computational fluid dynamics (cfd) modelling of transfer chutes: a study of the influence of model...
TRANSCRIPT
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 19
Computational Fluid Dynamics (CFD) modelling of transfer
chutes A study of the in1047298uence of model parameters
Xiaoling Chen n Craig Wheeler
Centre for Bulk Solids and Particulate Technologies The University of Newcastle NSW 2308 Australia
H I G H L I G H T S
The Two Fluid Model can be used to simulate the large particlendashgas 1047298ow in transfer chutes
Experiments were conducted to verify simulation results Air velocities are very sensitive to the specularity coef 1047297cient values less than 01 In the zone close to the base of the chute particle velocities cannot be predicted by any of the combinations of parameters investigated Best modelling parameters were determined to describe this particular application
a r t i c l e i n f o
Article history
Received 22 January 2013
Received in revised form
8 March 2013
Accepted 17 March 2013Available online 27 March 2013
Keywords
Transfer chutes
Computational Fluid Dynamics (CFD)
Dust emission
Particle Image Velocimetry (PIV)
a b s t r a c t
Transfer chutes are essential components in almost all bulk material handling plants Belt conveyors
utilize transfer chutes to load and redirect bulk materials from one conveyor to another with their
reliability and performance being critical to the overall operation of the plant While reliability and
operational performance is typically measured by availability and throughput transfer chutes must also
operate with minimal environmental impact Consequently the design of transfer chutes in relation to
entrained air 1047298ow with the aim of minimizing fugitive dust emissions is an area of ongoing research and
investigation requiring the analysis of complex multiphase systems Computational Fluid Dynamics (CFD)
has been successfully applied to evaluate potential dust emissions from transfer chutes with theimplementation of appropriate models and modelling parameters shown to be critical to the overall
accuracy of the simulation results A notable shortcoming in the current research is the lack of systematic
guidelines available to appropriately select model parameters The aim of this paper is to offer guidance
in this regard and provide a better understanding of the in1047298uence of the maximum fractional packing
limit and evaluate the most appropriate model parameters for simulations including particlendashparticle
restitution coef 1047297cient values and solid slip conditions For this purpose the two-phase three-dimensional
EulerndashEuler model of commercial CFD software Fluent has been selected to model the granular and air
1047298ow in the transfer chute Air velocities were measured experimentally by Particle Image Velocimetry
(PIV) with the simulated velocity pro1047297les being in good overall agreement with the experimental data
amp 2013 Elsevier Ltd All rights reserved
1 Introduction
Belt conveying systems rely extensively on transfer chutes toload and redirect bulk material The performance of a transfer
chute has a signi1047297cant impact not only on the ef 1047297ciency of the belt
conveyor system but also on the level of fugitive dust emission
across the bulk material handling plant This paper presents the
application of Computational Fluid Dynamics (CFD) to analyse and
evaluate potential dust emissions from transfer chutes and focuses
on the selection of appropriate modelling parameters within the CFD
software
There are two approaches commonly used to model bulkgranular 1047298ows namely the Discrete Element Method (DEM) and
the Continuum Method DEM follows the principle of simulating
real particle motion involving collisions with walls and other
particles and linear motion between these collisions In contrast
the continuum method models the granular 1047298ow as a continuous
1047298ow stream rather than a large number of individual particles in
space The continuum method has the advantage of utilising
experimentally determined bulk material properties measured
by well-established test methods and standards Both methods
facilitate the analysis of the forces and moments acting on
different elements of the transfer chute however DEM has the
advantage of being able to visualize the bulk material 1047298ow and the
Contents lists available at SciVerse ScienceDirect
journal homepage wwwelseviercomlocateces
Chemical Engineering Science
0009-2509$- see front matter amp 2013 Elsevier Ltd All rights reserved
httpdxdoiorg101016jces201303032
n Corresponding author Tel+61 040339129
E-mail addresses cxling612126com XiaolingChenuoneduau (X Chen)
CraigWheelernewcastleeduau (C Wheeler)
Chemical Engineering Science 95 (2013) 194ndash202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 29
capability of analysing considerably more complex 1047298ow patterns
but does so at the expense of considerably longer analysis times
When considering fugitive dust emissions from transfer chutes
neither the continuum nor DEM can be used to predict the air
1047298ow For this application CFD shows good potential for simulating
the multiphase 1047298ow and has proven accurate in predicting the air
velocity and location of 1047298ow recirculation zones due to the effect
of the bulk material stream (McILVENNA and MOSSAD 2003
Witt et al 1999 Donohue et al 2009) Previous studies (Chenet al 2012 Goniva et al 2012) undertaken by the authors indi-
cate that two-phase CFD simulations correlate well with scale
model experimental tests indicating that this approach can be
used to effectively predict likely dust emissions from transfer
chutes
The selection of appropriate models and modelling parameters is
found to be critical for the successful simulation of multiphase 1047298ow
Considerable research has been published on this topic but most
articles are focused on circulating 1047298uidized beds and spouted beds
with only 1047297ne or small particles Du et al (2006) investigated the
in1047298uence of the frictional stress maximum packing and coef 1047297cient of
restitution of particles for CFD simulations of spouted beds
Almuttahar and Taghipour (2008) conducted a study to evaluate
the effect of modelling parameters including different drag models
wall restitution coef 1047297cient values and solid slip conditions To the
authorsrsquo knowledge there are no known guidelines available for
the selection of models and model parameters (such as values of the
restitution and specularity coef 1047297cients) to simulate coarse granular
1047298ow in transfer chutes The aim of this paper is to investigate
the in1047298uence of the maximum fractional packing limit and evaluate
the most appropriate model parameters for simulations including
particlendashparticle restitution coef 1047297cient values and solid slip
conditions
It is well know that experimental studies are required to
evaluate any multiphase CFD model Particle Image Velocimetry
(PIV) which is used to obtain instantaneous measurements and
related properties in 1047298uids was used to determine the air velocity
pro1047297le at the outlet of a number of scale model transfer chutes PIV
is widely used to measure 1047298uid 1047298ow in both wind and water
tunnel experiments Furthermore PIV has also been successfully
used to measure granular 1047298ow named ldquogranular PIV rdquo (Du et al
2006 Almuttahar and Taghipour 2008 Ansart et al 2009 Ansart
et al 2011) Recently Ansart et al (2009 2011) used PIV to
investigate both the free falling particle plume and the in1047298uence of
the hopper outlet on the dust plume generated during free falling
Like other laser based measurement methods such as Laser-Doppler
Anemometry (LDA) and Particle Tracker Velocimetry (PTV) PIV is
minimally invasive and fast enough to measure velocities in turbu-
lent 1047298ow Furthermore unlike point measurement techniques PIV
provides velocity measurements across a whole plane in the 1047298uid at
any instant
The main objective of the current work is to develop a two-
phase three-dimensional CFD model to simulate the granular andentrained air 1047298ow in a transfer chute using the commercial CFD
simulation software FLUENT The solid particle behaviour near the
wall was investigated using different slip conditions the in1047298uence
of the particlendashparticle restitution coef 1047297cient studied and the
effect of frictional viscosity investigated by altering the maximum
packing limit Since the air velocity is considered to be one of the
main factors in1047298uencing dust generation the air 1047298ow patterns and
magnitudes of the velocities around the outlet of transfer chute
are used as one of the main criterion to evaluate the performance
of different model parameters PIV experiments were conducted to
visualize and measure the air 1047298ow at the outlet of scale model
transfer chutes The validity and accuracy of the CFD simulations
for different models and parameters were evaluated by qualitative
and quantitative comparison with experimental results obtained
from PIV Predicted particle velocities were also compared with
theoretical results calculated using a continuum approach
2 Experimental work
21 Experimental setup and procedure
Scale model experiments were undertaken to compare the
performance of a number of different transfer chute designs The
experimental procedure involved scale model testing of a number
of different transfer chute designs in an enclosure Bulk material
was fed through each chute and the velocity of the dust exiting the
transfer chute measured using PIV
The experimental procedure involved containing the transfer
chutes in a large enclosure to capture the fugitive dust to ensure
the repeatability of each experiment and eliminate any danger
caused by exposure to the class 4 laser and at the same time
minimize interference of external light The enclosure measured
25 m wide 25 m long and 30 m high and is shown pictorially in
