designs

Article
A Study of the Mixing Performance of Different
Impeller Designs in Stirred Vessels Using
Computational Fluid Dynamics
Ian Torotwa and Changying Ji *
College of Engineering, Nanjing Agricultural University Jiangsu Province, Nanjing 210031, China;
iantorotwa@live.com
* Correspondence: chyji@njau.edu.cn

Received: 8 January 2018; Accepted: 6 March 2018; Published: 8 March 2018

Abstract: Design and operation of mixing systems using agitated vessels is a difficult task due
to the challenge of obtaining accurate information on impeller-induced turbulence. The use of
Computational Fluid Dynamics (CFD) can provide detailed understanding of such systems. In this
study, experimental tests and computational fluid dynamics simulations were performed to examine
the flow characteristics of four impeller designs (anchor, saw-tooth, counter-flow and Rushton
turbine), in achieving solution homogeneity. The impellers were used to mix potassium sulfate
granules, from which values of electrical conductivity of the solution were measured and used to
estimate the distribution pattern of dissolved solid concentrations within the vessel. CFD models
were developed for similar mixing arrangement using commercial software, ANSYS Fluent 18.1 solver
and the standard k-epsilon (ε) turbulence model. The Multiple Reference Frame (MRF) approach was
used to simulate the impeller rotation. Velocity profiles generated from the simulations were in good
agreement with the experimental predictions, as well as with results from previous studies. It was
concluded that, through CFD analysis, detailed information can be obtained for optimal design of
mixing apparatus. These findings are relevant in choosing the best mixing equipment and provides a
basis for scaling up mixing operations in larger systems.

Keywords: impeller design; turbulent mixing; homogeneity; computational fluid dynamics; ANSYS
fluent; multiple reference frame; velocity profile

1. Introduction
Mixing is an essential operation in many engineering fields. It has central significance in
food processing, pharmaceutical production, chemical engineering, biotechnology, agri-chemical
preparations, paint manufacturing, water purification among countless other applications [1].
Many mixing schemes using stirred tanks have been developed to meet various production and
processing goals [2]. In agricultural systems, mixing set-ups have been used in a number of processes,
such as preparing farm chemical concentrations, balancing nutrient amounts in fertilizer tanks,
blending different substances and processing farm products.
One of the main aims of agitation systems using stirred vessels is to maintain balanced quantities of
substances in different phases based on concentration levels [3]. In cases where soluble solids are mixed,
stirrers are used to increase interaction between the particles and avoid uneven accumulation at one
point [4]. Flow streams in stirred vessels are known to be turbulent, chaotic and difficult to determine;
therefore attaining homogeneity in such mixing processes is a demanding task. Achieving uniform
concentrations of mixing products is paramount for efficient and economical use of the expensive
chemicals, fertilizers and other mixing agents in agricultural applications. A careful choice of
equipment that generates sufficient turbulence and flow in the mixing vessel is therefore necessary.

Designs 2018, 2, 10; doi:10.3390/designs2010010 www.mdpi.com/journal/designs

Designs 2018, 2, 10 2 of 16

The inconsistencies in mixing quality can be attributed to the lack of a clear understanding of the
mixing processes due to the complex nature of impeller-induced turbulence in agitated vessels [5,6].
There is a need for in-depth studies to determine mixing efficiency and accurately predict the overall
performance of these systems. It has been established that the performance of mixing processes is a
product of the mixing time, the type of impeller used, number of impeller blades, blade size, working
angular speeds, and vessel configurations [7].
In large-scale mixing plants, stirrers and the entire agitation set-up should be able to create faster
movement of substances and high turbulence. This makes the entire mixing process in large containers
complicated and impractical to study through experiments [8]. There is therefore a need to provide a
more workable method that will simplify the process.
Computational fluid dynamics (CFD) is instrumental in analyzing fluid flow systems using
numerical methods and simulations in computer-controlled programs [9]. With CFD, engineers can
easily study new and complex models in virtual environments, determine the design details, predict
possible sources of failure and optimize system operations. Simulations have been used to investigate
the hydrodynamics of many industrial processes and aeronautics among other numerous fields [10].
Furthermore, many researchers and industrialists have found the use of CFD helpful in reducing
the cost and time spent in creating large and complicated prototypes of trial systems. Using CFD
technique, satisfactory and crucial information can be obtained within a small working area and with
less effort.
In order to achieve excellent mixing results, CFD can be employed as a tool for gaining in-depth
knowledge of the turbulent dynamics of mixing operations in stirred vessels. Through numerical
simulations, the effects of diverse and irregular configurations of mixing tanks, impeller designs and
baffles can be modelled. The expected performance of using atypically large vessels [11] and under
extreme working conditions can also be projected. Some of the mixing parameters that have been
investigated using numerical methods in CFD include mixing times, power requirements, flow types,
and velocity patterns [12].
Since impeller design is the most important component for determining the performance of
mechanically agitated mixers [13], its design features and operational characteristics can be described
theoretically using CFD. Several researchers have performed CFD tests to examine the effects of
impeller type in mixing vessels. Cokljat et al. [14–16] did simulations of impeller flow behavior in
stirred tanks using CFD in order to study flow velocities and mixing time. Tatterson [17] emphasized
the importance of numerically modelling precise impeller flow characteristics.
The objective of this work was to use CFD technique alongside experimental tests to study the
mixing behavior of different impeller designs. This will add to the knowledge necessary for choosing
the best mixing designs. It was also expected that this work would give a justifiable basis for accurate
scale-up of mixing systems in industrial, field, and agricultural mixing operations.
In this study, four different types of impellers: anchor, counter-flow, saw-tooth, and Rushton
turbine impellers were designed and studied to determine how their distinct design features affected
flow characteristics in a stirred vessel. Experimental tests were performed to provide a comparative
reference to the simulation results. Velocity profiles generated in CFD were interpreted as the impeller
flow characteristics. Simulations were done using commercial software, ANSYS Fluent 18.1 solver.
The standard k-epsilon (ε) model was used to set up the turbulent flow process and the multiple
reference frames (MRF) approach used to model the impeller motion in a baffled tank.

2. Materials and Methods

Experiments were conducted in the engineering laboratory of Nanjing Agricultural University
in the month of August 2017, at room temperature and pressure. Four distinct, top-entering
agitation impeller designs were used to dissolve solid granules of potassium sulfate (K2 SO4 ) in
water. Electrical Conductivity (EC) measurements of the mixed solutions from different locations in
the liquid volume were used to predict the individual impeller performance.
Designs 2018, 2, 10 3 of 16

Designs 2018,
Designs 2, x2,FOR
2018, PEER
x FOR REVIEW
PEER REVIEW 3 of3 16
of 16
2.1. Experimental Set-Up
2.1.2.1.
Experimental Set-Up
Experimental Set-Up
A cylindrical tank with four baffles arranged symmetrically on the tank’s inner walls was designed
A cylindricaltank
A cylindrical tankwith
withfour
fourbaffles
baffles arranged
arranged symmetrically
symmetrically ononthe
thetank’s
tank’sinner
innerwalls
wallswaswas
for the experiment.
designed Figure 1 belowFigure
showsbelow
the experimental tank, with with
one baffle cutcutoff
offto
to show the
designed forfor
thethe experiment.
experiment. Figure 11below shows the experimental
shows experimentaltank,
tank, withone
one baffle
baffle cut off to
inner apparatus.
show
show thethe inner
inner apparatus.
apparatus.

