Effects of Size and Density Differences On Mixing of Binary Mixtures of Particles

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Effects of size and density differences on

mixing of binary mixtures of particles


Cite as: AIP Conference Proceedings 1542, 739 (2013); https://doi.org/10.1063/1.4812037
Published Online: 18 June 2013

H. Musha, G. R. Chandratilleke, S. L. I. Chan, et al.

ARTICLES YOU MAY BE INTERESTED IN

Mixing behaviour of cohesive and non-cohesive particle mixtures in a ribbon mixer


AIP Conference Proceedings 1542, 731 (2013); https://doi.org/10.1063/1.4812035

Granular segregation in the double-cone blender: Transitions and mechanisms


Physics of Fluids 13, 578 (2001); https://doi.org/10.1063/1.1347961

Competition of mixing and segregation in rotating cylinders


Physics of Fluids 11, 1387 (1999); https://doi.org/10.1063/1.870003

AIP Conference Proceedings 1542, 739 (2013); https://doi.org/10.1063/1.4812037 1542, 739

© 2013 AIP Publishing LLC.


Effects of Size and Density Differences on Mixing of Binary
Mixtures of Particles
H. Musha1, G.R. Chandratilleke1, S.L.I. Chan1, J. Bridgwater2 and A.B. Yu1
1
Laboratory for Simulation and Modeling of Particulate System, School of Materials Science and Engineering, The
University of New South Wales, Sydney, NSW 2052, Australia; 2Department of Chemical Engineering and
Biotechnology, University of Cambridge, Pembroke Street, Cambridge CB2 3RA, UK.

Abstract. Mixing of granular materials is an important operation in many industries such as the pharmaceutical
industry. Generally, a granular material to be mixed can have particles of different densities and sizes, which makes the
mixing difficult. Several studies have addressed the issue of finding the optimum combinations of the size and density
ratios of binary particles for mixing to happen, using rotating drums as the mixers. For mixtures of particles uniform in
either the size or density, we have shown previously that mixing index of the mixtures can be represented by a unique
function of the variable, Ȗȕ3 where β (<1) and γ (<1) are the size and density ratios respectively. This concept was
demonstrated for a volume fraction, α = 0.5 of the large (or denser) particles, using a vertically shafted cylindrical
mixer. The implication of this result is that the size effects can be represented by an equivalent density difference. Here,
this concept is further investigated for mixing of binary particles having different combinations of β and γ, at different
values of α. It is shown that the variation of the mixing index can be expressed as a function of the three variables,
β, γ and α. Such a relationship will be useful in predicting the outcomes of mixing of binary particles.
Keywords: Powder mixing, coordination number, mixing index, discrete element method, size difference, density
difference.
PACS: 83.80.Fg; 83.10.Pp; 82.20.Wt; 83.50.Xa; 81.20.Ev

INTRODUCTION Here, Discrete Element Method (DEM) is used to


study the combined effect of particle size and density
Particle properties can significantly affect the ratios on mixing of binary particles in a vertically-
mixing behavior of particles [1]. According to this shafted cylindrical mixer. The aim of the study is to
study, the size and density differences are the major express the mixing state of a binary mixture as a
influencing factors among many other factors. In a function of the three variables, the size and density
vertically-shafted cylindrical bladed mixer, an increase ratios and volume fraction of the components of the
in the difference in either the size or density has led to mixture. Such a relationship would prove that mixing
an increase in the rate of segregation according to behaviour of binary particles is predictable under a
Zhou et al. [2]. They showed that the segregation is wide range of particle and physical properties.
caused by the generation of a vertical driving force on
particles due to the differences in particle size or SIMULATION METHOD
density. However, their study mainly focused on
investigating the effects of size and density differences The DEM model used here is based on one
on particle segregation as separate effects. Conversely, previously developed [5, 6], and is an extension of the
Nitin et al. showed that the mixing in rotating drums original DEM model [7] to account for the rolling
can be improved under certain combinations of size friction of particles. The model has been validated and
and density differences [3]. Recently, Chandratilleke its details can be found elsewhere [6].
et al. [4] studied the independent effects of the size and Mixture quality is quantified by a particle scale
density differences on mixing in a vertical bladed mixing index (PSMI), which has been used
mixer and found that the two effects are successfully in the quantification of size and density
interchangeable. Namely, the two effects produce effects previously [4].
similar mixing states if the particle weight ratio, γβ3 is
matched, where γ and β are the density and size ratios, Simulation Conditions and Procedure
respectively. However, it is not clear whether different
combinations of size and density ratios would produce A vertically-shafted cylindrical mixer shown in
similar mixing states even if γβ3 is matched. Fig. 1 is used in the study of mixing. The blades of the

