Академический Документы
Профессиональный Документы
Культура Документы
Abstract: This report is a detailed analysis of the drag out of an infinite plate from a pool of viscous
liquid. Many industrial processes involve coating of surfaces that can be modeled on this setting. This
report is one such effort to explain such a setting and thus incorporate it on an industrial level. Benilov
and Zubkov (2008) attempted to solve the problem by simulation of Deraguin’s paper (1945).
Setback of Deraguin’s paper: In the case of Deraguin (1945), the thickness of the liquid film was
determined by the emptying of a liquid from a container. The problem admits to infinite solutions all
corresponding to different loads and it is unclear why only (1.1) is selected from the others. Secondly, the
formula does not have any rigorous mathematical proof; hence it is difficult to establish any physical
limitations. In this paper however, Benilov and Zubkov (2008) have attempted to solve the above
problems to an extent.
This report includes the analysis of drag out and the thickness of the liquid film without incorporating the
effects of surface tension and inertia, also the flow is considered to be under the ‘Creeping Flow’ regime.
It includes the comparison of the same plate being dragged out vertically and at an angle from an infinite
pool of viscous liquid and incorporating the ‘Capillary number’, which is a dimensionless no. that
compares the intensity of viscosity and surface tension. It is tough to neglect the effect of surface tension
paper. The fluid is assumed to be incompressible and is considered to be a Newtonian liquid. In this
paper the surface tension is assumed to be zero. Following Deraguin (1945) we shall assume that the
inertia and surface tension are negligible. Thus the flow is governed by Stokes’ equations
In the Cartesian co-ordinate system, Stokes’ equation is used and the flow equations obtained (2.1). Also
continuity is satisfied (2.2). No-slip boundary condition is applied at the plate (2.3). The other boundary
condition is the effect of zero stress on the surface of the film (2.5). The effect dewetting is considered to
be negligible. Dwetting is the process of rupture of the thin film and is characteristic of the spreading co-
efficient. This gives rise to the presence of a contact point which is considered to move with the plate
corresponding to the boundary condition (2.7) and (2.8). Due to the presence of the presence of the
boundary layer we introduce the non-dimensional variables (2.9) and (2.10) into the Stokes’ set and
Assuming that the slope of the plate is small i.e. ε<<1 lubrication analysis is used to treat equations
(2.11)-(2.15) as a boundary value problem and it is expressed as expansions in ε2 to obtain (2.19). Thus,
The paper is concerned with the effect of negligible surface tension, but it can be used to discuss the
capillary effects on this system by modifying (2.20). Thus the incorporation of surface tension gives
(2.23). The surface tension and capillarity effects are discussed in the critique and analysis part. Here
the discussion will continue without the inclusion of the surface tension or capillarity.
The steady state emerges only at the edge of the pool and the film’s tip remains unsteady at large times
t>>1, the two regions of the film can be separated and examined separately.
which matches (3.4). On simulating this equation the observed range of thickness of the liquid film
under steady state was found to be as shown in equation (3.5). It also turns out that the characteristic
This load can be found by assuming the solution to be of the form (3.6) and (3.7). After several
mathematical variations and simulations equation (3.17) is obtained which is a composite asymptotic
solution. If the solution outside the load were steady, then the mean thickness would be equal to that
of the load l, but, since the film is mainly unsteady and its thickness varies from l to 0, the mean is
less than l. To calculate the mean thickness for t>>1, one can approximate the profile of the solution
by taking into account that the film thickness grows linearly with time; the non-dimensional thickness
obtained is found to be ⅔. Dimensionally the film’s mean thickness is 1.5 times smaller than the load
If the slope of the film is not small then the Stokes’ equation does not involve parameters that can be
solved numerically as they would have to incorporate the effect of surface tension on the plate. This
work is carried out by Jin et al. (2005) for a more general case that has been covered in the critique
and analysis. In the present work the surface tension is neglected as the asymptotic results need to be
verified.
