Experimental and Theoretical Study of Twin-Screw

Extrusion of Polypropylene

O. S. CARNEIRO,1 J. A. COVAS,1 B. VERGNES2

1
Department of Polymer Engineering, University of Minho, Campus de Azurém, 4800 Guimaraes, Portugal

2
CEMEF, Ecole des Mines de Paris, UMR CNRS 7635, BP 207, 06904 Sophia-Antipolis Cedex, France

Received 9 November 1999; accepted 29 March 2000

ABSTRACT: The flow of a polypropylene in a self-wiping corotating twin-screw extruder

was characterized by measuring the pressure, temperature, and residence time along
the screw profile. The influence of the operating conditions (feed rate, screw speed,
barrel temperature) and screw profile was studied. Flow modeling was performed using
the Ludovic娀 software and measured and calculated pressure, temperature, residence
time, and energy consumption were compared. The values of the temperature close to
the melting zone were overestimated by the model, which considers instantaneous
melting upon the first restrictive screw element. If the program assumes that melting
occurs at the screw location identified experimentally, a correct description of the
temperatures along the screw profile is produced. The influence of the processing
conditions (feed rate, screw speed, barrel temperature, screw profile) is well described
by the model. These results put in evidence the importance of including an adequate
melting model in the modeling of the twin-screw extrusion process. © 2000 John Wiley &
Sons, Inc. J Appl Polym Sci 78: 1419 –1430, 2000

Key words: twin-screw extrusion; polypropylene; temperature measurement; mod-

eling

INTRODUCTION along the screw profile and how they are influenced
by variables such as screw geometry, feed rate,
Twin-screw extrusion has been used extensively by screw speed, or barrel temperature.
the polymer industry in various applications, in- Numerous experimental studies on twin-screw
cluding mixing, compounding, and reactive extru- extrusion have been reported in the literature.
sion.1,2 Among the various available system config- Their relevance is related to the reliability of the
urations, corotating self-wiping screws have gained technique used to measure the process parameters,
wide acceptance, due to their good balance between to the practical importance of the variables moni-
mixing efficiency and pressure-generation capacity. tored, and to the location of the sensors in the ex-
Process engineering and optimization requires the truder. Global residence time distribution functions
knowledge of the values of the main-flow parame- are relatively easy to obtain; they are calculated
ters (e.g., pressure, temperature, residence time, from the variation with time of the concentration of
shear rate, specific energy) at different locations a tracer that can be detected using different tech-
niques such as fluorescence emission,3,4 solids re-
Correspondence to: B. Vergnes. sidual weight measurement,5 ultrasonic attenua-
Contract grant sponsors: ICCTI (Portugal); Ambassade de tion,6 or light absorption.7 Melt-pressure measure-
France (Portugal).
Journal of Applied Polymer Science, Vol. 78, 1419 –1430 (2000)
ments are possible in the filled sections of the screw
© 2000 John Wiley & Sons, Inc. or in blocks of kneading discs.8,9 Local material
1419
1420 CARNEIRO, COVAS, AND VERGNES

performed in partially filled sections are not accu-

rate.10
In this work, measurements of pressure, melt
temperature, residence time, and energy con-
sumption were carried out on a laboratory-scale
twin-screw extruder. The use of sampling devices
developed by Machado et al.12 allowed us to per-
form these measurements at six different loca-
tions along the extruder.
Knowledge of the extrusion process can also be
gathered through numerical simulation. Recent
developments provided descriptions of the three-
dimensional flow field in filled screw elements13,14
and in blocks of kneading discs.15,16 Global (plas-
ticating) models, taking into account the whole
Figure 1 Complex viscosity master curve at 200°C. history of the extruded material from the hopper
to the die exit, have also been published.17–20 In
the present study of such a global model, the
temperature is another important parameter, but software Ludovic娀 was used to predict the flow
more difficult to measure with accuracy. Flush- conditions along the extruder. The direct compar-
mounted thermocouples measure the melt temper- ison between measured and computed values is
ature at the wall; for reliable measurements, pro- important for assessing the theoretical model and
truding probes insulated from the barrel wall are may suggest relevant modifications that may im-
recommended (although they affect the local veloc- prove the accuracy of its predictions.
ity field and, consequently, the actual melt temper-
atures). Infrared probes appear as an interesting
technique, as it allows measurements along the PROCEDURE
screw channel.10,11 However, the technique has
some limitations since measurements are carried Polymer
out on the melt surface or in the bulk, depending on The experiments were performed using a polypro-
the melt emissivity.10 Furthermore, measurements pylene homopolymer (HB121J, from Borealis,

