You are on page 1of 7

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

Contents lists available at ScienceDirect

Chemical Engineering Research and Design


journal homepage: www.elsevier.com/locate/cherd

Mixing analysis of a Newtonian uid in a 3D planetary


pin mixer
Robin Kay Connelly a, , James Valenti-Jordan b
a
b

Department of Food Science, University of Wisconsin-Madison, 1605 Linden Dr., Madison, WI 53706, USA
Campbell Soup Company, 1 Campbell Place, Mail Stop 210, Camden, NJ 08103, USA

a b s t r a c t
The mixograph is a planetary pin mixer that has been used for decades to evaluate the mixing tolerance and large
strain rheology of hydrated our. In this work, computational uid dynamics (CFD) has been used to gain greater
understanding of the mixing action of this mixer by evaluating both local and global measures of mixing using
particle tracking. In this study, mixing of a highly viscous, Newtonian corn syrup is simulated. Segregation scale,
length of stretch and efciency are used to evaluate the mixer. It is shown that this planetary pin mixer does not
experience as much axial mixing as cross-sectional mixing over the same time span. Additionally, it is observed that
some pin positions are more efcient than others. These results are being used to compare this mixer with other
mixers used for similar purposes in the food industry.
2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Planetary pin mixer; Mixograph; Computational uid dynamics; CFD; Mixing analysis; Mixing efciency

1.

Introduction

The mixograph (National Instruments, Lincoln, NE, USA) is a


planetary pin mixer used to evaluate the mixing tolerance
of hydrated wheat our dough in the food industry. It is a
four-pin planetary pin mixer with three stationary pins, where
the moving pins travel on an epitrochoid path. Two separate
groups have looked at the motion of the mixograph independent of the response curve (Buchholz, 1990; Steele et al.,
1990). In this study, the mixing prole of the mixograph-style
reomixer (Reomix Instruments, Lund, Sweden) is looked at
using a Newtonian corn syrup. The reomixer is used as an
alternative to the 10 g of our mixograph and differs from
the mixograph only in the specic spacing of the pins and
walls.
The results provided by this instrument consist of the reaction torque on the base of the mixing bowl caused by the forces
that have been applied to the stationary pins and walls of the
bowl by the dough as it is moved by the planetary pins. These
results are used to determine the point of peak development
and the mixing tolerance of the dough, which is related to
the development of the gluten protein structure during hydration and mixing. A more fundamental understanding of the

results is possible when types of forces acting on the dough


within the mixing chamber are known. One tool that is helpful in resolving these forces is computational uid dynamics
(CFD). Previous studies have looked at a 2D cross-section of
the mixograph (Jongen, 2000; Jongen et al., 2003). This study
continues into 3D where the effect of the angled bottom of
the bowl is incorporated and the full 3D mixing prole is
analysed.
CFD mixing studies with other mixing geometries have covered topics such as comparison of mixers through numerical
simulation (Rauline et al., 1998, 2000; Jongen, 2000; Jongen et
al., 2003; Connelly and Kokini, 2007a,b), assessment of mixing effectiveness (Tanguy et al., 1992, 1996, 1997; Bertrand et
al., 1994; Wong and Manas-Zloczower, 1994; Yang and ManasZloczower, 1994; Yang et al., 1994; Connelly and Kokini, 2004,
2006) and design and optimization of improved mixing elements (Liu et al., 2006). During these studies, various measures
of mixing have been used or developed to help assess how
well the mixing has taken place. In this study, we will be making use of some of these mixing assessment tools (Connelly
and Kokini, 2007b), which are available in the nite element
method (FEM) CFD program Polyow (Fluent Inc., Lebanon, NH,
USA).

Corresponding author. Tel.: +1 608 262 8033; fax: +1 608 262 6872.
E-mail address: rkconnelly@wisc.edu (R.K. Connelly).
Received 20 August 2008; Accepted 27 August 2008
0263-8762/$ see front matter 2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2008.08.023

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

Nomenclature
c
cj , cj
D
D
e
F
j
Ls

m
M
N
r
R(|r|)
S2
t
v
x
X

average concentration
concentration of the points in the jth pair
dimension
rate of strain tensor
efciency of mixing
deformation gradient tensor
index variable
scale of segregation
current orientation unit vector
number of pairs of material points
number of material points
distance between a pair of points
Eulerian coefcient of correlation
sample variance
time
velocity vector
position of a uid element at time, t
initial position of a uid element

Greek letters

distance beyond which there is no correlation

length of stretch

arithmetic mean of the length of stretch

motion as function of time and initial position

2.