The Stokes’ equation (2.11)-(2.18) was solved using the Incompressible Navier Stokes’ equation of
the COMSOL package. The parameters were chosen such that in all the runs the Reynolds no. was
10-4, as a result the inertia was negligible. The results are verified by Figure 8. It is observed that this
small slope approximation is applicable only α<35°. Also the range of the error is 10-3 which is better
It is observed that from the Figure 8 that at 35° there is an abrupt turn in the numerical curve from the
asymptotic curve. This can be attributed to the fact that the solution has to begin from an initial condition
of zero load and as a result the system cannot reach a high load situation without passing through a low-
load one. The low load situation is however the more stable one. Once this is reached, the system stops
Aim: To compare the cases of drag out where an infinite plate is being retracted vertically and the other
case where the same plate is being withdrawn at an angle with the inclusion of the capillary effects.
It should be noted that Deraguin (1945) assumed the plate to motionless and the level of the liquid to be
dropping at a constant speed. Benilov and Zubkov (2008) and Bo Jin et al (2005) make use of a time
dependant frame to obtain the value of the load. The drag-out problem of a plate under vertical conditions
and thus determination of the asymptotic film thickness is computed as a function of capillary no. under
creeping flow conditions. The characteristic length scale chosen is (l=µU/ρg) ½ unlike Deraguin (1945)
where the length scale is found to be (l=µU/ρgsin α) 1/2 it can be shown that the parameters can be grouped
into two independent dimensionless groups (a) Capillary number Ca= µU/σ, (b) Reynolds’s number
Re=ρUl/µ and (c) m= (ρσ3/µ4g) 1/2 which is a dimensionless parameter containing fluid properties. The
load was found to be q=ho-⅓h3o. This analysis revealed that at small Ca, q=0.946Ca1/6-0.389Ca1/2. There is
some uncertainty regarding the corresponding value of q as Ca→infinity. Deraguin (1945) proposed that
the load q should reach ⅔ of its maximum value. From the experiments however this proof is found to be
questionable, qmax=0.582. This result is attributed to the flow field around the stagnation point and it is
dominated by only viscous forces and gravitational forces with the effects of interfacial tension being
completely negligible everywhere in the limit Ca tends to infinity (or σ→0). Thus the introduction of the
interfacial tension and inertia gives rise to the formation of steady wavy patterns on the air-liquid
interface and accounts to the variation in the load from the conjectured value.
Similarly, for plates that are inclined, it must be noted that the plates are inclined differently than that of
the original setting used by Deraguin hence the length scale varies from the original equation (1.1) by a
cosine factor instead of a sine factor. By using the routine non-dimensionalization of the Stokes’ equation
to obtain boundary conditions that are exclusively independent of α. It was observed that as α decreased,
it was seemingly difficult to obtain a convergent solution for appropriate values of Ca. It was found that
the load increased monotonically with Ca. With Re=0, it is predicted that for small slopes the Deraguin’s
equation is valid and the load is found to approach ⅔ but it increases for α>45°. (Bo Jin et al. 2005)
In the original paper [1] αcritical=35° however in the other paper [2] the value of αcritical=45° there is about a 10°
difference between the two values. This is owing to the fact that (Benilov and Zubkov, 2008) neglect the
interfacial tension and inertia, but in the paper by (Bo Jin et al, 2005) interfacial tension and capillary
Conclusion: This analysis is crucial from an industrial point of view. There are several unanswered
questions in the paper. Benilov and Zubkov (2008) fail to discuss cases where the Reynolds no. is not in
the Creeping flow regime. The application of Lubrication analysis and boundary layer cannot be applied
unless the Re→0. The range the steady state load 1≤ho≤√2 is questionable. Since simulation and computer
parameters are a direct consequence of the input variables, it is unclear how and why the chosen
The paper is a laborious attempt to solve Deraguin’s equation and is quite a successful one, although there
are several unanswered questions. The paper as a whole has catered to a special case which can be
incorporated on an industrial level. Also the range of error 10-3 determined in this region is remarkably
References
1. Benilov, E.S., and Zubkov V.S. 2008. On the drag-out problem in liquid film theory. J Fluid
Mech. (2008). vol. 617. pg/s 283-289.
2. Jin Bo and Archivos Andreas. 2005. The drag out in film coating. Physics of fluids (2005). vol 17
103603
3. Deraguin B., 1945. On the thickness of the liquid film adhering to the walls of the vessel after
emptying. Acta Physicochem, USSR. Vol XX, No 3