Sie sind auf Seite 1von 10

chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036

Contents lists available at ScienceDirect


Chemical Engineering Research and Design
j our nal homepage: www. el sevi er . com/ l ocat e/ cher d
CFD simulation and PIV measurement of the ow eld
generated by modied pitched blade turbine impellers
Chun-Yan Ge, Jia-Jun Wang

, Xue-Ping Gu, Lian-Fang Feng


State Key Laboratory of Chemical Engineering, Department of Chemical and Biological Engineering, Zhejiang University, Hangzhou
310027, PR China
a b s t r a c t
The hydrodynamics generated by modied pitched blade turbine (m-PBT) impellers with down-pumping mode were
systematically investigated throughparticle image velocimetry (PIV) measurements and computational uid dynam-
ics simulations. The simulated meanaxial velocity, meanradial velocity, and turbulent kinetic energy by the standard
k turbulent model were validated against the measured PIV data. This shows that the standard k turbulent model
predicts mean velocity well, but underestimates turbulent kinetic energy near the blade. The ow eld and power
consumption as well as pumping number for the m-PBT and the standard PBT impeller were predicted. The sim-
ulation results demonstrate that a few simple changes of the blade shape inuence the velocity distribution, i.e.,
increasing the magnitude of mean velocity in the vicinity of impeller, and that the m-PBT impeller has a higher
pumping efciency than the standard one.
2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Computational uid dynamics; Particle image velocimetry; Power consumption; Stirred vessel; Turbulent
ow; Pitched blade turbine
1. Introduction
Stirred vessels are commonly used in chemical process
industries to carry out a variety of operations such as homog-
enization, gas dispersion, solid suspension and heat transfer.
In these processes, the required mixing quality varies over a
wide range. The type of impeller is one of the main factors
that determine the ow characteristics and achieve the goal
of mixing instirredtanks. Radial owimpellers are suitable for
applications ingasliquid, liquidliquid, and other multiphase
dispersions. Axial ow impellers are used for blending, heat
transfer, and solid suspension (Ranade et al., 1992). A calcula-
tionmethodwhichaccurately predicts the oweldaroundan
impeller of arbitrary shape would be an enormous benet to
the reactor designer (Kumaresan and Joshi, 2006; Patwardhan
and Joshi, 1999; Ranade et al., 1992). The information obtained
would be very useful for the design and operation of stirred
vessels.
Experimental investigation is a fundamental method to
reveal the ow characteristics in stirred tanks. The ow visu-
alization techniques used in stirred tanks include Pitot tube,

Corresponding author. Tel.: +86 57187951307; fax: +86 57187951612.


E-mail address: jiajunwang@zju.edu.cn (J.-J. Wang).
Received12 March2013; Receivedinrevisedform29 July 2013; Accepted26 August 2013
hot-wire anemometer, laser Doppler velocimetry (LDV), laser
induced uorescence (LIF), and particle image velocimetry
(PIV) (Mavros, 2001). PIV is convenient when obtaining the full
eld owcharacteristics ina specied region(Patwardhanand
Joshi, 1999). Inthe past, stirredtank designwas connedto the
use of empirical correlations, because more detailed analysis
of the ow elds by means of computational uid dynam-
ics (CFD) was not reliable due to limitations on computer
power and memory. However, advances in computer technol-
ogy andmathematical models have enabledresearchers touse
ner computational grids and smaller time-steps, as well as
more complex turbulence models (Delafosse et al., 2008). CFD
may not eliminate altogether the necessity for experiments
because of the complexity of the physics of ow.
Assessments of different turbulent models have beenmade
by many researchers (Hartmann et al., 2004; Joshi et al., 2011;
Wu, 2012). Hartmann et al. (2004) used the standard k tur-
bulent model, Reynolds stress model (RSM) and large eddy
simulation (LES) to study turbulence characteristics and com-
pared the results with LDV measurements. They found that
all the turbulent models predict the mean axial and tangential
0263-8762/$ see front matter 2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cherd.2013.08.024
1028 chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036
Nomenclature
C clearance between the impeller to the vessel
bottom(m)
D impeller diameter (m)
H height of vessel (m)
k local turbulent kinetic energy (m
2
s
2
)
N impeller rotational speed (rpm)
N
p
power number
N
Qc
circulation number
N
Qd
pumping number
r radial coordinate (m)
R
r
radial distance to vorticity center
T tank diameter (m)
U
tip
impeller tip velocity (ms
1
)
u
r
radial velocity (ms
1
)
u
t
tangential velocity (ms
1
)
u
z
axial velocity (ms
1
)
u

