Modeling of particle–resin suspension impregnation in compression resin transfer

molding of particle-filled, continuous fiber reinforced composites

Snehomoy Ghosh

Department of Mechanical Engineering, National Institute of Technology Rourkela, Odisha –

769008, India

Email address: 120me1117@nitrkl.ac.in; snehmay27@gmail.com

1) Introduction

Resin transfer molding (RTM) is a liquid molding technique for manufacturing continuous
fiber-reinforced advanced composites. The process offers near net-shape fabrication of
complex-shaped products. High fiber loadings can be imparted, yielding strong mechanical
behavior with a high strength-to-weight ratio due to light-weight composite constituents.
processing method for critical engineering applications, especially in aerospace and
automotive industries [1, 2]. In RTM, the composite product is formed by injection of a
thermosetting resin into a mold cavity that has preplaced layers of fiber fabric or a fiber
preform. The resin impregnates the fiber reinforcement and is subsequently cured to give the
solid shape to the composite. Properties such as improved toughness, modified thermal
properties, electromagnetic shielding and/or self-healing properties can be achieved [3] in the
composite with the inclusion of fillers. During production, the filler particles are mixed with
the resin and are introduced to the composite through the impregnation of the fibers by the
filler-resin suspension. Certain production issues may arise during the process such as
ineffective wetting of the fiber preform, higher injection pressures due to increased fluid
viscosity, and possible particle filtration especially for large particle size or agglomeration,
and high fiber volume fraction preforms (for high-performance composites). Compression
resin transfer molding (CRTM) is a variation of the RTM process to manufacture medium-
tolarge sized composites with high fiber volume fractions. In this process, the resin is injected
into preplaced layers of fiber fabric or a fiber preform, while the mold is partially closed and
the preform is not fully compressed to its final dimensions. Only some portion of the preform
is impregnated during the injection phase. Since the fiber volume fraction is smaller, the
resistance to flow and the flow pressures (for constant injection rate) are smaller during the
injection stage compared to RTM process [1]. The injection phase is followed by the
compression phase in which the mold closes and forces the injected resin to impregnate the
rest of the fiber preform as the preform thickness is brought to its design value (the design
fiber volume fraction is achieved). In RTM, due to higher fiber volume fractions throughout
the impregnation and/or the need to reduce the fill time for large composite molding, the
processing pressures may be quite high. This may lead to deformation of the mold/preform or
fiber wash-out, all of which affect the part quality adversely. As such, in CRTM, better fiber-
resin wetting, lower processing pressures and larger structural composite production are
aimed, compared to RTM [4, 5]. These features render the process desirable not only for neat
resin-fiber composites but also for the production of composites with particle fillers. Figure 1
presents the schematic of the CRTM process for particle-filled composite production. In this
process, the higher permeability of the fibrous medium during injection may help keep the
processing pressures low. In addition, the filler particle distribution may be more uniform
compared to the same composite produced via RTM. However, these issues have not been
investigated and the current work aims to analyze the CRTM process with the inclusion of
particle fillers using a particle–resin impregnation model. In composites processing, the use
of flow models are important in designing the mold and/or determining the appropriate
process parameters for producing a composite part with the desired properties. The more
conventional trial-and-error type process design approaches are costly and time consuming as
molds and composite material components are expensive, especially in high-performance
composite applications. Impregnation modeling tools, when coupled with experiments that
yield basic process parameter values such as preform permeability which are used as inputs in
the simulations, can present the possible outcomes of alternative processing scenarios for the
designer to choose before any manufacturing begins. In this work, the injection and the
compression stages of the CRTM process are modeled by coupling flow through porous
media with particle filtration in a continuously evolving flow domain with a free surface.
Flow modeling in RTM process has been well studied [2, 6–8]. The particle-filled resin
impregnation model with filtration was previously developed by the author [9] for RTM
process and variants of this model have been presented [10, 11]. Although CRTM
impregnation modeling work exists in the literature [1, 4, 5, 12, 13], resin impregnation in the
presence of fillers with particle filtration has not been applied to CRTM before. The flow
model is based on Darcy flow for macroscopic flow of resin through the fibrous medium with
neglected inertial effects, suitable for modeling resin flow in liquid molding of composites.
The flow model is coupled with a filtration kinetics equation to obtain particle distribution
during impregnation. Mass conservation equations complete the model. The resin flow is
assumed isothermal which is typical in many molding studies, as the majority of the cure
occurs after impregnation is complete [2, 6].

