Experimental Study of LDPE Melting in a Twin Screw Extruder using On-Line Visualization and Axial Pressure and Temperature MeasurementsMark D. Wetzel, E. I. du Pont de Nemours and Co., Inc.
This paper is also available in pdf format via download here:
BEST PAPER 2002.pdf
EditAbstract
The melting of polymers in a twin-screw (T/S)extruder is an important operation in many industrial processes. Recent research by Shih, Gogos, Geng and others has identified the physical phenomena that take place during the phase transition. This paper describes an experimental study of Low Density Polyethylene (LDPE) melting in a corotating, intermeshing T/S extruder using on-line visualization
and axial scanning of pressure and temperature techniques. The LDPE melting sequence observed included solid transportin a partially filled screw channel with conductive heating, compaction, pellet deformation, and viscous energy dissipation in the melt with un-melted solids. The effects of throughput (Q) and screw rotational speed (N) are examined. Low and high Q/N ratios have significantly different axial pressure profiles.
EditIntroduction
Twin-screw extruders are used in the plastics, food and allied industries to perform a variety of critical unit operations. Unit steps on a modular screw configuration include solid transport, melting, mixing, reaction, degassing and melt pumping. The melting function can consume the majority of mechanical energy input by the rotating screws.
Until recently, little was known about the melting process. Shih (1)described four fundamental steps in the melting of semi-crystalline polymers during compounding in a batch mixer using on-line visualization. Recently, Gogos and Kim (2, 3, 4) identified similar steps in the melting sequence in the T/S extruder using carcass analysis from samples extracted from the screws. These key steps, from an energy
input perspective are:
1. conductive heating in a partially-filled screw channel,
2. compaction and frictional heating in a fully-filled zone,
3. bulk plastic deformation (PED) and lubrication by the first formation of the melt phase
4. viscous energy dissipation (VED) in the liquid phase in the presence of un-melted solids with heat transfer into the solid phase, and
5. energy dissipation in the melt after the completion of the phase transformation (viscous heating).
Curry (5) examined the role of heat transfer from the barrel during melting in the compaction, PED and VED steps. Other researchers, including Potente (6) and Vergnes (7), have modeled melting using the VED and heat transfer mechanisms. In the current work, an experimental study was conducted in a T/S extruder running Low Density Polyethylene (LDPE) to quantify the axial melting profile and
relative contributions of each step in the sequence.
EditExperiment
McCullough (8) and Christiano (9) used a sliding barrel mechanism to generate dynamic axial and radial pressure scans in the filled liquid m ixing zone of a T/S extruder. Geng and Zhu (10) used a set of barrels fitted with glass windows contoured to the apex region to visualize solid transport and melting of HDPE in a T/S extruder. In this study, a Coperion W & P ZSK-40m m T/S extruder was fitted with a barrel slide device to enable axial movementof the barrels over the screws during a polymer compounding operation. The entire melting zone was examined with an on-line scanning method using four pressure transducers, a flush-wall infrared melt temperature probe (IR-TC) and one glass window as shown in Fig. 1. Pressure probes and the IR-TC were mounted in a spacer plate upstream of the barrel containing the glass window. Probes were located at several radial locations with tips mounted perpendicular and flush to the barrel surface. Two pressure
probes were located near the apex region on the up-turning and down-turning screws. The glass window was contoured
to the barrel shape in the apex region, with a slightincrease
in clearance to preventthe screws from contacting the surface.
The window provided a 30m m axial direction by
40m m wide view port into the extruder.
A standard m elting zone screw configuration wasused
as shown in Fig 1. A solid conveying region used 60m m
followed by 40m m lead forward pum ping elem ents. Tw o
forward 45o-stagger kneading blocks backed by a 40m m
lead reverse pum ping elem ent constituted the working section.
A short melt conveying zone was placed near the end
of the screw. The system was set up to run in an opendischarge
m ode. For each operating state, the slide was
started in the retracted (X=0mm) position. A setofsteady
state and dynam ic m easurem ents w ere recorded along with
video images. Probe signals were recorded at 200 sam -
ples/sec (though an elliptic anti-aliasing filter with 80hz
cutoff frequency) for a 60 second duration. Allpressure
probe signals were averaged to produce a single value at
each axial location. Video was recorded a 30 fram es per
second using standard equipm ent. The barrel slidewasthen
m oved 10 or20m m toward the discharge. The system was
allowed to stabilize and data were recorded again. The
m easurem ents were repeated until the probes were positioned
beyond the reverse pum ping elem ent. The test cycle
was run for the operating conditions listed in Table 1. Low throughput (Q) and screw speed (N) states were selected so
that the video equipm ent could capture the m otion of the
material in the window.
The polym er used in the experiments was Low Density
Polyethylene (LDPE) Petrothene NA from Quantum
Chemical.The melting point is between 104 and 115oC, the
M elt Index (M I) is 7.0 and the m eltdensityis0.918 gm /cc.
