Chinese Journal of Chemical Engineering 29 (2021) 57–66

Contents lists available at ScienceDirect

Chinese Journal of Chemical Engineering

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

Article

Numerical optimization for blades of Intermig impeller in solid–liquid

stirred tank
Xiaolong Li 1,2, Hongliang Zhao 3, Zimu Zhang 1,2, Yan Liu 1,2,⇑, Ting’an Zhang 1,2
1
Key Laboratory for Ecological Utilization of Multimetallic Mineral, Ministry of Education, Shenyang 110819, China
2
School of Metallurgy, Northeastern University, Shenyang 110819, China
3
School of Metallurgy & Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China

a r t i c l e i n f o a b s t r a c t

Article history: The multiphase flow in the solid–liquid tank stirred with a new structure of Intermig impeller was ana-
Received 18 November 2019 lyzed by computational fluid dynamics (CFD). The Eulerian multiphase model and standard k–e turbu-
Received in revised form 4 August 2020 lence model were adopted to simulate the fluid flow, turbulent kinetic energy distribution, mixing
Accepted 22 August 2020
performance and power consumption in a stirred tank. The simulation results were also verified by
Available online 8 October 2020
the water model experiments, and good agreement was achieved. The solid–liquid mixing performances
of Intermig impeller with different blade structures were compared in detail. The results show that the
Keywords:
improved Intermig impeller not only enhances the solid mixing and suspension, but also saves more than
Intermig impeller
Solid-liquid mixing
20% power compared with the standard one. The inner blades have relatively little influence on power
Stirred tank and the best angle of inner blades is 45°, while the outer blades affect greatly the power consumption
CFD and the optimized value is 45°.
Ó 2020 The Chemical Industry and Engineering Society of China, and Chemical Industry Press Co., Ltd. All
rights reserved.

1. Introduction minum Plant (in Guangxi, China) applied the Intermig impeller to
seed precipitation tanks [2] and configured the inner part of blade
The solid–liquid mechanical stirred reactors are applied widely inclined downward slightly. To avoid scabbing and depositing of
in chemical engineering and nonferrous metallurgy, such as crys- solid particles, Shenyang Aluminum Magnesium Design Institute,
tallization, leaching processes and catalytic reactions. The fluid China, then optimized blades and designed the improved Intermig
mixing, particle suspension, concentration distribution and power impeller [3] shown in Fig. 1(c). The improved Intermig impeller
consumption are mainly determined by the type, structure, and lengthens the outer blades, which promotes the scouring force sig-
operations of an impeller in a stirred tank. EKATO, a Germany com- nificantly. Shear force generated by the impeller and baffles also
pany, put the first Mig impeller into practical production as early as increases noticeably when the feed solution flows through baffles,
1964. Then, the Mig impeller was redesigned after absorbing the having positive effect on the decrease of deposition and scab at the
characteristics of the common pitch blade turbine. An outer blade bottom and corners of tanks, while enhancing mixing perfor-
inclined in the opposite direction to the inner blade was added to mances. Compared with the common pitch blade turbine, the
promote the overall circulation, as shown in Fig. 1a. Ever since, power consumption of the improved Intermig impeller reduces
EKATO altered its outer blade into two-layer structure and dramatically, to only about 1/5–1/4 of blade impeller’s [4].
obtained the Intermig impeller (shown in Fig. 1b) with a further Nowadays, a few investigations on the solid–liquid suspension
improvement on solid–liquid mixing performance. These two process by employing the Intermig impeller have been reported.
impellers belong to the multi-section counter-flow impeller and The suspension behaviors of solid particles in the Newtonian fluids
could generate axial flows, so they have the advantages of high cir- were studied by Ibrahim et al., who compared mixing perfor-
culation and low power consumption [1]. In 1994, PingGuo Alu- mances of five agitators (including the Intermig impeller) in low
viscosity liquids [5,6]. The results showed that fewer particles were
suspended in the case of higher viscosity. When the fluid viscosity
⇑ Corresponding author at: Department of Non-ferrous Metallurgy, School of
increased from 10
3 Pas to 0.01 Pas, the stirring performance of
Metallurgy, Northeastern University, Shenyang, China.
two-layer Intermig impeller improved remarkably, and the Inter-
E-mail addresses: zhaohl@ustb.edu.cn (H. Zhao), liuyan@smm.neu.edu.cn (Y.
Liu). mig impeller had the highest stirring efficiency when the viscosity

https://doi.org/10.1016/j.cjche.2020.08.044
1004-9541/Ó 2020 The Chemical Industry and Engineering Society of China, and Chemical Industry Press Co., Ltd. All rights reserved.
58 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66