Model description

2.1.
Geometry, pin motion, mesh and boundary
conditions
The mixer geometry is modelled as lled to 50 ml and the
meshes are shown in Fig. 1. The mesh elements are triangular
prisms. In the standard reference frame, the darker pins are
the moving pins, which follow an epitrochoid (planetary) pin
path, while the lighter pins are the stationary pins. The pin
motion is accounted for in the simulations using the mesh
superposition technique (Avalosse, 1996; Avalosse and Rubin,
2000). In order to simplify the pin paths so that the mesh could
be created to provide a consistent pin shape and volume at
every time step with this approach, the simulation is con-

ducted in the rotating reference frame (RRF). The wall of the


vessel and the stationary pins were set to rotate counterclockwise around the central vertical axis at the rate of the
primary gear (88 rpm), while the moving pins were set to
rotate clockwise around their vertical planetary axis at the
rate of the secondary gear (66 rpm). This approach also allows
the true epitrochoid (planetary) pin path to be modelled more
easily and the mesh to be better rened in the areas near the
pins.
The 2D horizontal cross-section mesh selected as a starting point for the 3D mesh was created such that the pin shape
remained a nearly constant hexagon made up of six triangular elements with a time step of 0.0227 s that moved the
pins one element along the pin path. The 2D mesh was also
subjected to a mesh discretization and time step study using
quadratic FEM interpolation for velocity and linear continuous
FEM interpolation for pressure, where it was found that the
difference in the solution was only 23% along the pin paths
when the mesh density was increased 16 times and only about
0.14% if a four times smaller time step is used. Therefore, this
mesh was deemed adequate for use in this study. Also, analysis of the difference between the initial time steps on start-up
of the simulation and the results when those pin positions
were repeated showed that the results from the rst 0.1 s of
the simulation should not be used for steady-state analysis,
due to start-up effects (Jordan, 2006).
Globe Corn Syrup 011420 (Corn Products, Westchester,
IL), was selected as the test uid because the single
proportionality constant of Newtonian viscosity was less computationally difcult to simulate. Also, its viscosity lies in the
range of observed viscosities exhibited by developing dough
(Dhanasekharan et al., 1999). This corn syrup has a Newtonian viscosity of 5400 cP and a density of 1.409 g/cm3 at 49 C.
A no slip boundary condition was used at the walls and on the
pins. The top surface was given a full slip boundary condition
but was otherwise constrained to maintain a at shape.
The ow and mixing simulation was done using the FEM
CFD package Polyow 3.10.2 (Fluent Inc., Lebanon, NH) with
a mesh created using Gambit 2.3 and post-processed using
Fieldview 10F (Intelligent Light, Rutherford, NJ). The simulation was run on an IBM IntelliStation Z Pro with dual Intel
Xeon 3.4 GHz64 bit processors and 6 GB RAM running Red
Hat Enterprise Linux WS 3. The 3D simulation was done using
mini-element velocity and linear pressure FEM interpolations.
With these FEM interpolation types, the problem size was
1.4 GB and took a CPU time of 12.68 h.

2.2.

Fig. 1 Geometry and mesh where the pins with the light
grey tops are stationary in the standard reference frame.

1435

Particle tracking and visualizations

In order to analyse the ability of the reomixer to fully distribute


the uid over the entire domain, the fate 10,000 initially randomly distributed material points were studied over 360 time
steps for a total simulation time of 8.182 s. In order to track
these points over time, Polyow and its mixing post-processor,
Polystat, were used to track the positions of the 10,000 material
points over all the time steps using the ow prole solutions.
The trajectories of the massless material points or particles
were calculated by the time integration of the equation x = v
using a fourth order explicit Runge-Kutta scheme within an
element with local rather than global coordinates. The time
step was sized such that the nal position in crossing an element is always on the element boundary so that the element
coordinates may be transformed to the local coordinates of the
next element to be crossed before continuing the integration.

