The effect of process parameters on ultraviolet cured out of die bent pultrusion process

Composites Part B 89 (2016) 9e17

Contents lists available at ScienceDirect

Composites Part B journal homepage: www.elsevier.com/locate/compositesb

The effect of process parameters on ultraviolet cured out of die bent pultrusion process I. Tena a, *, M. Sarrionandia a, J. Torre b, J. Aurrekoetxea a a b

n, Spain Mechanical and Industrial Production Department, Mondragon Unibertsitatea, Loramendi 4, 20500 Mondrago Irurena S.A., Ctra. de Tolosa s/n, 20730 Azpeitia, Spain

a r t i c l e i n f o

a b s t r a c t

Article history: Received 13 August 2015 Received in revised form 29 September 2015 Accepted 4 November 2015 Available online 30 November 2015

In this paper the effect of emitting intensity and pulling speed on ultraviolet (UV) cured out of die bent pultrusion process has been analysed. The curing process of the composite at the exit of the die has been characterised (kinetic model) through the analysis of the evolution of the electrical resistivity (DC sensor) of the material. Combining the studied pulling speeds and emitting intensities, the developed curing model can predict accurately the curing degree at the exit of the die of the profiles. In addition, through the analysis of the final quality of all the manufactured bent profiles in the pultrusion line, the experimental optimum process window has been defined. Indeed, in this study, the optimum process window is limited approximately to the parameters resulting in a curing degree from 5% to 12% at the nonexposed surface (at the exit of the die). © 2015 Elsevier Ltd. All rights reserved.

Keywords: E. Pultrusion E. Cure D. Electron microscopy D. Process monitoring

1. Introduction The pultrusion is a highly automated continuous process for manufacturing structural composite profiles. In conventional thermoset pultrusion process, continuous fibres are impregnated in a resin bath and pulled through a long heated die (traditionally 1 m long at least). Resin cures inside the die, causing high forces of friction along the die wall: 50e150 kN [1]. Traditional thermoset pultrusion process is geometrically limited to straight profiles of constant cross section. Recently, some variants of the traditional pultrusion have succeeded in obtaining curved profiles [2,3]. However, those processes are restricted to constant radii, low productivity rates and high pulling force since the profile continues being cured inside the die. Moreover, complex structures or frames and variable radii cannot be manufactured with those processes. As Britnell and co-workers [1] demonstrated, the restrictions to achieve this aim can be overcome if the profile is cured out of the die. In this new approach the die is only required to define the geometry of the fibre/resin bundle and to remove excess of resin. Thus, the pulling force is much reduced [4]. It is therefore possible to pull the fibres through the die by using a robot arm. By careful

* Corresponding author. Tel.: þ34 943 73 96 44. E-mail address: [email protected] (I. Tena). http://dx.doi.org/10.1016/j.compositesb.2015.11.027 1359-8368/© 2015 Elsevier Ltd. All rights reserved.

control of the robot and the curing conditions, it is also possible to manipulate the fibres so that a structure complete with radii and corners can be produced, without the need of any kind of additional tool. However, the curing of the composite out from the die is not possible using the traditional thermal curing method. Therefore, an alternative fast-curing method is needed [5]. One of those alternative routes is the ultraviolet (UV) curing [6e9]. The UV curing industry, using the energy of UV light in the formation of polymeric materials, has approached the last years, high degree of maturity. The development of monomers, oligomers, and photoinitiators during this time has allowed the technology to advance into very efficient formulations for a wide variety of applications [10]. Resins such as vinylester [7], epoxy [8] and polyester [9], when formulated with a proper photoinitiator, can be cured quickly under exposure to UV light. Hence, it can be stated that the combination of the pultrusion and UV curing can overcome the main limitations of the use of the traditional pultrusion. Literature searches reveal that previous works have been developed in order to apply the benefits offered by ultraviolet curing to the pultrusion process [1,4,11]. The only research work around non-linear out of die pultrusion is the study carried out by Britnell and co-workers [1]. This work demonstrates the capacity of the process to produce non-linear profiles. Anyway, aspects related to the process as the effect of the UV source and resin formulation, die design, force estimation, pulling speed, path design, process

