Vous êtes sur la page 1sur 23

ELSEVIER

J ournal of Food Engineering 27 (1996) 353-375


Copyright 0 1996 Elsevier Science Limited
Printed in Great Britain. All rights reserved
0260-8774/96 $15.00 fO.00
0260-8774(95)00016-X
Modeling of Flow in a Single Screw Extruder*
Y. Li & F. Hsieh?
Departments of Agricultural Engineering and Food Science and Human Nutrition,
University of Missouri, Columbia, MO 65211, USA
(Received 31 August 1994; revised version received 7 March 1995;
accepted 27 March 1995)
ABSTRACT
A new analytical solution of an isothermal, Newtonian flow in a single
screw extruder with a finite channel is developed with the actual
boundary conditions encountered. Down channel velocity distributions
are presented in three-dimensional plots. The boundary conditions,
velocity distributions, and screw characteristics predicted by the new
solution are tested using the experimental data from published literature
(Choo et al., 1980, Polymer Engineering and Science, 20, 349-56;
Grtffith, 1962, I ndustrial and Engineering Chemistry Fundamentals, 1,
180- 7). The results are found to be more accurate than existing
theories (Rowe11 & Finlayson, 1922, Engineering, 126, 249-87; Tadmor
& Gogos, 1979, Principles of Polymer Processing, Wiley; Rauwendaal,
1986, Polymer Extrusion, Hanser).
NOTATION
a,b
Factors defined in eqns (7) and (13)
Internal barrel diameter (m)
The flight width (m)
Shape factor of drag flow defined in eqn (19)
Fd*
Shape factor for overall drag flow defined in eqn (15)
FP
Shape factor of pressure flow defined in eqn (19)
FP
Shape factor for pressure flow defined in eqn (14)
F,*
Shape factor for pressure flow defined in eqn (15)
FS
The shape factor for drag flow caused by screw root surface
*Contribution from the Missouri Agricultural Experiment Station, Journal Series
No. 11,988.
tTo whom correspondence should be addressed.
353
Y Li, E Hsieh
defined in eqn (14)
The drag flow factor for the wall effect defined in eqn (14)
The maximum channel depth (m)
The number of leads
The dimensionless pressure
Pressure gradient in the down channel direction
Pressure gradient in the channel depth direction
Pressure gradient in the cross channel direction
Volumetric flow rate of leakage flow (m3/s)
Down channel volumetric flow rate (m3/s)
Dimensionless down channel volumetric flow rate
Internal barrel radius (m)
Screw root radius (m)=Rb-H
Velocity of the barrel, screw flights and screw root, respectively
(m/s)
Component of velocity in X, y and z directions, respectively (m/s)
Tangential velocity (m/s)
Axial velocity (m/s)
Width of channel at the internal radius of barrel
(m)=2nRb sin&-e
Average width of channel in the n direction (m)
Channel width coordinate (m)
Channel depth coordinate (m)
Down channel coordinate (m)
Viscosity of the Newtonian fluid (Pa s)
Helix angle of screw
Helix angle of screw at radius Rb
Rotational speed of screw (l/s)
INTRODUCTION
Over the last few decades, extrusion technology has played an increasingly
important role in many industries, such as polymer, food, and feed
processing, and its future is still very promising. Among all kinds of
extruders utilized in industries, single screw extruders are certainly a type
that should not be overlooked. The theories of a single-screw extruder have
been significantly advanced during the last 40 years. But the attempt of
applying these theories to screw design and simulation of the extrusion
process has not been entirely successful (Rauwendaal, 1989). First, this is
because the existing theories are not accurate enough for the purpose of
screw design and extrusion simulation. Secondly, these theories are often
misapplied to situations where their assumptions are not valid.
Using the simplified flow theory as an example, it is valid only for screws
with an infinite channel. Although this theory is based on incorrect
boundary conditions, the predictions of the volumetric flow rate will still be
reasonably accurate if it is only applied to screws with a channel depth to
radius ratio, H/ Rb, less than 0.05 (Li & Hsieh, 1994). Due to its simplicity,
this theory is commonly used in extrusion analysis, and is often used for
Modeling of J low in a single screw extruder 355
screws with a finite channel width where its assumptions are violated
severely (Squires, 1958).
