- Email: [email protected]
Understanding heat transfer mechanisms during the cooling phase of blow molding using infrared thermography
NDT&E International 38 (2005) 433–441 www.elsevier.com/locate/ndteint
Understanding heat transfer mechanisms during the cooling phase of blow molding using infrared thermography A. Bendadaa,*, F. Erchiquib, A. Kippingc a
National Research Council of Canada, Industrial Materials Institute, 75 De Mortgane, Boucherville, Que., Canada J4B 6Y4 b University of Quebec in Abitibi-Temiscamingue, 445 Universite´ Blvd., Rouyn-Noranda, Que., Canada J9X 5E4 c University of Siegen, Paul-Bonatz Strasse 9-11, Siegen 57068, Germany Received 15 June 2004; accepted 25 November 2004 Available online 24 February 2005
Abstract The cooling phase of the extrusion blow molding process has a large influence on the cycle time of the process as well as on the properties and quality of the molded products. A better understanding of the heat transfer mechanisms occurring during the cooling phase will help in the optimization of both mold cooling channels and operating conditions. A continuous extrusion blow molding machine and a rectangular bottle (motor oil type) mold were used to produce bottles. A high density polyethylene (HDPE) and a metallocene polyethylene (mPE) having different rheological properties were tested. Melt and mold temperatures, cooling time, inflating pressure and die gap were varied systematically. An infrared camera was used to measure the temperature distribution of the plastic part just after mold opening as well as after part ejection. The wall thickness and dimensions of the bottles of the finished parts were measured in order to determine the shrinkage and warpage. Finally, the infrared temperature fingerprints were used to explain what happens during the cooling phase and correlated with the final part characteristics. Crown Copyright q 2005 Published by Elsevier Ltd. All rights reserved. Keywords: Blow molding; Cooling; Infrared thermography; Polymer processing
1. Introduction In the extrusion blow molding of medium size bottles, the cooling stage represents a substantial part of the overall cycle and has a profound effect on the microstructure development and on the ultimate properties of the molded article. During the cooling of the blown part whilst in the mold, heat is removed both by forced convection (polymer/air interface at the internal wall) and conduction (polymer/metal interface at the mold wall). Once the mold opens, the part will continue to cool from both surfaces by natural convection. Several authors [1–4] have developed numerical algorithms for the predictions of the temperature profiles in the parts. Their efforts have been of limited value due to the uncertainty in the values of heat transfer coefficients used to represent the different heat transfer
* Corresponding author. Tel.: C1 450 641 5241; fax: C1 450 641 5106. E-mail address: [email protected] (A. Bendada).
mechanisms. For these reasons infrared thermography presents itself as a complementary technique capable of mapping the thermal history of blow molded parts during the different stages of the cooling process [5–7].
2. Experimental setup The parametric study was done on a continuous extrusion blow molding machine (Battenfeld-Fischer FBZ1000), equipped with a motor oil type bottle mold. The mold is modular and has five interchangeable parts. The cavity is made of a top block, a middle block and a bottom block, each with its own cooling circuit. The three blocks are mounted onto a back plate and within the middle block an exchangeable insert is located (Fig. 1). The mold cavity has a height of 220 mm, a width of 100 mm and a depth of 50 mm. In this study, a flat insert was used in the middle block to ensure a complete contact between plastic and mold. A diverging die (fZ30 mm) was used in this study.
0963-8695/$ - see front matter Crown Copyright q 2005 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ndteint.2004.11.007
434
A. Bendada et al. / NDT&E International 38 (2005) 433–441
Fig. 1. Mold configuration (cavity and cooling lines).
The machine set-up enables to vary several operating parameters such as die gap, melt temperature, inflation pressure, cooling time, and temperature of the coolant liquid. The temperature distribution of the plastic bottle was measured directly after mold opening, while remaining attached to the blow pin. An AGEMA 900 LW infrared camera was used. The thermographs covered an area of the bottle described as follows: short side-corner-wide side. The temperature measurement through an infrared camera is a non-contact technique, so that the recorded temperature
itself is not disturbed by the measurement [8]. With this technique, both the temperature level and its distribution on the part surface can be observed. The infrared camera is equipped with a HgCdTe sensor; its spectral response is centered around 8–12 mm in the far infrared spectrum. It has a sensitivity of 0.08 at 30 8C, and a repeatability of 0.5%. The infrared frames of 68 lines of 272 columns are stored in real time at a rate of 30 Hz on a hard disk for subsequent processing. Due to the polymer semi-transparency to infrared radiation, the infrared camera was only able to provide thermal information about the process in terms of an average or bulk temperature, but not about the surface temperature or the detailed temperature distribution inside the part [9]. During the measurements, precautions have been taken to overcome spurious radiation (transmitted or reflected) from nearby hot objects. The overall experimental set-up (blow molding machine and camera) is shown in Fig. 2. After demolding, the subsequent cooling of the bottles was monitored by placing the bottles in a reference support. Furthermore, the solid bottles were also photographed placed in this support (constant camera angle) in order to measure their dimensions and dimensional stability. This allowed the comparison between differently produced bottles by superposing the images on a grid. Finally, the wall thickness of the bottles is measured along the circumference, 80 mm from the bottom.