Fig. 1. Structures of Mig, Intermig and improved Intermig impellers.

of fluid was 1 Pas. Bujalski et al. adopted the discoloring and con- 2.2. Method to measure concentration
ductivity methods in high solid hold-up systems, to investigate the
mixing time of Lightnin 310, 315 and Intermig impellers [7,8]. The In cool water model experiments, a particle concentration mea-
mixing time for high solid hold-up system was usually more than surement device (PC6D, Institute of Process Engineering, Chinese
two orders of magnitude than that for single-phase liquid, but for Academy of Sciences) based on the optical method [15] is used
Intermig impellers to achieve suspension, the mixing time to measure local solid hold-up in the stirred tank. Two optical
increased by five-fold only. fibers are arranged parallelly in a probe and each of them can
Although investigations on solid–liquid flow systems in stirred transmit the laser beam. When particles pass by the probe, they
reactors have been carried out, most of them were focused on will reflect the laser, which is received by the reflex optical fiber
single-phase [9,10] or solid–liquid systems with low solid hold- to a photoelectric detector and then converted into a voltage signal
up [11]. There were few reports [12–14] on the multiphase stirring proportional to solid hold-up. Thus, the phase volume fraction of
systems with high solid hold-up by the improved Intermig solid particles is acquired by analyzing the average of voltage sig-
impeller. nals. Shan et al. applied the same instrument to measure the distri-
This paper adopts CFD methods to investigate the effects of the bution of solid hold-up in a stirred tank without baffle, with the
structures of the improved Intermig impeller on the solid–liquid relative error within 0.5% [16].
mixing performance, particle suspension and power consumption
in the high solid hold-up system. The simulation results are also 3. Modeling Details
verified by the cold model experiments. The optimized structure
of the Intermig impeller is obtained and the present results may The Eulerian multiphase model and the standard k–e turbulence
provide significant guidance to the design of impellers and the model [17] are selected to predict the three-dimensional unsteady
scale-up of stirred reactors. flow field and the solid–liquid mixing performance. The effects of
mass transfer, lift force and virtual mass force are ignored due to
their minor effects in the present situation.
2. Experimental
3.1. Governing equations
2.1. Water model experiment
The continuity equation for each phase (i = l or s) can be written
A lab-scale stirred tank of 425 mm in diameter was manufac- as:
tured according to the geometric structures of the industrial cylin- @
!

drical stirred tank with the diameter of 14 m. Fig. 2 shows the ðai qi Þ þ r  ai qi v i ¼ 0 ð1Þ
@t
experimental equipment with diameter (T) and height (H) are
425 mm and 500 mm, respectively. Two baffles are installed off The momentum equations for liquid and solid phases can be
the wall of the stirred tank with a gap of 10 mm, and the lower written as:
@


ends of two baffles stretch downstream the stirred suspension

al ql !
v l þ r  al ql ! v l ¼
al rp þ r  sl þ al ql !
v l!


with a radius of 50 mm. Top of the tank is equipped with a cover g
@t
plate, which has 48 holes for sampling and measuring particle con- !

! !
centration across the whole reactor. The diameter of the impeller þ F l þ K sl v s
v l ð2Þ
(D) is 0.624 T. Meanwhile, the ratio for the distance (between lower
@


blade and the bottom of reactor) and diameter of reactor is C/
as qs !
v s þ r  as qs !
v s!
vs
T = 0.071. Agitation speed is kept at N = 250 rmin
1. The impeller @t



rotates in a clockwise direction and baffles bend to the down-
! ! ! !
stream direction to allow high speed flow around the bottom cor-
¼
as rp þ r  ss þ as qs g þ F s þ K ls v l
v s ð3Þ
ner. The agitator pushes the fluid flow downwards. As the fluid

!

