendation, or
authorization to take any action that
would infringe any patent. The term
"Huntsman" is used herein for
convenience only, and refers to
Huntsman Corporation, its direct and
indirect affiliates, and their employees,
officers, and directors.
REFERENCES
1. Connolly, M., King, J.P., Shidaker,
T.A., Duncan, A.C., Pultruding
Polyurethane Composite Profiles:
Practical Guidelines for Injection
Box Design, Component Metering
Equipment and Processing,
Proceedings of Composites 2005,
American Composites
Manufacturers Association, Sept.
2005.
2. Connolly, M., King, J.P., Shidaker,
T.A., Duncan, A.C., Processing
and Characterization of Pultruded
Polyurethane Composites,
Proceedings of the 8 th World
Pultrusion Conference, European
Pultruders Technical Association,
March 2006.
3. Connolly, M., King, J.P., Shidaker,
T.A., Duncan, A.C.,
Characterization of Pultruded
Polyurethane Composites:
Environmental Exposure and
Component Assembly Testing,
Proceedings of Composites 2006,
American Composites
Manufacturers Association, Oct.
2006.
4. Jeswani, A.L., Roux, J.A.,
Vaughan, J.G., Impact of Axial
Location of the Injection Slot for
High Pull Speed Resin Injection
Pultrusion, Proceedings of the 1 st
Global Pultrusion Conference,
Baltimore, MD, June 2006.
5. Gebart, B.R., Permeability of
Unidirectional Reinforcements for
RTM, Journal of Composite
Materials, 26: 1100-1133 (1992).

Mixhead
Static Mixers
Pressure
Guage at
Tee Inlet
Digital
Pressure
Transducer
Fiberglass Rovings Mounting Bracket
Figure 1: Typical Injection Box Setup
Figure 2: CFD model of injection box

Figure 3: Porosity variation of the fiber bundle in the injection box
Figure 4: Stream wise and transverse permeability profiles in the injection box

Figure 5: Flow streamlines in the injection box colored with pressure for a line
speed of 1.2 m/min
Figure 6: Centerline pressure profiles for 45.7 cm (18 ) injection box at different
line speeds with a viscosity of 1200 cP

Figure 7: Influence of viscosity, glass fraction, box length and box height on final
pressure rise in the injection box
Figure 8: Development of resin flow front indicating fiber wet-out