Experimental
Figure 1. Figure apparatus
1. Experimental apparatusset-up showing
set-up showing the assembly
the assembly that was that was
used to used
perform theto perform the
mixing
mixing experiment.
experiment.
Figure 1. Experimental apparatus set-up showing the assembly that was used to perform the mixing
experiment.
The shaft holding the impellers had a diameter of 0.012 m and were positioned concentric to the
The shaft holding
axis of the tank.the impellers
Baffles had a in
were included diameter
the set-upof
to0.012 mthe
prevent and were
liquid frompositioned
spinning asconcentric
a single to the
The shaft holding the impellers had a diameter of 0.012 m and were positioned concentric to the
axis of the body. The outline of the experimental tank and the dimensions of the agitation components are
axistank.
of the Baffles werewere
tank. Baffles included ininthe
included theset-up
set-up to preventthethe
to prevent liquid
liquid fromfrom spinning
spinning as a single
as a single
shown in Figure 2 and Table 1 respectively.
body. The outline
body. of the experimental
The outline tank and
of the experimental tank the
anddimensions of the
the dimensions agitation
of the components
agitation components are
are shown
in Figure 2 and Table 1 respectively.
shown in Figure 2 and Table 1 respectively.

Figure 2. Diagram representing the experimental set-up dimensions.

2. Diagram
FigureFigure representing
2. Diagram representingthe
the experimental set-up
experimental set-up dimensions.
dimensions.

Table 1. Dimensions of the agitation apparatus.

Parameter Symbol Value (mm)

Tank diameter T 360
Tank height H 500
Impeller diameter D = T/2 180
Impeller blade height h 10
Baffle length L 440
Baffle width B = D/12 15
Impeller clearance C = D/3 60
 Tank height H 500
Impeller diameter D = T/2 180
Impeller blade height h 10
Baffle length L 440
Designs 2018, 2, 10
Baffle width B = D/12 15 4 of 16
Impeller clearance C = D/3 60

An
An electric
electric mixer
mixer (M12
(M12 ++7070cm,
cm,Wuyi
WuyiMingcheng
MingchengCo., Co.,Hengshui
HengshuiCity, City, China)
China) with
with aa capacity
capacity of
of
220
220 V,
V, 50
50 Hz,
Hz, 1050
1050 W Wand
andsix
sixadjustable
adjustablespeeds
speedsof ofbetween
between100–600
100–600rpm rpmwas wasused
usedto
torun
runthe
theimpellers.
impellers.
Water
Water was filled into the tank up to the 500 mm mark so that all baffles baffles were
were submerged.
submerged.
The
The four
four distinct
distinct designs
designs ofof impellers
impellers used
used forfor the
the experiment
experiment were were designed
designed inin Solid-works
Solid-works
2016 computeraided
2016 computer aideddesign
design (CAD)
(CAD) software
software and manufactured
and manufactured using Computer
using Computer NumericalNumerical
Machines
Machines
(CNC). They (CNC).
were They were
all open all open
left-hand left-hand
(LH) types and (LH)madetypes and made
of SUS304 of The
steel. SUS304
bladesteel. The blade
thickness of the
thickness
impellers of the2.5
was impellers
mm with was 2.5 mm with
a diameter a diameter
of 0.180 m (T/Dof = 0.180
0.5). m (T/D = 0.5).
The impellers were
The impellers wereplaced
placed 6060
mm mm above
above the tangential
the tangential line ofline
the of the bottom
liquid liquid bottom
surface. surface. A
A working
working angular of
angular velocity velocity
100 rpmof 100
wasrpm
usedwas used
for all thefor all the designs.
designs.
Figure
Figure 3 3below shows
below the design
shows details details
the design of the impellers
of the and the manufactured
impellers experimental
and the manufactured
designs.
experimental designs.

Figure 3.
Figure Experimentalimpeller
3. Experimental impellerdesigns
designs and
and thethe actual
actual impellers
impellers tested:
tested: (a) saw-tooth
(a) saw-tooth impeller
impeller (b)
(b) Rushton turbine (c) counter-flow impeller and (d) anchor impeller.
Rushton turbine (c) counter-flow impeller and (d.) anchor impeller

Three
Three hundred
hundred grams
grams of of potassium
potassium sulfate
sulfate granules
granules were
were dissolved
dissolved forfor aa period
period ofof two
two minutes
minutes
for each impeller construction. Eleven samples of the resultant solution were collected
for each impeller construction. Eleven samples of the resultant solution were collected at different at different
points
points as
as the
the solution
solution was
was drained
drained through
through the
the valve
valve atat the
the bottom
bottom of
of the
the tank.
tank. The
The points
points of of sample
sample
collection
collection were
were identified
identified when
when the
the solution
solution was
was atat an
an interval
interval of
of 5050 mm
mm (5 (5 cm)
cm) apart
apart along
along the
the
vertical
vertical cross-section
cross-section (i.e.,
(i.e., 00 mm,
mm, 50 mm, 100 mm .… . . 500
500 mm).
mm). ECEC values
values ofof the
the resultant
resultant sampled
sampled
solutions
solutions were
were measured
measured using an EC-meter. K22SO SO44 was
wasusedusedas
asthe
thetracer
tracerelement
elementin inthis
thisexperiment
experiment
because
because ititdissolves
dissolvesand
andionize
ionizereadily
readilyininwater
water[18].
[18].Figure 4 below
Figure 4 below shows
showsthethelaboratory
laboratory set-up of
set-up
of the experimental apparatus and the Electrical conductivity meter used to measure the electrical
conductivity values.
Designs 2018, 2, x FOR PEER REVIEW 5 of 16

the experimental apparatus and the Electrical conductivity meter used to measure the electrical
Designs 2018, 2, 10 5 of 16
conductivity values.

Figure 4.
Figure 4. (a)
(a) The
The laboratory
laboratory apparatus
apparatus set-up showing the mixer connected to a power power source, tank
tank
filled with
filled with water,
water,stand
standand
andaa shaft
shaft connecting
connectingthe
the motor
motor to
to the
the impeller
impeller inside
inside the
the tank.
tank. (b) Measuring
Measuring
electrical conductivity
electrical conductivity (EC)
(EC) using
using anan EC-meter.
EC-meter.

The samples tested represented the solution from the different specific regions of the tank and
The samples tested represented the solution from the different specific regions of the tank and
would be used to predict the concentration variability of the dissolved solids in the entire container.
would be used to predict the concentration variability of the dissolved solids in the entire container.

