Int J Mater Form (2008) Suppl 1:783 –786 DOI 10.1007/s12289-008-0292-9 © Springer/ESAFORM 2008
Optimization and modelling of rotational molding process E. Perot 1, K. Lamnawar 1, A. Maazouz1 1
Ingénierie des Matériaux Polymères-(IMP/LMM), UMR-CNRS 5223, INSA-Lyon, F-69621, 17 Avenue Jean Capelle, Villeurbanne Cedex, France e-mail:
[email protected]
ABSTRACT: Rotational Molding is the best method for producing large hollow plastic articles without weld lines, such as kayak, tanks, large bins, etc. But it is a quite complex and empirical process. Constant quality in technical parts requires the mastery of the process by controlling on line the main physical phenomena. One of these of first importance is heat transfers. During the processing time, polymer powder melts, then the phenomena of particle coalescence and melt densification occur. After cooling, the molded part is obtained. The understanding of sintering phenomenon, linked to polymer structure, may explain surface defects and bubbles in rotationally molded parts. This presentation is divided into two parts: the first part deals with the relationship between the material structure, the process and the final properties; the second part deals with the modelization of heat transfers during the process. Firstly, material properties such as polymer structure, rheological parameters and surface tension were studied and linked to sintering kinetics. The sintering phenomenon was investigated, including coalescence and densification, in order to examine the effect of particle size and shape. Moreover, existing sintering models were compared to experimental data and were improved. Secondly, sintering kinetics were bound to final parts properties. Indeed, samples were molded with a pilot-scale rotational molding machine. On these samples, some defects were detected such as inner roughness and bubbles. Samples mechanical properties were also studied. The effect of particle size and shape on samples properties was examined taking into account the sintering. Quantitative relationships were stablished. This work enabled us to model the sintering phenomenon and to bind its kinetics with polymer structure, rheological properties and final parts properties. An experimental analysis of heat transfer in rotational molding process was also lead. By using an instrumented mold associated with an original radio transmission data acquisition system, we demonstrate that the rotational nature of the process implies complex heat transfer evolutions in the mold. The crystallization of the material was modeled with accuracy by coupling heat transfer equation to a kinetic model determined by calorimetry. Moreover, a thermal model was developped by using a static heated plate in order to validate the numerical results. This modelization took into account the sintering phenomenon. Key words: rotational molding – sintering –coalescence – densification –thermal model – heat transfers. 1 INTRODUCTION Rotational Molding is the most adequate method for producing large hollow plastic articles without weld lines. The process involves at least four steps: i) melting of the polymer powder, ii) coalescence of the fine particles, iii) melt densification and then iv) the cooling period. Some of major problems encountered during such a process are the surface defects and inhomogeneous densification of the bulk polymer. This leads to final articles characterized by a rough surface along with micro-and-macro voids and bubbles. Such problems occur mainly during the second step governed by sintering phenomena. The sintering phenomenon has been first studied for metals and ceramic materials [1],[2] and then the
approach has been extended to polymers[3]. Various models have been proposed in this area. Frenkel[4] described the rate of coalescence occurring by viscous flow promoted by surface tension for two Newtonian spherical particles having the same diameter, during the early stages of sintering. However such a model violates the continuity equation, which was corrected later (1949) by Eshelby [5]. Extension of Frenkel’s model was given by Hopper [6]. Pokluda et al. [7] extended the Eshelby-Frenkel approach to include the complete coalescence process of two spherical particles. Recently, Bellehumeur et al. [8] improved the Pokluda model so as to better fit the experimental results of the polypropylene-ethylene copolymer sintering. The Bellehumeur sintering model has used
784
