Professional Documents
Culture Documents
To cite this article: Osuke Ishida, Junichi Kitada, Katsuhiko Nunotani & Kiyoshi Uzawa (2020):
Impregnation and resin flow analysis during compression process for thermoplastic composite
production, Advanced Composite Materials
Article views: 47
1. Introduction
Carbon fiber reinforced thermoplastics are gaining remarkable interest because they have
advantages such as short cycle times, toughness, and recyclability. In recent years,
various fundamental studies have been conducted on the application of this material
[1–3]. Thermoplastic composite sheet, also known as organo-sheet, is a promising
material for industrial and automobile parts because it enables a short cycle production
by stamp forming and welding [4–8]. Normally, organo-sheet can be produced by
compression molding using fiber reinforcements and thermoplastic resin. The key aims
are to achieve good impregnation quality and low cost. However, it is difficult to achieve
these properties together because the impregnation is difficult due to the high melt
viscosity. Therefore, the impregnation technology has been a subject of research interest
for thermoplastic composites [9–12].
A double-belt press is one of the solutions for organo-sheet production. The machine
consists of heat and pressure modules for impregnation and a subsequent cooling module
for solidification (Figure 1), enabling a continuous compression molding process.
Several investigations have been reported about this system [13–18]. There are two
types of press systems for a double-belt press, the isobaric press and the roller press
(Figure 2). The isobaric press provides constant and uniform pressure for the impregna-
tion, and the process parameters can be easily controlled by a machine setup. However,
the roller press is more suitable for the impregnation process in a double-belt press
© 2020 Japan Society for Composite Materials, Korean Society for Composite Materials and Informa UK Limited, trading as
Taylor & Francis Group
2 O. Ishida et al.
Figure 1. Schematic of the fixed rollers double-belt press for organo-sheet production.
Figure 2. Difference of the isobaric press system and the roller press system.
because the machine cost is relatively lower than isobaric press. Additionally, different
types of materials can be applied, such as discontinuous reinforcements and thick
composites. The material is compressed along the steel belt deflection under a roller
(Figure 3). As there is a pressure gradient between the outside of a roller and the region
underneath it, not only fabric impregnation but in-plane resin flow occurs as illustrated in
Figure 4. This phenomenon is prominent while using a film-stacking material. The
applied pressure is not constant under a roller, which is determined by the balance of
the impregnation and in-plane flow under compression. The difficulty of optimizing the
process parameters is a critical issue for the composite manufacturing process using
a double belt press. Thus, the understanding of the resin flow mechanism under a roller is
very important.
Resin impregnation process during compression molding has been studied by many
researchers [19–24]. In these studies, the resin flow through fiber network has been
Advanced Composite Materials 3
Figure 3. Material compression behaviour under a roller in the double belt press (in the case of
film-stacking material).
Figure 4. Impregnation and in-plane flow behavior under a roller in the double belt press.
modelled based on Darcy’s law. For example, Gutowski et al. [19] evaluated the
transverse permeability of fiber bundles based on fiber elastic deformation model.
Michaud et al. [20] developed the impregnation model of glass mat thermoplastic
(GMT) based on local resin flow in compressible porous media. Kobayashi et al. [21]
studied the impregnation behavior of Micro-Braided-Yarn based on the ellipsoidal model
of fiber yarn and evaluated the effects of the yarn conditions. For the compression resin
transfer molding (CRTM) process, Merrote et al. [22] modelled the resin in-plane flow in
fiber preform during compression process. Their works have successfully explained the
resin flow behavior through fiber reinforcements. However, these analyses cannot
describe the combined process of the fiber bundle impregnation and resin in-plane
flow of film stacking material. The aim of our study is to clarify this combined
impregnation and in-plane flow behavior.
In this study, model experiments were conducted to study this flow behavior. The
stack of carbon fiber woven fabrics and viscous fluid were compressed using a testing
machine equipped with a flat mold under several process conditions. The fabric
4 O. Ishida et al.
impregnation, in-plane flow, and pressure profiles were characterized from the experi-
mental observations. Furthermore, a theoretical model was developed to describe the
impregnation and in-plane flow behavior. The predicted result was compared with the
experimental data.