uctuating radial velocity (ms


1
)
v

uctuating axial velocity (ms


1
)
z axial coordinate (m)
energy dissipated per unit mass (m
2
s
3
)
j dynamic viscosity (pas)
, water density (kgm
3
)
velocities reasonably well, but underpredict the decay of mean
radial velocity away fromthe impeller. Joshi et al. (2011) con-
ducted a review on modeling of single-phase turbulent ow
in stirred tank reactors, in which the standard k, RNG k,
RSM, andLESwere involved. Wu (2012) evaluatedsixReynolds-
averaged NavierStokes (RANS) models and LES with three
subgrid scale models in a tank mixed by an axial impeller,
and observed the global ow patterns for all RANS simula-
tions being similar to the LES. The ow pattern in a dual
Rushton turbine stirred tank has been systematically investi-
gated by means of CFD and PIV (Li et al., 2011, 2012; Pan et al.,
2008). Theyreportedthat that bothphase-averagedandphase-
resolved results obtained by LES were in excellent agreement
with the PIV experimental data. An anisotropic analysis of the
impeller discharge region was done to help understand why
the k turbulent model fails to predict complex ow elds.
Unfortunately, the LES approach is out of reach in most practi-
cal settings due to the rather intense computations required.
Khopkar et al. (2003) investigated the gasliquid ow gener-
ated by a pitched-blade turbine throughPIVand CFDmethods,
and found that the standard k turbulence model correctly
captures the overall ow eld.
In stirred vessels, the mixing performance generated by
the impeller is mainly determined by the impeller design. The
desired ow pattern is dependent on the application. As the
owproceeds fromthe impeller and circulates within the ves-
sel, the meankinetic energy is convertedinto turbulent kinetic
energy and the relative distribution at any location depends
upon the design of the impeller, vessel and other internal con-
ditions. Kumaresan and Joshi (2006) undertook a systematic
investigation of the effect of various axial impeller designs
(blade angle, number of blades, blade width, blade twist, etc.)
on the ow pattern and mixing time. When the impellers
are compared at equal power consumption, the narrow blade
hydrofoil impeller with a blade twist of 15

yields the maxi-


mumownumber. Bugay et al. (2002) analyzedthe oweldin
a stirred tank induced by a Lightnin A310 axial impeller using
PIV technique with particular attention to turbulence (pro-
duction, transport, and dissipation rate of turbulent kinetic
energy). Ankamma and Sivashanmugam (2010) investigated
the power consumption for single and double rectangular and
V cuts in the conventional turbine agitator. They found that
the power consumption for the modied turbine agitator is
less than that of the conventional turbine agitator. However,
the pumping capacity for the energy-saving turbine agitator
was not taken into account.
However, the design and manufacture of high efciency
axial owimpellers are quite difcult to mold due to the com-
plicated shape, twist and bend of the blade. The standard
pitched blade turbine (PBT) impellers are made simply, but
provide low ow efciency. The overall goal of this study is
to design a new axial impeller (m-PBT) by only changing the
blade shape of the PBT impeller to improve the owefciency.
The specic objectives of this study were to: (1) carry out PIV
measurements and CFD simulations of mixing by the m-PBT
impellers, (2) propose a high efciency axial impeller, and (3)
reveal the relations among impeller shape, ow patterns and
mixing performance.
2. PIV experiment
Fig. 1 shows a schematic view of the experimental appara-
tus. In this experiment, a at bottom perspex tank with an
inner diameter of 0.34m was used, and four equally spaced
bafes with an identical width of 0.034mwere arranged along
the internal wall of the tank. The cylindrical tank was placed
inside a square tank to reduce the optical refractive index
effects at the cylindrical surface of the tank.
The m-PBT impellers were derived from the standard PBT
impeller by changing the impeller shape. The m-PBT with
three blades varied in blade angel (20

