Optimization of twin screw extrusion using

CFD for polymer/nanoclay composites

Cite as: AIP Conference Proceedings 1779, 030002 (2016); https://doi.org/10.1063/1.4965472
Published Online: 31 October 2016

Schuschnigg Stephan, Battisti Markus, Winkler-Ebner Joachim, Friesenbichler Walter, and Holzer Clemens

ARTICLES YOU MAY BE INTERESTED IN

3D-CFD-simulation of melting processes in a high-speed-extruder with solid-melt-separation

AIP Conference Proceedings 1593, 623 (2014); https://doi.org/10.1063/1.4873857

Fundamentals of twin-screw extrusion polymer melting: Common pitfalls and how to avoid
them
AIP Conference Proceedings 1664, 020007 (2015); https://doi.org/10.1063/1.4918387

Heat transfer in twin screw compounding extruders

AIP Conference Proceedings 1779, 030014 (2016); https://doi.org/10.1063/1.4965484

AIP Conference Proceedings 1779, 030002 (2016); https://doi.org/10.1063/1.4965472 1779, 030002

© 2016 Author(s).
 Optimization of Twin Screw Extrusion using CFD for
Polymer/Nanoclay Composites
Schuschnigg Stephan1, a), Battisti Markus2, b), Winkler-Ebner Joachim3,c),
Friesenbichler Walter2,d), Holzer Clemens1,e)
1
Department of Polymer Engineering and Science, Polymer Processing, Franz-Josef Strasse 18, 8700 Leoben -
Styria – Austria 2 Department of Polymer Engineering and Science, Injection Molding of Polymers - Styria –
Austria, 3 AGRU Kunststofftechnik GmbH - Upper Austria - Austria
a)
Corresponding author: [email protected]
b)
[email protected], c)[email protected], d)[email protected],
e)
[email protected]

Abstract. The aim of this research was the optimization of a co-rotating twin screw extruder for processing
polypropylene with layered silicates. The various input parameters of the process itself such as screw speed, flow rate,
temperature and pressure as well as the great variety of screw configurations lead to an enormous amount of experimental
runs for the optimization. By using CFD simulations some of these parameters can be varied without the need for
experiments. In order to verify the simulated results, experimental runs have been performed. The compounded materials
consisted of 90 wt% PP (Bormed DM55 pharm or BB 412 E both from Borealis) with 5 wt% compatibilizer (BYK Scona
TPPP 2112GA) and 5 wt% layered silicate (Rockwood Nanofil®5). The viscosity was measured using different
rheometers, as an input parameter for the isothermal simulations. The pressure profiles, mixing index and dissipative
energy input along the screw elements were simulated utilizing Polyflow simulation package from Ansys Inc.. The
regions where the simulations showed zero pressure where assumed to be starved regions. These regions were also be
verified in the “screw pull out” experiments. In addition the pressure was measured along the screw and compared to the
simulations. The experiments where done with the polymer nanocomposites injection molding compounder (PNC-IMC),
consisting of a Leistritz ZSE 27 MAXX 44D compounder and an Engel e-motion 740/180T injection molding machine.
The final part (tensile specimen or plates for SAXS measurements) were processed directly with the PNC-IMC with only
one plasticizing process. Instead of backward conveying elements, the optimized screw consisted of mixing and kneading
elements leading to lower pressures and lower dissipative energy inputs and a longer residence time. The exfoliation of
layered silicates and therefore the tensile strength of the material are often related to the shearing and residence time.
However, the longer residence time in these experiments, did not compensate the lower dissipative energy input. The
SAXS measurement also showed that the standard geometry had a higher exfoliation rate than the optimized screw.

INTRODUCTION
Many research groups are working on the topic of nano composites hoping to improve certain properties of
different polymers 1, such as mechanical 2,3, thermal, electrical 4, flammability 5,6 and rheological 7. The main
problem is to get a good dispersion of the nanoparticles in the highly viscous polymer melt, as many particles are not
in nanoscale when fed to the compounder.
The processing conditions in twin screw extrusion are very well investigated, either with analytical methods, and
recently more with CFD computational fluid dynamics 8,9, where the results can be seen in three dimensions and
therefore a better understanding of the actions in the screw channels is achieved. Kohlgrüber 10 wrote a
comprehensive book about the compounding with co-rotating twin screw extruders, where both mathematical
approaches are well described.
With an injection molding compounder 11 the polymer was molten and with a side feeder the layered silicates
was incorporated into the channels. The mixed compound was pumped to an injection molding machine, where

Proceedings of the Regional Conference Graz 2015 – Polymer Processing Society PPS
AIP Conf. Proc. 1779, 030002-1–030002-5; doi: 10.1063/1.4965472
Published by AIP Publishing. 978-0-7354-1441-9/$30.00

030002-1
either ISO tensile specimen or test plates for SAXS measurements could be fabricated. To investigate the influence
of the processing conditions the screw speed, the pressure at the screw tip and the mass flow rate was controlled in
this work 12. Two different screws where simulated using Polyflow from Ansys Inc. (Canonsburg, USA) and
verified in different experiments.