As the design and operations of extruders continue to evolve, commercial
extruders, especially for those that are used in food and feed processing
industries, are no longer restricted to screws with a very small channel depth
to width ratio which are favored by the simplified theory. The flow theory
applied for screws with a finite channel width is urgently needed. The first
such flow theory was published by Rowe11 and Finlayson (1922, 1928). Many
other approaches were attempted later on, but all proved to be almost
equivalent to that by Rowe11 and Finlayson (Rauwendaal, 1986).
A thorough examination of the boundary conditions used by Rowe11 and
Finlayson (1922) reveals that their solution is not a complete one. Two
important moving boundaries, the screw flights, were assumed to be
stationary. Therefore, their solution does not consider the effect of the
screw flights. Besides, like the simplified theory, the moving surface which
represents the barrel is assumed to have the velocity of the screw tip, &a.
Plenty of experimental results (Choo et al., 1980; McCarthy et al., 1992) have
shown that this assumption is wrong. The actual velocity of this moving
surface is the screw root velocity, R,w. A detailed examination has been
given elsewhere (Li & Hsieh, 1994).
As computer technology advances, solving a complicated analytical or
numerical problem is becoming less difficult. The effort of searching for a
more accurate solution should always be encouraged. As a reward, this
effort will provide us with more concrete foundations for extruder design
and extrusion simulation. A complete solution can provide more detailed
information of the effect of the screw geometry parameters on the
performance of the screw extruders, especially the effect of screw flights,
which has not been carefully studied before.
Campbell et al. (1992) designed a special screw extruder to test the
existing drag flow analysis. The barrel, screw core, and flights of this
extruder can be rotated separately or simultaneously in pairs. In their
experiments, they tested the following situations: (i) only the barrel was
rotated, (ii) only the screw core was rotated, (iii) only the screw flights were
rotated, (iv) the screw core and flights were rotated together. It was
concluded that the flow generated by rotation of the screw does not agree
with the existing theory and that screw flights are the major contributor to
the drag flow.
The objectives of this study are: (1) to develop a new analytical solution
for an isothermal, Newtonian flow in a single screw extruder using the actual
boundary conditions (i.e. rotating screw and stationary barrel); (2) to derive
simplified formulas and figures of shape factors for both drag and pressure
flows; (3) to verify the new solution using the experimental data from
published literature.
FORMULATION OF THE PROBLEM
In order to simplify the problem, the screw curvature is assumed to be small
so that the barrel surface and screw channels can be unwrapped and become
flat plates. The following assumptions are commonly used to further simplify
356 Y Li, E Hsieh
the problem: (1) the flow is laminar, (2) the flow is isothermal, (3) the fluid
is Newtonian and incompressible, (4) gravitational forces are negligible, (5)
the flow is fully developed, (6) there is no slip at the walls. Therefore, the
equation of motion in rectangular coordinates are reduced to:
2 component:
y component:
I
(1)
+ (!%+!%)
x component:
J!$ ($+A%)
(2)
(3)
where X, y and z are coordinates of channel width, channel depth and down
channel, respectively.
Assuming that v Y ~0, which is a reasonable assumption for screws with a
small channel depth to width ratio, H/ W, and using the continuity equation;
then, C$,lax z 0, au,/& z 0, and av,/ay z 0. Thus, eqns (l)-(3) become:
$ (z!k+%)
(4)
ap a%,
G ay*
(6)
The above simplified equation of motion can be found in Tadmor and
Gogos (1979) and Rauwendaal (1986). Existing solutions for eqn (4) were
based on the following boundary conditions (Rowe11 & Finlayson, 1922,
1928):
Barrel:
Screw root:
Screw flights:
u,(x, H) =&,o cos qb,,
v&,0)=0
&(O,y)=O
r&(W,y)=O
This set of boundary conditions is only valid for model extruders with
stationary screw and rotating barrel, not applicable for real extruders that
are always operated with a rotating screw and stationary barrel.
Experimental results obtained using extruders with a rotating screw have
clearly shown the error predicted by this existing solution (Rowe11 &
Modeling of flow in a single screw extruder 357
Finlayson, 1928; McKelvey, 1953; Middleman, 1977; Griffith, 1962; Choo et
al., 1980; Campbell et al., 1992). The source of the error was elucidated by
Li and Hsieh (1994) and a new solution for pure drag flow in extruders with
a rotating screw and stationary barrel was provided.
