Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e
i n f o
Article history:
Received 29 December 2011
Received in revised form 30 June 2012
Accepted 11 August 2012
Available online 20 August 2012
Keywords:
Heat and mass transfer
Pneumatic dryer
Two-phase
Numerical simulation
a b s t r a c t
Two-dimensional EulerianLagrangian model is presented for heat and mass transfer in pneumatic conveying dryer. The model takes into account the particleparticle and particlewall collisions, lift forces,
particle rotation, turbulence modulation and turbulence dispersion (i.e., four-way coupling). The drying
simulation is based on a two-stage drying model. Different correlations for heat transfer coefcient are
tested and assessed in terms of their accuracy. The model is validated against the available experimental
data and good agreement is obtained. The model predictions are compared to other models from literature and it produces better results than existing models. It is also found that the turbulence dispersion
has greater effect on the model predictions than particleparticle collision. However, neglecting either
particleparticle collision or turbulence dispersion results in a lower heat transfer and drying rates.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
Drying is an essential process in many industrial applications
such as food, agricultural, ceramic, polymers and plastic, pulp
and paper, pharmaceutical and wood processing industries. Drying
equipments are classied according to the heat transfer mechanism to convective and conductive and according to the handling
characteristics to batch and continuous operation [1]. Pneumatic
conveying dryers which can be classied as convective and continuous drying equipment is one of the most widely used equipment.
Pneumatic dryers are characterized by simultaneous momentum,
heat and mass transfer processes between the dispersed material
and the drying agent. The large surface area for heat and mass
transfer results in higher drying rate and higher drying capacity.
In these types of dryers the contact time between the drying medium and particulate material is relatively short (usually few seconds only). Therefore, these dryers are suitable for heat-sensitive
materials and also for removing external moisture. This allows
higher inlet temperatures to be used than in many other dryers
without unduly heating the product [1]. Pneumatic dryers are simple in construction and have low capital cost. Vertical type of construction, which facilitates installation in exiting buildings, is an
advantage of pneumatic dryer systems [2]. Among other dryers,
the pneumatic dryer shows the highest removal rate of the liquid
from the solid particle [3]. Drying calculations are based on the
knowledge of air and material properties. The successful
Corresponding author.
E-mail address: s_elbehery@yahoo.com (S.M. El-Behery).
0045-7930/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compuid.2012.08.006
160
Nomenclature
Symbols
Cp
Cl
D
Dv
dp
FD
FLR
FSL
Hfg
h
hh
hm
Ip
M
o
m
m
Nu
Pr
Prt
Rt
r
T
To
t
X
x
Y
Greek symbols
a
gas void fraction ()
b
solids void fraction ()
d
porosity ()
k
thermal conductivity (W/m K)
l
viscosity (kg/m s)
lt
turbulent viscosity (kg/m s)
q
density (kg/m3)
R
universal gas constant (J/kmole K)
sw
shear stress at wall (N/m2)
x
angular velocity (1/s)
Subscripts
Ave
average
cr
critical
da
dry air
g
gas
H2O
water vapor
p
particle
v
vapor
2. Mathematical model
Four-way coupling EulerianLagrangian approach is used in the
present study to predict the heat and mass transfer in gassolid
ow through pneumatic conveying dryer. The gas phase is simulated using Reynolds Averaged NavierStokes Equations. The turbulent viscosity is modeled by the low-Re ke model and the
particle tracking procedure is used for the solid phase. To provide
a reasonable solution for engineering objectives some simplifying
assumptions are taken as follow.
@
@/
1 @
@/
S/ S/p
aqu/ aC/
aqrv / aC/ r
@x
@x
r @r
@r
where S/ and S/p are source terms of gas and dispersed phases
respectively, while exchange coefcient, C/ is summarized in Table
1 for the dependent variable /.
161
C/
S/
Mass
Axial momentum
1
Ug
0.0
0.0
Radial momentum
Vg
leff
Energy equation
Tg
lg
Dissipation rate
leff
rk
leff
re
Y H2 O
Pr
f 2 1 0:3 exp R2t
f1 1:0;
D 2l
p!2
@ k
;
@y
E2
llt @U g
q @y
3
2
4
1
l
0:5882 0:228 t 0:0441
Prt
lg
lg
1 exp 5:165
lt
lg
!2
lt
!
dxp !
up
dt
Ip
aG qe D
a ke C e1 f1 G C e2 f2 qe aE
0.0
mp C pp
o
dT p
2
hh pdp T g T p md Hfg
dt
10
mp
Prtt
lt
qDv Sc
t
leff l lt ; lt C l fl q
h
i
ap
g
@
2
@
~
@@x
@x
aleff @U
ar leff @V@xg
1r @r
@x 3 r V g
h
i
ap
g
@
@
1r @r
@@r
@x
aleff @U
aleff r @V@rg 23 r ~
Vg
@r
V
Vg
2aleff r2g 23 alr e r ~
h
i
2
1
@
@
2C pg @x aqg U g U g V 2g 1r @r
r aqg V g U 2g V 2g
leff
md hm pdp
MH2 O pv o MH2 O pv g
RT p
RT g
11
where pvo and pvg are the partial pressures of water vapor at the particle surface and the gas phase.