, 30

, 40

, and 45

,
referred to as m-PBT20, m-PBT30, m-PBT40, and m-PBT45).
Their solidity ratios are 0.43, 0.40, 0.35 and 0.33, respectively.
The standard pitched blade turbine impellers with 45

angle
were also evaluated as comparison with both three and four
blades (referred to as PBT45-3, PBT45-4). The ratio of the blade
surface of the modied impellers to that of the standard
PBT45-4 is 1.21. As shown in Fig. 1, the diameter of each
impeller used in this study is 0.144m, with the center of the
impeller located 0.133m from the bottom. The relative blade
width of PBT45-3 and PBT45-4 (with respect of the impeller
diameter) is 0.174. Tap water was used as the working uid.
The rotational speedwas set as 150rpm. The Reynolds number
can be calculated as 50,000, indicating turbulent ow.
The two-dimensional (2D) PIV system(Dantec Inc.) used in
this work consistedof a dual Nd:YAG532nmpulsedlaser (New
Wave Research Solo, 200mJ, 15Hz), a frame-straddling CCD
camera with 16001200 pixels, a synchronizer, and Dynamic
Studio software. Ashaft encoder was usedto obtainthe phase-
resolved measurements. The TTL signals triggered by the
encoder once-per-revolution were used to synchronize the
blade angular position, image acquisition, and laser ring. The
vertical laser plane in all of the experiments was placed at a 5

angle fromthe nearest bafe, passing through the vessel axis


(Fig. 2). One phase, corresponding to the horizontal central line
of the blade being in the vertical central plane of bafe, was
used to yield the phase-resolved velocity eld statistics.
A cross-correlation algorithm was applied to the 3232
pixels interrogation windows with 50% overlap using the
Dynamic Studio software to obtain the instantaneous velocity
chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036 1029
Fig. 1 The geometry of the experimental tank and impeller designs: (a) m-PBT20, (b) m-PBT30, (c) m-PBT40, (d) m-PBT45,
(e) PBT45-3, and (f) PBT45-4.
distribution, so the vector resolution in this work was about
1.5mm. The time difference between laser pulses was set at
150s, which was optimized to ensure that the maximum
in-plane and out-of-plane displacements of seeding particles
would be less than one-quarter of the size of the interrogation
windows and greater than the thickness of the light sheet (Li
et al., 2011). 400 image pairs were recorded for each experi-
ment. The statistical reliability of the selection of the number
of image pairs was tested by comparing the mean velocity
between400 and500 image pairs withthe difference being less
than 1%. In a 2DPIVsystem, the tangential component of uc-
tuating velocity cannot be directly measured, but it usually can
be estimated using a pseudo-isotropic assumption, resulting
in the following equation estimating turbulent kinetic energy
(Gabriele et al., 2009):
k =
3
4
(u

2
+v

2
) (1)
Fig. 2 PIV measurement system.
where u

and v

are uctuating radial and axial velocities,


respectively.
3. Numerical simulation
3.1. Standard k turbulent model
Ina turbulent ow, if the turbulence responds rather quickly to
changes in the mean ow, the Reynolds stresses can be appro-
priately modeled. Acommon method employs the Boussinesq
hypothesis to relate the Reynolds stresses to the meanvelocity
gradients:
,u

