Printer Friendly

The Effects of Kneading Block Design and Operating Conditions on Distributive Mixing in Twin Screw Extruders.

A mixing limited interfacial reaction between polymer tracers was used to directly measure the distributive mixing performance of a co-rotating twin screw extruder during melt-melt blending of polypropylene. The reaction between the polymer tracers, which are low molecular weight succinic anhydride and primary amine terminally functionalized polymer chains, was followed using Fourier-Transform Infrared Spectroscopy (FT-IR). Experiments were completed to determine the effects of flow rate, screw speed, and kneading block design on the distributive mixing performance and the residence time distribution (RTD). The only RTD variable that was significantly affected by the experimental factors was the average residence time. Distributive mixing with neutral and reverse kneading blocks was controlled by the average residence time, the fully filled volume, and the shear rate. Conversely, the mixing performance of a forward kneading block did not follow the same trends.

INTRODUCTION

Dispersive mixing, or the breakup of cohesive agglomerates, during polymer blending in twin screw extruders has been studied using a screw pulling-out technique [1], direct sampling using specially designed barrel plates [2], and direct sampling from the screw channel using a clam shell barrel [3]. Reduction in the minor phase domain size is a direct measurement of the dispersive mixing performance of the extruder. Most research on dispersive mixing has focused on the morphology evolution from pellets to the final blend state. Sundararaj et al. [4] showed that morphology development involves the formation of sheets from pellets, which then break up into cylinders and droplets. Bourry and Favis [5] determined that morphology development was not solely determined by the melting process because rapid morphology development was also observed over a very short distance in the extruder during melt-melt blending. Melt-melt blending in single screw extruders with different mixing elements has also been recently inves tigated [6, 7]. Distributive mixing, which involves the homogenization of a mixture of two or more components, plays an important role in polymer blending, compounding, and reactive extrusion processes. Distributive mixing is more difficult to experimentally investigate because there is no direct indication of the mixing performance that is easy to quantitatively measure. To overcome this problem, special extruders with transparent barrels have been used to visually investigate the distributive mixing of model materials [8-10].

Research has also focused on comparing the mixing capabilities of different twin screw extruder mixing elements. Kalyon and Sangani [11] showed that interfacial area generation during the blending of a thermoplastic elastomer occurred faster with a reverse kneading block as compared with a forward kneading block. As well, the amount of axial mixing was greater for the forward kneading block. Christiano and Lindenfelzer [12] used dynamic pressure measurements to gain evidence of the flow patterns and mixing in kneading blocks staggered in the forward, neutral, and reverse directions. Lawal and Kalyon [13, 14] have numerically investigated the flow in kneading blocks, using residence time distributions and intensity of segregation as measures of mixing. Cheng and Manas-Zloczower [15] have performed FEM simulations of the flow in kneading discs to compare the dispersive mixing efficiency of trilobal and bilobal designs. Experimental and numerical results from the literature indicate that there are significant di fferences in the flow patterns and mixing capabilities of different kneading block designs, but experimental evidence is limited.

Mixing in conventional chemical reactors has been studied using fast bimolecular reactions [16] and special parallel reactions [17,18]. Application of these methods to the investigation of mixing during polymer processing is very limited. The reaction between oligomers with alcohol and primary amine functional groups and ethylene-propylene rubber with grafted maleic anhydride was used in an attempt to study the mixing rates in twin screw extruders [19]. Wu [20] used low molecular weight tracers to study the micromixing of a polyethylene wax in a block of trilobal kneading discs, but the effects of immiscibility between the polar tracers and the hydrophobic polymer were not discussed. Curry and Andersen [21] investigated the effects of extruder operating conditions on the mass transfer controlled reaction between oxazoline functionalized polystyrene and acrylic acid functionalized polyethylene Crosslink formation between the reactive polymers resulted in a viscosity increase, which was dependent on the screw speed and the residence time. Two different morphologies were observed at the same extent of reaction, and therefore, the crosslink density was not directly related to the quality of mixing.

The objective of this work was to use a chemical reaction between polymer tracers to directly measure the distributive mixing performance of a twin screw extruder during melt-melt blending of polypropylene (PP). This work is a continuation of previous research on melt-melt blending in twin screw extruders [22,23], with the current focus on determining the effects of flow rate, screw speed, and kneading block design on the distributive mixing. As well, residence time distribution measurements were completed to investigate possible relationships between macromixing and distributive mixing.

EXPERIMENTAL

Materials

The material used in the experiments was a PP resin supplied by Equistar (Petrothene PP8000GK) having a melt flow index of 5 g/10 min (ASTM D1238 condition L). The polymer tracers were based on Polymer C-SYN (Crowley Chemical Co., [[bar{M}].sub.n] = 3500, [[bar{M}].sub.w] = 12000, 5wt% ethylene/95 wt% propylene, vinylidine concentration = 3 x [10.sup.-4] mole/g polymer). The following chemicals were obtained for the polymer functionalization: maleic anhydride, hydroxylamine-o-sulfonic acid, and 9-BBN (0.5 M 9-borabicyclononane solution in THF).