Figure 2 Screw profiles (a) A and (b) B and location of the sampling devices (ports
1– 6).
 TWIN-SCREW EXTRUSION OF POLYPROPYLENE 1421

Table I Screw Profile A, From Hopper to Die

Length (mm)

Parameter 97.5 150 60 60 30 120 45 60 60 37.5 120 90 30

Pitch (mm) 45 30 20 KD90 ⫺30 30 KD⫺60 45 30 KD⫺30 60 30 20

A negative pitch indicates a left-handed element; KD ⫺30 denotes a block of left-handed kneading discs, with a staggering angle
of 30°.

Lyngby, Denmark). It has a density of 902 kg m⫺3 staggering angles of 30 and 60° (left-handed) and
and a melt-flow index of 1.9 g/10 min (ISO 1133: 90° (neutral), and a left-handed element located
230°C, 2.16 kg). Its rheological behavior was mea- after the first block of kneading discs. Six sam-
sured in small amplitude oscillatory shear (Rheo- pling devices are located in the melting and mix-
metrics RMS 800) at three temperatures (200, ing sections [ports 1– 6 in Fig. 2(a)]. In the second
220, and 240°C). The complex viscosity master profile, the melting section is restricted to one
curve of this resin is presented in Figure 1. We left-handed screw element, and all the mixing
assumed that the Cox–Merz rule is satisfied and discs have either positive or neutral staggering.
that the viscous behavior is adequately described Consequently, this profile is assumed to be less
using a Carreau–Yasuda law: severe than is the first one. The location of the
collection ports remained unchanged.
␩ ⫽ ␩ 0a T 关1 ⫹ 共 ␭␥˙ a T兲 a兴 共n⫺1兲/a (1) The sampling devices were developed by
Machado et al.12 They allow one not only to mea-
with sure local pressures and temperatures with PT
transducers, but also to collect in a few seconds
a T ⫽ exp关E/R共1/T ⫺ 1/T 0兲兴 (2) polymer samples for, for example, temperature
measurement, determination of residence time,
where the parameters assume the following val- and evolution of plastication. Originally, they
ues: ␩0 ⫽ 22200 Pa s, ␭ ⫽ 1.7 s, n ⫽ 0.35, a were designed to collect samples from reactive
⫽ 0.73, E/R ⫽ 5546 K, T 0 ⫽ 473 K. systems that were quenched in liquid nitrogen
and characterized subsquentially. The working
Twin-Screw Extrusion principle of the device is presented in Figure 3. A
The experiments were carried out on a Leistritz circular aperture in the barrel wall (1) allows the
LSM 30.34 corotating intermeshing twin-screw pressure flow of material out of the extruder. The
extruder. The screw profile (A) used for the main occurrence of this flow is controlled by the cylin-
part of the experimental work is presented in drical valve (2), containing cavities (3) and (4).
Figure 2(a) and explicitly described in Table I. A When the valve is positioned as shown in the
second less restrictive profile (B) was used for one figure, there is no flow out of the barrel. To collect
operating condition. Its characteristics are given a representative portion of material from inside
in Table II and Figure 2(b). the extruder, the valve is rotated sequentially to
The main screw profile includes conveying ele- expose hole (1) to cavity (3) where the accumu-
ments, different types of kneading blocks with lated material can be discarded and, upon further

Table II Screw Profile B, From Hopper to Die

Length (mm)

Parameter 97.5 150 120 30 120 45 60 60 37.5 120 90 30

Pitch (mm) 45 30 20 ⫺30 30 KD60 45 30 KD90 60 30 20

Same legend as in Table I; location of collecting ports is the same.