10

I. Tena et al. / Composites Part B 89 (2016) 9e17

simulation and monitoring techniques, final mechanical properties, and so on, are not studied yet. The pulling speed and the UV exposure conditions are directly related to the final curing degree achieved during the out of die UV cured pultrusion process. Moreover, trough-thickness cure is critical at the exit of the die as Tena et al. [4] have demonstrated. In addition, one of the critical issues in pultrusion and in composite manufacturing processes is also directly related to the troughthickness cure distribution: residual stresses and strains. As it has been analysed in some recent studies [12e14], the internal residual stresses and strains have a direct effect on the mechanical performance of the composite part. It has been probed that one of the main mechanisms that generate residual stresses/strains is the non-uniform through-thickness cure and temperature gradients. Hence, it can be stated that the trough-thickness cure is critical for the process. However, during bent pultrusion, the effect of the curing degree at the exit of the die has not been evaluated yet. The undercure or overcure of the composite at the exit of the die could be translated into defective bent profiles. Thus, in this paper the effect of the degree of cure at the exit of the die for UV bent pultrusion process has been studied. The combination of different curing conditions (UV intensity and pulling speed) has been analysed. In addition, the curing process of the composite has been characterised (kinetic model) through the analysis of the evolution of the electrical resistivity [15e21] of the composite at the exit of the die. 2. Experimental 2.1. Materials and light sources The composite used in this study is a glass/UV cured polyester composite. The reinforcement consists of 4800 TEX unidirectional E-glass roving. The resin is UV curable unsaturated polyester supplied by Irurena S.A., whose commercial name is FPC-7621 NA. The UV source used is a high intensity Phoseon FireFlex UV LED curing system with a maximum intensity of 8 W/cm2 (variable) and an emitting window of 75  50 mm2. The selected photoinitiator system is a combination of Bis (2,4,6-trimethylbenzoyl)-phenylphosphine oxide (BAPO) and 2-Dimethylamino-2-(4-methylbenzyl)-1-(4-morpholin-4-yl-phenyl)-butan-1-one (a aminoketone). This composite/UV source combination has been demonstrated to be suitable, as low porosity and high mechanical performance can be obtained [4]. 2.2. Bent pultrusion processing The bent pultrusion (Fig. 1a) line has been developed entirely by the research group. The impregnation was done by an open resin bath system and the non-linear pulling is made using a Staübli TX60 6 axis robot arm and a pneumatic gripper. A discontinuous pulling strategy is used: after bending the profile the gripper releases the cured part and the process is stopped. Afterwards, the robot grips again the profile and pulls a straight part. Finally, the profile is cut by a saw. A continuous pulling process would be achieved using a complementary robot arm. It must be remarked that the profile is only irradiated from one side. The die was designed to manufacture continuous rectangular sectioned profiles (10 mm width and 2 mm thickness). The die length is 100 mm. The manufactured bent specimen has a radius of 50 mm and the angle at the corner between the straight parts is 90 (Fig. 1b). In this study, three different pulling speeds have been used: 2, 3 and 4 /s (1.75, 2.62 and 3.50 mm/s limited by the robot arm system); and five emitting intensities from 0.8 to 8 W/cm2 have been analysed (0.8, 2.4, 4, 6 and 8 W/cm2). 3 specimens for each intensity

have been tested in order to ensure the repeatability of the tests. The samples quality was evaluated based on the geometrical defects as well as using a NOVA NANOSEM 450 scanning electron microscope. 2.3. DC sensor monitoring In order to analyse the experimental evolution of the degree of cure at the exit of the die the monitoring of the electrical resistance and temperature using a direct current, DC sensor (an Optimold system provided by Synthesites Innovative Technologies Ltd.) has been employed. This type of sensor has demonstrated to be a suitable option for analysing the curing process of composites [4,18,19]. DC sensors are based on correlations between resistivity and state of cure of the resin. Resistivity of a polymer is determined by measurement of the potential drop across the sample and the electric current applied to the sample [18]. Hence, the changes in the measured resistance reflect the changes in the degree of cure and glass transition temperature (Tg) [20]. Furthermore, the electrical resistance (R) is a function of the degree of cure and temperature [20]. Based on literature [16,22] it can be assumed that the electrical resistivity (R) could be expressed separately by a degree of cure (a)function f1 (a)and temperature (T) function f2 (T) as follows:

R ¼ f1 ðaÞf2 ðTÞ

(1)

To correlate the electrical resistivity to Tg and degree of cure, it is desirable to remove the temperature influence on the measured electrical resistivity. As the curing process is initiated by the UV radiation, temperature is not a controllable parameter. Temperature is the sum of the effects of the exothermic reaction and the heat from the UV lamp. Thus, is not possible to analyse isothermal curing cycles in order to remover the influence of the temperature on the measured electrical resistivity. Indeed, the correlation between resistance and temperature might change depending on the state of the material (liquid or cured resin). However, as the geltime in UV curing is negligible [6], the effect of the temperature in the liquid state of the resin can be neglected. Hence, the effect of the temperature for glassy polymers may be well represented by an Arrhenius equation [16,17]:


 b f2 ðTÞ ¼ a$exp T

(2)

where, a and b are experimentally obtained measuring the electrical resistivity during cooling process at room temperature after UV curing. Combining 1 and 2 equations, the temperature decoupled signal (Rdecoupled) can be obtained:



 b /f1 ðaÞ ¼
R ¼ f1 ðaÞ$ a$exp T

R


 ¼ Rdecoupled a$exp Tb (3)

Thus, the degree of cure (a) can be defined as the following equation [17,21]:

     log Rdecoupled  log Rdecoupled min     $afinal  a¼  log Rdecoupled log Rdecoupled max

(4)

min

where, ðRdecoupled Þmin and ðRdecoupled Þmax are the minimum and maximum values for Rdecoupled respectively; and, afinal is the maximum cure degree (obtained experimentally as it is described in Section 2.4).

I. Tena et al. / Composites Part B 89 (2016) 9e17

11

Fig. 1. Front view of the bent pultrusion line (a); details of the manufactured bent specimen (b).

2.4. Modulated Differential Scanning Calorimetry and photocalorimetry The final curing degree of the manufactured composites can be calculated as is presented in equation (5):

afinal ð%Þ ¼

DHA  DHB  100 DHA

(5)

where, afinal is the final curing degree of the manufactured profiles; DHA is the enthalpy value for a fully cured resin; and DHB is the remaining enthalpy to reach the complete conversion. In order to determine the enthalpy value for a fully cured resin (DHA) the Differential Scanning Calorimetry (Mettler Toledo DSC 1 equipment) technique can be used. In this case, the DSC equipment is fitted with a photocalorimetry accessory (photo-DSC). Thus, the integral under the photo-DSC curve peak (heat flow versus time curve), above the baseline, gives the total enthalpy change for the photopolymerisation process. The photo-DSC technique was carried out at intensity of 75 mW/cm2 and at 25  C under nitrogen atmosphere (resin samples typically weighed between 10 and 15 mg). Modulated Differential Scanning Calorimetry (M-DSC e TA Instruments DSC Q-100 DSC equipment) analysis will be used in order to determine the remaining enthalpy to reach the complete conversion of the manufactured composite profiles (DHB). The experimental conditions for the M-DSC analysis are the followings: all the samples were taken from the manufactured composite profiles at the non-exposed surface (typically weighed between 10 and 20 mg). The amplitude of the sinusoidal modulated heat flow was ±1  C and 60 s period. The heating was carried out from room temperature up to 320  C with a heating speed of 5 ºC/min.

sensor, the measurement can be successfully performed. The electrical resistance and temperature during the curing process of all specimens were measured from fully uncured state to the final maximum cure state. All specimens were considered to reach the final maximum conversion when the value of the electrical resistivity converges to a constant value. After this point, the UV light was stopped in order to measure the cooling of the composite up to room temperature. Once the whole curing process of the composite is characterised, the resulting curing degree under different pulling speeds can be determined. It can be assumed that each pulling speed is equivalent to a specific exposure time (the diameter of the sensor is 16 mm). So, if a curing model based on the electrical resistance measurements is developed, the exposure time and the resulting curing degree at the exit of the die can be correlated. 2.6. Curing kinetics In order to obtain the evolution of the degree of cure under different processing conditions from the electrical monitoring, a kinetic model is needed. In this study the autocatalytic model has been employed, as it has been previously successfully applied to photo-polymerisation process [23]. The autocatalytic model can be expressed as:

da ¼ kam ð1  aÞn /aðtÞ ¼ dt

Zt

kam ð1  aÞn dt

0

2.5. Experimental simulation set-up So as to measure the electrical resistance of the composite, a close contact between the sensor and the composite has to be ensured. For this reason, it is not possible to use the DC sensor placed at the exit of the die. Therefore, a new die has been designed in order to simulate the same curing conditions (Fig. 2). In this case, the composite remains static and the composite is irradiated from one side. The sensor is placed in contact with the non-exposed surface (which will be the last part of the thickness being cured); and as there is not relative movement between the profile and the

Fig. 2. Die for the experimental simulation set-up.