i
u

j
= j
t

u
i
x
j
+
u
j
x
i

2
3

,k +j
t
u
i
x
i

ij
(2)
The standard k turbulent model is essentially a high
Reynolds number model and assumes the existence of
isotropic turbulence and the spectral equilibrium(Murthy and
Joshi, 2008; Sahu et al., 1999). The k turbulence models suf-
fer from the necessity of modeling a number of quantities
for which reliable experimental data are desirable under a
large number of ow conditions. While this necessity is a
fundamental weakness of the k approach, a further uncer-
tainty lies in the assumption that the turbulent kinetic energy
and its dissipation rate are necessary and sufcient turbu-
lence variable for the simulation of turbulent ows. Due to
the advantage of the relatively lowcomputational cost associ-
ated with the computation of the turbulent viscosity (j
t
), the
model is commonly used in the calculation of the complex
ows in large-scale industrial reactors.
3.2. Simulation details
The geometrical details of the probleminvestigated are shown
in Fig. 1. The GAMBIT meshing tool was used to create geom-
etry and generate grids. The computational cells consisted of
1030 chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036
Fig. 3 Distributions of mean velocity for m-PBT40 obtained with (a) PIV and (b) CFD.
Fig. 4 Mean radial (a and c) and axial (b and d) velocities generated by 45

pitched blade turbine impellers at different axial


position: z =90mm(a and b), and z =130mm(c and d).
chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036 1031
Fig. 5 Turbulent kinetic energy distributions generated by 45

pitched blade turbine impellers at different axial position: (a)


z =90mm, and (b) z =130mm.
150,000 unstructured tetrahedrons and 450,000 hexahedrons
for the rotating and the stationary zone, respectively. In order
to ensure good quality meshing, impeller blades and bafes
were treated as zero-thickness solid walls. The cell sizes in
the impeller discharge region were 2mm, all the other cells
were 6mm. Both cell sizes were selected after an extensive
grid renement study.
The numerical simulations were conducted using the com-
mercial software package FLUENT 6.3.26, which is based on
the nite volume method. The discretized form of the gov-
erning equations for each cell was obtained such that the
conservation principles are obeyed on each cell. The sec-
ond order upwind was used for the spatial discretization of
momentumequations and the second-order accurate implicit
scheme was used for the time step for the turbulence model.
The coupling between the continuity and momentum equa-
tions was achieved using SIMPLE (semi-implicit method for
pressure-linked equations). The transient interaction of the
ow between impeller (rotation) and bafe (stationary) was
modeled using the sliding mesh (SM) approach. Time step was
set at 0.001s. To provide good initial values for the transient
simulation, a steady calculationwithmultiple reference frame
(MRF) was conducted.
4. Results and discussion
In this section, the radial, axial, and tangential components of
mean velocity are represented by u
r
, u
z
, and u
t
, respectively.
The origin of the coordinates is located at the center of the
tank bottom; the axial and radial coordinates are normal-
ized with respect to the tank diameter and tank height,
expressed by z/H and r/T, respectively. All average velocities
are phase-resolved results, and turbulent kinetic energies are
phase-averaged results.
4.1. Flow pattern for m-PBT
Fig. 3 shows the mean velocity vectors, in which velocity
magnitudes are indicated by the contour bar. The PIV mea-
surement is shown in Fig. 3(a). It can be observed that the
uid is discharged fromthe impeller bottomwith a maximum
velocity magnitude of 0.5U
tip
and is swept around the vessel
bottom and sides to extends to a height of z/H=0.6. Here,
it forms a relatively strong downward ow loop structure
with the velocities from 0.2 to 0.5U
tip
. Note that beneath
the impeller hub where the ow is reversed, there is a
region where solid suspension is difcult (Nienow, 1998).
The standard k simulation (Fig. 3(b)) provides a reasonable
model for the generation and location of the circulation
loop.
4.2. Effect of blade shape
CFD simulations with the standard k model and the slid-
ing mesh method and PIV measurements were conducted on
three different types of 45