MATERIALS
Two different types of polypropylene where used, one a PP BB412E and the other a BormedTM DM 55 pharm
both provided from the company Borealis (Linz, AUT). As a compatibilizer, to help to separate the clay, a maleic
anhydride grafted polypropylene from BYK (Wesel, GER) Scona TPPP 2112 GA with 5 w% and as the clay a
layered silicate Nanofil® 5 from Rockwood (Baton Rouge, USA) also with 5 w% was used.
The materials were analysed for the viscosity using a plate-plate rheometer Physica MCR 501 from Anton Paar
GmbH (Graz, AUT) at shear rates ranging from 0.1 to 500 s-1 and a high pressure capillary rheometer (70 to
20 000 s-1) from Göttfert (Buchen, GER). The viscosity was approximated with the Cross law for the viscosity (Eq.
1 left) and the Arrhenius law for the temperature shift (Eq. 1 right). The density was measured with a pvT100 from
SWO (Krefeld, GER).

TABLE 1. Material parameters for the different materials using the Cross and Arrhenius law 12

Parameter BB412E pure BB412E filled DM55 pure DM55 filled

A in Pa·s 13 751 16 763 7 018 8 593
B in s 0.3848 0.49 0.2774 0.3385
C 0.78668 0.79114 0.74576 0.75174
'U in kJ/mol 40.64 39.18 45.35 42.33
T0 in K 488.15
U in kg/m³
759 773 769 789
at 15 bar and 230°C

'U § 1 1 ·
˜¨  ¸¸
A R ¨© T T0
K J ,T  aT T  aT T  e ¹
(1)
1  B ˜ J 
C

SIMULATION
The simulations were carried out with the CFD program Polyflow using the mesh superposition technique 13
where the cylinder and the screws are separately meshed. The mesh was refined at the flight clearance (Fig. 1) to
obtain a good description of the gradients and a stable calculation schema. The simulations were done on either
screw elements alone or calculated over whole regions of the screws, so that the influence of the boundary
conditions could be investigated. As Polyflow can only deal with fully filled regions the simulation started at the
screw tip and calculated backwards until reaching a pressure below zero, where the assumption of the fully filled
region is no more valid. The next simulation where done in regions where fully filled regions existed. As results
from the simulation, the pressure distribution, the shear rate and viscosity and the velocities were obtained. The
mixing index 14 was also calculated to assess the dispersive mixing efficiency.

'p ˜ D V
Sp SV (2)
K ˜n˜L n ˜ D3

The calculations of the screw elements enabled us to calculate dimensionless pressure-volume rate curves (Eq 2),
so that the prediction of not simulated process conditions could be done. In Eq. 2 Si is the dimensionless number, 'p
the pressure difference in Pa, D the diameter of the screw in m, K the viscosity in Pa·s, n the screw speed in s-1, L the

030002-2
length of the elements in m and V the flow rate in m³/s. The shear energy per volume was calculated by multiplying
the viscosity with the square of the shear rate for a correlation of this value to the exfoliation of the nano-particles.

FIGURE 1. a) Meshing of the cylinder b) and the twin screw elements

EXPERIMENTAL
The experiments where done on an injection molding compounder (Fig. 2) consisting of a Leistritz compounder
(Nürnberg, GER), a melt pump, a melt accumulator and an Engel injection molding machine (Schwertberg, AUT).
The pressure at the tip of the screw could be controlled by the melt pump, while the feeding rate is determined by
the gravimetric dosing unit.
Two different molds were used, one for standard tensile specimen and the other molded plates for SAXS
measurements. The tensile tests were done on a Zwick Roell (Ulm, GER) Z250 according to DIN EN ISO 527-2 and
the SAXS measurements on a Nanostar from the company Bruker AXS GmbH (Billerica, USA).