2.2. Numerical
2.2. Numerical Modeling
Modeling
Simulations of
Simulations of the
the mixing
mixingprocess
processwere weredone
done in in
CFD CFDusing
usingan ANSYS
an ANSYS Fluent 18.1 18.1
Fluent solver. The
solver.
design model used was an agitated baffled-cylindrical vessel of identical
The design model used was an agitated baffled-cylindrical vessel of identical dimensions to the dimensions to the one used
in the
one usedexperimental tests.
in the experimental tests.
The aim of turbulence simulation
The aim of turbulence simulation is toispredict the physical
to predict behaviorbehavior
the physical of turbulent flow generated
of turbulent flow
in a system using numerical methods. Turbulent motions in engineering
generated in a system using numerical methods. Turbulent motions in engineering applications applications are three-
dimensional,
are non-homogenous
three-dimensional, and non-isotropic.
non-homogenous The various
and non-isotropic. Themethods of simulating
various methods this behavior
of simulating this
behavior allows for the statistical description of variable flow fields using the post-process. A
allows for the statistical description of variable flow fields using the post-process. A modeling
modeling
method used
method used should
should ensure
ensure accuracy,
accuracy,simplicity
simplicityand andcomputational
computationalefficiency
efficiency[19].
[19].
A number of approaches have been employed for turbulent flow
A number of approaches have been employed for turbulent flow simulation in stirred simulation in stirred tanks.
tanks. In
In the
the case of CFD, the Reynolds-averaged Navier–Stokes (RANS) equations,
case of CFD, the Reynolds-averaged Navier–Stokes (RANS) equations, the Large Eddy Simulation the Large Eddy Simulation
(LES) and
(LES) and the
the Direct
Direct Numerical
Numerical Simulation
Simulation (DNS)
(DNS)arearethethethree
threemain
mainmethods
methodscommonly
commonlyused. used.
In the RANS technique, the equations are averaged over a time interval
In the RANS technique, the equations are averaged over a time interval or across a collection or across a collection of
equivalent fields. RANS computations are extensively used in practical
of equivalent fields. RANS computations are extensively used in practical computations for computations for predicting
steady-statesteady-state
predicting solutions. solutions.
AnisotropyAnisotropy
in the nature in the of nature
flow introduces a key uncertainty
of flow introduces in the
a key uncertainty
computation.
in the computation.
The Navier-Stokes
The Navier-Stokes equations
equations are
are used
used toto represent
represent the the characteristics
characteristics of of turbulence
turbulence and and form
form the
the
basis of describing the flow phenomena. The chaotic nature of turbulent fluxes
basis of describing the flow phenomena. The chaotic nature of turbulent fluxes act as a direct result of act as a direct result
of non-linear
non-linear terms
terms in the
in the N-SN-S equations.
equations. These
These equations
equations are are based
based on the
on the conservation
conservation lawslaws namely
namely the
the continuity,
continuity, momentum
momentum and and energy
energy conservation
conservation lawslaws as respectively
as respectively givengiven
belowbelow
[20]. [20].
𝜕𝜌
∂ρ + 𝛻. (𝜌𝑢) = 𝑜 (1)
𝜕𝑡 + ∇·(ρu) = o (1)
∂t
∂ρu
+ ∇·(ρuu) = −∇· P (2)
∂t
Designs 2018, 2, 10 6 of 16

∂ρe
+ ∇·(eu) = −∇(u· P) − ∇·q (3)
∂t
where u, ρ, e and q are the velocity components, density, total energy per unit volume, and heat
flux, respectively.
The stress tensor, P for a Newtonian fluid is defined by:

2 h i
P = p(ρ, T ) I + µ(∇·u) I − µ (∇u) + (∇v) T (4)
3

where, p(ρ, T ) is the scalar pressure, I, is a unit diagonal tensor, T is the temperature, and µ is the
dynamic viscosity coefficient.
Thus, the Navier-Stokes equation [21,22] can be given by:
  !
∂Ui ∂U ∂ P ∂ ∂U
+ Uj i = − + v i (5)
∂t ∂x j ∂xi ρ ∂x j ∂x j

The k-ε model is a two-equation method under the RANS approach, where, the turbulent kinetic
energy (TKE) and its dissipation rate (ε) are used to describe the unsteady fields. These two parameters
are obtained in the flow field by solving their modeled partial differential transport equations.
The standard k-ε model solves for high Reynolds number scenarios. This model is formulated
on the assumption that the Reynolds stress is proportional to the mean velocity gradient [23,24].
The constant of proportionality is taken to be the eddy viscosity, given as:

k2
vt = Cµ · (6)
ε
where k is the kinetic energy, ε is the dissipation rate, and Cµ is a parameter that depends on the k-ε
turbulence model.
The equation showing the TKE for three-dimensional flows can be represented as:

1  2 
k= · u + v2 + w2 (7)
2
The governing transport equations (k-equation and ε-equation) for the Standard k-ε model are
given below by Equations (8) and (9) respectively,
 
∂(ρk) ∂(ρui k) ∂(ρui k) µt ∂k
+ = · + ρ·( P − ε) (8)
∂t ∂xi ∂xi σk ∂xi
 
∂(ρε) ∂(ρui ε) ∂(ρui k) µt ∂ε 1
+ = · + ρ· ·(C1,ε P − C2,ε ε) (9)
∂t ∂xi ∂xi σε ∂xi τd
where τd is the dissipation rate time scale that characterizes the dynamic process in the energy spectrum
and P is the evolution of turbulence, represented respectively as,

k
τd = (10)
ε
!
∂ui ∂u j ∂u j
P = vt + (11)
∂x j ∂xi ∂xi

The values of empirical constants of the Standard k-ε model are Cµ = 0.09, σk = 1, σε = 1.314,
C1,ε = 1.44, and C2,ε = 1.92.
The standard k-ε model combines reasonable accuracy, time economy, and robustness for a wide
range of turbulent flows [25,26]. To improve the predictive accuracy of k-ε models, more transport
Designs 2018, 2, 10 7 of 16

equations have been derived. These include the realizable k-ε and the k-ε RNG (renormalization
group) methods.
The realizable k-ε model contains an additional state of eddy viscosity and a transport equation for
the dissipation derived from an exact equation for the transport of the mean square vorticity variations.
A disadvantage of this model is that it produces non-physical turbulent viscosities in the turbulent
viscosity equation. Thus, the use of this model is limited.
The k-ε RNG model is a method where the smallest eddies are first resolved in the inertial range
and then represented in terms of the next smallest eddies. This process continues until a modified
set of the Navier Stokes equations is obtained which can then be solved. This approach still poses
modeling problems of imperfectly solved eddies.
Generally, the main weakness associated with the RANS models is that it fails to predict
satisfactorily the explicit characteristics of complex flows, since the k-ε model assumes the isotropy of
turbulence. Anisotropic models such as Reynolds Stress Model (RSM), DNS and the LES model have
been applied in the simulation of complex three-dimensional flows.
RSM presents good accuracy in predicting flows with swirl, rotation and high strain rates.
It consists of six transport equations for the Reynolds stresses and an equation for the dissipation rate,
making it computationally cumbersome. This model also lacks universality in its parameters and it
does not adequately capture the time dependent nature of flow.
The DNS is based on a three-dimensional and unsteady solution of the Navier-Stokes equations.
However, the drawback of this model is its Reynolds number limitations, since the resolution of all
the fine scales of a high Reynolds number flow requires enormous computing capability. In addition,
it is hard to prove if it yields fully resolved eddies, because it would be impractical with inadequate
computing ability.
LES has arisen as a possible choice for modeling, where the time-dependent behavior of the flow
is resolved. It is based on the idea that the big eddies produced in the mean flow are anisotropic
and have a lengthy lifespan. On the other hand, the small eddies produced from inertial transfer
have more universal properties and are isotropic with a short life span hence relatively easy to
model. Equations describing this model are derived by filtering the Navier–Stokes equation [27].
This effectively separates the eddies whose scales are smaller than the filter size used in meshing.
The resulting equations have the structure of the original equation and resultant subgrid scale stresses
(SGS). The large eddies are resolved directly, while the small eddies are modelled using available
subgrid-scale models.
The filtered Navier–Stokes and conservation equation of LES for incompressible flows are as
shown below:
∂ui ∂u u 1 ∂p ∂2 u i ∂τij
+ i i =− +v − (12)
∂t ∂x j ρ ∂xi ∂x j ∂x j ∂x j
∂ui
=0 (13)
∂xi
where the items with bars indicate the large scales obtained from grid filtering.
The effects of the SGS are reflected in the subgrid scale stress tensor represented as:

τij = ui u j − ui u j (14)