Figure 5. Experimental equipment using a testing machine. The mold setup is available with (a)
shear-edge and (b) open-edge configurations by attaching the side walls.
in the two directions (x, y) were measured with an embedded pressure sensor. The
permeabilities were calculated based on Darcy’s equation in polar coordinates from the
measured pressure drops. The fluid viscosity was measured to be 44 mPa·s and the injection
pressure was 0.28 MPa. The average value was obtained from four measurements.
Vf V i f
þ V i f þ Vr ¼ Vm (2)
vf
Advanced Composite Materials 7
0.6
0.5
where Vf is the fabric solid volume, Vif is the impregnated part of the fabric solid volume, Vr
is the fluid volume, and Vm is the mold cavity volume. In this equation, (Vf – Vif) indicates
the solid volume of fibers in the unimpregnated part. The fibers in the unimpregnated part
are compressed by the stress equivalent to resin pressure during the impregnation process.
Thus, the fiber volume fraction vf in the unimpregnated part can be given by Equation (1).
By dividing (Vf – V if) by vf, one can obtain the bulk volume of unimpregnated part that
consists of the unimpregnated fibers and voids. Finally, the left-hand side of Equation (2)
comprises of the volume of unimpregnated fibers and voids (the first term), the volume of
impregnated fibers (the second term) and the volume of resin (the third term).
Based on the cross-sectional observation in Figure 6, the impregnation process can be
modeled as illustrated in Figure 9. The resin fills up the space outside the fiber bundles at
the initial stage, and then it starts to infiltrate into the fiber bundles in one direction. The
time increment of the impregnation length l in the fiber bundle can be expressed
according to the following equation.
ΔV i f
Δl ¼ (3)
10Avf
where A is the base area of the mold cavity. There are total ten impregnation flows with both
surfaces of the five layers of fabrics, thus, which explains the division by 10 in Equation (3).
The impregnation behavior can be described by Darcy’s law [19–24,26-30]. In this study, the
equation is defined as follows.
Δl Kb P
1 vf ¼ (4)
Δt μ l
where P is the resin pressure, t is the time, and μ is the resin viscosity. We refer to Kb as
the apparent fiber bundle permeability in this study. The degree of impregnation is
defined as follows.
Advanced Composite Materials 9
(a) 1400
1200
1000
Force (N)
800
600
400
640 N
200
1280 N
0
0 20 40 60
Time (s)
(b) 2500
0.5 mm/min
2000 1 mm/min
Force (N)
1500
1000
500
0
0 40 80 120 160
Time (s)
(c) 2
640 N
1.8 1280 N
Thickness (mm)
1.6
1.4
1.2
1
0 20 40 60
Time (s)
(d) 2
0.5 mm/min
1.8 1 mm/min
Thickness (mm)
1.6
1.4
1.2
1
0 40 80 120 160
Time (s)
Figure 8. Experimental results of (a), (b) applied force change and (c), (d) material thickness
reduction under different process conditions using the testing machine.
10 O. Ishida et al.
V if
α¼ (5)
Vf
Using Equations (1), (2) and (5), one can calculate the degree of impregnation. Using
Equations (1)–(4), one can calculate the apparent bundle permeability. These are obtained at
a time step of Δt from the experimental data. The calculated results are shown in Figure 10.
The relationship between the impregnation rate, impregnation length, and applied pressure
can be explained based on Equation (4). If the applied pressure is constant, the impregnation
rate will decrease with increasing impregnation. On the other hand, if the impregnation rate
is constant, the applied pressure will increase with increasing impregnation. This explana-
tion is consistent with the experimental results of the force profiles in Figure 8(a,b) and the
degree of impregnation in Figure 10(c,d). It can be observed that the apparent fiber bundle
permeability decreases with time. This may be attributed to the following effects. The filling
process outside the fiber bundles can cause a very high permeability in the initial stage.