FIGURE 2. Injection molding compounder

030002-3
 To experimentally verify the filled regions five pressure transducers at different positions in the barrel measured
the pressure at the cylinder wall and screw pull out experiments were done. The minimal residence time for the
different conditions of the extruder and feed rates was obtained by adding a coloured masterbatch into the feeder and
stopping the time it needs to the 3-way-valve.

RESULTS AND DISCUSSION

After first trial runs a full factorial experimental design, with the parameters screw speed, pressure at the tip and
screw configuration was set up and the material data (K and pvT) was measured. The simulations were done with
different materials and processing conditions on either the screw elements or on whole sections of the screws. The
standard screw (Fig. 3 upper screw) consisted from four sections with kneading blocks and one backwards
conveying element, which consumes a high pressure, but has also the highest dissipative energy input of the
simulated screw elements.

FIGURE 3. Simulated pressure along the standard and optimized screws

For the optimized screw (Fig. 3 lower screw) a reduction of the pressure peak and a higher residence time was
chosen. Therefore the backward conveying element was changed to a forward conveying mixing element and
pressure consuming kneading elements, so that the mixing section was fully filled. The pressure peak in the region
was reduced drastically, for the setting 8 kg/h, 55 rpm and 45 bar it went down from 40 bar to 10 bar (Fig. 3). The
minimum residence time was raised with the optimized screw between 1 and 7 %, but the main effect on it has the
mass flow rate. The dissipative energy input was lowered about 25 % between the standard and optimized screw.
The yield stress and the Young’s module were measured and with statistical methods the influence of the
parameters were investigated. The Young’s module showed no statistical relevant changes, while the yield stress
(Fig. 4) showed dependencies to the screw speed and to the screw itself. The desired effect of a higher yield stress
with the optimized screw could not be reached.
The SAXS measurements showed that intercalated and exfoliated particles exist in the polymer matrix. The
tendencies of the yield stress could also be seen in the measurements. The optimized screw showed lower
exfoliation than the standard screw, but the main effect on the exfoliation is the screw speed.

030002-4
 FIGURE 4. Yield stress for different experimental setups

CONCLUSION
With statistical means an optimization of the twin screw extruder setup for Polymer Nanoclay Compounds was
done. A change of one screw section from a backward conveying to mixing and kneading blocks lowered the
pressure peak and therefore the dissipative energy, while raising the residence time. The exfoliation of the layered
silicates showed in SAXS and the yield stress measurements that the standard screw with the higher dissipation was
better suited than the optimized screw with a higher residence time. The residence time is important for the diffusion
process, while the shear energy correlates to the shearing and elongation flow and therefore for the exfoliation
process. The shear energy was more important than the diffusion process for the exfoliation of the layered silicates.

ACKNOWLEDGMENTS
The work was done by Joachim Winkler-Ebner as a Master Thesis with contribution of Markus Battisti, Stephan
Schuschnigg and with the experimental help of Eduard Leitner.

REFERENCES
1. S. Sinha Ray and M. Okamoto, Progress in Polymer Science 28 (11), 1539–1641 (2003).
2. Y. Zare and H. Garmabi, Applied Clay Science 105-106, 66–70 (2015).
3. Y. Zare, Composites Part B: Engineering 73, 111–117 (2015).
4. I. C. Amaechi, et al., Eur. Phys. J. Appl. Phys. 69 (3), 30901 (2015).
5. J. W. Gilman, T. Kashiwagi, and J. D. Lichtenhan, Sampe Journal 33, 40–46 (1997).
6. H. Mattausch, et al., in Proceedings of ANTEC 2014, 2014.
7. M. G. Battisti, et al., J. Phys.: Conf. Ser. 602, 012012 (2015).
8. T. Avalosse and Y. Rubin, IPP XV (2), 117–123 (2000).
9. A. Durin, et al., IPP 29 (5), 641–648 (2014).
10. K. Kohlgrüber, Der gleichläufige Doppelschneckenextruder. Grundlagen, Technologie, Anwendungen, Hanser
Verlag, München, 2007.
11. M. G. Battisti and W. Friesenbichler, in Proceedings of PPS-29, American Institute of Physics, 2014, pp. 195–
198.
12. J. Winkler-Ebner, Simulationsgestützte Schneckengeometrie-optimierung für die Herstellung von
nanoverstärkten Polypropylencompounds, Masterarbeit, Montanuniversität, 2014.
13. N.N., POLYFLOW in Workbench User's Guide, Ansys, 2011.
14. H.-H. Yang, and I. Manas-Zloczower, Analysis of Mixing Performance in a VIC Mixer, IPP, IX, 4, 291–302,
1994

030002-5