3. Materials Two different materials were used in this study. A typical blow molding high density polyethylene (HDPE)—
Fig. 2. Experimental setup.
A. Bendada et al. / NDT&E International 38 (2005) 433–441 Table 1 Material properties
3
Density (g/cm ) MI (g/10 min) Vicat softening point (8C) Ultimate elongation (%) Tensile strength (MPa)
435
Table 2 Operating conditions HDPE
mPE
0.951 0.14 129 600 27
0.92 0.85 105 650 53
(Dow/Polisur 70055L) and an enhanced polyethylene (mPE) produced using metallocene technology (Dow Elite 5100). The basic material properties are given in Table 1. Due to the different molecular structure of the materials, different parison swell and sagging behavior during extrusion is expected. A crude estimation of this behavior was obtained by measuring the diameter and thickness of the extruded parisons, both close to the die and at the bottom of the parison. A comparison of the cross-section of the parisons (at the top and at the bottom) is shown in Fig. 3 for HDPE and mPE, respectively. HDPE swells considerably and also appears to have low melt strength. This results in a parison having a wall thickness distribution along its length. On the other hand, the mPE exhibits lower swell and low melt strength, which results paradoxically in a parison
Die gap (mm) Tmelt (8C) Tmold (8C) Blowing pressure (bar) Cooling time (s)
HDPE
mPE
2–4 220–240 9–21 3–7
2–4 190–210 9–21 3–7
10–20
8–11
having a more uniform thickness along its length. These differences will have a strong influence on the absolute and thickness distributions values in the blow part and consequently on its cooling behavior.
4. Processing parameters A simple design of experiments scheme was used to vary the processing parameters. During the variation of one of the described parameters, all other parameters were kept constant at their mean value. A summary of the range of parameter variation is given in Table 2. Due to the different material behavior, different melt temperatures and cooling times were chosen for the HDPE and the mPE. The temperature of the HDPE melt is varied between 220 and 240 8C, while for the mPE melt a 190–210 8C temperature range was used. In the case of the blowing times, these were varied as follows: HDPE—10, 15 and 20 s; mPE—5, 8 and 11 s. The lower temperatures and shorter cooling times were used in order to accommodate the shorter drop times of the mPE. The mold temperature is controlled through the variation of the flow rate and temperature of the cooling liquid. These temperatures were set at 4, 10 and 21 8C. The actual mold temperature was measured by using a thermocouple inserted flush with the mold surface. The die gap is also changed with the control unit of the machine. By the use of a servo cylinder, the mandrel within the die can be moved axially. Because the mandrel has a conical shape, the die gap changes by moving the mandrel, Fig. 4. After every parameter change, the process was allowed to reach stable conditions. This stabilization period appears to be quite important when changing mold temperature and blowing pressure.
5. Results
Fig. 3. (a) Cross-section of a HDPE-parison. (b) Cross-section of a mPEparison.
Most of the studies that have been performed on the effect of cooling conditions on the final properties of blow molded parts have not been careful in identifying the temperature distributions both in the parison and in the mold prior to blowing. These two conditions are crucial for a comprehensive interpretation of the molding results. The bottom of the parison, although thicker in general, is exposed longer to the ambient air and therefore has a lower temperature. In our study, the HDPE parison exhibits
436
A. Bendada et al. / NDT&E International 38 (2005) 433–441
Fig. 4. Variation of the die gap.
a greater axial temperature difference (top to bottom) (24 8C) when compared with the mPE (15 8C). This is illustrated by the temperature distribution measured with the infrared camera for both, a HDPE parison and mPE parison in Figs. 5 and 6. In each case the die gap is 3 mm. For HDPE, the melt temperature is 230 8C, for mPE it is 200 8C. Fig. 5 shows the temperature image of the HDPE parison, while Fig. 6 the temperature image of the mPE parison. In Fig. 7, the temperature profiles along the shown vertical lines in the two previous images are compared. In the case of the mPE, the low swell and high sag of the mPE generates a more isothermal parison. This situation could be the cause of different inflation patterns. During inflation, the parison will touch the mold walls having surface temperature distributions determined by the location and characteristics of the cooling channels [10,11]. To detect differences within the cavity temperature, the mold is cooled down using a temperature control unit. The chosen coolant temperature is 4 8C. After the temperature distribution is homogeneous within the cavity, it is rapidly heated. During this heating period, discontinuities of the cavity temperature can be observed, because of differences in the heating speed in the different locations of the cavity. After a certain time, again
the temperature distribution is homogeneous within the cavity, due to heat transfer compensation. Before measuring the mold cavity with the infrared camera, the interesting area of the cavity has to be painted with a high-emissivity black paint, to increase the emissivity of the mold. Usually, the emissivity of the aluminum mold is very low [8,9]. Fig. 8 shows the temperature distribution of the cavity 360 s after the beginning of heating. At this moment the most inhomogeneous temperature distribution is observable. A comparison of the temperature development with time for two specific spots is shown in Fig. 9. The two spots, Spot 1 and Spot 2, are marked with two cross-symbols in Fig. 8. Spot 1 is located in the center of the cavity, while Spot 2 is located at the corner of the cavity. Fig. 9 shows that heat transfer is more efficient in the central area of the cavity (wide side) than in the corners. It can also be observed that the largest temperature difference within the mold cavity is about 10 8C and is reached 360 s after the beginning of heating. These temperature gradients are caused through the used insert at the middle of the cavity. The cooling channel within the insert is located closer to the cavity surface than within the rest of the mold. The shown difference in mold
Fig. 5. Temperature distribution in a HDPE-parison, for a die gap of 3 mm and a melt-temperature of 230 8C.
