Vous êtes sur la page 1sur 10

The Hybrid FEM/FDM Computer Model for

Analysis of the Metering Section of a


Single-Screw Extruder
RONG-YEU CHANG and KUEN-JANG LIN

Department of Chemical Engineering


National Tsing Hua University
Hsinchu, Taiwan 30043, Republic of China
A hybrid FEM/FDM computer model was employed in this study for simulating
the non-Newtonian, nonisothermal polymer melt felt in the metering section of a
single-screw extruder. The pressure distribution in the screw surface was obtained by solving the generalized Reynolds equation. Instead of using the energy
equation in Eulerian frame, a Lagrangian expression was involved for stabilizing
the numerical scheme. The temperature profiles were obtained by finite difference
discretization for the energy equation in such element. The screw surface with the
screw channels and the flight lands could be modeled as a surface divided into
small shell elements. To demonstrate applicability, the results provided by the
hybrid FEM/FDM were found to be similar to those of the 2D FDM for the
thermally developing flow, through Fenners example. It can also illustrate the
leakage flow and the cross-channel effect in the screw pumping problem. The
results from the Hybrid FEM/FDM revealed that if the clearance becomes too
large, the volumetric flow rate would considerably decrease and the exist melt
temperature would increase. In addition, when the clearance is close to the normal
design clearance, the leakage flow through the flight lands was found to be small.
These computational results were observed to correlate with those of other experimental studies. Finally, the hybrid FEM/FDM approach can in principle be
extended to the non-Newtonian, nonisothermal flow in a complex screw surface
such as the barrier screw and the Maddock mixing head.

INTRODUCTION

xtrusion is one of the most important operations


in the polymer processing industry. A substantial
part of every polymer passes through an extruder at
least once in its production path from the polymerization reactor to the finished product (1). The metering
section of a single-screw extruder, which controls the
throughput and the exit melt temperature of the
extruder, has been extensively studied. Understanding the transport phenomena under screw pumping
has been the focus of several comprehensive investigations (1-10). Rowell and Finalyson (1 1, 12) p r e
posed the first pumping model for the analysis of
Newtonian fluids in a rectangular channel. McKelvey
et aL (13,141 used a simple energy balance to derive
the analytical solution of Newtonian fluids under
isothermal and adiabatic extrusion. Rotem and Shinnar ( 15) obtained numerical solutions for one-dimensional isothermal power-law fluids in a rectangular

To whom correspondence should be addressed

1748

channel. Griffith (161, Zamoditis et al (17). and Fenner (18) obtained numerical solutions for the 2D,
fully developed, nonisothermal, and nowNewtonian
flow of melt in an infinitely wide rectangular screw
channel. When the thermal convection effects become significant during the extrusion process, the
thermally fully developed flow will not be achieved
even at the exit of the metering section (5).Generally,
the governing equations of the thermally developing
flow could not analytically be solved. Yate (19) applied
a perturbation expansion in the Brinkman number of
the thermally developing model: however, the resulting Brinkman number was within the range of 0 to
1.5 (21). Fenner (20) made detailed comparisons of
the thermally fully developed flow and the thermally
developing flow. The polymer melt may possibly possess a pressure backflow,when the extrusion operates at either a high die head pressure or low
throughput. Elbirli and Lindt (21) got the stable solution under an appreciable pressure backflow by a p
plying the coordinate transformation from a n Eulerian frame toward a Lagrangian frame. Bruker et al

