Professional Documents
Culture Documents
Journal of Chemical Engineering of Japan, Vol. 47, No. 5, pp. 373381, 2014
Introduction
O-surface suspension of floating solids, which float at
the liquid surface without agitation, is commonly encountered in the processes involving mechanical stirring, such
as fermentation, mineral flotation, sewage treatment and
polymerization reactions. In a stirred tank, floating solids
are firstly drawn down from the liquid surface and then
distributed by the impeller throughout the tank. Hence, the
most critical place in a stirred tank is the liquid surface and
the diculty of solids drawdown depends much on the flow
pattern (Khazam and Kresta, 2008), which is strongly aected by the impeller design (Kumaresan and Joshi, 2006). Two
typical flow patterns produced by dierent PBT design are
single-eight and double-eight (Sharma and Shaikh, 2003).
For the single-eight flow pattern, a single circulation loop
fills the entire tank. While for the double-eight pattern, a
secondary circulation loop is formed in addition to a primary loop. An important performance criteria, the standard
deviation () of local solid concentrations, is widely used to
quantify the mixing quality in solidliquid systems and evaluate the competing impeller design (Khopkar et al., 2006).
Current methods used for studying the impeller design
are computational fluid dynamics (CFD), experimental fluid
dynamics (EFD) and empirical correlation method. CFD
method permits the numerical modeling of varied geometries and can provide qualitative or quantitative flow hydrodynamics underlying the system, especially the velocity field
and turbulent flow field. Such information can be obtained
using EFD method, but only by some expensive and complex equipment like particle image velocimetry. Three emReceived on November 18, 2013; accepted on January 6, 2014
DOI: 10.1252/jcej.13we313
Correspondence concerning this article should be addressed to X. Yang
(E-mail address: xxytju@163.com).
Vol. 47 No.52014
Copyright
2014The Society of Chemical Engineers, Japan
1.CFD Modeling
1.1Model equation
A commercial software Fluent 6.3 (ANSYS, Inc.) was
used for CFD modeling. The multiphase flow was simulated
with the Euler granular multiphase (EGM) model based
on the EulerEuler calculation, in which dierent phases
are treated mathematically as the interpenetrating continua.
Only a brief account of CFD model was given as following
and further details could be found in the literature (Ranade,
2002). The motion of each phase is governed by respective
mass and momentum conservation equations without considering mass transfer:
( )+ (i i ui ) = 0
t i i
(1)
( u ) + (i i ui ui ) = ip + ( i )
t i i i
+ i i g + Mi
(2)
3 CD
u ul (us ul )
4 dp l s s
(3)
Where dp is the particle diameter and CD is the drag coefficient obtained from the Brucato model (Brucato et al.,
1998), which could accounts for the influence of the free
stream turbulence on solid distribution and had been successfully employed in solidliquid systems, especially for
low solid loadings (Kasat et al., 2008; Ochieng and Onyango,
2008). Thus CD is calculated by the following equations:
3
CD / CD0
= 8.76 10/4 ( dp / )
CD0
CD0 =
24
1+ 0.15Re p0.687
Rep
(4)
(5)
(mixture, dispersed and per-phase) are developed to simulate the turbulent motion in multiphase systems. For the
mixture k model, only a couple of k and equations with
the physical properties of the mixture are solved and two
phases are assumed to share the same values of k and .
Montante and Magelli (2005) compared the axial profiles of
solid concentration predicted by three models, and found
the mixture k model was the most proper turbulence
model due to its less computational demand and qualitatively fair representation of the solid distributions. Moreover, this model is very applicable to that the density ratio
between the dispersed phase and continuous phase is close
to 1 (ANSYS, 2009). The solid and liquid densities simulated
in this work were 840 and 996 kg/m3, respectively. Hence,
the mixture k model was adopted.
1.4Solution domain and system
The investigated system consisted of a flat-bottomed cylindrical tank with the diameter T=0.32 m and liquid height
H=T. Four full baes of width 1/10 T were symmetrically
mounted perpendicular to the wall. For an up- and downpumping PBT with a fixed blade height h=1/5 D and blade
thickness t=2 mm, the studied impeller design variables
were: impeller diameter, D=0.250.41 T; o-bottom clearance, C=0.330.66 H; blade inclined angle to the horizontal,
= 3060; blade number, n=4 and 6. To reduce the computational requirement and numerical eort, one quarter
and one half of the symmetric domain was modeled, respectively, for 4- and 6-bladed PBT. The simulated continuous
phase and dispersed phase were tap water (l =996 kg/m3;
l =1103 Pa s) and polyethylene particles (dp =205 m;
s =840 kg/m3). For all the simulations, the dilute particles
with a mean solid loading of 5 kg/m3 (0.6% v/v) were initially uniformly distributed in the liquid.
