Sage Journals: Discover world-class research

Abstract

During thermoplastic pultrusion, increasing the pulling speed generally leads to a melted polymer pressure rise in the pultrusion dies. Yet, it was not clear if the polymer pressure’s increase would result in a faster impregnation flow that would compensate for the lower residency time inside the heating dies. Thus, the aim of this study was to verify if the speed-induced pressure can counterbalance the decrease in residency time that is required for impregnation quality and to analyze possible pultrudate morphology’s reconfiguration. Three distant pulling speeds were selected: namely 50, 500 and 1000 mm/min. Using a model, the polymer pressure in the pultrusion dies was computed to reach up to 1.0, 10.42, and 20.6 MPa for the respective pultrusion pulling speed. The morphology’s characterization showed a reconfiguration of the pultrudate at a higher pulling speed leading to larger unimpregnated agglomerations and polymer rich areas. The highest tensile strength achieved was 233.4 ± 1.5 MPa at 50 mm/min and dropped by around 20% when the pulling speed was raised to 1000 mm/min. The pultrudate reconfiguration at higher speed, attributed to delayed melting of PP fibers in underfilled dies, is deemed responsible for the loss in impregnation and mechanical properties.

Keywords

Pultrusion high-speed process natural fibers thermoplastic

Introduction

Talc, glass fibers and mica are all examples of inorganic fibers and fillers that are frequently used to reinforce the thermoplastic polymers in the in the automotive industry.¹ However, the manufacturing and discarding of these materials have a hostile impact on the environment. Due to growing environmental awareness, new environmental regulations were announced to help decrease the issue of excessive inorganic waste.² These new initiatives are putting pressure on the auto businesses to substitute inorganic fillers with biodegradable alternatives. In addition, fuel consumption can be decreased by around 8% when a 10% weight reduction is achieved as lighter matters require less energy to get transported compared to heavier vehicles.³ Thus, lighter materials offer improved vehicle fuel efficiency as well as lessening greenhouse gas emissions. In response, several carmakers are using larger amounts of natural fiber-reinforced composite materials. Similarly, several other industries such as medical, pharmaceutical and food packaging are discovering more ways to use natural fiber-reinforced composite materials for both structural and non-structural applications.⁴

The use of flax fibers is becoming popular in industrial applications as a result of their low environmental impact as well as their light weight. Compared to other natural fibers, flax fibers are composed of relatively high cellulose content (65-83%), moderate hemicellulose content (11-23%), and low lignin content (0.9-6%).⁵ Typically, cellulose decomposition occurs at temperatures above 300°C while hemicellulose decomposition at temperature above 220°C. Lignin decomposition, on the contrary, occurs around 150°C.⁶ The high cellulose content makes flax more resistant to the thermal degradation relative to other natural fibers. Additionally, flax is also favoured due to its higher tensile strength attributed to its high cellulose content. In fact, it is said that flax fibers have the highest Young’s modulus among all natural fibers.⁷ Thus, the use of flax-reinforced composite materials is beneficial for sustainable industrial applications.

In comparison to thermoset polymers, thermoplastic polymers have very high melt viscosity. This high viscosity makes it difficult to impregnate reinforcement fibers and mould the final product in one-step processes.⁸ Consequently, an additional step such as bulk charge or pre-impregnated pellets is needed before the final production, which is usually done using compression or injection molding process. Fibre-reinforced thermoplastic composite materials are frequently categorized based on their fiber reinforcement lengths. For example, long fibre-reinforced thermoplastic (LFT) are composite materials which have thermoplastic polymer and discontinuous reinforcement fibres having a length-to-diameter aspect ratio longer than the critical aspect ratio.⁸ LFT pellets are frequently produced using a thermoplastic composites (TPC) pultrusion process.^9,10 Pultrusion has drawn recent interest from academics after overcoming persistent impregnation and cooling problems.^11–14 Through the TPC pultrusion process, strand precursors that contain the reinforcement and polymer fibers are wound into bobbins that are positioned in a creel. The reinforcement and polymer fibers are then dragged together via a cascade of dies that are heated in order to liquefy the polymer fibers. The yarns which enter the dies contain excess of polymer fibers relative to the die’s volume. The excess polymer, therefore, creates a backflow that helps raise the impregnation pressure. During pultrusion, damage to the reinforcement fibers can be avoided. As a result, the fibers stay continuous after they get processed. Angelov et al.¹⁵ produced Flax/Polypropylene (Flax/PP) pultrudates at different pulling speeds, heating-die temperatures, and fiber volume fractions. The Charpy test fractured coupons images showed pultrudates having dry surface finish and appeared unimpregnated in the fractured areas. The tensile and shear tests were not conducted on these Flax/PP pultrudates. Oswald¹⁶ produced Flax/PolyLactic-Acid (Flax/PLA) rods having a fiber volume fraction around 50% using a multi-die system. Parameters such as pulling speeds and heating-die temperatures were varied. The lowest void content achieved was 2.5% at a pulling speed of 50 mm/min. The pultrudates produced at that speed had a tensile strength of 300 MPa and shear strengths of 20 MPa. Linganiso et al.¹⁷ produced Flax/PLA rods having a volume fiber fraction of 40%. The highest tensile strength achieved was around 75 MPa and belonged to the samples that were produced at 500 mm/min.