1436

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

Fig. 2 Horizontal mixing analysis: data at (a) 0, (b) 120, (c) 240, and (d) 360 time steps.
The sum of the steps when transformed to global coordinates
gives the successive positions of the particles in real space
(Debbaut et al., 1997; Ishikawa et al., 2001). The default of
an average of three steps to cross an element was chosen
for the time integration, with the particle positions and kinematic parameters recorded at the same time intervals as used
between pin positions. The results were visually inspected to
illustrate the ow prole within the mixer, as well as quantitatively analysed to provide standard mixing indicators over
the entire ow eld.
One analysis studied the mixing in the horizontal plane.
The randomly distributed material points were initially
divided along the x = 0 plane and arbitrarily assigned a concentration value of 1 (light grey) or 0 (dark grey). A pictorial
representation of this division is shown in Fig. 2a. Completion of mixing is visually indicated by a random distribution
of light and dark grey points over the entire ow domain. In
a second analysis, several cutting planes were used to alternate between zones of randomly distributed points initially
assigned a concentration value of 1 (light grey) or 0 (dark grey)
in order to determine the extent of vertical mixing because
low vertical mixing was expected in this geometry. The cutting planes are located at a height of 1, 3, 5, 7, 9, 11, 13, 15,
17, and 19.698 mm, which created 11 sub-regions in the ow
domain, as shown in Fig. 3a.

2.3.

scale of segregation is a statistical measure of the mean distance from a point at which it is equally probable to nd a light
grey point as a dark grey point. The positions of the particles
at any given time were used to calculate the value of the scale
of segregation, which is dened as (Dankwertz, 1952; Tadmore
and Gogos, 1979; Brodkey, 1985; Chella, 1994):


Ls =

R(|r|) d|r|,
0

Measures of mixing

The information generated from producing the visual depictions of the mixing above were also analysed statistically. The

Fig. 3 Vertical mixing analysis: results at (a) 0 and (b) 360


time steps.

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

3.

where

M
j=1

R(|r|) =

(cj c ) (cj c )
MS2

/
1/2

(D : D)

D : m
1/2

(D : D)

D ln/Dt
1/2

(D : D)

is the current orientawhere D is the rate of strain tensor, m


m
according
tion unit vector and (D:D)1/2 is the limit of D : m
to CauchySchwarzs inequality (Ottino et al., 1979, 1981). The
efciency can be thought of as the fraction of the energy dissipated locally that is used to stretch a uid element at a given
instant in a purely viscous uid (Ottino, 1989) and falls in the
range {1, 1}. A value of 1 would indicate that all the energy
dissipated was used to shorten the length of the material line,
in effect completely unmixing it, while a value of 1 indicates
that all the energy dissipated was used to stretch the material
line. The time-averaged efciency is dened as (Ottino, 1989):

e  =

1
t

t
0

Results

3.1.
Topographical mixing analysis and scale of
segregation

R(|r|) is the Eulerian coefcient of correlation between concentrations of pairs of points in the mixer separated by |r|
where R(0) = 1 for points having the same correlation and
R() = 0 at large |r| where there is no correlation. The number of
pairs is M = N(N 1)/2 where N is the number of points and S2
is the sample variance. The concentration of the points in the
jth pair is cj and cj , while c is the average concentration. The
minimum value occurs when the initially segregated particles
become randomly distributed and is a function of the number
of particles tracked and the size of the ow domain. Calculation of the scale of segregation was done at each recorded time
step in order to track the evolution of this parameter over time.
If there are dead spots or faults in the ow that create areas
of the mixer where parts of the initially segregated material
cannot reach, this parameter will not be able to reduce to the
minimum value. In addition, the segregation scale is a global
average value that cannot pinpoint the exact location, size or
number of local ow defects (Dankwertz, 1952; Tadmore and
Gogos, 1979; Brodkey, 1985; Chella, 1994).
The 10,000 points were also used to follow the lamellar
mixing parameters, including length of stretch and efciency
(Ottino et al., 1979, 1981; Ottino and Chella, 1983; Ottino, 1989).
Given a motion x =  (X, t) where initially X =  (X, 0) for an
innitesimal material line segment dx = FdX located at position x at time t and the deformation tensor is F = , the length
of stretch of a material line is dened as  = |dx|/|dX|. The
has been shown
arithmetic mean of the length of stretch, ,
to be directly related to the geometric mean striation thickness and is a measure of the growth of the interfacial area
(Alvarez et al., 1998; Muzzio et al., 2000; Zalc et al., 2002). An
exponential increase in the length of stretch over time is a
necessary requirement for effective mixing (Ottino, 1989). The
local or instantaneous efciency of mixing for isochoric ows
is dened as
e =