(6)

12

I. Tena et al. / Composites Part B 89 (2016) 9e17

where, a is the degree of cure at any time, t; k is the rate constant; m is the autocatalytic exponent; and, n is the reaction order. Thus, combining 4 and 6 equations, the kinetic model can be expressed as:

0

     log Rdecoupled  log Rdecoupled min     $afinal  kam ð1  aÞn dt ¼  log Rdecoupled log Rdecoupled max

The values for rate constant k vary depending on the emitting intensity [23]. This dependence of the rate constant and the UV intensity can be expressed as:

k ¼ k0 $I b

(8)

where, I is the emitting UV intensity; k0 is a constant value for all intensities; and, b is a constant exponent which is linked to the termination mechanism of the reaction [24]. 3. Results and discussion 3.1. Evolution of the degree of cure at the exit of the die In order to determine the evolution of the degree of cure at the exit of the die under different process conditions, the simulation set-up has been used. Composite profiles under all the different emitting intensities have been manufactured and monitored. The first step to analyse the evolution of the degree of cure at the exit of the die is to decouple the effect of the temperature from the signal of the DC sensor. An example of the measured signals (resistance and temperature) of the DC sensor is shown in Fig. 3a. In order to determine the effect of the temperature in the measurement, the experimental values for a and b parameters have to be determined (from equation (3)). Those values can be obtained from the cooling process after the UV curing process. Fig. 3b illustrates an example of the comparison between the evolution of the resistance during the cooling process and the predicted Arrhenius model for an intensity of 0.8 W/cm2. It can be observed that the predicted resistance fits accurately the measured resistance.

issues during cooling, the values for 8 W/m2 can not be shown): a is between 2  1023 and 2.5  1023; and, b was found to have a value very close to 21.8  103. The next step is to determine the maximum degree of cure achieved in each curing condition. For this purpose, M-DSC analysis was carry out for all the profiles manufactured in the simulation set-up. Fig. 5 presents a representative example (in this case for an intensity of 4 W/cm2) of the heat flow versus temperature curve obtained in M-DSC tests. In this curve 3 different main peaks can be identified: the first peak, around 160  C, is related to Tg [4]; the next peak, around 190  C, is related to the curing of the composite (this is the area to analyse in order to define the final degree of cure of the composite); and, the last peak, located at approximately 300  C is related to the degradation of the composite. Therefore, the area to be analysed in the M-DSC test is the peak located near 190  C. As it was described before, photo-DSC test will define the enthalpy for the fully cured resin. As it can be observed in Table 1 the enthalpy referred to a fully cured resin is 208 J/g. This table also shows the final curing degree of the manufactured profiles and the initial (Tinit) and peak (Tpeak) temperatures of the curing area of the M-DSC curves. It can be observed that the final curing degree of all composites is very similar (approximately 99%) as well as recorded initial and peak temperatures. Once all those parameters are determined, the kinetic model that describes the curing reaction can be developed. The experimental curves for the conversion rate as function of the conversion of the composite profiles at three different intensities are shown in Fig. 6. The same figure presents the predicted conversion rate

60

18

107

50

17.5

40

17

105

Cooling

UV stop

30 104 20

103 102 101

R (MΩ) T(ºC)

UV start

10 0

0

50

100

150

200

250

300

350

Temperature (ºC)

108

106 R (MΩ)

(7)

min

lnR

Zt

The values of a and b parameters are linked to the behaviour of the fully cured composite. Thus, those values have to be independent of the intensity within the experimental scatter. Fig. 4 shows those experimental values for a and b parameters (due to recording

16.5

b lna lnR = 21696(1/T) - 52.157 R² = 0.994

16

Ln (R) Linear (lnR)

15.5 15 0.00314

t (s)

0.00316

0.00318 1/T (K-1)

(a)

(b)

0.00320

0.00322

Fig. 3. Example of the measured signals (resistance and temperature) of the DC sensor (a); Example of the comparison between the evolution of the resistance during the cooling process and the predicted Arrhenius model for an intensity of 0.8 W/cm2 (b).