When pultruding at high pulling speeds, attaining low void contents in the pultrudates can be challenging since there appears to be a trade-off between increasing the pulling speed and decreasing the void contents.¹⁸ This also can be attributed to the high viscosities of the thermoplastic polymers. Lapointe et al.¹⁸ observed a significant rise in void contents in Carbon/Polyetheretherketone (C/PEEK) pultrudates after the TPC pultrusion’s pulling speed was increased from 50 to 100 mm/min. In fact, the void contents increased by more than 85%. Wiedmer et al.¹⁹ pultruded Carbon/Polymide-12 (C/PA-12) and tested the pultrusion process at different pulling speeds. The void content increased by around 300% after the pulling speed increased from 100 to 400 mm/min. Ghaedsharaf et al.²⁰ pultruded Carbon/Polyetherimide (C/PEI) of commingled braided yarns. The void content surged after the pulling speed was raised from 50 to 150 mm/min. This finding implies that the void rise was likely due to the polymer flow time which became too short to fully infiltrate through the carbon fiber agglomerations after the pulling speed was raised to 150 mm/min.

Interlacing is a process that periodically interlocks fibers by blowing aligned fiber bundles using compressed air. Saito et al.²¹ used Glass-reinforced-polyamide-66 (Glass/PA-66) braided yarns and the interlacing process was applied to some points of the braided yarn. Thus, there were still non-interlaced regions within the yarn. Interlacing was found to reduce the likelihood resin-rich areas. However, the pultrusion speed was limited to 150 mm/min. Linganiso et al.¹⁷ manufactured Flax/PLA rectangular “sandwich” beams using die temperature up to 290°C and speed of 400 mm/min without reporting flax degradation. However, they reported insufficient bonding between fibre and polymer, as well as insufficient polymer penetration in some areas. Bechtold et al.²² used pultrusion to process Glass/PP commingled braids. Contact heating was tested to increase the pulling speed while maintaining a low void content. Yet, frictional forces were increased, and this resulted in reducing the pulling speed. They suggested that the reason of the high pulling forces was related to the die length which was too long. In order to increase the impregnation level and pulling speed while maintaining the pulling friction low, they recommended replacing the long die with a cascade of shorter dies.

Babeau et al.²³ modeled the heat transfer and polymer flow during a TPC pultrusion process. Their model predicted that the time of residency required for impregnation is inversely proportional to the speed. In other words, when the pulling speed increases, a balance between the increase of pressure and the decrease of the residency time occurs; therefore, the impregnation quality would stay the same as long as the thermal conditions remain the same. Kim et al.²⁴ proposed a model to predict the pressure inside the heating die as well as impregnation quality in a TPC pultrusion process. This model was subdivided into two polymer flow sub-models, the axial backflow around yarns and the radial impregnation flow sub-models. They were both based on Darcy’s law and each sub-model had a different permeability. The axial backflow sub-model was used to predict the pressure along the die longitudinal axis by modeling the polymer’s backflow induced by the excess polymer. On the other hand, the radial impregnation sub-model was utilized to determine impregnation polymer flow through the fibers’ agglomerations. The models were verified and tested experimentally, and they were proven to be accurate in determining the impregnation quality in TPC pultrusion. Babeau et al.²³ and Kim et al.²⁴ models both showed that the pressure-speed relationship was linear. Later, Lessard et al.²⁵ proposed a model, based on Kim et al.’s²⁴ approach, to predict the pulling force and degree of impregnation of a multi-die pultrusion system. They used Carbon/Polyetherimide (C/PEI) commingled fibers for their validation experiments. In their model, they assumed that backflow did not completely fill the die, thus creating an effective die length shorter than the actual die length. The accuracy of the model was proven to be adequate such that the predicted pulling forces and degree of impregnation agreed with experimental data for low pulling speeds. However, at higher pulling speeds and higher taper angles, their model did not match the experimental data due to non-isothermal conditions.

During a typical TPC process, the preheating stage and dies are heated according to a set process temperature. However, the actual temperature reached by the pultrudate depends on the apparatus’ geometry and residency time. This can create significant discrepancies between the set and the actual material temperatures especially when the pulling speed is high. Alsinani et al.²⁶ pultruded Flax/PP rods at four distant speed between 50 to 1000 mm/min. The pultruded rods were then cut into pellets for subsequent injection molding. The pulling speeds of 50 and 250 mm/min led to void content increase up to 8% and the rods resulted in well consolidated pellets. Flax/PP rods produced at 500 mm/min and 1000 mm/min led to void contents above 15% as the polymer did not fully melt during the pultrusion process. While a follow-on thermocouple was not used to confirm if the polymer had the same thermal history at all pulling speeds, the microscopic images showed that middle region of the pultrudates produced at 500 mm/min and above did not melt. Thus, at high pulling speeds, it was concluded that the polymer temperature did not reach the melting point and had a different thermal history compared to the low-pulling-speed experiments. Conversely, Babeau et al.²³ adjusted their pultrusion line set temperatures to obtain the same thermal conditions irrespective of the speed. They confirmed the consistency of the thermal conditions at different speeds; namely 100, 500 and 1000 mm/min, using a follow-on thermocouple inserted with the pultruded materials. They observed that an increase in pulling speed caused an increase in pressure that induced local compaction of the tows. This compaction decreased the tows’ permeability since individual reinforcement fibers were closer to each other in the fiber agglomerations. These observations indicate that the reconfiguration of the fiber agglomerations during pultrusion must be considered to achieve proper consolidation.

The objective of this study is to verify if the speed-induced pressure can offset the decrease in residency time that is required for impregnation quality and observe possible pultrudate morphology’s reconfiguration. Flax/PP precursors were pultruded at 50, 500 and 1000 mm/min. To ensure consistency in the thermal conditions at different pulling speeds, the dies’ temperatures were adjusted to obtain similar pultrusion thermal conditions at every pulling speed. The pressure in the pultrusion dies was estimated using an approach developed by Lessard et al.²⁵ The pressure estimations were used to support the pultrudates’ cross-section morphology, tensile and shear strength variations for different pultrusion pulling speeds.