Fig 1 The laser head of the PIV system was mounted at the end of
the enclosure facing the chute outlet A high speed camera wasmounted to the side using a sliding panel to facilitate adjustment
while concealing the light
While a range of different transfer chute designs were tested
during the course of the experimental test program Fig 2 shows
two of the scale model chutes tested These chutes are labelled A
and B and include A ndash baseline case B ndash a chute with an inbuilt
restrictor plate in the vertical leg in addition to the lower removable
cover 1047297tted and lower openings closed
The test procedure involved the use of a forklift and hydraulic
kibble (not show in the 1047297gure) to load the bulk material into the
top of the vertical section that feeds the transfer chute The 1047298ow
rate of bulk material from the kibble was controlled by a slide gate
at the discharge point and remained in a 1047297xed position for all
tests with the average mass 1047298ow rate calculated from the total
time of discharge Once the material passes through the chute it is
directed onto a conveyor belt located along the length of the
enclosure that transports the bulk material into a storage drum in
preparation for the following test
The bulk material used in all tests was screened Iron Ore with a
particle size distribution of 59 in the range of 40ndash475 mm 28
in the range of 475ndash60 mm 11 in the range of 6ndash11 mm with
the remaining 2 sub 4 mm
Fig 1 PIV Experimental setup
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 195
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 39
22 Data acquisition
Particle velocity measurements of the entrained dust particles
exiting the chute were measured using a two-dimensional digital
PIV system The PIV system illuminates a particle-seeded 1047298ow 1047297eld
with two laser sheet pulses separated by a time delay and captures the
images with a charge-coupled device (CCD) camera The PIV system
used in this study is an Oxford Fire1047298y diode laser and a high
performance digital 12bit CCD camera with a resolution of 13921040 pixels The PIV system synchronizes the camera and laser to
obtain a pair of images separated by a small time delay (400 μs) Pairs
of images were obtained at a frequency of 10 Hz The camerarsquos 1047297eld of
view is set to capture a region about 180 250 mm To minimize the
random error at least three tests were run for each chute
PIV is an indirect measurement technique where the particle
velocity is determined instead of the 1047298uid velocity The selection
and concentration of the seeding particles used in PIV work is
critical to the success and quality of the measurements In this case
the iron ore dust itself served as ideal seeding particles Further-
more very often the seeding particles must be injected into the
1047298ow shortly before the gaseous medium enters the test section Inthe present work it was possible to adequately mix the dust and
the bulk material so the seeding particles 1047298owed through the
transfer chute with the bulk material ensuring a more accurate
velocity measurement of the entrained dust particles
Seeding particle size and concentration are responsible for the
particlersquos 1047298ow tracing quali1047297cations and optical characteristics
(Schmitt et al 1995) There are some compromises that need to
be considered when choosing the seeding particles The 1047297rst one is
the size of the seeding particle A smaller particle will more
faithfully follow the 1047298uid 1047298ow increasing measurement accuracy
while a larger particle will scatter more light increasing signal
strength and result in greater measurement precision It
is a compromise between a quick response of the tracer particles
in the 1047298uid requiring small diameters and a high signal-to-noise
ratio (SNR) of the particle images necessitating large diameters
The second compromise is the concentration of the seeding
particles More particles will provide a good correlation
signal and increase measurement accuracy while too many
particles will increase the background noise and absorb the light
resulting in measurement failure (Melling 1997 Meinhart et al
2000)
3 CFD simulation
There are two commonly used approaches to model gasndashsolid
1047298ow namely the discrete phase model (DPM) and the two-1047298uid
model (TFM) These approaches are represented as the Eulerndash
Lagrangian model and EulerndashEuler model respectively in FLUENT
The former model is suitable for dilute gasndashsolid 1047298ow and the
later for dense phase simulations being much more applicable to
complex multiphase 1047298ows For the TFM approach the different
phases are mathematically treated as interpenetrating continua
and the conservation equations have similar structure for each
phase The air is assumed to be incompressible and at a constant
density while the particles are assumed to be spherical and have a
uniform density Due to the complexity of the particle volume
fraction in the transfer chute problem a TFM approach was
adopted
31 Model equation
We propose in the current work to solve the governing equa-
tions of mass and momentum by means of a multiphase Eulerian
model incorporating the Kinetic Theory of Granular Flow (KTGF)
available in the software Fluent The energy conservation equation
was ignored as the 1047298ow is isothermal A brief summary of the model
equations are list below
Fig 2 Scale Model Transfer Chutes (a) Chute A (b) Chute B
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202196
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 49
311 Mass conservation equations
The conservation of mass for the gas and solid phase can be
written as
part
partt ethε g ρ g THORN thornnablasdotethε g ρ g v
g THORN frac14 0 eth1THORN
part
partt ethεs ρsTHORN thornnablasdotethεs ρs v
sTHORN frac14 0 eth2THORN
where ε ρ and v are the volume fraction the density and theinstantaneous velocity respectively The volume fraction con-
straint requires ε g thorn εs frac14 1
312 Momentum conservation equations
Assuming no mass transfer between phases and no lift and
virtual mass forces due to the large difference in their densities
only the drag force and gravity are considered in this study thus
the conservation of momentum for each phase can be expressed
as
part
partt ε g ρ g v
g
thorn nablasdot ε g ρ g v
g v
g
frac14nablasdotτ g minusε g nablaP thorn ε g ρ g g thorn K sg v
sminus v
g
eth3THORN
part
partt εs ρs v
s
thorn nablasdot εs ρs v
s v
s
frac14nablasdotτ sminusnablaP sminus εsnablaP thorn εs ρs g thorn K sg v
sminus v
g
eth4THORN
where τ is the stress tensor P is the pressure g is gravity and K sg is
the momentum exchange coef 1047297cient that represents the drag force
between the gas and the solid phase Details of these parameters
are explained below
The stress tensor for each phase is given by
τ g frac14 ε g μ g nabla v
g thorn nabla v
g
T
thorn ε g λ g minus2
3 μ g
nablasdot v
g I eth5THORN
τ s frac14 εs μs nabla vs thorn nabla vs T thorn εs λsminus
2
3 μs
nablasdot vsI eth6THORN
where λs and λ g are solid and gas viscosity respectively The
granular bulk viscosity λs represents the resistance of granular
particles to compression or expansion and is modelled by Lun
et al 1984 as
λs frac14 4
5εs ρsdseth1 thorn eTHORN
ffiffiffiffiffiffiΘs
π
r eth7THORN
where ds is the particle diameter e is the particlendashparticle restitu-
tion coef 1047297cient and Θs is the granular temperature
The granular pressure P s is derived from the kinetic theory of
granular 1047298ow and is composed of a kinetic term and a term due to
particle collisions In the current work the model proposed by Lun
et al 1984 is used and is given byP s frac14 εs ρsΘs thorn 2 ρseth1 thorn eTHORNε2
s g0sΘs eth8THORNwhere g 0s is the radial distribution function that modi1047297es the
probability of collisions between particles and is given by Lun et al
(1984) and expressed as
g 0s frac14 1minusεs
εsmax
13
minus1
eth9THORN
The in1047298uence of the drag model is not covered in this study
with the Gidaspow et al (1992) model selected for all work The
Gidaspow et al (1992) model is given by
K sg frac14 150ε2
g μ g
ε g d
2
s
thorn 175ε g ρ g j v
sminus v
g jd
s
ε g o08 eth10THORN
K sg frac14 3
4C D
ε g εsjvsminusv g jds
εminus265 g ε g ge08 eth11THORN
where C D
C D frac14 24
Res1 thorn 015ethResTHORN0687h i
Reso1000 eth12THORN
C D frac14 044 Res41000 eth13THORN
The Reynolds number of the particles is given by
Res frac14 ρ g j v
g minus v
sjds
μ g
eth14THORN
313 Granular shear viscosity
The solids stress tensor contains shear and bulk viscosities
arising from particle momentum exchange due to translation and
collision In other words the solids shear viscosity consists of a
collision term a kinetic term and a friction term and is given by
μs frac14 μscol thorn μskin thorn μsfric eth15THORNThe collision viscosity is taken from the kinetic theory of
granular 1047298ow by Gidaspow et al (1992)
μscol frac14 45ε2
s ρsds g 0seth1 thorn esTHORN ffiffiffiffiffiffiΘs
π
r eth16THORN
The kinetic viscosity is expressed by Gidaspow et al (1992) as
μskin frac14 10 ρsds
ffiffiffiffiffiffiffiffiπ Θs
p
96eth1 thorn esTHORN g 0s
1 thorn 4
5εs g 0seth1 thorn esTHORN
2
eth17THORN
The friction viscosity is expressed by Schaeffer (1987) as
μsfric frac14 P s sin ϕ
2 ffiffiffiffiffiffiffi
I 2D
p eth18THORN
where the constant ϕfrac14 300007 is the default angle of internal
friction value and I 2D is the second invariant of the deviatoric
stress tensor which can be written as
I 2D frac14 16 ethDs11minusDs22THORN2 thorn ethDs22minusDs33THORN2 thorn ethDs33minusDs11THORN2h i
thornD2s12 thorn D2
s23 thorn D2s31 eth19THORN
Dsij frac14 1
2
partusi
part x j
thorn partus j
part xi
eth20THORN
32 Boundary conditions and numerical procedure
At the inlet all velocities and volume fractions of both phases
were speci1047297ed The inlet velocity pro1047297le is assumed to be uniform