Fig. 6. Temperature distribution mPE-parison, for a die-gap of 3 mm and a melt-temperature of 200 8C.
A. Bendada et al. / NDT&E International 38 (2005) 433–441
437
Fig. 7. Temperature vertical profiles of the parison (mPE and HDPE). Fig. 10. Bottle wall thickness vs. cooling time for HDPE and mPE.
Fig. 8. Temperature distribution of the mold cavity after 360 s of heating time.
the bottle. Due to the magnetic behavior of measurement head and ball, the ball follows the measurement head, and the gap between ball and measurement head is determined. The wall thickness is measured along a specific line 80 mm above the bottom of the mold as shown in Fig. 10. Within the figure, the length of the line is normalized. It starts at the mid-wideside of the bottle (spot marked 1 in Fig. 11). The bottles wall thickness results show that the wall thickness (for both HDPE and mPE) is independent of the processing conditions with the exception of the cooling time (Fig. 10). As expected (due to the use of a cylindrical die), the shorter side (higher blow ratio) has a lower thickness than the wider side (lower blow ratio). It can also be seen that the thickness profiles for the HDPE and mPE are very different, with the mPE having a
cooling will influence also the temperature distribution of the plastic part itself. After the production of the bottles, the wall thickness of the finished parts is measured with an inductive wall thickness instrument. It consists of a measurement head and a ball. The ball is put inside the bottle while the measurement head is driven along the outer surface of
Fig. 9. Temperature development for two specific spots of the mold cavity.
Fig. 11. Line of wall thickness measurement.
438
A. Bendada et al. / NDT&E International 38 (2005) 433–441
short side with almost uniform thickness. This is probably due to the mPE being able to stretch into the corners even after having touched the mold wall at the large side of the bottle. Furthermore, due to its low swell the mPE has lower overall thickness values. The temperature distribution of the plastic parts is measured directly after mold opening, while the bottle is remaining for some seconds at the blow pin. As for the wall thickness, the influence of the cooling conditions for HDPE and mPE is investigated. To get a qualitative comparison, the infrared images are recorded and compared. A typical infrared image of the HDPE bottle produced during the investigation of the inflating pressure effect is reported in Fig. 12. The quantitative temperature measurement of the bottles surface was made along a line covering about a quarter of its circumference. This specific line is shown in the infrared image of Fig. 12. The line starts at the middle of the short site of the bottle and ends at the middle of the wide side of the bottle. A nondimensional distance (0Zmid-short-side; 1Zmid-wideside) is used to display the results. The wall thickness along this line is shown together with the temperature in the same diagram to give an idea of its influence on the temperature distribution. It was observed that the temperature distribution was directly proportional to the die gap used. For smaller die gaps, a more homogeneous temperature distribution at mold opening is obtained. Higher melt temperatures only resulted in a shift of the temperature at mold opening without any change in the shape of the spatial distribution. Figs. 13 and 14 show the temperature and the wall thickness distributions as function of various inflation pressures for HDPE and mPE, respectively. These results show the potential of infrared thermography to detect material characteristics and flaws in the process. As expected from a non-axisymmetric blowing (motor oil bottle), the short side (thinner) is colder than the wider side (thicker). The temperature distribution can also locate the thin spots corresponding to the edges of the part (minimum temperature). Furthermore, it can pinpoint areas in
the mold where cooling is inefficient [10,11]. As was mentioned before, the location of the cooling channels creates a more efficient cooling in the wide side of the bottle. This explains the higher temperatures at the thinner sections. From the material point of view, the temperature distributions was able to detect the behavior of the material during blowing. MPE requires lower pressures (stresses) to stretch and therefore the temperature distribution is independent of the blowing pressure. On the other hand, the HDPE is much stiffer and at the lower pressures it cannot stretch enough to make full contact with the mold wall. Air trapped between the plastic and the mold could render the analysis of the results even more complicated. The effect of cooling time on the temperature distribution of the part is shown in Fig. 15. Longer cooling times produce parts having a more uniform temperature at demolding. This is particularly the case for parts having thinner walls (mPE vs. HDPE). The different stretching of the material and the different polymer/mold contact conditions will have an effect on the warpage of the parts. An analysis of the temperature development after
Fig. 12. A typical infrared image of the HDPE bottle as it is ejected out of the mold.
Fig. 14. Temperature and wall thickness vs. inflation pressure for mPE.
Fig. 13. Temperature and wall thickness vs. inflation pressure for HDPE.
A. Bendada et al. / NDT&E International 38 (2005) 433–441
439
Fig. 18. Illustration of the temperature compensation.
