See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230555671

Numerical Modeling of the Flow of a Power Law Ceramic Slurry in the Tape
Casting Process

Conference Paper · June 2012

CITATIONS READS

5 3,312

2 authors:

Masoud Jabbari J.H. Hattel

The University of Manchester Technical University of Denmark
50 PUBLICATIONS   443 CITATIONS    293 PUBLICATIONS   4,806 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

BladeRunner View project

Numerical Modelling of Local Climate inside Electronic Enclosures- Humidity and Thermal Management View project

All content following this page was uploaded by Masoud Jabbari on 21 May 2014.

The user has requested enhancement of the downloaded file.

 M. Jabbari and J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.

ISBN 978-0-9866504-9-9 / ISBN 978-0-9879945-0-9

ATEMA’2012
AES Montreal, CA

CAN’2012 TENTH AES-ATEMA INTERNATIONAL CONFERENCE

AES-ATEMA’ 2012 Tenth International Conference on Advances and Trends in Engineering
Materials and their Applications
(Montreal, CANADA: June 18 – 22, 2012)

Numerical Modeling of the Flow of a Power Law Ceramic Slurry in

the Tape Casting Process
M. Jabbari 1*, J. H. Hattel 2
1
PhD Student, Dept. of Mechanical Engineering, Technical University of Denmark, Nils Koppels
Allé, 2800 Kgs. Lyngby, Denmark.
(Email:mjab@mek.dtu.dk)
2
Professor, Dept. of Mechanical Engineering, Technical University of Denmark, Nils Koppels Allé,
2800 Kgs. Lyngby, Denmark.
(Email:jhat@mek.dtu.dk)
*Corresponding Author

Abstract numerical modelling for the tape thickness.

Multilayer ceramics and their application have This model is then used for different values of
increased recently. One of the most common substrate velocity and material load in the
ways used to produce these products is tape reservoir and compared with experimental
casting. In this process the wet tape thickness findings of the wet tape thickness and good
is one of the most determining parameters agreement is found.
affecting the final properties of the product and
it is therefore of great interest to be able to Keywords
Tape casting, Doctor Blade, Fluid Flow, Non-
control it. In the present work the flow of
Newtonian, Power Law.
La0.85Sr0.15MnO3 (LSM) material in the doctor
blade region is modelled numerically with 1 Introduction
ANSYS Fluent in combination with an Ostwald Tape casting is a forming method that has
power law constitutive equation. Based on mainly been used in the electronics industry to
151

rheometer experiments the constants in the produce multi-layer capacitors and electronic
Ostwald power law are identified for the substrates [1,2]. This method basically starts
Page

considered LSM material and applied in the with a specially produced slurry which can be

© 2012, Advanced Engineering Solutions, Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
cast by a blade to a flat sheet or layer, then deposition and fluid jetting devices in which
dried into a flexible solid tape which can be material interfaces are present. Tracking of free
sintered subsequently into a hard ceramic surface is also considered in multi-material
substrate layer [3]. This technique is a well- flows with sharp immiscible interfaces [20].
established process which is used to produce Several CFD methods have been developed in
ceramic layers and multi-layered ceramics recent years with the aim of simulating such
(MLC). The general schematic of the process is complex flows. Among them are the volume of
illustrated in Figure 1. The parallel (doctor) fluid (VOF) and level set methods which are
blade process was first used in preparing able to cope with the presence of material
ceramic tapes in the 1940s and it has a key role interface.
in producing thin and flat ceramic tapes [4,5].
In the tape casting process, the ceramic slurry
is mostly categorized as a non-Newtonian flow
with relatively high viscosity. A viscoplastic
constitutive law was used by Huang et al. [6] to
model the flow field in tape casting. In their 2D
analytical model, the effect of pressure Figure 1. Schematic illustration of the tape
gradient, substrate velocity and resultant tape casting process.
thickness was evaluated. Their work combine
Binhgam plasticity with a shear thinning
behavior described by a power law. The effect
of different rheological behaviors of the tape
slurry (Newtonian, power law and Bingham
plasticity) for a generalized pressure flow in
tape casting was investigated by Joshi and et al.
[7]. They estimated the tape thickness
analytically and controlled the size of the
parallel channel in tape casting accordingly. In
most engineering applications, the flow of non-
Newtonian fluids are characterized by the so-
called Bingham and Herschel-Bulkley Figure 2. Rheological classification of fluids.
(viscoplastic) and Ostwald de Waele power law
(pseudoplastic) constitutive models [7-19], In the present work the flow of
which are shown in Figure 2. La0.85Sr0.15MnO3 (LSM) ceramic material is
Flow processes often involve the presence of modeled numerically with ANSYS Fluent in
free surfaces, the tracking of which has combination with an Ostwald power law
significant impact on the manufacturing and
152

constitutive equation. This model is then used

the final quality of the product. Examples for different values of substrate velocity and
abound, e.g., casting processes, mold filling, material load in the reservoir and compared
Page

thin film processes, extrusion, coatings, spray with experimental findings of the wet tape