moves to the wall baffles, it turns upwards along the bending baf- where v i and sl;s are the average velocity and the Reynolds stress,
fles, alleviating the probability of particle settling. Water and glass respectively. The latter item can be given as follows:
powder are utilized in the liquid–solid system, which maintains

 
2
sl;s ¼ al;s ll;s r!
v l;s þ r!
v l;s
Ir  !


v l;s
T
the average of solid hold-up (Cavg) at 800 gL
1 in the reactor (the ð4Þ
volume fraction is 0.33). Table 1 shows the geometric structure
3
of impeller and physical properties of test materials.
 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66 59

Fig. 2. Experimental system of cold water model (unit: mm).

Table 1 When al is below 0.8, Ksl can be calculated as:

Geometry of impeller and physical properties of test materials
 
Geometric structures of tank and impeller as ð1
al Þll as ql !
vs
!
v l
K ls ¼ K sl ¼ 150 þ 1:75 ð9Þ
Diameter/mm Tank 425 al ds
2 ds
Impeller 265
Height of liquid/mm 400 The standard k–e turbulent model is used for turbulence mod-
Distance between impeller and tank base/ 30 elling. The following transport equations for the turbulent kinetic
mm energy (k) and the turbulent energy dissipation rate (e) are
Width of baffle/mm 42.5 adopted.
Angle of inner impeller/(°) 60
   
Inner blade angle/(°) 15–60
@ ! lt
Outer blade angle/(°) 15–60 ðqkÞ þ r q v k
rk ¼ al ðGl
ql eÞ þ as ðGs
qs eÞ ð10Þ
@t rk
Physical properties of water and glass powder
Density/kgm
3    
Water (5℃) 1000
@ ! lt
Glass 2380 ðqeÞ þ r q v e
re
powder @t re
Viscosity/mPas Water 1.52 e e
Particle size/lm Glass 176 ¼ al ðC 1 Gl
C 2 ql eÞ þ as ðC 1 Gs
C 2 qs eÞ ð11Þ
k k
powder (average)
where G corresponds to the production of the turbulent kinetic
energy. The turbulent viscosity of the liquid phase, lt, is determined
The volume fractions of the phases add up to unity in each con- by:
trol volume, namely:
2
k
al þ as ¼ 1 ð5Þ lt ¼ qC l ð12Þ
e
Gidaspow model [18] is adopted to calculate the momentum
The parameters for turbulent model are chosen as follows:
exchange coefficient in the high solid hold-up system in this paper.
Cl = 0.09, C1 = 1.44, C2 = 1.92, rk = 1.0, re = 1.3.
When the liquid volume fraction al is bigger than 0.8, the momen-
tum exchange coefficient Ksl can be calculated as:
! ! 3.2. Boundary condition
3 as al ql  v s
v l 
2:65
K ls ¼ K sl ¼ C D al ð6Þ
4 ds The whole fluid region in the stirred tank is selected as compu-
tational domain, and the value of each variable on the solid wall is
where CD represents the drag force coefficient:
calculated with the standard wall function. The liquid level is set as
24 h i
symmetric interface, namely gradients of all physical variables are
CD ¼ 1 þ 0:15ðal Res Þ0:687 ð7Þ
al Res zero in the normal direction on the free liquid surface. Unsteady
where Res is the relative Reynolds number: sliding mesh method is employed to handle the rotation of the stir-
!  rer zone [19]. The SIMPLE algorithm based on the couple of the
ql ds v s
v l 
!
pressure and velocity, and the second-order upwind scheme is
Res ¼ ð8Þ adopted to discretize the governing equations. All residuals need
ll converge to 10
4.
60 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66

