Вы находитесь на странице: 1из 11

COMPUTER SIMULATION OF FILM BLOWING

V. Sidiropoulos, J. J. Tian, J. Vlachopoulos


Centre for Advanced Polymer Processing and Design (CAPPA-D).
Department of Chemical Engineering,
McMaster University
Hamilton, ONT., L8S4L7

CANADA

ABSTRACT
The present work is concerned with the numerical simulation of the blown film process. The bubble formation and the
biaxial stretching of the film were studied using a non-isothermal, purely viscous, temperature dependent model. The
model is incorporated into a software package called B-FILMCAD. Special attention has been given to the importance of
the temperature in the modeling of the blown film process. It was found that temperature is by far the most important
modeling parameter. Thus, the inclusion of the temperature effects is imperative for any blown film calculation.
The results show that the majority of the process parameters can be successfully predicted by the employed model,
showing a great proximity between experimental data and theoretical predictions. This was valid for most of the studied
cases despite the diversity of the experimental data. Therefore, with the use of the present model many useful conclusions
can be drawn about blown film production lines.

MATHEMATICAL MODELING
Mass And Force Equilibria
Mathematical models for isothermal film blowing were developed by Pearson and Petrie [1]. In this approach the
bubble is regarded as a thin shell in tension. The tension is produced by both axial traction and internal air pressure.
Gravity and inertia forces are neglected as the bubble is extremely thin. Surface tension and momentum transfer from the
air are neglected as well.
We can further assume that the molten part of the bubble is axisymmetric. Then the problem can be conveniently
referred to intrinsic coordinates. Fig. 1 presents a blown film bubble as well as the prime directions and the intrinsic
coordinate system (meridional, tangential and normal directions).
From mass conservation we obtain:

(1)

The tangential velocity is always zero (axisymmertic problem). The normal velocity is small, thus it can be neglected
without error. By differentiating (1) with respect to m, we obtain the rate of strain tensor, which can be written as [2]:

(2)
where:

(3)
(4)

(5)
From continuity we have:

(6)
Equations 3-6 finally give:

(7)
Equation (7) relates the stretch rates in the three directions. The left-hand side represents the rate of stretching in the
machine direction, while on the right-hand the terms represent the stretch rate in the thickness and tangential directions
respectively [2].
For any film element there should be an equilibrium of forces in all directions. The equilibrium in the normal
direction can be expressed:

(8)
where m, t are the principal curvatures at the position of the element:

(9)

(10)
We can also write an equation describing the force balance on any plane z [2],

(11)
Equation (11) does not include the weight of the film or the cooling air drag on the film. It has been proposed that
gravity force and air drag are of the same order of magnitude. Therefore, since they act on opposite directions, we may
assume that they cancel one another.

The Purely Viscous Approach


For the prediction of the bubble characteristics the material is considered to be Newtonian. Therefore the viscosity is
only a function of temperature.
It is known that the stresses of the film in the m & t directions are much higher than the pressure difference on the two
sides of the film. Therefore we can assume that n0, without much error. Then we can use the Newtonian constitutive
equation, which gives:

(12)
(13)
From (12) and (13), we get:

(14)
Using (14) we obtain the total stress in the meridional and tangential directions:

(15)

(16)
In the same manner we obtain t:

(17)
By substituting (16) and (17) to (8) and (11) we obtain a system of two non-linear ordinary differential equations.
By introducing dimensionless parameters r=R/R0, w=h/h0, equation (11) reduces to [2]:

(18)
where:

(19)
and,

(20)
By substituting (18) into (8) we finally obtain a differential equation containing only the dimensionless radius r [2]:

(21)
Integration of equation (21), provides a prediction of the bubble shape. At the die, the dimensionless radius and
thickness have unit value. Therefore the following boundary conditions can be applied:

(22)