PO1.YMER ENGINEERING AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

Hybrid FEM/ FDM Computer Model


(22) compared the experimental data with the results
of the thermally developing flow, concluding that the
thermally developing flow analysis provided a n accurate flow description of polymer melt in the metering
section. The 2D FDM (21-25) only predicted the flow
fields along the screw channel and neglected the
flight lands, assuming that the temperature, shear
rate, and sear stress remained constant across the
cross-channel direction. Pittman et a1 (26) studied
the nonisothermal non-Newtonian model with the
leakage flow. Rauwendaal et aL (27) found that the
effects of the leakage flow on total throughput became rather large for small values of the power law
index. Increased flight clearance in the melt conveying section having a pronounced effect on extruder
performance was confirmed by other experimental
reports (28-30).
Two approaches have been proposed in regard to
the entire flow domain in the screw extruder, i.e., the
3D finite element flow analysis and the simplified 2D
flow analysis in the screw surface via solving the
generalized Reynolds equation. Masberg and Menges
(31)proposed the usage of the 3 D FEM for the numerical simulation of the 3 D flow problems. The 3 D
FEM model (32-36) used a solid element to approximate the screw channel: hence, this model could
predict the gapwise pressure gradient and the crosschannel flow effect. Rauwendaal (37) and Spalding
et aL (38) utilized the 3 D FEM for simulating the
polymer melt flow in the screw channel and the flight
lands. A comparison of the 3 D FEM results with
those of the 2D FDM revealed that these results were
similar to each other when the screw channel with a
width to depth ratio higher than 10 (7). Fortunately,
the aspect ratio of the metering section of a singlescrew extruder is almost > 10. Two numerical methods have been proposed to solve the generalized
Reynolds equation, i.e., the flow analysis network
(FAN) and the hybrid FEM/FDM. The FAN is the
powerful method for solving the flow problems in
polymer processing, which includes pin barrel screw
(39), tangential counter-rotating twin-screw (40-4 1).
and intermeshing co-rotating twin screw (42-43). The
hybrid FEM/FDM for solving the generalized
Reynolds equation has been successfully used to simulate an unsteady filling process, which includes injection molding, compression molding, and resin
transfer molding (44-48). Matsuoka and Takahashi
(49) applied the hybrid FEM/FDM for simulating the
isothermal, non-Newtonian flow in the steady profile
extrusion coating die process. The purpose of this
investigation lies in applying the hybrid FEM/FDM to
simulate the flow in the metering section of a singlescrew extruder. The generalized Reynolds equation is
solved by the FEM, and the energy equation is discretized by the FDM. The leakage flow is also discussed later.

much smaller than the other physical dimensions. A


screw surface is assumed for the mathematical
derivations. The following simplified assumptions are
made in order to obtain a solution to the problem (5),
i.e.,
1) The flow is steady.
2) The flow is incompressible.
3) The interia force is negligible from the order of
magnitude.
4) The body force is neglected.
5) The polymer melt is purely viscous fluid.
6) The velocity and pressure gradient in a gapwise
direction are neglected.
7) The lubrication approximation is made.
8) Heat conduction in the direction of flow is negligible if compared with the conduction in the g a p
wise direction.

The simplified governing equations for any planar


geometry are as follows:
du

(1)
(2)

(3)

BC1

z=O

U=U,

U=U,

T=T,

(5)

BC2

Z = f f

u=ub

U=Ub

T=T,

(6)

BC3

inlet

p=po

T=To

(7)

BC4

outlet

p=p,

(8)

where x,y are the planar coordinates (e.g. x is the


axial coordinate and y the circumferential coordinate
for the unwound screw surface): z is the gapwise
coordinate u , u are the velocity components in x,y
directions, respectively: p is the pressure: T is the
temperature: v ( j , T ) is the shear viscosity; and
is
the shear rate, i.e.,

(9)
In addition, p is the density: C, is the specific heat: k
is the thermal conductivity: H is the thickness;
(T,, Tb,
T o ) are the screw, barrel and inlet temperature: ( p o ,p,) are the inlet and outlet pressure: and
( u s ,us) and (u,, u,) are the screw and barrel velocity
components in ( x, y). respectively. The shear viscosity is represented in this work by the power-law model
(5).
r]=

THEORETLCAL ANALYSIS

Most screw channels and flight lands have shell-like


geometrical configurations in which its thickness is

du

-d+x - =doy

'exp[-b(T-TreJ)l

(10)

where mo is the power-law constant: n is the powerlaw index: b is the temperature sensitivity: and Trcr

POLYMER ENGINEERING AND SCIENCE, NOVEMBER 7995, Vol. 35, No. 22

1749

R.-Y. C h a n g and K . J . Lin


is the reference temperature. The screw temperature
in most cases is unknown (5). Usually, two types of
the screw thermal boundary conditions, i.e., constant
screw temperature (i.e., T, = Tb) and adiabatic thermal condition at the screw surface (i.e. a n adiabatic
screw), are considered.

tional domain. Fortunately, the pressure boundary


conditions have little effects on the flow field in the
screw pumping problem. They will be discussed later.
The generalized Reynolds Eq 20 is solved by the
Galerkin weighted residual method, i.e.