1422 CARNEIRO, COVAS, AND VERGNES

locations (ports 4 and 6 and die exit). Two grams

of a powder silica tracer (Degussa Ultrasil VN3)
were fed through the hopper throat. Samples
were collected in one specific port every 5 s. The
tracer concentration was inferred from the weight
of the calcinated samples. The average residence
time, t៮ , was computed from21

冘 tC⌬t
⬁

0
t៮ ⫽ (3)

冘 C⌬t
⬁

where C is the concentration of the tracer at time

t, and ⌬t, the time interval.
The feed rate Q (6, 12, and 18 kg/h), the rota-
tion speed N (100, 150, and 200 rpm), and the
barrel temperature T b (190, 210, and 230°C) were
Figure 3 Sampling device: (a) working concept; (b) varied. For each run, we measured the motor
schematic. current (to estimate the mechanical power con-
sumption), the pressure at ports 2– 6 (no pressure
buildup was observed at port 1) and in the die,
rotation, hole (1) is now exposed to cavity (4), and the temperature at the same locations, both
which is rapidly filled with fresh material. The in the flow (PT transducer) and in the collected
valve is rotated again to the position shown in the samples. The operating conditions used in the
figure, where about 1 g of the sample can be various runs and the corresponding experimen-
removed for subsequent analysis. tally measured parameters are reported in Tables
Pressures along the extruder and at the die III and IV, respectively.
entrance were monitored using flush-mounted
Dynisco-type PT transducers connected to a PC Software
interface. Temperatures were measured with the
same flush-mounted PT transducers (at locations Various theoretical approaches to the flow of a
2–5) and also by immediately sticking a fast-re- molten polymer in a twin-screw extruder may be
sponse thermocouple into the samples collected found in the literature. However, here we are
with the devices (at ports 2– 6) and at the extru- interested in a global plasticating model able to
date emerging from the die. The average resi- describe the extrusion process from the hopper to
dence time was determined for two runs at three the die. Such models have been proposed previ-

Table III Operating Conditions Used in Extrusion Experiments

Screw Feed Rate Screw Speed Barrel Set

Run Configuration (kg/h) (rpm) Temperature (°C)

1 A 12 150 190
2 A 12 150 230
3 A 12 150 210
4 A 12 100 210
5 A 12 200 210
6 A 6 150 210
7 A 18 150 210
8 B 12 150 210
 TWIN-SCREW EXTRUSION OF POLYPROPYLENE 1423

Table IV Measured Extruder Motor Currenty Intensity, Average Pressures, Melt Temperatures,
and Average Residence Time

(a) Measured Extruder Motor Current Intensity and Average Pressures

Pressure (MPa) (Average and Standard Deviation)

Motor Current
Run Intensity (A) Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Die

1 21 2.54 7.04 3.49 0.71 0.5–1.0 4.2

0.25 0.08 0.05 0.04 0.1
2 16 2.73 5.57 2.79 0.54 0.6–1.1 3.2
0.23 0.05 0.04 0.01 0.1
3 16 2.48 5.71 2.88 0.58 0.5–1.0 3.7
0.11 0.06 0.02 0.02 0.0
4 18 3.55 6.33 3.01 0.56 0.4–1.0 3.8
0.18 0.04 0.02 0.02 0.0
5 15 2.04 5.45 2.82 0.52 0.2–1.0 3.5
0.13 0.05 0.02 0.02 0.1
6 13 1.66 4.42 2.41 0.51 0.5–1.0 3.0
0.10 0.07 0.03 0.05 0.1
7 21 4.08 7.73 3.68 0.86 0.4–1.0 4.3
0.09 0.05 0.02 0.01 0.0
8 14 0.22 0.10 5.37 0.13 0.0–0.2 3.8
0.02 0.00 0.11 0.03 0.1