Experimental

Materials

Pure PP yarns (Normal tenacity PP multifilament yarn, Guangzhou Chemical Fiber Co.) in the form multifilament fiber yarns of 100 Tex size were used for the pultrusion. The glass transition temperature of PP was in the range of −18 to −13°C.²⁷ The melting point of the PP was 166.5 $℃$ as obtained by DSC at heating rate of 10 $° C / \min$ . Details about the DSC procedure can be found in Alsinani et al.²⁶ The PP fiber’s density was considered to be 0.91 g/cm³.²⁸ After melting, the PP density was considered to decrease to 0.75 g/cm³.²⁹ Flax yarns (Flax Low Twist Roving, Safilin) having a 200 Tex size and a solid density of 1.53 g/cm³ were used.

Pultrusion experiments

Figure 1 presents the schematic for the pultrusion system used in this study. The pultrusion apparatus contained a creel, a guide, a preheater, seven heating dies, a cooling die, a load-cell and a pulling system. The creel allowed for individual adjustment for each of the bobbins’ tensions. One Newton tension was applied to each bobbin that contained flax while no tension was applied to the bobbins that contained the PP fibers. The 1 m-long preheater had a square cross-section of 300 mm in width and was equipped with three Hot-Air Blowers (HAB, VT-752C, Varitemp Heat Gun) installed in sequence. The HAB were all manually controlled, and their temperature was measured using thermocouples at HAB exit. All seven heating dies had a 31.75 mm-long conical cavity followed by a straight 6.35 mm cylindrical cavity (total length of 38.1 mm). The dies’ conical opening angle ( $θ_{D i e}$ ) was $10 °$ , that is, $5 °$ relative to the longitudinal axis. The diameter of each die’s cylindrical exit ( $ϕ_{D i e}$ ) was 7.91, 7.01, 6.40, 5.13, 5.00, 4.90 and 4.76 mm respectively. The white triangles in Figure 1 denote the thermocouple’s locations in the heating dies. Additionally, the dark grey circles represent the location of the heating cartridges whereas the light grey circles represent the location of the air-cooling channels. A thin-walled tube protruded from the last pultrusion die toward the cooling die. Details about the cooling system can be found in Alsinani et al.¹² The pultrusion pulling speed was controlled by a puller throughout the process. The precursor included 97 PP yarns and 68 flax yarns distributed into several bobbins: 24 bobbins contained four PP yarns, one bobbin contained one PP yarn and 17 bobbins contained four flax yarns. To ensure proper mixing in the creel, every flax bobbin was placed beside a PP bobbin. The filling ratios ( $R_{f}$ ) is calculated as

R_{f} = \frac{A_{i n}}{A_{D i e}}

(1)

Where

A_{i n}

is the material area entering the die and

A_{D i e}

is the die exit area. The nominal

R_{f}

in the first die uses the flax and PP material areas pulled from the creel. These areas are obtained by dividing the flax and PP yarn sizes (in Tex, or g/1000m) by their corresponding densities (in g/cc). The melt density was considered for the PP material area. The other dies nominal

R_{f}

were calculated using the same flax area as in the first die. For the PP area, in case of under filling (

R_{f} < 100 %

), all the PP is assumed to follow the pultrudate. In case of over filling (

R_{f} > 100 %

), the PP area was considered to be the non-flax area from the previous die exit area. There were three experiments in total. The pultrudate’s final flax fiber volume content was 50%, while the pulling speeds and set temperatures varied. Table 1 shows the filling ratios, pulling speeds and set temperatures that were used to produce the Flax/PP pultrudates. The sample’s name follows the following format: Flax-<Puling speed>. For example, Flax-50 referrers to the pultrudates produced at 50 mm/min. In order to avoid premature melting which causes polymer dripping inside the preheater, the preheater’s HAB temperatures were adjusted such that the temperature would be less than melting temperature of PP before entering the first heating die.³⁰ In addition, both preheater’s and heating die’s temperatures were adjusted so the pultrudates reached and maintained 200°C from the fourth die to the seventh die.

Figure 1.

A schematic of the multi-die pultrusion system used in this study. The pultrusion system included a creel, a pre-heater, seven pultrusion dies with a 10° taper angle and reducing exit diameter, a thin-walled tube, a cooling die, a load-cell and a pulling system.

Table 1.

Die filling ratios and set temperatures of the preheater’s three Hot-Air Blowers (HAB) and seven dies for the three pultrusion speed experiments.

HAB/Die number		HAB 1	HAB 2	HAB 3	Die 1	Die 2	Die 3	Die 4	Die 5	Die 6	Die 7
Filling ratio		N/A	N/A	N/A	44%	57%	68%	106%	105%	104%	106%
Set temp. ( $℃$ )	Flax-50	50	100	100	200	200	200	200	200	200	200
	Flax-500	100	120	120	240	240	240	240	240	240	240
	Flax-1000	100	150	250	265	265	265	265	265	265	265

Characterization

Cylindrical samples were made using a precision saw. The cylindrical samples were polished and analyzed under a microscope (Metallovert, Leitz). Photos of 200 X magnification were taken and then stitched to form the whole cross-section of pultrudate. The void content was measured according to ASTM2734-09 Method C. The shear tests were conducted according to ASTM D3914-02 and the tensile tests were conducted according to ASTM D3916-08. More details on the testing procedures can be found in Oswald.¹⁶

