Professional Documents
Culture Documents
‘)=mi 4 $ ~ ~ * ) r (4)
T angential flow in cylindrical annuli is of interest in
the extrusion theory of polymer melts. Polymer melts
in flow exhibit a non-Newtonian viscosity, normal
The two constants m and n are temperature and pres-
sure dependent. The flaws in this model and its insuffi-
ciency to represent the shear stress and shear rate rela-
stresses and time dependent effects. Inclusion of the last tionship over the whole range of shear rates were pre-
two phenomena in a screw extrusion theory leads to viously discussed (3, 4a). An important deficiency is
great diaculties due partly to the complex geometry that at zero shear rate it predicts infinite apparent vis-
that is involved. It is, therefore, not surprising that most cosity for a pseudoplastic fluid, while polymer melts
of the attempts to further elaborate on the existing tend to a Newtonian behavior at low shear rates. There-
theories concentrate upon the inclusion of the first phe- fore, for those flow situations where the shear rate at-
nomena, i.e., non-Newtonian viscosity, although normal tains very low values or passes a value of zero, a certain
stresses may have a great significance due to their un- error can be expected in the flow rate calculations. A
symmetrical distribution across the channel, while visco- rough approximation of this error is carried out in Ap-
elastic effects may be especially important in a compres- pendix 1. Pressure flows between two infinite parallel
sion-relief type screw. plates of two fluids are compared: one fluid is assumed
A constitutive equation that describes a non-New- to obey the “power law” over the whole range of shzar
tonian viscosity but excludes normal stresses and time rate, while the other obeys it until the shear rate be-
dependent viscoelastic effects is ( l a ) : comes smaller than a certain value after which the fluid
is assumed to be Newtonian. The results indicate that
even for a very pseudoplastic fluid (n = 0.25) the error
- becomes significant only when the Newtonian range
where 7 and aare the stress and rate of deformation covers more than half of the cross section. Considering
tensors respectively and 9 is a scalar, which for an in- the shear rate profile in a real extruder, it often will
compressible fluid is a function of the second and third happen that the shear rate will not pass a zero value
-
invariants of 2 defined in Equations 2 and 3, respec- nor will it attain very low values. This is shown in
tively. Appendix 2. Somewhat encouraged by these results, and
- -
- - furthermore, being aware of the fact that only a rela-
lg=A:A (2) tively simple model such as the “power law” provides
- results that can be presented and plotted in a simple
I, = det (3) form, the “power law” model was applied in the calcu-
In simple flows, such as axial flow in a tube, Couette lations presented below. Relatively simple results that
flow, I8 vanishes. However, Tanner (2) has shown re- can be plotted as graphs are important for practical de-
cently that for an axisymmetrical flow in a duct of vary- sign purposes.
ing diameter, I, will vary. Such a situation exists in the The conventional screw extrusion theory is based upon
compression zone of a screw extruder. But, since the a Newtonian flow between infinite parallel plates. The
dependence of 7 on I, is unknown, it is usually assumed solution of the flow equations results in two independent
that r ) is a function of I, only. Experimental observations terms-ne for the drag flow and the other for the pres-
led to several models for this function with two or more sure flow. These independent terms are then separately
adjustable parameters. Bird (3) reviewed these models corrected for wall effect, curvature, and radial tempera-
in relation to the rheological behavior of polymer melts. ture distribution in order to account for some of the sim-
d
p-(">'O
dP RIP
The absolute value in Equation 12 can be eliminated
by rewriting it as:
[ =--(--1)
1 d P e (15)
When the value of r for a given fi and s is such that 5"
= 1, the shear stress at the inner cylinder is zero. These
dP RIP 2m dB p- I' values can be calculated from Equation 21 by substi-
which after some rearrangement and integration of the tuting 5" = 1. If the pressure gradient is increased to
left hand side gives: very high values or if the velocity 1%is decreased to-
P ward zero, a pure pressure flow is obtained. For this
case Equations 19, 20, and 21 become:
-v
=u- 2pr A(C-l)'dp (16)
vu 1 P P -
m
1 dP (- 1)1"
where s = l / n and the dimensionless group r is defined
re-(--)
Ri 1 dP '
2V, 2m dB
The integral in Equation 16 cannot be solved in
closed form for all real s, but by expansion into an in- a
1 dP 1
finite series an open form solution can be obtained. De-
%?