In numerical simulation, the method of surface integral for all 3.4.2. Test for time step independence
force components on the blade is utilized to calculate the resultant In unsteady numerical calculation, setting a suitable time step is
moment. Then the stirred power consumption can be expressed as: also vital. When the time step is too short, computing time will
increase significantly and round-off error may accumulate. While
2pMN
P ¼ Mx ¼ ð13Þ if the time step is overlong, it will lead to a greater cut-off error,
60 which makes the residual error difficult to converge. Now, some
methods rely on choosing time step by trial and error. For instance,
where P/W: the stirring power consumption; M/Nm: torque; x/
the ratio of the minimum length of grid cell and mean velocity of
rads
1: angular velocity. N/rmin
1: stirring speed.
flow field is set as t1. The ratio of characteristic length and charac-
teristic velocity is defined as t2. When t1 is lower than t2 two or
3.3. Mesh model more orders of magnitude in unsteady sliding mesh methods, some
scholars believe it suitable to choose 10% of the reciprocal of rotat-
Gambit was utilized for modeling and meshing of the stirred ing speed as the time step. The rotating speed varies from 150 to
tank and impeller. Unstructured grids, including the tetrahedron 250 rmin
1 in this work, so the longest time step will not exceed
and hexahedron are employed to divide the stirred tank. The com- 0.006 s. Therefore, three time steps, 0.0005 s, 0.002 s and 0.005 s,
plex region near the solid walls is discretized by tetrahedral mesh are chosen in the initial computation. The results are shown in
while hexahedral mesh occupies the inner zones. The thickness of Fig. 5. The distributions of velocity and solid hold-up in both axial
baffles is neglected during model building and meshing, mean- and radial directions show less difference (<3%) with different time
while the grid is more refined around the baffle surfaces. Fig. 3 steps. For a solid–liquid suspension system of high solid hold-up
shows the meshing for the stirred tank and impeller. and small particle sizes, particles normally fall slowly, which
means that particles need 1–2 min to reach steady motion. In this
work a short time step of 0.002 s was chosen at the beginning of
3.4. Independence verification
calculation, leading to convergent results. Then the time step
increased to 0.005 s to accelerate computing processes.
The effects of grid size, time step on the simulation results were
investigated, followed by the verification of CFD results against the
experiments under the same conditions. The main structures of the
tank and physical properties of solid–liquid two-phase system are
described in Table 1. The stirring speed of impeller is set at 250
3.4.3. Validation against water model experiment
rmin
1 and the ratio of liquid level to tank diameter is 0.94. The
After the tests on grid and time step independence, water model
volume fraction of solid particles is controlled at 0.33. The impeller
experiments were used to further verify the reliability of simula-
with the inner impeller angle of 60°, the inner blade angle of 30°
tion. Solid-liquid two-phase flow in a stirred tank is simulated by
and the outer blade angle of 30° is employed in the calculation
using water as the continuous liquid phase and glass powder as
to verify the simulation results. Those operating conditions and
the solid phase. Both calculations and experiments are imple-
geometric parameters are suitable for the results in Figs. 4–6.
mented in the same conditions. Three impellers with 0.624, 0.67
and 0.714 in D/T are used in water model experiments. Fig. 6 com-
3.4.1. Verification for grid independence pares experimental and simulation results in terms of solid hold-
Grids consisting of 520 k, 390 k and 260 k cells on the whole up distributions. Distributions of particles become more homoge-
computing domain are employed to simulate the liquid–solid neous as the diameter of impeller increases when the rotational
two-phase flow in the stirred tank. The liquid phase velocity and speed is same. The increase of diameter not only shrinks the parti-
solid hold-up distributions are compared in the context of different cle accumulation area near the bottom and wall of tank, but also
grids [20]. What we can see from Fig. 4 is that differences of liquid improves mixing effects in the axial direction, contributing to the
phase velocity and solid hold-up distributions are very small (less decrease of concentration gradient on the surface of liquid. For
than 5%) under different grids. Therefore, we decide to adopt the the solid hold-up in the axial direction, the results of simulation
260 k grid for the subsequent simulation, which cuts down compu- and experiments are identical, which verifies the reliability of
tational loads but also obtains reliable results. CFD model.

Fig. 3. Mesh models.

 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66 61

Fig. 4. Velocity and solid hold-up distributions with different grids.

Fig. 5. Velocity and solid hold-up distributions with different time steps.

Fig. 6. Comparison of solid hold-up distributions between experiment and simulation.