Reactive Tracer Preparation

Functionalization of Polymer C-SYN with succinic anhydride and primary amine functional groups was completed according to previously published techniques [22, 23]. Terminal anhydride functionalization was completed by a melt-phase Alder Ene reaction with maleic anhydride. Primary amine functionalization was completed in refluxing THF by performing hydroboration with 9-BBN followed by amination with hydroxylamine-o-sulfonic acid. A detailed characterization of the reactive tracers and their coupling reaction can be found elsewhere [23]. FT-IR analysis of the interfacial reaction between the polymer tracers was completed using the anhydride functional group characteristic peak located at 1793 [cm.sup.-1] [24].

Equipment for Melt-Melt Blending

The reactive tracers were blended into PP at a concentration of 25 wt% using a batch mixer (Haake Rheomix 3000), and then blended down in a twin screw extruder to a concentration of 5 wt%. Two extruders were required to perform the melt-melt blending experiments. One of the PP streams was metered to the first barrel position of the Leistritz LSM 30.34 co-rotating twin screw extruder (D = 34 mm, L/D = 35) using a K-Tron feeder. The second PP stream was melt fed at a desired downstream position using a single screw extruder (Haake Rheomex 252, D = 19 mm, L/D = 25), with an adapter (inner diameter = 1.9 cm, injection point located directly over the intermeshing region). The screw configuration and feeding arrangement is depicted in Fig. 1. The screw configuration of the melt-melt blending section consisted of conveying elements and a kneading block (16 bilobal kneading discs, thickness = 7.5 mm, followed by a reverse conveying element, length = 30 mm, pitch = 20 mm). Pressure transducers were used to determine the locations of fully filled regions in the extruder, and the transducer positions are indicated in Fig. 1. In all experiments, the pressure was observed to be zero immediately after the reverse conveying element (P4), which indicated that the screw channel was partially filled at that location. Therefore, the degree of fill and the flow patterns in the kneading block were not affected by the polymer backup length prior to the die. The polymer was extruded through a slit die into a water bath, and the produced film (average thickness = 1 mm) was used directly for analysis of the interfacial reaction using a Nicolet 520 mid-range FT-IR spectrometer (500-4000 [cm.sup.-1]).

Experimental Factors and Procedures

The experiments were completed using the same PP in both melt streams, which eliminated the effect of interfacial tension on the mixing. The three experimental factors, which are machine parameters, were the design of the kneading block, the screw speed (50-200 rpm), and the total flow rate (50 g/min, 100 g/min). The stagger angle of the kneading discs was varied between 30[degrees] (forward), 90[degrees] (neutral), and -30[degrees] (reverse). In all experiments, the barrel temperatures in the melt-melt blending section were set to 220[degrees]C, and equal flow rates were used in the two segregated streams (volumetric ratio of one).

The extruders were purged with pure PP prior to each experiment, after which, the desired flow rate and screw speed were set. The extruders were run for a minimum of 15 min prior to sampling the polymer extrudate, and then five films were collected and analyzed using the FT-IR spectrometer. From the FT-IR spectrum, the relative peak height (RPH) of the anhydride functional group (1793 [cm.sup.-1]) was calculated using the peak at 528.5 [cm.sup.-1] as an internal reference, which corresponds to the C-C-C bending, C-[CH.sub.3] stretching, and [CH.sub.2] rocking vibrations of the PP backbone [25]. To calculate anhydride conversion, the RPH of the anhydride functional group for the conditions of zero conversion and full conversion were required. For zero conversion, the PP/anhydride tracer melt stream was blended with a pure PP melt stream. To obtain full conversion of the interfacial reaction, 20 randomly selected polymer samples from the melt-melt blending experiments were heated in a vacuum oven at 120[degrees]C for 3 days. The polymer samples were then pressed into films using a hot press (5 min at 200[degrees]C and 7 MPa, film thickness = 0.9 mm). The obtained values were (presented as 95% confidence intervals) 2.170 [pm] 0.020 for zero conversion and 1.451 [pm] 0.023 for full conversion. Using these values, the following equation was developed to calculate the anhydride conversion from the measured RPH:

Anhydride Conversion = [frac{2.170 - RPH@1793}{2.170 - 1.451}] (1)

RESULTS AND DISCUSSION

Distributive Mixing Measurements

Melt-melt blending of the two segregated streams results in the generation of interfacial area. As shown in previous work [23], the interfacial reaction between the polymer tracers is controlled by mixing and not chemical kinetics. The reactive tracers come into contact and react at the growing interface, and therefore, the conversion measured at the die is directly related to the total amount of interfacial area generated, or the distributive mixing performance of the twin screw extruder. The terminal coupling of the low molecular weight tracers does not affect the properties of the interface. Conversely, the crosslinks formed between high molecular weight reactive polymers, as in the work of Curry and Andersen [21], will affect the interfacial properties and its deformabiity. Therefore, high molecular weight reactive polymers cannot be used to gain evidence of the mixing performance of a twin screw extruder because their reaction affects the mixing process.