© 2012, Advanced Engineering Solutions , Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
thickness and good agreement is found.
(6)

2 Problem Formulation 2.2 Discretization

2.1 Governing equations A schematic illustration of the computational
When dealing with flow problems, the Navier- domain with design parameters is shown in
Stokes equations should be solved. This system Figure 3. The calculation domain is designed in
in the form of the momentum and continuity 3D, but in the z direction the length is equal to
equations is as below: a cell size with symmetry plane in both sides.
By this simplification, the model turns out to
(1)
be in 2D dimensions and hence also the side
effect of flow outside the doctor blade region
(2)
which has relatively low influence on the wet
where is density, is velocity, is pressure, tape thickness is neglected. The dimensions are
is stress tensor and is the contribution from chosen relevant to the machine design of the
external forces. Here, the momentum equation tape caster. The domain is discretized with a
(1) expresses Newton’s second law of motion, relatively fine and structured mesh (the cell
and the continuity equation (2) ensures size is equal to ). To compare the
conservation of mass. For the power law fluid numerical results with corresponding
the shear stress, , is given by the constitutive experiment ones, the doctor blade width ( )
equation: and doctor blade height ( ) are kept constant.

(3)

In the VOF model, a single set of momentum

equations is shared by the fluids, and the
volume fraction of each of the fluids in each
computational cell is tracked throughout the
domain. The VOF method was first proposed
by Hirt and Nichols [21]. The VOF formulation Figure 3. Computational domain and the
relies on the fact that two or more fluids (or different configurations at t=0.
phases) are not interpenetrating. The properties
appearing in the transport equation are 2.3 Boundary conditions
determined by the presence of the component For the inlet boundary condition the velocity is
phases in each control volume. In a two-phase set to a very small value ( ), just to
system, for example, if the phases are avoid the initial zero boundary condition. A
represented by the subscripts 1 and 2, and if the fixed velocity in the -direction ( ) is
volume fraction of the second of these is being implemented for the substrate as the velocity of
tracked, the density in each cell is given by the peeling belt with no slip condition. Zero
153

(4) gradient is assumed for all flow properties in

The evolution of the scalar (volume fraction) the outlet boundary condition. All other
Page

is governed by the simple advection equation: boundaries are implemented as wall boundaries

© 2012, Advanced Engineering Solutions, Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
with no slip condition. 4 Results and Discussion
4.1 Power law constants
3 Experiments The results of the rheology experiment showed
Experiments were conducted using a one- that the LSM slurry follows the Ostwald power
doctor blade continuous type of casting bench, law fluid behavior. This is illustrated in Figure
i.e. the tank with doctor blade is fixed, and the 4 and the relationship between shear rate and
plastic carrier (in the present case Mylar) is shear stress is found to be:
driven by a stainless steel roller. The doctor (7)
blade gap distance was set by using etalon
sticks and a micrometer screw with an
accuracy of .
The rheological parameters of the slurry were
obtained using a narrow-gap parallel plate
sensor system in the Rheometer HAAKE Rheo
Stress 600 (Haake, Germany).
Experiments were conducted in two different
cases, which are summarized in Table 1. For
the first set of experiments, the same amount of
ceramic slurry (constant ) were casted with
three different substrate velocities. Figure 4. Rheology behaviour of LSM ceramic
slurry.
Table 1. Summary of two experimental case
studies. From Eq. (7) the constant which is the
Parameter Case 1 Case 2 consistency of the fluid is found to be equal to
600 600 3.3127. Moreover, the constant which is the
1000 400 amount of deviation from a Newtonian fluid is
3.67, 10, equal to 0.9038 (indicating close to a
3.67
14.67 Newtonian behavior). These parameters are
3536, 7074, implemented in the numerical calculations for
1200 the behavior of the non-Newtonian LSM
21222
ceramic.
Then in case 2, during the constant substrate 4.2 Case 1
velocity the material load in the reservoir ( ) The aim of this first case study is to evaluate
is varied. These experiments were carried out the effect of velocity changes in the peeling
to investigate the effect of two important substrate on the tape thickness both
parameters, hydrostatic pressure and drag experimentally and numerically. The results of
the numerical modeling and the experiments
154

force, on the tape thickness ( ).