Powders and Grains 2013


AIP Conf. Proc. 1542, 739-742 (2013); doi: 10.1063/1.4812037
© 2013 AIP Publishing LLC 978-0-7354-1166-1/$30.00

739
mixer are located diametrically opposite and their Combined Effect of Size and Density
wider surfaces are oriented vertically. The mixer is Ratios
roughly filled up to the top edge of the blades as
shown Fig. 1(b) to keep the total volume of a mixture
Figure 2 shows a comparison of the combined
constant. Particles are loaded into the mixer such that
effect of ȕ and Ȗ of the present work with the
large (and heavy) particles are on top of the small (and
individual effects of ȕ and Ȗ for the Ȗ=1 and ȕ=1
light) ones as shown Fig. 1(b). After the particles have
systems respectively, of a previous work [4]. Note that
settled down, the impeller is rotated from the
x-axis represents the weight fraction γβ3 of binary
stationary state at a constant acceleration until it
particles. The figure shows that mixing index reduces
reaches a pre-set speed, at which point the impeller
with a decrease in the weight fraction in the uniform
continues to rotate at that constant speed. The size
systems. However, in binary systems, size and density
ratio ȕ is defined as d s / d l and density ratio Ȗ as
effects interact with each other, and a maximum
ρ s / ρ l , where s and l refer to small (light) and large mixing state is observed for an optimum weight ratio
(heavy) particles respectively. as seen from the dotted-line in the figure. When the
weight fraction is either smaller or larger than this
The simulations conducted are of three types. In optimum value, the initially top-loaded particles either
the first, ȕ is kept fixed while Ȗ is changed. In the sink to bottom or remain on top of the mixture
second, ȕ is changed while Ȗ is kept fixed. In the third, respectively causing more segregation than at the
ȕ is varied at different values of Ȗ at a volume fraction optimum condition. Below the effects of ȕ and γ are
Į = 0.9. More details about the study can be found
investigated at different volume fractions α.
elsewhere [8].
Mixing index at steady-state 1.0

0.8

0.6

0.4
Density effect (binary, β fixed)
Size effect (binary, γ fixed)
0.2 Density effect (β=1)
Size effect (γ=1)

(a) (b) 0.0


0.01 0.1 1
3
FIGURE 1. Mixer details: (a) geometry, and (b) initial Weight fraction, γ β
loading.
FIGURE 2. Steady-state mixing index as a function of
weight ratio for uniform and binary systems at Į=0.5. Fixed
The particle properties used are as follows: values of β and γ are 0.5 and 0.22 respectively.
Young’s modulus=1×107 N/m2, Poisson’s ratio=0.3,
damping coefficient=0.3, static sliding friction
coefficient=0.3, density of large particles=6040 kg/m3 Size effect at a fixed density ratio Ȗ
and rolling friction coefficient=0.001d, where d is the
particle diameter of small particles if particle sizes are Previous works suggest that volume fraction Į of
different. the mixing particles play some role on the final/steady
state of mixing achieved [4, 9]. Therefore, the size
effect is investigated at several α values, keeping Ȗ at
RESULTS AND DISCUSSION 0.22. The results are shown in Fig. 3 for Į = 0.1, 0.5,
0.7 and 0.9.
It has been reported that large particles in a binary The figure shows that mixing behaviour is similar
particle mixture of different sizes but uniform density for a fixed volume fraction, with a peak value
(or similarly, light particles of binary particles of appearing at certain size ratio, which seems to vary
different density, but uniform size) are subjected to somewhat. It can be also seen that the peak value
large upward forces during the initial stages of mixing, varies also with the volume fraction: either too small
causing segregation [2]. Here, the density of the large or large Į value causes the peak to decrease. The
particles was set, therefore to 6040 kg/m3 to counter maximum mixing index is achieved at about Į =0.5.
this upward force, compared to 2500 kg/m3 of the Note that β has been varied keeping the diameter of
study of Zhou et al [2]. Below the results of the the large particle fixed at either 8 or 9 mm.
combined effect of γ and β are discussed.