4. Results recorded when the stirring time is 60 s. After that, the distributions
of solid hold-up and power consumption change very little.
The mixing performances of blades with different angles
(shown in Fig. 7) are simulated with numerical methods. a1, a2, 4.1. The inclination of inner impeller
a3 are the angle of inner impeller, inner and outer blade, respec-
tively. It needs a long time for solid particles to achieve the final The inner impeller of a standard Intermig impeller is installed
stable state because of high solid hold-up (average solid volume horizontally in industry, while the improved Intermig impeller
fraction of 0.33). Hence, the present simulation results are inclines the inner impeller downwards by 30° (a1 = 60°). From
62 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66

Fig. 7. Setting angle of the blades.

streamlines showed in Fig. 8(a) and (b), we can see that the
improved Intermig impeller (a1 = 60°, a2 = 30°, a3 = 30°) can form
more intensive axial circulations which will be beneficial to axial
mixing for particles at bottom. Two small circulations at the end
of the impeller form on both two cases. The structure of the baffle
also affects the flow state. The inclined baffle in the Fig. 8(b) allows
the incoming flow to climb up along the baffle, enhancing the axial Fig. 9. Solid hold-up distributions with different types of impellers.
circulation of the fluid. This also promotes suspension and reduces
deposition of particles at the bottom of central regions. The main vectors of standard and improved Intermig impellers. The
function of vertical baffle is to obstruct the flow of suspension improved impeller combined with the bottom inclined baffles
and make it flow around the baffle, with little contribution to the formed a stronger axial circulation, which is beneficial to the sus-
axial circulation of the fluid. Fig. 8(c) and (d) compares the velocity pension of bottom particles.

Fig. 8. Streamlines and velocity vector under the interaction between impeller and baffles.
 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66 63

Fig. 9 shows changes of solid hold-up distributions with differ- thus it will go against to developed axial circulation. Fig. 11 illus-
ent sorts of impellers. The standard impeller is matched with 4 trates the turbulent kinetic energy with different inner blade
upright baffles, so the thickness of clear liquid layer at surface is angles. With the increase of the inner blade inclination, the effec-
smaller and particles deposit seriously on the bottom. By contrast, tive area driving the fluid will also increase, leading to the
the improved impeller has positive effect on particle suspension enhancement of turbulent kinetic energy. The turbulent energy
although the region of clear fluid is larger than that of the standard around the baffle is higher when the a2 ranges from 30° to 45°,
impeller. causing more particles to float up along the baffle.
Power consumption for agitator is meaningful in practice. From solid hold-up distributions in Fig. 12, we can see that with
Although the power consumption is mainly decided by the diame- the increase of a2, mixing effects are improved in both axial and
ter and rotational speed, the structure of agitator also has a signif- radial directions. When a2 reaches 45° (a1 = 60°, a3 = 30°), a better
icant effect. Power consumption for the standard impeller is mixing performance forms in tank where particle deposition
40.6 W, while the improved impeller is only 31.7 W, saving decreases significantly at the bottom. As a2 increases to 60°,
approximately 20% energy. despite the enhanced disturbance of the fluid, it does not obviously
improve the axial movement of particles from flow field analysis
above. Hence, particle concentration distribution has little change
4.2. Inner blade angle
at the surface of fluid. Simultaneously, owing to the fact that axial
pumping effects of blades are weakened in the vertical direction, it
Fig. 10 shows that axial circulations are strengthened with a2
will have adverse impact on particle upward motion. As a result,
increasing gradually from 15° to 45° (where a1 = 60°, a3 = 30°),
solid hold-up is larger at the bottom of blades, and deposition is
before weakened when a2 reaches 60°. Combined with the inner
more serious.
blade and the inclined baffles, the axial pumping performance
Fig. 13 compares power consumption at different a2. When the
becomes more forceful as a2 increases to 45°. The axial circulation
inner blade angle is 30°, the power consumption is relatively lower.
occupies the whole stirred tank and brings sufficient momentum
With the increase of inner blade angle, the effective agitation area
exchanges for fluid. When a2 further increases to 60°, although
increases, so the power consumption rises slightly. Each increment
the effective agitation area enlarges, circulations are strengthened
of 15° is added to blades, power consumption normally rises less
in the peripheral direction near sides of blade. This may have little
than 2%. As a2 increases to 60°, blades suffer greater torques, and
or even a negative effect on the upward movement of fluid, and

Fig. 10. Streamlines with different a2.

Fig. 11. Turbulent kinetic energy with different a2.

