PROFILE EXTRUSION DIE DESIGN: THE EFFECT OF WALL

SLIP

L. L. Ferrás1, J. M. Nóbrega1, F. T. Pinho2,3 , O. S. Carneiro1

1
IPC – Institute for Polymers and Composites, University of Minho, Campus de Azurém, 4800-058 Guimarães,
Portugal – luis.ferras@dep.uminho.pt, mnobrega@dep.uminho.pt, olgasc@dep.uminho.pt; 2Centro de Estudos de
Fenómenos de Transporte, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, 4200-465,
Porto, Portugal - fpinho@fe.up.pt; 3University of Minho, Campus de Azurém, 4800-058 Guimarães, Portugal

Abstract - This work describes the implementation of the slip boundary condition at the flow channel walls in a 3D
finite volume numerical code that is under development. Several phenomenological models relating the velocity and the
shear stress at the wall were implemented. The potentialities of the new numerical code are illustrated performing the
flow simulation of a polymer melt in a complex flow channel where variations of the slip velocity are expected to occur.
The results obtained are qualitatively in accordance with the theoretical expectations.

Introduction

In the past, the design of extrusion dies was usually based on trial-and-error procedures, which strongly depend on the
designer knowledge and experience and often require a large number of iterations. During the last decades, the
progressive development of computational fluid dynamics and of computer technology (enabling the implementation of
more realistic complex rheological models together with more accurate discretization and interpolation schemes)
allowed the progressive replacement of the empirical trial-and-error process by a computational systematic method,
leading to significant savings in time and human and material resources. More recently, this research team successfully
implemented and validated an optimisation methodology [1], which encompasses the numerical solution of flow and
heat transfer equations by a finite-volume based three-dimensional code, aimed at balancing automatically the flow in
complex profile extrusion dies. In a first stage, in this numerical method the no-slip boundary condition at the flow
channel walls was assumed. In fact, the majority of studies concerning the flow in extruders, extrusion dies and
rheometers normally proceed from the assumption that the flowing melt adheres to the wall. However, there are certain
plastic melts such as poly(vinyl chloride) (PVC), high-density polyethylene (HDPE), polypropylene and elastomers,
often used in the production of plastic profiles, that show wall-slipping under certain conditions. Since wall slip within
the die affects the overall velocity field and post-extrusion behaviour, it was decided to implement the wall-slip
boundary condition in the above referred numerical code.
Wall slip of polymer melts has been studied by several authors [2-5], resulting in a comprehensive description and
development of phenomenological models to handle the problem.
The well known incompressible Navier-Stokes equations with no-slip boundary condition [4,6] are:

∇.u = 0 in Ω (1)
∂ρ u (2)
+ ∇.ρ uu = −∇p + ∇.τ + ρ g in Ω
∂t
u.n = 0 in ∂Ω (3)
ut = 0 in ∂Ω (4)

where u is the velocity vector, ut is the velocity vector tangent to the wall , n is the unit normal vector to the surface,
3
ρ is the fluid density, g is the gravitational acceleration, Ω is a simply-connected domain in R , ∂Ω is the boundary of
the domain Ω and the stress tensor σ has been decomposed in the sum of the pressure term p and the traceless stress
term τ [7] (deviatoric stress tensor), σ = − pδ + τ . Therefore, in order to include the wall slip boundary condition,
Equation 4 must be replaced by the Navier slip boundary condition at solid walls [8-10].

Slip Models

With slippage, the tangential velocity can be given by one of the three slip laws [11], listed below, relating the shear
force with the tangential relative velocity. These models represent the relation between the Euclidean norms of slip
velocity and shear force vectors.

The Polymer Processing Society 23rd Annual Meeting

Generalized Navier’s law

eslip −1
f s = Fslip ( vwall − vs ) vs − vwall (5)

where Fslip and eslip are material parameters, f s is the shear force, vs is the tangential velocity of the fluid and vwall is
the tangential velocity of the wall. This law becomes linear if eslip = 1 and a power law if 0 < eslip < 1 . To solve this
equation in order to get vs as a dependent variable, care must be taken in order to avoid unrealistic values.