the upper convected Maxwell model[9], assuming quasi-steady state flow, to obtain a non-linear differential equation. In this model, the sintering rate is a function of the material relaxation time and the characteristic sintering time: ηa0/γ, where η is the viscosity, a0 the initial particle radius and γ the surface tension. Studies about sintering[8, 10], densification[11] or rotomolded sample properties[12], rarely take into account the entire process chain, from the beginning to the end, i.e. from the polymer properties to final samples properties. In this work, polypropylene-ethylene copolymers have been studied. Although 90% of rotomolded products are made of polyethylene[13], the study of other materials will permit to develop new applications, such as automotive application and food packaging and reservoirs for solvents storage. Indeed, polypropylene-ethylene copolymers offer excellent environmental stress cracking resistance, high heat distorsion temperatures and scratch resistance, while having improved toughness, compared to homopolymer [12]. Sintering process has been studied [13] on various polypropylene-ethylene copolymer powders and pellets differing both in their shape, size and molecular weight. Rheological properties, surface tension and thermal properties during the rotomolding cycle were also examined. 2 EXPERIMENTAL PART 2.1 Materials Three commercial ethylene-propene copolymers (C1, C2 and C3) from Borealis were used in powder and micropellets form. The granules of the copolymer C2 were supplied in four different forms: fine powder C2a, medium size powder C2b, large size powder C2c, and micropellets C2d. C1 and C3 had the same mean size as the powder C2b. The particle size and particle size distribution were determined with a Rotap testing sieve shaker, according to ASTM D1921. The sieves mesh sizes ranged from 1108 microns to 106 microns. 100 g of each powder were shaken for 20 minutes and then the weight percentage of the powder retained on each sieve was calculated to determine the mean particles size. The size distribution was also determined by using a laser granulometer MasterSizer 2000® of MALVERN INSTRUMENTS®. The molecular weights and the
polymolecularity were measured by Size Exclusion Chromatography (SEC). The material composition and the blockiness were studied by 13C Nuclear Magnetic Resonance (NMR) spectroscopy. The melting points, crystallization points, heats of fusion, heats of crystallization and crystallinity were determined by Differential Scanning Calorimetry (DSC) analysis. 2.2 Rheological Properties Small amplitude oscillatory shear measurements were carried out using the strain controlled “Rheometric Scientific ARES®” with a parallel-plates geometry. The dynamic complex shear viscosity, the storage modulus and the loss modulus were determined in frequencies ranging from 10-1 rad/s to 102 rad/s. All rheological measurements were carried out in the linear viscoelastic region in small amplitude oscillatory shear mode as was verified by preliminary strain sweep measurements. Polymer degradation was avoided by continuous nitrogen purge in the oven. Zero shear viscosities were estimated by fitting the experimental data using the Yasuda-Carreau model. 2.3.
Sintering experiments
The sintering experiments were conducted by using a homemade regulated heat chamber equipped with a Zeiss® optical microscope and a Dalstar Pantera® CDD camera. Digital image analysis software was used to analyze the obtained images. Measurements were taken on two particles placed on a glass slide, under isothermal and non-isothermal conditions. Images of neck growth evolution were taken as a function of time to obtain the dimensionless neck radius (x/r), where x is the length of the neck between the two particles and r is the mean radius of the particles. Moreover, the Matlab® program permitted us to compare the experimental results with the Pokluda model and the Bellehumeur sintering model. 2.4.
Rotational Molding Experiments
A Rotoline® pilot-scale shuttle rotational molding machine was used to produce cylinders, with 200 mm in diameter and 300 mm in length, and also boxes with a volume of 250x250x100 mm3. The Datapaq® Tracker Telemetry System was used to measure, in real time, the temperatures in the oven, the mold, the molten polymer and the intern air
785
during the processing cycle.
during rotational molding, we can conclude that our experiments, with only one glass slide, are closer of the reality than the measurements that were usually performed.
3 MAIN RESULTS AND DISCUSSION 3.1 Sintering Figure 1 presents coalescence images as a function of time and temperature, during a non-isothermal experiment. Digital images for quantitative treatment are shown in parallel. Sintering curves of each copolymer are shown in Figure 2 (x/r as a function of time). Coalescence rate was calculated for each copolymer and it can be shown that the higher the copolymer viscosity, the lower is the coalescence rate. For each experiment, the Bellehumeur’s model was plotted, solving by the Runge Kutta method the Bellehumeur’s differential equation: ηa K 2 2 8 (αλ K1 ) θ ′2 + 2αλ K1 + 0 1 θ ′ − 1 = 0 γ K2 ε
(8)
ϖ
Fig. 1. Sintering images with Matlab image processing.
