Pergamon Press.
I/
1. INTRODUCTION
-L
3 -
RapionI
1. This is
region
of the liquid in the bath where the film thickness becomes constant at ho, and in which only viscous and
gravity forces are involved.
Region 2. This is normally referred to as the
dynamic meniscus region in which the film thickness, h, varies with height above the bath, x, and in
which viscous, gravity and surface tension forces
have all to be taken into account.
Region 3. In this region, close to the surface of
the bath, flow effects can be neglected to a first
approximation
and only surface tension forces
considered.
from the
389
R. P.
390
SPIERS,
C. V.
SUBBARAMAN,
W. L.
WILKINSON
conditions:
3. THEORETICAL
(i) y=O;u=U
(ii) y = ho; (duldy)
(2)
=0
Uhn(l-!g).
(4)
TREATMENT
=cL
(f-9
The boundary
are:
conditions
(i) u(x, 0) = U
CT
(S)
stress at the free surface:
@ii)
(7)
This result reduces to the Landau-Levich
equation for low speed withdrawal, i.e. where both Ca
and To approach zero, (say Ca < OeOl), and for high
values of Ca (say Cu > lo), it reduces to To = 1,
which is the result obtained by Deryagin[3] based
@iii)
where the stresses on the atmospheric side of the
interface are considered negligible compared to
those in the fluid, 0 is the angle between the normal
to the surface and the y-axis, and R is the radius of
curvature of the free surface.
du
(lli)
dh
(llii)
391
mh3
>
(12)
(3Ca).(l-
To2/3)$
(+
g)
++(L-l+T,,(l-L3)/3)=0.
(18)
and
L=l,
u = U+3(;
-+)(?&
hy).
$0
(13)
at a position
&=-p
-2p$(x,
h).
L+m
(14)
and
(19)
giving:
$x,h)=-&$
(15)
Differentiating Eq. (14) with respect to x, equating with the pressure gradient of Eq. (12) and substituting Eq. (15), we obtain:
dh
w-3p+($$)+~(Uh-q-$=0.
(16)
and hence
d2L
G+mas
L+m.
392
R. P. SPIERS.C. V. SUBBARAMAN,
W. L. WILKINSON
(18) becoming:
(20)
dx
Integration of Eq. (21) is found to satisfy boundary condition (19), and Eq. (20) is also found to
satisfy it. We may then apply the curvature matching procedure of Landau and Levich which requires that the curvature of the free surface is continuous from the constant film thickness region to
the static meniscus region.
From Eq. (20) we can find the limit of the second
derivative at the lower limit of the dynamic meniscus region where it merges with the static meniscus
region in the form:
dZL
m -@ = cu(Ca, To).
(22)
$4
(23)
m as x + 0, and
following
condition
Landau
and
dx
Continuity of the second derivative is then ensured by equating the limits given by Eqs. (22) and
(24) from which we obtain:
To =
Ca
TO
ToC2'
OGol
0.003
0.006
0.01
0.03
0.06
0.1
0.2
0.3
0.6
1.0
0.2829
0.3322
0.3663
0.3926
0.4525
0.4920
0.5215
0.5617
0.5852
0.6246
0.6530
0.2839
0.3346
0.3702
0.3984
0.4651
0.5120
0.5494
0.6043
0.6387
0.7008
0.7480
;:;
4.0
8:;
0.7107
0.6900
0.7249
0.7575
0.7443
0.8455
0.8108
0.8686
0.9155
0.8973
10-o
0.7674
0.9280
region, we have
$/[l+($r]112=5.
Integrating
Further details of the numerical methods employed are given in the Appendix. The result for To
as a function of Ca is given in tabular form in Table
1, and is shown graphically in Figs. 3-7, where it is
(21)
dx
L lim
(26)
dx
(25)
393
Table 2. Experimental
Capacitance
_j_p_ro,bee
Roller to remove
Voriable
speed
drive
Solution
No.
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
Glycerol-water
0.0180
2
3
4
5
6
7
8
9
10
0.0195
0.0280
0.0648
0.1695
0.0628
0.0156
0.0097
0.0062
06408
for each
N.s!m.