I. Tena et al. / Composites Part B 89 (2016) 9e17

22

3.5

conversion rates and lower total curing time. On the other hand, as it was commented before, different pulling speeds can be translated into different exposure times. So, using the developed autocatalytic model the resulting curing degree at the exit of the die with those different exposure times (equivalent to different pulling speeds) can be determined.

× 103

× 10-23

4

21.5

a (-)

a

21

b (-)

b 3

13

3.2. Bent pultrusion processing

2.5

20.5

2

1.5

20

0

1

2

3 4 Intensity (W/cm2)

5

6

7

Fig. 4. Experimental values for a and b parameters.

(obtained with the kinetic model, equation (7)) for each intensity. It can be observed that theoretical curves for the lowest intensity and the intermediate intensity fit accurately the experimental curves (Fig. 6a and b). However, in the case of the highest intensity (Fig. 6c) the accuracy is lower. As the intensity is very elevated (8 W/cm2) the curing reaction is very fast. This fact is translated into less accuracy in the initial part of the conversion rate curve. Nevertheless, it can be concluded that the overall fitting of the predicted curves and the experimental data is adequate. Fig. 7 presents the resulting parameters of the autocatalytic model. It can be seen that the autoacceleration exponent m was found to have a value close to 0.5; while the reaction order n was close to 1.5, with a slight decrease as the intensity increases. The same figure presents the values of the rate constant k at different emitting intensities. The dependence of the rate constant k is evident, being higher as the intensity increases. The exponent b linked to the termination mechanism of the reaction [23]. As b < 0.5, it can be stated that the predominant termination mechanism of the reaction is primary radical termination. Finally, the evolution of the degree of cure versus time at different intensities is plotted in Fig. 8 (0.55 and 1.25 values are chosen for m and n parameters respectively). It can be concluded that the developed autocatalytic model predicts accurately the experimental data obtained from DC monitoring under different UV exposure conditions. Even all the specimens reach almost the same final degree of cure, higher intensities are translated into higher

Bent profiles with all the possible combinations of pulling speed and emitting intensity have been manufactured. Indeed, the inspection of the final quality of these profiles will allow to define the optimum process window for the process. In order to define this optimum process window, the typology of defects found in the manufactured specimens have been identified. Analysing all the specimens two main types of defects have been identified: undercure and overcure defects. First, detached fibres have been identified in the bent area of some profiles. This error typology is related to a too low curing degree during bending process, producing detached fibres and the deformation of crosssection, as the cohesion of the composite is not ensured (Fig. 9a and b). On the other hand, if the curing degree of the composite is too high during the bending process, the resulting stiffness of the composite at the exit of the die is also elevated. Thus, it could produce damage in the profile due to the bending process. This defect type is shown in Fig. 9c and d. It can be observed that the bending process could inflict damage in the matrix, which results in the debonding of the fibres. The damage is located at the inner region of the bent corner instead of the outer region. The outer region is under tension whereas the inner region is under compression during the bent process. Based on the studies carried out by Baran et al. [25,26], it can be stated that the process parameters have direct influence in the possibility of producing defects and process induced residual stresses. In addition, the residual stresses that may be produced during the troughthickness curing process may have an important effect on the failure mechanism of the pultruded composite. This effect will be evaluated in future work. Finally, Fig. 9e and f shows an example of an optimum specimen. The bending process do not produce any type of damage if the correct curing and pulling conditions are properly chosen. The determination of the optimal pulling conditions and the limits of the process are analysed in the next section. In addition, Fig. 9g shows clearly the 90 angle of a successfully manufactured profile. 3.3. Optimal process window

0.15

Degradation

Heat flow (W/g)

0.4035 J/g Tg 0.1

Table 1 Results of photo-DSC and M-DSC tests.

