Anda di halaman 1dari 13

Three-dimensional modelling of pneumatic drying process

M. Mezhericher
a,b
, A. Levy
a,
, I. Borde
a
a
Pearlstone Center for Aeronautical Engineering Studies, Department of Mechanical Engineering, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
b
Department of Mechanical Engineering, Sami Shamoon College of Engineering, Bialik/Basel Sts., Beer-Sheva 84100, Israel
a b s t r a c t a r t i c l e i n f o
Article history:
Received 11 September 2009
Received in revised form 16 May 2010
Accepted 27 May 2010
Available online 4 June 2010
Keywords:
CFD
Drying kinetics
Heat and mass transfer
Pneumatic drying
Steady-state three-dimensional calculations of heat and mass transfer in vertical pneumatic dryer were
performed. The theoretical model of the drying process is based on two-phase EulerianLagrangian approach
for gas-particles ow and incorporates advanced drying kinetics for wet particles. The model was utilized for
simulation of the drying process of wet PVC and silica particles in a large-scale vertical pneumatic dryer. The
inuence of wall thermal boundary conditions was investigated by assuming either known value of the wall
temperature or adiabatic ow in the dryer. Analyzing the predicted particle drying kinetics, an uneven
product quality was predicted due to non-uniform drying conditions in the central and peripheral zones of
the pneumatic dryer. Moreover, for the case of non-insulated chamber walls such quality unevenness was
estimated to be substantially greater than for the case with thermally insulated drying chamber. The
examination of the predicted temperature proles within the silica and PVC wet particles showed that the
latter is subjected to higher temperature gradients potentially resulting in the greater rate of thermally-
degraded nal product.
2010 Elsevier B.V. All rights reserved.
1. Introduction
Pneumatic (ash) drying is one of the widely used technological
processes in food, chemical, agricultural and pharmaceutical indus-
tries. Despite its apparent simplicity, the process of pneumatic drying
is a complex multi-phase transport phenomenon involving turbu-
lence ow of humid compressed gas and multicomponent wet
particles, heat and mass transfer between the wet particles and
drying gas, drying kinetics of the wet particles and thermal and
mechanical stresses in the dried particulate material.
Due to growing demand for the pneumatic dryer units and tough
design requirements regarding their efciency and low resource
consumption, extensive experimental and theoretical studies of the
pneumatic drying process have been performed during the last years
[110]. Two-uid theory is a typical approach used to model the
pneumatic drying. The two-uid model is based on EulerianEulerian
formulation which considers both the drying gas and the wet particles
as two pseudo continuous phases occupying each point of the
computational domain with their own volume fractions. Another
way to model the pneumatic drying is utilizing an Eulerian
Lagrangian approach like Discrete Element Modelling (DEM) [11,12]
or Discrete Phase Model (DPM) [1316]. The Discrete Phase Model is a
kind of Discrete Particle Model [1722] and it is suitable for systems
with large amounts of particles owing to the usage of the concept of
parcels: each parcel contains a number of identical discrete particles
with the same parameters simultaneously injected into the domain. In
contrast to the EulerianEulerian formulation, the EulerianLagrang-
ian models allowparticle trajectories tracking by treating the particles
as situated in discrete points of the domain whilst the drying gas is
assumed to be a continuous phase.
The aim of this study was to develop a steady-state three-
dimensional theoretical model of the pneumatic drying process using
Discrete Phase Model and Computational Fluid Dynamics technique.
The drying kinetics of wet particles is described with the help of an
advanced theoretical model validated for single wet particle drying
[22,23]. For comprehensive literature survey on drying of droplets and
wet particles the reader is referred to the paper [24].
2. Problem setup
For the purposes of the theoretical study the geometry of Baeyens
et al. [1] experimental setup is adopted. Hot dry air and wet particles
are supplied to the bottom of vertical pneumatic dryer with 1.25 m
internal diameter and 25 m height (see Fig. 1).
The following assumptions are used in the study:
steady-state drying process is considered;
gas-particle ow is dilute;
mass, momentum and heat transfer between particles themselves
are negligible;
continuous phase is an ideal mixture of vapour and dry gas;
Powder Technology 203 (2010) 371383
Corresponding author.
E-mail addresses: mezher@bgu.ac.il (M. Mezhericher), avi@bgu.ac.il (A. Levy),
borde@bgu.ac.il (I. Borde).
0032-5910/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2010.05.032
Contents lists available at ScienceDirect
Powder Technology
j our nal homepage: www. el sevi er. com/ l ocat e/ powt ec
continuous phase behaves like an ideal gas;
dryer walls are either thermally insulated or wall temperature
distribution is known.
3. Theoretical model
The continuous phase (drying air) is assumed to be an ideal gas
and it is treated by an Eulerian approach using the k- model for
turbulence description. The discrete phase of spherical particles is
considered using DPM Lagrangian formulation.
3.1. Continuous phase
For the continuous phase of drying air, three-dimensional steady-
state conservation equations of continuity, momentum, energy,
turbulent kinetic energy, dissipation rate of turbulence kinetic energy
and species are applied [25]:
continuity