Modeling

The model to determine the overfill-induced pressure and forces is similar to the approach presented in Lessard et al.²⁵ This model is an implementation of Kim et al.’s pultrusion model for a multi-die pultrusion. The model is structured so that in each die, the axial backflow and the radial impregnation flow were solved by two separate sub-models. The model starts by taking the inputs of material properties. Then, the first die geometry is defined by its length and cross-sectional area in function of length. The modelling work is based on Darcy’s law which describes the flow of a single-phase fluid in a porous media. This law has been widely used by academics for composite material impregnation for many decades.³¹ Both flows are solved using Darcy’s equation of flow in a porous media with a specific media permeability. The permeability is modeled using Kozeny-Carman’s equation with specific Kozeny constant for the axial backflow and the radial impregnation flow. Euler’s method was employed to determine the pressure profile from the axial backflow sub-model. In this study, only the last four heated dies, i.e. Dies 4 to 7, were considered since the first three dies were underfilled with the pultrudate (see Table 1). Similar to Lessard et al.,²⁵ the backflow polymer was assumed to partly occupy the die and thus creating an effective die length. Table 2 shows the parameters used to obtain the pressure distribution in the last four dies. The initial fiber volume fraction (

v_{f, i})

value of 0.45 corresponds to the initial fiber volume fraction at the fourth heating dies’ inlet. The final fiber volume fraction (

v_{f, \infty})

of 0.5 corresponds to the pultrudate’s final volume fraction at the seventh die’s outlet. The radial impregnation Kozney (

K_{r})

and axial backflow Kozney (

K_{z})

constants are 12 and 1.2 respectively. They are fitting constants and were taken from Kim et al.²⁴ since the model in this study is similar to theirs. The model assumes that the flax yarn radius at die inlet,

R_{0}

, is calculated according to

R_{0} = \frac{\sqrt{\frac{N_{Y} f}{ρ}}}{π α}

(2)

Where

ρ

is the flax density and

α

is the yarn compaction factor. In this study,

α

is assumed to be 60%. The values of flax density

ρ

and the flax yarn sizes

f

can be found in the experimental section.

N_{Y}

is the number of yarns bobbined together. Using (Equation 2),

R_{0}

is set at 527 µm since the flax bobbins contained four yarns. The radius of the flax fiber for radial permeability was 7 µm.⁵ As hypothesized by Babeau et al.,²³ Kim et al.,²⁴ and Lessard et al.,²⁵ the radial impregnation flow is modelled using a constant size agglomeration throughout the pultrusion. Only the axial backflow considers that the inter-agglomeration distance is increasing with increasing

V_{f}

. It is important note that melted PP exhibits a non-Newtonian fluid behavior,³² where its viscosity decreases with increasing shear rate. Therefore, the melted PP viscosity during pultrusion is greatly affected by the pulling speed changes. It is possible to estimate the shear rate between the die surface. However, the micro-flow shear rate remains unknow and is expected be much lower than the shear rate near the die surface. By assuming that the pulling speed is the only factor affecting shear rate, the micro-flow viscosity at a lower shear rate would be underestimated. In this study, the polymer’s viscosity was assumed to be Newtonian, i.e. constant irrespective of the pulling speed and was calculated to be 637 Pa.s according to Kim et al.²⁴ Since the pultrusion was modeled for the isothermal Dies 4 to 7, the impact of temperature on the viscosity was neglected. More details about the pultrusion modeling approach used herein can be found in Lessard et al.²⁵ The effective die length (

L_{e f f, i}

) at each die was calculated as follows

L_{e f f, i} = \frac{ϕ_{D i e, i - 1} - ϕ_{D i e, i}}{2 \tan (\frac{θ_{D i e, i}}{2})}

(3)

Table 2.

Models’ constants and parameters.

Property	Value
Die number	Die 4	Die 5	Die 6	Die 7
Initial fiber volume fraction $v_{f, i}$	0.28	0.43	0.45	0.47
Final fiber volume fraction $v_{f, \infty}$	0.43	0.45	0.47	0.50
Radial impregnation Kozney constant, $K_{r}$ ²⁴	12
Backflow Kozney constant, $K_{z}$ ²⁴	1.2
Agglomeration radius, $R_{0}$ ( $μ m)$	527
Flax fiber radius $(μ m)$ ⁵	7
Polymer viscosity (Pa.s) ²⁴	637

Figure 2 shows the computed pressure evolution in the tapered region of the last four dies, which are the only dies that had overfill for the three pultrusion speeds. As seen in Figure 2, the effective die length was calculated to be 7.26, 0.74, 0.57 mm and 0.80 for Dies 4 to 7 respectively. Despite having a larger exit diameter, the pressure in Die 4 reached higher pressures compared to Dies 5 and 6, for each pulling speed. This can be attributed to the longer effective die length of Die 4 compared to Dies 5 and 6. Die 7 resulted in the highest pressure. From Dies 5 to 7, the effective die length values are similar when compared to the effective die length of Die 4. For a given pultrusion speed, it is seen that progressive outlet diameter reduction from Dies 5 to 7 resulted in progressive maximal pressure increase. Finally, it is seen that faster pulling speeds made the computed maximal pressure in the seventh die to increase. The polymer pressure in the pultrusion dies was computed to reach only 1.0 MPa when the pulling speed was 50 mm/min. This pressure raised to 10.4 and 20.6 MPa when the speed was increased to 500 and 1000 mm/min, respectively. The pulling force was considered here to be the force opposite to the polymer pressure application onto the filled dies’ surface ( $S_{D i e, i}$ ), projected onto the dies’ axisymmetric axis. Thus, the pulling forces generated in the last four dies were summed according to

F = \sum_{D i e 4}^{D i e 7} \int_{S_{D i e, i}} P d S \cdot \sin (\frac{θ_{D i e, i}}{2})

(4)

Figure 2.