64 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66

Fig. 12. Solid hold-up distributions with different a2.

thus power consumption will have a significant increase, rising 4.3. Outer blade angle
about 10% compared with the figure of 30°.
Fig. 14 shows streamlines with outer blade angles (a3) ranging
from 15° to 60° and keeping a1 = 60° and a2 = 30°. As the agitator
moves clockwise, the outer blades push the fluid downward. With
the increase of a3, the downward pressure on fluid at bottom
enhances, in conjunction with baffles, promoting the development
of axial circulation flow. When outer blade angle is 15°, a certain
area of circulations forms only in the central lower parts of the stir-
red tank. Flow field is relatively flat at the surface of fluid, where
particles do not have enough suspending kinetic energy and it will
produce more easily a clear liquid layer. As a3 increases to 30°,
axial circulation flows are enhanced and the flat area at surface
has largely disappeared. When outer blade angle is 45° or larger,
stable axial circulation flows form in the whole tank, and fluid
mixes more fully, so it has little effect on flow field with an increas-
ing a3. Turbulent kinetic energy distributions with different a3 are
shown in Fig. 15. The changes of structure and size for outer blade
have a great influence on the distribution of turbulent kinetic
energy in the whole tank. The turbulent kinetic energy increases
obviously, especially in the region around impeller, when the outer
blade angle increases.
Fig. 13. Power consumption with different a2.

Fig. 14. Streamlines with different a3.

 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66 65

Fig. 15. Turbulent kinetic energy with different a3.

Fig. 16. Solid hold-up with different a3.

From solid hold-up distributions as shown in Fig. 16, we can 45°, the solid–liquid system is mixed uniformly except for the clear
know that outer blades affect overall solid hold-up distributions fluid in the central of axis. When a3 is further increased to 60°, the
significantly. When a3 is smaller, the effective agitation area in solid–liquid mixing effects are improved slightly, but the distribu-
the rotating direction is also smaller. Thus, particles are not mixed tion of particles tends to be uniform in whole tank.
well and deposit at the bottom, while a thick clear fluid layer Fig. 17 shows that power consumption tends to grow exponen-
appears at the surface. The performance of particle suspension is tially with the increase of outer blade angle. The outer blade angle
improved with the increase of outer blade angle. As a3 reaches is the primary factor that affects the power consumption in the
whole tank. This is mainly because outer blades are installed at
the end of impeller and the radius is larger, leading to larger tor-
ques. In the meanwhile, considering mixing effects, outer blades
make the fluid flow downward, which brings more fluid into the
impeller zone. As a3 increases to 45°, the system achieves the per-
fect solid–liquid mixing, although power consumption almost dou-
bled. With the further increase of a3, the mixing performance
improves little. Overall, outer blade angle should not be too large,
and 45° is more suitable.

5. Conclusions

The Eulerian multiphase model, standard k–e turbulence model

and unsteady sliding mesh method are selected to simulate the
solid–liquid mixing, suspension, and power consumption in the
stirred tank. The simulated model with independence of mesh size
Fig. 17. Power consumption with different a3. and time step is verified by water model experiment and good
66 X. Li et al. / Chinese Journal of Chemical Engineering 29 (2021) 57–66