Fig. 15. Temperature and wall thickness vs. cooling time for mPE.
subsequent cooling in ambient air provides no additional information. Comparison of infrared images of the HDPE bottle recorded 11 and 131 s after mold opening (Figs. 16 and 17), shows that just temperature compensation takes place. This can be clearly seen on the temperature horizontal profiles extracted from the previous infrared images and plotted in Fig. 18. Indeed, the profiles show that there is not
only a heat transfer to the ambient air, which would lead to a decrease of temperature all over the line; there is also a heat transfer in the surface of the bottle direction. The temperature decreases at the middle of the line, normalized length 0.45, but increases at the ends of the line due to heat conduction. The influence of the operating conditions on the shrinkage and warpage of the finished bottles is shown in Figs. 19 and 20 for HDPE and mPE, respectively. The dimensions of the bottle were obtained from pictures taken with a digital camera and compared with the cavity dimensions via image superposition on a grid. It is clear
Fig. 16. Temperature distribution 11 s after mold opening, HDPE.
Fig. 17. Temperature distribution 131 s after mold opening, HDPE.
Fig. 19. Part geometry vs. cooling time (HDPE).
440
A. Bendada et al. / NDT&E International 38 (2005) 433–441
the process. The temperature distribution can locate the thin spots corresponding to the edges of the part (minimum temperature). Furthermore, it can pinpoint areas in the mold where cooling is inefficient. From the material point of view, the temperature distributions were able to detect the behavior of the material during blowing. mPE requires lower pressures (stresses) to stretch and therefore the temperature distribution is independent of the blowing pressure. On the other hand, the HDPE is much stiffer and at the lower pressures it cannot stretch enough to make full contact with the mold wall. It is clear that non-uniform temperature distributions (shorter cooling times) at demolding will favor warping of the bottles, particularly for materials that have been under relative high stresses (HDPE). Shorter cooling times increase the time over which the parts cool down under stress free conditions. Finally, the results can also be used in the validation of process simulation models. They will certainly be helpful in establishing initial values for the cooling and shrinkage and warpage simulations. Fig. 20. Part geometry vs. cooling time (mPE).
that non-uniform temperature distributions (shorter cooling times) at demolding will favor warping of the bottles, particularly for materials that have been under relative high stresses (HDPE). Shorter cooling times increases the time over which the parts cool down under stress free conditions.
Acknowledgements The authors would like to thank the financial support of the DAAD (Deutscher Akademischer Austauschdienst/German Scientific Exchange Service). Also, they wish to thank M.A. Rainville and M. Carmel of the Industrial Materials Institute (IMI) of the National Research Council of Canada (NRC) for their help during the molding trials.
6. Conclusions The influence of different operating conditions on the cooling phase of the extrusion blow molding process of HDPE and mPE has been investigated. The temperature distribution of the bottles, just after mold opening and after subsequent cooling in ambient air has been measured using infrared thermography. The cooling design of the mold has also been investigated with infrared thermography. It has been shown that in order to have a reliable analysis of the cooling phase of the blow molding process, an accurate description of the dimensions of the parison (swell and sag effects) as well as the temperature distribution in the mold (cooling channels effect) is required. The temperature distribution was found to be directly proportional to the die gap used. For smaller die gaps, a more homogeneous temperature distribution at mold opening is obtained. Higher melt temperatures only resulted in a shift of the temperature at mold opening without any change in the shape of the spatial distribution. The results have also shown the potential of infrared thermography to detect material characteristics and flaws in
References [1] Delorenzi HG, Nied HF. Blow molding and thermoforming of plastics: finite element modeling. Comput Struct 1987;26(1–2): 197–206. [2] Laroche D, Erchiqui F. 3D modeling of the blow molding process. In: Proceedings of NUMIFORM’98 conference, Enschede, Netherlands; 1998, p. 483–8. [3] Laroche D, Erchiqui F. Experimental and theoretical study of the thermoformability of industrual polymers. J Reinf Plast Compos 2000;19(3):231–9. [4] Derdouri A, Erchiqui F, Bendada A, Verron E, Peseux B. Viscoelastic behavior of polymer membranes under inflation. In: Proceedings of rheology’2000, XIII international congress on rheology, Cambridge, UK; 2000, p. 394–6. [5] Prystay M, Garcia-Rejon A. Application of thermography for the optimization of the blow molding process. In: Proceedings of ANTEC conference, vol. 1. Brookfield, CT, USA: Society of Plastics Engineers; 1996, p. 996–1000. [6] Prystay M, Garcia-Rejon A. Using thermography to optimize the blow-molding process. Plast-Eng 1996;52(12):41–3. [7] Prystay M, Wang H, Garcia-Rejon A. Application of thermographic temperature measurements in injection molding and blow molding of plastics. In: Proceedings of SPIE, vol. 2766. Bellingham, WA, USA: Society of Photo-Optical Instrumentation Engineers; 1996, p. 5–14.
A. Bendada et al. / NDT&E International 38 (2005) 433–441 [8] Gaussorgues G. Infrared thermography. London: Chapman & Hall; 1994. [9] Dewitt DP, Nutter GD. Theory and practice of radiation thermometry. New York: Wiley; 1988. [10] Prystay M, Loong CA, Nguyen K. Optimization of cooling channel design and spray patterns in aluminum die casting using infrared
441
thermography. In: Proceedings of SPIE, vol. 2766. Bellingham, WA, USA: Society of Photo-Optical Instrumentation Engineers; 1996, p. 15–4. [11] Attar A, Bendada A, Connolly R, De´nomme´ Y, Dallaire E, Aitcin PC. Novel ultra-high-performance concrete as mold material for the blow molding process. SAMPE 2000;36(3):45–9.