740
Np ki
1.0 1 (1)
α = 0.5
fz =
Np
¦ [m g + ¦ ( F
i =1
i
j =1
c ,ij + Fd ,ij )] z
Mixing index at steady-state 0.8
Nt
α = 0.7 1
¦f (2)
t
fz = z
0.6 α = 0.9
Nt 1

0.4 α = 0.1
Here Fc,ij and Fd,ij are the contact and damping
0.2 force vectors on particle i due to contact with particle
j; mi is mass of particle i; Np is the number of particles
0.0
0.4 0.5 0.6 0.7 0.8 0.9 1
of one type of particles, g is the acceleration due to
Size Ratio, β gravity; suffix z represents the force components are
being considered in the vertical direction; and Nt is the
FIGURE 3. Effects of size ratio β and volume fraction α on number divisions in the time axis for averaging.
steady-state mixing for a density ratio of Ȗ =0.22.
Figure 5 shows the time-averaged vertical forces
on small and large particles for the size ratio ȕ=0.875
Density Effect at a Fixed Size Ratio β at a volume fraction Į=0.9.

Figure 4 shows the effect of density on steady-state 0.20

mixing at three volume fractions. At a given volume

Time Averaged Verticle Force (N)


0.15
fraction, mixing index increases with a decrease in Ȗ, 0.10
Small particles

achieved by reducing ρs while keeping ρl at 6040 0.05


kg/m3. When Ȗ=1, only the effect of size difference 0.00
exists, and large particles remain on top of the small
-0.05
particles at the bottom. On the other hand, when ρs is
-0.10
decreased (i.e. Ȗ decreases), the weight of large Large particles
-0.15
particles become dominant, and large particles would
be able to overcome the upward forces on them to -0.20
0 5 10 15 20 25 30
Time(sec)
move into small particle region at the bottom.
Therefore, mixing is improved with Ȗ reduced. To (a) Ȗ=1
illustrate this effect, the forces on large and small 0.20

particles are analyzed following the method reported


Time averaged verticle force (N)

0.15
Small particles
previously (Zhou et al, 2003). 0.10

1.0 0.05

0.00
Steady-state mixing index

0.8
α = 0.5 -0.05

-0.10
0.6
-0.15
Large particles
0.4 -0.20
α = 0.9 0 5 10 15 20 25 30 35
Time(sec)
0.2
α = 0.1 (b) Ȗ=0.22
0.0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 FIGURE 5. Illustration of the interaction of the size and
Density Ratio, γ density effects considering the vertical forces on the two
types of particles for the case of ȕ= 0.875 with Į=0.9.
FIGURE 4. Effects of density ratio γ at different volume
fractions α on steady-state mixing for a size ratio of β =0.5;
with dl fixed at 8mm. ȕ=0.875 is chosen here because mixing index is close
to the peak value than ȕ=0.5 for Į=0.9 (see Fig. 3).
The figure shows that the vertical force on small
First, following them, the instantaneous vertical
particles, for both ȕ values, are positive from the
forces averaged for each particle group, f z are found beginning of mixing and continue to be so throughout.
as a function of time using Eq.(1). Next, f z is However, ȕ=0.875 has a larger vertical force on small
averaged over time using Eq.(2) to obtain the time- particles in comparison, and thus, more of small
particles are expected to be raised to top of the mixer
average vertical force on a group of particles, f z t .
promoting mixing. The results are supported by Fig. 3,

741
where ȕ=0.875 has a better mixing state than ȕ=1 at predict the general effects of size, density and volume
Į=0.9. fraction of particles of a binary mixture.