Fiber alignment at the interface between the resin layer and fiber bundle is non-uniform as
shown in Figure 6. This causes a higher permeability compared to the core region of the
fiber bundle. Another effect is air entrapment inside the fiber bundles, which especially
occurs in woven fabric architectures [14,26]. The entrapped air generates internal pressure
due to the advancing flow front. Consequently, the apparent fiber bundle permeability
decreases with time. Based on these considerations, we thought that the apparent fiber
bundle permeability could be described as a function of the degree of impregnation. The
result is shown in Figure 11. The resulting curves are similar across the different process
conditions. Slight differences between the curves may be attributed to the difference of the
fiber volume fractions, which vary based on the different pressure conditions as suggested
by the fabric compression behavior in Figure 7. It is known that the transverse permeability
of fiber bundles depends on the fiber volume fraction [19–21]. However, the permeability
difference is not significant in this experimental range when it is compared with the
variation of the fiber volume fraction. The curves are approximated using a Gaussian
function as shown in Equation (6). This approximation is used for the impregnation
simulation in the following discussion.
14 α2
Kb ¼ 4:3 10 exp (6)
0:28
Advanced Composite Materials 11
(a) 4E-14
640 N
1280 N
2E-14
1E-14
0
0 20 40 60
Time (s)
(b) 4E-14
0.5 mm/min
1 mm/min
Bundle permeability (m2)
3E-14
2E-14
1E-14
0
0 40 80 120 160
Time (s)
(c) 1
0.8
Degree of impregnation
0.6
0.4
0.2 640 N
1280 N
0
0 20 40 60
Time (s)
(d) 1
0.8
Degree of impregnation
0.6
0.4
Figure 10. (a), (b) Apparent fiber bundle permeability and (c), (d) the degree of impregnation
calculated from the experimental data under different process conditions.
12 O. Ishida et al.
4E-14
640 N
1280 N
1E-14
0
0.1 0.3 0.5 0.7 0.9
Degree of impregnation
Figure 11. Relationship between the apparent fiber bundle permeability and the degree of
impregnation under different process conditions.
@P @2u
¼μ 2 (7)
@x @z
where u is the velocity in the x-direction (in-plane flow direction) and z is the thickness
direction. The no-slip boundary conditions are:
where h is the fluid thickness, u1 and u2 are the velocities in the x-direction at the top
and bottom surfaces. By integrating Equation (7) twice with the above boundary condi-
tions, we get:
1 @P 2 z z
u¼ z zh þ 1 u1 þ u2 (8)
2μ @x h h
Advanced Composite Materials 13
In this experiment, u1 and u2 are zero at the mold surfaces. There is no flow in the width
direction, thus, the continuity equation can be formulated as follows.
@u @v
þ ¼0 (9)
@x @z
where v is the velocity in the z-direction. By substituting Equation (8) to (9) and by
integrating across the thickness from z = 0 to z = h, one can obtain the following
equation.
@ h3 @P
V ¼ (10)
@x 12μ @x
where V is the compression speed. The pressure field inside the mold can be determined
by applying the following two boundary conditions at the center position (symmetry
boundary condition) and the edge position Le in the mold cavity.
@P
¼ 0 and Px¼Le ¼ 0
@x x¼0
6μV 2
Pð x Þ ¼ Le x2 (11)
h3
The force acting on the mold plate F can be obtained by integrating the fluid pressure on
the surface.
ð Le
8wμLe 3 V
F ¼ 2w Pð xÞdx ¼ (12)
0 h3
where w is the width of the fluid. The applied force F can be calculated at each time step
using Equation (12) when the compression speed is given.
The comparison of the experimental and prediction force profiles is shown in Figure 12.
The theoretical model achieves a good fit with the data. A slight discrepancy might have been
caused due to the theoretical assumption of no-slip boundary condition at the mold surface.
2500
2000
Load (N)
1500
1000
1.6E-10
Experiment
Fitting
In-plane permeability (m2)
1.2E-10
8E-11
4E-11
0
0.4 0.45 0.5 0.55 0.6
Fiber volume fraction
Figure 13. Measured in-plane permeability of the carbon fiber woven fabrics.
3.6. Analysis of the combination of impregnation and in-plane flow using a testing
machine
The combination experiments of impregnation and in-plane flow have been conducted
using a testing machine equipped with an open-edge mold. When the mold edge is under
atmospheric pressure, there is a pressure gradient between the center position and the
edge in the mold, which causes in-plane flow in the resin layer and fabric layer.
Advanced Composite Materials 15
Qb Δl Kb P
¼ 1 vf ¼ (14)
wd Δt μ l
where w is the width of the in-plane flow, d is the in-plane length of each area, l is the
impregnation length in the fiber bundle, and P is the resin pressure.