The LES model can solve all eddying scales in a complex flow: however, the challenge of
limited computing power still prevails, and thus not suitable for practical industrial applications.
Moreover, there is excessive dissipation in flows produced by growth of initially small agitation to
fully turbulent flow which ought to be resolved [28].
For these reasons, it is deemed that the RANS equations for turbulence modeling are the most
fitting CFD tool to use for realistic and economical study of turbulent mixing schemes. The results
obtained from the simulations using this model were an estimate prediction of the expected behavior
Designs 2018, 2, 10 8 of 16

of the impellers under study and were not fully validated in the experiments due to the scale
of the computations and inadequacy of apparatus. To this effect, future work should employ
reliable experimental procedures such as the laser Doppler velocimetry (LDV) or particle image
Designs 2018, 2, x FOR PEER REVIEW 8 of 16
velocimetry (PIV).
2.2.1. Meshing
2.2.1. Meshing and
and Pre-Processing
Pre-Processing
The different
The different impeller
impeller configurations,
configurations, fluid volume, and
fluid volume, the baffles
and the baffles were
were modelled
modelled as as separate
separate
regions in Solid Works 2016 software, before being imported into ANSYS Fluent for
regions in Solid Works 2016 software, before being imported into ANSYS Fluent for pre-processing pre-processing
and meshing.
and meshing.Elaborate
Elaborateinterfaces
interfacesbetween
between thethe contacting
contacting fluid
fluid regions
regions and and boundary
boundary conditions
conditions were
were created in ANSYS workbench
created in ANSYS workbench design modeler.design modeler.
A mesh
A mesh was was generated
generated to to discretize
discretize the
the domains
domains into
into small
small control
control volumes,
volumes, where
where thethe
conservation equations
conservation equations were
weretotobe
beapproximated
approximatedbybycomputer
computer numerical
numerical calculations. TheThe
calculations. mesh for
mesh
the mixing simulation set-up contained two main zones, tank-fluid region, and impeller
for the mixing simulation set-up contained two main zones, tank-fluid region, and impeller region, region,
modelled as
modelled separate interacting
as separate interacting fluid domains. A
fluid domains. A fine
fine mesh
mesh was used to
was used to enhance
enhance the
the stability
stability and
and
accuracy of the computation
accuracy of the computation [9]. [9].
Figure 55 below
Figure below shows
shows the
the boundary
boundary interfaces
interfaces created
created and
and meshed
meshed regions
regions of
of the
the agitation
agitation
assembly. A
assembly. A compact
compact mesh
mesh can
can be
be seen
seen at
at the
the impeller
impeller and
and shaft
shaft region.
region.

(a) (b) (c)

Figure 5. (a) Boundary conditions between impeller and fluid interfaces set in ANSYS workbench
Figure 5. (a) Boundary conditions between impeller and fluid interfaces set in ANSYS workbench
design modeler;
design modeler; (b)
(b) Sectional
Sectional view
view of
of the
the meshed
meshed model
model showing
showing the
the impeller
impeller and
and shaft
shaft region
region at
at the
the
center; and
center; and (c)
(c) Meshing
Meshing the
the fluid
fluid volume.
volume.

2.2.2. Simulation Set up and Computation

2.2.2. Simulation Set up and Computation
The simulations were prepared in fluent solver, using the pressure-based steady state and
The simulations were prepared in fluent solver, using the pressure-based steady state and absolute
absolute velocity conditions with gravity acting in the negative y-axis direction. The created fluid
velocity conditions with gravity acting in the negative y-axis direction. The created fluid regions were
regions were then set to viscous type in the k-ε standard model with standard wall functions. The
then set to viscous type in the k-ε standard model with standard wall functions. The material was
material was chosen as water-liquid with a density, ρ, of 998.2 kg/m3 and constant viscosity, µ, of
chosen as water-liquid with a density, ρ, of 998.2 kg/m3 and constant viscosity, µ, of 0.001 kg/m.s.
0.001 kg/m.s. Cell zone conditions entailed the impeller-fluid interface, which consisted of the
Cell zone conditions entailed the impeller-fluid interface, which consisted of the impeller surface and
impeller surface and the fluid regions around the impeller. Mesh interfaces and contact regions were
the fluid regions around the impeller. Mesh interfaces and contact regions were confirmed to be the
confirmed to be the exact points where interactions occurred.
exact points where interactions occurred.
The movement of the impeller zone in the tank-fluid region was modeled using a Multiple
The movement of the impeller zone in the tank-fluid region was modeled using a Multiple
Reference Frame (MRF) approach that combines the computation of both stationary and moving
Reference Frame (MRF) approach that combines the computation of both stationary and moving
frames. The two zones consists of well-defined boundaries. The moving zone comprised of the
frames. The two zones consists of well-defined boundaries. The moving zone comprised of the
impeller and the shaft domains, rotating with an angular velocity of 600 rpm along the y-axis [29].
impeller and the shaft domains, rotating with an angular velocity of 600 rpm along the y-axis [29].
The tank-fluid zone together with the baffles and tank walls were set to the stationary frame [30].
The tank-fluid zone together with the baffles and tank walls were set to the stationary frame [30].
The simulation was configured using Hybrid initialization technique before running the
The simulation was configured using Hybrid initialization technique before running the
calculations with 200 iterations at a reporting interval of six and profile update of four cycles. The
calculations with 200 iterations at a reporting interval of six and profile update of four cycles.
calculations were ascertained to have converged for all the computations within 1–3 h of
computational time. Velocity profiles (contours, velocity streamlines and velocity vectors) were
finally generated in CFD-post process to represent the effects of each impeller type. The simulations
were executed using a 2.30 GHz, 4 GB RAM, (Lenovo TianYi 300-15ISK, Intel core i5) laptop computer
(Lenovo Group Ltd., Beijing, China).
Designs 2018, 2, 10 9 of 16

The calculations were ascertained to have converged for all the computations within 1–3 h of
computational time. Velocity profiles (contours, velocity streamlines and velocity vectors) were
finally generated in CFD-post process to represent the effects of each impeller type. The simulations
were executed using a 2.30 GHz, 4 GB RAM, (Lenovo TianYi 300-15ISK, Intel core i5) laptop computer
(Lenovo Group
Designs 2018, Ltd.,
2, x FOR Beijing,
PEER REVIEW China). 9 of 16

3. Results and Discussion

The values of electrical conductivity obtained through experiments
experiments were used to predict the
mixing performance of the different impellers. CFD results were compared to the experimental data
obtained on the distribution of the granules concentration
concentration in
in the
the tank.
tank.

3.1. Analysis of
3.1. Analysis of the
the Experimental
Experimental Concentration
Concentration Distribution
Distribution
Figure
Figure 66 below
below shows
shows aa graphical
graphical representation
representation of
of the
the distribution pattern of
distribution pattern of the
the EC
EC values of
values of
dissolved K SO in the solution. These values were assumed to be the amounts of solids at
dissolved K2SO4 in the solution. These values were assumed to be the amounts of solids at the specific
2 4 the specific
regions. The horizontal
regions. The horizontal axis
axis represents
represents the
the sample
sample point
point in
in the
the tank
tank from
from which the solution
which the sample
solution sample
was
was collected.
collected. The vertical scale
The vertical scale shows
shows the
the values
values of
of the measured EC
the measured EC at
at the
the specific
specific points. The lines
points. The lines
show the trend of concentrations in the tank from bottom
show the trend of concentrations in the tank from bottom to top. to top.