/QNz.cjddl = 0

Finite Element Approximation for Pressure


Under the boundary conditions 5 and 6, Eqs 2 and
3 are integrated in the gapwise direction so a s to
obtain the velocity u and u (50, 51). respectively:

u=u,+-

ub-u,
SO

z l

-&+-

L 4

s,
/
-&--/
[ z 7z
dx

ub-u,

SO 0

s,

-&+-

SO L n

where Ni is the shape function of the linear triangular element and dl denotes the area of the domain of
interest.

1
I-&--/
1
0

1
-&

Finite Difference Approximation for Temperature

The temperature gradient in the gapwise direction


could not be ignored when the viscous dissipation
effects become significant as a result of either high
viscosity or a high screw rotation speed. After the
pressure field is obtained at each vertex node, the
finite difference formulation for the energy equation
is represented at the gapwise direction at the centroid of each element. The location of temperature
nodes is chosen to be the centroid of each element
where the velocity, viscosity, and shear rate are most
accurate and less averaging is required for the temperature dependent properties (48). Owing to the low
thermal conductivity of melt and subsequently high
Peclet numbers, an upwind technique is employed
for enhancing the numerical stability in the calculation. Therefore, the energy equation in Eulerian frame
is transferred here to a Lagrangian expression from
the instability point of view (2 1, 52)

(11)

u=u,+-

(21)

J yp [ oz n
d
z

SO 0

1
-&
4
(12)

where

so=/H -1 &
0 4

A second integration of Eqs 11 and 12 with respect to


z results in the flow rate per unit width qx and q!,
H

where D/Dt denotes the material derivative. The finite difference discretization of the energy equation at
the centroid of a triangle element is given as

where

The continuity equation averaged across the gapwise


direction could be written a s
(19)

where q = qxj + q J is the flow rate per unit width. A


substitGtion of the flow rate per unit width, Eqs 15
and 16, into the averaged continuity equation yields
the pressure equation, which is also called the generalized Reynolds equation (50, 51)

where i;, = u,,! + ud, g , = us!+ UJ. The generalized


Reynolds equation is a nonlinear elliptic PDE for
non-Newtonian fluids. The boundary conditions
should be given at all boundaries of the computa1750

where
denotes the temperature of layer j at the
centroid of element: TJ denotes the upstream temperature of TJ; k,+ ,,2 = k((q+,
+ T,)/2): and A l denotes
the length of each triangle element. Figure 1 is a
schematic representation of the upstream temperature in each local element. The physical meaning of
the Lagrangian expression could be referred to the
monotone streamline upwind (MSU) approximation
(52). The MSU approximation yields little physical
spatial oscillations and possess little numerically false
diffusion (52). So far, the governing equations have
been derived for a two-dimensional geometry. The
system equation for each element defined in the
three-dimensional coordinate is derived on the local
coordinate. Therefore, additional coordinate transfor-

TJ

POLYMER ENGlNEERlNG AND SCIENCE, NOVEMBER 7995, Vol. 35,No. 22

Hybrid FEM/ FDM Computer Model


I

I -

AI

-4

Fg. !. A schematic representationof the upstream temperature 7; in each element.

is shown in Fig. 2. The pressure and temperature


fields of the 2D FEM are shown in Figs. 3 and 4. The
magnitudes of the pressure in the pushing flight and
the trailing flight at the exit of the metering section
are 6.2 and 5.0 MPa, respectively. At the entrance,
the melt is assumed to have a uniform temperature
of 220C.The thermal boundary conditions of the set
barrel and screw temperature are also assumed to
have an equal temperature profile of 220C (i.e. a cool
screw T, = Tb= Toin the present case). Based on these
assumptions, the averaged temperature elevation of
the melt depends only on the vicious dissipation. The
metal temperature profiles continuously vary from
220C at the entrance to 235C around the exit:
consequently, the flow at the delivery end is still far
from being thermally fully developed as shown in
Flg. 4.
In order to obtain the mesh-independent solution,
two types of mesh are designed. The MESH 1 ( R g .5a)

mations are required to transform the nodal coordinates of a n element from a global system to a local
system (48). More details can be found from Refs. 48
and 53. The complex screw surface could be modeled
by the finite element mesh without using the unwound screw approximation.
RESULTS AND DISCUSSION