Computed pressure evolution in the tapered region of the Dies 4 to 7. Higher effective die length and smaller exit diameters made the pressure go higher.

The solid line in Figure 3 shows the computed force using (Equation (4)). The computed pultrusion force values are 2.4 N, 23.6 N, and 47.2 N for 50 mm/min, 500 mm/min, and 1000 mm/min, respectively. Moreover, the pulling force shows a linear increase with respect to pulling speed.

Figure 3.

A comparison between measured pultrusion force and computed pultrusion force in Dies 4 to 7. The measured pultrusion force is higher and it’s slope with respect to pultrusion speed steeper than the computed one. The measured force slope changes before and after the 500 mm/min point, while the computed one is constant.

Results and discussions

Pultrusion experiments’ observation

During all the three experiments, there was no incident of fiber breakage. It is also important to note that the pulling speed was gradually raised especially during the high pulling speeds experiments. This reduced the probability of having fiber breakage. Figure 4 (a), (b) and (c) show the temperature measurements taken during Flax-50, Flax-500 and Flax-1000 experiments respectively. Before entering Die 1, the temperature in all cases was well below the melting point. The highest temperature reached was 133, 108 and 96°C for Flax-50, Flax-500, Flax-1000 respectively. This implies that the polymer did not melt before entering Die 1. Furthermore, the curves demonstrate that the pultrudates reached the processing temperature, i. e. 200°C by the fourth die, for all speeds. Table 3 shows the measured temperatures during the pultrusion experiments. The average temperatures in the last four dies were 201.3 ± 1.3, 199.8 ± 2.1, 201.2 ± 2.6°C for Flax-50, Flax-500, and Flax-1000 respectively. At the higher pulling speed experiments, the set temperatures were 245 for Flax-500, and 265°C for Flax-1000 (see Table 1). The temperature stopped raising towards the set temperatures once the precursor entered the fourth die. The temperature stabilization is attributed to the overfilled state of the pultrudate beginning at the fourth die. The thermocouples were placed in the middle of precursors. Once it entered the last four dies, it became encapsulated in the low thermally conductive Flax/PP composite. The first three dies worked as preheaters with filling ratios below 100%. Since the pultrudate was unconsolidated, it is supposed that heating could reach the separated fibers. The temperature gradient between the die surface and the pultrudate’s core, where the thermocouple was located, was not characterized due to technical difficulties. The pultrudate temperature was considered constant along the radial direction. Future work should consider the temperature gradient in the radial direction at high pulling and study its influence on impregnation and pulling force.

Figure 4.

The temprature measurement during a typical: (a) Flax-50. (b) Flax-500. (c) Flax-1000. The process temperature is reached at Die 2 for Flax-50 while being reached at Die 4 for Flax-500 and Flax-1000. All three experiments reached 200°C process temperature in Dies 4 to 7.

Table 3.

Measured temperatures during the pultrusion experiments.

Sample	Temperature ( $℃$ )
Flax-50	201.3 ± 1.3
Flax-500	199.8 ± 2.1
Flax-1000	201.2 ± 2.6

Flax-50 pultrudate spent 3.2 min in the last four dies while Flax-500 spent 0.32 min and Flax-1000 spent 0.16 min. Figure 4 also indicates the exit temperature (T_out) of the pultrudate for the different pultrusion speed. Flax-50’s T_out was 47.6°C, Flax-500s was 101.4°C and Flax-1000s was 156.8°C. The increasing pultrudate’s T_out with respect to pultrusion speed is attributed to improper adjustment of the cooling die temperature. The compressed air cooling flowing through the cooling channels was insufficient to keep the same cooling conditions irrespective of pultrudate speed. Water cooling could be used to improve the cooling efficiency. Figure 3 also shows the measured pultrusion force for all experiments compared to the computed force using equation (3). Flax-50 resulted in 47.1 ± 0.1 N pulling forces while Flax-500 and Flax-1000 resulted in 160.2 ± 0.5 N and 266.5 ± 1.1 N respectively. The measured forces are one order of magnitude higher than the computed forces at their corresponding speed. The primary explanation is related to the fact that the model neglected the fiber-to-die-wall friction forces in all pultrusion and cooling dies. Friction forces are independent of relative speed between friction surfaces. However, in Figure 3, the measured pulling force differences with the computed force is seen to increase with increasing pulling speed. At 50 mm/min, the measured pulling force is 44 N higher than computed pulling force, while at 1000 mm/min, it was 217 N higher than the computed forces. According to Figure 4, the polymer temperature rise was steeper at 50 mm/min than the other speeds. It is suspected that the polymer melted earlier at 50 mm/min than at the faster experiments. Therefore, the polymer acted as a lubricant that lowered the friction forces. Moreover, at higher speed, the unmelted polymer fibers in the first dies exerted higher pressure on the die surfaces upon compaction. Finally, it suspected that the cooling section also contributed to the increasing measured-to-computed difference in pulling force with respect to speed. As seen in Alsinani et al.,¹² the polymer adheres to the die surface when cooling from the processing temperatures until it shrinks and separates. The adhesion length is directly related to the distance on which the polymer cools. At higher speed, the pultrudate adhesion length with cooling die’s surface became higher. This likely added more frictional forces opposing the pulling action. The significantly higher actual pultrusion force at 1000 mm/min is an indication that the model cannot capture all phenomena occurring during pultrusion.