Figure 6. Graph comparing the performance of the different impellers under study (Anchor, Rushton
Figure 6. Graph comparing the performance of the different impellers under study (Anchor, Rushton
turbine, Saw-tooth
turbine, Saw-toothandandCounter-flow
Counter-flowimpellers).
impellers).The lines
The show
lines showthethe
trend of the
trend dissolved
of the solids
dissolved (EC
solids
values) within the vessel. This represents the amounts of solid granules that were broken
(EC values) within the vessel. This represents the amounts of solid granules that were broken downdown into
solution
into and distributed
solution by the
and distributed byturbulence generated
the turbulence by the
generated byimpeller action.
the impeller action.

From the graph, it can be observed that the saw-tooth impeller and the counter flow impellers
From the graph, it can be observed that the saw-tooth impeller and the counter flow impellers
performed better than the anchor and the Rushton turbine in breaking down the solids into solution.
performed better than the anchor and the Rushton turbine in breaking down the solids into solution.
The counter-flow and Rushton turbine impellers attained more uniform solution concentrations,
The counter-flow and Rushton turbine impellers attained more uniform solution concentrations,
though the mixing ability of the latter ranked lower than the former impeller.
though the mixing ability of the latter ranked lower than the former impeller.
The saw-tooth impeller was able to dissolve more granules and distribute them evenly within
The saw-tooth impeller was able to dissolve more granules and distribute them evenly within the
the solution as shown by the high values in the averages derived from the experiment (Table 2). The
solution as shown by the high values in the averages derived from the experiment (Table 2). The anchor
anchor impeller performed the least, both in the total amount of solids it dissolved and in level of
homogeneity, it attained.

Table 2. Mean values of the measured Electrical Conductivity (EC).

Type of Impeller Mean EC

saw tooth impeller 4113.5
Designs 2018, 2, 10 10 of 16

impeller performed the least, both in the total amount of solids it dissolved and in level of homogeneity,
it attained.

Table 2. Mean values of the measured Electrical Conductivity (EC).

Type of Impeller Mean EC

saw tooth impeller 4113.5
Rushton turbine 3571.4
Anchor impeller 3119.0
counter flow impeller 4101.9
Designs 2018, 2, x FOR PEER REVIEW 10 of 16

These
These results
results clearly
clearly point to a possibility
possibility ofof combining
combining different
different flow
flowpatterns
patternssuchsuchasasthe
the
counter-axial
counter-axialflow
flowand thethe
and slotsslots
effects of theofsaw-tooth
effects impellers.
the saw-tooth This willThis
impellers. increase
will mixing
increaseturbulence
mixing
and provideand
turbulence more particle
provide moreinteraction, which willwhich
particle interaction, improve
willthe mixingthe
improve performance of the impellers.
mixing performance of the
Commonly used impeller
impellers. Commonly usedtypes can thus
impeller typesbecan
redesigned to attain better
thus be redesigned standards.
to attain better standards.

3.2.
3.2.CFD
CFD Post-Process
Post-Process Analysis
Analysis
From
From the
the numerical
numerical simulation
simulation procedures, velocity contours,
procedures, velocity contours,streamlines
streamlinesand andvelocity
velocityvector
vector
representations
representationswere
weregenerated
generated inin CFD post process. The The results
resultsrepresented
representedthetheflow
flowbehavior
behaviorofofthe the
different mixing impellers and helped
different mixing impellers and helped in explaining the experimental outcome.
experimental outcome.
For
Forthe
the anchor
anchor impeller, the performance
performance around
aroundthetheimpeller
impellerregions
regionswas
wasasasininFigure
Figure7 7below.
below.
There
Therewas
wasmore
morecontact
contactbetween
betweenthe
thefluid
fluidand
andthe
theimpeller
impellerasas
the impeller
the cutcut
impeller through
through thethe
fluid. In In
fluid. the
velocity vector diagram it was observed that there was a zero-flow region generated behind
the velocity vector diagram it was observed that there was a zero-flow region generated behind the the blades
due to the
blades duerelatively wide blade
to the relatively widecross-sectional area of area
blade cross-sectional this impeller.
of this impeller.

Figure 7.
Figure 7. Diagrams
Diagrams showing
showing the
the (a)
(a) velocity
velocity vectors
vectorsand
and(b)
(b)contours
contoursgenerated
generatedby
byanchor
anchorimpeller.
impeller.

The velocity streamlines of this impeller in (Figure 8), shows that the blades drove the fluid to
The velocity streamlines of this impeller in (Figure 8), shows that the blades drove the fluid to
the walls, which then splashed back and was directed vertically and in opposite directions toward
the walls, which then splashed back and was directed vertically and in opposite directions toward
the center of the tank. There was more turbulence experienced at the impeller region, due to the large
the center of the tank. There was more turbulence experienced at the impeller region, due to the
blade area. This type of impeller was not very efficient in distributing the fluid all over the tank. The
large blade area. This type of impeller was not very efficient in distributing the fluid all over the tank.
generated flow is expected to cause higher concentrations of the solution in the lower parts of the
The generated flow is expected to cause higher concentrations of the solution in the lower parts of the
tank as compared to the upper regions. This observation clearly concurs with the experimental results
tank as compared to the upper regions. This observation clearly concurs with the experimental results
performed in this work as well as conclusion of Akiti, and Bai [31], who earlier proved that the anchor
performed in this work as well as conclusion of Akiti, and Bai [31], who earlier proved that the anchor
impeller produces less flow and turbulence regardless of the configuration of the mixing vessel.
impeller produces less flow and turbulence regardless of the configuration of the mixing vessel.
The counter-flow impeller design is based on the idea that the blades can be modeled to produce
axial flow in opposite directions. This is with the view that the counter flows generated will increase
turbulence and mixing performance. From the CFD post process results, it was observed that the
design was able to distribute the fluid towards the lower and upper parts of the tank efficiently.
 The velocity streamlines of this impeller in (Figure 8), shows that the blades drove the fluid to
the walls, which then splashed back and was directed vertically and in opposite directions toward
the center of the tank. There was more turbulence experienced at the impeller region, due to the large
blade area. This type of impeller was not very efficient in distributing the fluid all over the tank. The
Designs 2018, 2, 10 11 of 16
generated flow is expected to cause higher concentrations of the solution in the lower parts of the
tank as compared to the upper regions. This observation clearly concurs with the experimental results
performed
From in this
the velocity workcontour
vector as well as conclusion
(Figure of Akiti,
9), it can andthat
be seen Bai [31], who is
the fluid earlier
well proved that
projected the anchor
axially in both
impeller produces less flow and turbulence regardless of the configuration of the
directions. The vector diagram shows a relatively even distribution around the impeller region. mixing vessel.

Designs 2018, 2, x FOR PEER REVIEW 11 of 16

The 2018,
Designs counter-flow impeller
2, x FOR PEER REVIEWdesign is based on the idea that the blades can be modeled to11
produce
of 16
axial flow in opposite directions. This is with the view that the counter flows generated will increase
The counter-flow
turbulence and mixing impeller designFrom
performance. is based
theon the idea
CFD postthat the blades
process canitbewas
results, modeled to produce
observed that the
axial flow in opposite directions. This is with the view that the counter flows generated will increase
design was able to distribute the fluid towards the lower and upper parts of the tank efficiently. From
turbulence
Figure8. 8. and mixing
Velocity performance.
streamlines generatedFrom theanchor
bythe
the CFD post process results, it was observed thatatthe
theFigure
velocity Velocity
vector streamlines
contour generated
(Figure 9), it by
can beanchor impeller.
impeller.
seen that More
More
the fluid turbulence
isturbulence can
can
well projected bebeobserved
observed
axially at
in both
design
thethe was
lower
lower able to
regions
regions distribute
than
than the
the the
upper
upper fluid towards
regions
regions of
of the
the the lower
tank.
tank. and upper parts of the tank efficiently. From
directions. The vector diagram shows a relatively even distribution around the impeller region.
the velocity vector contour (Figure 9), it can be seen that the fluid is well projected axially in both
directions. The vector diagram shows a relatively even distribution around the impeller region.