To demonstrate the applicability and reliability of


the hybrid FEM/FDM, the thermally developing flow
in the screw pumping problem is studied. The screw
geometrical configuration, material properties of the
polystyrene melt, and operating conditions of
Fenner's example ( 5 ) are listed in Table 1 . In order to
see the reliability of this approach, this problem is
solved again by the 2D FDM, which follows Fenner's
method with the negligible leakage flow assumption.
Another reason for solving this problem by the 2D
FDM is try to find the boundary conditions for the
hybrid FEM/FDM. The computational domain for the
unwound screw channel surface used by the 2D FDM

x(cm)
2. Computational domain for the unwound screw channel used via the 20 FDM.

Q.

Table 1. Screw Geometrical Configurationand


Material Properties.
(a) Extruder Geometry and Operating
- Conditions
Barrell diameter, Db
mm
120
Screw L/D ratio
8
17.66
Screw helix, angle, H
Screw channel depth, H
6
mm
Screw channel width, W
mm
102
Screw flight width, e
mm
12
Barrell temperature, Tb
"C
220
Screw temperature, Ts
"C
220
Inlet melt temperature, To
"C
220
Screw rotation speed
100
rPm
Flow rate, Q
143
cm3/sec
(b) Properties of PS Melt
Power-law index, n
0.36
Power-law constant, m,
10,800
Pa. sec "
Reference temperature, T,,,
"C
200
"C - 1
Temperature sensitivity, b
0.022
Thermal conductivity, k
W/m-"K
0.21
Density, p
990
kg/m3
j / kg .OK
Specific heat, Cp
2000

Fig. 3. The pressure profiles predicted by the 2 0 FDM.

Fig. 4. The averaged melt temperature predicted by the 2 0


FDM.

POLYMER ENGINEERING AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

1751

R.-Y. Chang a n d K . J . Lin

(b)
Fig. 5. (d Finite element mesh MESHl for the 30 screw
channels and theflight lands. (b)MESHZ.
-1

'
0

and MESH2 (Fig. 5b) contain 1748 and 3836 elements, respectively. The thickness of the flight lands
is assumed here to be 0.1 mm. The hybrid FEM/ FDM
results are demonstrated by mapping the numerical
results from the 3D shell screw surface into the 2D
unwound screw surface, a s shown in Figs. 6a and
6b. The computational domain of the hybrid
FEM/FDM has some discrepancies with that of the
2 D FDM; in addition, the mesh used in Figs. 5a and
5b has a resemblance to the screw surface in the real
extrusion process. The inlet and outlet pressure
boundary conditions for the hybrid FEM/FDM are
assumed to be 0 and 5.0 MPa from the 2 D FDM
results. It is easy to find that the solutions of the
pressure fields are almost mesh independent from
FYg. 7.The discrepancy between the melt temperatures provided by the hybrid FEM/FDM from MESH 1
and MESH2 is within 1C. as shown in Fig. 8. In
addition, the discrepancy between the melt temperatures provided by the hybrid FEM/FDM from MESH2
and the 2 D FDM is within 4C. Therefore, the results
from the MESH1 and MESH2 seem to be acceptable.
More precisely, the MESH2 is therefore chosen instead of MESH 1 in the following calculation from the
numerical point of view.

I
2

to

X/D
Fig. 7. The pressure proJles along the screw axis predicted
by the hybrid FEM/ FDM from MESHl and MESHZ.