Pelletswere half-spheres with a diam eter of approximately
4.0 to 4.5mm.
Discussion of Results
Steady-State M easurem ents
Five melting states were run atseveralthroughputsand
screw speeds as show n in Table 1. Steady state energy input,
feed tem perature and extrudate m elt tem peratures are
listed. M elt tem peratures were obtained with a hand-held
thermocoupleimmersed into the molten polym er and a noncontact
infrared sensor. As throughput is increased atconstant
screw speed (120RPM ), specific energy inputand m elt
tem peratures are decreased as expected. At constant
throughput and increasing screw speed,the specificenergy
decreases at 90RPM before increasing at120RPM . M elt
TC m easurem entsand IR surface tem peratures move in opposite
directions. The steady-state data is not sufficient to
quantify changes in melting perform ance with sm allchanges
in operating conditions.
Visualizing M elting Sequence
Figs 2a and 2b show the m elting sequence observed
through the glass window at 27kg/hr (60lb/hr) /60RPM . In
the partially filled solid conveying zone, the predom inant
flow m echanism w as axial displacem ent. Some pellets did
move from screw to screw in the figure-8 channel path. A
sm allam ountof liquid film was formed on the surface of the
glass indicating that som e m elting occurred by conductive
heating from the barrel walls. A t the X=20m m slide position
(window location W =110m m ) the screw channelwas
fully filled and solids flow shifted to the figure-8 channel
direction. This state represented a highly filled m ode where
the screw channel filled well upstream of the firstkneading
block disk. A t X=40mm thefullyfilled solid agglom eration
is underwent deformation. At X=60m m deformation continued
and a m eltphase appeared in the m atrix, indicating
the lubrication step. Islands ofun-melted solids in molten
polym erappeared by X=100m m . The VED m elting m ode
dom inated and continued for the rem aining m elting zone
length. There was evidence of localized deformation and
squeezing flow in the nip region as the kneading block volum
e was com pressed, but pellets slipped axially to theadjacent
channels in the downstream and upstream disks. O ver
the reverse pum ping elem ent at X=160m m therewasasignificant
am ount of un-melted solids. The downstream portion
of the reverse elem ent was not filled atthe interface
with the forward conveying bushing. Un-meltwasobserved
at the screw tip.
The visualization for a low-fill state, 27kg/hr (60lb/hr)
at 90RPM show ed that a slight change in screw speed had a
large effecton the filled length of the m elting zone. The
degree of fill at X=0mm and 20m m wasdecreased from the
60RPM state as expected. M elt streaking was observed
indicating surface m elting with heat conduction from the
barrel. At X=60mm, the conveying channelfilled withpellets.
D eformation and the onset of the melt phase wereobserved
over the first kneading block at X=80m m . The
transition to VED melting occurred between X=100 and
120m m . The VED m elting m echanism progressed through
the entire m elting zone. Although a significantam ountof
un-meltrem ained,there w as slightly less than the 60RPM
state.
The 55kg/hr (120lb/hr) at 120RPM represented a high
degree of fill with the sam e Q/N ratio as the
27kg/hr/60RPM condition. Solid conveying at X=0mm and
filled length observed at X=20mm are similar to the
27kg/hr/60RPM state. Less melt streaking on the glass surface
occurred due to reduced residence time at the higher
screw speed decreasing heat transfer from the barrel. Compaction
and deformation were seen at X=40 and 60m m .
Lubrication and VED m elting began in the first kneading
block at X=80mm. The VED m ode continued withareduction
in pellet size. The un-melted solid volum e wasgreater
than that observed in the low rate/RPM state.
Axial Pressure and Tem perature Profiles
The m elting steps observed in the visualization were
used to interpret the average axial pressure (PAVG) and IR
tem perature (TIR) profiles. Fig 2c shows the high-fill
27kg/hr at 60RPM state with the melting steps indicated. A
large 300Psig pressure peak was observed at X=140m m
followed by a rapid reduction to 210Psig. This is interpreted
as the high-stress deformation region, followed by
the lubrication of the m elt phase. The rem ainder of the
kneading block length and reverse pum ping elem ent follow
a pressure profile consistent with liquid flow theory
7.
The TIR profile had a rapid decrease to a m inimum at
X=130m m ,the location where the com paction began. This
indicated that the pellets filling the channel had intimate
contact with the barrel and drew m ore heat. Asthe deformation,
lubrication and VED melting steps occurred, the IR
tem perature increased significantly. This transition is consistent
with the heat flux phenom ena observed by Curry
5,
where there is a high heat transfer rate from the barrel to
coldpelletsduring com paction and from the melt film to the
barrel during VED melting.
The low-fillstate, 27kg/hr at 120RPM , pressure profile
is shown in Fig 3. The plot illustrates the reduced filled
length and m ovem ent of the com paction region to the first
kneading block interface. Although the m elting steps are
the sam e as other states, the large pressure peak isgreatly
reduced. This m ay indicate that lubrication occurs very early, significantly decreasing the energy input from the
deformation m ode.