(b) Measured Melt Temperatures

Temperature (°C)

Measured in the Sample (Average and Measured in the Flow (PT) (Average and
Standard Deviation) Standard Deviation)

Run Port 2 Port 3 Port 4 Port 5 Port 6 Extrudate Zone 2 Zone 3 Zone 4 Zone 5 Die

1 145.5 163.7 192.1 203.2 211.3 202.4 185.6 190.4 187 181 —
5.9 1.2 3.7 3.5 1.0 4.4 2.3 2.6 2 1
2 145.0 178.8 209.1 232.8 240.5 232.2 223.6 239.2 233 226 —
2.5 1.0 0.9 2.5 1.7 0.8 1.8 2.3 2 3
3 148.7 176.0 201.6 221.8 227.0 213.0 206.2 216.5 213 200 212
1.3 4.4 1.5 1.3 1.1 1.5 1.9 1.3 1 8 1
4 140.0 162.8 191.8 212.8 217.5 211.6 205.6 218.0 212 200 210
4.8 3.2 2.1 0.9 1.0 1.4 0.2 0.3 0 1 1
5 148.1 181.9 210.3 229.8 233.2 226.3 209.5 217.1 215 200 214
4.5 5.1 2.1 0.5 1.4 1.4 2.5 2.8 3 1 1
6 149.4 180.2 213.7 227.5 227.5 217.2 207.4 216.3 209 199 208
2.9 2.5 3.2 1.5 1.6 4.5 0.5 0.4 1 8 2
7 129.5 164.3 189.7 214.8 222.2 223.3 206.4 216.3 210 195 214
5.8 1.2 2.0 3.7 1.6 2.3 0.4 0.4 1 3 1
8 — — 155.6 191.1 210.8 216.9 215.8 217.8 213 201 213
1.7 2.6 0.8 3.2 0.5 0.5 1 1 1

(c) Measured Average Residence Time

Average Residence Time (s)

Run Port 4 Port 6 Die Exit

4 23.5 38.7 61.8

5 22.5 33.8 53.5
1424 CARNEIRO, COVAS, AND VERGNES

techniques, that is, flush-mounted PT transduc-

ers and a manual thermocouple stuck into the
collected melt nut. Significant differences of up to
80°C can be observed. Moreover, it appears that
the PT transducers, which are not insulated from
the barrel wall (which is the most common indus-
trial practice), are largely influenced by the barrel
temperature. Such results were also reported by
Karwe and Godavarti.11 Consequently, since
flush-mounted PT transducers do not seem capa-
ble of monitoring adequately the melt tempera-
ture, only the temperatures measured using the
sampling devices will be considered in the re-
mainder of this work.
The main results concerning temperature and
Figure 4 Comparison between temperatures mea- pressure measurements are depicted in Figures
sured in the flow and in the samples (F) T b ⫽ 190°C; 5– 8. For all runs, the temperature increases rap-
(E) T b ⫽ 210°C; (■) T b ⫽ 230°C.
idly in the melting section (ports 2– 4). At port 2
ously, for example, by White and Szydlowski,17 (upstream the first block of kneading discs), the
Meijer and Elemans,18 or Potente et al.19 Com-
mercial softwares derived from these studies (e.g.,
Akroco Twin Screw娀 or Sigma娀) are now avail-
able. The Ludovic娀 software was developed at the
Ecole des Mines de Paris–CEMEF and was de-
scribed in detail elsewhere.20 It is a global model,
based on local 1D approximations, which allows
one to compute the evolution of the main-flow
parameters (e.g., pressure, temperature, resi-
dence time, filling ratio) along the screw profile,
between the hopper and the die. Nonisothermal
flow conditions, including dissipated power and
heat transfer, are assumed. The polymer viscous
behavior may be described by a power law or by a
Carreau–Yasuda law.
Despite some recent efforts,22,23 a really satis-
factory melting model for twin-screw extrusion is
not yet available. Consequently, it was prescribed
that melting is instantaneous and that it takes
place just prior to the first restrictive element
(i.e., a left-handed screw element or a block of
kneading discs) exposed to the solid pellets. The
global computation follows an iterative scheme,
since the calculations proceed backward, starting
from an arbitrary initial value of the die exit
temperature. This software has been successfully
used for many applications, such as the study of
starch transformation,24 of blend morphology evo-
lution,25 or reactive extrusion.26,27