sin θ with K1 = and (1 + cos θ )( 2 − cos θ ) K2 =
2−5 3 cos θ sin θ
(1 + cos θ ) ( 2 − cos θ ) 43
53
and sin θ =
x r
where x is the length of the neck between the two particles; r is the mean radius of the particles, α is the coefficient for convected Maxwell model, equal to -1, 1 or 0; λ is the relaxation time; θ is the sintering angle (Figure 3). Comparison between the experimental results and Bellehumeur’s model is shown in Figure 4. The model captures the general trend of the variation of x/r as a function of time, but important disagreement with the experimental results is obtained. The effect of initial particle size on the time evolution x/r for C2 particles is shown in Figure 5. According to the Bellehumeur’s model, the finest particles should have the faster sintering rate. Our results show however the inverse trend; the finer particles showed the slower sintering rate. On the Figures 4, 5, we can observe that the Bellehumeur’s model is not in good agreement, quantitatively, with the experimental data. It can be due to the fact that we put the powder particles on a glass slide and not between two glass slides. Nevertheless, according to what happens
Fig. 2. Sintering curves of the three copolymers.
Fig. 3. Schematic sintering of two particles.
786
powders which have different sizes (fine: C2a, medium: C2b and large: C2c) are shown in Figure 9.
Fig. 4. Effects of the initial particle size on sintering kinetics.
Fig. 9. Temperature curves during rotational molding of C2a, C2b and C2c.
Fig. 1. Effects of particle shape on sintering kinetics.
3.2
Rotomolded Samples Properties and defects
Typical variations of temperature in time during the rotational molding process as measured by DATAPAQ telemetry tracker are obtained for samples. For all the experiments, identical process parameters were used: 12 minutes of heating, 12 minutes of stabilization and 30 minutes of cooling, with a rotation speeds equal to 4 rpm for the minor axis and 5 rpm for the major axis. The maximum temperature reached by the internal air was between 235 and 240 °C. The material temperature picks ranged from 235 to 238°C. Temperatures of various parts of the rotational mold are shown. The temperature of the sample during the whole process lies between the temperature of the surface of the mold and that of the inner air. During the heating step, the sample has quite the same temperature as that of the inner air during the first stages of heating, the temperature of the surface of the mold being much higher. However, after a certain time needed for thermal equilibrium, the three temperatures merge at the same value. The time required for thermal equilibrium depends obviously on thermal conductivity of the polymer under study. In contrast, during the cooling step, the sample temperature is almost a simple average between the two temperatures. Variations of temperatures in time for
The obtained data show material effects on the intern air temperature curves. Indeed, the fusion time, from the moment when the first particles melt from the moment when all the particles are molten, is different according to the material size. For example, with C2a, the first particles begin to melt earlier than with the C2b and C2c. Molded samples with very good appearance were obtained. They did not have porosities on the outer surface and their thickness was homogeneous. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
G. C. Kuczynski, Adv. Colloid Interface Sci., 3, 275 (1972). N. M. Rahaman, Ceramic Processing and Sintering, Marcel Dekker, New York (2003). S. Mazur, Polymer Powder Technology, John Wiley and Sohn, New York (1995). J. Frenkel, J. Phys., 9, 385 (1945). Eshelby, J. Metals, 1, 806 (1949). Hopper, J. Am. Ceram. Soc., 67, C262-C264 (1984). O. Pokluda, C.T. Bellehumeur, J. Vlachopoulos, AlChE J., 43, 3253 (1997). C. T. Bellehumeur, M. Kontopoulou, J. Vlachopoulos, Rheological Acta, 37, 270 (1998). D. D. Joseph, Fluid dynamics of viscoelastic liquids, Springer, New York (1990). A. R. Tinson, PhD Thesis, McMaster University (2004) A. G. Spence, R.J. Crawford, Polym. Eng. Sci., 36, 993 (1996). (a) M. Kontopoulou, M. Bisaria, J. Vlachopoulos, Intern. Polym. Proc., 12, 165 (1997) . E. PEROT, A Maazouz, Journal of Polymer Engineering p 207 Vol. 27, No. 4, (2007).