Understanding heat transfer mechanisms during the cooling phase of blow molding using infrared thermography A. Bendadaa,*, F. Erchiquib, A. Kippingc a
National Research Council of Canada, Industrial Materials Institute, 75 De Mortgane, Boucherville, Que., Canada J4B 6Y4 b University of Quebec in Abitibi-Temiscamingue, 445 Universite´ Blvd., Rouyn-Noranda, Que., Canada J9X 5E4 c University of Siegen, Paul-Bonatz Strasse 9-11, Siegen 57068, Germany Received 15 June 2004; accepted 25 November 2004 Available online 24 February 2005
Abstract The cooling phase of the extrusion blow molding process has a large influence on the cycle time of the process as well as on the properties and quality of the molded products. A better understanding of the heat transfer mechanisms occurring during the cooling phase will help in the optimization of both mold cooling channels and operating conditions. A continuous extrusion blow molding machine and a rectangular bottle (motor oil type) mold were used to produce bottles. A high density polyethylene (HDPE) and a metallocene polyethylene (mPE) having different rheological properties were tested. Melt and mold temperatures, cooling time, inflating pressure and die gap were varied systematically. An infrared camera was used to measure the temperature distribution of the plastic part just after mold opening as well as after part ejection. The wall thickness and dimensions of the bottles of the finished parts were measured in order to determine the shrinkage and warpage. Finally, the infrared temperature fingerprints were used to explain what happens during the cooling phase and correlated with the final part characteristics. Crown Copyright q 2005 Published by Elsevier Ltd. All rights reserved. Keywords: Blow molding; Cooling; Infrared thermography; Polymer processing
1. Introduction In the extrusion blow molding of medium size bottles, the cooling stage represents a substantial part of the overall cycle and has a profound effect on the microstructure development and on the ultimate properties of the molded article. During the cooling of the blown part whilst in the mold, heat is removed both by forced convection (polymer/air interface at the internal wall) and conduction (polymer/metal interface at the mold wall). Once the mold opens, the part will continue to cool from both surfaces by natural convection. Several authors [1–4] have developed numerical algorithms for the predictions of the temperature profiles in the parts. Their efforts have been of limited value due to the uncertainty in the values of heat transfer coefficients used to represent the different heat transfer
* Corresponding author. Tel.: C1 450 641 5241; fax: C1 450 641 5106. E-mail address: [email protected] (A. Bendada).
mechanisms. For these reasons infrared thermography presents itself as a complementary technique capable of mapping the thermal history of blow molded parts during the different stages of the cooling process [5–7].
2. Experimental setup The parametric study was done on a continuous extrusion blow molding machine (Battenfeld-Fischer FBZ1000), equipped with a motor oil type bottle mold. The mold is modular and has five interchangeable parts. The cavity is made of a top block, a middle block and a bottom block, each with its own cooling circuit. The three blocks are mounted onto a back plate and within the middle block an exchangeable insert is located (Fig. 1). The mold cavity has a height of 220 mm, a width of 100 mm and a depth of 50 mm. In this study, a flat insert was used in the middle block to ensure a complete contact between plastic and mold. A diverging die (fZ30 mm) was used in this study.
0963-8695/$ - see front matter Crown Copyright q 2005 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ndteint.2004.11.007
434
A. Bendada et al. / NDT&E International 38 (2005) 433–441
Fig. 1. Mold configuration (cavity and cooling lines).
The machine set-up enables to vary several operating parameters such as die gap, melt temperature, inflation pressure, cooling time, and temperature of the coolant liquid. The temperature distribution of the plastic bottle was measured directly after mold opening, while remaining attached to the blow pin. An AGEMA 900 LW infrared camera was used. The thermographs covered an area of the bottle described as follows: short side-corner-wide side. The temperature measurement through an infrared camera is a non-contact technique, so that the recorded temperature
itself is not disturbed by the measurement [8]. With this technique, both the temperature level and its distribution on the part surface can be observed. The infrared camera is equipped with a HgCdTe sensor; its spectral response is centered around 8–12 mm in the far infrared spectrum. It has a sensitivity of 0.08 at 30 8C, and a repeatability of 0.5%. The infrared frames of 68 lines of 272 columns are stored in real time at a rate of 30 Hz on a hard disk for subsequent processing. Due to the polymer semi-transparency to infrared radiation, the infrared camera was only able to provide thermal information about the process in terms of an average or bulk temperature, but not about the surface temperature or the detailed temperature distribution inside the part [9]. During the measurements, precautions have been taken to overcome spurious radiation (transmitted or reflected) from nearby hot objects. The overall experimental set-up (blow molding machine and camera) is shown in Fig. 2. After demolding, the subsequent cooling of the bottles was monitored by placing the bottles in a reference support. Furthermore, the solid bottles were also photographed placed in this support (constant camera angle) in order to measure their dimensions and dimensional stability. This allowed the comparison between differently produced bottles by superposing the images on a grid. Finally, the wall thickness of the bottles is measured along the circumference, 80 mm from the bottom.
3. Materials Two different materials were used in this study. A typical blow molding high density polyethylene (HDPE)—
Fig. 2. Experimental setup.