1437

Fig. 2a shows the initial conditions of the 10,000 data points


discussed above. Fig. 2b and c shows the progression of the
mixing as bulk rotational ow has started in the mixer and
areas where the pins have travelled are showing signicantly
more blending of the differently shaded points. In Fig. 2d, the
mixing has approached completion at 360 time steps, based on
optical verication. The main effect holding back the completion of the mixing is that the points at the walls have not been
displaced much. This lack of movement is due to uid boundary layers near the wall where there is a no slip boundary
condition, and has been seen experimentally in similar mixers (Gouillart et al., 2007; Thiffeault et al., 2008). This has been
shown to reduce the overall mixing effectiveness for similar
mixers from exponential to power law (Gouillart et al., 2007;
Thiffeault et al., 2008).
When the mixing is analysed using the segregation scale,
the results in Fig. 4 show that most of the mixing takes
place early in the simulation. When a single dark grey particle invades a large volume of light grey particles, the average
distance for the light grey points to the nearest dark grey
point drops drastically. Therefore, the initial large drop in
segregation scale is expected and observed in Fig. 3. Overall, the segregation scale reduces to a nearly constant value
of 0.8 mm by 270 time steps, which is near the minimum
possible with the level of discrimination available with 10,000
points.
Next, the vertical mixing capability of the reomixer is analysed. The 11 zones of alternating colour at time zero are
shown in Fig. 3a. As mixing progresses, only minimal mixing
is observed. In Fig. 3b, the points are just beginning to show
blending after 360 time steps. These results show that there is
relatively little vertical mixing in comparison to the amount
of horizontal mixing in the reomixer occurs during the 8.182 s
simulated here. However, the slow vertical mixing seen in the
simulation should still provide signicant vertical distributive
mixing in the normal run time of this mixer, which is on the
order to 10 min.
The initial segregation scale value for the axial mixing case
shown in Fig. 5 was much lower than initial value for the horizontal mixing analysis due to the effect of the initial multiple
layers. However, the initial slope was much lower, so that the

e dt.

Typical behaviour of the time averaged mixing efciency


ranges from the decay of the efciency with time as t1 for
ows with no reorientation such as the simple shearing ows,
to ows with some periodic reorientation but still decaying
on average with time as t1 , to the optimum case for mixing,
which is ows with strong reorientation that have a positive,
constant average value of the efciency (Ottino, 1989).

Fig. 4 Segregation scale of vertically divided particles for


analysis of horizontal mixing.

1438

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

Fig. 5 Segregation scale of horizontally divided particles


for analysis of axial mixing.

Fig. 6 Log of the length of stretch.

value after 360 time steps was higher than that of the horizontal mixing analysis case. There was also little indication of
levelling off to a constant value within the timeframe of the
simulation.
Fig. 5 also shows that the segregation scale data was mildly
oscillating along its decline. The period of the oscillations was
every 20 time steps, which was the same frequency that the
mixer takes to produce a geometrically similar orientation.
This indicates some periodic reorientation or un-mixing of
the material being mixed.
The oscillations were also present in the results shown in
Fig. 4, but are much easier to see with this narrow segregation scale range. These results were not unexpected since the
straight vertical pins did not provide any vertical pumping
and the simplication of a xed, at shape for the top surface used in the simulation further constrains vertical motion
of the uid.

3.2.

