INTRODUCTION
his paper is concerned with the thickness of liquid
T c o a t i n g which can be entrained by a wire or filament
withdrawn vertically through the free surface of a large
quiescent liquid bath. Figure 1 shows the features of the
system of interest, which might form a model of industrial systems for use in producing solid plastic coatings
EXPERIMENTAL
meniscus
region
I
I - 4
coating
bath
Stanley Middleman
100
feed reel
C n : : p
reel
10
coated
wire
u)
.-
-a
0
Ly
camera
It
I
0.
I00
10
coating bath
Paraffin oil
Glycerine F
Glycerine E
80% Glycerine
Polyhall 295
0.1%
0.25%
0.5%
0.75%
1.5%
POlyOX 301
0.5%
0.75%
1.0%
P9
(T,
poise
glcm3 dynelcm
0.46
10
8.6
0.61
10
48
160
350
820
1.1
6.6
23
0.86
1.26
1.26
1.21
1
1
1
1
1
1
1
1
Symbol on
Fig. 5
26.5
64.9
64.9
70.7
72.5
72.5
68.6
66.5
67.0
62.4
62.0
62.0
v
A
n
U
e
iLi
A
RESULTS
Data are correlated in dimensionless terms using the
following definitions:
Coating thickness
D,
= H,(pg/a)
(1)
NCa = p U / a
(2)
Figure 5 shows the theoretical curves of White and
Tallmadge ( 3 ) ,believed to be the best available theory
relating coating thickness to speed of withdrawal. The
theory is limited in two respects. It is, first of all, a low
Capillary number
356
10
i(s-1)
I o2
I o3
speed (non-inertial) theory, which cannot be extrapolated with any confidence to high speed coating beyond
a Capillary number of approximately 0.5. All of the data
presented herein, and the region of dominant commerPOLYMER ENGINEERING AND SCIENCE, APRIL, 1978, Vol. 18, No. 5
0.01
0.I
100
10
Nca
1000
=*
0-
Fig.5 . Dimelisionless correlation ofcoating thickness as a function of wire speed. See Tuble 1 f o r key to .symbols.
R(pg/2a)l2
(3)
DISCUSSION
Newtonian Data
As the Capillary number gets large the data show a
power dependence which may be written as
D,
= 0.15N:S
(N(.,,> 3)
(4)
No other wire coating data are available for comparison to or confirmation of this result. Data are available,
however, for coating entrained by withdrawal of a flat
film from a bath (4)which also suggest a similar square
root dependence ofD, on Capillary number, and which
may be represented by
D,
0.8Nii,2
( N c , > 3)
(5)
(6)
Both the theory and supporting data (5)are for the flat
POLYMER ENGINEERING AND SCIENCE, APRIL, 1978, Vol. 18, No. 5
ci
film case, but one would presume that the idea pertains
to wire coating as well. We see no evidence of inertial
effects in our data, presumably because of the relatively
high viscosity of the fluids used. At the highest Capillary
numbers we calculate a value of y of 0.02 which is too
small, according to theory and evidence, to exhibit this
inertial effect.
It would appear, then, that the Newtonian data
presented are consistent with available related observations. The data are limited in two respects:
Only a single filament diameter was investigated,
giving a Goucher number of 0.08.
The high Capillary number data were obtained
with high viscosity fluids, and so show no inertial effects.
Viscoelastic Data
From the normal stress data ofFig. 4 , it is apparent
that the polyacrylamide solutions are generally more
elastic than the polyethylene oxide solutions. Normal
stresses were not measurable in t h e 0.5 percent
polyethylene oxide solution, which may also be seen
(Fig. 3 ) to be only mildly non-Newtonian. We note in
Fig. 5 that the 0.5 percent polyethylene oxide data fall in
with the Newtonian data, when the zero-shear viscosity
is used in calculating N c n .