and the volume fraction constant The average solids velocity at
the inlet is calculated from the shape of the kibble and it is
assumed that the inlet air has the same velocity as the solid phase
At the outlet the pressure is assumed to be atmospheric while theremaining opening areas are de1047297ned as pressure inlets The
specularity coef 1047297cient is speci1047297cally used in the multiphase
analysis with granular 1047298ow to express the slip coef 1047297cient At the
wall the gas phase was set to the no-slip condition meaning the
air velocity is zero at the surface of the wall while different slip
conditions between the solid and the wall were conducted to
assess the effect of slip at the wall
In the present investigation the interaction of gas and solid
phases was represented by the well-known Gidaspow drag model
The Gidaspow model has been successfully applied in similar
applications such as spouted bed simulations (Lan et al 2012)
While the turbulence model selected for this study was the
standard kminusε model Unsteady simulations were performed with
a small initial time step of 1 10minus4
s while 20 iterations per time
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 197
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 59
step were used to ensure numerical stability A convergence
criterion of 110minus3 was speci1047297ed for the relative error between
two successive iterations The typical computational time for each
simulation was 5ndash10 days on a 267 GHz workstation
33 Grid independency
To con1047297rm that the CFD results are independent of the mesh
size simulations of Chute A with four different meshes were
performed A coarse mesh was created using a maximum face
size of 15 mm an intermediate mesh using a maximum face size of
12 mm a 1047297ne mesh using a maximum face size of 10 mm and a
1047297ner mesh using a maximum face size of 8 mm The air velocity is
an important characteristic parameter that can be used to deter-
mine how well numerical models capture the behaviour of the
transfer chute As shown in Fig 3 the 1047297ne and 1047297ner mesh cases
predicted similar outlet air velocities however considering the
minor variation in results and computational time a 1047297ne mesh
with a maximum face size of 10 mm was used in subsequent
simulations
4 Results and discussion
To provide guidance on the selection of values for the particlendash
particle restitution coef 1047297cient and the specularity coef 1047297cient when
simulating granular 1047298ow in transfer chutes an investigation into
the most appropriate model parameters was performed Chute Awas used to perform the parametric studies while Chute B was
used to verify the applicability of the chosen model parameters
41 In 1047298uence of the particle-wall slip condition
The specularity coef 1047297cient is a measure of the fraction of
collisions which transfer momentum to the wall and varies from
zero (smooth walls) to one (rough walls) The effect of the slip
condition between the solid phase and wall was investigated by
altering the specularity coef 1047297cient for the solid phase The free-slip
condition for the solid phase at the wall is usually obtained by
setting the shear condition to zero In total eight different
specularity coef 1047297cients were analysed to study the effect of the
particlendash
wall slip condition using a default particlendash
particle
restitution coef 1047297cient value of 09 The air velocities along the
centreline of the chute outlet for both the simulation and experi-
mental results are shown in Fig 4 with the specularity coef 1047297cient
and particlendashparticle restitution coef 1047297cient referenced as R and S
respectively There are no PIV measurements for the lower section
of the chute since the chute is 1047297tted with metal gussets in the
corners
Comparison between simulation results and experimental data
at the outlet of the scale model transfer chute showed signi1047297cant
variance with different specularity coef 1047297cients Differences mainly
occurred for specularity coef 1047297cients less than 01 with values less
than 001showing signi1047297cant differences both in the magnitude
and distribution from that of the experimentally measured values
From further inspection of Fig 4 the specularity coef 1047297cient values
less than 01 show the maximum air velocity position occurring
further away from the wall and increasing with an increase in the
value of the specularity coef 1047297cient Qualitatively speaking the
overall trends of the predicted air velocity pro1047297les are quite similar
for the data obtained from the specularity coef 1047297cients larger than
005 in spite of some quantitative differences In reference to the
experimental data it is clear the simulation results are lower
overall potentially resulting from a different particle 1047298ow depth
between the simulations and experiment
The specularity coef 1047297cient indirectly affects the air velocity by
affecting the particle behaviour The simulated particle velocity
pro1047297les and particle volume fractions at the outlet for the range of
specularity coef 1047297cients are illustrated in Fig 5(a) and (b) respec-
tively Due to dif 1047297culties in measuring the outlet particle velocitythis velocity was calculated using the classic continuum method of
Roberts (2003) An equivalent friction coef 1047297cient of 08 was used
based on the average measured value of the particle wall friction for
iron ore on PVC and steel Based on the continuum method
calculation a schematic drawing of particle velocity distribution is
given in Fig 5(a) for comparison to the other simulation results It is
noted that the continuum method calculation gives a single velocity
for the entire bulk material stream rather than a pro1047297le throughout
the stream thickness
As shown in Fig 5(a) the simulated particle velocity pro1047297les
indicate the particle velocities are gradually decreasing with an
increase in the specularity coef 1047297cient For specularity coef 1047297cient
values of 02 and 03 the results show similar velocity pro1047297le
predictions The simulated particle velocities sharply decrease to
00 10 20 30 40 50 60
000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
Coase Intermediate Fine Finer
Fig 3 Outlet air velocity pro1047297le of different mesh for Chute A
00 10 20 30 40 50 60000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
R=09Free slip R=09S=0005 R=09S=001 R=09S=003 R=09S=005 R=09S=01 R=09S=02 R=09S=03
Experiments
Fig 4 Outlet air velocities for different specularity coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202198
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 29
capability of analysing considerably more complex 1047298ow patterns
but does so at the expense of considerably longer analysis times
When considering fugitive dust emissions from transfer chutes
neither the continuum nor DEM can be used to predict the air
1047298ow For this application CFD shows good potential for simulating
the multiphase 1047298ow and has proven accurate in predicting the air
velocity and location of 1047298ow recirculation zones due to the effect
of the bulk material stream (McILVENNA and MOSSAD 2003
Witt et al 1999 Donohue et al 2009) Previous studies (Chenet al 2012 Goniva et al 2012) undertaken by the authors indi-
cate that two-phase CFD simulations correlate well with scale
model experimental tests indicating that this approach can be
used to effectively predict likely dust emissions from transfer
chutes
The selection of appropriate models and modelling parameters is
found to be critical for the successful simulation of multiphase 1047298ow
Considerable research has been published on this topic but most
articles are focused on circulating 1047298uidized beds and spouted beds
with only 1047297ne or small particles Du et al (2006) investigated the
in1047298uence of the frictional stress maximum packing and coef 1047297cient of
restitution of particles for CFD simulations of spouted beds
Almuttahar and Taghipour (2008) conducted a study to evaluate
the effect of modelling parameters including different drag models
wall restitution coef 1047297cient values and solid slip conditions To the
authorsrsquo knowledge there are no known guidelines available for
the selection of models and model parameters (such as values of the
restitution and specularity coef 1047297cients) to simulate coarse granular
1047298ow in transfer chutes The aim of this paper is to investigate
the in1047298uence of the maximum fractional packing limit and evaluate
the most appropriate model parameters for simulations including
particlendashparticle restitution coef 1047297cient values and solid slip
conditions
It is well know that experimental studies are required to
evaluate any multiphase CFD model Particle Image Velocimetry
(PIV) which is used to obtain instantaneous measurements and
related properties in 1047298uids was used to determine the air velocity
pro1047297le at the outlet of a number of scale model transfer chutes PIV
is widely used to measure 1047298uid 1047298ow in both wind and water
tunnel experiments Furthermore PIV has also been successfully
used to measure granular 1047298ow named ldquogranular PIV rdquo (Du et al
2006 Almuttahar and Taghipour 2008 Ansart et al 2009 Ansart
et al 2011) Recently Ansart et al (2009 2011) used PIV to
investigate both the free falling particle plume and the in1047298uence of
the hopper outlet on the dust plume generated during free falling
Like other laser based measurement methods such as Laser-Doppler
Anemometry (LDA) and Particle Tracker Velocimetry (PTV) PIV is
minimally invasive and fast enough to measure velocities in turbu-
lent 1047298ow Furthermore unlike point measurement techniques PIV
provides velocity measurements across a whole plane in the 1047298uid at
any instant
The main objective of the current work is to develop a two-