Exo

2 4 W/cm 2.4 W/cm2

0.05

0

50

100

150 200 Temperature (ºC)

250

300

The proper selection of the combination of UV curing parameters and pulling conditions is critical to obtain successfully manufactured bent profiles. The variation of the emitting UV intensity and the pulling speed vary the curing degree at the exit of the die of the bent profile. Consequently, as it has been commented before, the resulting quality of the manufactured profiles depends on those process parameters. This dependence is represented in Fig. 10. This

350

Fig. 5. Example (intensity of 4 W/cm2) of the heat flow versus temperature curve obtained in M-DSC tests.

Specimen

DHA (J/g)

DHB (J/g)

afinal (%)

Tinit (ºC)

Tpeak (ºC)

0.8 W/cm2 2.4 W/cm2 4 W/cm2 6 W/cm2 8 W/cm2

208

0.4167 0.4035 0.5300 0.5190 0.4883

99.3 99.3 99.1 99.1 99.1

177.7 178.7 178.3 180.8 176.5

194.9 195.4 197.7 198.0 197.8

14

I. Tena et al. / Composites Part B 89 (2016) 9e17

0.05

Predicted conversion rate

Experimental

4 W/cm2

Predicted conversion rate

0,04 Conversion rate (s-1)

0.04

Conversion rate (s-1)

0,05

Experimental

0.8 W/cm2

0.03

0.02

0.01

0,03

0,02

0,01

0

0

0

0.2

0.4 0.6 Conversion (-)

0.8

1

0

0,2

0,4 0,6 Conversion (-)

(a)

0,8

1

(b) 0.05

Experimental

8 W/cm2

Predicted conversion rate

Conversion rate (s-1)

0.04

0.03

0.02

0.01

0 0

0.2

0.4 0.6 Conversion (-)

0.8

1

(c) Fig. 6. Comparison of experimental and predicted curves for the conversion rate as function of the conversion of the composite profiles at 0.8 W/cm2 (a), 4 W/cm2 (b) and 8 W/cm2 (c) intensities.

figure presents the optimum process window for the process described in this study. The relationship between the emitting intensity, pulling speed and the resulting curing degree at the exit of the die is presented. Combining the studied pulling speeds and

2

a)

m, n (-)

1.5 1 0.5

n m

k (s-1)

0 01 1

b)

0.1 k = 0.0646·I 0.2751 0.01 0.1

1 Intensity (W/cm2) Fig. 7. Resulting parameters of the autocatalytic model.

10

emitting intensities, the developed curing model can predict the curing degree of the profiles at the exit of the die. In addition, through the analysis of the final quality of all the manufactured bent profiles in the pultrusion line, the experimental optimum process window can be defined. This optimum process window shows that as the pulling speed increases, the emitting intensity has to increase in order to obtain successfully manufactured profiles. In the same way, if the pulling speed is decreased, lower intensities are needed in order not to find overcure defects. The experimental results show that if the pulling speed is 4 /s or higher, the emitting intensity of 8 W/cm2 is not enough to manufacture an optimum bent profile. However, higher intensities may allow manufacturing optimum bent profiles. In the same way, if the pulling speed is under 2 /s, the emitting intensity has to be decreased in order not to have overcure defects. If the degree of cure at the exit of the die is analysed, the optimum process window is limited approximately from 5% to 12% at the non-exposed surface. Values of degree of cure under 5% will produce undercure defects; whereas values above 12% may be translated into overcure defects. Based on the developed autocatalytic model, a predicted process window can be also roughly delimited. It can be stated that higher intensities than 8 W/cm2 may allow to reach the minimum curing degree value of 5% at the non-exposed surface. In the same way, the emitting intensity has to decrease as the pulling speed decreases bellow 2 /s in order not to have overcure defects if the curing degree value exceeds 12%.