The increase in pultrusion force indicates higher impregnation pressure with increasing pulling speed. However, contrary to the computed pultrusion forces, the increase in measured force does not follow a linear relationship with speed. The apparent force increase rate was 0.25 N/(mm/min) from 50 mm/min to 500 mm/min, and 0.21 N/(mm/min) from 500 mm/min to 1000 mm/min. In comparison, the computed force in Dies 4 to 7 was linear at an increase rate of 0.06 N/(mm/min). The non-linear increase observed experimentally is contradictory with behaviors predicted by Kim et al.,²⁴ Babeau et al. ²³ as well as the modeling done in this study. Future improvements to pultrusion models should address this contradiction. For example, the polymer viscosity should be considered non-Newtonian. Moreover, the viscosity should be modeled differently for the highly sheared polymer at die wall and the lower speed micro-flow within the fiber bundles.

Morphological analysis of the pultrudates

Table 4 shows pultruded rods’ measured porosities. The void content characterization indicated that the 50 mm/min pultrusion speed led to a 2.5 ± 0.9% void content, the 500 mm/min to a 2.6 ± 0.7% void content and that 1000 mm/min led to the highest void content of 5.4 ± 0.4%. Figure 5 presents the typical cross-section microscopic photos of Flax/PP pultrudates with respect to the pulling speeds. Contrary to Alsinani et al.,²⁶ the polymer in all cases had fully melted. The pultrudates’ cross-section is round in all cases. The pultudate’s diameter was measured to be 4.66 ± 0.04 mm, 4.70 ± 0.07 mm, and 4.76 ± 0.10 mm for Flax-50, Flax-500, and Flax-1000 respectively. The similar pultrudate diameters, irrespective of the pulling speed, is a clear indication that deconsolidation did not occur at cooling die exit. The absence of deconsolidation shows that the polymer reached sufficient stiffness at cooling die exit to counteract the compacted yarns’ spring-back action.¹² Since in all cases,

T_{o u t}

is well above the glass transition temperature of polypropylene (below 0°C), it is suspected that the polypropylene reached a sufficient crystallization content for all cooling conditions. This sufficient crystallization hypothesis was confirmed in Alsinani et al.¹² where carbon fiber and semi-crystalline polyetheretherketone pultrudates did not deconsolidate when exiting the cooling die between the glass transition temperature and the melting point. Figure 6 shows a PP build-up at the heating die cavity’s inlet after a Flax/PP pultrusion. Build-ups were seen after all experiments, whatever the selected pultrusion speed. These build-ups are occurring since the filling ratio of the dies was greater than 100% in Dies 4 to 7 (see Table 1). The excess PP volume of the pultrudate was flushed at the die entrance due to backflow. The quantity of PP polymer in these backflow build-ups was not measured quantitatively. However, it was observed that the size of PP build-ups were larger for pultrusion experiments that were done at higher pulling speeds. The larger PP build-up explains the lack of polymer in the pultrusion experiments at higher pulling speeds which resulted in higher void contents.

Table 4.

Tensile and shear tests results.

Sample	Void content (%)	Shear strength (MPa)	Tensile strength (MPa)
Flax-50	2.5 ± 0.9	12.7 ± 1.4	233.4 ± 1.5
Flax-500	2.6 ± 0.7	3.7 ± 0.7	193.2 ± 1.3
Flax-1000	5.4 ± 0.4	3.2 ± 1.2	189.3 ± 4.5

Figure 5.

The microscopic photos of flax/PP pultrudates. (a) Flax-50. (b) Flax-500. (c) Flax-1000. Flax-500 and Flax-1000 are showing large polymer rich areas.

Figure 6.

PP build-up at the heating die cavity’s inlet in a typical Flax/PP pultrusion process.

Flax-50 microscopic image shows well distributed and well impregnated fibers. Figure 5 (b) and c) shows pultrudates produced at higher pulling speeds. Large polymer rich areas are seen. It seems that the increase in pulling forces, induced by higher pulling speeds, led to an increase in the fiber agglomeration compaction. The agglomeration compaction was not considered in the model. This observation indicates that it must be included to improve model accuracy. Also, the pultrudates produced at high pulling speed experiments were reconfigured in large groups of yarns. Therefore, the effective agglomeration radius to be impregnated was larger at higher speed than at lower speed. A plausible explanation for this rearrangement can be deduced by analysing the temperature reading of the follow-on thermocouple in Figure 4. The temperature rise in the first three dies show that the maximum temperature is reached much faster in Flax-50 than the Flax-1000 experiment. This created a delay in polymer melting at higher speed. Moreover, as observed in Nawaf et al. ²⁶ the PP polymer melts progressively from the die surface to the center of the pultrudate. Therefore, since un-melted polymer fibers cannot flow around flax agglomerations, the area they occupied effectively compressed and pushed the flax yarns into the melted PP in contact with the die. Then, the PP fiber rich areas finally melted before entering the fourth die leading to polymer rich regions seen in Figure 5 (b) and (c). The above explanation behind the fiber reconfiguration will be investigated thoroughly in future studies. In any case, it can be affirmed that a co-wound yarn precursor structure is more likely to result in such a rearrangement since the unimpregnated fiber agglomeration movement relative to each other is not secured by any mechanical feature once the polymer is melted. Future work must develop a precursor structure for high speed pultrusion that is structured in a way that prevents polymer rich creation. Another possible solution would be to use preheating dies with low or no taper angle. These preheating dies would bring the polymer fibers to melt without compression, while securing them to prevent their separation from the pultrudate at die entrance.