When the air ring effect is included in the simulation, the pressure difference on the two sides of the film (P) is not
considered constant throughout the bubble (due to Venturi effects the pressure difference is higher inside the air ring).
However, we also have an additional boundary condition: the bubble diameter at the end of the air ring is almost equal to
the air ring diameter.
Energy Equilibrium
It is possible to further improve the model by introducing non-isothermal conditions. The viscous dissipation terms
can be neglected [2]. Then, for any temperature higher than the solidification temperature, the heat balance gives [3]:

(23)
and eventually:

(24)
By making equation (24) dimensionless, we finally get [2]:

(25)

where: (26)

(27)
and T/T0 = 1 at the die lips (boundary condition).
The Viscosity Function
The well known exponential function has been used extensively to describe the temperature effect on melt viscosity.

(28)
Equation (28) gives adequate results as long as the temperature is not in the vicinity of the melting (or glass transition)
range. However when the temperature is low enough, cannot be considered constant (its value becomes higher as the
temperature falls). Therefore equation (28) cannot be used over the whole range of the temperatures involved in film
blowing.
In order to achieve a better agreement of the Arrhenius function with viscosity-temperature data (especially around
the solidification point), another term was introduced in the exponent. The temperature dependent viscosity function
becomes:
when T>Ts:

(29)

when T Ts:
In this form the function predicts better the behavior of semi-crystalline polymers in temperatures near the
solidification point. It should be noted that for temperatures far from the solidification point (29) reduces to (28).
COMPARISON BETWEEN SIMULATIONS AND EXPERIMENTAL RESULTS
The experiments were carried out by Butler et al. [4]. They involved an LLDPE resin with narrow molecular weight
distribution and short chain branching. Table 1 presents the parameters of the viscosity function (eq. 29). The spiral die
that had been used, had a lip gap of 1.778mm and diameter of 20.2cm.
The known variables were set to be the frost line height, the final thickness and BUR. The rest of the variables were
calculated. The input variables can be seen in table 2, as reported by Butler et al [4]. For the FLH Butler et al. report a
value of 62.5cm. However their own data indicate that there was film deformation up to 70cm from the die. Previous
simulations [5] which used a FLH value of 62.5cm showed some inconsistencies with the data .
B-FILMCAD was used to obtain predictions with and without the inclusion of the air ring effect. Figure 2 presents the
radius profile (bubble shape) for those cases. It can be seen that both predictions were very close to the real bubble shape.
Figure 3 shows the results of an additional simulation which was performed using the exponential viscosity function
(eq. 28). The improvement of the simulation results with the use of eq. 29 is obvious.
Figures 4-7, present the results of the temperature, time, thickness, and velocity simulations. The temperature profile
and time were predicted very well. The thickness prediction is not as accurate showing an early thinning. The same
applies for the prediction of the film velocity (early acceleration).
Figures 8 and 9 present the strain rate history for the machine and normal directions. The predicted strain rate
maximums are very close to the experimental ones. There are some differences concerning the distance from the die
where these maximum values occur. The inclusion of the air ring effect improves things a little bit.
Parametric Study of the Experiment.
In this part the FLH and BUR were changed in order to observe changes in the processing conditions and film
properties. The die gap was also slightly reduced to 1.688mm. The input variables for those simulations can be seen in
table 3.
Fig. 10 presents the results of the B-FILMCAD simulation with respect to the maximum strain rate in the machine
direction. Fig. 11 presents the experimental results obtained by Butler et al. [4]. The simulation predicts correctly that the
maximum strain rate in the machine direction reduces with the increase of BUR and that it also reduces with the increase
of FLH. Although the actual shape of the surface in fig. 11 is not adequately approximated, the trends are correct.
Other simulations [5] involved data from Gupta [6], Farber [7], and Tas [8]. Comparisons show that the present model
is successful in predicting many important parameters but less accurate for stresses and the stress related parameters --
tension force and inflation pressure.

CONCLUSIONS
The purely viscous, non-isothermal model that has been used in B-FILMCAD gives good results when compared to
experiments. The reason is highly likely to be the accurate prediction of the temperature drop along the bubble, as well as
the improved estimation of the viscosity as a function of temperature.