Figure 5 Influence of screw speed on the (a) temper-

RESULTS AND DISCUSSION ature and (b) pressure measured at the sampling ports
(7 corresponds to the die, 8 to the extrudate). Q ⫽ 12
Figure 4 compares, for different processing condi- kg/h; T b ⫽ 210°C; (䊐) N ⫽ 100 rpm; (z) N ⫽ 150 rpm;
tions, the temperatures measured using the two ( ) N ⫽ 200 rpm.
 TWIN-SCREW EXTRUSION OF POLYPROPYLENE 1425

temperature is lower than the polymer melting

temperature, as confirmed by visual examination
of the samples collected, which shows that, in
most cases, solid granules are still visible at this
location. This is demonstrated in Figure 9, which
illustrates the evolution of plastication along the
barrel for the various runs, as inferred from the
analysis of the samples collected. Further down-
stream, the temperature profile is influenced
mainly by the screw speed, feed rate, and barrel
temperature. Melt temperature increases with in-
creasing rotation speed [Fig. 5(a)] due to viscous
dissipation, with a decreasing feed rate [Fig. 6(a)]
due to higher specific energy and with an increas-
ing barrel temperature [Fig. 7(a)] due to more
heat conduction. Thus, all the processing param-
eters have a clear influence on the temperature
profile. Moreover, when using the same process-
ing conditions [Figs. 8(a) and 9], melting occurs

Figure 7 Influence of barrel temperature on (a) tem-

perature and (b) pressure, measured at the sampling
ports (7 corresponds to the die, 8 to the extrudate). N
⫽ 150 rpm; Q ⫽ 12 kg/h; (䊐) T b ⫽ 190°C; (z) T b
⫽ 210°C; ( ) T b ⫽ 230°C.

earlier in screw A (port 3) than in screw B (port 5).

In screw B, solid granules are present at ports 2
and 3, that is, at the screw elements preceding the
left-handed element. Here, the majority of the
material melts, but some granules are still visi-
ble. Full melting in screw B is only achieved in the
block of right-handed kneading discs (port 5).
Further downstream, the temperature increase
rate diminishes since the screw profile is less
severe, although the outlet temperatures of both
screws are very similar.
For screw A, the pressure reaches a maximum
of 4.4 –7.7 MPa at port 3 (before the left-handed
element), depending on the operating conditions.
Figure 6 Influence of feed rate on (a) temperature This pressure increases sharply with increasing
and (b) pressure measured at the sampling ports (7 feed rate [Fig. 6(b)] or decreasing barrel temper-
corresponds to the die, 8 to the extrudate). N ⫽ 150 ature [Fig. 7(b)] and decreases slightly with in-
rpm; T b ⫽ 210°C; (䊐) Q ⫽ 6 kg/h; (z) Q ⫽ 12 kg/h; ( ) creasing screw speed [Fig. 5(b)]. Conversely, the
Q ⫽ 18 kg/h. pressure in the mixing elements (ports 5 and 6) is
1426 CARNEIRO, COVAS, AND VERGNES