pitched-blade down ow axial


Fig. 6 Normalized local specic energy dissipation rate in the blades for (a) m-PBT45 and (b) PBT45-3.
1032 chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036
Fig. 7 Mean radial (a and c) and axial (b and d) velocities generated by m-PBT20 against PBT45-4: z =90mm(a and b), and
z =130mm(c and d).
impellers, i.e., m-PBT45, PBT45-3, and PBT45-4. The com-
parisons of the mean radial and axial velocity proles at
different axial positions (belowand above the blade, i.e., z =90
and 130mm) are shown in Fig. 4. There is a negligible dif-
ference in both the mean axial and radial velocity trends
between PBT45-3 and PBT45-4. The non-randomvelocity uc-
tuation caused by the periodic passage of the m-PBT45 blades
makes the ow structure in the region close to the impeller
(0.1<r/T<0.3) more complicated in comparison to the two
others. The PBT45-4 and PBT45-3 reach a maximum veloc-
ity of 0.4U
tip
whereas m-PBT45 with its different blade shape
reaches to 0.47U
tip
at the same location. It can be thought that
the change of the blade shape by increasing the area of rear
corner for the m-PBT45 increases the magnitudes of radial
velocity and axial velocity in the impeller vicinity, and then
improves the axial pumping ow rate.
Fig. 8 Mean radial (a and c) and axial (b and d) velocities generated by m-PBT30 against PBT45-4: z =90mm(a and b), and
z =130mm(c and d).
chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036 1033
Fig. 9 Mean radial (a and c) and axial (b and d) velocities generated by m-PBT40 against PBT45-4: z =90mm(a and b), and
z =130mm(c and d).
A quantitative comparison of the predicted results and the
experimental dataof meanvelocityfor PBT45-4 are alsoshown
in Fig. 4. It can be seen that the comparison between the pre-
dicted values and experimental data of the axial and radial
velocity is satisfactory except for the value of radial velocity
in the vicinity of the impeller.
Fig. 5 shows the comparisons of the turbulent kinetic
energy (k) for the three impellers at two radial locations. From
the predictedresults at the two locations (z =90 and130mm), it
can be seen that the turbulent kinetic energy for the m-PBT45
is always greater thanthose for the PBT45-4 and PBT45-3 espe-
cially at the zone below the impeller, and that the turbulent
kinetic energy for m-PBT45 is much greater above the impeller
than below the impeller, which explains the different mixing
characteristics inthese zones withinthe same vessel. This low
turbulent kinetic energy belowthe impeller is not obvious with
the standard PBT. It is evident that the standard k turbulent
model underestimates turbulent kinetic energy for PBT45-4,
which agrees with the results of Hartmann et al. (2004). Li
et al. (2011) found that the regions with high turbulent kinetic
energy are affected by the movement of trailing vortices gen-
erated behind the impeller blades, but the k turbulent model
is unable to capture the chaotic and three-dimensional nature
of the trailing vortices (Harminder et al., 2011).
Khan et al. (2006) made angle-resolved stereo-PIV mea-
surements close to a down-pumping PBT impeller and found
that the average rate of dissipation of kinetic energy is 40
times higher in the impeller region than the average dissi-
pation rate in the vessel. Therefore, it is important to study
the energy dissipation in the impeller region for the impeller
optimization. In Fig. 6, the local specic energy dissipation
rate (
T
) on the blade surfaces of m-PBT45 and PBT45-3,
which calculated from the standard k turbulent model,
are compared. The value of
T
was normalized by the tank
averaged specic energy dissipation rate (
av
) as

T
=
T
,
av
.
in which
av
was determined from the calculation of power
consumption via the volume integration of turbulence energy
dissipation rate (Eq. (3)) obtained with the CFD model. Fig. 6(a)
and (b) shows that the value of

T
in the top edge of the
blades especially around the corners much higher than other
regions. Besides, the value of

T
in the bottom edge is very
small. The area with high

T
(>2000) for the m-PBT45 is
smaller than that for the PBT45-3. Therefore, one of the blade
optimization methods was to cut off top straight corner and
formarc shape to reduce local specic energy dissipation rate.
4.3. Effect of blade angle
The effect of blade angel on the ow pattern prevailing in the
stirred tank, comparing the mean axial and radial velocity for
m-PBT20, m-PBT30, m-PBT40 and m-PBT45 against PBT45-4
are shown in Figs. 79 and Fig. 4. At two axial positions, an
increase in the blade angle provides more power in pumping
the uid at r/T=0.150.25, especially below the impeller. The
maximummean radial velocity for m-PBT45 is 0.25U
tip
which
is muchhigher thanthe others belowthe impeller. Kumaresan
and Joshi (2006) observed the ow patterns of 45