One of the difficulties in presenting Fig. 5 for nonNewtonian fluids lies in the lack of a well-defined shear
rate for this flow field. It is not clear how to define the
viscosity appropriate to this flow, and so the magnitude
of the Capillary number is subject to some arbitrary
definition of the shear rate at which the viscosity should
be evaluated. In Fig. 5, all non-Newtonian data are
plotted using the zero-shear viscosity.
The use of any other viscosity would shift any particular set of viscoelastic data horizontally and toward the
left along the N c a axis. Since viscosity does not appear
in D , there would be no vertical shift of the data. Thus,
while it might be possible to displace some sets of viscoelastic data into coincidence with the Newtonian data,
it is not clear that this could be done in an unequivocal
manner, based on an a priori rational choice of viscosity
for each fluid. Indeed, it is clear that no shift in the data
for 1.5 percent polyacrylamide will cause these data to
resemble a Newtonian fluid, especially if the data are
shifted toward the left.
It would appear, then, that the effect of significant
viscoelasticity is to reduce entrainment of liquid by the
filament. The extent of reduction can be as much as an
order of magnitude. Perhaps the most interesting aspect
of the behavior of viscoelastic fluids is the attainment, at
high speed, of an essentially constant, limiting, value of
coating thickness, independent o f f l a m e n t speed. Ifsuch
a result turns out to be generally observed in other
viscoelastic fluids, it has significant implications for
process operability.
Of course, the presence of an asymptotic upper limit
to coating thickness has obvious consequences in process design. Independence ofD, from N ( . , , however, has
additional consequences. For example, one would expect the process to be insensitive to fluctuations in line
speed U and, to a large extent, insensitive to modest
357
Stanley Middleman
changes in temperature. Normally, temperature
changes would affect viscosity strongly, altering the
Capillary number. For a Newtonian fluid this would
lead to significant changes in coating thickness.
Despite the fact that the polyacrylamide solutions
cover a wide range of viscous and elastic properties, the
data suggest an asymptotic value of D, within a very
narrow range, nearly independent of polymer concentration. More data will be required before this feature
can be elucidated clearly and quantitatively. The qualitative implications and features are apparent, however.
One might speculate, at this point, on the mechanism
responsible for the observed results. It may lie in the
character of the entrainment region, which may be regarded as a stretching flow with a short timescale Lagrangian transient. Over a distance of the order of the
meniscus height, fluid elements are accelerated from
near rest to the filament speed. We may estimate an
order of magnitude of the stretch rates that the fluids
studied have been subjected to. The static meniscus
height is given roughly by the value of
(2)
1 /2
Hu
10-2s
(9)
N i m = utd
(10)
358
ACKNOWLEDGMENT
The bulk of the data presented here was obtained in
undergraduate research projects carried out by Thomas
Mumley and Samuel November. Financial support was
provide by the Materials Research Laboratory of the
Polymer Science and Engineering Department, and by
Eastman Kodak Company.
NOMENCLATURE
D,
h/tDDeborah number
R(pg/2u)lRGoucher number
= wire radius (cm)
= time scale of deformation (s)
= wire speed (cm/s)
= 0-/(p/p4g)lGinertial parameter
= stretch rate (s-l)
= fluid relaxation time (s)
= viscosity (poise)
= density (g/cm3)
= surface tension (dyne/cm)
=
=
REFERENCES
1. J. A . Tallrnaclge and C. Gutfinger, Znd. E n g . Chem., 59, 19
(1967).
2. D. A. White and J. A. Tallmadge, A Z C h E J . , 12, 333 (1966).
3. D. A. White and J. A. Tallmadge, AZChEJ., 13, 745 (1967).
4. R. P. Spiers, C. V. Subbaraman, a n d W. L. Wilkinson, Cheni.
E n g . Sci., 29, 389 (1974); 30, 379 (1975).
5. M. N. Esmail arid R. L. Hrimmel, AZChEJ., 21, 958 (1975).