A. Bendada et al. / NDT&E International 38 (2005) 433–441 Table 1 Material properties
3
Density (g/cm ) MI (g/10 min) Vicat softening point (8C) Ultimate elongation (%) Tensile strength (MPa)
435
Table 2 Operating conditions HDPE
mPE
0.951 0.14 129 600 27
0.92 0.85 105 650 53
(Dow/Polisur 70055L) and an enhanced polyethylene (mPE) produced using metallocene technology (Dow Elite 5100). The basic material properties are given in Table 1. Due to the different molecular structure of the materials, different parison swell and sagging behavior during extrusion is expected. A crude estimation of this behavior was obtained by measuring the diameter and thickness of the extruded parisons, both close to the die and at the bottom of the parison. A comparison of the cross-section of the parisons (at the top and at the bottom) is shown in Fig. 3 for HDPE and mPE, respectively. HDPE swells considerably and also appears to have low melt strength. This results in a parison having a wall thickness distribution along its length. On the other hand, the mPE exhibits lower swell and low melt strength, which results paradoxically in a parison
Die gap (mm) Tmelt (8C) Tmold (8C) Blowing pressure (bar) Cooling time (s)
HDPE
mPE
2–4 220–240 9–21 3–7
2–4 190–210 9–21 3–7
10–20
8–11
having a more uniform thickness along its length. These differences will have a strong influence on the absolute and thickness distributions values in the blow part and consequently on its cooling behavior.
4. Processing parameters A simple design of experiments scheme was used to vary the processing parameters. During the variation of one of the described parameters, all other parameters were kept constant at their mean value. A summary of the range of parameter variation is given in Table 2. Due to the different material behavior, different melt temperatures and cooling times were chosen for the HDPE and the mPE. The temperature of the HDPE melt is varied between 220 and 240 8C, while for the mPE melt a 190–210 8C temperature range was used. In the case of the blowing times, these were varied as follows: HDPE—10, 15 and 20 s; mPE—5, 8 and 11 s. The lower temperatures and shorter cooling times were used in order to accommodate the shorter drop times of the mPE. The mold temperature is controlled through the variation of the flow rate and temperature of the cooling liquid. These temperatures were set at 4, 10 and 21 8C. The actual mold temperature was measured by using a thermocouple inserted flush with the mold surface. The die gap is also changed with the control unit of the machine. By the use of a servo cylinder, the mandrel within the die can be moved axially. Because the mandrel has a conical shape, the die gap changes by moving the mandrel, Fig. 4. After every parameter change, the process was allowed to reach stable conditions. This stabilization period appears to be quite important when changing mold temperature and blowing pressure.
5. Results
Fig. 3. (a) Cross-section of a HDPE-parison. (b) Cross-section of a mPEparison.
Most of the studies that have been performed on the effect of cooling conditions on the final properties of blow molded parts have not been careful in identifying the temperature distributions both in the parison and in the mold prior to blowing. These two conditions are crucial for a comprehensive interpretation of the molding results. The bottom of the parison, although thicker in general, is exposed longer to the ambient air and therefore has a lower temperature. In our study, the HDPE parison exhibits
436
A. Bendada et al. / NDT&E International 38 (2005) 433–441
Fig. 4. Variation of the die gap.
a greater axial temperature difference (top to bottom) (24 8C) when compared with the mPE (15 8C). This is illustrated by the temperature distribution measured with the infrared camera for both, a HDPE parison and mPE parison in Figs. 5 and 6. In each case the die gap is 3 mm. For HDPE, the melt temperature is 230 8C, for mPE it is 200 8C. Fig. 5 shows the temperature image of the HDPE parison, while Fig. 6 the temperature image of the mPE parison. In Fig. 7, the temperature profiles along the shown vertical lines in the two previous images are compared. In the case of the mPE, the low swell and high sag of the mPE generates a more isothermal parison. This situation could be the cause of different inflation patterns. During inflation, the parison will touch the mold walls having surface temperature distributions determined by the location and characteristics of the cooling channels [10,11]. To detect differences within the cavity temperature, the mold is cooled down using a temperature control unit. The chosen coolant temperature is 4 8C. After the temperature distribution is homogeneous within the cavity, it is rapidly heated. During this heating period, discontinuities of the cavity temperature can be observed, because of differences in the heating speed in the different locations of the cavity. After a certain time, again
the temperature distribution is homogeneous within the cavity, due to heat transfer compensation. Before measuring the mold cavity with the infrared camera, the interesting area of the cavity has to be painted with a high-emissivity black paint, to increase the emissivity of the mold. Usually, the emissivity of the aluminum mold is very low [8,9]. Fig. 8 shows the temperature distribution of the cavity 360 s after the beginning of heating. At this moment the most inhomogeneous temperature distribution is observable. A comparison of the temperature development with time for two specific spots is shown in Fig. 9. The two spots, Spot 1 and Spot 2, are marked with two cross-symbols in Fig. 8. Spot 1 is located in the center of the cavity, while Spot 2 is located at the corner of the cavity. Fig. 9 shows that heat transfer is more efficient in the central area of the cavity (wide side) than in the corners. It can also be observed that the largest temperature difference within the mold cavity is about 10 8C and is reached 360 s after the beginning of heating. These temperature gradients are caused through the used insert at the middle of the cavity. The cooling channel within the insert is located closer to the cavity surface than within the rest of the mold. The shown difference in mold
Fig. 5. Temperature distribution in a HDPE-parison, for a die gap of 3 mm and a melt-temperature of 230 8C.