1.5Solution method
The relative motion between rotating impeller and stationary baes was modeled by the MRF approach. The divided solution domain consisted of a central cylinder region
associated with the impeller discharge stream in a rotating
frame of reference and the rest region in a stationary frame
of reference. The diameter and height of the central cylinder region were 0.6 T and 4 h, respectively. Hexahedral and
tetrahedral elements were used to mesh the computational
domain, and finer grid was used in the impeller region to
ensure an adequate prediction of turbulent characteristic
produced by the impeller.
The periodic boundary condition was used for the left
and right halves of the symmetry plane. The symmetry
boundary condition (zero normal velocity gratitude) was assumed on the free liquid surface without considering the air
entrainment from the headspace and large surface vortex,
which is suppressed by the full baes. The no-slip boundary condition with standard wall function was assumed on
the tank wall, the surfaces of the baes, impeller and shaft.
All terms of governing equations were discretized by
a second-order upwind scheme and the phase coupled
Journal of Chemical Engineering of Japan
1
m
ci
i=1
avg
( c
1)2
(6)
Fig. 2 Solid distributions for (a) PBTD and (b) PBTU at C=0.33 H
(from left to right, D1 =0.25 T, D2 =0.33 T, D3 =0.41 T)
375
Fig. 5 Velocity fields for (a) PBTD and (b) PBTU at C=0.33 H (from
left to right, D1 =0.25 T, D2 =0.33 T, D3 =0.41 T)
for larger impeller (D2 and D3) is higher than smaller one
(D1).
For PBTU of D3 =0.41 T, complete drawdown can be
achieved at a much lower N around 500 rpm. But the ultimate is almost twice higher than the other two. The mixing quality of floating solids seems sensitive to D for PBTU.
This is attributed to the double-eight flow pattern produced
by the up-pumping mode of PBT. Seeing from the velocity
fields produced by PBTU at 700 rpm shown with the same
scale in Figure 5(b), the impeller discharge stream hits the
tank wall to generate two loops and an increase in D makes
the primary loop in the lower part of the tank become stronger and its eye moves down and outward. The particles are
more easily trapped in the downward flow of the primary
loop once drawn down from the liquid surface, and such an
eect becomes more noticeable as D is increased.
Comparisons of for three D of PBTU and PBTD indicate that PBTU is overall less ecient for o-surface suspension of floating solids on the condition of C=0.33 H.
For PBTU, the diculty to achieve complete drawdown
increases due to the double-eight flow pattern. Moreover,
this flow pattern requires more P to achieve complete drawdown. In previous EFD studies, some authors (Gray, 1987;
Sharma and Shaikh, 2003; Khazam and Kresta, 2009) found
the similar phenomena when the double-eight flow pattern
was responsible for o-surface suspension of floating solids
or o-bottom suspension of sinking solids.
2.3Effect of impeller clearance
The location of the impeller in the stirred tank, that
is, o-bottom clearance may also aect the solid distribution. So the eects were studied with PBTU and PBTD
of D=0.106 m (D/T=0.33) positioned at three clearances:
C1 =0.106 m (C1/H=0.33), C2 =0.157 m (C2/H=0.49) and
C3 =0.212 m (C3/H=0.66).
Journal of Chemical Engineering of Japan
Fig. 9 Velocity fields for (a) PBTU and (b) PBTD of D=0.33 T (from
left to right, C1 =0.33 H, C2 =0.49 H, C3 =0.66 H)
(7)
(8)
(9)
NQax
[o]
30
45
60
PBTU
PBTD
PBTU
PBTD
0.569
0.917
1.307
0.531
0.965
1.180
0.081
0.177
0.207
0.064
0.150
0.181
(10)
Where Vax and Vrad are axial and radial liquid velocity; the
axial integral region (Aax) is a circular surface with the same
diameter as the impeller just above and below the blade
edge, respectively, for PBTU and PBTD; the radial integral
region (Arad) is a cylindrical surface with the same height as
the blade at the impeller tip for both PBTU and PBTD.