The experimental results of the anhydride conversion are presented for the three screw configurations in Fig. 2. The experimental results are compared with regression models, and the linear regression results are discussed at the end of this section of the paper. The error bars correspond to the standard errors obtained from the five films collected at every set of operating conditions (error bars smaller than the data symbols are not visible in the figures). The average estimate of the experimental error for the anhydride conversion was 0.019. As well, replicate experiments were completed with the screw configuration containing the neutral kneading block at the conditions of 50 g/min and 75 and 100 rpm. Comparison of the replicate averages indicated that the error between separate experiments was less than 2%. Screw speed and flow rate exhibited negative effects on the mixing performance of the screw configuration with the forward kneading block. Conversely, screw speed exhibited a non-linear effect on the m ixing performances of the screw configurations with the neutral and the reverse kneading blocks. As well, flow rate exhibited a positive effect on the mixing performances of these two configurations. Clearly, mixing performance was dependent on the combination of the kneading block design and the operating conditions.

Efficient mixing occurs predominately in kneading blocks and fully filled conveying elements, which exhibit flow recirculation because of the negative pressure flow component. The pressure (P2) upstream of the kneading block (transducer P2 is located 15 mm upstream of the first kneading disc) is presented in Fig. 3. A higher pressure upstream of the kneading block corresponds to a longer polymer backup length, or a larger fully filled volume prior to the kneading block. As expected, the fully filled volume prior to the kneading block followed the trend of: reverse [greater than] neutral [greater than] forward. The pressure at the reverse conveying element following the kneading block (P3) was the same for all three screw configurations at equal operating conditions. This observation suggested that the melt temperature at the end of the kneading block was independent of the stagger angle for the investigated operating conditions. Screw speed and flow rate exhibited small positive effects on the pressure at P3, with the pressure increasing from a minimum of 1.1 MPa at 50g/min and 50 rpm to a maximum of 1.38 MPa at 100 g/min and 200 rpm. The pressure difference between P2 and P3 indicated that the forward kneading block generated pressure, and as expected, the neutral and the reverse kneading blocks consumed pressure.

Screw speed affects the mixing process by the following contributions: shear rate, residence time, and fully filled volume. The relationship between interfacial area growth and total applied shear strain is well known [26]. Total applied strain depends on the residence times and the shear rates in the extruder. The effect of increasing screw speed on distributive mixing is complicated by the competing effects of increased shear rate and decreased residence time. A more detailed explanation of the effect of average residence time is discussed in the final section of this paper. From Fig. 3, the fully filled volume in the melt-melt blending section with the neutral and the reverse kneading blocks decreased with an increase in screw speed. The non-linear screw speed effects (Figs. 2b and c) observed for the screw configurations with the neutral and the reverse kneading blocks were attributed in part to the competing effects of shear rate, residence time, and fully filled volume. Conversely, the mixing performan ce of the screw configuration with the forward kneading block may have been dominated by the effect of reduced residence time.

Flow rate affects the average residence time and the fully filled volume, which exhibit competing effects on the mixing process. The screw configurations with the restrictive neutral and reverse kneading blocks exhibited improved mixing at the higher flow rate, which suggested that the increase in the fully filled volume dominated the effect of reduced residence time. Conversely, the screw configuration with the forward kneading block exhibited poor mixing at the higher flow rate. Thus, there is a fundamental difference between the effects of operating conditions on the mixing performances of pressure consuming and pressure generating kneading blocks.

All three screw configurations exhibited viscous heating at screw speeds between 125 and 200 rpm. Viscous heating was manifested by an increase of 8 to 30[degrees]C in the barrel temperature at the location of the kneading block. As well, the melt temperature at the die (measured using an infrared probe) increased by 6 to 12[degrees]C. This temperature increase affected the reaction kinetics, but the interfacial reaction remained mixing limited. The magnitude of the temperature increase depended on the screw speed and the flow rate. Viscous heating may have contributed to the decrease in the mixing performances that were observed for all three screw configurations at the higher screw speeds.