are shown in Figure 5. As seen, an increased
substrate velocity results in decreasing of the
Page

tape thickness. By further increase of the

© 2012, Advanced Engineering Solutions , Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
substrate velocity it is found that the reduction decrease the final height of the tape especially
in tape thickness decreases and it seems to for higher pressures. However, the slurry
become constant. The drag force is increased height in the reservoir is decreasing over time
by increasing the substrate velocity and it which causes a non-uniform tape with some
becomes more dominant compared to the oscillation in the interface. These variations in
pressure force which results in more stretching the tape thickness are measured along the
of the slurry over the peeling belt. On the casting direction. Results of the measurements
contrary, increasing the substrate velocity for a are shown in Figure 7(a) and (b) for the slurry
constant pressure force, the thickness of tape is height of 7074 and 21222 , respectively.
decreased in the exit and vice versa.

Figure 5. Comparison of the numerical Figure 6. Comparison of the numerical

predictions and experiments for the effect of predictions and experiments for the effect of
the substrate velocity on the tape thickness the slurry height on the tape thickness (doctor
(doctor blade height is equal to ). blade height is equal to ).

4.3 Case 2 Two types of information can be observed from

The effect of the material load (slurry height) Figure 7. In the early stage of the process, the
in the reservoir ( ) on the tape thickness is tape thickness is higher than that of the end due
investigated in this case study. Results of both to a decrease in the slurry height in the
numerical simulations and experiments are reservoir over time. Moreover, this difference
illustrated in Figure 6. As the slurry height between the two ends of the strip gets higher as
increases in the reservoir the hydrostatic the initial slurry height is increased. On the
pressure on the doctor blade region will be other hand the more the height in the reservoir,
increased. Consequently, the tape thickness the longer the strip which will be produced in
increases when it is conveyed out of the doctor the process.
blade region. The further increase in the As earlier mentioned the increased level of
155

numerical result for as compared material in the reservoir forces the ceramic to
to the experiment may be caused by the flow in the -direction (and – , side flow, see
Page

neglecting of the side effect flow which can Figure 3) which is neglected in the numerical

© 2012, Advanced Engineering Solutions, Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
modelling in both sides of the tape. Increasing increasing the material load (the height of the
the slurry height will increase this side flow slurry in the reservoir) the aforementioned
and decrease the tape thickness. However, differences between the beginning and the end
increasing the slurry height will also increase of strip and the length of the strip will be
the tape thickness due to the increased increased. The results show that the model has
pressure. These two competing phenomena a reasonable agreement with corresponding
obviously work against each other, but the experimental data. The model contains the
latter will be most predominant. major parameters which influence the process
and it has the flexibility to be used for different
slurries and different machine designs.

Acknowledgments
The authors would like to acknowledge the
support of the Scientific Research Council for
Technology and Production Sciences (FTP)
(Contract No. 09-072888, OPTIMAC), which
is part of the Danish Council for Independent
Research (DFF).
Our thanks are also directed to Miss R.
Bulatova and Dr. Christian R. H. Bahl for
Figure 7. Oscillations in the tape interface in providing us with experimental data.
the casting direction with the slurry height of
(a) 7074 and (b) 21222 . References
[1] Annika Kristoffersson and Elis Carlstrom
Conclusions (1997) Tape Casting of Alumina in Water
Numerical modelling of the non-Newtonian with an Acrylic Latex Binder, J. Eur. Cer.
slurry of La0.85Sr0.15MnO3 (LSM) was Soc., 17, 289-97.
conducted by ANSYS Fluent, and the effect of [2] J. Gurauskis, A.J. Sanchez-Herencia, and
substrate velocity and slurry height in the C. Baudin (2006) Al2O3/Y-TZP and Y-
reservoir on the final tape thickness was TZP materials fabricated by stacking layers
evaluated. The results show that increasing the obtained by aqueous tape casting, J. Eur.
substrate velocity (casting speed) causes a Cer. Soc., 26, 1489–96.
decrease in the tape thickness due to the [3] Alfred I. Y. Tok, Freddy Y.C. Boey, and K.
dominance of drag force over hydrostatic A. Khor (1999) Tape casting of high
pressure. On the other hand, an increased value dielectric ceramic composite substrates for
of the slurry height in the reservoir, results in microelectronics application, J. Mater.
increasing the final tape thickness. However, it Proc. Tech., 89-90, 508-12.
156

is observed that the height of the tape in the [4] Matthe John M. Krane, Kevin P. Trumble,
end point of the strip is higher as compared to Keith J. Bowman Hyun Jun Kim (2006)
Page

the beginning of the strip. Moreover, by Analytical Fluid Flow Models for Tape

© 2012, Advanced Engineering Solutions , Ottawa, Canada.

