Processability of Pultrusion Using Natural Fiber and Thermoplastic Matrix

Fundamental mechanisms of the pultrusion process using commingled yarns of polypropylene matrix and discontinuous flax fiber to produce thermoplastic profiles were investigated in numerical and experimental manners. Essential issue is the fact that all natural fibers are discontinuous by nature, which may negatively influence the processability. The pultrusion process will be only successful if the pulling force exerted on the solidified pultrudates can be transmitted to the regions of unmelted commingled yarns by “bridging over” those melted regions within the die. This can be realized by applying a sufficient number of small yarn bundles of high compactness rather than a thicker single bundle of lower compactness as the raw material. Furthermore, the possibility of adding extra melt into the yarn bundles by side-fed extrusion has been investigated showing that the impregnation can be improved only for the outer layers of yarns, which is owed to the high viscosity of the thermoplastic melt and the limited length of the die.


INTRODUCTION
Pultrusion is a continuous process for manufacturing fiber reinforced profiles of constant cross-section.So far, the process has been mainly applied for thermosetting matrices as there are inherent difficulties associated with the thermoplastic matrices, such as high processing temperatures and high melt viscosities.However, with recent developments of preforms to facilitate impregnation, such as pre-impregnated tapes, commingled yarns, and powder-impregnated bundles [1,2], thermoplastic pultrusion has gained a greater interest.
Figure 1 shows the heating die as an opening and steadystate thermodynamic system with reinforcing filaments and polymer melt fed separately (case A) or in term of commingled yarns (case B).The drag flow is a result of the movement of the filaments and the no-slip condition.Moreover, in case A the drag flow is superposed by the pressure-driven flow.With a proper design of the die inlet, the drag flow can compensate the pressure driven flow so that no sealing is needed.In case B no precaution is required since the melt is not pressurized.
The pultrusion process will be only stable if the force flux is properly transmitted from the outlet to the inlet, which is possible using continuous fibers like glass fiber rovings.Natural fibers are, however, not endless by nature.Their average length is normally shorter than the minimum length of the die which is required to ensure sufficient heat transfer and proper impregnation.So the only way to maintain the force flux between the outlet and inlet of the die is to make use of special textile technique to spin the separate fibers together into a bundle or to build extra core cords into the yarns.In this work, the fundamental mechanisms of the pultrusion of natural fibers reinforced thermoplastics were investigated on commingled yarns of polypropylene and flax fibers, whereby the possibility of adding extra melt by side-fed extrusion to improve impregnation was also considered.The objective of the paper is to find out a proper way to maintain the force transmission along the discontinuous yarns while at the same time ensuring a proper impregnation of the commingled yarns.

SIMULATION
Numerical simulation was carried out for a two-dimensional rectangular geometry, that is, effects caused by the side edges were neglected.First, it was considered that the yarns act like a continuous fiber bundle.The calculated temperature field will serve as an assumption for the temperature in further simulations.Although this may not meet the reality, it is a good way to deal with such unstable processes, where a coupled steady simulation would end up with no convergence.In a next step, the flow situation of the discontinuous yarns containing virtual cords was considered, whereby the velocities at the corresponding cord positions were fixed (Figure 2).In both cases, the commingled yarns were modeled as a fluid; and inertia, gravitation, and compressibility were neglected so that the equations of mass, momentum, and energy can be written as follows: where t, v, T, p, τ , γ , ρ, c p , and k denote time, velocity vector, temperature, hydrostatic pressure, extra-stress tensor, rate of deformation tensor, density, specific heat, and heat conductivity, respectively.The constitutive equation for a generalized Newtonian fluid is The viscosity was approximated using the Bird-Carreau model [3]: where η 0 is the zero-shear-rate viscosity, n is the power law index, and λ is the time constant.The Arrhenius model was used for the viscosities at different temperatures: with the reference temperature T 0 and the Arrhenius coefficient α.The thermal and rheological properties of polypropylene and commingled yarns in the melted state are listed in Table 1.Finally, the situation using polypropylene as an extra melt to additionally impregnate the yarns was considered, whereby the flow in a porous media is governed by the Darcy law: ( The longitudinal component K xx of the permeability tensor K is calculated using the Kozeny-Carman model [5]: where D f is the fiber diameter, V f is the fiber volume fraction (Table 2), and C is the Kozeny-Carman constant.The value of the Kozeny-Carman constant C was taken as 0.093 [6].The transversal component K yy was given by the Bruschke model [7]: The following boundary conditions were set (Figure 2): symmetry condition at the symmetry, slip wall along the yarn/air interface, no-slip along the die wall, constant velocity v p along the so-called virtual cords which may be added into the cavity to simulate the solid areas within the melt, constant temperatures at the yarn and melt inlet.At the interface with the die and the air, heat flux densities were defined as follows: with constant die and air temperature T d and T a as well as constant heat transfer coefficients α d and α a .The die temperature T d was kept constant.At the melt inlet, the velocity was either zero or constant, depending on whether extra melt was considered.
The governing equations together with the constitutive equation were solved numerically using the FE-code FIDAP [8].A mixed formulation was applied for pressure and velocity.In transient case, the implicit backward Euler approach was used for the time integration.