phase three-dimensional CFD model to simulate the granular andentrained air 1047298ow in a transfer chute using the commercial CFD
simulation software FLUENT The solid particle behaviour near the
wall was investigated using different slip conditions the in1047298uence
of the particlendashparticle restitution coef 1047297cient studied and the
effect of frictional viscosity investigated by altering the maximum
packing limit Since the air velocity is considered to be one of the
main factors in1047298uencing dust generation the air 1047298ow patterns and
magnitudes of the velocities around the outlet of transfer chute
are used as one of the main criterion to evaluate the performance
of different model parameters PIV experiments were conducted to
visualize and measure the air 1047298ow at the outlet of scale model
transfer chutes The validity and accuracy of the CFD simulations
for different models and parameters were evaluated by qualitative
and quantitative comparison with experimental results obtained
from PIV Predicted particle velocities were also compared with
theoretical results calculated using a continuum approach
2 Experimental work
21 Experimental setup and procedure
Scale model experiments were undertaken to compare the
performance of a number of different transfer chute designs The
experimental procedure involved scale model testing of a number
of different transfer chute designs in an enclosure Bulk material
was fed through each chute and the velocity of the dust exiting the
transfer chute measured using PIV
The experimental procedure involved containing the transfer
chutes in a large enclosure to capture the fugitive dust to ensure
the repeatability of each experiment and eliminate any danger
caused by exposure to the class 4 laser and at the same time
minimize interference of external light The enclosure measured
25 m wide 25 m long and 30 m high and is shown pictorially in
Fig 1 The laser head of the PIV system was mounted at the end of
the enclosure facing the chute outlet A high speed camera wasmounted to the side using a sliding panel to facilitate adjustment
while concealing the light
While a range of different transfer chute designs were tested
during the course of the experimental test program Fig 2 shows
two of the scale model chutes tested These chutes are labelled A
and B and include A ndash baseline case B ndash a chute with an inbuilt
restrictor plate in the vertical leg in addition to the lower removable
cover 1047297tted and lower openings closed
The test procedure involved the use of a forklift and hydraulic
kibble (not show in the 1047297gure) to load the bulk material into the
top of the vertical section that feeds the transfer chute The 1047298ow
rate of bulk material from the kibble was controlled by a slide gate
at the discharge point and remained in a 1047297xed position for all
tests with the average mass 1047298ow rate calculated from the total
time of discharge Once the material passes through the chute it is
directed onto a conveyor belt located along the length of the
enclosure that transports the bulk material into a storage drum in
preparation for the following test
The bulk material used in all tests was screened Iron Ore with a
particle size distribution of 59 in the range of 40ndash475 mm 28
in the range of 475ndash60 mm 11 in the range of 6ndash11 mm with
the remaining 2 sub 4 mm
Fig 1 PIV Experimental setup
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 195
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 39
22 Data acquisition
Particle velocity measurements of the entrained dust particles
exiting the chute were measured using a two-dimensional digital
PIV system The PIV system illuminates a particle-seeded 1047298ow 1047297eld
with two laser sheet pulses separated by a time delay and captures the
images with a charge-coupled device (CCD) camera The PIV system
used in this study is an Oxford Fire1047298y diode laser and a high
performance digital 12bit CCD camera with a resolution of 13921040 pixels The PIV system synchronizes the camera and laser to
obtain a pair of images separated by a small time delay (400 μs) Pairs
of images were obtained at a frequency of 10 Hz The camerarsquos 1047297eld of
view is set to capture a region about 180 250 mm To minimize the
random error at least three tests were run for each chute
PIV is an indirect measurement technique where the particle
velocity is determined instead of the 1047298uid velocity The selection
and concentration of the seeding particles used in PIV work is
critical to the success and quality of the measurements In this case
the iron ore dust itself served as ideal seeding particles Further-
more very often the seeding particles must be injected into the
1047298ow shortly before the gaseous medium enters the test section Inthe present work it was possible to adequately mix the dust and
the bulk material so the seeding particles 1047298owed through the
transfer chute with the bulk material ensuring a more accurate
velocity measurement of the entrained dust particles
Seeding particle size and concentration are responsible for the
particlersquos 1047298ow tracing quali1047297cations and optical characteristics
(Schmitt et al 1995) There are some compromises that need to
be considered when choosing the seeding particles The 1047297rst one is
the size of the seeding particle A smaller particle will more
faithfully follow the 1047298uid 1047298ow increasing measurement accuracy
while a larger particle will scatter more light increasing signal
strength and result in greater measurement precision It
is a compromise between a quick response of the tracer particles
in the 1047298uid requiring small diameters and a high signal-to-noise
ratio (SNR) of the particle images necessitating large diameters
The second compromise is the concentration of the seeding
particles More particles will provide a good correlation
signal and increase measurement accuracy while too many
particles will increase the background noise and absorb the light
resulting in measurement failure (Melling 1997 Meinhart et al
2000)
3 CFD simulation
There are two commonly used approaches to model gasndashsolid
1047298ow namely the discrete phase model (DPM) and the two-1047298uid
model (TFM) These approaches are represented as the Eulerndash
Lagrangian model and EulerndashEuler model respectively in FLUENT
The former model is suitable for dilute gasndashsolid 1047298ow and the
later for dense phase simulations being much more applicable to
complex multiphase 1047298ows For the TFM approach the different
phases are mathematically treated as interpenetrating continua
and the conservation equations have similar structure for each
phase The air is assumed to be incompressible and at a constant
density while the particles are assumed to be spherical and have a
uniform density Due to the complexity of the particle volume
fraction in the transfer chute problem a TFM approach was
adopted
31 Model equation
We propose in the current work to solve the governing equa-
tions of mass and momentum by means of a multiphase Eulerian
model incorporating the Kinetic Theory of Granular Flow (KTGF)
available in the software Fluent The energy conservation equation
was ignored as the 1047298ow is isothermal A brief summary of the model
equations are list below
Fig 2 Scale Model Transfer Chutes (a) Chute A (b) Chute B
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202196
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 49
311 Mass conservation equations
The conservation of mass for the gas and solid phase can be
written as
part
partt ethε g ρ g THORN thornnablasdotethε g ρ g v
g THORN frac14 0 eth1THORN
part
partt ethεs ρsTHORN thornnablasdotethεs ρs v
sTHORN frac14 0 eth2THORN
where ε ρ and v are the volume fraction the density and theinstantaneous velocity respectively The volume fraction con-
straint requires ε g thorn εs frac14 1
312 Momentum conservation equations
Assuming no mass transfer between phases and no lift and
virtual mass forces due to the large difference in their densities
only the drag force and gravity are considered in this study thus
the conservation of momentum for each phase can be expressed
as
part
partt ε g ρ g v
g
thorn nablasdot ε g ρ g v
g v
g
frac14nablasdotτ g minusε g nablaP thorn ε g ρ g g thorn K sg v
sminus v
g
eth3THORN
part
partt εs ρs v
s
thorn nablasdot εs ρs v
s v
s
frac14nablasdotτ sminusnablaP sminus εsnablaP thorn εs ρs g thorn K sg v
sminus v
g
eth4THORN
where τ is the stress tensor P is the pressure g is gravity and K sg is
the momentum exchange coef 1047297cient that represents the drag force
between the gas and the solid phase Details of these parameters
are explained below
The stress tensor for each phase is given by
τ g frac14 ε g μ g nabla v
g thorn nabla v
g
T
thorn ε g λ g minus2
3 μ g
nablasdot v
g I eth5THORN
τ s frac14 εs μs nabla vs thorn nabla vs T thorn εs λsminus
2
3 μs
nablasdot vsI eth6THORN
where λs and λ g are solid and gas viscosity respectively The
granular bulk viscosity λs represents the resistance of granular
particles to compression or expansion and is modelled by Lun
et al 1984 as
λs frac14 4
5εs ρsdseth1 thorn eTHORN
ffiffiffiffiffiffiΘs
π
r eth7THORN
where ds is the particle diameter e is the particlendashparticle restitu-
tion coef 1047297cient and Θs is the granular temperature
The granular pressure P s is derived from the kinetic theory of
granular 1047298ow and is composed of a kinetic term and a term due to
particle collisions In the current work the model proposed by Lun
et al 1984 is used and is given byP s frac14 εs ρsΘs thorn 2 ρseth1 thorn eTHORNε2
s g0sΘs eth8THORNwhere g 0s is the radial distribution function that modi1047297es the
probability of collisions between particles and is given by Lun et al
(1984) and expressed as
g 0s frac14 1minusεs
εsmax
13
minus1
eth9THORN
The in1047298uence of the drag model is not covered in this study