In extruders with rotating screw and stationary barrel, the following
boundary conditions should be used to solve eqn (4) for the down channel
flow:
Barrel:
Screw root:
Screw flights:
z&,H)=O,
z&(x, O)=R,o cos &,
t:(O,y)=(Rs+y)o c os4hr
& (W,y)=(R,+y)(~ c os+h
Thus, different from the conventional parallel plate model, the screw
channel in this new model is kept moving with its actual velocity and the
barrel surface is stationary. This set of boundary conditions are defined
based on the absolute velocities of the rotating screw. The velocity profiles
based on this set of boundary conditions can be integrated directly to obtain
the net volumetric flow rate.
Notice that the flight velocities are not zero but a function of channel
depth. In existing solutions (Rowe11 & Finlayson, 1922, 1928; Tadmor &
Gogos, 1979; Rauwendaal, 1986) the flight velocities were all set to zero by
assuming the screw was stationary. Since the screw is rotating, the velocities
from the screw center to the tip are not constant but a function of the
radius.
Since eqn (4) is a nonhomogeneous, second-order partial differential
equation with nonhomogeneous boundary conditions, it needs to be
separated into two parts with two sets of boundary conditions. The finite
sine transformation method can then be used to transform the problem into
ordinary differential equations, and solve z;,(x,y). This is shown in the
Appendix. The final result is
1 aP
~z=Rstr,cos(bhfvl+(2Rh-H)wcos~b~l+~ z.(aW2f,3+hH2f,4) (7)
where
sin $$ sinh in(H-Y)
W
71 i =l . 3, 5, . . .
i i7cH
sinh -
W
i71y in(W-x) i7cx
sin - sinh
f+2 5 H
H + sinh H
71 r=l.3,.5....
i i7rW
sinh -
H
358 I: Li, l! Hsieh
im rsinh i7c(H-y)
sin -
W
+ sinh 5
-1
inH
sinh -
W
i7T( W-x) inx
sinh
H + sinh H
-1
i71W
sinh -
H
The constants a and b in eqn (7) are dependent on the H/ W ratio. When the
H/ W ratio is small, a=0 and b=l; when the H/ W ratio is large, a=1 and
b=O.
For the cross channel flow, u,, eqn (6) can be integrated twice to yield:
aP y*
c,=--+c,y+c*
ax 2~
(8)
The constants cl and c2 are solved with the following boundary conditions:
u,(H)=&
uJO)= -R,w sin 4b
Then, the equation for U, becomes:
aP (y2-Hy)
H-Y
v,=-
ax 2/l
- R,w sin &, -
H
(9)
In eqn (9) the pressure gradient aP/ ax is usually not known. In order to
evaluate this pressure gradient, v, can be integrated over the channel depth
H:
s
H
H3 aP R,w sin&H
Ux dy= _- --
0
12~ ax 2
=Q leakage
(10)
or
ap
lb
Q
R,o sin &,H
-= _-
ax H3
leakage+
2
1
(11)
Clearly, aPli3x depends not only on the screw geometry and viscosity, but
also the clearance between the screw tip and barrel. Substituting eqn (11)
Modeling of flow in a single screw extruder
back into eqn (9) the cross channel velocity IJ,, becomes
359
Rsw sin 4bW-y>
vx=
H
+6QleakageW91
H3
(12)
If the screw geometry is provided and the leakage flow Qlrakage known, the
velocity profile v, can be determined. Unfortunately, Qleakage is difficult to
evaluate. Equations for calculating Qleakage are available in the literature by
Rauwendaal (1986, 1988) as well as Tadmor and Gogos (1979); however,
these equations have not been verified by experimental data.
FLOW RATE CHARACTERISTICS
For extruder design and extrusion simulation, the flow rate-pressure
relationship is of great interest. By integrating the down channel velocity,
CX(~Y) [eqn ( 7) 1,
over the cross-sectional area of the channel, the overall
flow rate can be obtained:
Qz=R,cl, cosq4,W2fc,, +(2Rh-H)w cos&H2fQ2+$ $
!