Continuous yarns
Figure 3 shows exemplarily the temperature along the symmetry line in case the yarns were considered as being continuous.At a low pulling velocity v p = 16 cm/min, the section containing the melting yarns, that is, where their temperatures are above the melting point T m of the polypropylene matrix (T m = 162 • C), reaches a length of approximately 80 mm, which means that a proper melting of the yarns throughout the commingled yarns is generally possible.The main problem is therefore to find a proper way to maintain the force transmission along the yarns, while enabling a proper melting of the yarns.This can be done by increasing the pulling speed above 44 cm/min to reduce the melting section below the average length of the natural fibers (L f ≈ 40 mm) so that it is possible to transmit the pulling force along the symmetry of the die.

Discontinuous yarns
When the commingled yarns are partially melted, they cannot be assumed as being continuous.However, virtual cords may be added along the unmelted area of the yarns to enforce the transmission of the pulling velocity.In the following cases, either one virtual cord was added along the symmetry line (C1) or up to three other ones are distributed equidistantly over half of the die thickness (C2 to C4).Along the cords, the velocity is kept equal to the pulling velocity.
In Figure 4 the corresponding longitudinal velocity at the die outlet is shown respectively.It can be seen clearly that with one cord at the symmetry line (C1), the longitudinal velocity obtained is far away from this ideal situation of a uniform velocity across the thickness.As the number of the cords increases, the velocity profiles at the outlet show a wider region of constant velocity in the middle of the die.With four extra cords (C4) a good shaping of the finished product may be expected, since the velocity profile at the outlet is not so much different from a typical velocity profile in an extrusion die for a shear-thinning polymer.Since significant improvement of the velocity profile can be still achieved as the cord number over half of the cavity thickness was increased from 2 to 3, it is recommendable that at least 5 cords should be used for the whole cavity thickness to ensure a good shape of the pultrudate.

Extra melt
When extra melt was inserted to improve the impregnation of the yarns which were modeled as a porous media, the movement of the yarns was only considered in the temperature calculation, but not in the flow simulation.Figures 5  and 6 show contour plots of the velocity parallel and across the flow direction within the contact region yarn/melt.Based on the transversal velocity v y at a representative cross-section (e.g., along the line CDE) as shown in Figure 6, the time needed for the extra melt to fully impregnate the whole thickness of the yarns was estimated.In current case, the time required for a full impregnation is much longer than the time the yarns stay in the contact region, which means that only the outer layers of the commingled yarns can be properly impregnated.In order to improve the impregnation of the inner layers, the length of the contact region should be significantly increased.Moreover, the contact region should end at a certain distance away from the outlet to avoid such a peak of the longitudinal velocity v x occurred at the gap between the yarns and the die wall (Figure 5, bottom), which may negatively influence the shape of the finished product.

EXPERIMENTS
Experiments were conducted on commingled yarns of polypropylene and flax fibers [9], which were spun by the Thüringisches Institut für Textil-und Kunststoff-Forschung (Rudolstadt, Germany).Rectangular profiles of 3.5 mm × 10 mm were manufactured using the self-designed facility at the Institut für Verbundwerkstoffe (Kaiserslautern, Germany) (Figure 7).Several bundles of yarn were stored in a creel stand (Figure 8) and then guided into the preheating chamber, where it was heated to a temperature close to the melting temperature of the polymeric fibers.Directly after preheating, the yarn bundles were pulled through the heating die where the polymer is supposed to melt and impregnate the flax fibers before being consolidated in the cooling die.Table 3 shows some of the process conditions which led to successful experiments.Some successful examples of the finished profiles can be seen in Figure 9, whereby 80 bundles of the yarns have been used to fill the die cavity.Assuming that the cavity was totally filled and each bundle is of rectangular geometry, the edge length of each bundle can be calculated, which is approximately 0.66 mm.In other words, 5 to 6 yarn bundles were needed for a die thickness of 3.5 mm to ensure the success of the pultrusion process, which verified well the theoretical requirements for the number of yarns as revealed by the numerical simulation.
This way, it can be shown that the pultrusion process is stable when a sufficient number of yarn bundles of high   compactness are used instead of a single bundle of less compactness.The reason might be that due to the compactness some spots of the compact bundles remain unmelted, which behave like core cords so that a throughout force transmission is possible.Moreover, the compact bundles are less sensitive to break and also the force required to pull the compact bundles is smaller since there is no need to compress the bundle into the die.

CONCLUSIONS
The simulations and experiments have revealed some of the fundamental issues about the pultrusion of discontinuously fiber reinforced thermoplastics.In general, it is possible to produce pultrusion profiles by using a sufficient number of compact bundles of commingled yarns as raw material.The requirement for the minimal number of yarn bundles to ensure a good shape of the finished pultrudate may be theo-retically calculated using numerical simulation.Extra melt inserted into the yarn bundles by side-fed extrusion to improve the impregnation effects only the outer layers of yarns, which is owed to the high viscosity of the thermoplastic melt and the limited length of the die.

Figure 1 :
Figure 1: Heating die and the flow situation at the inlet and outlet.

Figure 5 :
Figure 5: Longitudinal velocity in the die (top) and along the outlet (bottom).

Figure 6 :Figure 7 :
Figure 6: Transversal velocity in the die (top) and along the line CDE (bottom).

Table 3 :
Pultrusion parameters using commingled yarns of polypropylene and flax.