The pressure distribution in the whole domain predicted by the hybrid FEM/FDM from MESH2 is provided in Fig. 9; it is noted that the pressure in the
pushing flight is higher than that in the trailing
flight. This phenomenon suggests that (a) the melt
flows, from the pushing flight toward the trailing
flight in the screw channel and (b) the melt leaks
from the pushing flight backward the trailing flight in
the flight lands. It is discovered that the larger shear
rate appears in the thinner element from Fig. 10.
There is not much variation in the screw cross-channel direction for the shear rate, as indicated from the
comparison with a high shear rate in the flight lands.
The averaged shear rate can be approximated by the
shear rate caused via the pure drag flow (i.e.,
Nn-D,/H, where N is the screw rotation speed; and
D , is the barrel diameter). The value of the shear rate
caused via the pure drag flow is of the order of 105
sec-' in the screw channel, while it reaches a peak of
the order of 6283 sec-' in the flight land. Figure 11

"

-L
220.0

230.0

MESH2

240.0

250.0

T e m p e r a t u r e ("C)
(b)
Fig. 6. (d Finite element mesh MESHl for the unwound
screw surface mapped.frorn Fig. 5a. fb) MESHZ.
1752

Fig. 8. Comparison between the 2 D FDM and the hybrid


F E M / FDM predicted melt temperature across the depth of
the channel at L / 2 and L

POLYMER ENGINEERING AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

Hybrid FEM/ FDM Computer Model

t-

-1

Fig. 9. The pressure proj?les predicted b y the hybrid F E M / F D M under a cool screw.

Fig. 10. The aueiraged shear rate predicted b y the hybrid F E M / F D M under a cool screw.

shows that the averaged melt temperature profiles


under a cool screw gradually increase from 220C at
the entrance to 233C at the end of the metering
section. The convection in the energy equation has
been considered both in the down-channel and the
cross-channel direction when using the hybrid
FEM/FDM; whereas only the down-channel convection is considered by the 2 D FDM. However, the discrepancies between the melt temperatures produced
by the hybrid FEM/FDM and the 2 D FDM are apparently small in light of the fact that the down-channel
velocities are typically three times larger than the

cross-channel velocities (37).The volumetric flow rate


predicted by the hybrid FEM/FDM with 0.1 mm
clearance is 1 4 2 . 1 cm3/sec, a s listed in Table 2. From
the above results, it can be conchded that the pressure and melt temperature profiles obtained by the
hybrid FEM/FDM are similar to those of the 2 D FDM
in this case.
The results described above are based on isothermal screw temperature. Another special case, an adiabatic screw, is considered (2). Figure 1 2 shows that
the averaged melt temperatures under an adiabatic
condition gradually vary from 220C at the entrance

POLYMER ENGINEERING AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

1753

R:Y. Chang and K . J . Lin

.
.

1
.
.
.
.

F g . 1 I . T h e averaged melt temperature predicted b y the hybrid FEM/ FDM under a cool screw.

Table 2. Comparison of the Flow Rate (cm3/sec) at 100 rpm.

FDM

d=0.6
mm

d=0.3
mm

FEM
d=0.15
mm

143.0

116.4

135.5

142.0

d=0.1
mm

d=O.O6
mm

143.4

144.3

to 239C at the exist of the metering section. The melt


temperature elevation under such a process is higher
than that under a isothermal condition (i.e., a cool
screw) because the viscous dissipation is conducted
away to the barrel and screw under a cool screw. In

Q.

1754

order to compare the different thermal situation in


the screw channel and the flight clearance, the exit
melt temperature profiles across the depth of the
screw channel and the flight clearance are plotted in
fig. 13. The melt temperature profiles in the flight
clearance are lower than those in the screw channel
no matter what a cool screw or an adiabatic screw is
considered, as shown in Fig. 13. Similar numerical
results have been reported by Rauwendaal(37).
The throughput versus the screw rotation speed
with various clearances produced by the hybrid
FEM/FDM under a cool screw is provided in Fig.

12. The averaged melt temperature predicted b y the hybrid FEM/ FDM under an adiabatic screw.
POLYMER ENGINEERING AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

llybrid FEM/ FDM Computer Model


1.0

0 8

0.6

0.4

0.2

0.0
220

250

240

230

260

T e m p e r a t u r e ("C)
Fig. 13. The exit melt temperature profrles across the depth
oj-lhe screw channels andflight lands.