WH
W4fc,3-12
b i3P H3W
+- -
P az
H4fQ4-
12
where
71-j-135 I-
, _ ,
;, ~
_ sinh -
w .
i71W
cash - - 1
H
I
i7cW
sinh -
H
I
inH
fQ4=? f 11;
72
i=l,3,.5...
i7rW
cash - - 1
H
I
i7cW
sinh -
H
I
(13)
360 X Li, l? Hsieh
When H/W< 1, a common case for single-screw extruders, the coefficient
f
a2, will become 0.271377272 and fQ4 converges to 0.05252075 (Rowe11 &
Finlayson, 1928). Therefore, by setting a=0 and b=l and rearranging eqn
(13), the flow rate can be simplified to:
R; i3P
Qz=R;co(Fs+Fw) -- -F,
IJ az
(14)
where
F = (1 -H/R,) sin2&,
s
nt
For better accuracy, fpl is evaluated using the average radius, or
ircH
where
W,=
7C(2Rb -H) tan &,
-e cos&,
nt )
An equation suggested by Rowe11 and Finlayson (1928) can be used to
estimate fnI as well:
&* =0*5 (;)-0.3151(;)
Detailed analyses of F, and F, can be found in Li and Hsieh (1994). For
the case of H/W> 1, it is convenient to set a=1 and b=O in eqn (13) and
the factors, &I, fez and&,,
can be calculated in a similar manner.
By defining the dimensionless flow rate Qzand pressure P, as:
Q;=
1aP H=
RbW COS4bwH
and P,=- -
p & R,,co COS 4b
the equation for dimensionless screw characteristics can be obtained:
QL=F&-F;Pz
(15)
Modeling of flow in a single screw atruder
where
361
Fd*=
1 - HIR, 27ctan& e
@271377272H/R,(2 -H/R,)
HI&,
Q
-R l+
b 1
2ntan& e
-- cos&
& Rh >
27ctan& e
-- cos&,
& Rh >
The shape factor, F 2, in eqn (15) indicates the maximum dimensionless
flow rate, or open discharge flow rate for any given screw. It is plotted in
Fig. 1 as a function of H/R, ratio at various helix angles &,. As the channel
depth to radius ratio, H/R,, increases, this maximum flow rate decreases in
a slightly nonlinear manner. For a small helix angle, $h, a higher
dimensionless flow rate will be expected at the same H/ R, ratio. But this
helix angle effect will diminish quickly as the helix angle increases, especially
for a H/ R, ratio less than 0.2. The shape factor, F,*, in eqn (15) is the slope
of the dimensionless screw curves and an indication of pressure flow effect.
A similar plot for Fp* is given in Fig. 2. The relationship between Fp* and
H/ R, is linear; Fp* decreases with increases in H/ R,. The effect of helix
angle, &,, on Fp* is opposite to that on F$. For a smaller helix angle, &,
F,* decreases faster with increases in H/ R, than for a larger helix angle &.
Using Figs 1 and 2, the screw characteristics can be constructed. If the
H/ R, ratio and helix angle, $h, are given, the corresponding maximum flow
rate, F$ and the slope of the screw curve, Fp*, can be located from Figs 1
and 2, respectively. A corresponding screw characteristic can then be
plotted. The intercept of the curve with the x axis, or the ratio of
Fd
0. 510
0. 459
0. 408
0. 357
0. 306
0. 255
0. 204
0. 153
0. 102
0. 0 0. 1 0. 2 0. 3 0. 4 0. 5 0. 6 0. 7 0. 8 0. 9 1. 0
H/ R,
Fig. 1. Drag flow factor, F$, as a function of H/ R, for overall drag flow at various
helix angles & (e/ R,,=O.l, n,=l).
362
2: Li, E Hsieh
F$F,*, is the maximum dimensionless pressure, defined in eqn (15). Four
screw curves are shown in Fig. 3. For a very small H/ Rb ratio (O-05) the
effect of the helix angle is trivial. But for a large H/ R, ratio (O-.5), the effect
of the helix angle &,, is significant. On the other hand, no matter what the
helix angle is, the effect of H/R, is always considerable.