The various runs of Table III were computed

with Ludovic娀 using a heat-transfer coefficient
evaluated as 1500 W m⫺2 °C⫺1. Figure 10 pre-
sents, as an example, the predicted and experi-
mental pressure and temperature profiles for run
3, corresponding to the central experimental con-
ditions (Q ⫽ 12 kg/h, N ⫽ 150 rpm, T b
⫽ 210°C). The screws are partially filled, with
only the die and the restrictive elements being
under pressure. The global agreement between
computation and experiments is good. The order
of magnitude of the pressures is well described by
the model. The melt temperature increases as the
material flows through the mixing elements due
to viscous dissipation and decreases in the con-
veying sections due to heat transfer toward the
colder barrel. The agreement between the predic-
tions and the experiments is poorer in the first
part of the extruder, where the temperature is
significantly overestimated. In fact, samples ex-
tracted from the collecting devices evidenced the
survival of granules up to port 3 (axial position:
0.62 m, see Fig. 1), as illustrated in Figure 9. This
means that melting is not instantaneous as as-
sumed theoretically, but develops along the first
kneading block, inducing a progressive increase
in the average material temperature. Prescribing
that full melting occurs between ports 3 and 4
Figure 8 Influence of screw profile on the (a) temper- produces very good agreement with the experi-
ature and (b) pressure measured at the sampling ports mental data (Fig. 11). Therefore, in the remainder
(7 corresponds to the die, 8 to the extrudate). N ⫽ 150 of this work, all computations will be carried out
rpm; Q ⫽ 12 kg/h; T b ⫽ 210°C; (䊐) screw A; ( ) screw
assuming the experimentally observed location of
B.
the onset of melting.
quite insensitive to variations in the processing The influence of the barrel temperature can be
conditions and is characterized by low values, observed in Figure 12. The extrudate tempera-
ranging between 0.4 and 0.9 MPa. These values ture is highly dependent on this variable, which
are in agreement with data previously published demonstrates the importance of heat transfer in
in the literature.8,9 As expected, the pressure pro- the overall process. The results also show that, in
file for screw B is significantly different [Fig. industrial practice, even when the exit tempera-
8(b)]. It is also characterized by a peak in the ture seems to exhibit the correct values, it is
left-handed element (port 4), but the pressures in possible that, at various locations along the screw
the kneading discs are much lower than for screw profile (depending on the local screw configura-
A, as we have positive (60°) or neutral (90°) stag- tion), hot spots occur which could lead to polymer
gering angles. Die pressures are quite identical, degradation problems. Again, the simulation pro-
as the final temperatures and throughputs are vides a good description of the experiments.
similar. Screw speed seems to have a stronger influence
The average residence times shown in Table on the process than does the feed rate, as shown
IV(c) clarify the influence of the screw speed. An in Figure 13. For this particular case, it seems
increase in the value of this parameter (from 100 that the former only influences viscous dissipa-
to 200 rpm) promotes a decrease in the average tion, while the latter affects both viscous dissipa-
residence time in all the locations considered, tion and heat transfer, which have conflicting ef-
since the material advances with a higher speed fects. This sensitivity is correctly described by the
in partially filled sections.28 model. Again, temperatures reached inside the
 TWIN-SCREW EXTRUSION OF POLYPROPYLENE 1427

Figure 9 Evolution of plastication (experimental data).