t
+

x
j
u
j
= S
c
: 1
Here the mass source termS
c
= n
p
dm
p
dt
, where n
p
and m
p
are num-
ber density andsingle particle mass of the discrete phase respectively.
momentum

t
u
i
+

x
j
u
j
u
i
=
p
x
i
+

x
j

e
u
i
x
j
+
u
j
x
i
_ _ _ _
+ g
i
+ U
pi
S
c
+ F
gp
2
energy conservation

t
h +

x
j
u
j
h =

x
j

h
h
x
j
_ _
q
r
+ S
h
; 3
where the energy source term S
h
=hS
c
.
species conservation

t
Y
v
+

x
j
u
j
Y
v
=

x
j

Y
Y
v
x
j
_ _
+ S
c
: 4
turbulence kinetic energy

t
k +

x
j
u
j
k
_ _
=

x
j

k
k
x
j
_ _
+ G
k
+ G
b
5
dissipation rate of turbulence kinetic energy

t
+

x
j
u
j

_ _
=

x
j

x
j
_ _
+

k
C
1
G
k
C
2

_ _
6
The production of turbulence kinetic energy due to mean velocity
gradients is equal to:
G
k
=
T
u
i
x
j
+
u
j
x
i
_ _
u
i
x
j
: 7
The production of turbulence kinetic energy due to buoyancy is
given by:
G
b
=g
j

T
T
x
j
; 8
where is the coefcient of thermal expansion:
=
1

T
_ _
p
; 9
Here the model constants are equal to: C
1
=1.44, C
2
=1.92,

k
=
h
=
Y
=
T
=0.9 and

=1.3. The effective viscosity,


e
, is
calculated by:

e
= +
T
; 10
where
T
is turbulent viscosity

T
= C

k
2

: 11
In the above expression C

=0.09.
The relationship between humid air temperature, pressure and
density is given by the ideal gas law:
p =

M
T: 12
3.2. Discrete phase
The motion equation of the discrete phase of wet particles is as
follows:
d

U
p
dt
=

g +

F
p
m
p
: 13
Fig. 1. Schematic illustration of vertical pneumatic dryer [1].
372 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
Here F

p
is the sumof the forces acting on the considered wet particle
from the gas phase and from other droplets/particles and walls of the
drying chamber as well. Since we deal with steady-state pneumatic
drying of non-charged micron-sized spherical particles with inlet
velocity close to terminal one, we can neglect all the forces arising
from eld gradients in the gas phase as well as from particle rotation,
acceleration and electric charging. Therefore:

F
p
=

F
D
+

F
A
+

F
B
+

F
C
; 14
where F

D
is the drag force, F

A
is the virtual (added) mass force, F

B
is the
buoyancy force and F

C
is the contact force.
The drag force is determined by [26]:

F
D
=

8
C
D
j

U
p
j

u

U
p
_ _
d
2
p
: 15
The drag coefcient, C
D
, is evaluated according to empirical
correlations [25].
The components of added mass (virtual mass) force, required to
accelerate the gas surrounding the droplet/particle, are determined as
follows [27]:

F
A
=
d
3
p
12
D

u
Dt

D

U
p
Dt
_
_
_
_
: 16
The buoyancy force is given by:

F
B
=
d
3
p
6

g: 17
The contact force due to collisions of a droplet/particle with the
walls of the drying chamber, F

C
, is calculated in terms of the
corresponding normal and tangential coefcients of restitution [15].
3.3. Drying kinetics model
The internal transport phenomena within the dried particles are
described with the help of original two-stage drying kinetics model
[23], which was successfully validated by comparison with the
published experimental and theoretical data in the case of single
droplet/wet particle drying in still air. The adopted drying kinetics
model is briey described below.
The drying process of wet particle containing solids is divided in
two drying stages. In the rst stage of drying, an excess of moisture
forms a liquid envelope around the particle solid fraction, and
unhindered drying similar to pure liquid droplet evaporation results
in the shrinkage of the outer diameter. At a certain time, the moisture
excess is completely evaporated and the second stage of a hindered
drying begins. In this second drying stage, two regions of wet particle
can be identied: layer of dry porous crust and internal wet core. The
drying process is controlled by the rate of moisture diffusion from the
particle wet core through the crust pores towards the particle outer
surface. As a result of the hindered drying, the particle wet core
shrinks and the thickness of the crust region increases. The particle
outer diameter is assumed to remain unchanged during the second
drying stage. After the point when the particle moisture content
decreases to a minimal possible value (determined either as an
equilibriummoisture content or as a bounded moisture that cannot be
removed by convective drying), the particle is treated as a dry non-
evaporating solid sphere. The concept of two-stage drying kinetics is
illustrated by Fig. 2.
To conserve the space, only basic equations of the two-stage
drying kinetics model are presented in the present report. For more
detailed description and validation of the drying kinetics model, see
Mezhericher et al. [22,23].
3.3.1. First drying stage
In the rst drying stage, the temperature of wet particle is assumed
to be uniformly distributed. The corresponding equation of energy
conservation is given by:
h
fg
m
v
+ c
p;d
m
d
dT
d
dt
= h T
g
T
d
_ _
4R
2
d
: 18
The time-change of wet particle outer radius is determined by [28]:
dR
d
dt
=
1