160

I20
h

L)

1
1

d=0.06 m n

Ec:

L
-

20

60

40

80

100

120

RPM
(a)
240

Lj

"
235
,

d = 0.6
d = 0.3
d = 0.1
d=O.06

mm
mm
mm
mm

230

i
CI

Ee:

225

220

LAI

20

40

60

60

100

120

RPM
(b)
Fig. 14. (4, The throughput us. rpm with vanomflight clearances provided by the hybrid F E M / F D M under a cool
screw.lb) The averaged exit melt temperature us. rpm with
various flight clearances provided by the hybrid FEM/ FDM
under a cool screw.

14a It is shown that the relationship between the


throughput and the screw rotation speed is almost
linear. and the throughput, as expected, decreases as
the clearance increases. The leakage flow through the
flight lands is quite small under any screw rotation
speed if the clearance is less than 0.1 mm. I t implies
that this screw with a 0.1 mm clearance in the present case could provide efficient pumping. This value
is very close to the normal design clearance (27). i.e.,
0.001 Db. If the clearance is 0.6 mm and the screw
rotation speed is 100 rpm, much reduction of
throughput can be achieved (say, 18%). In addition,
the reduction of throughput with increasing clearance becomes larger for higher screw rotation speed.
The same qualitative result was also reported in other
studies (54-58).
The averaged melt temperature a t the exit of the
metering section versus the screw rotation speed obtained via the hybrid FEM/FDM under a cool screw
is illustrated in Fig. 14b. For constant flight clearance, the melt temperature strongly increases with
the screw rotation speed. I t is not surprising that the
melt temperature rise with the screw rotation speed
is not a linear function since the viscous dissipation
of power-law fluids is a nonlinear form. For constant
screw rotation speed, the melt temperature rise has
two opposite effects as the clearance is increased. On
the one hand, the shear rate in the clearance is
reduced, which subsequently leads to lower rates of
viscous dissipation. On the other hand, the length of
the radial conduction path in the clearance is increased, which consequently leads to a lower heat
transfer between the melt in the screw and barrel.
The former contribution to the melt temperature is
represented in volume weighting and the latter is
represented in area weighting when dealing with the
energy equation. I t could be confirmed that the latter
is the key role in the present case from Fig. 14b.
The same trend that the melt temperature increases by increasing the clearance has been reported by Rauwendall (29) and Pittman e t al (26). If
the clearance becomes too large, however, the leakage flow increases and the exit melt temperature
rises. The same remarks are also available to the
throughput and the averaged exit melt temperature
under a n adiabatic screw, which are shown in Figs.
15a and 15b. A conclusion can be made that the
screw with a normal clearance can provide efficient
pumping from Figs. 14a and 15a The above results
provided by the hybrid FEM/FDM are based on constant inlet and outlet pressure boundary conditions
(BCs); however, the pressures a t the screw channel
and the flight lands would not be the same. Therefore, the influence of the pressure BCs is examined
by either interpolating or extrapolating the pressure
profiles of the 2D FDM as the inlet and outlet pressure BCs of the hybrid FEM/FDM. Although the constant pressure BCs are replaced by linear-like pressure BCs a t the inlet and outlet, the pressure profile
also gradually increases as a zigzag pattern, which is

POLYMER ENGINEERING AND SCIENCE, NOVEMBER 1995, VoI. 35, No. 22

1755

R:Y. Chang and K . J . Lin


160

shown in Fig. 16. The pressure BCs do not apparently disturb the pressure profiles in the inner regions, as indicated in Figs. 9 and 16.The throughput
predicted by the hybrid FEM/FDM with linear-like
pressures BCs is 146.1 cm3/sec, and the averaged
exit melt temperature is 233C. The results are similar to those using constant pressure BCs. Therefore,
those results will not be addressed again.

---

d= 0 6 mrn
d = 0 3 mrn
d= 0 I mm
d=O06 mrn

CONCLUSIONS

20

40

60

80

100

The hybrid FEM/FDM computer model was employed in this investigation for simulating the polymer melt flow in the metering section of a single-screw
extruder. With fewer assumptions, the hybrid
FEM/FDM can provide similar results to those of a
conventional 2 D FDM modeling for the thermally fully
developed flow. The hybrid FEM/FDM can predict the
flow over the flight lands. Additionally, the results
suggest that the flights are capable of providing efficient pumping as long as the clearance is close to the
normal design clearance, i.e., 0.001 Db. If the clearance becomes too large, the leakage flow would increase and the exist melt temperature would rise. The
hybrid FEM/FDM uses the 3 D shell elements to r e p
resent the screw surface. Therefore, the hybrid FEM/
FDM approach can be easily extended towards the
non-Newtonian, nonisothermal flow in a complex
screw surface such as the barrier screw and the
Maddock mixing head. In addition, this work can be
extended to other continuous polymer processes, e.g.
extrusion dies.