with the Gidaspow et al (1992) model selected for all work The
Gidaspow et al (1992) model is given by
K sg frac14 150ε2
g μ g
ε g d
2
s
thorn 175ε g ρ g j v
sminus v
g jd
s
ε g o08 eth10THORN
K sg frac14 3
4C D
ε g εsjvsminusv g jds
εminus265 g ε g ge08 eth11THORN
where C D
C D frac14 24
Res1 thorn 015ethResTHORN0687h i
Reso1000 eth12THORN
C D frac14 044 Res41000 eth13THORN
The Reynolds number of the particles is given by
Res frac14 ρ g j v
g minus v
sjds
μ g
eth14THORN
313 Granular shear viscosity
The solids stress tensor contains shear and bulk viscosities
arising from particle momentum exchange due to translation and
collision In other words the solids shear viscosity consists of a
collision term a kinetic term and a friction term and is given by
μs frac14 μscol thorn μskin thorn μsfric eth15THORNThe collision viscosity is taken from the kinetic theory of
granular 1047298ow by Gidaspow et al (1992)
μscol frac14 45ε2
s ρsds g 0seth1 thorn esTHORN ffiffiffiffiffiffiΘs
π
r eth16THORN
The kinetic viscosity is expressed by Gidaspow et al (1992) as
μskin frac14 10 ρsds
ffiffiffiffiffiffiffiffiπ Θs
p
96eth1 thorn esTHORN g 0s
1 thorn 4
5εs g 0seth1 thorn esTHORN
2
eth17THORN
The friction viscosity is expressed by Schaeffer (1987) as
μsfric frac14 P s sin ϕ
2 ffiffiffiffiffiffiffi
I 2D
p eth18THORN
where the constant ϕfrac14 300007 is the default angle of internal
friction value and I 2D is the second invariant of the deviatoric
stress tensor which can be written as
I 2D frac14 16 ethDs11minusDs22THORN2 thorn ethDs22minusDs33THORN2 thorn ethDs33minusDs11THORN2h i
thornD2s12 thorn D2
s23 thorn D2s31 eth19THORN
Dsij frac14 1
2
partusi
part x j
thorn partus j
part xi
eth20THORN
32 Boundary conditions and numerical procedure
At the inlet all velocities and volume fractions of both phases
were speci1047297ed The inlet velocity pro1047297le is assumed to be uniform
and the volume fraction constant The average solids velocity at
the inlet is calculated from the shape of the kibble and it is
assumed that the inlet air has the same velocity as the solid phase
At the outlet the pressure is assumed to be atmospheric while theremaining opening areas are de1047297ned as pressure inlets The
specularity coef 1047297cient is speci1047297cally used in the multiphase
analysis with granular 1047298ow to express the slip coef 1047297cient At the
wall the gas phase was set to the no-slip condition meaning the
air velocity is zero at the surface of the wall while different slip
conditions between the solid and the wall were conducted to
assess the effect of slip at the wall
In the present investigation the interaction of gas and solid
phases was represented by the well-known Gidaspow drag model
The Gidaspow model has been successfully applied in similar
applications such as spouted bed simulations (Lan et al 2012)
While the turbulence model selected for this study was the
standard kminusε model Unsteady simulations were performed with
a small initial time step of 1 10minus4
s while 20 iterations per time
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 197
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 59
step were used to ensure numerical stability A convergence
criterion of 110minus3 was speci1047297ed for the relative error between
two successive iterations The typical computational time for each
simulation was 5ndash10 days on a 267 GHz workstation
33 Grid independency
To con1047297rm that the CFD results are independent of the mesh
size simulations of Chute A with four different meshes were
performed A coarse mesh was created using a maximum face
size of 15 mm an intermediate mesh using a maximum face size of
12 mm a 1047297ne mesh using a maximum face size of 10 mm and a
1047297ner mesh using a maximum face size of 8 mm The air velocity is
an important characteristic parameter that can be used to deter-
mine how well numerical models capture the behaviour of the
transfer chute As shown in Fig 3 the 1047297ne and 1047297ner mesh cases
predicted similar outlet air velocities however considering the
minor variation in results and computational time a 1047297ne mesh
with a maximum face size of 10 mm was used in subsequent
simulations
4 Results and discussion
To provide guidance on the selection of values for the particlendash
particle restitution coef 1047297cient and the specularity coef 1047297cient when
simulating granular 1047298ow in transfer chutes an investigation into
the most appropriate model parameters was performed Chute Awas used to perform the parametric studies while Chute B was
used to verify the applicability of the chosen model parameters
41 In 1047298uence of the particle-wall slip condition
The specularity coef 1047297cient is a measure of the fraction of
collisions which transfer momentum to the wall and varies from
zero (smooth walls) to one (rough walls) The effect of the slip
condition between the solid phase and wall was investigated by
altering the specularity coef 1047297cient for the solid phase The free-slip
condition for the solid phase at the wall is usually obtained by
setting the shear condition to zero In total eight different
specularity coef 1047297cients were analysed to study the effect of the
particlendash
wall slip condition using a default particlendash
particle
restitution coef 1047297cient value of 09 The air velocities along the
centreline of the chute outlet for both the simulation and experi-
mental results are shown in Fig 4 with the specularity coef 1047297cient
and particlendashparticle restitution coef 1047297cient referenced as R and S
respectively There are no PIV measurements for the lower section
of the chute since the chute is 1047297tted with metal gussets in the
corners
Comparison between simulation results and experimental data
at the outlet of the scale model transfer chute showed signi1047297cant
variance with different specularity coef 1047297cients Differences mainly
occurred for specularity coef 1047297cients less than 01 with values less
than 001showing signi1047297cant differences both in the magnitude
and distribution from that of the experimentally measured values
From further inspection of Fig 4 the specularity coef 1047297cient values
less than 01 show the maximum air velocity position occurring
further away from the wall and increasing with an increase in the
value of the specularity coef 1047297cient Qualitatively speaking the
overall trends of the predicted air velocity pro1047297les are quite similar
for the data obtained from the specularity coef 1047297cients larger than
005 in spite of some quantitative differences In reference to the
experimental data it is clear the simulation results are lower
overall potentially resulting from a different particle 1047298ow depth
between the simulations and experiment
The specularity coef 1047297cient indirectly affects the air velocity by
affecting the particle behaviour The simulated particle velocity
pro1047297les and particle volume fractions at the outlet for the range of
specularity coef 1047297cients are illustrated in Fig 5(a) and (b) respec-
tively Due to dif 1047297culties in measuring the outlet particle velocitythis velocity was calculated using the classic continuum method of
Roberts (2003) An equivalent friction coef 1047297cient of 08 was used
based on the average measured value of the particle wall friction for
iron ore on PVC and steel Based on the continuum method
calculation a schematic drawing of particle velocity distribution is
given in Fig 5(a) for comparison to the other simulation results It is
noted that the continuum method calculation gives a single velocity
for the entire bulk material stream rather than a pro1047297le throughout
the stream thickness
As shown in Fig 5(a) the simulated particle velocity pro1047297les
indicate the particle velocities are gradually decreasing with an
increase in the specularity coef 1047297cient For specularity coef 1047297cient
values of 02 and 03 the results show similar velocity pro1047297le
predictions The simulated particle velocities sharply decrease to
00 10 20 30 40 50 60
000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
Coase Intermediate Fine Finer
Fig 3 Outlet air velocity pro1047297le of different mesh for Chute A
00 10 20 30 40 50 60000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
R=09Free slip R=09S=0005 R=09S=001 R=09S=003 R=09S=005 R=09S=01 R=09S=02 R=09S=03
Experiments
Fig 4 Outlet air velocities for different specularity coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202198
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 39
22 Data acquisition
Particle velocity measurements of the entrained dust particles
exiting the chute were measured using a two-dimensional digital
PIV system The PIV system illuminates a particle-seeded 1047298ow 1047297eld
with two laser sheet pulses separated by a time delay and captures the
images with a charge-coupled device (CCD) camera The PIV system
used in this study is an Oxford Fire1047298y diode laser and a high
performance digital 12bit CCD camera with a resolution of 13921040 pixels The PIV system synchronizes the camera and laser to
obtain a pair of images separated by a small time delay (400 μs) Pairs
of images were obtained at a frequency of 10 Hz The camerarsquos 1047297eld of
view is set to capture a region about 180 250 mm To minimize the
random error at least three tests were run for each chute
PIV is an indirect measurement technique where the particle
velocity is determined instead of the 1047298uid velocity The selection
and concentration of the seeding particles used in PIV work is
critical to the success and quality of the measurements In this case
the iron ore dust itself served as ideal seeding particles Further-
more very often the seeding particles must be injected into the
1047298ow shortly before the gaseous medium enters the test section Inthe present work it was possible to adequately mix the dust and