Mezhericher et al. [17] introduced the following differential
equation to calculate the time change of particle radius during
the constant rate period.
o
dRp
1
md
dt
qw 4pR2p
12
!
!
!
!
dup !
F D F LS F LR F g
dt
!
dxp !
T
dt
!
!
3
!
To pldp 0:5r U g x p
! !
!
where x p is the particle position vector, U g ; u p are the gas and par!
ticle velocity vectors, x p is the particle angular velocity vector, To is
2
162
dp do
1=3
qs
X X cr 1
qw
13
md
1
1
4pd Ro Ri
1
M H2 O D v P P P v i
ln
P Pv o
RT av e
14
The mass transfer rate from the wet core, Eq. (14), must be equal
to that from particle surface to the gas, Eq. (11). Eliminating the partial pressure of water vapor at the dry crust, pvo from Eq. (14) using
Eq. (11) yields an implicit equation for the evaporation rate:
o
md
1
1
1
4pd Ro Ri
ln
P
1
M H2 O D v P
RT av e
P Pv i
o
RT p
MH
h
2O m
d2po
md pv g T pg
15
T g T p
!1
1
2
hh pdo
Tg Ti
do di
2
hpdo 2pdo di ks
!1
16
where Tp, Ti are temperatures of the outer dry crust and the inner
wet core, respectively, and Tave = (Tp + Ti)/2.
In general, during the second drying period, the outer shape of
the particle might be changed due to shrinkage of both outer and
core diameters. However, to simplify the model, it is assumed that
the particle outer diameter remains constant during the second
drying period. Thus, only the change of the wet core diameter, Di
is considered.
dDi
2 o
md
dt
dpD2i
17
Nu
hdp
;
kg
Rep
!
qg dp j !
ug u p j
lg
Pr
l g C pg
kg
0:667
Nu 0:015Re1:6
p Pr
20
Nu 0:035Re1:15
Pr0:333
p
21
Nu 0:15Rep
22
Nu
0:333
2 0:6Re0:5
p Pr
1 B0:7
C pH2 O T g T p
Hfg
23
24
0:8
Nu 2 0:5Re0:5
Pr0:333
p 0:06Rep
25
Sh
hm dp
;
Dv
Sc
lg
qg Dv
26
Eqs. (19)(25) have been tested and used to calculate the mass
transfer coefcient in the present study.
2.7. Coupling between the two phases
The particles occupy the computational cell and reduce the gas
volume fraction. They also exert interaction forces on the surrounding gas phase. Thus, the two phases are coupled through
the gas volume fraction and through the total source/sink term,
S/P that accounts for the momentum, heat and mass exchange between continuous and dispersed phases. The void fractions of dispersed phase, b and gas phase, a are calculated respectively using
trajectory method as depicted in [32] by:
X N k Dt k V k
traj
18
19
V Cell
a1b
27
where Nk is the number of actual particles in the computational particle parcel (k), Vk is the volume of the particle, Vcell is the volume of
163
SUp i
1 X
mk N k
V Cell traj
"
! #
Nt
X
q
n1
n
upi k upi k g i 1
Dt L
qp
n1
28
where DtL is the Lagrangian time step used in the solution of Eqs.
(3)(8), and summing over n indicates averaging along particle trajectory (time average).
The energy source term, STS , which represents the convective
heat exchange between the dispersed phase and the continuous
phase as well as the energy transferred to the gas phase by water
vapor, is given by:
STp
Nt h
i
X
o
1 X
2
Nk
hh pdp T g T p md hH2 O DtL
C pg V Cell traj
n1
29
Smass
S
mH2 O
P
Rg T g
35
R
;
Mg
36
kg Y H2 O kH2 O 1 Y H2 O kda
31
34
where
Rg
30
Y H2 O
Y H2 O 1 Y H2 O M H2 O =M da
qg
Nk
md Dt L
V C traj
n1
37
38
39
Sp C e3
e
k
Skp
32
where Ce3 is a model constant, its value varies in the literature from
1.1 to 2. Zhang and Reese [34] accounted for the reduction in the gas
turbulence length scale due to the presence of a second phase in Ce3.