DOWN CHANNEL VELOCITY PROFILES
The velocity profile, v,, as a function of the channel depth, y [i.e. u,=v, (y)],
in the screw channel of single screw extruders has been discussed extensively
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.6 0.9 1.0
H/R,
Fig. 2. Pressure flow factor, Fp*, as a function of H/R, at various helix angles &,
(e/R,=O.l, n,=l).
0.5
2
c 0.4
t
oz
g 0.3
i;
ti
6 0.2
E
f
; 0.1
z
0.0
- &= lo, H/R,= 0.05
..----- 60. H/R,= 0.5
0 1 2 3 4 5 6 7 6
Dlmenrionless Pressure. P, or Fi/Fl
Fig. 3. Dimensionless screw characteristics predicted by eqn (15).
Modeling offlow in a single screw extruder 363
in published literature (Tadmor & Gogos, 1979; Rauwendaal, 1986, 1988).
However, little information has been given on the effect of screw flights.
From this analytical solution [eqn (7)], the dependency of the down channel
velocity component v, on the cross channel position, x, and channel depth
position, y (~=v&,y)) can be analyzed. This provides information that
cannot be revealed by one-dimensional solutions.
If the leakage flow is ignored, the flow in a screw channel inside a single
screw extruder becomes a combination of drag and pressure flow. Usually,
the pressure flow is caused by the restriction of the die mounted at the end
of an extruder. If there is no die attached to the extruder, the back pressure
flow will vanish. The drag flow is introduced by the movement of the screw,
or specifically, the screw root surface and the screw flights.
In the conventional studies, only the drag flow caused by the screw root
surface is investigated. The argument for ignoring the flight effect is based
on the assumption of an infinite channel width. But in practical situations,
screws with an infinite channel width do not exist. Therefore, investigation
of the flight effect is imperative. The analysis given here can provide us
information on under what conditions a screw channel can be treated as
infinite. Figure 4 is the pure drag velocity distribution introduced by the
screw root surface itself, the first term on the right-hand side of eqn (7).
Figure 5 is the velocity distribution introduced by the flight surfaces only,
the second term on the right-hand side of eqn (7). The overall drag flow is
a combination of these two contributions, and is presented in Fig. 6. At the
boundaries (screw root and screw flight surfaces), the velocities satisfy no-
slip conditions. In the cross channel direction (x direction), the velocity is
very much affected by the movement of the flights. At any given cross
channel position x, a convex velocity profile is seen except at x=0 and x= W,
where the velocity profiles are linear. This phenomenon was also reported
by Choo et al. (1980). This effect is caused by the drag effect of the flights.
This is why it was not revealed by the simplified and Rowe11 and Finlaysons
solutions (1922, 1928) in which this effect was ignored. In Rowe11 and
Finlaysons solution, the velocities at the flights are set to be zero implicitly.
Therefore, their velocity profile is similar to that shown in Fig. 4.
Figure 7 shows the corresponding velocity distribution of combined drag,
the first two terms on the right-hand side of eqn (7) and pressure flow, the
third term on the right-hand side of eqn (7). Due to the influence of
pressure flow, the convex profile changes to concave shape in the central
part of the screw channel. Consequently, the overall flow rate is reduced.
From these three-dimensional plots, it is evident that, as long as the channel
width is finite, the velocity distribution along the cross channel direction is
not uniform, and the maximum velocities are located at the flight
boundaries. Thus, the simplified theory and the exact solution by Rowe11
and Finlayson (1922, 1928) Tadmor and Gogos (1979) and Rauwendaal
(1986) cannot provide sufficient accuracy when the flight effect is significant.
COMPARISON WITH PUBLISHED EXPERIMENTAL RESULTS
The predicted tangential and axial velocities, v(-, and v,, are compared with
the published data in Figs 8-11. The equations used to calculate vg and cZ3
364 IT Li, E Hsieh
are as follows:
uO=Rs~ - [uz cos & - V, sin &]
(16)
u, = u, sin & + v, cos q&
(17)
Since the clearance of the screw used by Choo et al. (1980) was over four-
fold the standard clearance (0.001 I&), the leakage flow would have been
considerable (Rauwendaal, 1992). However, neither the cross channel
1.0
0.8
Fig. 4. Velocity distribution of a drag flow by screw root surface.