extruder may be 30°C higher than the ones mea-

sured at the die exit.
Residence time and residence time distribution
are other important parameters in twin-screw ex-
trusion, particularly when chemical reactions oc-
cur inside the machine. Since the Ludovic娀 soft-
ware is based on simple 1D approximations, it is
not possible to compute a residence time distribu-
tion. However, the time spent in each successive
screw element provides an estimation of the local
average residence time, which can then be com-
pared with the experimental mean values ob-
tained for ports 4 and 6 and at the die exit [see
Table IV(c)]. Figure 14 shows that the software is
able to capture adequately this information, that
is, the cumulative residence times along the
screw and die outlet are correctly estimated.
The results obtained with the second screw
profile [denoted B, see Fig. 2(b)] are shown in
Figure 15. As expected, this profile is less severe
than is screw A (see also Fig. 9). The left-handed
element and the die are the only filled parts of the
machine. As for screw A, the pressure evolution is
correctly predicted, whereas the temperature is
largely overestimated. Again, the original as-
sumption that melting takes place in front of the
left-handed element proves to be incorrect (see Figure 10 Evolution of the computed (a) pressure
Fig. 9). During the experiment, partially melted and (b) temperature along the screw profile. Q ⫽ 12
granules were collected at port 4 (axial position: kg/h; N ⫽ 150 rpm; T b ⫽ 210°C; symbols (F) are
0.58 m), that is, while the material was flowing experimental values measured at the ports.
1428 CARNEIRO, COVAS, AND VERGNES

Figure 11 Comparison between (F) measured and

computed temperature profiles Q ⫽ 12 kg/h; N ⫽ 150
rpm; T b ⫽ 210°C; (full line) melting at experimental
location; (dotted line) melting in front of the first re-
strictive element.

through the left-handed element. As before, when

setting the correct location of melting in the sim-
ulation, the predictions improve substantially, as
illustrated in Figure 16. The need for inserting a
more precise melting model in the software is
thus confirmed and is currently in progress.29 Figure 13 Comparison between measured and com-
Finally, Figures 17 and 18 present a general puted temperature profiles for different values of (a)
comparison between computed and measured pa- feed rate [N ⫽ 150 rpm; (F) Q ⫽ 6 kg/h; (䊐) Q ⫽ 12
kg/h; (F) Q ⫽ 18 kg/h] and (b) screw speed [Q ⫽ 12
rameters for the various runs of Table III. The
kg/h; (F) N ⫽ 100 rpm; (䊐) N ⫽ 150 rpm; (■) N ⫽ 200
predicted temperatures at different screw loca-
rpm].
tions are within ⫾10% of the measured values.
The pressures seem slightly overestimated, but
this could simply be due to the very sharp gradi-
ents observed (see, e.g., Fig. 10), which originate correctly predicted by the model (Fig. 18). What-
large pressure differences even for small errors in ever the processing conditions and the screw con-
the axial location of the transducer. The specific figuration, Ludovic娀 software seems able to pro-
energy (derived from the motor current) is also

Figure 12 Comparison between measured and com-

puted temperature profiles for different values of barrel Figure 14 Comparison between (F) measured and
temperature. Q ⫽ 12 kg/h; N ⫽ 150 rpm; (F) T b computed cumulative residence time. Run 4; Q ⫽ 12
⫽ 190°C; (E) T b ⫽ 210°C; (■) T b ⫽ 230°C. kg/h; N ⫽ 100 rpm; T b ⫽ 210°C.
 TWIN-SCREW EXTRUSION OF POLYPROPYLENE 1429

Figure 17 Comparison between measured and com-

Figure 15 Evolution of the computed (a) pressure
puted (a) temperatures and, (b) pressures: (■) port 4;
and (b) temperature along the second screw profile B. Q
(F) port 5; (E) port 6; (䊐) die.
⫽ 12 kg/h; N ⫽ 150 rpm; T b ⫽ 210°C. Symbols (F) are
experimental values measured at the ports.

CONCLUSIONS
vide a correct estimation of the main thermome- Experiments using novel sampling devices and
chanical process parameters. In practice, this flow modeling were used as complementary pow-
makes it useful for optimizing screw profiles or for erful tools for a better understanding of the twin-
solving scale-up problems. screw extrusion process. It was confirmed that
the use of flush-mounted PT transducers does not
guarantee an adequate temperature measure-