d;w
4 R
d

2
m
v
: 19
The mass transfer rate from the particle surface is determined
according to the mass convection law:

m
v
= h
D

v;s

v;
_ _
4 R
d

2
: 20
The coefcients of heat and mass transfer are calculated in terms of
the corresponding Nusselt and Sherwood numbers that are given by
following modied RanzMarshall correlations:
Nu
d
= 2 + 0:6Re
1 = 2
d
Pr
1=3
_ _
1 + B
0:7
; 21
Sh
d
= 2 + 0:6Re
1 = 2
d
Sc
1=3
_ _
1 + B
0:7
: 22
The particle mass is found by integration of Eq. (19):
m
d
= m
d;0

4
3

d;w
R
d;0
_ _
3
R
d

3
_ _
: 23
Finally, the value of particle moisture content on dry basis, X
d
, is
equal to:
X
d
= m
d;w
= m
d;s
= m
d
1 + X
d;0
_ _
= m
d;0
1: 24
The transition between the rst and second drying stages occurs in
the moment when the particle moisture content reduces to a critical
value. In the present study, this critical moisture content is evaluated
using the condition of formation of nal particle without internal void
[15]:
X
cr
=

w

1
: 25
Fig. 2. Two-stage drying kinetics of wet particle. D
0
initial particle diameter, D
p
nal
particle diameter, D
i
wet core diameter.
373 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
3.3.2. Second drying stage
In the second drying stage, the wet particle is considered as a
sphere with isotropic physical properties and temperature-indepen-
dent crust thermal conductivity. The crust region is assumed to be
pierced by a large number of identical straight cylindrical capillaries,
and the wet core region is considered to be a sphere with liquid and
solid fractions. The equations of energy conservation for the wet core
and crust regions are as follows:

wc
c
p;wc
T
wc
t
=
1
r
2

r
k
wc
r
2
T
wc
r
_ _
; 0 r R
i
t ; 26
T
cr
t
=

cr
r
2

r
r
2
T
cr
r
_ _
; R
i
t r R
p
: 27
Here R
i
is the internal radius of the interface between the crust and
wet core and R
p
is outer particle radius.
The boundary conditions for the above set of equations are:
T
wc
r
= 0; r = 0;
T
wc
= T
cr
; r = R
i
t ;
k
cr
T
cr
r
= k
wc
T
wc
r
+ h
fg
mv
A
i
; r = R
i
t ;
h T
g
T
cr
_ _
= k
cr
T
cr
r
; r = R
p
;
_

_
28
The crustwet core interface receding rate is tracked by the
following relationship:
d R
i

dt
=
1

wc;w
4R
2
i
m
v
: 29
The total mass transfer rate through the crust pores is the sum of
corresponding diffusion and forced mass ow rates:
m
v
= m
v;diff
+ m
v;flow
= h
D

v;s

v;
_ _
A
p
: 30
The mass ow rate of vapour diffusion is dened by:
m
v;diff
=
8

D
v;cr
M
w
R
p
R
i
T
cr;s
+ T
wc;s
_ _
p
m
p
v

p
v
p
m
r
p
m
p
v
r
_ _
31
The mass ow rate of forced vapour owis determined as follows:
m
v;flow
=
B
k

m
8R
p
R
i
M
m
p
m
T
cr;s
+ T
wc;s
_ _
p
m
r
: 32
The permeability B
k
is calculated according to well-known
CarmanKozeny equation:
B
k
= d
2
p

3
= 180 1
2
_ _
: 33
Similarly to the rst drying stage, the coefcients of heat and mass
transfer are calculated in terms of the corresponding Nusselt and
Sherwood numbers that are given by the modied RanzMarshall
correlations (see Eqs. (21) and (22)).
The particle mass is tracked using the following expression:
m
p
= m
d;0
= 1 + X
d;0
_ _
1
wc;w
=
wc;s
_ _
+ 4= 3
wc;w
R
3
i
+ 1 R
3
p
_ _
:
34
Finally, the particle moisture content is given by:
X
p
= m
w
= m
s
= m
p
1 + X
d;0
_ _
= m
d;0
1: 35
3.3.3. Final sensible heating
When the particle moisture content decreases to a minimal value
attainable under the given drying conditions, the wet particle turns
into a non-evaporating dry particle. This non-evaporating particle and
surrounding ow of the drying air continue the interaction by heat
transfer:
T
p
t
=