2. 0
0. 5
0.0
Fig. 5. Velocity distribution of a drag flow by screw flights.
Modeling of flow in a single screw extruder
Fig. 6. Overall velocity distribution of a drag flow.
365
Fig. 7. Overall velocity distribution of a combined drag and pressure flow
pressure gradient, i3PBx, not the leakage flow, Qleakage is available (Choo et
al., 1980). In order to estimate oX, Qleakagr=(6/H)R,o sin &Wr,H was used as
the first guess, which turned out to be close to the final values used. At each
given condition, the estimated value was adjusted to fit one of the velocity
profiles, such as vg, obtained experimentally by Choo et al. (1980). The same
&akage
was used to predict the axial velocity profile (v,) in eqn (17). For
three out of four cases with dimensionless pressure, P, ~3.33 (Choo et al.,
366 Y. Li, E Hsieh
V, (mm/s)
4 6 6 10 12
- 0.6
<
c
-! 0.4
0.2
0.0
- i
.
. .
I
#
. /
/
l V,-Choor data .
. .
/
- V,-pmdlctnd
/ .
1
VII-Choos data
/
/
- V,,-Pradlcted
0 4 6 12 16 20 24
V, (mm/s)
Fig. 8. Comparison of the predicted v o and v, and the experimental results by
Choo et al. (1980) for corn syrup (P,=-2.01, H/R,=0.505, &,=45.2, R,=19mm,
H=9.6 mm, W=54.9 mm, n,=l, e=ll.9 mm).
v, (mm/s>
6 8 10 12 14 16
.
V,-Choos data
- Ve-Predlctrd
-.-. V#-PredIcted
6 8 10
Va (mm/s>
16
Fig. 9. Comparison of the predicted v
o and v, and the experimental results by
Choo et al. (1980) for corn syrup (P,=1.25, H/Rb=0.505, &,=45.2, Rb=19 mm,
H=9.6 mm, W=54*9 mm, q=l, e=11.9 mm).
Modeling of flow in a single screw txtmder
V, (mm/s)
0 5 10 15 20 25 30 35 40
. VB-Chooa data
- VB-Prodlctod
V,,-Choos data _
-.- Vo-Pradlctod
0 5 10 15 20 25
367
Fig. 10. Comparison of the predicted u o and u, and the experimental results by
Choo et al. (1980) for corn syrup (P,=3.33, H/R,=O+505, &=45.2, R,,=19 mm,
H=9*6 mm, W=54.9 mm, nt=l, e=ll*9 mm).
V, (mm/s)
0 20 40 60 60 100 120
- I
1 .o
0.6
G 0.6
>
Y
- 0. 4
l I,-Choor data
- Vo-Pradlctod
. V,-Choos data
l -.- V_-Predictad
V, (mm/s)
Fig. 11. Comparison of the predicted u o and u, and the experimental results by
Choo et al. (1980) for corn syrup (P,=4.08, H/Rb=@505, &,=45+2, Rb=19mm,
H=9.6 mm, W=54.9 mm, nf=l, e=ll.9 mm).
1980), the predicted values are in good agreement with the data from Choo
et al. (1980) for both ug and v, (Figs 8-10). When P,=4*08, a case near the
close discharge condition, the predicted ug is still reasonable, but u, is
slightly off (Fig. 11).
368 Y Li, E Hsieh
L 0.6
0
: 0.5
0
L, 0.4
r
:
r 0.3
0
z
2 0.2
E
ii 0.1
.
.
. .
. _
\ . . . . .
.
.~~~.~~ The slmpllflrd theory
- - Rowell and Flnlayrons theory
.
:
,....
- This analytlcal rolutlon
l Choos axperlmrntal data
0.0
-2 -1 0 1 2 3 4 5 6 7
Dimensionless Pressure
Fig. 12. Comparison of the predicted screw characteristics by this solution, the
simplified theory, Rowe11 and Finlaysons solution and the experimental data by
Choo et al. (1980) for corn syrup (H/ &,=0.505, &,=45.2, I&=19 mm, H=9.6 mm,
W=54.9 mm, n,=l, e=ll.9 mm).
0.5 I I I / I
_
q.