Treshold law

 yc
− Fslip1(vs − vwall ) se vs − vwall <
 Fslip1 (6)
fs = 
− y − Fslip 2(vs − vwall − yc ) se vs − vwall ≥
yc
 c Fslip1 Fslip1

where Fslip1 and Fslip 2 are two slip coefficients and yc is the critical shear force at which slip starts to occur. Note
that this relation is linear being easy to solve the equation in order to get vs as the dependent variable.

Asymptotic law

  vs − vwall   (7)
f s = − Fslip 1 − exp  
  vc 

where vc is a scale factor (with velocity units) affecting the slope of the slip velocity curve. As in the previous model, it
is easy to get vs as the dependent variable.

Numerical code development

The 3D flow fields are calculated with a computational code based on the finite volume method that is being developed
by the authors [12,13]. It comprises a set of routines to model the relevant physical process and uses the non-staggered
grid arrangement, in which all dependent variables are located at the centre of the control volumes. This greatly
simplifies the adoption of general curvilinear coordinates for the mesh [13], being the Cartesian ( x1 , x2 , x3 , t )
coordinates converted into general curvilinear coordinates (ξ1, ξ 2, ξ3 ,ι ) that fit to the complex geometry using the
transformations x1 = x1 (ξ1, ξ 2, ξ 3 ,ι ), x2 = x2 (ξ1, ξ 2, ξ3 ,ι ) , x3 = x3 (ξ1, ξ 2, ξ3 ,ι ) , t = t (ι ) [14].
The main modifications performed in the numerical code in order to account for slippage are related to the stress term in
the momentum equation (Equation 2) for the generalized Newtonian fluid. The components of the stress tensor can be
written in a generic way as:

µ ∂ui ∂u j  2 µ ∂u
τ ij =  βlj + βli − δ ij βlk k , (8)
J  ∂ξ l ∂ξl  3 J ∂ξ l

where i, j, k = 1, 2,3 are the Cartesian directions, β lj is the cofactor of the lj position in the Jacobian (of the coordinate
change) matrix and l = 1, 2,3 are the directions on the new coordinates. The discretization of the above equation results
in:

f
 µ   3 f 
)
3 3
τ ijf = 
 δ V   l =1
(f f f f 2
  ∑ blj [ ∆ui ]l + bli  ∆u j  l − δ ij ∑∑ blk [ ∆uk ]l 
3 l =1 k =1
f


, (9)

where f = 1, 2,3, 4, 5, 6 is related to each face of the hexahedral elementary cell. It should be noted that after the
discretization the derivatives, e.g. ∂ui , are transformed in a finite difference , e.g. [ ∆ui ] f .
l
∂ξl

The Polymer Processing Society 23rd Annual Meeting

The problem is solved in an iterative way, being necessary to define the initial values for the variables. In this case, the
velocity field near the flow channel wall must be set. When the no-slip boundary condition is used a null tangential
velocity is employed, but the velocity at the wall does not need to be null. In fact, any constant real value can be set for
the initial velocity field since while evaluating velocity differences along the face, a null variation is achieved (if
f
ui = cnt then [ ∆ui ]l =0). On the other hand, when the slip boundary condition is used, the iterative scheme illustrated in
Figure 1 is adopted. The process will stop when convergence is achieved.

Figure 1 – Iterative scheme used to compute the slip velocity.

Case study

In order to access the adequateness of the new boundary condition implemented, the numerical code was used to
simulate the flow in a channel with variable cross-section, shown in Figure 2, where the slip velocity at the wall was,
therefore, expected to vary. The slip model used in this case study was the generalized Navier’s law (Equation 5), with
-2
Fslip = 0.6 and eslip = 3.50831x10 . Other relevant input parameters were the following: melt inlet temperature:
230ºC; channel wall temperature: 250ºC; inlet melt average velocity: 0.1 ms-1, imposed as a rectangular (plug) velocity
profile. The constitutive equation used was the Bird-Carreau

η0 − η ∞ (10)
η (γ& ) = η ∞ + 1−n
(1 + (λγ& ) )
2 2

with η 0 = 53060 Pa.s, η ∞ = 0 , n=0.26 and l=2.27s.

As can be seen in Figure 3, when slippage is taken into account, the velocity profile deviates from parabolic to plug-
like. This effect is also illustrated in Figure 4, where, in a first stage the velocity profile is not yet fully developed and,
in the subsequent instants, become more plug-like due to the progressive increase in the slip velocity at the wall
originated by the reduction of the flow channel cross-section, and consequent increase of the average flow velocity.