Figure
Figure 9. 9.
(a)(a) Velocityvectors
Velocity vectorsand
and(b)(b)velocity
velocity contours
contours of
of counter-flow
counter-flowimpeller.
impeller.More
More particles areare
particles
Figurethe
seen 9. axial
(a) Velocity vectors as
and (b) are
velocity contours of counter-flow impeller. More particles are
seen onon
the axial bladesurface
blade surface they
as they projected
are projected vertically.
vertically.
seen on the axial blade surface as they are projected vertically.
The velocity streamlines (Figure 10), are observed to have dispersed to further regions of the
The velocity streamlines
The velocity (Figure
streamlines 10),10),
(Figure areare
observed to to
observed have dispersed
have dispersedtotofurther
furtherregions
regions of
of the
the tank.
tank. When mixing substances, this type of impeller will be able to efficiently distribute the particles
When tank. When mixing substances, this type of impeller will be able to efficiently distribute the particles
and mixing
create a substances, this type
more homogenous of impeller will be able to efficiently distribute the particles and
solution.
createand createhomogenous
a more a more homogenous solution.
solution.

Figure 10. Flow streamlines of counter-flow impeller.

Figure 10. Flow streamlines of counter-flow impeller.
Figure 10. Flow streamlines of counter-flow impeller.
The counter flow impeller was, thus expected to increase mixing performance due to the
The counter flow impeller was, thus expected to increase mixing performance due to the
counter-axial flow, which improved turbulence in the vessel. This outcome agrees well with the
counter-axial
experimental flow, which
findings improved
as well turbulence given
as the explanation in thebyvessel. This outcome
McDonough agrees
[32] on the well withofthe
characteristics
experimental findings as
the axial flow impellers.well as the explanation given by McDonough [32] on the characteristics of
the axial flow studies
Fewer impellers.
have been done to study the saw tooth impeller. This type of impeller uses a
Fewer studies have been done to study the saw tooth impeller. This type of impeller uses a
Designs 2018, 2, 10 12 of 16

The counter flow impeller was, thus expected to increase mixing performance due to the
counter-axial flow, which improved turbulence in the vessel. This outcome agrees well with the
experimental findings as well as the explanation given by McDonough [32] on the characteristics of
the axial flow impellers.
Fewer studies have been done to study the saw tooth impeller. This type of impeller uses a
circular disc with protruding slots on its edges, modified to act like saw blades that cut through the
fluid. Through experiments, this impeller was seen to have performed well in dissolving the solid
Designs
Designs 2018,
granules. 2,
2, xx FOR
2018,This PEER
is also
FOR PEER REVIEW
proven
REVIEW in CFD simulation as observed in Figure 11 below. The velocity contours 12
12 of
of 16
16
and vectors indicate that this impeller produced a uniform mixing result around the impeller regions
as
as seen
seen byby the evenly
the evenly distributed
distributed patterns.
evenly distributed patterns. The
The slots in this disc
disc are
are accredited
accredited to
to increasing
increasing contact
as seen by the patterns. The slots
slots in
in this
this disc are accredited to increasing contact
contact
area
area and
and inducing
inducing circular
circular flow
flow in
in the
the fluid.
fluid.
area and inducing circular flow in the fluid.
The
The velocity
velocity streamlines (Figure
(Figure 12)
12) shows
shows that
that there was great turbulence generated as
as the fluid
The velocity streamlines
streamlines (Figure 12) shows that there
there was
was great
great turbulence
turbulence generated
generated as the
the fluid
fluid
was caused
was caused
caused to to swirl
to swirl
swirl inin the
in the vessel.
the vessel. This type of impeller was, thus expected to distribute the fluid more
was vessel. This
This type
type of
of impeller
impeller was,
was, thus
thus expected
expected to
to distribute
distribute the
the fluid
fluid more
more
evenly
evenly throughout
throughout the the entire
entire volume.
volume. This
This validates the excellent performance of such a construction
evenly throughout the entire volume. This validates
validates the
the excellent
excellent performance
performance ofof such
such aa construction
construction
as established
as established through
established through experiments.
through experiments.
experiments.
as

Figure
Figure 11.
11. (a)
(a) Velocity vectors and (b) velocity contours of saw-tooth
saw-tooth impeller.
impeller. Uniform
Uniform patterns
patterns can
Figure 11. (a) Velocity vectors and
Velocity vectors and (b)
(b) velocity
velocity contours
contours of
of saw-tooth impeller. Uniform patterns can
can
be seen around the impeller.
impeller.
be seen around the impeller.

Figure
Figure 12.
12. Velocity
Velocity streamlines
Velocity streamlines of
of the
the saw-tooth
the saw-tooth impeller
saw-tooth impeller showing
impeller showing the
showing the swirling
the swirling motion
swirling motionof
motion ofthe
of theimpeller,
the impeller,
impeller,
which
which is
is spread
spread throughout
throughout the
the vessel.
vessel.

The
The Rushton
Rushton turbine
turbine isis one
one of
of the
the most
most studied
studied agitation
agitation impellers.
impellers. It
It has
has provided
provided the
the basis
basis for
for
comparative
comparative studies for many researchers [33]. The flow patterns around the impeller region were
studies for many researchers [33]. The flow patterns around the impeller region were
generally
generally well
well distributed
distributed due
due to
to its
its radial
radial action
action on
on the
the fluid
fluid (Figure
(Figure 13).
13). This
This impeller
impeller was
was able
able to
to
thrust the fluid from the center towards the walls of the tank and as the flow approached
thrust the fluid from the center towards the walls of the tank and as the flow approached the wall the wall
regions,
regions, the
the velocity
velocity was
was observed
observed to to steadily
steadily drop.
drop.
Designs 2018, 2, 10 13 of 16

thrust the fluid from the center towards the walls of the tank and as the flow approached the wall
Designs 2018, 2,2,xxFOR
FORPEER REVIEW 13
13 of
of 16
regions,
Designs the
2018, velocity was
PEER observed
REVIEWto steadily drop. 16

Figure
Figure 13.
13. (a)
(a) Velocity
Velocity vectors
vectors and
vectors and (b)
(b) velocity
(b) velocity contours
velocity contours of
contours of Rushton
Rushton turbine.
turbine. The
The radial
radial action
action of
of the
the
impeller on the fluid can be observed at the blades.
impeller on the fluid can be observed at the blades.

The
The streamlines were
wereobserved
observed(Figure
(Figure14)14)to
tomove
movetowards
towardsthethewall
wallofofthe stirred tank and then
The streamlines
streamlines were observed (Figure 14) to move towards the wall of the
the stirred
stirred tank
tank and
and then
then
split
splitinto
into upward
upward and
and downward
downward directions.
directions. The
The mixing
mixing was
was better
better distributed
distributed at
at the
the central
central parts
parts of
of
split into upward and downward directions. The mixing was better distributed at the central parts
the
the vessel
vessel as
as compared
compared to
to the
the outer
outer regions
regions as
as seen
seen from
from the
the vector
vector diagram.
diagram. The
The intensity
intensity of
of the
the
of the vessel as compared to the outer regions as seen from the vector diagram. The intensity of the
velocity
velocity currents
currents in
in the
the regions
regions below
belowthethe impeller
impeller was
was stronger
stronger than
than that
that above
above the
the impeller [34].
impeller[34].
[34].
velocity currents in the regions below the impeller was stronger than that above the impeller

Velocity streamlines of
Figure 14.Velocity
Figure of Rushton turbine.
turbine.
Figure14.
14. Velocitystreamlines
streamlines ofRushton
Rushton turbine.