....... The slmpllfled theory
2
- - Rowell and Flnlaysons theory
l.E
0.4
- Thls analytlcal solutlon
:
Grlfflths experlmental data
z 0.3
:
0
$ 0.2
si
:
E 0.1
0.0
0 1 2 3 4 5 6 7
Dimensionless Pressure
Fig. 13. Comparison of the predicted characteristics by this solution, the simplified
theory, Rowe11 and Finlaysons solution and the experimental data by Griffith (1962)
for corn syrup (H&,=O~128, q&=30, Rr,=25.4 mm, H=3.25 mm, W=69.19 mm,
n,=l, e=6.35 mm).
Modeling of flow in a single screw extruder 369
Figure 12 is a comparison of the screw characteristics predicted by eqn
(15) the simplified theory [eqn (18)], Rowe11 and Finlaysons theoretical
solution [eqn (19)], and the experimental results of Choo et al. (1980).
(19)
where
izH
tanh -
FDXS i 2w
7-r3
i=l
i7
inH
tanh -
H 16 %
F,=---
12H
c i2w
7? ;+,
Over the whole pressure and flow rate ranges, this new analytical solution
predicted the experimental data in sufficient accuracy while the other two
theories produced significant overpredictions. Figure 13 is a comparison of
the screw curves predicted by the simplified flow theory, Rowe11 and
Finlaysons two-dimensional solution, this new analytical solution [eqn (15)],
and the experimental results given by Griffith (1962). Again, the predicted
screw curve by eqn (15) is more accurate than others as shown. In addition,
the slopes predicted by the other two solutions also deviate from the
experimental data.
CONCLUSION
The analytical solution obtained satisfies the actual boundary conditions
encountered in single screw extruders for screws with a finite rectangular
channel. The simplified formulas derived and figures of shape factors
presented in this study can be used for the construction of the screw
characteristics and the analysis of extruder performance. The velocity
profiles presented in three-dimensional plots reveal important information
about the effect of the screw flights. This new analytical solution is found to
be more accurate than the existing solutions using the experimental results
from published literature. Therefore, the new analytical solution developed
is expected to give better results for the purposes of screw design and
extrusion simulation.
370 Y Li, F Hsieh
REFERENCES
Campbell, G. A., Sweeney, P. A. & Felton, J. N. (1992). Experimental investigation
of the drag flow assumption in extruder analysis. Polymer Engineering and Science,
32. 1765-70.
Chad, K. P., Neelakantan, N. R. & Pittman, J. F. T. (1980). Experimental deep-
channel velocity profiles and operating characteristics for a single-screw extruder.
Polymer Engineering and Science, 20,349-56.
Griffith, R. M. (1962). Fully developed flow in screw extruders. I ndustrial and
Engineering Chemistry Fundamentals, 1, 180-7.
Li, Y. & Hsieh, F. (1994). New melt conveying models for a single screw extruder.
J ournal of Food Process Engineering, 17, 299-324.
McCarthy, K. L., Kauten, R. J. & Agemura, C. K. (1992). Application of NMR
imaging to the study of velocity profiles during extrusion processing. Trends in
Food Science and Technology, 3,21519.
McKelvey, J. M. (1953). Experimental studies of melt extrusion. I ndustrial and
Engineering Chemistry, 45,982-6.
Middleman, S. (1977). Fundamentals of Polymer Processing. McGraw-Hill, New
York.
Rauwendaal, C. (1986). Pot)mer Extrusion. Hanser Publishers, New York.
Rauwendaal, C. (1988). Leakage flow of an isothermal power law fluid. Advances in
Polymer Technology, 8,289-316.
Rauwendaal, C. (1989). The ABC of extruder screw design. Advances in Polymer
Technology, 9,301-8.
Rowell, H. S. & Finlayson, D. (1922). Screw viscosity pumps. Engineering, 114,
606-7.
Rowell, H. S. & Finlayson, D. (1928). Screw viscosity pumps. Engineering, 126,
249-387.
Squires, P. H. (1958). Screw-extruder pumping efficiency. SPE J ournal, 14, 24-30.
Tadmor, Z. & Gogos, C. G. (1979). Principles of Polymer Processing. John Wiley,
New York.