Therefore, this model is adopted in the present study:
"
C e3
6b
1:95 1
pbm
1=3 #
33
Cpp
X
1
Cpw
Cps
1X
1X
40
qp qsa 1 X qs 1 d1 X
41
In the present study, there are four types of boundary conditions. At inlet, the gas velocity, temperature, mass fraction of water
vapor and turbulent kinetic energy and its dissipation rate are
specied. In compressible ow computations, the gas mass ow
rate is specied at inlet instead of the gas velocity. The inlet velocity prole is assumed to be uniform and the turbulent kinetic energy and its dissipation rate are calculated by:
3=2
ein C3=4
l
kin
0:03D
42
164
400
Tg (K)
380
Table 3
Coefcient of determination of the tested heat and mass transfer correlations.
Debrand [31]
Frantz [9]
Coefcient of determination, R2
Correlation
De Brandt [1]
Frantz [27]
De Brandt [1,11]
Debrand [20]
Baeyens et al. [1]
Modied RanzMarshall [11]
Modied Weber [17]
360
340
Gas temperature
0.4836
0.8172
0.7237
0.9927a
0.9795
0.9227
0.1268
0.7121
0.4163
0.9866a
0.8186
0.8916
320
a
300
10
15
20
25
x (m)
400
0.4
390
Present 2D Num.
0.3
Tg (K)
380
370
360
0.2
350
340
0.1
330
0.0
10
15
20
25
Rocha
[43]
Case I
Case II
1.25
25.0
190 700
PVC
140
1195
980
400
0.26
0.125
12.81
1.25
25.0
190 700
PVC
180
1116
980
399
0.4
0.125
12.9
0.0525
4.0
100 150
Sand
380
2622
799.7
382.4
0.0468
0.0381
0.03947
20
25
1.58
1.85
0.00474
300
0.01
300
0.01
312.9
0.0469
0.25
Table 2
Flow conditions and physical properties of drying test cases.
25
0.30
20
15
x (m)
x (m)
10
0.20
0.15
0.10
0.05
10
15
x (m)
165
390
322
Exp., Ref. [43]
Present 2D Num.
380
320
Tp (K)
Tg (K)
370
318
316
360
314
350
312
0
x (m)
x (m)
0.054
0.05
0.04
0.052
0.050
0.03
0.02
0.048
0.01
0.046
0.00
0
x (m)
x (m)
Fig. 4. Comparisons between present prediction and predictions of Refs. [11,16,18] with experimental data of Ref. [43].
by Skuratovsky et al. [15,16], Mezhericher et al. [17] and Jamaleddine and Ray [18], the wall temperature is varied linearly form
325 K at inlet to 320 K at outlet. Other conditions for this test case
are given in Table 2.
For direct comparison with experimental data, mass weighted
average technique is used to obtain the average value of various
solution properties (except temperature), Refs. [15,16], as follows:
2p
RR
2p
qU/r dr
qUr dr
0
R0 R
43
The average temperature of each phase (the mean bulk temperature) can be calculated as given [15,16] as:
2p
RR
qUC p Tr dr
2p 0 qUC p r dr
0
RR
44
166
0.4
Present 2D, r = 0.0 m
Present 2D, r = 0.62 m
Exp., Ref. [1]
0.3
0.2
0.1
0.0
10
15
20
25
20
25
x (m)
Tg (K)
380
360
340
320
10
15
x (m)
R2 1 Pi1
n
360
/i;num /i;exp
i1 /i;exp
Complete model
Without particle-particle collision
Without turbulence dispersion
380
/exp 2
45
The results presented in Table 3 indicate that the best prediction (the highest R2) is obtained when Baeyens et al. correlation
is applied. In addition, modied RanzMarshall and modied Weber correlations can be applied to the cases involving heat transfer
only. Overall, however, the correlation proposed by Baeyens et al.
[1] is recommended in the present study for the heat and mass
transfer coefcients.
Figs. 3 and 4 present comparisons between present predictions
and experimental data of Baeyens et al. [1] (Case II) and Rocha
[43], respectively. The experimental data of Rocha are reported in
Refs. [11,1418]. The pipe wall temperature for Rocha test case is
falling linearly from 360 K at inlet to 354 K at outlet. Other conditions for these test cases are given in Table 2. The gures also present direct comparison between the present predictions and other
computations from literature. The selected numerical results for
comparison are the one-dimensional prediction of Levy and Borde
[11], two-dimensional numerical results of Skuratovsky et al.
[15,16] and FLUENT three-dimensional predictions of Jamaleddine
and Ray [18]. These models are based on EulerianEulerian approach. The gures show that the present model agrees well with
experimental data for both cases. In addition, the present model
predicts the temperature and water content better than other
two-dimensional and three-dimensional EulerianEulerian models.