the bulk material so the seeding particles 1047298owed through the
transfer chute with the bulk material ensuring a more accurate
velocity measurement of the entrained dust particles
Seeding particle size and concentration are responsible for the
particlersquos 1047298ow tracing quali1047297cations and optical characteristics
(Schmitt et al 1995) There are some compromises that need to
be considered when choosing the seeding particles The 1047297rst one is
the size of the seeding particle A smaller particle will more
faithfully follow the 1047298uid 1047298ow increasing measurement accuracy
while a larger particle will scatter more light increasing signal
strength and result in greater measurement precision It
is a compromise between a quick response of the tracer particles
in the 1047298uid requiring small diameters and a high signal-to-noise
ratio (SNR) of the particle images necessitating large diameters
The second compromise is the concentration of the seeding
particles More particles will provide a good correlation
signal and increase measurement accuracy while too many
particles will increase the background noise and absorb the light
resulting in measurement failure (Melling 1997 Meinhart et al
2000)
3 CFD simulation
There are two commonly used approaches to model gasndashsolid
1047298ow namely the discrete phase model (DPM) and the two-1047298uid
model (TFM) These approaches are represented as the Eulerndash
Lagrangian model and EulerndashEuler model respectively in FLUENT
The former model is suitable for dilute gasndashsolid 1047298ow and the
later for dense phase simulations being much more applicable to
complex multiphase 1047298ows For the TFM approach the different
phases are mathematically treated as interpenetrating continua
and the conservation equations have similar structure for each
phase The air is assumed to be incompressible and at a constant
density while the particles are assumed to be spherical and have a
uniform density Due to the complexity of the particle volume
fraction in the transfer chute problem a TFM approach was
adopted
31 Model equation
We propose in the current work to solve the governing equa-
tions of mass and momentum by means of a multiphase Eulerian
model incorporating the Kinetic Theory of Granular Flow (KTGF)
available in the software Fluent The energy conservation equation
was ignored as the 1047298ow is isothermal A brief summary of the model
equations are list below
Fig 2 Scale Model Transfer Chutes (a) Chute A (b) Chute B
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202196
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 49
311 Mass conservation equations
The conservation of mass for the gas and solid phase can be
written as
part
partt ethε g ρ g THORN thornnablasdotethε g ρ g v
g THORN frac14 0 eth1THORN
part
partt ethεs ρsTHORN thornnablasdotethεs ρs v
sTHORN frac14 0 eth2THORN
where ε ρ and v are the volume fraction the density and theinstantaneous velocity respectively The volume fraction con-
straint requires ε g thorn εs frac14 1
312 Momentum conservation equations
Assuming no mass transfer between phases and no lift and
virtual mass forces due to the large difference in their densities
only the drag force and gravity are considered in this study thus
the conservation of momentum for each phase can be expressed
as
part
partt ε g ρ g v
g
thorn nablasdot ε g ρ g v
g v
g
frac14nablasdotτ g minusε g nablaP thorn ε g ρ g g thorn K sg v
sminus v
g
eth3THORN
part
partt εs ρs v
s
thorn nablasdot εs ρs v
s v
s
frac14nablasdotτ sminusnablaP sminus εsnablaP thorn εs ρs g thorn K sg v
sminus v
g
eth4THORN
where τ is the stress tensor P is the pressure g is gravity and K sg is
the momentum exchange coef 1047297cient that represents the drag force
between the gas and the solid phase Details of these parameters
are explained below
The stress tensor for each phase is given by
τ g frac14 ε g μ g nabla v
g thorn nabla v
g
T
thorn ε g λ g minus2
3 μ g
nablasdot v
g I eth5THORN
τ s frac14 εs μs nabla vs thorn nabla vs T thorn εs λsminus
2
3 μs
nablasdot vsI eth6THORN
where λs and λ g are solid and gas viscosity respectively The
granular bulk viscosity λs represents the resistance of granular
particles to compression or expansion and is modelled by Lun
et al 1984 as
λs frac14 4
5εs ρsdseth1 thorn eTHORN
ffiffiffiffiffiffiΘs
π
r eth7THORN
where ds is the particle diameter e is the particlendashparticle restitu-
tion coef 1047297cient and Θs is the granular temperature
The granular pressure P s is derived from the kinetic theory of
granular 1047298ow and is composed of a kinetic term and a term due to
particle collisions In the current work the model proposed by Lun
et al 1984 is used and is given byP s frac14 εs ρsΘs thorn 2 ρseth1 thorn eTHORNε2
s g0sΘs eth8THORNwhere g 0s is the radial distribution function that modi1047297es the
probability of collisions between particles and is given by Lun et al
(1984) and expressed as
g 0s frac14 1minusεs
εsmax
13
minus1
eth9THORN
The in1047298uence of the drag model is not covered in this study
with the Gidaspow et al (1992) model selected for all work The
Gidaspow et al (1992) model is given by
K sg frac14 150ε2
g μ g
ε g d
2
s
thorn 175ε g ρ g j v
sminus v
g jd
s
ε g o08 eth10THORN
K sg frac14 3
4C D
ε g εsjvsminusv g jds
εminus265 g ε g ge08 eth11THORN
where C D
C D frac14 24
Res1 thorn 015ethResTHORN0687h i
Reso1000 eth12THORN
C D frac14 044 Res41000 eth13THORN
The Reynolds number of the particles is given by
Res frac14 ρ g j v
g minus v
sjds
μ g
eth14THORN
313 Granular shear viscosity
The solids stress tensor contains shear and bulk viscosities
arising from particle momentum exchange due to translation and
collision In other words the solids shear viscosity consists of a
collision term a kinetic term and a friction term and is given by
μs frac14 μscol thorn μskin thorn μsfric eth15THORNThe collision viscosity is taken from the kinetic theory of
granular 1047298ow by Gidaspow et al (1992)
μscol frac14 45ε2
s ρsds g 0seth1 thorn esTHORN ffiffiffiffiffiffiΘs
π
r eth16THORN
The kinetic viscosity is expressed by Gidaspow et al (1992) as
μskin frac14 10 ρsds
ffiffiffiffiffiffiffiffiπ Θs
p
96eth1 thorn esTHORN g 0s
1 thorn 4
5εs g 0seth1 thorn esTHORN
2
eth17THORN
The friction viscosity is expressed by Schaeffer (1987) as
μsfric frac14 P s sin ϕ
2 ffiffiffiffiffiffiffi
I 2D
p eth18THORN
where the constant ϕfrac14 300007 is the default angle of internal
friction value and I 2D is the second invariant of the deviatoric
stress tensor which can be written as
I 2D frac14 16 ethDs11minusDs22THORN2 thorn ethDs22minusDs33THORN2 thorn ethDs33minusDs11THORN2h i
thornD2s12 thorn D2
s23 thorn D2s31 eth19THORN
Dsij frac14 1
2
partusi
part x j
thorn partus j
part xi
eth20THORN
32 Boundary conditions and numerical procedure
At the inlet all velocities and volume fractions of both phases
were speci1047297ed The inlet velocity pro1047297le is assumed to be uniform
and the volume fraction constant The average solids velocity at
the inlet is calculated from the shape of the kibble and it is
assumed that the inlet air has the same velocity as the solid phase
At the outlet the pressure is assumed to be atmospheric while theremaining opening areas are de1047297ned as pressure inlets The
specularity coef 1047297cient is speci1047297cally used in the multiphase
analysis with granular 1047298ow to express the slip coef 1047297cient At the
wall the gas phase was set to the no-slip condition meaning the
air velocity is zero at the surface of the wall while different slip
conditions between the solid and the wall were conducted to
assess the effect of slip at the wall
In the present investigation the interaction of gas and solid
phases was represented by the well-known Gidaspow drag model
The Gidaspow model has been successfully applied in similar
applications such as spouted bed simulations (Lan et al 2012)
While the turbulence model selected for this study was the
standard kminusε model Unsteady simulations were performed with
a small initial time step of 1 10minus4
s while 20 iterations per time
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 197
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 59
step were used to ensure numerical stability A convergence
criterion of 110minus3 was speci1047297ed for the relative error between
two successive iterations The typical computational time for each
simulation was 5ndash10 days on a 267 GHz workstation
33 Grid independency
To con1047297rm that the CFD results are independent of the mesh
size simulations of Chute A with four different meshes were
performed A coarse mesh was created using a maximum face
size of 15 mm an intermediate mesh using a maximum face size of
12 mm a 1047297ne mesh using a maximum face size of 10 mm and a
1047297ner mesh using a maximum face size of 8 mm The air velocity is
an important characteristic parameter that can be used to deter-
mine how well numerical models capture the behaviour of the
transfer chute As shown in Fig 3 the 1047297ne and 1047297ner mesh cases
predicted similar outlet air velocities however considering the
minor variation in results and computational time a 1047297ne mesh
with a maximum face size of 10 mm was used in subsequent
simulations
4 Results and discussion
To provide guidance on the selection of values for the particlendash
particle restitution coef 1047297cient and the specularity coef 1047297cient when
simulating granular 1047298ow in transfer chutes an investigation into
the most appropriate model parameters was performed Chute Awas used to perform the parametric studies while Chute B was
used to verify the applicability of the chosen model parameters
41 In 1047298uence of the particle-wall slip condition
The specularity coef 1047297cient is a measure of the fraction of
collisions which transfer momentum to the wall and varies from
zero (smooth walls) to one (rough walls) The effect of the slip