p
r
2

r
r
2
T
p
r
_ _
;0 r R
p
: 36
4. Numerical simulations
The developed theoretical model was numerically solved with the
help of Finite Volume Method and 3D simulations of pneumatic
drying were performed using the CFD package FLUENT 12.0.7. For
these purposes, the 3D numerical grid with 9078 distributed
hexahedral/wedge cell volumes was generated in GAMBIT 2.2.30
using the Cooper scheme (see Fig. 3).
The set of differential equations for drying air (continuous phase)
was numerically solved utilizing the 3D pressure-based steady-state
solver of FLUENT package. In particular, the pressurevelocity
coupling was realized by means of the SIMPLE algorithm [29].
Furthermore, the pressure equation was spatially discretized by the
PRESTO! scheme [29]. For all transport equations, a second-order
upwind spatial discretization method was applied and the turbulent
effects were taken into consideration by the standard k- turbulence
model.
The ow of wet particles in the pneumatic dryer was simulated
through 89 injections of spherical particles. Each injection began on
the bottom of the dryer at the centroid of one of the 89 bottom plane
mesh elements. The particle injections were normal to the dryer
bottom plane and parallel to each other.
The numerical simulations were performed in the following way.
First, the ow of drying air was simulated without the discrete phase
until converged solution. At the next step, wet particles were injected
into the domain and two-way coupled simulations were performed
until convergence.
The calculation of the discrete phase drying kinetics was
accomplished using the concept of user dened functions (UDF)
Fig. 3. Three-dimensional numerical grid.
374 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
which were attached to FLUENT. In this way, the rst drying stage and
the nal heating of non-evaporating dry particles (Eqs. (18)-(25))
were simulated using the FLUENT built-in UDFs responsible for
evaporation and sensible heating of droplets/particles. For the second
drying stage, Eqs. (26)(35) were solved numerically using an
original numerical algorithm developed by Mezhericher [15,22,23],
which is based on implicit nite differences method. The particle
temperature prole, diameter, mass, moisture content, positions of
inter-particle crustwet core interface and material characteristics
were stored in the computer memory for each particle at every
calculation time-step. This numerical procedure was implemented
and linked to FLUENT package via a set of original UDFs. The
transitions between the various drying stages for each particle were
made automatically, based on the local values of the particle moisture
Table 1
Properties of PVC and silica particles.
Name Diameter,
m
Primary
particles
diameter, m
Density,
kg/m
3
Specic
heat,
J/(kg K)
Thermal
conductivity,
W/(m K)
Thermal
diffusivity,
10
7
m
2
/s
Porosity Initial moisture
content, %
(dry basis)
Critical moisture
content (dry basis,
Eq. 25)
Final moisture
content, %
(assumed)
PVC 140 [1] 0.272 1195 [1] 980 [1] 0.14 [1] 1.195 0.3 36 [1] 0.357 5
Silica 140 0.272 [31] 1950 [31] 750 [31] 1.40 [32] 9.573 0.4 [31] 36 0.341 5
Fig. 4. 3D air ow elds for the case of silica drying (adiabatic ow). (a) velocity magnitude, m/s (b) temperature, K, (c) vapour mass fraction, kg/kg, (d) density, kg/m
3
.
375 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
content. It is worth noting that for each drying stage, the thermo-
physical properties of wet particles were tracked as dependent on
both particle temperature and moisture content.
The convergence of the numerical simulations was monitored by
means of residuals of the transport equations. Particularly, the
converged values of the scaled residuals were ensured to be lower
than 10
6
for the energy equation and 10
3
for the rest of equations.
The convergence was also veried by negligibly small values of the
global mass and energy imbalances.
Finally, the received numerical solution was tested for grid-
independency by reducing the size of original meshes. For two grids
consisting of 9078 (original) and 42,330 (reduced) cell volumes, the
maximum difference in the calculated ow elds of air velocity,
pressure, density, temperature and vapour mass fraction was
observed to be smaller than 5%.
5. Results and discussion
5.1. Three-dimensional drying simulations
The processes of pneumatic drying of PVC and silica particles
were simulated using the developed model. The particles of 140 m
in diameter were introduced from the bottom of the dryer via 89
injections with zero initial velocity. The injections were placed
normally to the bottom plane at the centroids of its grid cells. The
mass ow rates of the injections were assumed to be identical and
the overall ow rate of the discrete phase was set to 1.583 kg/s.
Particles hitting the chamber walls were reected with
corresponding normal and tangential restitution coefcients set to
0.9. The applied properties of PVC and silica particles are consoli-
dated in Table 1.
Fig. 5. 3D air ow elds for the case of PVC drying (adiabatic ow). (a) velocity magnitude, m/s (b) temperature, K, (c) vapour mass fraction, kg/kg, (d) density, kg/m
3
.
376 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
Dry atmospheric air at 400 K was supplied to the dryer bottom
with mass ux of 10.44 kg/(m
2
s). The air outlet pressure was
assumed to be a subatmospheric and equal to 25 Pa. Furthermore,
the effect of air turbulence was considered by assuming 5% inlet
turbulence intensity and 1.25 m hydraulic diameter. Chamber walls
were assumed to be made of 2 mm steel. Finally, the ambient
temperature was set to 300 K.
The inuence of wall thermal boundary conditions was investi-
gated by assuming either adiabatic ow in the dryer (insulated
chamber walls) or linear drop of the wall temperature from325 at the
inlet to 320 K at the outlet (non-insulated chamber with known wall
temperature) [4,6].
The predicted steady-state 3D ow elds of velocity, temperature,
vapour mass fraction and density of the drying air are illustrated by
Figs. 47 (all results are in SI units). The results demonstrate three-
dimensionality and complexity of the ow in the pneumatic dryer.
Inbothdryingcases thedeveloping core of the air velocityis observed