340
320
10
15
20
25
20
25
x (m)
Pn
Tg (K)
squares and explained sum of squares (i.e., total variation in experimental data), see Cameron and Windmeijer [44]. R2 can be calculated by:
400
0.3
0.2
0.1
0.0
10
15
x (m)
167
[19] Jamaleddine TJ, Ray MB. Drying of sludge in a pneumatic dryer using
computational uid dynamics. Dry Technol 2011;29:30822.
[20] NamKung W, Cho M. Pneumatic drying of iron ore particles in a vertical tube.
Dry Technol 2004;22:87791.
[21] Kaensup W, Kulwong S, Wongwises S. A small-scale pneumatic conveying
dryer of rough rice. Dry Technol 2006;24:10513.
[22] Kaensup W, Kulwong S, Wongwises S. Comparison of drying kinetics of paddy
using a pneumatic dryer with and without a cyclone. Dry Technol
2006;24:103945.
[23] Chagras V, Oesterl B, Boulet P. On heat transfer in gassolid pipe ows: effects
of collision induced alterations of the ow dynamics. Int J Heat Mass Transfer
2005;48:164961.
[24] Boulet P, Moissette S. Inuence of the particleturbulence modulation
modeling in the simulation of a non-isothermal gassolid ow. Int J Heat
Mass Transfer 2002;45:420116.
[25] Mansoori Z, Saffar-Avval M, Tabrizi HB, Ahmadi G. Modeling of heat
transfer in turbulent gassolid ow. Int J Heat Mass Transfer 2002;45:
117384.
[26] El-Behery SM, El-Askary WA, Hamed MH, Ibrahim KA. Hydrodynamic and
thermal elds analysis in gassolid two-phase ow. Int J Heat Fluid Flow
2011;32:74054.
[27] Mansoori Z, Saffar-Avval M, Tabrizi HB, Dabir B, Ahmadi G. Inter-particle heat
transfer in a riser of gassolid turbulent ows. Powder Technol
2005;159:3545.
[28] Launder BE, Sharma BI. Application of the energy-dissipation model of
turbulence to the calculation of ow near a spinning disc. Lett Heat Mass
Transfer 1974;1:1318.
[29] Kays WM. Turbulent Prandtl number where are we? J Heat Transfer
1994;116:28495.
[30] Shuen JS, Chen LD, Faeth GM. Evaluation of a stochastic model of particle
dispersion in a turbulent round jet. AIChE J 1983;29:16770.
[31] Debrand S. Heat transfer during a ash drying process. Ind Eng Chem Process
Des Develop 1974;13:396404.
[32] Crowe C, Sommerfeld M, Tsuji Y. Multiphase ow with droplets and
particles. Florida, USA: CRC Press; 1998.
[33] Lain S, Sommerfeld M. Turbulence modulation in dispersed two-phase ow
laden with solids from a Lagrangian perspective. Int J Heat Fluid Flow
2003;24:61625.
[34] Zhang YH, Reese JM. Gas turbulence modulation in a two-uid model for gas
solid ows. AICHE J 2003;49:304865.
[35] Oesterl B, Petitjean A. Simulation of particle-to-particle interactions in gas
solid ows. Int J Multiphase Flow 1993;19:199211.
[36] Sommerfeld M. Modelling of particlewall collisions in conned gasparticle
ows. Int J Multiphase Flow 1992;18:90526.
[37] Reynolds WC. Thermodynamic properties in SI: graphs, tables and
computational equations for forty substances. Published by the Department
of Mechanical Engineering. Stanford, CA 94305: Stanford University;
1979.
[38] Patankar SV. Numerical heat transfer and uid ow. New York, USA: McGrawHill; 1980.
[39] Karki KC. A calculation procedure for viscous ows at all speeds in complex
geometries. Ph.D. thesis. University of Minnesota; 1986.
[40] El-Behery SM, Hamed MH, El-Kadi MA, Ibrahim KA. CFD prediction of airsolid
ow in 180 curved duct. Powder Technol 2009;191:13042.
[41] El-Behery SM, Hamed MH, Ibrahim KA, El-Kadi MA. CFD evaluation of
solid particles erosion in curved ducts. ASME J Fluids Eng 2010;132:
07130313.
[42] El-Behery SM, El-Askary WA, Hamed MH, Ibrahim KA. Numerical and
experimental studies of heat transfer in particle-laden gas ows through a
vertical riser. Int J Heat Fluid Flow 2012;33:11830.
[43] Rocha SCS. Contribution to the study of pneumatic drying: simulation and
inuence of the gasparticle heat transfer coefcient. Ph.D. thesis. Sao Paulo:
EPUSP, Sao Paulo University; 1988.
[44] Cameron AC, Windmeijer FAG. An R-squared measure of goodness of t for
some common nonlinear regression models. J Economet 1997;77:32942.