agreements are obtained, verifying the reliability of the simulation [5] S. Ibrahim, A.W. Nienow, Comparing impeller performance for solid-
suspension in the transitional flow regime with Newtonian fluids, Chem. Eng.
method.
Res. Des. 77 (1999) 721–727.
The improved Intermig impeller not only enhances the solid [6] S. Ibrahim, A.W. Nienow, Particle suspension in the turbulent regime: The
mixing and suspension but also saves more than 20% power com- effect of impeller type and impeller vessel configuration, Chem. Eng. Res. Des.
pared with the standard one. The inner blades have relatively little 74 (1996) 679–688.
[7] W. Bujalski, S. Paolini, M. Jahoda, A. Paglianti, K. Takahashi, A.W. Nienow, A.W.
influence on power and the best angle of inner blades is 45°. While Etchells, K. Takenaka, Suspension and liquid homogenization in high solids
the outer blades greatly affect the power consumption. The best concentration stirred chemical reactors, Chem. Eng. Res. Des. 77 (1999) 241–
angle of outer blades is 45°, which has a relatively uniform 247.
[8] Li H.T., Study of interference multistage counterflow impeller in wide viscosity
solid–liquid distribution. domain, Master Thesis, Guizhou Univ., China, 2009 (in Chinese).
[9] Y.H. Zhang, C. Yang, Z.S. Mao, Large eddy simulation of turbulent flow in a
stirred tank, CIESC J. 58 (2007) 2474–2479 (in Chinese).
Declaration of Competing Interest [10] A. Bakker, H. Akker, Single-phase flow in stirred reactors, Chem. Eng. Res. Des.
72 (1994) 583–593.
The authors declare that they have no known competing finan- [11] G.R. Kasat, A.R. Khopkar, A.B. Pandit, CFD simulation of liquid-phase mixing in
solid-liquid stirred reactor, Chem. Eng. Sci. 63 (2008) 3877–3885.
cial interests or personal relationships that could have appeared [12] H.L. Zhao, Y. Liu, T.A. Zhang, S.Q. Gu, C. Zhang, Computational fluid dynamics
to influence the work reported in this paper. (CFD) simulations on multiphase flow in mechanically agitated seed
precipitation tank, JOM 66 (2014) 1218–1226.
[13] H.L. Zhao, Z.M. Zhang, T.A. Zhang, Y. Liu, S.Q. Gu, C. Zhang, Experimental and
Acknowledgements CFD studies of solid-liquid slurry tank stirred with an improved Intermig
impeller, Trans. Nonferrous Met. Soc. China 24 (2014) 2650–2659.
[14] H.L. Zhao, X. Zhao, L.F. Zhang, P. Yin, Experimental study on scale-up of solid-
This work was supported by the National Natural Science Foun-
liquid stirred tank with an Intermig impeller, JOM 69 (2017) 301–306.
dation of China (U1760120, U1508217), National Key R&D Program [15] H.A. Nasr-El-Din, R.S.M. Taggart, J.H. Masliyah, Local solids concentration
of China (2017YFC0210403, 2017YFC0210404) and Shenyang measurement in a slurry mixing tank, Chem. Eng. Sci. 51 (1996) 1209–1220.
Science & Technology Project (17-500-8-01). [16] X.G. Shan, G.Z. Yu, C. Yang, Z.S. Mao, W.G. Zhang, Numerical simulation of
liquid-solid flow in an unbaffled stirred tank with a pitched-blade turbine
downflow, Ind. Eng. Chem. Res. 47 (2008) 2926–2940.
References [17] B.E. Launder, D.B. Spalding, Lectures in Mathematical Models of Turbulence,
Academic Press, London, England, 1972.
[18] D. Gidaspow, R. Bezburuah, J. Ding, Hydrodynamics of circulating fluidized
[1] E.S. Szalai, P. Arratia, K. Johnson, F.J. Muzzio, Mixing analysis in a tank stirred beds, kinetic theory approach, Proceedings of the 7th Engineering Foundation
with ekato intermigÒ impellers, Chem. Eng. Sci. 59 (2004) 3793–3805.
Conference on Fluidization, Brisbane, Australia, 1992.
[2] D. Chen, Selection of central cooling schemes for 4500m3 large mechanically- [19] A. Bakker, R.D. Laroche, M.H. Wang, R.V. Calabrese, Sliding mesh simulation of
agitated digestion tank, Non-ferrous Smelt. 6 (1998) 84–86.
laminar flow in stirred reactors, Chem. Eng. Res. Des. 75 (1997) 42–44.
[3] Zhang T.S., Zhang C., Wang Y.C., A type of impellers for seed precipitation [20] H.L. Zhao, Z.M. Zhang, T.A. Zhang, Y. Liu, S.Q. Gu, C. Zhang, Experimental and
agitated tank, China Patent, ZL200520145940.6, 2006-12-20. CFD studies of solid
liquid slurry tank stirred with an improved Intermig
[4] Zhang C., Shan Y., Liu C.L., A type of large stirred tank, CN Patent,
impeller, Trans. Nonferrous Met. Soc. China 24 (2014) 2650–2659.
ZL200710158974.2, 2009-6-24.