(Figs. 4a, 5a, 6a and 7a). The air temperature gradually decreases in the
core(Figs. 4b, 5b, 6band7b) while thevapour fractionincreases (Figs. 4c,
5c, 6c and 7c) as a result of heat and mass transfer from the discrete
phase of particles. An interesting behaviour is demonstrated by the air
density (Figs. 4d, 5d, 6dand7d). Initially, the air density decreases due to
fast increase of vapour fraction in the region of prevalent heat and mass
transfer fromthe wet particles. Then, whenmost of the moisture content
of wet particles was evaporated, the rate of mass transfer reduces andthe
value of vapour fraction in the air stabilizes. However, the air
temperature decrease continues due to particles heating up to
equilibrium. Such air temperature drop and constant vapour fraction
value result in the observed rise of the air density.
A valuable one-dimensional representation of actual three-
dimensional air ow elds can be obtained by mass-weighted
averaging. Using this technique, the computational domain was cut
Fig. 6. 3D air owelds for the case of silica drying (known wall temperature). (a) velocity magnitude, m/s (b) temperature, K, (c) vapour mass fraction, kg/kg, (d) density,
kg/m
3
.
377 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
along z axis in XY plane by 100 slices. For each slice, mass-weighted
averages of the ow parameters were calculated. Combining the
corresponding values for all slices, 1D representations of the owelds
were obtained. The simulated mass-weighted averages of 3Dowelds
of velocity, temperature, vapour mass fraction, density and pressure of
the drying air are shown in Fig. 8. For the case of insulated chamber, the
wall temperature was inthe range of 376394 K, whereas for the case of
non-insulated chamber the wall temperature decreased linearly from
325 to 320 Kover the dryer length. It canbe observedthat these 1Dow
elds qualitatively illustrate the 3Dbehaviour of owparameters along
z axis and this behaviour is in accordance with the conclusions drawn
above. At the same time, the usage of the 1D representations for exact
quantitative comparison with the published predictions of 1D pneu-
matic drying models [1,3,4] is inappropriate since such models do not
take into account numerous 3D space effects like turbulence and
transversal components of velocity. Moreover, the 1D averaged repre-
sentations of the 3D ow elds cannot be adequately validated by the
publishedexperimental dataobtainedusing1Dor 2Dsamplingtechnique
[1]. Duetothesereasons, thedevelopment of proper validationprocedure
for the present 3D model is a prospect work. Nevertheless, a qualitative
validation of the presented averaged ow elds (Fig. 8) demonstrates
similarity between the predicted trends of air velocity, temperature,
vapour mass fraction, density and pressure with their behaviour
calculated by Baeyens et al. [1], Levy and Borde [3,4], Skuratovsky and
Levy [5] andSkuratovsky et al. [6]. Further comparisonwithexperimental
meauserments of Baeyens et al. [1] is given in the next section.
Fig. 9 illustrates the predicted drying kinetics of two PVC particles
injected at different locations. The rst was injected close to the dryer
centerline at the distance of 0.077 m from it and the second was
introduced at the dryer peripheral zone of 0.585 m form the
Fig. 7. 3D air ow elds for the case of PVC drying (known wall temperature). (a) velocity magnitude, m/s (b) temperature, K, (c) vapour mass fraction, kg/kg, (d) density,
kg/m
3
.
378 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
Fig. 8. Mass-weighted averages of actual 3D air ow elds.
379 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
centerline. For both drying cases, a considerable difference between
the central and peripheral particles temperatures is observed, reach-
ing up to 10 K for the case of adiabatic owand up to 20 K for the case
with known wall temperature. Furthermore, in the periphery region
the particle transformation from a wet particle into a dry particle
occurs with relative displacement equal about 1 m for the case of
adiabatic ow and 2 m for the case with known wall temperature.
These results are consequences of different drying conditions in the
central and peripheral zones of the pneumatic dryer which eventually
lead to an uneven quality of the obtained product. Also not presented
here, the same trends were observed for silica particles drying as well.
The 1D mass-weighted averages of particle drying kinetics
predicted for the silica and PVC particles are shown in Fig. 10. It can
be found that the drying process of silica particles goes slower than
PVC particles. This phenomenon can be explained by greater heat
capacity of the silica solids (see Table 1).
Analyzing Figs. 9 and 10, it can be found that the average particle
moisture content remains constant and equal to the assumed nal
value (see Table 1) in about 80% of the dryer length. Therefore these
results indicate a possibility of increasing the initial product moisture
content or shortening the length of the adopted drying chamber for all
the studied cases.
Figs. 11 and 12 demonstrate the capability of the utilized drying
kinetics model to predict the evolution of temperature proles within
wet particles. In these gures the mass-weighted averages of tempera-
tures of wet particle center and outer surface are illustrated in the rst
and second drying stages. In the subsequent period of dry particle
heating (not shown here) the temperature in the particle center
approaches the value at the particle surface. Analyzing the calculated
data, it is concluded that the duration of the rst drying stage (including
initial heating and equilibrium evaporation periods) is negligible for
both types of the wet particles. Furthermore, it is found that the
maximum temperature difference in PVC particles attains up to 12 K
whilst inthe silica particles this valueis upto3 K. Theseobservations are
explained by lowthermal conductivity of the PVCparticles solidcontent
(see Table 1). It is worth noting that the presence of such steep
temperature gradients in micron-sized PVC particles can lead to
signicant thermal stresses in the particles and eventually result in