Mechanical properties of pultrudates

Figure 7 (a) presents a pultruded sample prepared for shear test; (b) presents a typical shear stress versus displacement result for pultrudates produced at 50, 500, 1000 mm/min. The 50 mm/min curve exhibited the highest peak value which was 12.7 ± 1.4 MPa. The pultrudates, produced at higher speeds, achieved much lower shear strengths of 3.7 ± 0.7 and 3.2 ± 1.2 for Flax-500 and Flax-1000 respectively. The shear strength variation between Flax-500 and Flax-50 can be calculated by normalizing their difference with respect to Flax-50 result. The variation is −71%, while measured void contents for the same experiments do not show a significant difference when considering the standard deviation. This −71% variation in shear strength is attributed to the apparition of large unreinforced resin-rich regions in the Flax-500 pultrudates, as seen in Figure 5(b). The large resin rich regions are mostly located in the center of the pultrudate, where the larger shear stress is occurring when testing per ASTM D3914-02. When comparing Flax-1000 to Flax-50, the shear strength variation is −75%. In this case, the void content difference is significant, and the void content of the Flax-1000 is approximately twice the Flax-50 value. The other contributing factors could be the lower impregnation of the flax fiber bundles combined with large voids in the resin rich regions, as seen in Figure 5(c). Table 4 presents the tensile and shear strength values. Following the same trend, the highest tensile strength value was achieved for Flax-50 and was 233.4 ± 1.5 MPa. The higher speed pultrudates achieved much lower tensile strengths: 193.2 ± 1.3 and 189.3 ± 4.5 MPa for 500 and 1000 mm/min respectively. These values are significantly lower than the Flax-50. The lower tensile and shear strengths can be attributed the polymer rich regions and bad impregnation quality as reported in the Results and Discussions Section.

Figure 7.

(a) Pultruded sample prepared for shear test. (b) A typical Shear stress-displacement result of shear test for 50, 500, 1000 mm/min. The 50 mm/min sample is showing a higher shear strength value compared to the other samples.

Conclusions

Three different pulling speeds were selected to study the effect of high pulling speeds on the morphologies of pultrudates in a TPC pultrusion process. To ensure consistency in the thermal conditions at different pulling speeds, the dies’ temperatures were adjusted to obtain similar pultrusion thermal conditions at every pulling speed. The cross-section morphology characterization showed well-distributed reinforcement fibers for the low pulling speed experiment of 50 mm/min. On the contrary, the pultrudates that were produced at high pulling speed had compacted and clustered flax yarns, which resulted in polymer rich areas and more voids. Also, during Flax-500 and Flax-1000 experiments and when the polymer fibers melted, the melted polymer was not able impregnate the reinforcement fibers since they were more compacted when compared to Flax-50. It is important to note that the pulling forces during Flax-500 and Flax-1000 experiments were significantly higher than Flax-50. At every speed, the measured pulling forces were higher than the computed value. The significantly higher actual pultrusion force at every speed is an indication that the model cannot capture all phenomena occurring during pultrusion; it is also indicative that more realistic modeling approaches must be developed. The microscopic images showed well distrusted yarns for pultrudates produced at 50 mm/min. While at higher pulling speeds, reconfiguration of the reinforcement fibers occurred. The pultrudate reconfiguration at higher speed, attributed to delayed melting of PP fibers in underfilled dies, is deemed responsible for the loss in impregnation. The reconfiguration was confirmed by mechanical property characterization. Tensile properties deteriorated when a faster pulling speed was used. The highest tensile strength achieved was 233.4 ± 1.5 MPa at 50 mm/min and dropped by around 20% when the pulling speed was raised to 1000 mm/min. The shear strength also dropped by 75% when the pulling speed was raised from 50 to 1000 mm/min. Future work should find mitigation strategies to counter the reconfiguration mechanism. For example, the use of the interlacing technique could be investigated. Also, using low-angle Flax/PP braids might help secure the position of the flax yarns in the pultrudate, while still allowing fiber separation in LFT pellets when they are being processed during injection molding.

Footnotes

Acknowledgements

The authors would like to thank Danielle Szydlowski for her experimental help. We would like also to thank Bombardier Aerospace, Pultrusion Technique.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Saudi Arabia’s Saline Water Conversion Corporation and Ministry of Education (KSP12014087), NSERC (CRDPJ488387-15) and Prima Quebec (R10-009).

Data availability statement

The raw/processed data required to reproduce these findings cannot be shared at this time due to technical or time limitations

ORCID iD

Louis Laberge Lebel

References

Fogorasi

Barbu

. The potential of natural fibres for automotive sector-reviewIOP Conference Series: Materials Science and Engineering. IOP Publishing 2017; 252(1): 012044.

Zhu

Njuguna

, et al. Recent development of flax fibres and their reinforced composites based on different polymeric matrices. Materials 2013; 6(11): 5171–5198.

Cheah

. Cars on a Diet: The Material and Energy Impacts of Passenger Vehicle Weight Reduction in the US. Massachusetts Institute of Technology, 2010.

Ngo

. Natural fibers for sustainable bio-composites. In: Natural and Artificial Fiber-Reinforced Composites as Renewable Sources. IntechOpen, 2017.

Yan

Chouw

Jayaraman

. Flax fibre and its composites–A review. Composites Part B: Eng 2014; 56: 296–317.

Yang

Yan

Hangping

Dong

Chuguang

. Characteristics of hemicellulose, cellulose and lignin pyrolysis. Fuel 2007; 86: 1781–1788.

Van de Velde

Kiekens

. Thermoplastic pultrusion of natural fibre reinforced composites. Compos Structures 2001; 54(2–3): 355–360.

Ning

Hassen

, et al. A review of Long fibre thermoplastic (LFT) composites. Int Mater Rev 2020; 65(3): 164–188.