Fig. 7 Length of stretch distribution at 360 time steps.


some points that did not experience much stretch, especially
along the walls as shown in Fig. 7.
The oscillation observed in Fig. 6 implies that the planetary nature of the path of the pins impacts the progression of
stretch. This oscillation repeats every 20 time steps with the
maximum increase occurring approximately halfway between
the repetition at 10, 30, 50, etc. time steps. The pin conguration and velocity vectors at one of these odd multiples of ten
are shown in Fig. 8.
This pin conguration was referred to as the efcient
conguration because it causes the highest increase in the
amount of stretch. When at the efcient conguration, the
stretch increased the quickest because the manner by which
the moving pins approach and straddle a stationary pin created a high elongational situation in the ow as the moving
pin pulls away from the stationary pin.
The instantaneous efciency is a powerful tool because it
tells an investigator where/what time step to look for the most
efcient mixing. In Fig. 9, the instantaneous efciency mean
and standard deviation data were shown to repeat for each set
of 20 time steps. The most efcient time step was found precisely in the middle of the bounds of the 20 time steps. This

Length of stretch and efciency

The purpose of tracking the length of stretch is to understand how much the material is being stretched in a particular
region. The length of stretch is a measure of how much a material point, conceptualized as an innitesimal line, is stretched
over time. The length of stretch was tracked over time and
recorded in Fig. 6 as the logarithm of the value of the stretch.
It showed good mixing represented by a logarithmic increase
in the length of stretch (Avalosse and Rubin, 2000). The standard deviation also increased with time because there were

Fig. 8 Pin position and velocity vectors at an efcient


conguration (130 time steps).

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

Fig. 9 Mean instantaneous efciency.


result agrees with the conclusion based on the analysis of the
length of stretch results. The pin position in Figs. 8 and 10b was
the most efcient time step with respect to instantaneous efciency, while the least efcient position was where the pin and
wall were closest as shown in Fig. 10a. The standard deviation
of the data in Fig. 9 indicates a broad spectrum of data, which
means that the efciency values for some material points may
be much higher.
In Fig. 10, the areas of high instantaneous efciency,
denoted in red, were shown to be located near the stationary
pins and in areas of high moving pin velocity in the starting
pin orientation. High efciency was noted around the stationary pins because the inertia in the uid was causing the
stationary pins to incite mixing by being a disturbance in the

Fig. 11 Mean time averaged efciency.

translational ow pattern. Fig. 8 supports this claim as a pictorial representation of the velocities at this pin position, which
shows the bulk of the uid moving with some inertial velocity and high gradients in some areas of high efciency seen in
Fig. 10b.
It is important to note the presence of low values as well in
Fig. 10, as these contribute to the standard deviation in Fig. 9.
The standard deviation in Fig. 9 also reveals another trend in
the data, because the oscillation of the standard deviation was
out of phase with the instantaneous efciency. Occasionally,
the instantaneous efciency curve produces steps at 5 or
15 time steps after the repetition around 13% efciency. The
orientation of the bottom pins at ve steps after the repetition
was the same as the efcient orientation, but the value of the
efciency maxed out 10 steps after. The most efcient form
of mixing in this mixer was when passing a stationary pin
between two moving pins without having either pass inside
of the stationary pin.
The time averaged efciency is used for describing the net
efciency of a mixer as time passes. In Fig. 11, the time averaged efciency was shown to increase quickly and then level
off to a value of 10% in an oscillating fashion. This value was
signicant because it represented the net efciency of mixing
over time. Since the standard deviation falls to a level of 6%,
it can be assumed that the bulk of the uid is experiencing
mixing.

4.

Fig. 10 Instantaneous efciency at (a) 120 and (b) 130 time


steps.

1439

Conclusion

By analysing the results from the 3D simulation, information


was gathered about the intensity and quality of mixing taking place within the reomixer. This information provided new
insight into the mechanics of how the reomixer, and subsequently all planetary pin mixers, achieve mixing.
In summary, the results show that most of the mixing takes
place horizontally in the bulk of the uid and not along the
walls or near the oor. Vertical mixing does take place slowly,
but may be signicantly limited in the simulation by the nondeforming top boundary condition. A particular orientation
of the seven pins within the mixer was shown to produce
faster/better mixing in the segregation scale, stretch, and efciency analyses. The likely cause of the effectiveness of the
mixing at this pin position is the effect of inertial velocity
taking place within the bulk of the mixer ow domain.
Future studies of this mixer will attempt to analyse the
impact of the deforming top surface on the vertical ow. In
addition, rheological models for more complex uids will be

1440

chemical engineering research and design 8 6 ( 2 0 0 8 ) 14341440

used in the future to observe the effects on the mixing of properties such as shear thinning or yield stress that are common
in food materials such as dough.
Synergy exists between this work and that of Connelly and
Kokini (2006) on the twin sigma blade farinograph, another
common our mixing tolerance evaluation instrument. When
that work is combined with this, as well as experimental and
future research on dough mixing, the nature of the effect of
mixing on dough development may be unlocked.