In Table 1, NQ for PBTU is higher than that for PBTD and
NQax is much higher than NQrad for both PBTU and PBTD
at 600 rpm. Moreover, both pumping numbers, especially
NQax, increase as is increased, which agrees qualitatively
with EFD results of Ranade and Joshi (1989). zcan-Takin
and Wei (2003) reported that it was the axial flow to be
mostly responsible for o-surface suspension of floating
solids. Thus the diculty of floating solids mixing decreases
as is increased. However, the associated energy dissipation
behind the blades increases.
2.5Effect of blade number
4- and 6-bladed PBT are most widely used in stirred vessels. Thus the eect of blade number on floating solids mixing in stirred tanks was investigated by simulating the flow
produced by 4- and 6-bladed PBT. The D/T and C/H ratios
are both 0.33.
The curves of versus N show that 6-bladed PBT perform
better to achieve higher degree of homogeneity in Figure
14. For PBTU, even decrease from 0.93 to 0.23 at 400 rpm.
The maximum homogeneity is achieved at 500600 rpm,
beyond which the homogeneity has no improvement and
even begins to decrease. This can be attributed to the vigorous circulation flow discharged from 6-bladed PBTU. Seeing
from the solid distributions in Figure 15, both the primary
Vol. 47 No. 5 2014
Fig. 16Velocity fields for 4- and 6-bladed (a) PBTU (b) PBTD
the particles down from the liquid surface at lower N. However, an increase in blade number makes the energy dissipation behind the blades also increase.
Thus although 6-bladed PBT can achieve better performance, it requires more P. So were plotted against P to
further compare the mixing ability between 4- and 6-bladed
PBT in Figure 17. The results point to that 6-bladed PBT are
more energy ecient than 4-bladed PBT.
Conclusions
CFD simulations based on the EulerEuler approach
along with mixture k turbulence model were performed
to qualitatively investigate eects of impeller design on the
mixing quality of floating solids in stirred tanks with PBTU
and PBTD. The mean solid loading is 5 kg/m3 (0.6% v/v).
The main conclusions are as follows:
With an increase in D, as well as n, both and NCS decrease, but the associated P increases due to more energy
dissipation behind the blades.
The double-eight flow pattern requires more P for floating solids mixing.
For the impellers of dierent diameters positioned at
C=0.33 H, PBTU is overall less ecient than PBTD, especially for large impeller D3 =0.41 T.
Once the maximum homogeneity is achieved, further
increase in N is not beneficial and even be detrimental.
The drawdown performance is sensitive to C for both
PBTU and PBTD, but much more for PBTU. Further
increase in C above 0.66 H cannot further improve the
energy eciency for both PBT impellers.
has a more significant eect for PBTD and the impeller
pumping capacity can explain the eect of .
6-Bladed PBT are more energy ecient than 4-bladed
PBT.
CFD simulations show many promising results, which are
useful for extending the application of CFD to two-phase
flow involving floating solids and providing references for
impeller design for floating solids mixing in stirred vessels.
Nomenclature
A
C
CD
CD0
c
cavg
D
dp
FD
g
H
h
k
M
N
NCS
380
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
integral area
clearance from bottom to center of blades
drag coecient in turbulent liquid
drag coecient in quiescent liquid
local solid concentration
average solid concentration
impeller diameter
particle diameter
drag force
gravitational constant
liquid height
blade height
turbulent kinetic energy
inter-phase force
impeller speed
critical impeller speed
[m2]
[m]
[]
[]
[kg/m3]
[kg/m3]
[m]
[m]
[N/m3]
[m/s2]
[m]
[m]
[m2/s2]
[N/m3]
[rpm]
[rpm]
NQ
n
P
Q
Rep
t
T
V
v/v
=
=
=
=
=
=
=
=
=
pumping number
blade number
power input
volume flow rate
particle Reynolds number
blade thickness
tank diameter
velocity vector
solid volume/total volume
=
=
=
=
=
=
=
=
[]
[]
[W]
[m3/s]
[]
[m]
[m]
[m/s]
[]
[]
[o]
[m2/s3]
[m]
[kg/m3]
[Pa s]
[N/m]
[]
Subscripts
l
= liquid phase
s
= solid phase
ax
= axial direction
rad
= radial direction
tip
= blade tip
Literature Cited
ANSYS, Inc.; Ansys Fluent Theory Guide, Canonsburg, U.S.A. (2009)
Bakker, A. and J. J. Frijlink; The Drawdown and Dispersion of Floating
Solids in Aerated and Unaerated Stirred Vessels, Chem. Eng. Res.