2) Formulation of the problem and numerical implementation

The resin impregnation in CRTM process occurs over two stages: Injection and compression.
The mold is partially closed during injection and the suspension flows directly into the fiber
preform. The modeling of the injection stage of CRTM is similar to the modeling of resin
impregnation in RTM process [9] whereas for the compression stage, the change in preform
porosity due to the closing of the mold and the subsequent deformation of the flow domain
are taken into account. The conservation equations and the constitutive relations are presented
for both stages, with the necessary boundary and initial conditions.

Fig. 1 Injection and compression phases of the particle-filled CRTM Process

2.1 Injection of particle-filled resin into uncompressed fibrous medium

The mass conservation for the particle-filled, incompressible resin flow through a fibrous medium is
presented for three different species: the particles suspended in the resin, the neat resin and the
particles filtered to the preform. The suspended and filtered particles are treated as different species
for the purpose of derivation of the conservation equations, only. The conservation of mass for the
particles suspended in the resin is given by

where is an artificial term that corresponds to the rate of change of particle mass in
the suspension due to filtration (negative, when particles are filtered) in an infinitesimal
control volume 8o in the mold cavity. It does not actually imply generation of new mass and
is included to represent the transport of filler particles from the suspension to the fiber
preform (filtering) and vice versa. The mass conservation for the filtered particles is
expressed as

where represents the rate of increase of filtered particle mass and is related to the
particles in the suspension through

The continuity equation for the resin flow can be obtained as:

Noting that

where ϵ0 is the constant porosity of the fibrous medium (without filtered particles), and
combining Eqs. (1) through (5), the bulk continuity equation is obtained as

This is the same form of continuity equation as that for resin impregnation with no filler
particles. Combining Eqs. (1)–(3) yields the mass conservation for the particles:
The particle filtration during impregnation of fiber preform is modeled based on deep bed
filtration which relates the time rate of change of filtered particles to the volume flux of the
incoming particles by an amount α, the filtration coefficient [9],

The term σu is the maximum possible specific deposit of particles that can be filtered
(ultimate specific deposit) and is included in the model to avoid further filtration once a
clogging stage is reached. The suspension flow during impregnation is modeled by Darcy’s
law, which is commonly used in macroscopic modeling of resin impregnation in liquid
molding processes for continuous fiber reinforced composites.

Darcy’s law relates the velocity of the impregnating liquid to the pressure gradient in the
porous medium through the fluid viscosity and a macroscopic permeability representative of
the continuous porous medium of the flow domain. In this case, the permeability is a tensor
due to the multidimensionality of the flow. A permeability element K ij induces flow in i
direction due to the pressure gradient in j direction. If the coordinate system is chosen such
that the principal permeability directions align with the coordinate axes, the permeability
tensor becomes a diagonal matrix and the diagonal elements represent the principal
permeabilities of the fiber preform. In the current analysis, the principal permeability
directions coincide with the coordinate axes. The domain permeability will change with
respect to the amount of particles filtered onto the fiber preform. The permeability is modeled
with respect to the changing porous medium specific surface via Kozeny–Carman relation as

where Ko is the permeability of the fiber preform in the absence of particle deposit. During
the injection phase of the process, the local porosity e changes due to particle deposit via Eq.
(5), which in return, changes the local domain permeability via Eq. (10). In the case of
anisotropic fiber preform which has different permeability values in different directions, Eq.
(10) can be applied to modify the permeability in each direction to account for the effect of
particle deposit. The particle–resin suspension viscosity is related to the particle
concentration in the suspension as

where, μo is the neat resin viscosity and A is an empirical model constant. It is possible to
replace this model with other models such as in [10], specific to the type of particle–resin
suspension used in the process. The four equations; Darcy’s law (9), the bulk continuity (6),
the particle conservation Eq. (7) and the filtration Eq. (8), along with the viscosity and
permeability models of Eqs. (10) and (11), are solved to determine the four unknowns of the
flow: pressure (p), the velocity field (V), particle concentration (C) and specific particle
deposit (r). During solution, there is no flow across the solid boundaries (mold walls)
however no-slip condition on the mold walls is not imposed due to the first-order nature of
Darcy’s law. At the injection gate, either the injection velocity or the injection pressure is
specified. The pressure at the flow front is taken as atmospheric pressure since the mold
cavity has air vents to depose air replaced by the impregnating resin. The particle
concentration in the injected suspension at the inlet gate is specified.