The in-plane flow rate in fabric layer Qf can be expressed as Darcy flow. The in-plane
permeability Ki was previously obtained in Section 3.5. The equation is defined as follows.
Qf Ki ΔP
¼ (15)
wl μ 2d
where ΔP indicates the pressure difference between the adjacent areas. There is
a pressure gradient from the resin pressure in the resin layer P to the atmospheric
pressure in the unimpregnated part of a fiber bundle due to the fiber bundle impregnation
as illustrated in Figure 9. We took the mean pressure (P/2) to define the resin pressure
Qr P1 P2 P3 P4 P5 h In-plane flow
Outflow
Qf Fiber bundle
l impregnation
Qb
z
d Mold
x
Resin pressure
Figure 14. The combination model of impregnation and in-plane flow during compression
process.
16 O. Ishida et al.
inside the fabric in Equation (15) for simplicity, which explains the division by two in the
equation.
The in-plane flow rate in the resin layer Qr can be expressed as a thin-layer Stokes
flow between the fabric surfaces based on the discussion in Section 3.4. It can be
calculated by integrating Equation (8) across the resin layer thickness h. It has been
reported that slippage occurs at the interface between the fluid velocity in the resin layer
and Darcian velocity in the fabric layer, however, the effect can be negligible due to the
small transition distance [33]. Thus, the fluid velocity at the fabric surface is assumed to
be the same as that of the in-plane flow velocity in the fabric layer. Consequently, the
flow can be expressed as:
wh3 ΔP h
Qr ¼ þ Qf (16)
12μ d l
ΔVr
¼ Qrin Qrout þ Qfin Qfout (17)
Δt
where ΔVr/Δt is the resin volume change with time in each area, which is determined by
the inflow rate Qrin and outflow rate Qrout in the resin layer, and the inflow rate Qfin and
outflow rate Qfout in the fabric layer. When Equation (2) and Equations (14)-(17) are
combined, one can formulate the system of equations for the mass conservation of the
whole area. Using the solver tool of Microsoft Excel, the system of equations has been
solved at a time step of 1 s. Consequently, we obtained the resin pressure in each area
and the compression speed at each time step. Figure 15 shows the compression speed
obtained from the model and experiment under a controlled force of 640 N (0.4 MPa).
Figure 16 shows the applied force profiles obtained from the model and experiment at the
controlled compression speeds of 0.5 mm/min and 1 mm/min. It is found that the
0.005
Model
Compression speed (mm/s)
0.004 Experiment
0.003
0.002
0.001
0
0 20 40 60
Time (s)
Figure 15. Comparison of compression speed obtained from the model and experiment under
a controlled force of 640 N (0.4 MPa).
Advanced Composite Materials 17
experimental data is well represented by the theoretical model. The degree of impreg-
nation obtained from the model and experiment is shown in Figure 17. The result shows
reasonable agreement. The values of in-plane outflow obtained from the model and
experiment are shown in Figure 18. The values of outflow are less than one-tenth of
the values of total fiber bundle impregnation (Σ Qb) and there are differences between the
predicted and experimental values. The possible causes are the variations of the experi-
mental setup and the inaccuracies of the approximation Equations (1), (6), and (13) as
discussed before.
2500
0.5 mm/min
1 mm/min
2000
Load (N)
1500
1000
500
0
0 20 40 60 80 100 120 140
Time (s)
Figure 16. Comparison of the force profiles obtained from the model and experiments under
controlled compression speeds of 0.5 mm/min and 1 mm/min: the solid line is the prediction, the
dotted line is experimental data.
1
Model Experiment
0.8
Degree of impregnation
0.6
0.4
0.2
0
(a) Speed: (b) Speed: (c) Force:
0.5 mm/min 1 mm/min 640 N
Figure 17. Comparison of the degree of impregnation obtained from the model and experiment
under different compression conditions: (a) controlled speed of 0.5 mm/min, (b) controlled speed
of 1 mm/min, (c) controlled force of 640 N.
18 O. Ishida et al.
50
Model Experiment
40
20
10
0
(a) Speed: (b) Speed: (c) Force:
0.5 mm/min 1 mm/min 640 N
Figure 18. Comparison of the values of in-plane outflow obtained from the model and experi-
ment under different compression conditions: (a) controlled speed of 0.5 mm/min, (b) controlled
speed of 1 mm/min, (c) controlled force of 640 N.