product thermal degradation, see Mezhericher et al. [23,30]. Therefore,
the drying of PVC wet particles potentially results in greater rate of
thermal degradation in the nal product than drying of the silica.
5.2. Comparison with experimental measurements
For the pneumatic drying of PVC particles with 140 m diameter
(PVC-140), the predicted proles of air temperature and particle
average moisture content can be compared to the Baeyens et al. [1]
experimental data. To obtain a good agreement with the experimental
data, Baeyens et al. [1] and Levy and Borde [3] recommend using the
following correlations of Nusselt and Sherwood numbers:
Nu
p
= 0:15Re; 37
Sh
p
= 0:15Re: 38
It is worthnoting, that Baeyens et al. [1] announced5 radial (r=0 m,
0.38 m, 0.48 m, 0.57 mand 0.62 m) and 8 height (z=0 m, 4.5 m, 7.5 m,
Fig. 9. Drying kinetics of PVC particles in the central and peripheral injections. (a)
adiabatic ow, (b) known wall temperature.
Fig. 10. Averaged drying kinetics of wet particles in pneumatic dryer. (a) adiabatic
ow, (b) known wall temperature.
380 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
10.5 m, 13.5 m, 16.5 m, 19.5 m and 22.5 m) sampling positions in their
experimental setup. However, in the subsequent parts of the article [1]
the measured air temperature and particle moisture content were
illustrated only for 7 experimental points located at various dryer
heights (z=0 m, 4.5 m, 7.5 m, 10.5 m, 13.5 m, 16.5 m and 19.5 m).
Moreover, Baeyens et al. [1] did not indicate explicitly for which of the
declared radial positions the presented experimental data were
obtained. Taking into account these uncertainties, and three-dimen-
sionality of air and particle ow elds in the dryer observed in the
previous section, the predicted longitude distributions of air tempera-
ture and particle moisture content at various for radial sampling
positions are compared with Baeyens et al. [1] experimental data (see
Figs. 13 and 14). It can be found that all the experimental temperature
points are lying within the range of the calculated upper and lower
temperature curves; however no one of these curves corresponds
exactly to the measured data found in [1]. The calculated curves of
particle moisture content are in satisfactory agreement with experi-
mental points for the case when Baeyens et al. [1] correlations were
utilized (Eqs. 37 and 38) and substantially over predict the experiments
when RanzMarshall Eqs. (21) and (22) were implemented. It is worth
noting that the same trends for particle moisture content were reported
by Baeyens et al. [1] and Levy and Borde [3].
Summarizing Figs. 13 and 14, due to many uncertainties in the
Baeyens et al. experimental data, only a satisfactory agreement between
the calculations and the measurements can be obtained. For better
comparison of the model predictions with the experimental points and
answer thequestionwhichsemi-empirical correlations of heat andmass
transfer coefcients are better for 3Dpneumatic drying modelling, more
information is necessary and thus a separate validation study involving
sets of experiments on the pneumatic dryer is a prospect work.
Fig. 11. Averaged temperature evolution at center and surface of PVC wet particles
during 1st and 2nd drying stages. (a) adiabatic ow, (b) known wall temperature.
Fig. 12. Averaged temperature evolution at center and surface of silica wet particles
during 1st and 2nd drying stages. (a) adiabatic ow, (b) known wall temperature.
Fig. 13. Comparison between the predicted proles of drying air temperature and
Baeyens et al. [1] experimental points for PVC-140 (known wall temperature). The
calculation results are given for different radial positions, solid lines RanzMarshall
correlations (Eqs. 21 and 22) and dashed lines Baeyens et al. correlations (Eqs. 37 and
38).
381 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
6. Conclusions
Steady-state three-dimensional calculations of the pneumatic
drying processes in vertical dryer were performed. The presented
theoretical model of the process is based on two-phase Eulerian
Lagrangian approach for gas-particles ow. The two-stage drying
kinetics of wet particles is described with the help of an advanced
theoretical model validated for single wet particle drying. The
developed model was numerically solved using the 3D pressure-
based steady-state solver of FLUENT package. The processes of
pneumatic drying of 140 m PVC and silica particles in a large-scale
vertical pneumatic dryer were numerically simulated. The inuence
of wall thermal boundary conditions was investigated by assuming
either adiabatic ow in the dryer or linear drop of the wall
temperature from 325 at the inlet to 320 K at the outlet. The obtained
3D results demonstrated three-dimensionality and complexity of the
ow in the pneumatic dryer. For qualitative illustration of the main
ow features, 1D mass-weighted averaging of 3D ow elds has been
utilized. Analyzing the predicted particle drying kinetics, a consider-
able difference between the central and peripheral particles temper-
ature was observed. Such a difference is a sequence of non-uniform
drying conditions in the central and peripheral zones of the
pneumatic dryer and can lead to an uneven quality of the obtained
product. Moreover, for the case of non-insulated chamber walls this
quality unevenness is estimated to be substantially greater than for
the case with thermally insulated drying chamber. The examination of
the predicted temperature proles within the silica and PVC wet
particles showed that the latter are subjected to higher temperature
gradients that potentially result in greater thermal stresses and could
lead to thermal degradation in the nal product.
Notation
A surface area, m
2
B Spalding number
B
k
crust permeability, m
2
C
D
drag coefcient
c
p
specic heat under constant pressure, J kg
1
K
1
d diameter, m
D
v
coefcient of vapour diffusion, m
2
s
1
F
A
virtual mass force, N
F
B
buoyancy force, N
F
C
contact force, N
F
D
drag force, N
G gravity acceleration, m s
2
h heat transfer coefcient, W m
2
K
1
; specic enthalpy, J kg
1
h
D
mass transfer coefcient, m s
1
h
fg
specic heat of evaporation, J kg
1
k thermal conductivity, W m
1
K
1
; turbulence kinetic energy,
m
2
s
2
m mass, kg