fining a new variable Z = --r - 1 Equation 16 becomes
P
[ ( I-$)"' -(1-$)'"] E"p'/? (23)
1
The term -
1+z
can be expanded into an infinite, uni-
Shear Stress
Profile Tangential Velocity Profile Equation to Solve for p
i + integer
_ _ ~ ~ ~ _ _ ~
Table II. Stress Profiles, Velocity Profiles, and Flow Rate for Tangential Combined Pressure and Drag Flow
When the Inner Cylinder Wet Surface has a Tangential Velocity Ve and a Constant Angular Pressure
dP
Gradient -
d6
is Maintained
Shear Stress
Profile Tangential Velocity Profile Equation to Solve for p
___
m
vs
-
v a =p- pr C -[s + I
'=I
(-11'+1
( p - 1 >'+I -('- P 1)"' 3
1 L p L E e"<2
m
[ + (-l)'(r-l)s+q (29)
16F<2
Finally, for a Newtonian fluid, Equation 29 with Equa-
tion 25 converges, as before, to the independently ob-
tainable solution:
-
Figure 2. Tangential oelocity profiles in a combined drag and pressure flow between two con-
centric cylinders. Full line shows the exact profile, broken line shows the profile calculated by
superposition of drug and pressure flow. Q P d / Q d l/i, and /3 = 1.6.
2.5
2.0
3 1.5
3
5
:
UI
10
0.5
0.0
-0.5
-1.0 -0.5 0.0 0.5 1.0
(od+Qp)/Qd
Figure 5. Correction factor to account for curuature in a Figure 6. Correction factor to account for curvature in a
pure pressure flow versus p. The angular pressure gradient pure pressure flow versus 8. The angutar pressure gradient
was compared to the linear pressure gradient at the circum- was compared to the linear pressure gradient at the arithmetic
ference of the outer cylinder. mean circumference of the cylinders.
tures, /3 = 1.02 and /3 = 1.6 which shows that the in- ders, therefore, an additional correction factor 5 was
fluence of curvature is mild in general except when drag derived defined as:
flow is approached. Experimental quantitative verifica-
tion of the error in superposition in extrusion calcula- 5 = Qp/Qp, between parallel plates (33)
tions is not easy because the pressure gradient varies Qp is shown in Equation 29 and pressure flow of a
along the screw (10) and the flow is not isothermil. power law fluid model between parallel plates can be
However, if the exact experimental temperature and easily derived (9b). The value of 5 depends on the cir-
pressure profiles along the screw are available, one can cumference along which the angular pressure gradient is
select a portion of the screw where isothermal conditions compared to the linear one between parallel plates.
prevail and a linear pressure gradient exists and coin- Figure 5 shows [o versus /3 when the comparison was
pare the calculated flow rate in that section with the made at the outer cylinder circumference, and Figure 6
experimentally measured flow rate. This work is in prog- shows tY versus /3 when the comparison is made at the
ress in our laboratories. Essentially the same is being arithmetic mean circumference of the two cylinders. In
done by including the superposition correction factor in the later case the correction factor is close to 1.0 except
a computer program that simulates screw extrusion per- for high values of /3 and very pseudoplastic fluids. This
formance (11).In this case the correction factor has was concluded previously by b o y (8) for the specific
to be evaluated for negative pressure gradients as well case of Newtonian fluids.
since this is the case in different sections in the extruder.
This would mean an extension of the curves in Figure 4
APPENDIX 1.
+
for (Qd QD)/Qd > 1.