References
Alvarez, M.M., Muzzio, F.J., Cerbelli, S., Adrover, A. and Giona, M.,
1998, Self-similar spatiotemporal structure of intermaterial
boundaries in chaotic ows. Phys Rev Lett, 81(16): 33953398.
Avalosse, T., 1996, Numerical simulation of distributive mixing in
3-D ows. Macromol Symp, 12: 9198.
Avalosse, T. and Rubin, Y., 2000, Analysis of mixing in corotating
twin screw extruders through numerical simulation. Int
Polym Proc, XV(2): 117123.
Bertrand, F., Thibault, F., Tanguy, P.A. and Choplin, L., 1994, 3D
Modelling of the mixing of highly viscous polymers with
intermeshing impellers, in AIChE Symposium Series no. 299,
Process Mixing -Chemical and Biochemical Applications,
Tatterson, G. and Calabrese, R., Calabrese, R. (eds) , pp.
106116.
Brodkey, R.S., 1985, Fundamentals of turbulent mixing and
kinetics, In Ulbrecht, J.J. and Patterson, G.K., Patterson, G.K.
(Eds.), Mixing of Liquids by Mechanical Agitation (Gordon and
Breach Publishers, New York, NY), pp. 2958.
Buchholz, R.H., 1990, An epitrochoidal mixer. Math Sci, 15(1): 18.
Chella, R., 1994, Laminar mixing of miscible uids, in Mixing and
Compounding of Polymers: Theory and Practice,
Manas-Zloczower, I. and Tadmor, Z., Tadmor, Z. (eds) (Carl
Hanser Verlag, New York, NY), pp. 125.
Connelly, R.K. and Kokini, J.L., 2004, Analysis of mixing in a
model mixer using 2-D numerical simulation of differential
viscoelastic uids with particle tracking. J Non-Newton Fluid
Mech, 123: 117.
Connelly, R.K. and Kokini, J.L., 2006, Mixing simulation of a
viscous Newtonian liquid in a twin sigma blade mixer. AIChE
J, 52(10): 33833393, cover
Connelly, R.K. and Kokini, J.L., 2007a, Examination of the mixing
ability of single and twin screw mixers using 2D nite
element method simulation with particle tracking. J Food Eng,
79(3): 956969.
Connelly, R.K. and Kokini, J.L., 2007b, Analysis of mixing
processes using CFD, in Computational Fluid Dynamics in
Food Processing, Da-Wen Sun, (ed) (Taylor and Francis Group,
LLC, Boca Raton, FL), pp. 555588. Chapter 23
Dankwertz, P.V., 1952, The denition and measurement of some
characteristics of mixtures. Appl Sci Res, 3(A): 279296.
Debbaut, B., Avalosse, T., Dooley, J. and Hughes, K., 1997, On the
development of secondary motions in straight channels
induced by the second normal stress difference: experiments
and simulations. J Non-Newton Fluid Mech, 69: 255271.
Dhanasekharan, M., Huang, H. and Kokini, J.L., 1999, Comparison
of the observed rheological properties of hard wheat our
dough with the predictions of the Giesekus-Leonov, the
White-Metzner, and the Phan-Thien Tanner models. J Text
Studies, 30(5): 603623.
Gouillart, E., Kuncio, N., Dauchot, O., Dubrulle, B., Roux, S. and
Thiffeault, J.-L., 2007, Walls inhibit chaotic mixing. Phys Rev
Lett, 99: 114501.
Ishikawa, T., Kihara, S.-I. and Funatsu, K., 2001, 3-D
non-isothermal ow eld analysis and mixing performance