Fig. 6. Temperature distribution mPE-parison, for a die-gap of 3 mm and a melt-temperature of 200 8C.
A. Bendada et al. / NDT&E International 38 (2005) 433–441
437
Fig. 7. Temperature vertical profiles of the parison (mPE and HDPE). Fig. 10. Bottle wall thickness vs. cooling time for HDPE and mPE.
Fig. 8. Temperature distribution of the mold cavity after 360 s of heating time.
the bottle. Due to the magnetic behavior of measurement head and ball, the ball follows the measurement head, and the gap between ball and measurement head is determined. The wall thickness is measured along a specific line 80 mm above the bottom of the mold as shown in Fig. 10. Within the figure, the length of the line is normalized. It starts at the mid-wideside of the bottle (spot marked 1 in Fig. 11). The bottles wall thickness results show that the wall thickness (for both HDPE and mPE) is independent of the processing conditions with the exception of the cooling time (Fig. 10). As expected (due to the use of a cylindrical die), the shorter side (higher blow ratio) has a lower thickness than the wider side (lower blow ratio). It can also be seen that the thickness profiles for the HDPE and mPE are very different, with the mPE having a
cooling will influence also the temperature distribution of the plastic part itself. After the production of the bottles, the wall thickness of the finished parts is measured with an inductive wall thickness instrument. It consists of a measurement head and a ball. The ball is put inside the bottle while the measurement head is driven along the outer surface of
Fig. 9. Temperature development for two specific spots of the mold cavity.
Fig. 11. Line of wall thickness measurement.
438
A. Bendada et al. / NDT&E International 38 (2005) 433–441
short side with almost uniform thickness. This is probably due to the mPE being able to stretch into the corners even after having touched the mold wall at the large side of the bottle. Furthermore, due to its low swell the mPE has lower overall thickness values. The temperature distribution of the plastic parts is measured directly after mold opening, while the bottle is remaining for some seconds at the blow pin. As for the wall thickness, the influence of the cooling conditions for HDPE and mPE is investigated. To get a qualitative comparison, the infrared images are recorded and compared. A typical infrared image of the HDPE bottle produced during the investigation of the inflating pressure effect is reported in Fig. 12. The quantitative temperature measurement of the bottles surface was made along a line covering about a quarter of its circumference. This specific line is shown in the infrared image of Fig. 12. The line starts at the middle of the short site of the bottle and ends at the middle of the wide side of the bottle. A nondimensional distance (0Zmid-short-side; 1Zmid-wideside) is used to display the results. The wall thickness along this line is shown together with the temperature in the same diagram to give an idea of its influence on the temperature distribution. It was observed that the temperature distribution was directly proportional to the die gap used. For smaller die gaps, a more homogeneous temperature distribution at mold opening is obtained. Higher melt temperatures only resulted in a shift of the temperature at mold opening without any change in the shape of the spatial distribution. Figs. 13 and 14 show the temperature and the wall thickness distributions as function of various inflation pressures for HDPE and mPE, respectively. These results show the potential of infrared thermography to detect material characteristics and flaws in the process. As expected from a non-axisymmetric blowing (motor oil bottle), the short side (thinner) is colder than the wider side (thicker). The temperature distribution can also locate the thin spots corresponding to the edges of the part (minimum temperature). Furthermore, it can pinpoint areas in
the mold where cooling is inefficient [10,11]. As was mentioned before, the location of the cooling channels creates a more efficient cooling in the wide side of the bottle. This explains the higher temperatures at the thinner sections. From the material point of view, the temperature distributions was able to detect the behavior of the material during blowing. MPE requires lower pressures (stresses) to stretch and therefore the temperature distribution is independent of the blowing pressure. On the other hand, the HDPE is much stiffer and at the lower pressures it cannot stretch enough to make full contact with the mold wall. Air trapped between the plastic and the mold could render the analysis of the results even more complicated. The effect of cooling time on the temperature distribution of the part is shown in Fig. 15. Longer cooling times produce parts having a more uniform temperature at demolding. This is particularly the case for parts having thinner walls (mPE vs. HDPE). The different stretching of the material and the different polymer/mold contact conditions will have an effect on the warpage of the parts. An analysis of the temperature development after
Fig. 12. A typical infrared image of the HDPE bottle as it is ejected out of the mold.
Fig. 14. Temperature and wall thickness vs. inflation pressure for mPE.
Fig. 13. Temperature and wall thickness vs. inflation pressure for HDPE.
A. Bendada et al. / NDT&E International 38 (2005) 433–441
439
Fig. 18. Illustration of the temperature compensation.