six straight
blade pitched blade turbines with D/T=0.3 and found that the
mean radial velocity reached 0.2U
tip
.
Figs. 79 and Fig. 4 show the comparisons of the mean
axial velocity for m-PBT with different blade angles at two
axial locations. An increase in blade angle also increases the
axial velocity. The maximumaxial velocity at r/T=0.18 for m-
PBT45, m-PBT40, m-PBT30, and m-PBT20 are 0.52U
tip
, 0.4U
tip
,
0.35U
tip
, and 0.25U
tip
, respectively. Figs. 79 and Fig. 4 also
show a quantitative comparison of the predicted results and
the experimental data of mean velocity for each m-PBT. The
standard k turbulent model predictions of each component
of the mean velocity are in good agreement with the experi-
ment data except for a slight deviation for the m-PBT20.
The proles of turbulent kinetic energy are shown in
Figs. 1012 and Fig. 5. Fromthe simulation results, just below
1034 chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036
Fig. 10 Turbulent kinetic energy distributions generated by m-PBT20 against PBT45-4: (a) z =90mm, and (b) z =130mm.
Fig. 11 Turbulent kinetic energy distributions generated by m-PBT30 against PBT45-4: (a) z =90mm, and (b) z =130mm.
the impeller (z =90mm) the value of k increases slightly up to
r/T=0.17, increases sharply at r/T=0.23, and then decreases
drastically, eventually the turbulent kinetic energy is very
small and constant. Above the impeller (z =130mm), turbu-
lent kinetic energy is always small and varies less. It can be
seenthat the values of turbulent kinetic energy behave similar
to the mean velocity with respect to blade angle. The modi-
ed pitched blade turbine with blade angle of 45

generates
an intensive turbine ow. The maximumvalue of axial veloc-
ity is 0.52U
tip
, whereas it is only 0.25U
tip
for the impeller with
blade angle of 20

(Fig. 7b). Murthy and Joshi (2008) observed


the same peak for a k with a pitched blade down owturbine.
The standard k model fails to predict k in the impeller
vicinity belowthe impeller. Li et al. (2011) useda relatively sim-
ple method proposed by Lee and Yianneskis (1998) to conduct
anisotropic analysis in Rushton turbine stirred tanks. They
found that the turbulence deviates fromisotropic, which can
explain why the standard k model, based on the assumption
of local isotropic conditions, cannot correctly predict the ow
near the blade.
4.4. Power number and pumping number
Generally, power input and pumping capacity are used to
characterize the agitation in stirred reactors. These global
parameters are particularly useful when comparing the
mixing in agitated tanks equipped with different types of
impellers. The power consumedby the impeller during stirring
should be equal to the power dissipated by the impeller in the
liquid. Hence, the impeller power number can be calculated
fromthe volume integration of turbulence energy dissipation
rate predicated fromthe CFD model, in the following manner
(Devals et al., 2008):
N
p
=

R
0

H
0

2
0
,drdzd
,N
3
D
5
(3)
Using the impeller torque is another method to calculate
power number. However the thickness of the blade must be
accounted for and resolved properly by the mesh in the sim-
ulations (Singh et al., 2011).
Fig. 12 Turbulent kinetic energy distributions generated by m-PBT40 against PBT45-4: (a) z =90mm, and (b) z =130mm.
chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036 1035
Table 1 Power number (N
p
), circulation number (N
Qc
) and pumping number (N
Qd
) for different impellers.
Impeller N
p
(CFD) N
Qc
(CFD) N
Qc
(PIV) N
Qd
(CFD) N
Qd
(PIV) N
Qd
/N
p
(CFD)
m-PBT45 1.83 1.57 1.52 0.83
m-PBT40 1.36 1.47 1.6 1.44 1.54 1.06
m-PBT30 0.66 1.26 1.10 1.24 1.07 1.88
m-PBT20 0.27 0.92 0.88 0.91 0.84 3.41
PBT45-4 1.27 1.38 1.44 1.35 1.35 1.02
PBT45-3 1.03 1.25 1.23 1.19
The predicted and experimental axial velocity eld was
used for calculating the pumping number and circulation
number.
N
Qd
=
2