All rights are reserved.
 M. Jabbari, J. H. Hattel / Proceedings, CAN’2012 Tenth AES-ATEMA International Conference,
Montreal, Canada, June 18 - 22, 2012, pp. 151 – 157.
Casting, J. Am. Ceram. Soc., 89, 2769-75. method to steady pipe flows of Bingham,
[5] T. Chartier, M.deF. Granja, F. Doreau, J.M. Casson and Herschel-Bulkley fluids, J.
Ferreira, J.F. Baumard C. Pagnoux (1998) Nonnewton. Fluid Mech., 128, 126-43.
Aqueous suspensions for tape-casting [14] Georgios C. Florides, Andreas N.
based on acrylic binders, J. Eur. Cer. Soc., Alexandro, and Georgios C. Georgiou
18, 241–47. (2007) Flow development in compression
[6] X. Y. Huang, C. Y. Liu, and H. Q. Gong of a finite amount of a Bingham plastic, J.
(1997) A Viscoplastic Flow Modeling of Nonnewton. Fluid Mech., 143, 38-47.
Ceramic Tape Casting, Mater. Man. Proc., [15] Eleni Taliadorou, Georgios C. Georgiou,
12, 935-43. and Irene Moulitsas (2009) Weakly
[7] Sunil C. Joshi, Y. C. Lam, F. Y. C. Boey, compressible Poiseuille flows of a
and A. I. Y. Tok (2002) Power law fluids Herschel-Bulkley fluid, J. Nonnewton.
and Bingham plastics flow models for Fluid Mech., 158, 162-69.
ceramic tape casting, J. Mater. Proc. Tech., [16] Alfred I. Y. Tok, Freddy Y. C. Boey, and
120, 215-25. Y. C. Lam (2000) Non-Newtonian fluid
[8] C. R. Beverly and R. I. Tanner (1992) flow model for ceramic tape casting,
Numerical analysis of three-dimensional Mater. Sci. Eng. A, 280, 282-88.
Bingham plastic flow, J. Nonnewton. Fluid [17] T. G. Howell, D. R. Jeng, and K. J.
Mech., 42, 85-115. Dewitt (1996) Momentum and heat
[9] D. D. Atapattu, R. P. Chhabra, and P. H. T. transfer on a continuous moving surface in
Uhlherr (1995) Creeping sphere motion in a power law fluid, Int. J. . Heat. . Mass.
Herschel-Bulkley fluids: flow field and Trans., 40, 1853-61.
drag, J. Nonnewton. Fluid Mech., 59, 245- [18] Susana Gabbanelli, German Drazer, and
65. Joel Koplik (2005) Lattice Boltzmann
[10] J. Blackery and E. Mitsoulis (1997) method for non-Newtonian (power-law)
Creeping motion of a sphere in tubes filled fluids, Phys. Rev. E, 72, 046312.
with a Bingham plastic material, J. [19] W. Kelly and B. Gigas (2003) Using
Nonnewton. Fluid Mech., 70, 59-77. CFD to predict the behavior of power law
[11] Andreas N. Alexandrou, Emmanuel Duc, fluids near axial-flow impellers operating
and Vladimir Entov (2001), Inertial, in the transitional flow regime, Chem. Eng.
viscous and yield stress effects in Bingham Sci., 58, 2141-52.
fluid filling of a 2-D cavity, J. Nonnewton. [20] Hao Tang, L. C. Wrobel, and Z. Fan
Fluid Mech., 96, 383-403. (2004) Tracking of immiscible interfaces
[12] Andreas N. Alexandrou, Philippe Le in multiple-material mixing processes,
Menn, Georgios Georgiou, and Vladimir Comput. Mater. Sci., 29, 103-18.
Entov (2003) Flow instabilities of [21] C. W. Hirt and B. D. Nichols, "Volume
Herschel-Bulkley fluids, J. Nonnewton. of Fluid (VOF) Method for the Dynamics
157

Fluid Mech., 116, 19-32. of Free Boundaries, J. Comput. Phys., 39,

[13] R. R. Huilgol and Z. You (2005) 201-25.
Page

Application of the augmented Lagrangian

© 2012, Advanced Engineering Solutions, Ottawa, Canada.

All rights are reserved.

View publication stats