3) Results and Discussion

Particle-filled impregnation for the injection and compression stages of CRTM in a

rectangular mold cavity with a center gate has been simulated. The mold and gate
configuration for CRTM are shown in Fig. 9. The model is solved for two-dimensional flow
in the cavity with no change of properties along the mold width, in y-direction. The
remaining process parameters are the same as those given in Table 1, except for the
compression speed, which is U = 0.1 mm/s and the filtration coefficient, a = 0.05 cm -1 . The
preform permeability is isotropic having the initial value given in Table 1 for both x and z
directions. Figure 10 presents the progression of the suspension flow front during the CRTM
process. The flow fronts in injection and compression stages are plotted with different time
increments in order to present the progression in each stage clearly. Injection lasts for 45 s
followed by the compression stage which lasts 300 s for the given compression speed. The
cavity height is constant during the injection phase and decreases during the compression
stage until the design composite thickness is reached. The latter is reflected in the flow front
patterns of compression. The isotropic nature of the fiber preform is apparent from the freely
expanding semi-circular flow front patterns near the gate. Although the flow is two-
dimensional during most of injection, it is approximately one-dimensional during the
compression stage with a nearly uniform profile moving along the mold length. The
production of the same composite (same design fiber volume fraction and final dimensions)
with RTM process has also been simulated to compare the two processes. In this case, the
mold cavity geometry and fiber preform properties are fixed at the values shown in Fig. 9b.
The injection is through the same center gate, however there is no compression and the
injection (which is at the same rate as that in CRTM) continues until the whole cavity has
been filled. The processing parameters are the same as those for CRTM except the
permeability of the isotropic preform is set to 3.75×10 -11 m2 which corresponds to the fiber
permeability at the end of the compression stage of CRTM via Eq. (10). The flow front
progression during impregnation is presented in Fig. 11. Without compression and with the
small cavity height compared to mold length, the impregnating flow is nearly one-
dimensional within the majority of the mold except during the initial stages of injection. The
isotropic nature of the preform is again visible near the gate. The total particle distribution
(particle volume per cavity volume = (ϵC+σ) at various instances of flow front progression
for RTM and CRTM processes are presented in Figs. 12 and 13. Both results are scaled in the
same way
.
based on the maximum particle content which occurs in RTM. It is seen that the particle
distribution in the composite at the end of the CRTM process exhibits a more homogeneous
trend compared to the part produced through RTM. In CRTM, even though the non-
homogeneity in particle distribution is significant at the end of the injection phase, the peak
concentration regions are smoothed out by the end of compression. The largest particle
volume fraction in the composite is found as 0.358 near the injection gate at the end of the
process, whereas for the part produced through RTM, the largest value is 0.556, again at the
gate. The flow front locations are the regions of lowest particle content in both cases. Figure
14 presents the pressure distribution at various instances during the CRTM process, and the
pressure distribution at the end of RTM process. In RTM, the pressure continuously increases
with injection until the end of the process, therefore only the conclusion of the process is
shown to observe the maximum pressure during the process. The maximum pressure values
in the mold during CRTM process at the presented time instances are also given. In CRTM,
the maximum pressure occurs at the end of the injection stage. After injection is terminated,
the pressure first drops but then continuously increases while the mold closes until the end of
compression stage. Comparing the two processes, the maximum pressure in RTM process is
seen to be significantly higher—nearly one order of magnitude—than that in CRTM process
when the same composite part is produced. This is expected since the continuous injection in
RTM occurs against a fibrous preform that has a low permeability and high fiber volume
fraction (design values). In CRTM, the design permeability and fiber volume fraction values
are reached only at the conclusion of the process, i.e. the end of compression, and the
majority of the impregnation process takes place through a less resistant preform. From the
simulation results, CRTM appears to be the better process in terms of imparting a more
homogenous filler particle distribution within the composite and lower processing pressures,
even though the process time is longer (345 s for CRTM compared to 45 s in RTM).
However, the processing parameters are influential in the results. For instance, when the
simulations are repeated for the same process parameters except for a lower filtration
coefficient of a = 0.01 cm-1 (which will depend on the fiber preform architecture as well as
the filler particles), the total particle distributions in the resulting composite produced through
the two alternative processes (RTM and CRTM) do not exhibit significant differences, even
showing a less homogeneous distribution in the case of CRTM (Fig. 15). The maximum
pressure (not shown) still occurs during the RTM process. CRTM process becomes a more
advantageous process compared RTM in yielding a more homogenous microstructure when
the particle filtration amount increases, although the process pressure is still highest in RTM.
Thus, whether one process is advantageous over the other will depend on the specific
composite material components and process parameters.