v
vapour mass transfer rate, kg s
1
M molecular weight, kg mol
1
n
p
number density of discrete phase
Nu Nusselt number
p pressure, Pa
Pr Prandtl number
r radial space coordinate, m
R radius, m
universal gas constant, J mol
1
K
1
Sc Schmidt number
S
c
mass source term
Sh Sherwood number
S
h
energy source term
t time, s
T temperature, K
u velocity of drying agent, m s
1
U

p
vector of particle velocity, m s
1
u

vector of gas velocity, m s


1
x space coordinate, m
X moisture content (dry basis), kg kg
1
y space coordinate, m
z space coordinate, m
Greek symbols
thermal diffusivity, m
2
s
1
coefcient of thermal expansion, K
1
; empirical coefcient
particle crust porosity; dissipation of turbulence kinetic
energy, m
2
s
3
dynamic viscosity, kg m
1
s
1
kinematic viscosity, m
2
s
1
density, kg m
3
velocity, m s
1

v
vapour mass fraction
Subscripts
a air, dry air fraction
atm atmospheric
cr particle crust; critical
d droplet
diff diffusion
f nal point of drying process
ow forced ow
g drying agent
i crustwet core interface
m airvapour mixture
p particle, discrete phase
pores crust pores
r radial direction
s solid fraction; surface
v vapour
w water
wc particle wet core
0 initial point of drying process
1 initial point of droplet evaporation period
2 contributor
bulk of drying agent
Fig. 14. Comparison between the predicted proles of particle moisture content and
Baeyens et al. [1] experimental points for PVC-140 (known wall temperature). The
calculation results are given for periphery and center of the dryer, solid lines Ranz
Marshall correlations (Eqs. 21 and 22) and dashed lines Baeyens et al. correlations
(Eqs. 37 and 38).
382 M. Mezhericher et al. / Powder Technology 203 (2010) 371383
Acknowledgement
The authors appreciatively acknowledge the nancial support of
the present work from GIF German-Israeli Foundation for Scientic
Research and Development under grant no. 952-19.10/2007.
References
[1] J. Baeyens, D. van Gauwbergen, I. Vinckier, Pneumatic drying: the use of large-
scale experimental data in a design procedure, Powder Technology 83 (1995)
139148.
[2] I.C. Kemp, D.E. Oakley, R.E. Bahu, Computational uid dynamics modelling of
vertical pneumatic conveying dryers, Powder Technology 65 (1991) 477484.
[3] A. Levy, I. Borde, Steady state one dimensional ow model for a pneumatic dryer,
Chemical Engineering and Processing 38 (1999) 121130.
[4] A. Levy, I. Borde, Two-uid model for pneumatic drying of particulate materials,
Drying Technology 19 (8) (2001) 17731788.
[5] I. Skuratovsky, A. Levy, Finite volume approach for solving multiphase ows in
vertical pneumatic dryers, International Journal of Numerical Methods for Heat
and Fluid Flow 14 (8) (2004) 9801001.
[6] I. Skuratovsky, A. Levy, I. Borde, Two-uid, two-dimensional model for pneumatic
drying, Drying Technology 21 (9) (2003) 16451668.
[7] I. Skuratovsky, A. Levy, I. Borde, Two-dimensional numerical simulations of the
pneumatic drying in vertical pipes, Chemical Engineering and Processing 44 (2005)
187192.
[8] K.S. Rajan, K. Dhasandhan, S.N. Srivastava, B. Pitchumani, Studies on gassolid
heat transfer during pneumatic conveying, International Journal of Heat and Mass
Transfer 51 (2008) 28012813.
[9] F. Tanaka, Y. Maeda, T. Uchino, D. Hamanaka, G.G. Atungulu, Monte Carlo
simulation of the collective behavior of food particles in pneumatic drying
operation, LWT-Food Science and Technology 41 (9) (2008) 15671574.
[10] F. Tanaka, T. Uchino, D. Hamanaka, G.G. Atungulu, Mathematical modeling of