Flow
inlet

Figure 2 – Geometry of the flow channel (4 mm and 2 mm height at the entrance

and exit, respectively, and total length of 20 mm).

The Polymer Processing Society 23rd Annual Meeting

 Figure 3 – Fully developed velocity profiles computed at the entrance (parallel zone) with and without slip.

Figure 4 – Velocity profiles along the axial length of the flow channel

As shown in Figure 5, the velocity at the wall is almost constant in the initial parallel zone of the flow channel (between
locations A and B) where the velocity profile is fully developed. In the convergent zone (from C to D), the melt average
velocity increases and, as a consequence of an increase in the wall shear stress, there is also a progressive increase in the
slip velocity. On the contrary, in the divergent zone (from D to E, exit of the channel), the opposite effect is observed.
The steep decrease in the slip velocity observed in the vicinity of the transition zone (location C) can be justified by the
path adopted by the melt in this region. In fact, and as expected, when an abrupt change in the geometry of the flow
channel occurs, the streamlines deviate from the boundary contour of the channel, resulting in a decrease of velocity
near the walls. This hypothesis can be confirmed in Figure 6, where a streamline corresponding to the melt flowing in
the vicinity of the channel walls in the transition zone is shown. As it can be seen, the distance of this line (which is an
iso-velocity line) to the channel wall is higher close to the abrupt transition (location C).

The Polymer Processing Society 23rd Annual Meeting

 Figure 5 – Slip velocities at the wall, along the axial length of the flow channel.

Figure 6 – Streamline at the transition zone (location C).

Conclusion

In this work the implementation of the slip boundary condition at the flow channel walls in a 3D finite volume
numerical code previously developed was described. The potential of the new numerical code was illustrated with a
case study, where the observed variations of the slip velocity were expected. It was shown that the computed velocity
profiles were qualitatively in accordance with the theoretical expectations, thus making the computational tool more
adequate to design extrusion dies for polymer melts that exhibit slip at the walls of the flow channel.

Acknowledgements

The authors gratefully acknowledge funding by FEDER via FCT, Fundação para a Ciência e Tecnologia, under the
POCI 2010 (project POCI/EME/58657/2004) and Plurianual programs.

The Polymer Processing Society 23rd Annual Meeting

References

1. J. M. Nóbrega; O. S. Carneiro; F. T. Pinho; P. J. Oliveira International Polymer Processing 2004, 19(3), 225.
2. H. Potente; H. Ridder; R. V. Cunha Macromolecular Materials and Engineering 2002, 11: 287, 836.
3. S. G. Hatzikiriakos International Polymer Processing 1993, 8, 135.
4. E. Mitsoulis; I. B. Kazatchkov; S. G. Hatzikiriakos Rheol Acta 2005, 44, 418.
5. L. A. Archer; R. G. Larson; Y. -L. Chen. J. Fluid Mech. 1995, 301, 133.
6. M-N. Sabry; A-R. Abdel-Rahim; M. H. Mansour Effect of Slip on Transient Liquid Flow Development in
Microchannels 5th Int. Conf. on the Thermal and Mechanical Simulation and Experiments in Micro-electronics and
Micro-Systems, 2004, 257.
7. P. J. Oliveira, Numerical treatment of the stress terms in non-staggered computational meshes Internal report.
Departamento de Engenharia Electromecânica, Universidade da Beira Interior, 6020 Covilhã, Portugal, 1996.
8. W. J. Silliman; L. E. Scriven Journal of Computational Physics 1980, 34, 287.
9. A. Liakos Computers and Mathematics with Aplications 2004, 48, 1153.
10. A. Liakos Computers and Mathematics with Aplications 2005, 49, 281.
11. Polyflow user manual.
12. P. J. Oliveira; F. T. Pinho; G. A. Pinto J. Non-Newtonian Fluid Mech. 1998, 79, 1.
13. P. J. Oliveira, F. T. Pinho e-rheo.pt 2001, 1, 1.
14. M. Peric, Ph.D. Thesis, University of London, 1985.

The Polymer Processing Society 23rd Annual Meeting