The mixing performances of the three screw configurations are compared in Fig. 4. At 50 g/min, the mixing performance followed the trend of: forward [greater than] reverse [greater than] neutral. Conversely, at 100 g/min, the mixing performance followed the trend of: reverse [greater than] neutral [greater than] forward at the screw speeds of 50 and 100 rpm. At the screw speeds of 150 and 200 rpm, the mixing performances of all three screw configurations were approximately equal. From Fig. 4, the same level of distributive mixing can be obtained from many different combinations of the experimental factors. For example, an anhydride conversion of approximately 80% can be obtained using a forward kneading block with a flow rate of 100 g/min and a screw speed of 75 rpm, or a flow rate of 50 g/min and a screw speed of 130 rpm. The same level of mixing can be obtained using a neutral kneading block with a flow rate of 100 g/min and a screw speed of 140 rpm, or a flow rate of 50 g/min and a screw speed of 75 rpm. Finally, the same level of mixing can be obtained using a reverse kneading block with a flow rate of 100 g/min and a screw speed of 155 rpm, or a flow rate of 50 g/min and a screw speed of 115 rpm.

As compared with the screw configurations containing the pressure consuming kneading blocks, the mixing performance of the screw configuration with the forward kneading block was superior at 50 g/min and inferior at 100 g/min (at screw speeds of 50 and 100 rpm). From Fig. 3, pressure measurements suggested that the forward kneading block was partially filled for all the combinations of the operating conditions expect at 100 g/min and 50 rpm. Therefore, the break position between the fully filled and partially filled channels would be located in the forward kneading block. Conversely, the break position would be located in the conveying elements prior to the restrictive neutral and reverse kneading blocks. As suggested in previous work [23], the break position in the forward kneading block may explain its superior mixing performance. The new results at the high flow rate contradict this explanation, and alternative explanations for the mixing behavior of the forward kneading block are offered in the final sec tion of this paper.

Linear regression analysis was completed to quantify the effects of screw speed (N) and flow rate (Q) on the mixing performance of each screw configuration. The regression models are compared with the experimental results in Fig. 2, and the model equations are presented in Table 1. All the terms included in the model equations were significant at the 95% confidence level. In the case of the neutral kneading block experimental results, regression analysis was completed using all the data in a single model (1-piece) and with data partitioning (4-piece). As indicated in Fig. 2b, the 1-piece model cannot account for the sharp decrease in the mixing performance immediately after the maximum conversion. As well, the 1piece model cannot account for the effect of flow rate on the location of the maximum conversion. Therefore, the experimental results were partitioned into 4 pieces to improve the fit quality by including the discontinuities in the results at the maxima. The residual sum of squares for the 1-piece and 4-piece models were 4.85 X [10.sup.-3] and 2.12 X [10.sup.-3] respectively. The 56.3% reduction in the residual sum of squares indicated that data partitioning resulted in a much better fit to the experimental results.

The non-linear effect of screw speed on mixing with the neutral kneading block was incorporated into the 1-piece model with the 1/[N.sup.4] term, but the screw speed coefficients lacked physical meaning. Conversely, data partitioning revealed the true non-linear screw speed effects. At each flow rate, screw speed exhibited an initial positive effect on the mixing performance, which was attributed to an increase in shear rate. The sharp drop in the mixing performance after the maximum conversion was attributed to the decreases in residence time and fully filled volume, which may have dominated the effect of an increase in shear rate. The position of the maximum conversion increased from 75 rpm at 50 g/min to 100 rpm at 100 g/min. This shifting was attributed to an increase in the fully filled volume at the higher flow rate, which may have delayed the sharp decrease in mixing. The only disadvantage with data partitioning was that the effect of flow rate cannot be quantified.

The non-linear effect of screw speed on mixing with the reverse kneading block was not as well defined as in the case of the neutral kneading block. As well, no maximum conversion was observed at the high flow rate. Removal of the non-linear screw speed term (1/[N.sup.4]) resulted in a 24% increase in the standard error. Therefore, the non-linear term was significant as it accounted for the maximum conversion at the low flow rate. In the case of the forward kneading block, the regression analysis indicated that the effects of screw speed and flow rate were linear, but their interaction was also significant. Regression analysis confirmed the fundamental difference between the mixing behavior of the forward kneading block and the restrictive kneading block designs.