condition between the solid phase and wall was investigated by
altering the specularity coef 1047297cient for the solid phase The free-slip
condition for the solid phase at the wall is usually obtained by
setting the shear condition to zero In total eight different
specularity coef 1047297cients were analysed to study the effect of the
particlendash
wall slip condition using a default particlendash
particle
restitution coef 1047297cient value of 09 The air velocities along the
centreline of the chute outlet for both the simulation and experi-
mental results are shown in Fig 4 with the specularity coef 1047297cient
and particlendashparticle restitution coef 1047297cient referenced as R and S
respectively There are no PIV measurements for the lower section
of the chute since the chute is 1047297tted with metal gussets in the
corners
Comparison between simulation results and experimental data
at the outlet of the scale model transfer chute showed signi1047297cant
variance with different specularity coef 1047297cients Differences mainly
occurred for specularity coef 1047297cients less than 01 with values less
than 001showing signi1047297cant differences both in the magnitude
and distribution from that of the experimentally measured values
From further inspection of Fig 4 the specularity coef 1047297cient values
less than 01 show the maximum air velocity position occurring
further away from the wall and increasing with an increase in the
value of the specularity coef 1047297cient Qualitatively speaking the
overall trends of the predicted air velocity pro1047297les are quite similar
for the data obtained from the specularity coef 1047297cients larger than
005 in spite of some quantitative differences In reference to the
experimental data it is clear the simulation results are lower
overall potentially resulting from a different particle 1047298ow depth
between the simulations and experiment
The specularity coef 1047297cient indirectly affects the air velocity by
affecting the particle behaviour The simulated particle velocity
pro1047297les and particle volume fractions at the outlet for the range of
specularity coef 1047297cients are illustrated in Fig 5(a) and (b) respec-
tively Due to dif 1047297culties in measuring the outlet particle velocitythis velocity was calculated using the classic continuum method of
Roberts (2003) An equivalent friction coef 1047297cient of 08 was used
based on the average measured value of the particle wall friction for
iron ore on PVC and steel Based on the continuum method
calculation a schematic drawing of particle velocity distribution is
given in Fig 5(a) for comparison to the other simulation results It is
noted that the continuum method calculation gives a single velocity
for the entire bulk material stream rather than a pro1047297le throughout
the stream thickness
As shown in Fig 5(a) the simulated particle velocity pro1047297les
indicate the particle velocities are gradually decreasing with an
increase in the specularity coef 1047297cient For specularity coef 1047297cient
values of 02 and 03 the results show similar velocity pro1047297le
predictions The simulated particle velocities sharply decrease to
00 10 20 30 40 50 60
000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
Coase Intermediate Fine Finer
Fig 3 Outlet air velocity pro1047297le of different mesh for Chute A
00 10 20 30 40 50 60000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
R=09Free slip R=09S=0005 R=09S=001 R=09S=003 R=09S=005 R=09S=01 R=09S=02 R=09S=03
Experiments
Fig 4 Outlet air velocities for different specularity coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202198
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 49
311 Mass conservation equations
The conservation of mass for the gas and solid phase can be
written as
part
partt ethε g ρ g THORN thornnablasdotethε g ρ g v
g THORN frac14 0 eth1THORN
part
partt ethεs ρsTHORN thornnablasdotethεs ρs v
sTHORN frac14 0 eth2THORN
where ε ρ and v are the volume fraction the density and theinstantaneous velocity respectively The volume fraction con-
straint requires ε g thorn εs frac14 1
312 Momentum conservation equations
Assuming no mass transfer between phases and no lift and
virtual mass forces due to the large difference in their densities
only the drag force and gravity are considered in this study thus
the conservation of momentum for each phase can be expressed
as
part
partt ε g ρ g v
g
thorn nablasdot ε g ρ g v
g v
g
frac14nablasdotτ g minusε g nablaP thorn ε g ρ g g thorn K sg v
sminus v
g
eth3THORN
part
partt εs ρs v
s
thorn nablasdot εs ρs v
s v
s
frac14nablasdotτ sminusnablaP sminus εsnablaP thorn εs ρs g thorn K sg v
sminus v
g
eth4THORN
where τ is the stress tensor P is the pressure g is gravity and K sg is
the momentum exchange coef 1047297cient that represents the drag force
between the gas and the solid phase Details of these parameters
are explained below
The stress tensor for each phase is given by
τ g frac14 ε g μ g nabla v
g thorn nabla v
g
T
thorn ε g λ g minus2
3 μ g
nablasdot v
g I eth5THORN
τ s frac14 εs μs nabla vs thorn nabla vs T thorn εs λsminus
2
3 μs
nablasdot vsI eth6THORN
where λs and λ g are solid and gas viscosity respectively The
granular bulk viscosity λs represents the resistance of granular
particles to compression or expansion and is modelled by Lun
et al 1984 as
λs frac14 4
5εs ρsdseth1 thorn eTHORN
ffiffiffiffiffiffiΘs
π
r eth7THORN
where ds is the particle diameter e is the particlendashparticle restitu-
tion coef 1047297cient and Θs is the granular temperature
The granular pressure P s is derived from the kinetic theory of
granular 1047298ow and is composed of a kinetic term and a term due to
particle collisions In the current work the model proposed by Lun
et al 1984 is used and is given byP s frac14 εs ρsΘs thorn 2 ρseth1 thorn eTHORNε2
s g0sΘs eth8THORNwhere g 0s is the radial distribution function that modi1047297es the
probability of collisions between particles and is given by Lun et al
(1984) and expressed as
g 0s frac14 1minusεs
εsmax
13
minus1
eth9THORN
The in1047298uence of the drag model is not covered in this study
with the Gidaspow et al (1992) model selected for all work The
Gidaspow et al (1992) model is given by
K sg frac14 150ε2
g μ g
ε g d
2
s
thorn 175ε g ρ g j v
sminus v
g jd
s
ε g o08 eth10THORN
K sg frac14 3
4C D
ε g εsjvsminusv g jds
εminus265 g ε g ge08 eth11THORN
where C D
C D frac14 24
Res1 thorn 015ethResTHORN0687h i
Reso1000 eth12THORN
C D frac14 044 Res41000 eth13THORN
The Reynolds number of the particles is given by
Res frac14 ρ g j v
g minus v
sjds
μ g
eth14THORN
313 Granular shear viscosity
The solids stress tensor contains shear and bulk viscosities
arising from particle momentum exchange due to translation and
collision In other words the solids shear viscosity consists of a
collision term a kinetic term and a friction term and is given by
μs frac14 μscol thorn μskin thorn μsfric eth15THORNThe collision viscosity is taken from the kinetic theory of
granular 1047298ow by Gidaspow et al (1992)
μscol frac14 45ε2
s ρsds g 0seth1 thorn esTHORN ffiffiffiffiffiffiΘs
π
r eth16THORN
The kinetic viscosity is expressed by Gidaspow et al (1992) as
μskin frac14 10 ρsds
ffiffiffiffiffiffiffiffiπ Θs
p
96eth1 thorn esTHORN g 0s
1 thorn 4
5εs g 0seth1 thorn esTHORN
2
eth17THORN
The friction viscosity is expressed by Schaeffer (1987) as
μsfric frac14 P s sin ϕ
2 ffiffiffiffiffiffiffi
I 2D
p eth18THORN
where the constant ϕfrac14 300007 is the default angle of internal
friction value and I 2D is the second invariant of the deviatoric
stress tensor which can be written as
I 2D frac14 16 ethDs11minusDs22THORN2 thorn ethDs22minusDs33THORN2 thorn ethDs33minusDs11THORN2h i
thornD2s12 thorn D2
s23 thorn D2s31 eth19THORN
Dsij frac14 1
2
partusi
part x j
thorn partus j
part xi
eth20THORN
32 Boundary conditions and numerical procedure
At the inlet all velocities and volume fractions of both phases
were speci1047297ed The inlet velocity pro1047297le is assumed to be uniform
and the volume fraction constant The average solids velocity at
the inlet is calculated from the shape of the kibble and it is
assumed that the inlet air has the same velocity as the solid phase
At the outlet the pressure is assumed to be atmospheric while theremaining opening areas are de1047297ned as pressure inlets The
specularity coef 1047297cient is speci1047297cally used in the multiphase
analysis with granular 1047298ow to express the slip coef 1047297cient At the
wall the gas phase was set to the no-slip condition meaning the
air velocity is zero at the surface of the wall while different slip
conditions between the solid and the wall were conducted to
assess the effect of slip at the wall
In the present investigation the interaction of gas and solid
phases was represented by the well-known Gidaspow drag model
The Gidaspow model has been successfully applied in similar
applications such as spouted bed simulations (Lan et al 2012)
While the turbulence model selected for this study was the
standard kminusε model Unsteady simulations were performed with
a small initial time step of 1 10minus4
s while 20 iterations per time
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 197
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 59
step were used to ensure numerical stability A convergence
criterion of 110minus3 was speci1047297ed for the relative error between
two successive iterations The typical computational time for each
simulation was 5ndash10 days on a 267 GHz workstation
33 Grid independency
To con1047297rm that the CFD results are independent of the mesh
size simulations of Chute A with four different meshes were
performed A coarse mesh was created using a maximum face
size of 15 mm an intermediate mesh using a maximum face size of
12 mm a 1047297ne mesh using a maximum face size of 10 mm and a
1047297ner mesh using a maximum face size of 8 mm The air velocity is