From
Fromthe theobtained
obtainedresults,
results,itit was
wasevident
evidentthatthatthe theunconventional
unconventionalimpeller impellertypes
types(saw
(sawtoothtoothand
and
counter
counterflow flowimpellers)
impellers)exhibited
exhibitedbetter
bettermixing
mixingresults,
results, which
whichisisessential
essentialininachieving
achievingconcentration
concentration
homogeneity
homogeneityin
homogeneity ininagitation
agitation
agitation tanks.
tanks. Therefore,
tanks. to
Therefore,
Therefore, torealize the
thebest
to realize
realize results
the
best bestfrom
results mixing
results
from from
mixing processes, impellers
mixing processes,
processes, impellers
need
impellers
need to be
to beneedmodified
modified to
to be to generate
modified adequate
generatetoadequate energy
generate energy
adequate for creating
for energy turbulence
creatingforturbulence in the
creating turbulence vessel.
in the vessel. Optimum
in the vessel.
Optimum
designs
Optimum of mixing
designs impellers
of mixingcan be established
impellers can be by methods
established such
by as
methods
designs of mixing impellers can be established by methods such as turbulent diffusivity. This methodturbulent
such diffusivity.
as turbulent This method
diffusivity.
allows
This
allows for
for numerical
method allows for
numerical calculations
numericalof
calculations turbulent
turbulent mixing
calculations
of scenarios.
of turbulent
mixing mixing
scenarios. Turbulent diffusion
scenarios.
Turbulent isis defined
Turbulent
diffusion diffusion
defined by
by ais a
turbulent
defined bymass
turbulent mass diffusion
a turbulent mass
diffusion coefficient, D
Dt,t, which
whichDisis
diffusion coefficient,
coefficient, t, aa factor
which is aof
factor the
the turbulent
offactor of Schmidt
the turbulent
turbulent number,
Schmidt
Schmidt number,
number, SSctct,,
represented
S ct , represented
represented as:
as: as:
µt
S = 𝜇𝜇𝑡𝑡 (15)
𝑆𝑆𝑐𝑡ct == ρD t (15)
(15)
𝑐𝑡 𝜌𝐷
𝜌𝐷𝑡𝑡
where, µt is the turbulent viscosity.
where,
where, μμttisis thethe turbulent
turbulentviscosity.
viscosity.
The
The Schmidtnumber
Schmidt number determines
determinesthe the relative
relativedistribution
distribution of ofimpeller-induced
impeller-inducedmotion motionand andmass
mass
caused by turbulence.
caused by turbulence.
For
For efficient
efficient mixing
mixing processes,
processes, precise
precise prediction
prediction of of mixing
mixing behavior
behavior of of impellers
impellers in in stirred
stirred
vessels is necessary. This will be beneficial when designing or selecting
vessels is necessary. This will be beneficial when designing or selecting the most suitable apparatus. the most suitable apparatus.
The
The use use ofof the
the turbulence
turbulence models
models can can provide
provide aa good good basis
basis for
for detailed
detailed understanding
understanding of of generated
generated
Designs 2018, 2, 10 14 of 16

The Schmidt number determines the relative distribution of impeller-induced motion and mass
caused by turbulence.
For efficient mixing processes, precise prediction of mixing behavior of impellers in stirred vessels
is necessary. This will be beneficial when designing or selecting the most suitable apparatus. The use
of the turbulence models can provide a good basis for detailed understanding of generated flow in
mixing vessels. Nonetheless, the RANS, k-ε model used in this work is characterized by unsteadiness
because it assumes that the flow is statistically steady [35]. This assumption leads to poor estimates
for swirling and rotational flows. RANS model is valid only for fully turbulent flows, requires wall
function implementation, and leads to over-prediction of turbulence in highly strained flows.
More advanced methods such as DNS model can provide better simulation results.
However, they are expensive and demand immense computer capability in regards to both processor
speed and capacity for storing intermediate results than in RANS methods. A more realistic alternative
would be the large eddies simulation model, which has higher accuracy. It is much more economical
in terms of computational power requirements than DNS. It is applicable in resolving time-dependent
behavior of turbulent flow and allows computation of large and complex vortexes [36]. Like in
DNS, LES have high computational cost as compared to RANS model, and is affected by viscous
near-wall features.
The performance of the four impeller designs in this experiment were clearly captured by the
CFD simulation model and was in good agreement with all the experimental results and prior studies
conducted on the same. The good concurrence demonstrated in both the simulation and experimental
procedures is an evidence that CFD is a reliable tool for explaining turbulence in mixing systems.
It was also noted that numerical aspects play an important role in turbulent-flow computations in
terms of both accuracy and efficiency. The future significance of CFD turbulent mixing systems will be
guided by how accurate complex flows can be estimated.
For successful simulation and prediction of flow characteristics in turbulent systems,
a compromise between accuracy and computing demands ought to be made. Although it was
acknowledged that complex flow may not be predicted adequately with RANS models, the emphasis
on simplicity, practicality, computational speed and robustness in an industrial setting influences
against the adoption of more advanced models.
The ANSYS solver and numerical approach used in this work can therefore aid in the development
of optimal mixing schemes in regards to the equipment geometry and related parameters. It is expected
that improving the impeller designs from the conventional types will offer cheaper modes of agitation,
while minimizing costs and improving on the quality of the resultant products in agricultural chemical
preparations and many other fields.

4. Conclusions
The approach used in this study clearly depicted the mixing dynamics in mechanically stirred
tanks, which are often difficult to predict. This technique provided relevant information about the
turbulent flow characteristics. The flow patterns of the different impeller designs used in this work
were satisfactorily described through experiments and CFD analysis. The results obtained from CFD
simulations were confirmed to be in good agreement with experimental values. It was proven that
the impeller-blade configuration significantly affected the performance of a mechanically agitated
mixer. Achieving the best mixing designs will improve the quality of mixing and establish a good
degree of homogeneity, as a function of the design. This work will be useful in selecting the right
type of impellers that will guarantee optimum results and economical use of expensive chemicals and
other mixing agents. It will also provide a basis on which large mixing systems can be designed and
controlled using minimal costs, time and space.

Acknowledgments: Support from the Department of Agricultural Mechanization Engineering of Nanjing

Agricultural University facilitated the successful completion of this work.
Designs 2018, 2, 10 15 of 16

Author Contributions: Ian Torotwa conceived and designed the experiments. Changying Ji supervised the work
and facilitated the acquisition and installation of the apparatus. Ian Torotwa performed and analyzed the data
with extensive input from Changying Ji. Ian Torotwa wrote the paper and Changying Ji verified and approved it.
Conflicts of Interest: The authors declares no conflict of interest.