evaluation of kneading blocks in a co-rotating twin screw


extruders. Polym Eng Sci, 41(5): 840849.
Jongen, T.R., 2000, Characterization of batch mixers using
numerical ow simulations. AIChE J, 46(11): 21402150.
Jongen, T.R., Bruschke, M.V. and Dekker, J.G., 2003, Analysis of
dough kneaders using numerical ow simulations. Cereal
Chem, 84(4): 383389.
Jordan., J.B., 2006, CFD Simulation Development And Mixing
Analysis of a Newtonian Fluid in a 3D Mixograph Style Mixer.
MS Thesis. University of Wisconsin-Madison.
Liu, S., Hrymak, A.N. and Wood, P.E., 2006, Design modications
to SMX static mixer for improving mixing. AIChE J, 52(1):
150157.
Muzzio, F.J., Alvarez, M.M., Cerbelli, S., Giona, M. and Adrover, A.,
2000, The intermaterial area density generated by time- and
spatially-periodic 2D chaotic ows. Chem Eng Sci, 55:
14971508.
Ottino, J.M., (1989). The Kinematics of Mixing: Stretching, Chaos and
Transport. (Cambridge University Press).
Ottino, J.M. and Chella, R., 1983, Laminar mixing of polymeric
liquids: a brief review and recent theoretical developments.
Polym Eng Sci, 23: 357359.
Ottino, J.M., Ranz, W.E. and Macosko, C.W., 1979, A lamellar model
for analysis of liquidliquid mixing. Chem Eng Sci, 34: 877890.
Ottino, J.M., Ranz, W.E. and Macosko, C.W., 1981, A framework for
description of mechanical mixing of uids. AIChE J, 27(4):
565577.
Rauline, D., Tanguy, P.A., Le Blvec, J.-M. and Bousquet, J., 1998,
Numerical investigation of the performance of several static
mixers. Can J Chem Eng, 76: 527535.
Rauline, D., Le Blvec, J.-M., Bousquet, J. and Tanguy, P.A., 2000, A
comparative assessment of the performance of the kenics
and SMX static mixers. Trans IChemE, 78(A): 389396.
Steele, J.L., Brabec, D.L. and Walker, D.E., 1990, Mixograph
instrumentation for moving bowl and xed bowl comparisons
of wheat our performance, In Proceedings of the Food Processing
Automation Conference , pp. 224234.
Tadmore, Z. and Gogos, C.G., (1979). Principles of Polymer Processing
(1st ed.). (John Wiley and Sons, Hoboken).
Tanguy, P.A., Lacroix, R., Bertrand, F., Choplin, L. and Brito-de la
Fuente, E., 1992, Finite element analysis of viscous mixing
with a helical ribbon-screw impeller. AIChE J, 38: 939944.
Tanguy, P.A., Bertrand, F., Labrie, R. and Brito-de la Fuente, E.,
1996, Numerical modelling of the mixing of viscoplastic
slurries in a twin-blade planetary mixer. Chem Eng Res Des,
74: 499504.
Tanguy, P.A., Thibault, F., Brito-de la Fuente, E., Espinosa-Solares,
T. and Tecante, A., 1997, Mixing performance induced by
coaxial at blade-helical ribbon impellers rotating at different
speeds. Che Eng Sci, 52(11): 17331741.
Thiffeault, J.-L., Gouillart, E. and Dauchot, O., 2008, Speeding Up
Mixing with Moving Walls, Paper was presented at the
International Congress of Theoretical and Applied
Mathematics (Adelaide, Australia) August 2529.
Wong, T.H. and Manas-Zloczower, I., 1994, Two-dimensional
dynamic study of the distributive mixing in an internal mixer.
Int Polym Proc, IX(1): 310.
Yang, H.-H. and Manas-Zloczower, I., 1994, Analysis of mixing
performance in a VIC mixer. Int Polym Proc, IX(4): 291302.
Yang, H.-H., Wong, T.H. and Manas-Zloczower, I., 1994, Flow eld
analysis of a banbury mixer, in Mixing and Compounding of
Polymers: Theory and Practice, Manas-Zloczower, I. and
Tadmor, Z., Tadmor, Z. (eds) (Carl Hanser Verlag, New York,
NY), pp. 187223.
Zalc, J.M., Szalai, E.S., Alvarez, M.M. and Muzzio, F.J., 2002, Using
CFD to understand chaotic mixing in laminar stirred tanks.
AIChE J, 48(10): 21242134.

You might also like