pneumatic drying of rice powder, Journal of Food Engineering 88 (4) (2008)
492498.
[11] J.T. Li, D.J. Mason, Application of the discrete element modelling in air drying of
particulate solids, Drying Technology 20 (2002) 255282.
[12] T. Brosh, A. Levy, Modelling of heat transfer in pneumatic conveyer using a
combined DEM-CFD numerical code, Drying Technology 28 (2010) 155164.
[13] ANSYS FLUENT 12 Documentation. Ansys Inc., 2008; www.uent.com.
[14] M. Mezhericher, A. Levy, I. Borde, Dropletdroplet interactions in spray drying
using 2D computational uid dynamics, Drying Technology 26 (3) (2008)
265282.
[15] Mezhericher, M. Drying of slurries in spray dryers. Ph.D. Thesis, Ben Gurion
University of the Negev, Beer Sheva, Israel, 2008.
[16] M. Mezhericher, A. Levy, I. Borde, Modelling of droplet drying in spray chambers
using 2Dand 3DComputational Fluid Dynamics, Drying Technology 27 (3) (2009)
359370.
[17] Y. Tsuji, T. Kawaguchi, T. Tanaka, Discrete particles simulation of two dimensional
uidized bed, Powder Technology 77 (1993) 7987.
[18] B.P.B. Hoomans, J.A.M. Kuipers, W.J. Briels, W.P.M. Van Swaaij, Discrete particle
simulation of bubble and slug formation in a two-dimensional gas-uidized bed: a
hard-sphere approach, Chemical Engineering Science 51 (1996) 99118.
[19] B.H. Xu, A.B. Yu, Numerical simulation of the gassolid ow in a uidized bed by
combing discrete particle method with computational uid dynamics, Chemical
Engineering Science 52 (1997) 27852809.
[20] J. Ouyang, J.H. Li, Discrete simulations of heterogeneous structure and dynamic
behavior in gassolid uidization, Chemical Engineering Science 54 (1999)
54275440.
[21] M. Xu, W. Ge, J. Li, A discrete particle model for particleuid ow with
considerations of sub-grid structures, Chemical Engineering Science 62 (2007)
23022308.
[22] M. Mezhericher, A. Levy, I. Borde, Theoretical drying model of single droplets
containing insoluble or dissolved solids, Drying Technology 25 (6) (2007)
10251032.
[23] M. Mezhericher, A. Levy, I. Borde, Modelling of particle breakage during drying,
Chemical Engineering and Processing: Process Intensication 47 (8) (2008)
14041411.
[24] M. Mezhericher, A. Levy, I. Borde, Theoretical models of single droplet drying
kinetics: a review, Drying Technology 28 (2010) 278293.
[25] L. Zhou, Theory and Numerical Modelling of Turbulent Gas-Particle Flows and
Combustion, Science Press and CRC Press Inc., Hong Kong, 1993.
[26] L.S. Fan, C. Zhu, Principles of GasSolid Flows, Cambridge University Press, 1998.
[27] C. Crowe, M. Sommerfeld, Y. Tsuji, Multiphase Flows with Droplets and Particles,
CRC Press, 1998.
[28] D. Levi-Hevroni, A. Levy, I. Borde, Mathematical modelling of drying of liquid/solid
slurries in steady state one-dimensional ow, Drying Technology 13 (57) (1995)
11871201.
[29] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere, Washington,
DC, 1980.
[30] M. Mezhericher, A. Levy, I. Borde, Heat and mass transfer and breakage of particles
in drying processes, Drying Technology 27 (7&8) (2009) 870877.
[31] H. Minoshima, K. Matsushima, H. Liang, K. Shinohara, Estimation of diameter of
granule prepared by spray drying of slurry with fast and easy evaporation, Journal
of Chemical Engineering Japan 35 (9) (2002) 880885.
[32] R.H. Perry, D.W. Green, J.O. Maloney (Eds.), Perry's Chemical Engineers'
Handbook, McGraw-Hill, New York, 1997.
383 M. Mezhericher et al. / Powder Technology 203 (2010) 371383