1I
12
Syrup-water
Syrup-water
Syrup-water
Syrup-water
Syrup-water
13
14
15
0.1302
0.0980
0.0830
0~0500
0.0220
Liquid Paraffin-I
Liquid Paraffin-II
16
17
0.1695
0.1932
Lubricating
Lubricating
Lubricating
Lubricating
18
19
20
21
0.1318
0.2115
0.7712
2.0580
Sugar
Sugar
Sugar
Sugar
Sugar
5.
fluids
Oil I
Oil II
Oil III
Oil IV
DISCUSSION
OF RESULTS
394
-Tallmadge
II
0
Symbol
12
A
I3
l
14
A
I5
0
0.9
9
6-5
I
co
I
e*
t.0
Landau-Levich
0.6
0.7
0.6
0.5
0.4
0.3
Fluid
Symbol0
2 3 4 5 6 7 6 9 IO
l q A 4 v v l o o
0.2
395
I
0.9
0.6
3;
0.7
O-6
P
0.5
,I
tz
04
16
17
lcJ.arv.
0.3
18
19
20
21
02
.
0.9t
0.6
C&L!
Q
twenty times for each withdrawal speed, and his results, though very few,
agree will with the present theory at low Capillary
Numbers. It is worthwhile to point out that the present work was attempted initially with a micrometer similar to that of Gutfinger, and Soroka. Using
this technique we experienced many of the difficulties encountered by Gutfinger[7] these being:
(i) the accurate measurement of thin films,
(ii) the inaccuracy caused by the inevitable lateral movements of the belt , and
(iii) the formation of surface waves, particularly
as the speed increased.
The experimental set-up used here reduces these
difficulties considerably, yielding accurate, reproducible data for a wide range of Newtonian liquids.
A similar range of experimental data obtained by
Groenveld[9] also showed a levelling off but at
lower values of Cu. This may be due to the experimental arrangement since a large rotating disc was
used and not a vertical belt. His data are not, therefore, compared with the present theory.
The discrepancy between the theory and experiment at high values of Cu is to be expected since
the assumption of one-dimensional
flow in the
meniscus region is no longer valid, and capillary
statics no longer adequately describe the shape of
the film near to the bath surface. For many problems of industrial interest this limitation is unlikely
to be serious.
6. CONCLUSIONS
The modified
gravity-corrected
withdrawal
theory gives predictions which are significantly
different from previous treatments and accurately
predicts the relationship between TO and Cu for
Newtonian fluids up to a value of Ca of about 2.
The experimental
techniques used in this work
have been shown to be capable of higher accuracy
than those previously used for studies of the withdrawal problem.
3%
it
ha
L
P
4
R
TO
;
x
Y
Greek symbols
APPENDIX
NUMERICAL SOLUTION OF EQ. (20)
As the limiting boundary conditions given by Eq. (26)
for Eq. (20) are not suitable for the numerical integration
of the latter, we choose some position above the bath such
that:
L=l+E
(Al-l)
d2L
-=e
dx
where e is a small parameter. These are the starting values of Landau and Levich[l]. It should also be noted that
an origin is not uniquely defined for Eq. (20) and we are
thus free to choose it such that the boundary conditions
(Al-l) are given at x = 0. We may then let x decrease
through the numerical integration. The Merson version of
the fourth order Runge-Kutta method[lO] was used,
enabling the error at each step and the accumulated error
to be examined as the integration proceeded.
To perform the integration it is necessary to select suitable values of l and Ax,, the initial step length. Values of
l and AX, were therefore varied over a twenty fold range,
keeping Ax,/r constant. Little variation was found in the
final value of a. This is in contrast to the second order
method employed by White and Tallmadge[2], details of
which are given by White [ 111,where extrapolation to zero
step length was required to obtain a. The final values for
e and AX, chosen were:
AX, = 5 x lo-
(Al-2)
6 = 5 x lo-
Because the initial steps of the integration need to be
co-ordinate,
&(l-
accurate
REFERENCES
(Al-3)
The use of the variation leads to a considerable reduction in the overall computation time compared to the use
of a fixed step length. It was found that approximately 500
steps were required to achieve a constant second derivative $$
Newton-Raphson
techniques.
to find s
for