Residence Time Distribution (RTD) Measurements

RTD measurements were completed with the objective of relating the distributive mixing results to differences in macromixing, or the axial mixing between particles of different ages. For comparison with the distributive mixing experiments, only the RTD in the melt-melt blending section was required. The same tandem extruder system was used in the RTD measurements, with equal mass flow rates in the two segregated melt streams. Pure Petrothene PP was metered to the beginning of the twin screw extruder, and a 5 wt% blend of the anhydride tracer in Petrothene PP was melt fed using the single screw extruder. To measure the RTD, a washout of the anhydride tracer was implemented. After melt-melt blending for 20 mm, the mass flow rate of the pure PP stream was doubled. A pressure transducer was positioned immediately prior to the melt feeding location. A sharp increase in this pressure indicated when the new flow rate had reached the melt feeding location. Immediately, the flow from the single screw extruder was sto pped, and a stop watch was started. No surging in the flow at the die was observed in any of the experiments, which indicated that the flow rate remained constant. Sampling of the polymer film at the die was completed every 5 s for 5 min, and then every 30 s for an additional 5 mm for a total of 70 measurements for each distribution. The sampling was completed using a knife to make a small etch on the side of the film at the die face. FT-IR analysis was performed using the collected film, and the washout of the anhydride tracer was followed using the relative peak height at 1793 [cm.sup.-1]. The RTD measurements were completed with the same factor levels used in the distributive mixing experiments. It is important to note that the distributive mixing and RTD experiments were not completed under isothermal conditions. Increasing the screw speed results in greater viscous dissipation and a higher melt temperature. The melt temperature may affect the RTD results through the pressures and degree of fill in the ex truder.

RTD measurements are most commonly completed using a pulse test because it requires much less tracer material. To measure the RTD in the melt-melt blending section, a pulse test was not used because it would be impossible to inject a perfect pulse of the tracer uniformly across the polymer melt in the screw channels. The washout technique is also the best method for determining the moments of a RTD that has a significant tail [27]. Therefore, the selected experimental procedure should result in the best method for calculating the breadth of the RTD. The washout distributions were calculated from the FT-IR results using Eq 2, and the average residence times and variances were calculated using the well known moment Eq 3 and 4 [27].

W(t) = [frac{RPH@1793 (t) - RPH@1793 (final t)}{RPH@1793 (t = 0) - RPH@1793 (final t)}] (2)

[bar{t}] = [[[integral of].sup.final t].sub.t=0] W(t) dt (3)

[[sigma].sup.2] = 2 [[[integral of].sup.final t].sub.t=0] tW(t)dt - [([bar{t}]).sup.2] (4)

The RPH at 1793 [cm.sup.-1] for the time equal to zero and the final time were average values from the first and last five data points, respectively.

A disadvantage of the washout technique is that numerical differentiation of the experimental results is required to obtain the residence time density function. Any small errors in the experimental measurements of the washout function can lead to large errors upon differentiating. Therefore, comparison of the RTD results was completed using the cumulative distributions only. The cumulative distribution is not as sensitive as the residence time density function to small changes in the degree of macromixing. As well, it is more difficult to identify stagnation and recirculation using the cumulative distribution.

Examples of the measured distributions are presented in Fig. 5 for the screw configuration with the forward kneading block at a flow rate of 50 g/min. The small fluctuations in the washout curves were within the expected experimental error of the FT-IR measurements. The shifting of the washout curves to lower times indicated a decrease in average residence time with increasing screw speed. The dimensionless results for the same conditions are presented in Fig. 6. The differences in the slopes of the dimensionless washout curves indicated small differences in the degree of macromixing. A comparison of the dimensionless washout curves for the three screw configurations is presented in Fig. 7 for a screw speed of 50 rpm. The dimensionless curves overlapped significantly, which indicated that the kneading block design did not exhibit a significant effect on the degree of macromixing. This observation was consistent for all the investigated values of the screw speed.

The average residence times are presented in Fig. 8 for all three screw configurations. Replicate experiments were completed with the screw configuration containing the neutral kneading block at 50 g/min and 50 rpm. The calculated experimental error for the average residence time was less than 1%, which indicated that the experiment procedure was adequate. Linear regression was performed to quantify the factor effects, and the results are presented in Table 2. Comparison of the t-statistics with the corresponding values from the t-distribution indicated that all the factors were significant at the 95% confidence level. The kneading block design exhibited a larger effect on the average residence time at 50 g/min. This observation suggested that the degree of fill in the conveying section prior to the kneading block was a less significant factor for the average residence time at 100 g/min. As expected, an increase in the flow rate decreased the average residence time. The non-linear effect of screw speed on ave rage residence time was incorporated into the specific throughput (Q/N) term, which is an indication of the degree of channel fill. The trends for the screw configurations with the neutral and the reverse kneading blocks were similar, with slightly higher average residence times for the more restrictive reverse kneading block. The average residence times did not follow the same trend for the screw configuration with the forward kneading block. Along with the effects of specific throughput and flow rate, their interaction was also significant. Without the interaction term, the linear regression fit quality decreased significantly as indicated by the decrease in [r.sup.2]. As previously observed with the distributive mixing measurements, the average residence time results indicated a significant difference between the pressure generating kneading block and the pressure consuming kneading block designs.