4) Conclusion

A particle–resin suspension impregnation model was used to analyze mold filling process in
CRTM of particle-filled, continuous fiber-reinforced composites. In the process, the mold is
partially closed during injection and the particle– resin suspension is injected directly into the
preform. The model is based on Darcy flow coupled with particle filtration and is capable of
simulating 2-D flow in irregular domains with a free surface. Suspension and fiber preform
properties such as viscosity and permeability were related to the changing particle content
that varied within the suspension and the fiber preform. The numerical solution was validated
by comparisons with simple analytical solutions and experimental results from the literature.
Processing simulations for the same composite with CRTM and RTM processes showed that
CRTM was more advantageous in terms of yielding smaller particle distribution non-
homogeneity in the composite, only when the particle filtration was significant, i.e. the
filtration coefficient associated with the fiber preform was high. The mold pressure was
found to be significantly lower in CRTM compared RTM. A closer analysis revealed that an
injection flow rate lower limit and a mold closing speed upper limit exist beyond which the
mold pressures in CRTM become comparable with or more than those in RTM. In CRTM, the
mold gate configuration was found to be very influential in the particle distribution of the
composite; the unconstricted one dimensional flow in single line-gate mold yielded
significantly lower particle distribution nonhomogeneity compared to two other mold
configurations. The mold pressures in the single line-gate mold were one order of magnitude
smaller than that of center-gate mold and comparable with that of the double line-gate mold.
The anisotropy of the fiber preform did not affect the range of particle volume fraction
distribution in the composite compared to the isotropic preform, except for the high particle
content near the injection gate. However, the distribution trend of the particles within the
composite was quite different, mirroring the two-dimensional flow pattern of anisotropic
impregnation. The expected increase in pressure due to smaller through-the-thickness
permeability in the freely expanding anisotropic flow did not yield higher molding pressure
than that in the isotropic flow which was constricted by contact with the mold wall, during
injection. The developed modeling tool can be used to analyze other mold configurations and
variations of CRTM process such as simultaneous injection-compression, in order to explore
the process limitations and to compare processing scenarios for choosing appropriate
processing method and/ or process parameter window. It can be extended, with appropriate
modifications on the particle filtration model, to any composite liquid molding process
involving particle fillers..

7) References

1. Shojaei A (2006) Numerical simulation of three-dimensional flow and analysis of

filling process in compression resin transfer moulding. Compos Part A Appl S
37(9):1434–1450

2. Coulter JP, Guceri SI (1988) Resin impregnation during the manufacturing of

composite materials subject to prescribed injection rate. J Reinf Plast Comp 7(3):200–
219
3. Nordlund M, Fernberg SP et al (2007) Particle deposition mechanisms during
processing of advanced composite materials. Compos Part A Appl S 38(10):2182–
2193

4. Pham XT, Tirochu F (1999) Simulation of compression resin transfer molding to

manufacture thin composite shells. Polym Compos 20(3):446–459
5. Chang C-Y (2006) Simulation of mold filling in simultaneous resin
injection/compression molding. J Reinf Plast Comp 25(12):1255–1268

6. Simacek P, Advani SG (2004) Desirable features in mold filling simulations for

liquid composite molding processes. Polym Compos 25(4):355–367

7. Tan CP, Springer GS (1999) Composite manufacturing: simulation of 3-D resin

transfer molding. J Compos Mater 33(18):1716–1742

8. Minaie B, Chen YF, Mescher AM (2002) A methodology to obtain a desired filling

pattern during resin transfer molding. J Compos Mater 36(14):1677–1692

9. Erdal M, Guceri SI, Danforth S (1999) Impregnation molding of particle–filled

preceramic polymers: process modeling. J Am Ceram Soc 82(8):2017–2028

10. Lefevre D, Comas-Cardona S et al (2009) Coupling filtration and flow during

liquid composite molding: experimental investigation and simulation. Compos Sci
Technol 69(13):2127–2134

11. Lefevre D, Comas-Cardona S et al (2007) Modeling the flow of particle–filled

resin through a fibrous preform in liquid composite molding technologies. Compos
Part A Appl 38:2154–2163

12. Shojaei A (2006) A numerical study of filling process through multilayer preforms
in resin injection/compression molding. Compos Sci Technol 66:1546–1557

13. Bhat P, Merotte J, Simacek P, Advani SG (2009) Process analysis of compression

resin transfer molding. Compos Part A Appl 40(4):431–441
14. Wirth S, Gauvin R (1998) Experimental analysis of mold filling in compression
resin transfer molding. J Reinf Plast Comp 17(16):1414–1430