I. Tena et al. / Composites Part B 89 (2016) 9e17

1

2.4 W/cm2

0.9

0.8

0.8

0.7

0.7 Conversion (-)

Conversion (-)

1

0.8 W/cm2

0.9

0.6 0.5

Experimental

0.4

Predicted conversion

0.6 0.5

Experimental

0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

15

Predicted conversion

0

0

50

100 Time (s)

150

200

0

20

40

60

(a) 1

120

140

6 W/cm2

0.9

0.8

0.8

0.7

0.7

Conversion (-)

Conversion (-)

100

(b) 1

4 W/cm2

0.9

80 Time (s)

0.6 0.5

Experimental

0.4

Predicted conversion

0.6 0.5

0.3

0.3

0.2

0.2

0.1

0.1

0

Experimental

0.4

Predicted conversion

0 0

20

40

60

80

100

0

20

40

Time (s)

60

80

100

Time (s)

(c)

(d) 1

8 W/cm2

0.9

0.8 Conversion (-)

0.7 0.6 0.5

Experimental

0.4

Predicted conversion

0.3 0.2 0.1 0

0

20

40

60

80

100

Time (s)

(e) Fig. 8. Evolution of the degree of cure versus time at different intensities: 0.8 W/cm2 (a); 2.4 W/cm2 (b); 4 W/cm2 (c); 6 W/cm2 (d); 8 W/cm2 (e).

4. Conclusions The effect of intensity and pulling speed on UV cured out of die bent pultrusion process has been characterised. Moreover, the curing process of the composite has been characterised (kinetic model) through the analysis of the evolution of the electrical resistivity (degree of cure) of the composite at the exit of the die.

Firstly, the autocatalytic model that correctly describes the curing reaction has been developed. It can be concluded that this autocatalytic model fits accurately the experimental data obtained from DC monitoring. Secondly, the relationship between the emitting intensity, pulling speed and the resulting curing degree at the exit of the die is presented. The experimental results show that the maximum pulling speed is 3 /s if the maximum emitting

16

I. Tena et al. / Composites Part B 89 (2016) 9e17

Fig. 9. Detailed view of undercure defect (a) and SEM section view (b); detailed view of overcure defect (c) and SEM section view (d); detailed view of an optimum specimen (e) and SEM section view (f); detailed view of the 90 angle of a successfully manufactured specimen (g).

intensity is 8 W/cm2 to manufacture an optimum bent profile. However, higher intensities may allow manufacturing optimum bent profiles. Thirdly, if the resulting degree of cure at the exit of the die is analysed, values of degree of cure under 5% will produce

undercure defects; whereas values above 12% may be translated into overcure defects. Hence, in this study, the optimum process window is limited approximately from 5% to 12% of degree of cure at the non-exposed surface (at the exit of the die). Finally, the

I. Tena et al. / Composites Part B 89 (2016) 9e17

17

Pulling speed (°/s)

5

4

1.2

1.6

1.9

2.2

2.4

2.7

2.8

3.0

3.2

3.4

3

2.2

3.0

3.6

4.1

4.5

4.9

5.2

5.6

5.9

6.2

2

5.2

7.2

8.6

9.7

10.7

11.6

12.3

13.1

13.7

14.3

Undercure Overcure

1 0

0.8

1.6

2.4

3.2 4 4.8 5.6 Intensity (W/cm2)

6.4

7.2

8

Experimental process window

8.8

Predicted process window

DC

Predicted degree of cure (%)

Fig. 10. The optimum process window for ultraviolet cured out of die bent pultrusion process.

analysis of the residual stresses/strains induced during the through-thickness cure it seems to be an important aspect to be analysed in future work. Acknowledgements I. Tena thanks the Basque Government for the grant (BFI 2011228). The authors also thank the Basque Government for providing financial support (ACTIMAT IE10-272 & GAITEK 2011/0000540, 2012/0001110 and 2013/0000447) for this study. The authors gratefully acknowledge the assistance provided provided in the MDSC tests by Noelia Nieto from Titania, Ensayos y Proyectos Industriales and the support of Nikos Pantelelis from Synthesites Innovative Technologies Ltd. in the DC sensor measurements. References [1] Britnell D, Tucker N, Smith G, Wong S. Bent pultrusionda method for the manufacture of pultrudate with controlled variation in curvature. J Mater Process Technol 2003;138(1):311e5. [2] Jansen K, Weidler D, Hoffmann M. In: GmbH Thomas, Technik Co, Innovation KG, editors. Method and device for the production of a plastic Profile; 2009. B29C70/28, B29C47/34. [3] Goupy M. In: Renault, editor. Energy absorbing curved sections; 1983. B60R19/04, B60R19/18, B60R19/36, B60R19/28, B60R19/30, B60R19/02. [4] Tena I, Esnaola A, Sarrionandia M, Ulacia I, Torre J, Aurrekoetxea J. Out of die ultraviolet cured pultrusion for automotive crash structures. Compos Part B Eng 2015;79(0):209e16. [5] Bader MG. Selection of composite materials and manufacturing routes for cost-effective performance. Compos Part A Appl Sci Manuf 2002;33(7): 913e34. [6] Endruweit A, Johnson MS, Long AC. Curing of composite components ultraviolet radiation: a review RID B-4506-2009. Polym Compos 2006;27(2). [7] Compston P, Schiemer J, Cvetanovska A. Mechanical properties and styrene emission levels of a UV-cured glass-fibre/vinylester composite. Compos Struct 2008;86(1e3). [8] Park J, Kong J, Kim D, Lee J. Non-destructive damage sensing and cure monitoring of carbon fiber/epoxyacrylate composites with UV and thermal curing using electro-micromechanical techniques. Compos Sci Technol 2004;64(16). [9] Shi W, Ranby B. UV curing of composites based on modified unsaturated polyester. J Appl Polym Sci 1994;51(6):1129e39.