Estimation of the error involved in using the
In conclusion, for engineering design purposes, the 11
drag and pressure flows can be superimposed provided power law” model fluid for a polymer melt.
that the so obtained net flow rate is multiplied by the A pressure flow between two infinite parallel plates
correction factor E . It must be remembered, however, is considered. Figure 7a shows the velocity profile for a
that other correction factors commonly used, such as “power law” fluid model. Figure 7b shows the velocity
those which account for channel shape, and for radial profile when the fluid behavior is described by the
temperature distribution, were derived separately for “power law” model in the region h 4 1x1 6 H and is a
drag and pressure flows and may introduce additional Newtonian in the region 0 6 1x1 6 h. In other words it
errors. is assumed that at 1x1 < h the shear rate becomes small
Pressure flow calculations between parallel plates are so that the fluid behaves as a Newtonian fluid. The
much simpler than tangential flow in concentric cylin- higher the pressure gradient the smaller the ratio h/H
X=-H
X=-h-
x=- n
Newtonian
"power law'
z
jj
I
at which this transfer occurs. The stress profile for the
case in Figure 7h is:
T
rill = - -X
~ =' -~
dP
dZ
dP
-~
dZ
x
h L 1x1 6 H
T~~~
(34)
(35)
= 0.
'
The Newtonian viscosity can be expressed in terms of
the stress and shear rate at x = h
0.5 1 I
0.0 0.2 0.4 0.6 0.8 I.0
h/ H
0 h
-I .o
0.0 0.5 I .o
Vz Vz.mox.
(Q)h=o = 2 [
-
s+2
;(-%)Is H" (45)
l o o 0 lloo
50
By dividing Equation 44 with Equation 45 the following
result is obtained:
r,8 = 49.82 (-
0.8589
Pa
- 1) t 47)
The shear rate profile is positive over the whole range
and it is:
rre and Are are plotted in Figure 10. The shear rate
z""
at any point falls within the range of shear rates in 2 b-1 I- 1)'dZ (50)
which the polymer obeys the "power law" model.
+
By expanding 1/(1 Z) in the integral in Equation 50
into an infinite series followed by a termwise integration,
a series is obtained for each term.
z"t?
z"
dZ = [ sfl
z""
- -+
s+2
--
z"+S
s+3
+ ... 1'
<?-,
Substituting Equation 52 into Equation 49 followed by
a similar solution of the second integral in Equation 49
gives the flow rate presented in Equation 28.
2nd
z"'?
dZ=[ --
s+2
+ References
--
Z"':'
s+3
+...I" f2-I
1. Bird, R. B.,,, Stewart W. E., and Lightfoot, E. M., "Transport
Phenomena John Wiley & Sons, Inc., New York, 1963. (a) p.
102 (b) p 's5 (c) 89
2. Tadner, R: I.,'Ind.%ng.'Chem. Fundamentals, 5, 55 (1966).
3. Bird. R. B.. A.I.Ch.E., Chem. Eng. Progress Symposmm Series,
3rd 61 86 (1965).
4. B e h a r d t E C., "Processing of Thermoplastic Materials," Rein-
z"'2 hold Pubiishing Corporation, New York, 1960. (a) p. 29, (b) p.
+
j - d Z1= [ z
Gl + 171, (c) p. 173.
5. Jacobi, H. R., "Screw Extrusion of Plastics," Iliffe Books, Lon-
don 1963.
--
z"+a
s+3
+...I0 +,
6. Krohsser, F. W. and Middleman, S., Polymer Eng. and Science, 5,
231 (1965).
7. Rotem Z. and Shinnar R Chem. Eng. Science 15 130 (1061).
8. Booy, M. i.,Soc. Plasti& I&.Journal:, 19, 176 '( 1983).
9, McKelve J M "Polymer Processing John Wiley & Sons, Inc.,
New Y o r c i96z" (a) p. 232 (b) p. 106.
The solution of the integral will be obtained by summing 10, Marsball D I., Klein, I., aGd Uhl, R. H.,SOC.Plastics. Eng. Jour-
nal 21 i192 (1965).
all the terms 11. Imhch'Klein and D. 1. Marshall, July 1966 Polymer Eng. & Sci.
(111).
. . m 12. Klein, I.. Marshall. D. I., and Friehe, C. A., SOC.Plastics Eng.
Journal, 21, 1299 (1965).
(- I)'+' dZ =
a
Acknowledgments
The author wishes to express his gratitude to Drs. D. I. Marshall
and I. Klein for helpful discussions, to MI. I. Duvdevani for his helpful
criticism and su gesting the calculations in Appendix 1, to Mr. K. E.
Larson for proofreading the mathematical derivations and to Messrs.
Substituting Equation 51 into Equation 50 gives: F. R. Cadic and F. J. Cammarano for aid on numerical computations.