Des., 67, 208210 (1989)
Bohnet, M. and G. Niesmak; Distribution of Solids in Stirred Suspensions, Ger. Chem. Eng., 3, 5765 (1980)
Brucato, A., F. Grisafi and G. Montante; Particles Drag Coecient in
Turbulence Fluids, Chem. Eng. Sci., 53, 32953314 (1998)
Driss, Z., G. Bouzgarrou, W. Chtourou, H. Kchaou and M. S. Abid;
Computational Studies of the Pitched Blade Turbines Design
Eect on the Stirred Tank Flow Characteristics, Eur. J. Mech. B/
Fluids, 29, 236245 (2010)
Gray, D. J.; Impeller Clearance Eect on o Bottom Particle Suspension in Agitated Vessels, Chem. Eng. Commun., 61, 151158
(1987)
Hosseini, S.; SolidLiquid Mixing in Agitated Tanks: Experimental and
CFD Analysis, M.A.Sc., Thesis, Ryerson University, Canada (2008)
Joosten, G. E. H., J. G. M. Schilder and A. M. Broere; The Suspension
of Floating Solids in Stirred Vessels, Trans. Inst. Chem. Eng., 55,
220223 (1977)
Kasat, G. R., A. R. Khopkar, V. V. Ranade and A. B. Pandit; CFD Simulation of LiquidPhase Mixing in SolidLiquid Stirred Reactor,
Chem. Eng. Sci., 63, 38773885 (2008)
Khazam, O. and S. M. Kresta; Mechanisms of Solids Drawdown in
Stirred Tanks, Can. J. Chem. Eng., 86, 622634 (2008)
Khazam, O. and S. M. Kresta; A Novel Geometry for Solids Drawdown
in Stirred Tanks, Chem. Eng. Res. Des., 87, 280290 (2009)
Khopkar, A. R., G. R. Kasat, A. B. Pandit and V. V. Ranade; Computational Fluid Dynamics Simulation of the Solid Suspension in a
Stirred Slurry Reactor, Ind. Eng. Chem. Res., 45, 44164428 (2006)
Kresta, S. M. and P. E. Wood; The Mean Flow Field Produced by a 45
Pitched Blade Turbine: Changes in the Circulation Pattern due to
O Bottom Clearance, Can. J. Chem. Eng., 71, 4253 (1993)
Kumaresan, T. and J. B. Joshi; Eect of Impeller Design on the Flow
Pattern and Mixing in Stirred Tanks, Chem. Eng. J., 115, 173193
(2006)
Kuzmani, N. and B. Ljubii; Suspension of Floating Solids with UpPumping Pitched Blade Impellers; Mixing Time and Power Characteristics, Chem. Eng. J., 84, 325333 (2001)
Kuzmani, N. and R. aneti; Influence of Floating Suspended Solids
on the Homogenization of the Liquid Phase in a Mixing Vessel,
Chem. Eng. Technol., 22, 943950 (1999)
Ljungqvist, M. and A. Rasmuson; Numerical Simulation of the TwoPhase Flow in an Axially Stirred Vessel, Chem. Eng. Res. Des., 79,
533546 (2001)
Mao, D., L. Feng, K. Wang and Y. Li; The Mean Flow Field Generated
by a Pitched Blade Turbine: Changes in the Circulation Pattern
due to Impeller Geometry, Can. J. Chem. Eng., 75, 307316 (1997)
Montante, G. and F. Magelli; Modelling of Solids Distribution in
Stirred Tanks: Analysis of Simulation Strategies and Comparison
with Experimental Data, Int. J. Comput. Fluid Dyn., 19, 253262
(2005)
Montante, G., G. Micale, F. Magelli and A. Brucato; Experiments and
CFD Predictions of Solid Particle Distribution in a Vessel Agitated with Four Pitched Blade Turbines, Chem. Eng. Res. Des., 79,
10051010 (2001)
Murthy, B. N., R. S. Ghadge and J. B. Joshi; CFD Simulations of Gas
LiquidSolid Stirred Reactor: Prediction of Critical Impeller Speed
for Solid Suspension, Chem. Eng. Sci., 62, 71847195 (2007)
381