Although a few improvements are possible, the theoretical model explains well, the
impregnation and in-plane flow of the film-stacking material under compression process.
In the future, this model will be applied to investigate the impregnation behavior under
rollers in a double-belt press process and optimize the process parameters.
4. Conclusions
In this study, the impregnation and in-plane flow of the film-stacking material under
compression process have been investigated based on experiments and a theoretical
model. The motivation of this study is to understand the impregnation behavior under
pressure rollers in a double-belt press for organo-sheet production. As there is a pressure
gradient between the outside of the rollers and the region underneath it, not only fabric
impregnation, but also in-plane resin flow occurs during the compression process. The
applied pressure is determined by these combination flows.
At first, the impregnation process has been investigated by the cross-section micro-
scopy of composites fabricated by compression molding. Subsequently, impregnation
experiments have been conducted using a testing machine equipped with a shear-edge
mold in several compression conditions. Consequently, the fiber bundle impregnation has
been characterized based on Darcy’s law. For the next step, the in-plane flow behavior
has been studied using a testing machine equipped with an open-edge mold. The
prediction model based on thin-layer Stokes flow has shown good agreement with the
experimental data. Finally, the combination behavior of fiber bundle impregnation and
in-plane flow has been studied through similar experiments. The experimental data has
been well represented by the prediction model, although a few improvements are
possible. In the future, the developed model will be applied to study the double-belt
press process in order to optimize the process parameters.
Advanced Composite Materials 19
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This work was supported by the JST COI Grant Number [JPMJCE1315].
References
[1] Taketa I, Kalinka G, Gorbatikh L, et al. Influence of cooling rate on the properties of carbon
fiber unidirectional composites with polypropylene, polyamide 6, and polyphenylene sulfide
matrices. Adv Compos Mater. 2020;29:101–113.
[2] Wei H, Nagatsuka W, Ohsawa I, et al. Influence of small amount of glass fibers on mechanical
properties of discontinuous recycled carbon fiber-reinforced thermoplastics. Adv Compos
Mater. 2019;28:321–334.
[3] Guo Q, Ohsawa I, Takahashi J. L-shaped structures made from ultrathin chopped carbon fiber
tape reinforced thermoplastics: delamination behavior and optimization. Adv Compos Mater.
2019;28:479–489.
[4] Hou M. Stamp forming of continuous glass fiber reinforced polypropylene. Compos Part A.
1997;28A:695–702.
[5] Vieille B, Albouy W, Chevalier L, et al. About the influence of stamping on
thermoplastic-based composites for aeronautical applications. Compos Part B.
2013;45:821–834.
[6] Wang P, Hamila N, Boisse P. Thermoforming simulation of multilayer composites with
continuous fibres and thermoplastic matrix. Compos Part B. 2013;52:127–136.
[7] Yousefpour A, Hojjtai M, Immarigeon J-P. Fusion bonding/welding of thermoplastic
composites. Compos Part A. 2006;37(10):1638–1651.
[8] Shirai T, Uzawa K. Relationship between tensile properties and fiber orientation after press
forming of discontinuous carbon fiber reinforced thermoplastic composite. 16th Japan
International SAMPE Symposium; 2019 Sep 2–4;Tokyo, Japan. p. 3A04.
[9] Gibson AG, Månson J-A. Impregnation technology for thermoplastic matrix composites.
Compos Manuf. 1992;3(4):223–233.
[10] Khondker OA, Ishiaku US, Nakai A, et al. A novel processing technique for thermoplastic
manufacturing of unidirectional composites reinforced with jute yarns. Compos Part A.
2006;37:2274–2284.
[11] Nishida N, Nunotani K, Uzawa K. In situ-polymerizing thermoplastic epoxy resin which
enables molding processes corresponding to various forms of thermoplastic composites. 18th
European conference on composite materials; 2018 June 24-28; Athens, Greece. P-5.19.
[12] Ishida O, Uzawa K, Kinpara I. Production method of carbon fiber woven fabric/PA6 compo-
site using PA6 solution impregnation process. Reinf Plast. 2016;62(8): 318–323. Japanese.