Fig. 15. Temperature and wall thickness vs. cooling time for mPE.
subsequent cooling in ambient air provides no additional information. Comparison of infrared images of the HDPE bottle recorded 11 and 131 s after mold opening (Figs. 16 and 17), shows that just temperature compensation takes place. This can be clearly seen on the temperature horizontal profiles extracted from the previous infrared images and plotted in Fig. 18. Indeed, the profiles show that there is not
only a heat transfer to the ambient air, which would lead to a decrease of temperature all over the line; there is also a heat transfer in the surface of the bottle direction. The temperature decreases at the middle of the line, normalized length 0.45, but increases at the ends of the line due to heat conduction. The influence of the operating conditions on the shrinkage and warpage of the finished bottles is shown in Figs. 19 and 20 for HDPE and mPE, respectively. The dimensions of the bottle were obtained from pictures taken with a digital camera and compared with the cavity dimensions via image superposition on a grid. It is clear
Fig. 16. Temperature distribution 11 s after mold opening, HDPE.
Fig. 17. Temperature distribution 131 s after mold opening, HDPE.
Fig. 19. Part geometry vs. cooling time (HDPE).
440
A. Bendada et al. / NDT&E International 38 (2005) 433–441
the process. The temperature distribution can locate the thin spots corresponding to the edges of the part (minimum temperature). Furthermore, it can pinpoint areas in the mold where cooling is inefficient. From the material point of view, the temperature distributions were able to detect the behavior of the material during blowing. mPE requires lower pressures (stresses) to stretch and therefore the temperature distribution is independent of the blowing pressure. On the other hand, the HDPE is much stiffer and at the lower pressures it cannot stretch enough to make full contact with the mold wall. It is clear that non-uniform temperature distributions (shorter cooling times) at demolding will favor warping of the bottles, particularly for materials that have been under relative high stresses (HDPE). Shorter cooling times increase the time over which the parts cool down under stress free conditions. Finally, the results can also be used in the validation of process simulation models. They will certainly be helpful in establishing initial values for the cooling and shrinkage and warpage simulations. Fig. 20. Part geometry vs. cooling time (mPE).
that non-uniform temperature distributions (shorter cooling times) at demolding will favor warping of the bottles, particularly for materials that have been under relative high stresses (HDPE). Shorter cooling times increases the time over which the parts cool down under stress free conditions.
Acknowledgements The authors would like to thank the financial support of the DAAD (Deutscher Akademischer Austauschdienst/German Scientific Exchange Service). Also, they wish to thank M.A. Rainville and M. Carmel of the Industrial Materials Institute (IMI) of the National Research Council of Canada (NRC) for their help during the molding trials.
6. Conclusions The influence of different operating conditions on the cooling phase of the extrusion blow molding process of HDPE and mPE has been investigated. The temperature distribution of the bottles, just after mold opening and after subsequent cooling in ambient air has been measured using infrared thermography. The cooling design of the mold has also been investigated with infrared thermography. It has been shown that in order to have a reliable analysis of the cooling phase of the blow molding process, an accurate description of the dimensions of the parison (swell and sag effects) as well as the temperature distribution in the mold (cooling channels effect) is required. The temperature distribution was found to be directly proportional to the die gap used. For smaller die gaps, a more homogeneous temperature distribution at mold opening is obtained. Higher melt temperatures only resulted in a shift of the temperature at mold opening without any change in the shape of the spatial distribution. The results have also shown the potential of infrared thermography to detect material characteristics and flaws in
References [1] Delorenzi HG, Nied HF. Blow molding and thermoforming of plastics: finite element modeling. Comput Struct 1987;26(1–2): 197–206. [2] Laroche D, Erchiqui F. 3D modeling of the blow molding process. In: Proceedings of NUMIFORM’98 conference, Enschede, Netherlands; 1998, p. 483–8. [3] Laroche D, Erchiqui F. Experimental and theoretical study of the thermoformability of industrual polymers. J Reinf Plast Compos 2000;19(3):231–9. [4] Derdouri A, Erchiqui F, Bendada A, Verron E, Peseux B. Viscoelastic behavior of polymer membranes under inflation. In: Proceedings of rheology’2000, XIII international congress on rheology, Cambridge, UK; 2000, p. 394–6. [5] Prystay M, Garcia-Rejon A. Application of thermography for the optimization of the blow molding process. In: Proceedings of ANTEC conference, vol. 1. Brookfield, CT, USA: Society of Plastics Engineers; 1996, p. 996–1000. [6] Prystay M, Garcia-Rejon A. Using thermography to optimize the blow-molding process. Plast-Eng 1996;52(12):41–3. [7] Prystay M, Wang H, Garcia-Rejon A. Application of thermographic temperature measurements in injection molding and blow molding of plastics. In: Proceedings of SPIE, vol. 2766. Bellingham, WA, USA: Society of Photo-Optical Instrumentation Engineers; 1996, p. 5–14.
A. Bendada et al. / NDT&E International 38 (2005) 433–441 [8] Gaussorgues G. Infrared thermography. London: Chapman & Hall; 1994. [9] Dewitt DP, Nutter GD. Theory and practice of radiation thermometry. New York: Wiley; 1988. [10] Prystay M, Loong CA, Nguyen K. Optimization of cooling channel design and spray patterns in aluminum die casting using infrared
441
thermography. In: Proceedings of SPIE, vol. 2766. Bellingham, WA, USA: Society of Photo-Optical Instrumentation Engineers; 1996, p. 15–4. [11] Attar A, Bendada A, Connolly R, De´nomme´ Y, Dallaire E, Aitcin PC. Novel ultra-high-performance concrete as mold material for the blow molding process. SAMPE 2000;36(3):45–9.