The Root M ean Square (RM S) values of the m eancentered
dynam ic pressure signals w ere com puted foreach
state. This RM S calculation (PRM S) represents the dynam ic
peak-to-peak pressureorstress as the screw rotates. Figs 4
and 6 show the PRM S profiles for all states. The large PRM S
increase corresponds to the com paction region and the
maximum value in the deformation zone. The changes occur
over very short axial lengths. The subtle effects of
changes in geometry are evident as well. For exam ple, a
sm all peak is observed at the double-lengthdisk attheinterface
between kneading blocks.
Effects of Throughput and Screw Speed
Figs 3 and 4 show the PAVG and PRM S at 120RPM for
different throughputs. The deformation PAVG peak is nearly
absent in the low Q/N state. It increases and m oves upstream
with increasing rate reflecting the filled length
change. The initial PRM S rise and peaks moveupstream with
throughput. For all rates, thePRM S values overlap starting at
the second kneading block. This indicates that lubrication
and VED melting dom inate and that flow is in the liquid
regime.The screw speed effects are illustrated in Figs 5 and
6. As screw speed is increased, peak PAVG decreases and
m oves downstream . The PRM S initial rise and peak also
move downstream . Thus, the transition from low filltohigh
fill states is captured.
Conclusions
An on-line m ethod to quantify melting during extrusion
using visualization and axial dynam ic pressure and tem perature
m easurem ents has been dem onstrated. Thekey m elting
steps observed for the LDPE system through the glass windows
included solid transport with conductive heating,
com paction,deformation,lubrication and VED in the m elt
with heat transfer to the un-melted solids. Average axial
pressure, RM S pressure and tem perature profiles showed
two distinct melting m odes. For large a Q/N ratio,com paction
and deformation are initiated in the conveying region.
A large, narrow pressure peak is observed in the high-stress
deformation region that increases at higher throughputs.
The RM S pressure increases prior to the pressure peak. For
low Q/N states, the pressure peak is absent, indicating the
onset of lubrication in the first kneading block. Infrared
probe scans show a significant drop in tem perature in the
com paction region due to efficient heatconduction from the
barrel to the cold pellets in full contact with the metal surface.
This is followed by a tem peraturerise w ith pellet-tobarrel
and pellet-to-pellet friction and as the m elt phase is
formed by VED heating.
Visualization and axial scan data are usefulforcharacterizing
melting performance as a function of materialproperties,
screw geom etry and operating conditions. By linking
the visualization with the axialpressureandtem peratureand
profiles in a m odel system , quantitative data for melting
zone evaluation is possible for operating states or materials
where glass windows cannot be used.
Nom enclature
N = Screw Speed
Q = Flow Rate
P = Pressure
RM S = Root m ean square
T = Tem perature
References
1) C. K. Shih, D. G. Tynan, D. A. Denelsbeck, “Rheological
Properties of M ulticom ponent Polym er System s
Undergoing M elting or Softening During Com pounding”,
Polym er Engineering and Science, M id-
D ecem ber, 1991, Vol. 31, No. 23, pp. 1670-1673.
2) C. G. Gogos, M . H. Kim, “M elting Phenom ena and
M echanism in Polym er Processing Equipm ent”, SPE
ANTEC 2000.
3) M . H. Kim, C. G. Gogos, “The Heating/M elting
M echanism of Plastic Energy Dissipation”, SPE
ANTEC 2000.
4) C. G. Gogos, Z. Tadm or, M . H. Kim, “M elting Phenom
ena and M echanism s in Polym er Processing
Equipm ent”, Adv. Polym . Tech., 17, 1998, pp. 285-
305.
5) J. Curry, “M elting M echanism s in ZSK Extruders”,
SPE ANTEC 95, pp. 92-97.
6) H. Potente, U. M elisch, “Theoretical and Experimental
Investigations of the M elting of Pellets in Co-Rotating
Twin-Screw Extruders”, International Polym er Processing
XI, 1996, pp. 101-108.
7) B. Vergnes, M . L. Delacour, G. Souveton, J. M . Bouvier,
“A Study of Polym er M elting in a Co-Rotating
Twin Screw Extruder”, PPS 15, June 1999.
8) T. M cCullough, B. Hilton, SPE ANTEC,1993, 3372-
3379.
9) J. P. Christiano, M . Lindenflezer. “Investigation of
M ixing Patterns in Co-Rotating Fully Intermeshing
Twin Screw Extruders M ixing Elem ents using Dynamic
Pressure Distributions”, SPE ANTEC ’97, 501.
10) L. Zhu, X. Geng, “Physical M odel of Polym er Pellets
M elting In Co-Rotating Twin-Screw Extrusion”, SPE
ANTEC 2000.
K eyw ords
Extrusion
M elting
Visualization