[10] Green WA. Industrial photoinitiators: a technical guide. CRC Press; 2010. [11] Lackey E, Vaughan J, Patki R. Enhanced pultrusion using photocure to supplement standard thermal cure. In: Proceedings of the convention and trade show composites fabricators association; 2003. California USA. [12] Baran I, Tutum CC, Nielsen MW, Hattel JH. Process induced residual stresses and distortions in pultrusion. Compos Part B Eng 2013;51:148e61. [13] Baran I, Akkerman R, Hattel JH. Modelling the pultrusion process of an industrial L-shaped composite profile. Compos Struct 2014;118:37e48. [14] Baran I, Hattel JH, Akkerman R. Investigation of process induced warpage for pultrusion of a rectangular hollow profile. Compos Part B Eng 2015;68: 365e74. [15] Chen J, Octeau M, Hojjati M, Yousefpour A. Cure cycle optimisation for composite panels fabricated by RTM using dielectric sensors. In: 17th international conference on composite materials, ICCM-17; 2009. [16] Chen J, Hojjati M. Microdielectric analysis and curing kinetics of an epoxy resin system. Polym Eng Sci 2007;47(2):150e8. [17] Maffezzoli A, Trivisano A, Opalicki M, Mijovic J, Kenny JM. Correlation between dielectric and chemorheological properties during cure of epoxy-based composites. J Mater Sci 1994;29(3):800e8. [18] Meier R, Zaremba S, Springl F, Drechsler K, Gaille F, Weimer C. Online process monitoring systems - benchmark and test study. Auckland, New Zealand. In: Proceedings of 11th international conference on flow processes in composite materials (FPCM11); 2012. [19] Pantelelis N, Nedelec G, Amosse Y, Toitgans M, Beigbeder A, Harismendy I, et al. Industrial cure monitoring and control of the RTM production of a CFRP automotive component. In: Proceedings of European conference on composite materials (ECCM15). Venice, Italy; 2012. [20] Pantelelis NG, Bistekos E. Process monitoring and control for the production of CFRP components. In: SAMPE Conference; 2010. Seattle. [21] Zhang J, Pantelelis NG. Modelling and control of reactive polymer composite moulding using bootstrap aggregated neural network models. Chem Prod Process Model 2011;6(2). [22] Kim HG. Dielectric cure monitoring for glass/polyester prepreg composites. Compos Struct 2002;57(1):91e9. [23] Dalle Vacche S, Geiser V, Leterrier Y, Månson JE. Time-intensity superposition for photoinitiated polymerization of fluorinated and hyperbranched acrylate nanocomposites. Polymer 2010;51(2):334e41. [24] Andrzejewska E. Photopolymerization kinetics of multifunctional monomers. Prog Polym Sci 2001;26(4):605e65. [25] Baran I, Tutum CC, Hattel JH, Akkerman R. Pultrusion of a vertical axis wind turbine blade part-I: 3D thermo-chemical process simulation. Int J Mater Form 2014:1e11. [26] Baran I, Hattel JH, Tutum CC, Akkerman R. Pultrusion of a vertical axis wind turbine blade part-II: combining the manufacturing process simulation with a subsequent loading scenario. Int J Mater Form 2014:1e12.