APPENDIX
a2uz a2vz 1 aP
-+-- -
ax2 ay -; az
Based on eqns (4) and (5), it is known that the pressure gradient aPb3.z is
either a constant or only a function of z. The unwrapped screw and barrel
system, the coordinates defined in this study, and the absolute velocities of
the barrel, screw root and screw flights are shown in Fig. Al. The actual
boundary conditions for this problem are defined based on the absolute
velocities:
n* (x, H) =fi (x) = 0,
o<x<w
uz(x, O)=f,(x)=&m COS4b,
o<x<w
~~(o,y)=gl~)=(~,+Y)~cos~b, O<y<H
u~(~y)=g2~)=(~,+y)~cos~b, O<y-=H
Modeling of flow in a single screw e&ruder 371
Fig. Al. Schematic representation of an unwrapped screw and barrel system.
This is a nonhomogeneous partial differential equation with
nonhomogeneous boundary conditions. This problem can be solved by
separating it into two problems with homogeneous boundary conditions.
Assuming v,(x,y)=u,,(x,y) + u&,y) and separating the boundary
condition into two sets, then the problem becomes:
a*~,, a%, , a aP
-+-=- -
ax2 ay2
P az
The first set of boundary conditions are:
1;Z,(x,H)=f,(x), o<x<w
~zI(~,q=fz(x), o<x<w
%l(O,Y)=O,
O<y<H
rzI(w?Y)=O,
O<y<H
aQz2 aGz2 b ap
- +- =- -
a.2 ay2
P a2
The second set of boundary conditions are:
cz2(x,H)=0, O<x< W
fI z2(x,0)=0, o<x< w
21,2(O,y)=g1Cy)> O<Y<H
uz@,y)=g&)~ O<Y <H
372
Y Li, F Hsieh
First, vZ1(x,y) is solved using the first set of boundary conditions. The
solution of vZl(x,y) is as assumed
k(4y)= 5 My) sin(b)
i=l
In order to satisfy the boundary conditions for x=0 and x=W, let k=i TC/ W
and define the following:
For a non-zero solution, i= 1,3,5,. . . Therefore,
Now the nonhomogeneous PDE has been transformed into a
nonhomogeneous ODE with the form
The boundary conditions become:
b.(H)=~
s
W
uZl (x,H) sin
0 (. >
; &CO
Modeling of flow in a single screw extruder
373
Solving the above ODE, the result is:
sinh ia-Y 1
bib)=
4R,o cos & W a aP 4w2
+- --
i7c
p dz (in)
sinh
in(H-y) .
W
+ sinh $
-1
inH
sinh -
W
Therefore, the general solution of vzI(x,y) can be presented as:
Vzl (qJ)= i:
i=1,3,5...
4R,w cos &
i71
a 8P 4W2
sinh wf-Y) +sinh iXY
W W
+- --
p CLz (in)
-1
ixH
sinh -
W
i7cx
sin -
W
The method used for solving u&x,y) is the same and the second set of
boundary conditions were used. The final solution for ~~~(11,~) is presented
as:
i
4&Gy>= i:
4&o co@, + b CI P 4H2
;=1,3,5... in p dz (i7r)
sinh (s) + sinh ( i~Ux) b ap 4H2
--
i7cW
sinh -
( )
-1 82 (in)3
H
i7ty
sin -
( 1
H
where
R,=
& +K
--zR~_~
2
374 Y Li, E Hsieh
The volumetric flow rate for the overall drag flow is obtained by
integrating the overall velocity distribution over the cross channel area of
the screw:
w H
WH
Qz=n,
ss
Qx,Y) dy dX=nt
ss
Q,I (X,Y) dy d.x
0 0 0 0
WH
+nt
ss
&,Y> dy h=Qs+Qw+Q,
0 0
Substituting the expression for R, and W and rearranging the equation,
the final results can be presented as:
i sHI W
Qs=ntR,~W2cos&,
inH,w,:
27cRb tan&, 2
=nt(Rb -H)o -e cos3 fjb
nt
where
F
)
2
s
where
Modeling of flow in a single screw extruder
ixH
+LP f
I_( a.2 r=1,3,5
ircW
16H4
cash - - 1
H 8HW
_p
(i7c) sinh i7cW (i7c)*
H
375

Vous aimerez peut-être aussi