Vaidya

Thattaiparthasarthy

Pillay

, et al. Colored inorganic-pigmented long-fiber thermoplastics. J Thermoplastic Compos Mater 2015; 28(12): 1627–1643.

10.

Bartus

Vaidya

Ulven

. Design and development of a long fiber thermoplastic bus seat. J Thermoplastic Compos Mater 2006; 19(2): 131–154.

11.

Carlsson

Åström

. Experimental investigation of pultrusion of glass fibre reinforced polypropylene composites. Composites Part A: Appl Sci Manufacturing 1998; 29(5–6): 585–593.

12.

Alsinani

Ghaedsharaf

Lebel

. Effect of cooling temperature on deconsolidation and pulling forces in a thermoplastic pultrusion process. Composites Part B: Engineering 2021; 219: 108889.

13.

Ghaedsharaf

Brunel

Lebel

. Multiscale numerical simulation of the forming process of biaxial braids during thermoplastic braid-trusion: Predicting 3D and internal geometry and fiber orientation distribution. Composites Part A: Appl Sci Manufacturing 2021; 150: 106637.

14.

Absi

Alsinani

Laberge Lebel

. Carbon fiber reinforced poly (ether ether ketone) rivets for fastening composite structures. Composite Structures. This is not a book! It is a journal article, 2021, p. 114877.

15.

Angelov

Wiedmer

Evstatiev

, et al. Pultrusion of a flax/polypropylene yarn. Composites A: Appl Sci Manufacturing 2007; 38(5): 1431–1438.

16.

Oswald

. Pultrusion multi-filière assistée par le vide de LIN/PLA. In: Master of Applied Science (M.A.SC), Mechanical Engineering. Montreal (Canada): Ecole Polytechnique, 2016.

17.

Linganiso

Bezerra

Bhat

, et al. Pultrusion of flax/poly (lactic acid) commingled yarns and nonwoven fabrics. J Thermoplastic Compos Mater 2014; 27(11): 1553–1572.

18.

Lapointe

Laberge Lebel

. Fiber damage and impregnation during multi‐die vacuum assisted pultrusion of carbon/PEEK hybrid yarns. Polym Composites 2019; 40(S2): E1015–E1028.

19.

Wiedmer

Manolesos

. An experimental study of the pultrusion of carbon fiber-polyamide 12 yarn. J Thermoplastic Composite Materials 2006; 19(1): 97–112.

20.

Ghaedsharaf

Brunel

Laberge Lebel

. Thermoplastic composite rod manufacturing using biaxial braid-trusion. In: Proceedings of the ECCM18–18th European Conference on Composite Materials. Athens, Greece, 2018, pp. 24–28.

21.

Saito

Nakai

Ohtani

. Effects of fabrication method of composite yarn intermediate material and resin melt viscosity on the impregnation properties and mechanical properties of PA66 fiber and glass fiber composites. In: Journal of Thermoplastic Composite Materials, 0892705720964559, 2020.

22.

Bechtold

Wiedmer

Friedrich

. Pultrusion of thermoplastic composites-new developments and modelling studies. J Thermoplastic Compos Mater 2002; 15(5): 443–465.

23.

Babeau

Comas-Cardona

Binetruy

, et al. Modeling of heat transfer and unsaturated flow in woven fiber reinforcements during direct injection-pultrusion process of thermoplastic composites. Composites Part A: Appl Sci Manufacturing 2015; 77: 310–318. DOI: 10.1016/j.compositesa.2015.04.017

24.

Kim

Lee

Friedrich

. A model for a thermoplastic pultrusion process using commingled yarns. Composites Sci Tech 2001; 61(8): 106–1077.

25.

Lessard

Dubé

Lebel

. A model of multi-die thermoplastic pultrusion. Composite Part B: Engineering, 2022, 110319, In press.

26.

Alsinani

Laberge Lebel

. Effect of Flax/PP pellets’ morphology on the mechanical properties of injection-molded parts. J Thermoplastic Compos Mater 2022; 0(0): 08927057221095394. DOI: 10.1177/08927057221095394

27.

Mileva

Androsch

Zhuravlev

, et al. Homogeneous nucleation and mesophase formation in glassy isotactic polypropylene. Polymer 2012; 53(2): 277–282.

28.

Dai Lam

Hoang

Quang

, et al. Effect of nanosized and surface-modified precipitated calcium carbonate on properties of CaCO3/polypropylene nanocomposites. Mater Sci Eng A 2009; 501(1–2): 87–93.

29.

Barbosa

Kenny

. Processing of short‐fiber reinforced polypropylene. I. Influence of processing conditions on the morphology of extruded filaments. Polym Eng Sci 2000; 40(1): 11–22.

30.

Novo

Nunes

Silva

, et al. Production of thermoplastics matrix preimpregnated materials to manufacture composite pultruded profiles. Ciência & Tecnologia Dos Materiais 2013; 25(2): 85–91. DOI: 10.1016/j.ctmat.2014.03.004.

31.

Sharma

Siginer

. Permeability measurement methods in porous media of fiber reinforced composites, 2010.

32.

Kobayashi

Kitamaru

Tsuji

. The Viscous Flow of Molten Polypropylene (Special Issue on Polymer Chemistry, II). Bull Inst Chem Res Kyoto Univ 1965; 43(2): 179–192.

Effect of high pulling speeds on the morphologies of pultrudates in a thermoplastic pultrusion process

Abstract

Keywords

Introduction

Experimental

Materials

Pultrusion experiments

Characterization

Modeling

Results and discussions

Pultrusion experiments’ observation

Morphological analysis of the pultrudates

Mechanical properties of pultrudates

Conclusions

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

Data availability statement

ORCID iD

References