D,2
0
rU
z
dr
ND
3
(4)
N
Qc
=
2

Rr
0
rU
z
dr
ND
3
(5)
where R
r
represents the radial distance to vorticity center.
Table 1 shows that the predicted results are in agreement
with PIV data. It can be seen that as the blade angle increases,
the power number (N
p
), circulationnumber (N
Qc
) andpumping
number (N
Qd
) increase for the new axial impeller. This can be
explained by the mean velocity proles. However, the N
Qd
/N
p
ratio gradually decreases with increasing pitch angle from20
to 45. Fo rt (2011) developed a description of pitched blade
impeller as an axial pump and also found that the hydraulic
efciency of pitched blade impeller increases signicant with
decrease pitch angle. With the increase in number of blades
fromthree to four, N
p
, N
Qc
and N
Qd
increase while the N
Qd
/N
p
ratio decreases.
The N
Qd
/N
p
ratio of m-PBT20 is the highest. The value of
N
Qd
for m-PBT30 is close to that of PBT45-3, but the value of
N
p
for m-PBT30 is 36%less thanthat of PBT45-3. This indicates
that m-PBT30 has a higher efciency in pumping capacity.
5. Conclusion
In the present work, the experimental PIV data was used to
validate the simulated results obtained by the standard k
model. The simulation shows a good agreement with the PIV
experiment for the mean velocity, but underestimates turbu-
lent kinetic energy near the blade.
When comparing the hydrodynamics at equal Reynolds
numbers, the change of the blade shape with a pitched blade
turbine helps to increase the magnitude of radial velocity in
the impeller vicinity and axial velocity in the poorly mixed
center region of the tank. The change of the liquid facing
edge of blades contributes to the lower energy dissipationrate,
while the increase of the area of rear corner may improve
axial ow performance with the higher power consumption.
An increase in blade angle increase radial and axial velocity,
power number and pumping number, but decreases N
Qd
/N
p
ratio. Taking these factors into account, the optimized axial
ow impellers can be obtained.
In summary, the relationship among the impeller shapes,
the local details of ow elds and macroscopic performance,
is signicant, which is helpful to the impeller optimization.
Acknowledgements
This work was supported nancially by the National Basic
Research Program of China (2011CB606001), Program for
Changjiang Scholars and Innovative Research Team in the
University (IRT0942) and State Key Laboratory of Chemical
Engineering (SKL-ChE-13D01).
References
Ankamma, R.D., Sivashanmugam, P., 2010. Experimental and CFD
simulation studies on power consumption in mixing using
energy saving turbine agitator. J. Ind. Eng. Chem. 16,
157161.
Bugay, S., Escudie, R., Line, A., 2002. Experimental analysis of
hydrodynamics in axially agitated tank. AIChE J. 48,
463475.
Delafosse, A., Line, A., Morchain, J., Guiraud, P., 2008. LES and
URANS simulations of hydrodynamics in mixing tank:
comparison to PIV experiments. Chem. Eng. Res. Des. 86,
13221330.
Devals, C., Heniche, M., Takenaka, K., Tanguy, P.A., 2008. CFD
analysis of several design parameters affecting the
performance of the Maxblend impeller. Comput. Chem. Eng.
32, 18311841.
Fo rt, I., 2011. On hydraulic efciency of pitched blade impellers.
Chem. Eng. Res. Des. 89, 611615.
Gabriele, A., Nienow, A.W., Simmons, M.J.H., 2009. Use of angle