The RTD spreads ([sigma]) and relative axial dispersions ([bar{t}]/[sigma]) are presented in Figs. 9 and 10, respectively. The RTD spread indicates the magnitude of the breadth of the distribution, while the relative axial dispersion, which is the ratio of the spread and the average residence time, indicates the level of macromixing. In the extreme cases, the relative axial dispersion is equal to zero for plug flow, and it is equal to one for perfect macromixing. From the replicate experiments completed with the screw configuration containing the neutral kneading block, the experimental errors in the RTD spread and relative axial dispersion were less than 2%. The RTD spread significantly decreased for each screw configuration as the flow rate was increased from 50 g/min to 100 g/min. Conversely, the spread exhibited no clear relationship with screw speed or specific throughput at each flow rate. Similar results have been discussed in the literature for the RID in twin screw extruders [28]. In contrast to the RTD spread, the relative axial dispersion increased with the flow rate, which may have contributed to the improved distributive mixing with the neutral and reverse kneading blocks at the high flow rate.

The RTD spreads and relative axial dispersions were averaged for each flow rate and screw configuration, and the results are presented in Table 3. The results suggested that at 50 g/min, the average breadth of the RTD followed the trend of: forward [greater than] reverse [greater than] neutral. This trend is identical to the distributive mixing performance trend at the low flow rate. Comparison of the dimensionless washout distributions did not confirm these differences in the degree of macro-mixing, but as previously mentioned, the cumulative distribution is not as sensitive as the residence time density function to small amounts of stagnation or re-circulation. At 100 g/min, the average breadth of the RTD was not significantly dependent on the kneading block design. Therefore, no general relationship existed between the distributive mixing results and the degree of macromixing.

Variables Controlling Distributive Mixing in Twin Screw Extruders

Distributive mixing and average residence time were shown to be dependent on the combination of the kneading block design and the operating conditions. The relationship between the measured distributive mixing performances and the average residence times is presented in Fig. 11. The mixing results from the three screw configurations form a group at each flow rate, which indicated that average residence time was not the only important factor. Two lines highlighting the experimental trends have been included in Fig. 11, which show that the distributive mixing performances of the screw configurations with the neutral and the reverse kneading blocks followed the same trend with respect to average residence time over some interval at each flow rate. The overlapping of the mixing results from these two screw configurations suggested that average residence time was the controlling variable in those intervals. After the maximum conversions for the screw configurations with the neutral and the reverse kneading blocks , the mixing results deviated from the linear trends. The deviations suggested that the effect of reduced shear rate dominated the effect of increased residence time at low screw speeds. After the maxima, the screw configuration with the reverse kneading block exhibited better mixing because of its larger fully filled volume. At the low flow rate, the screw configuration with the forward kneading block exhibited superior mixing in the same range of average residence times.

Steady-state macroscopic numerical simulations of the twin screw extruder process were performed using a program developed by Strutt et al. [29, 30]. The program, which is based on the simulation method of Potente et al. [31], yields pressure, filling level, melt temperature, and solid fraction profiles along the extruder for the non-isothermal extrusion of a power-law fluid. Shear viscosity measurements were completed using a Kayeness Galaxy V capillary rheometer (capillary diameter = 0.762 mm, L/D = 40) at 190, 220, and 250[degrees]C. A temperature dependent power-law viscosity model was fitted to the data ([r.sup.2] = 0.9952) , which is presented in Eq 5.

[eta] = 35985 exp(-8.52*[10.sup.-3]*T) [[dot{[gamma]}].sup.(0.348-1)]

[eta][=] Pa.s

where T[=] C (5)

[dot{[gamma]}][=] [S.sup.-1]

The other material properties required in the simulations were taken from the work of Tzoganakis et al. [32] and Wang et al. [33-35] for a generic PP resin.

Simulations were completed for all three screw configurations with the values of the operating conditions used in the experiments. Compared to the experimentally measured values, the simulations predicted higher pressures, which will result in over estimations of the fully filled volume. The simulations predicted that the forward kneading block was always fully filled, which was not supported by the experimental pressure measurements. Even though the predicted pressure values were higher, they followed the same trends as presented in Fig. 3 for the screw configurations with the neutral and the reverse kneading blocks. Therefore, the predicted trends of the fully filled volume remain valid.

The relationship between the experimental anhydride conversion results and the predicted fraction of fully filled channels in the melt-melt blending section is presented in Fig. 12. As expected the fully filled volume followed the trend of: reverse [greater than] neutral [greater than] forward. The results from the two flow rates for the screw configurations with the neutral and the reverse kneading blocks overlapped (2 lines highlighting the trends have been included in Fig. 12), which suggested that the flow rate affected the mixing performances through the fully filled volume. Slightly improved mixing was observed for the screw configuration with the reverse kneading block, which was attributed to its corresponding larger fully filled volume. The deviations in the mixing performances after the maximum conversions were attributed to the decrease in shear rate dominating the increase in fully filled volume at low screw speeds. Once again, the results for the screw configuration with the forward kneading blo ck did not follow a similar trend.