120

RPM
(a)

d = 0 8 rnm
d= 0 3 m r n

a,
4

!
K

at:

ra

230

225

3
I

220

2-1-

20

40

60

80

100

120

RPM
(W

ACKNOWLEDGMENTS

Fig. 15. fd The throughput us. rpm with uariousJight clearances provided by the hybrid FEM/ FDM under an adiabatic
screw. fb) The averaged exit melt temperature us. rprn with
various flight clearances provided by the hybrid FEM/ FDM
under an adiabatic screw.

The authors would like to thank the National Science Council of the Republic of China for its financial
support under Contract No. NSC 8 1-0405-E-007-586.

.
.. .

1
.

1
.

-2

L:.

/-/----

-.

1
.

.---.-I-.
.
...
....

2sL-\.

50

x(cm)

--.
,

.,\.

-\

75

-------.I,->/-loo 0

, , . .

..&

20

_-

30

--->

40

Y (cm)

Rg. J 6. The pressure projles predicted by the hybrid FEM/ FDM with linear-likepressure BCs.
1756

POLYMER ENGlNEERlNG AND SCIENCE, NOVEMBER 1995, Vol. 35, No. 22

Hybrid FEM/ FDM Computer Model

REFERENCES
1. 2. Tadmor and I. Klein, Engineering Principles of PlastG

cating Extrusion, Van Nostrand Reinhold Company, New


York ( 1970).
2. J. M. McKelvey, Polymer Processing, J o h n Wiley a n d
Sons, New York (1962).
3. S. Middleman, Fundamentals of Polymer Processing, McGraw-Hill, New York (1977).
4. 2. Tadmor a n d C. G. Gogos, Principles of Polymer Pre
cessing, Wiley, New York (1979).
5. R. T Fenner, Principles of Polymer Processing, Macmillan, London ( 1979).
6. M. J . Stevens, Extruder Principles and Operation, Elsevier, London (1985).
7. C. Rauwendall, Polymer Extrusion, Hanser. Munich
( 1985).
8. J . L. White, Principles of Polymer Engineering Rheology,
Wiley, New York (1990).
9. J . F. Agassant. P. Avenas, J. Ph. Sergent, a n d P. J.
Carreau, Polymer Processing Principles and Modeling,
Hanser, Munich (1991).
10. K. T. OBrien, Applications of Computer Modeling f o r
Extrusion and Other Continuous Polymer Processes,
Hanser, Munich (1992).
11. H. S. Rowell a n d D. Finlayson, Engineering, 114. 606
( 1922).
12. H. S. Rowell a n d D. Finlayson. Engineering, 126, 249
( 1928).
13. J. F. Carley and J. M. McKelvey, Ind Eng. Chern, 45.
989 (1953).
14. J. M. McKelvey, Ind Eng. Chern, 45,982 (1953).
15. Z. Rotem a n d R. Shinnar, C h e m Eng. Sci., 15, 130
(1961).
16. R. M. Griffth, Ind Eng. C h e m Fundam. 1, 180 (1962).
17. H. J . Zamodits and J . R. A. Pearson, Trans. SOC. Rheo l ,
13,357 (1969).
18. R T. Fenner, Polymer, 16, 298 (1975).
19. B. Yate, PhD thesis, Department of Chemical Engineering, University of Cambridge, Britain (1967).
20. R. T. Fenner, Polymer, 18. 617 (1977).
21. B. Elbirli a n d J . T. Lindt, Polyrn Eng. Sci., 24. 482
( 1984).
22. I. Bruker, C. Miaw, A. Hasson, and G. Balch, Polyrn Eng.
Sci., 27. 504 (1987).
23. C. D. Han, Polyrn Eng. Sci., 28. 1227(1988).
24. M. V. Karwe and Y. Jaluria, Numer. Heat Transfer, Part
A, 17, 167(1990).
25. S. Gopalakrishna, Y. Jaluria, and M. V. Karwe, Int. J .
Heat M a s s Transfer, 35,22 1 ( 1992).
26. J. F. T. Pittman a n d K. Rashid, J . Polyrn Eng.. 5, 1
(1985).
27. C. Rauwendaal a n d J . F. Ingen Housz, Adu. Polym
Technol. 8,290 (1988).
28. R. A. Barr and C. I. Chung, SPE J., J u n e 1966, p. 71.
29. C. Rauwendaal, SPE ANTEC Tech Papers, 35, 108
( 1989).