resolved PIV to estimate local specic energy dissipation rates
for up- and down-pumping pitched blade agitators in a stirred
tank. Chem. Eng. Sci. 64, 126143.
Harminder, S., David, F.F., Justin, J.N., 2011. An assessment of
different turbulence models for predicting ow in a bafed
tank stirred with a Rushton turbine. Chem. Eng. Sci. 66,
59765988.
Hartmann, H., Derksen, J.J., Montavon, C., Pearson, J., Hamill, I.S.,
van den Akker, H.E.A., 2004. Assessment of large eddy and
RANS stirred tank simulations by means of LDA. Chem. Eng.
Sci. 59, 24192432.
Joshi, J.B., Nere, N.K., Rane, C.V., Murthy, B.N., Mathpati, C.S.,
Patwardhan, A.W., Ranade, V.V., 2011. CFD simulation stirred
tanks: comparison of turbulence models. Part I: radial ow
impellers. Can. J. Chem. Eng. 89, 2382.
Khan, F.R., Rielly, C.D., Brown, D.A.R., 2006. Angle-resolved
stereo-PIV measurements close to a down-pumping
pitched-blade turbine. Chem. Eng. Sci. 61, 27992806.
Khopkar, A.R., Aubin, J., Xuereb, C., Le Sauze, N., Bertrand, J.,
Ranade, V.V., 2003. Gasliquid ow generated by a
pitched-blade turbine: particle image velocimetry
measurements and computational uid dynamics
simulations. Ind. Eng. Chem. Res. 42, 53185332.
Kumaresan, T., Joshi, J.B., 2006. Effect of impeller design on the
ow pattern and mixing in stirred tanks. Chem. Eng. J. 115,
173193.
Lee, K.C., Yianneskis, M., 1998. Turbulence properties of the
impeller streamof a Rushton turbine. AIChE J. 44, 1324.
Li, Z., Bao, Y., Gao, Z., 2011. PIV experiments and large eddy
simulations of single-loop ow elds in Rushton turbine
stirred tanks. Chem. Eng. Sci. 66, 12191231.
1036 chemical engineering research and design 9 2 ( 2 0 1 4 ) 10271036
Li, Z., Hu, M., Bao, Y., Gao, Z., 2012. Particle image velocimetry
experiments and large Eddy simulations of merging ow
characteristics in dual rushton turbine stirred tanks. Ind. Eng.
Chem. Res. 51, 24382450.
Mavros, P., 2001. Flow visualization in stirred vessels a review of
experimental techniques. Chem. Eng. Res. Des. 79, 113127.
Murthy, B.N., Joshi, J.B., 2008. Assessment of standard, RSM and
LES turbulence models in a bafed stirred vessel agitated by
various impeller designs. Chem. Eng. Sci. 63, 54685495.
Nienow, A.W., 1998. Hydrodynamics of stirred bioreactors. In:
Pohorecki, R. (Ed.), Fluid Mechanics Problems in
Biotechnology. App. Mech. Rev., vol. 51, pp. 332.
Pan, C.M., Min, J., Liu, X.H., Gao, Z.M., 2008. Investigation of uid
ow in a dual Rushton impeller stirred tank using particle
image velocimetry. Chin. J. Chem. Eng. 16, 693699.
Patwardhan, A.W., Joshi, J.B., 1999. Relation between ow pattern
and blending in stirred tanks. Ind. Eng. Chem. Res. 38,
31313143.
Ranade, V.V., Mishra, V.P., Saraph, V.S., Deshpande, G.B., Joshi, J.B.,
1992. Comparison of axial-ow impellers using a laser
doppler anemometer. Ind. Eng. Chem. Res. 31, 23702379.
Sahu, A.K., Kumar, P., Patwardhan, A.W., Joshi, J.B., 1999. CFD
modelling and mixing in stirred tanks. Chem. Eng. Sci. 54,
22852293.
Singh, H., Fletcher, D.F., Nijdam, J.J., 2011. An assessment of
different turbulence models for predicting ow in a bafed
tank stirred with a Rushton turbine. Chem. Eng. Sci. 66,
59765988.
Wu, B.X., 2012. Large eddy simulation of mechanical mixing in
anaerobic digesters. Biotechnol. Bioeng. 109, 804812.

Das könnte Ihnen auch gefallen