As discussed previously, the screw configuration with the forward kneading block exhibited surprisingly good distributive mixing at the low flow rate. The fully filled volume prior to the forward kneading block was significantly less than that prior to the pressure consuming kneading blocks. At the same time. the average residence times for the screw configuration with the forward kneading block were similar to the values for the screw configurations with the neutral and the reverse kneading blocks. The overall residence time can be considered to consist of the following three contributions: the time spent in the conveying elements prior to the kneading block, the time spent in the kneading block, and the time spent in the pumping section prior to the die. The fraction of the overall residence time spent in the pumping section prior to the die was independent of the kneading block design. A higher fully filled volume prior to the neutral and the reverse kneading blocks corresponded to a larger fraction of th e overall residence time spent in that section. Therefore, the local residence time in the forward kneading block may have been higher than the local residence times in the neutral and the reverse kneading blocks. The most efficient mixing will occur in the kneading block, which may explain the superior mixing performance of the screw configuration with the forward kneading block at the low flow rate.

The average values of the spread and relative axial dispersion suggested that the screw configuration with the forward kneading block had a broader RTD at the low flow rate, although the differences were not observed in the dimensionless washout distributions. Using numerical simulations, Lawal and Kalyon [13] observed the possibility of stagnation in forward kneading blocks with small staggering angles. The improved mixing may be related to a small amount of material which remains in the forward kneading block for significantly longer times. At the higher flow rate, the inferior mixing performance of the screw configuration with the forward kneading block (at screw speeds of 50 and 100 rpm) may have been caused by its lower fully filled volume and shorter average residence time as compared with the other two screw configurations. As well, the averages of the relative axial dispersion of all three screw configurations were approximately equal at the higher flow rate, which suggested that the stagnation in th e forward kneading block existed only at the lower flow rate.

In this work, the total amount of interfacial area generated during melt-melt blending was related to the kneading block design and the operating conditions. In the future, analysis of the mixing along the extruder channels will be completed using melt sampling plates. Following the generation of interfacial area along the extruder channels will allow for comparison of the earliness of mixing between different screw configurations. As well, the earliness of mixing will be related to the experimentally measured fill distribution in the melt-melt blending section. In addition, local residence time measurements will be attempted to investigate the possible flow stagnation in the forward kneading block and to confirm the residence time trends.

CONCLUDING REMARKS

An interfacial reaction between polymer tracers was used as a microscopic probe to gain direct evidence on the effects of kneading block design, screw speed, and flow rate on the distributive mixing performance of a twin screw extruder. To obtain good distributive mixing. the selection of the screw configuration and the operating conditions cannot be made independently. Experimental results indicated significant differences in the distributive mixing performances of pressure generating and pressure consuming kneading block designs. The only RTD parameter that was significantly affected by all the experimental factors was the average residence time. The average residence time and the fully filled volume were shown to be controlling factors for the distributive mixing with neutral and reverse kneading blocks prior to maximum conversions at each flow rate. After the maxima, reductions in shear rate dominated the effects of increased average residence time and fully filled volume. The superior mixing of the screw configuration with the forward kneading block at the low flow rate was attributed to a possible stagnation and a longer local residence time in the kneading block.

ACKNOWLEDGMENTS

Financial assistance from the Natural Sciences and Engineering Research Council of Canada (NSERC) is gratefully acknowledged.

REFERENCES

(1.) S. Lee and J. L. White, Intern. Polym. Proc., 12, 316 (1997).

(2.) O. Franzheim, M. Stephan, T. Rische, P. Heidemeyer, U. Burkhardt, and A. Kiani, Adv. Polym. Technol., 16, 1 (1997).

(3.) C. G. Gogos, M. Esseghir, D. B. Todd, and D. W. Yu. Macromol. Symp., 101, 185 (1996).

(4.) U. Sundararaj, Y. Dori, and C. W. Macosko, Polymer, 36, 1957 (1995).

(5.) D. Bourry and B. D. Favis, Polymer, 39, 1851 (1998).

(6.) M. Gale, Adv. Polym. Tech, 16, 251 (1997).

(7.) M. Esseghir, C. G. Gogos, D. Yu, D. B. Todd, and B. David, Adv. Polym. Tech., 17, 1 (1998).

(8.) R. Sanchez, D. I. Bigio, and S. Karnik, SPE ANTEC Tech. Papers. 43, 3724 (1997).

(9.) D. Bigio, W. Baim, and K. Wang. SPE ANTEC Tech. Papers, 40, 256 (1994).

(10.) D. Bigio and L. Erwin, Polym. Eng. Sci., 32. 760 (1992).

(11.) D. M. Kalyon and H. N. Sangani, Polym. Eng. Sci., 29, 1018 (1989).