Figure 16 Comparison between (F) measured and

computed temperature profiles when melting is as-
sumed at the end of the left-handed element. Q ⫽ 12 Figure 18 Comparison between measured and com-
kg/h; N ⫽ 150 rpm; T b ⫽ 210°C. puted specific energy.
1430 CARNEIRO, COVAS, AND VERGNES

ment. Generally, the extrusion experiments car- 11. Karwe, M. V.; Godavarti, S. J Food Sci 1997, 62,
ried out validated the theoretical model, which, in 367.
turn, is capable of providing additional informa- 12. Machado, A. V.; Covas, J. A.; van Duin, M. J Appl
tion on local parameters such as degree of filling, Polym Sci 1999, 71, 135.
shear rate, deformation, and viscosity, which is 13. Kajiwara, T.; Nagashima, Y.; Nakano, Y.; Funatsu,
difficult to obtain in practice. Finally, it has been K. Polym Eng Sci 1998, 36, 275.
shown that a better melting model should be de- 14. Cheng, H. F.; Manas-Zloczower, I. Polym Eng Sci
1998, 38, 926.
veloped, if the predictions of the software are to be
15. Ishikawa, T.; Kihara, S. I.; Funatsu, K. Polym Eng
improved.
Sci 2000, 40, 357.
16. Cheng, H. F.; Manas-Zloczower, I. Polym Eng Sci
This work was carried out through a French–Portu-
1997, 37, 1082.
guese collaboration between École des Mines de Paris–
17. White, J. L.; Szydlowski, W. Adv Polym Tech 1987,
CEMEF and the University of Minho, Department of
7, 419.
Polymer Engineering. Financial support from ICCTI
18. Meijer, H. E. H.; Elemans, P. H. M. Polym Eng Sci
and Ambassade de France, both in Portugal, is greatly
acknowledged. 1988, 28, 275.
19. Potente, H.; Ansahl, J.; Klarholtz, B. Int Polym
Proc 1994, 9, 11.
20. Vergnes, B.; Della Valle, G.; Delamare, L. Polym
REFERENCES Eng Sci 1998, 38, 1781.
21. Wolf, D.; White, D. H. AIChE J 1976, 22, 122.
1. White, J. L. Twin Screw Extrusion, Technology and 22. Potente, H.; Melish, U. Int Polym Proc 1996, 11,
Principles; Hanser: Munich, 1990.
101.
2. Xanthos, M. Reactive Extrusion, Principles and
23. Bawiskar, S.; White, J. L. Polym Eng Sci 1998, 38,
Practice; Hanser: Munich, 1992.
727.
3. Hu, G. H.; Kadri, I.; Picot, C. Polym Eng Sci 1999,
24. Della Valle, G.; Barrès, C.; Plewa, J.; Tayeb, J.;
39, 930.
4. Oberlehener, J.; Cassagnau, P.; Michel, A. Chem Vergnes, B. J Food Eng 1993, 19, 1.
Eng Sci 1994, 49, 3897. 25. Delamare, L.; Vergnes, B. Polym Eng Sci 1996, 36,
5. Kim, P. J.; White, J. L. Int Polym Proc 1994, 9, 108. 1685.
6. Chalamet, Y.; Taha, M. Polym Eng Sci 1999, 39, 347. 26. Berzin, F.; Vergnes, B. Int Polym Proc 1998, 13, 13.
7. Chen, T.; Patterson, W. I.; Dealy, J. M. Int Polym 27. Berzin, F.; Vergnes, B.; Dufossé, P.; Delamare, L.
Proc 1995, 10, 3. Polym Eng Sci 2000, 40, 344.
8. Huneault, M. SPE ANTEC Tech Pap 1997, 100. 28. Carneiro, O. S.; Caldeira, G.; Covas, J. A. Mat Proc
9. McCullough, T. W.; Hilton, B. T. SPE ANTEC Tech Tech, in press.
Pap 1993, 3372. 29. Vergnes, B.; Delecour, M. L.; Souveton, G.; Bou-
10. Nietsch, T.; Cassagnau, P.; Michel, A. Int Polym vier, J. M. In PPS-15 Annual Meeting, ’s Hertogen-
Proc 1997, 12, 307. bosch, 1999.