References
1. Zadghaffari, R.; Moghaddas, J.S.; Revstedt, J. A Study on Liquid-Liquid Mixing in a Stirred Tank with a
6-Blade Rushton Turbine. Iran. J. Chem. Eng. 2008, 5, 12–22.
2. Aubin, J.; Kresta, S.M.; Bertrand, J.; Xuereb, C.; Fletcher, D.F. Alternate Operating Methods for Improving
the Performance of a Continuous Stirred Tank Reactor. Chem. Eng. Res. Des. 2006, 84, 569–582. [CrossRef]
3. Tatterson, G.B. Fluid Mixing and Gas Dispersion in Agitated Tanks; McGraw-Hill: New York, NY, USA, 1991.
4. Mak, A.T.C. Solid-Liquid Mixing in Mechanically Agitated Vessels. Ph.D. Thesis, University of London,
London, UK, 1992.
5. Delvigne, F.; Destain, J.; Thonart, P. Structured Mixing Model for Stirred Bioreactors: An Extension to the
Stochastic Approach. Chem. Eng. J. 2005, 113, 1–12. [CrossRef]
6. Raju, R.; Balachandar, S.; Hill, D.F.; Adriana, R.J. Reynolds Number Scaling of Flow in a Stirred Tank with
Rushton Turbine. Part II—Eigen Decomposition of Fluctuation. Chem. Eng. Sci. 2005, 60, 3185–3198.
[CrossRef]
7. Isabela, M.P.; Leandro, S.O. CDF Modelling and Simulation of Transesterification Reactions of Vegetable Oils
with an Alcohol in Baffled Stirred Tank Reactor. Appl. Mech. Mater. 2013, 390, 86–90.
8. Seyed, H.; Dineshkumar, P.; Farhad, E.M.; Mehrab, M. Study of Solid-Liquid Mixing in Agitated Tanks
through Computational Fluid Dynamics Modeling. Ind. Eng. Chem. Res. 2010, 49, 4426–4435.
9. Versteeg, H.K.; Malalasekera, W. An Introduction to Computational Fluid Dynamics. In The Finite Volume
Method, 2nd ed.; Pearson Education: New York, NY, USA, 2007.
10. Zhang, Z.; Liu, H.; Zhu, S.P.; Zhao, F. Application of CFD in ship engineering design practice and ship
hydrodynamics. J. Hydrodyn. 2006, 18, 315–322. [CrossRef]
11. Zlokarnik, M. Scale-Up in Chemical Engineering, 2nd ed.; Wiley-VCH: Weinheim, Germany, 2006.
12. Pakzad, L.; Ein-Mozaffari, F.; Chan, P. Using computational fluid dynamics modeling to study the mixing of
pseudoplastic fluids with a Scaba 6SRGT impeller. Chem. Eng. Process. 2008, 47, 2218–2227. [CrossRef]
13. Ge, C.-Y.; Wang, J.-J.; Gu, X.-P.; Feng, L.-F. CFD simulation and PIV measurement of the flow field generated
by modified pitched blade turbine impellers. Chem. Eng. Res. Des. 2014, 92, 1027–1036. [CrossRef]
14. Cokljat, D.; Slack, M.; Vasquez, S.A.; Bakker, A.; Montante, G. Reynolds-Stress Model for Eulerian Multiphase.
Prog. Comput. Fluid Dyn. 2006, 6, 168–178. [CrossRef]
15. Khopkar, A.R.; Kasat, G.R.; Pandit, A.B.; Ranade, V.V. Computational Fluid Dynamics Simulation of the
Solid Suspension in a Stirred Slurry Reactor. Ind. Eng. Chem. Res. 2006, 45, 4416–4428. [CrossRef]
16. Zalc, J.M.; Szalai, E.S.; Alvarez, M.M.; Muzzio, F.J. Using CFD to understand chaotic mixing in laminar
stirred tanks. AIChE J. 2002, 48, 2124–2134. [CrossRef]
17. Tatterson, G.B. Scale-Up and Design of Industrial Mixing Processes; McGraw-Hill: New York, NY, USA, 1994.
18. Jiusheng, L.; Yibin, M.; Bei, L. Field evaluation of fertigation uniformity as affected by injector type and
manufacturing variability of emitters. Irrig. Sci. 2007, 25, 117–125.
19. Leschziner, M.A.; Drikakis, D. Turbulence and turbulent flow computation in aeronautics. Aeronaut. J. 2002,
106, 349–384.
20. ANSYS Fluent User Guide. Available online: http://www.ansys.fem.ir/ansys_fluent (accessed on 12
November 2017).
21. Paul, E.L.; Atiemo-Obeng, V.A.; Kresta, S.M. Handbook of Industrial Mixing: Science and Practice;
Wiley-Interscience: New York, NY, USA, 2004.
22. Dagadu, C.P.K.; Stegowsk, Z.; Furman, L.; Akaho, E.H.K.; Danso, K.A. Determination of Flow Structure in a
Gold Leaching Tank by CFD Simulation. J. Appl. Math. Phys. 2014, 2, 510–519. [CrossRef]
23. Launder, B.E.; Spalding, D.B. The Numerical Computation of Turbulent Flows. Comput. Meth. Appl.
Mech. Eng. 1974, 3, 269–289. [CrossRef]
24. Kresta, S.M.; Wood, P.E. Prediction of the Three-Dimensional Turbulent Flow in Stirred Tanks. AIChE J. 1991,
37, 448–460. [CrossRef]
Designs 2018, 2, 10 16 of 16

25. Barrue, H.; Bertrand, J.; Cristol, B.; Xuereb, C. Eulerian Simulation of Dense Solid-Liquid Suspension in
Multi-Stage Stirred Vessel. J. Chem. Eng. Jpn. 2001, 34, 585–594. [CrossRef]
26. Oshinowo, L.M.; Bakker, A. CFD Modeling of solids Suspensions in Stirred Tanks. Presented at the
Symposium on Computational Modeling of Metals, Minerals and Materials, TMS Annual Meeting, Seattle,
WA, USA, 17–21 February 2002.
27. Leonard, A. Energy cascade in large-eddy simulations of turbulent fluid flow. Adv. Geophys. 1974, 18,
237–248.
28. Pope, S. Turbulent Flows; Cambridge University Press: Cambridge, UK, 2000.
29. Luo, J.Y.; Issa, R.I.; Gosman, A.D. Prediction of Impeller Induced Flows in Mixing Vessels Using Multiple
Frames of Reference. Inst. Chem. Eng. Symp. Ser. 1994, 136, 549–556.
30. Khopkar, A.R.; Mavros, P.; Ranade, V.V.; Bertrand, J. Simulation of Flow Generated by an Axial Flow Impeller:
Batch and Continuous Operation. Chem. Eng. Res. Des. 2004, 82, 737–751. [CrossRef]
31. Akiti, O.; Yeboah, A.; Bai, G. Hydrodynamic effects on mixing and competitive reactions in laboratory
reactors. Chem. Eng. Sci. 2005, 60, 2341–2354. [CrossRef]
32. McDonough, R.J. Mixing for the Process Industries; Van Nostrand Reinhold: New York, NY, USA, 1992.
33. Musgrove, M.; Ruszkowski, S.; Van den Akker, H.; Derksen, J. Influence of impeller type and agitation
conditions on the drop size of immiscible liquid dispersions. In Proceedings of the 10th European Conference
on Mixing, Delft, The Netherlands, 2–5 July 2000; Elsevier Science: Amsterdam, The Netherlands, 2000;
pp. 165–172.
34. Divyamaan, W.; Ranjeet, P.U.; Moses, O.T.; Vishnu, K.P. CFD simulation of solid-liquid stirred tanks.
Adv. Powder Technol. 2012, 23, 445–453.
35. Drikakis, D. Advances in turbulent flow computations using high-resolution methods. Prog. Aerosp. Sci.
2003, 39, 405–424. [CrossRef]
36. Som, S.; Senecal, P.K.; Pomraning, E. Comparison of RANS and LES Turbulence Models against Constant
Volume Diesel Experiments. In Proceedings of the 24th Annual Conference on Liquid Atomization and
Spray Systems (ILASS Americas), San Antonio, TX, USA, 20–23 May 2012.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).