[13] Wang X, Mayer C, Neitzel M. Some issues on impregnation in manufacturing of thermo-
plastic composites by using a double belt press. Polym Compos. 1997;18(6):701–710.
[14] Mayer C, Wang X, Neitzel M. Macro-and micro-impregnation phenomena in continuous
manufacturing of fabric reinforced thermoplastic composites. Compos Part A.
1998;29:783–793.
[15] Liu D, Zhu Y, Ding J, et al. Experimental investigation of carbon fiber reinforced poly
(phenylene sulfide) composites prepared using a double-belt press. Compos Part B.
2015;77:363–370.
[16] Trende A, Astrom BT, Woginger A, et al. Modeling of heat transfer in thermoplastic
composites manufacturing: double-belt press lamination. Compos Part A.
1999;30:935–943.
[17] Ishida O, Fukushima G, Kitada J, et al. Continuous impregnation process by fixed rollers
double belt press. 17th European conference on composite materials; 2016 June 26-30;
Munich, Germany. STG-5.09-04.
[18] Ishida O, Kitada J, Uzawa K. Investigation of continuous impregnation process by fixed
rollers double belt press. Mater Syst. 2017;35: 53–59. Japanese.
20 O. Ishida et al.
[19] Gutowski TG, Cai Z, Bauer S, et al. Consolidation experiments for laminate composites.
J Compos Mater. 1987;21:650–669.
[20] Michaud V, Manson J-AE. Impregnation of compressible fiber mats with a thermoplastic
resin. Part I: theory. J Compos Mater. 2001;35(13):1150–1173.
[21] Kobayashi S, Tsukada T, Morimoto T. Resin impregnation behavior in carbon fiber reinforced
polyamide 6 composite: effects of yarn thickness, fabric lamination and sizing agent. Compos
Part A. 2017;101:283–289.
[22] Merotte J, Simacek P, Advani SG. Flow analysis during compression of partially impregnated
fiber preform under controlled force. Compos Sci Technol. 2010;70:725–733.
[23] Ye L, Klinkmüller V, Friedrich K. Impregnation and consolidation in composites made of GF/
PP powder impregnated bundles. J Thermoplast Compos Mater. 1992;5:32–48.
[24] Lekakou C, Johafu MAKB, Bader MG. Compressibility and flow permeability of
two-dimensional woven reinforcements in the processing of composites. Polym Compos.
1996;17(5):666–672.
[25] Yenilmez B, Caglar B, Sozer EM. Pressure-controlled compaction characterization of fiber
preforms suitable for viscoelastic modelling in the vacuum infusion process. J Compos Mater.
2017;51(9):1209–1224.
[26] Foley ME, Gillespie JW Jr. Modeling the effect of fiber diameter and fiber bundle count on
tow impregnation during liquid molding processes. J Compos Mater. 2005;39(12):1045–1065.
[27] Murphy CS, Simacek P, Advani SG, et al. A model for thermoplastic melt impregnation of
fiber bundles during consolidation of powder-impregnated continuous fiber composites.
Compos Part A. 2010;41:93–100.
[28] Koubaa S, Le Corre S, Burtin C. Thermoplastic pultrusion process: modeling and optimal
conditions for fibers impregnation. J Reinf Plast Comp. 2013;32(17):1285–1294.
[29] Jespersen ST, Wakeman MD, Michaud V, et al. Film stacking impregnation model for a novel
net shape thermoplastic composite preforming process. Compos Sci Technol.
2008;68:1822–1830.
[30] Studer J, Dransfeld C, Cano JJ, et al. Effect of fabric architecture, compaction and perme-
ability on through thickness thermoplastic melt impregnation. Compos Part A.
2019;122:45–53.
[31] de Vicente J, Ruiz-López JA, Andablo-Reyes E, et al. Squeeze flow magnetorheology.
J Rheol. 2011;55(4):753–779.
[32] Ren F, Yu Y, Yang J, et al. A mathematical model for continuous fiber reinforced thermo-
plastic composite in melt impregnation. Appl Compos Mater. 2017;24:675–690.
[33] Hammami A, Gauvin R, Trochu F. Modeling the edge effect of liquid composites molding.
Compos Part A. 1998;29A:603–609.