ACKNOWLEDGMENT
The authors would like to express their appreciation to Ms. Elizabeth Takacs, for her help with the viscosity
measurements and Mr. John Perdikoulias of Brampton Eng., for providing useful insights and information into the
blown film process.

REFERENCES
1. Pearson, J. R. A., and Petrie, C. J. S., Journal of Fluid Mechanics, “The flow of a tubular film. Part 1. Formal
mathematical representation”, vol. 40, part 1, p.1-19, 1970.
2. Tanner, R. I., “Engineering Rheology”, Clarendon Press, Oxford, 1985.
3. Han, C. D. and J. Y. Park, Journal of Applied Polymer Sci., "Studies on blown film extrusion. 2. Analysis of the
Deformation and Heat transfer Process", vol.19, 3277-3290, 1975.
4. Butler, T. I., Patel, R., Lai, S. and Spuria, J. E., TAPPI Proceedings, “Blown film frost line-freeze line interactions”,
13-25, 1993.
5. Sidiropoulos, V., “Comparison of experiments with a model of the blown film process”, M. Eng. Thesis, McMaster
University, Hamilton, 1995.
6. Gupta, R. K., “A non-isothermal rheological constitutive equation and its application to industrial film blowing
process” Ph.D. thesis, University of Delaware, 1980.
7. Farber, R., ”Measurement of deformation rates in the film blowing of polyethylene”, Master’s Thesis, McGill
University, Montreal, 1973.
8. Tas, P. P., “Film blowing: from polymer to Product”, Ph.D. thesis, Eindhoven University of Technology, 1994.

NOMENCLATURE
α : slope angle
β : constants
ε : emissivity
η : viscosity
θ : slope of bubble profile
Κ : Stefan-Boltzmann constant
ρ : film density
a, c, d : constants
B : dimensionless pressure
Fc : dimensionless draw force
D, E : dimensionless numbers
FT : total force on the shell
F : draw force
ht : heat transfer coefficient
Q : volumetric flow rate
∆P : inflation pressure
R : bubble radius
r : dimensionless radius (r=R/R0)
T : temperature
Ta : ambient air temperature
Tf : reference temperature
Ts : solidification temperature
h : film thickness
w : dimensionless thickness (w=h/h0).

Subscript 0 denotes values at z=0


Subscript L denotes values at z=L=FLH
Prime (‘) denotes derivative with respect to z (i.e. α’=dα/dz)
FLH stands for “Frost line height”
BUR stands for “Blow-up ratio”
Fig. 1: Molten zone of bubble and intrinsic coordinate system (n, m, t)

Fig. 2: Comparison between data and prediction, for the bubble radius.
Fig. 3: Comparison of bubble radius predictions with the use of eq. 28 and eq. 29.

Fig. 4: Comparison between data and prediction, for the temperature profile.

Fig. 5: Comparison between data and prediction, for the elapsed time.
Fig. 6: Comparison between data and prediction, for the film thickness.

Fig. 7: Comparison between data and prediction, for the bubble velocity.

Fig. 8: Comparison between data and prediction, for the strain rates history in the Machine direction.
Fig. 9: Comparison between data and prediction, for the strain rates history in the Normal direction.

Fig. 10: Prediction for the maximum strain rate in the machine direction, as a function of BUR and FLH.

Fig.11. Data for the maximum strain in the machine direction [4].
F , at Tf=190°C, (Pa.s) 15,000
a 0.005
c 16.95
d 1
Ts (°C) 107.2
Table 1. Material parameters.

Mass Flux (kg/hr) 125


BUR 3
T0 (°C) 246
FLH (cm) 70
Film thickness (m) 25
Table 2: Input variables for the simulation

Mass Flux (kg/hr) 125


BUR 2.5-3.5
T0 (°C) 246
FLH (cm) 50.8-76.2
Film thickness (m) 25
Table 3: Input variables for parametric study.

Вам также может понравиться