30. S. J. Derezinski. SPE ANTEC Tech Papers, 39, 2243


( 1993).
31. W. Masberg and G. Menges, SPE ANTEC T e c h Papers,
26. 96 (1980).
32. M. Viriyayuthakom a n d B. Kassahun, SPE ANTEC T e c h
Papers, 30. 81 (1984).
33.A. Kiani, R. Rakos, and D. H. Sebastian, SPE ANTEC
T e c h Papers, 35. 62 (1989).
34. K. Fraser, D. J. Coyle, a n d I. Bruker, SPE ANTEC T e c h
Papers, 35,214 (1989).
35. R. Y. Chang and K. J. Lin. Proc. 14th ROC Polymer
S y m p o s i u m 274 (1991).
36. N. Dontula a n d G. A. Campbell, SPE ANTEC T e c h Pu
pers, 39, 1524 ( 1993).
37. C. Rauwendaal, doctoral thesis, Twente University, The
Netherlands (1988).
38. M. A. Spalding, J . Dooley, K. S. Hyun, and S. R. Strand,
SPE ANTEC T e c h Papers, 39. 1533(1993).
39. R. Brzoskowski. J . L. White. W. Szydlowski, N. Nakajima.
a n d K. Min, Int. Polyrn Proc., 3, 134 (1988).
40. D. S. Bank a n d J. L. White, SPE ANTEC T e c h Papers,
39. 2763 (1993).
4 1. T. Fukuoka a n d K. Min, SPE ANTEC T e c h Papers, 38,
1317 (1992).
42. Y. Wang a n d J. L. White, J . Non-Newtonian Fluid Mech.
32. 19 11989).
43. J . L. White and Z. Chen, Polym Eng. Sci., 34, 229
(1994).
44. C. A. Hieber a n d S. F. Shen, J. Non-Newtonian Fluid
M e c k . 7, 1 ( 1980).
45. V. W. Wang, C. A. Hieber, and K. K. Wang, J. Polym
Eng.. 7,21 (1986).
46. H. H. Chiang, C. A. Hieber, and K. K. Wang, Polyrn Eng.
Sci., 31. 116 (1991).
47. E. W. Liang, H. P. Wang. a n d E. M. Prrry. Adu. Polym
Technol. 3,243 (1993).
48. C. L. Tucker, Fundamentals of Computer Modeling .for
Polymer Processing, Hanser Publishers, New York ( 1989).
49. T. Matsuoka and H. Takahashi, Int. Polyrn Proc., 3. 183
(1991).
50. R. I. Tanner, Engineering Rheology, Clarendon Press.
Oxford, England (1985).
51. C. Dorier a n d J. Tichy, J . Non-Newtonian Fluid Mech.
45,291 (1992).
52. F. Shemirani and K. J a m b u n a t h a n , Int. J. Numer. M e tk
o d s Eng., 14, 1245 (1992).
53. V. W. Wang, PhD thesis, Cornell University. Ithaca. N.Y.
(1985).
54. L. L. Chu a n d K. Min, SPE ANTEC T e c h Papers, 39. 30
(1993).
55. R. A. Honstrater, Plastics Engineering. Dec. 1981, p. 35.
56. D. C. Lounsbury, Plastics Engineering, Sept. 1982,p. 31.
57. E. L. Steward, Plastics Engineering. J a n . 1985, p. 53.
58. J. C. Miller, Plastics Engineering, Ort. 1989. p. 37.

POLYMER ENGiNEERiNG AND SCIENCE, NOVEMBER 7995, VoL 35, No. 22

Revision received September 1994

1757

Vous aimerez peut-être aussi