(12.) J. P. Christiano and M. Lindenfelzer, SPE ANTEC Tech. Papers, 43, 78 (1997).

(13.) A. Lawal and D. M. Kalyon, Polym. Eng. Sci., 35, 1325 (1995).

(14.) A. Lawal and D. M. Kalyon, J. Appl. Polym. Sci., 58, 1501 (1995).

(15.) H. Cheng and I. Manas-Zloczower, Polym. Eng. Sci., 37, 1082 (1997).

(16.) U. V. Shenoy and H. L. Toor, AIChE J., 36, 227 (1990).

(17.) J. R Bourne, F. Kozicki, and P. Rys, Chem. Eng. Sci., 36, 1643 (1981).

(18.) M.-C. Fournier, L. Falk, and J. Villerrnaux, Chem. Eng. Sci., 51, 5053 (1996).

(19.) C. Maier, M. Lambla, and D. Bigio, SPE ANTEC Tech. Papers, 40, 123 (1994).

(20.) J. Wu, PhD Dissertation, University of Delaware, Delaware (1994).

(21.) J. Curry and P. Andersen. SPE ANTEC Tech. Papers, 36, 1938 (1990).

(22.) G. Shearer and C. Tzoganakis, SPE ANTEC Tech. Papers, 44, 55 (1998).

(23.) G. Shearer and C. Tzoganakis, Polym. Eng. Sci., 39, 1584 (1999).

(24.) L. J. Bellamy, The Infrared Spectra of Complex Molecules, John Wiley and Sons, Ltd., New York (1958).

(25.) P. C. Painter, M. M. Coleman, and J. L. Koenig. The Theory of Vibrational Spectroscopy and its Application to Polymeric Materials, John Wiley and Sons, Ltd., New York (1982).

(26.) H. E. H. Meijer and J. M. H. Janssen, in Mixing and Compounding of Polymers: Theory and Practice, Ch. 4, I. Manas-Zloczower and Z. Tadmor, eds., Hanser Publishers, Munich (1994).

(27.) E. B. Nauman and B. A. Buffham, Mixing in Continuous Flow Systems, Wiley-Interscience Publication, New York (1983).

(28.) R. Gendron, L. E. Daigneault, J. Tatibou[ddot{e}]t, and M. M. Dumoulin, Adv. Polym. Tech., 15, 111 (1996).

(29.) D. Strutt, PhD Dissertation, University of Waterloo, Waterloo, Ontario, Canada (1998).

(30.) D. Strutt, C. Tzoganakis, and T. A. Duever, SPE ANTEC Tech. Papers, 43, 146 (1997).

(31.) H. Potente, J. Ansahi, and B. Klarholz, Intern. Polym. Proc., 9, 11 (1994).

(32.) C. Tzoganakis, J. Vlachopoulos, and A. E. Hamielec, Intern. Polym. Proc., 3, 141 (1988).

(33.) X. C. Wang, PhD Dissertation, University of Waterloo, Waterloo, Ontario, Canada (1996).

(34.) X. Wang, C. Tzoganakis, and G. L. Rempel, J. Appl. Polym. Sci., 61, 1395 (1996).

(35.) X. Wang, C. Tzoganakis, and G. L. Rempel, SPE ANTEC Tech. Papers, 43, 2066 (1997).
COPYRIGHT 2000 Society of Plastics Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

 Reader Opinion

Title:

Comment:



 

Article Details
Printer friendly Cite/link Email Feedback
Author:SHEARER, GIFFORD; TZOGANAKIS, COSTAS
Publication:Polymer Engineering and Science
Geographic Code:1USA
Date:May 1, 2000
Words:6867
Previous Article:Experimental Study on Extrudate Swell and Die Geometry of Profile Extrusion.
Next Article:In-Line Monitoring of Polyethylene Density Using Near Infrared (NIR) Spectroscopy.
Topics:


Related Articles
Numerical nonisothermal flow analysis of non-Newtonian fluid in a nonintermeshing counter-rotating twin screw extruder.
Transport in a twin-screw extruder for the processing of polymers.
Polymer blend mixing and dispersion in the kneading section of a twin-screw extruder.
Scale-up rules for mixing in a non-intermeshing twin-screw extruder.
Simulation of thermal degradation, peroxide induced degradation, and maleation of polypropylene in a modular co-rotating twin screw extruder.
Study of mixing efficiency in kneading discs of co-rotating twin-screw extruders.
A new approach to analyzing residence time and mixing in a co-rotating twin screw extruder.
Mathematical modeling and experimental studies of twin-screw extrusion of filled polymers.
Temporal Distributions: The Basis for the Development of Mixing Indexes for Scale-up of Polymer Processing Equipment.
Relationship between local residence time and distributive mixing in sections of a twin-screw extruder.

Terms of use | Copyright © 2014 Farlex, Inc. | Feedback | For webmasters