Combined Effect of β, γ and α CONCLUSIONS

In order to illustrate the effects of α, β and Ȗ, The combined effect of differences in particle size
several methods were tested. One was to use the and density and volume fraction on the mixing
weight fraction γβ3 as a variable as used in Fig. 2. behaviour of a binary particle mixture was studied by
However, the density effect of Fig. 4 remains means of the discrete element method considering a
essentially the same, ȕ being fixed. Therefore, γβ3n is vertically-shafted bladed mixer. The following
used to account for the variation of particle number conclusions are made.
with changes in α, where n is the number ratio of the For a fixed density ratio γ, the mixing index attains
two particle types. Figure 6(a) replots the effect of a peak value at an optimum size ratio β if the volume
density shown in Fig. 4, using γβ3n as the x-axis fraction α of the mixing particles is kept fixed.
variable. It can be seen that the data can be correlated Similarly, this peak value varies with the volume
fraction, the maximum mixing index being achieved
0.8
for a volume fraction of about 0.5. The interaction of
0.7
the size and density differences could be explained
using the concept of vertical buoyancy forces on the
Steady-state mixing index

0.6
α=0.5 particles, based on a previous work.
0.5
α=0.9
It was also shown that effects of size, density and
0.4 volume fraction can be correlated by considering the
0.3 total weight ratio γβ3n.
0.2 α=0.1
0.1 ACKNOWLEDGMENTS
0.0
0.001 0.01 0.1 1 10
Weight fraction, γ β 3n The authors would like to acknowledge funding for
(a) this work by the University New South Wales, ARC
and Chinese Council Scholarship.
1.0

REFERENCES
Steady-state mixing index, M

0.8

0.6 1. L. T. Fan, Y. M. Chen and F. S. Lai, Powder Technol. 61


255 - 287 (1990).
0.4 2. Y. C. Zhou, A. B. Yu and J. Bridgwater, J. Chem.
β=0.5
Technol. Biot. 78 (2), 187-193 (2003).
β=0.6
0.2
β=0.85
3. J. Nitin, J. M. Ottino and R. M. Lueptow, Granular
Fit Matter 7 (2-3), 69-81 (2005).
0.0
0.02 0.04 0.06 0.08 0.1
4. G. R. Chandratilleke, A. B. Yu, J. Bridgwater and K.
Weight fraction, γ β 3n
Shinohara, AIChE J. 58 (4), 1099-1118 (2012).
5. Y. C. Zhou, B. D. Wright, R. Y. Yang, B. H. Xu and A.
(b) B. Yu, Physica A 269, 536-553 (1999).
FIGURE 6. Combined effects of α, β and γ: (a), effect of γ 6. Y. C. Zhou, A. B. Yu, R. L. Stewart and J. Bridgwater,
at different α; and (b), effect of γ and β at α=0.9. Chem. Eng. Sci. 59, 1343-1364 (2004).
7. P. A. Cundall and O. D. L. Strack, Geotechnique 29 (1),
47-65 (1979).
by a single curve. Figure 6(b) shows the effect of 8. H. Musha, G. R. Chandratilleke, A. B. Yu, S. L. I. Chan
density at different β for α=0.9. The figure shows that and J. Bridgwater, (To be published).
the trend of the data points can be described by a 9. M. Alonso, M. Satoh and K. Miyanami, Powder Technol.
single curve of the following form, with a correlation 68 (2), 145-152 (1991).
coefficient of about 0.84.
M = c1 − c 2 log(γβ 3 n ) (3)

The fitting curve of Fig. 6(a) also has the same form as
Eq.(4). Therefore, the above equation should be able to

742

You might also like