an important characteristic parameter that can be used to deter-
mine how well numerical models capture the behaviour of the
transfer chute As shown in Fig 3 the 1047297ne and 1047297ner mesh cases
predicted similar outlet air velocities however considering the
minor variation in results and computational time a 1047297ne mesh
with a maximum face size of 10 mm was used in subsequent
simulations
4 Results and discussion
To provide guidance on the selection of values for the particlendash
particle restitution coef 1047297cient and the specularity coef 1047297cient when
simulating granular 1047298ow in transfer chutes an investigation into
the most appropriate model parameters was performed Chute Awas used to perform the parametric studies while Chute B was
used to verify the applicability of the chosen model parameters
41 In 1047298uence of the particle-wall slip condition
The specularity coef 1047297cient is a measure of the fraction of
collisions which transfer momentum to the wall and varies from
zero (smooth walls) to one (rough walls) The effect of the slip
condition between the solid phase and wall was investigated by
altering the specularity coef 1047297cient for the solid phase The free-slip
condition for the solid phase at the wall is usually obtained by
setting the shear condition to zero In total eight different
specularity coef 1047297cients were analysed to study the effect of the
particlendash
wall slip condition using a default particlendash
particle
restitution coef 1047297cient value of 09 The air velocities along the
centreline of the chute outlet for both the simulation and experi-
mental results are shown in Fig 4 with the specularity coef 1047297cient
and particlendashparticle restitution coef 1047297cient referenced as R and S
respectively There are no PIV measurements for the lower section
of the chute since the chute is 1047297tted with metal gussets in the
corners
Comparison between simulation results and experimental data
at the outlet of the scale model transfer chute showed signi1047297cant
variance with different specularity coef 1047297cients Differences mainly
occurred for specularity coef 1047297cients less than 01 with values less
than 001showing signi1047297cant differences both in the magnitude
and distribution from that of the experimentally measured values
From further inspection of Fig 4 the specularity coef 1047297cient values
less than 01 show the maximum air velocity position occurring
further away from the wall and increasing with an increase in the
value of the specularity coef 1047297cient Qualitatively speaking the
overall trends of the predicted air velocity pro1047297les are quite similar
for the data obtained from the specularity coef 1047297cients larger than
005 in spite of some quantitative differences In reference to the
experimental data it is clear the simulation results are lower
overall potentially resulting from a different particle 1047298ow depth
between the simulations and experiment
The specularity coef 1047297cient indirectly affects the air velocity by
affecting the particle behaviour The simulated particle velocity
pro1047297les and particle volume fractions at the outlet for the range of
specularity coef 1047297cients are illustrated in Fig 5(a) and (b) respec-
tively Due to dif 1047297culties in measuring the outlet particle velocitythis velocity was calculated using the classic continuum method of
Roberts (2003) An equivalent friction coef 1047297cient of 08 was used
based on the average measured value of the particle wall friction for
iron ore on PVC and steel Based on the continuum method
calculation a schematic drawing of particle velocity distribution is
given in Fig 5(a) for comparison to the other simulation results It is
noted that the continuum method calculation gives a single velocity
for the entire bulk material stream rather than a pro1047297le throughout
the stream thickness
As shown in Fig 5(a) the simulated particle velocity pro1047297les
indicate the particle velocities are gradually decreasing with an
increase in the specularity coef 1047297cient For specularity coef 1047297cient
values of 02 and 03 the results show similar velocity pro1047297le
predictions The simulated particle velocities sharply decrease to
00 10 20 30 40 50 60
000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
Coase Intermediate Fine Finer
Fig 3 Outlet air velocity pro1047297le of different mesh for Chute A
00 10 20 30 40 50 60000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
R=09Free slip R=09S=0005 R=09S=001 R=09S=003 R=09S=005 R=09S=01 R=09S=02 R=09S=03
Experiments
Fig 4 Outlet air velocities for different specularity coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202198
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 59
step were used to ensure numerical stability A convergence
criterion of 110minus3 was speci1047297ed for the relative error between
two successive iterations The typical computational time for each
simulation was 5ndash10 days on a 267 GHz workstation
33 Grid independency
To con1047297rm that the CFD results are independent of the mesh
size simulations of Chute A with four different meshes were
performed A coarse mesh was created using a maximum face
size of 15 mm an intermediate mesh using a maximum face size of
12 mm a 1047297ne mesh using a maximum face size of 10 mm and a
1047297ner mesh using a maximum face size of 8 mm The air velocity is
an important characteristic parameter that can be used to deter-
mine how well numerical models capture the behaviour of the
transfer chute As shown in Fig 3 the 1047297ne and 1047297ner mesh cases
predicted similar outlet air velocities however considering the
minor variation in results and computational time a 1047297ne mesh
with a maximum face size of 10 mm was used in subsequent
simulations
4 Results and discussion
To provide guidance on the selection of values for the particlendash
particle restitution coef 1047297cient and the specularity coef 1047297cient when
simulating granular 1047298ow in transfer chutes an investigation into
the most appropriate model parameters was performed Chute Awas used to perform the parametric studies while Chute B was
used to verify the applicability of the chosen model parameters
41 In 1047298uence of the particle-wall slip condition
The specularity coef 1047297cient is a measure of the fraction of
collisions which transfer momentum to the wall and varies from
zero (smooth walls) to one (rough walls) The effect of the slip
condition between the solid phase and wall was investigated by
altering the specularity coef 1047297cient for the solid phase The free-slip
condition for the solid phase at the wall is usually obtained by
setting the shear condition to zero In total eight different
specularity coef 1047297cients were analysed to study the effect of the
particlendash
wall slip condition using a default particlendash
particle
restitution coef 1047297cient value of 09 The air velocities along the
centreline of the chute outlet for both the simulation and experi-
mental results are shown in Fig 4 with the specularity coef 1047297cient
and particlendashparticle restitution coef 1047297cient referenced as R and S
respectively There are no PIV measurements for the lower section
of the chute since the chute is 1047297tted with metal gussets in the
corners
Comparison between simulation results and experimental data
at the outlet of the scale model transfer chute showed signi1047297cant
variance with different specularity coef 1047297cients Differences mainly
occurred for specularity coef 1047297cients less than 01 with values less
than 001showing signi1047297cant differences both in the magnitude
and distribution from that of the experimentally measured values
From further inspection of Fig 4 the specularity coef 1047297cient values
less than 01 show the maximum air velocity position occurring
further away from the wall and increasing with an increase in the
value of the specularity coef 1047297cient Qualitatively speaking the
overall trends of the predicted air velocity pro1047297les are quite similar
for the data obtained from the specularity coef 1047297cients larger than
005 in spite of some quantitative differences In reference to the
experimental data it is clear the simulation results are lower
overall potentially resulting from a different particle 1047298ow depth
between the simulations and experiment
The specularity coef 1047297cient indirectly affects the air velocity by
affecting the particle behaviour The simulated particle velocity
pro1047297les and particle volume fractions at the outlet for the range of
specularity coef 1047297cients are illustrated in Fig 5(a) and (b) respec-
tively Due to dif 1047297culties in measuring the outlet particle velocitythis velocity was calculated using the classic continuum method of
Roberts (2003) An equivalent friction coef 1047297cient of 08 was used
based on the average measured value of the particle wall friction for
iron ore on PVC and steel Based on the continuum method
calculation a schematic drawing of particle velocity distribution is
given in Fig 5(a) for comparison to the other simulation results It is
noted that the continuum method calculation gives a single velocity
for the entire bulk material stream rather than a pro1047297le throughout
the stream thickness
As shown in Fig 5(a) the simulated particle velocity pro1047297les
indicate the particle velocities are gradually decreasing with an
increase in the specularity coef 1047297cient For specularity coef 1047297cient
values of 02 and 03 the results show similar velocity pro1047297le
predictions The simulated particle velocities sharply decrease to
00 10 20 30 40 50 60
000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
Coase Intermediate Fine Finer
Fig 3 Outlet air velocity pro1047297le of different mesh for Chute A
00 10 20 30 40 50 60000
002
004
006
008
010
012
H e i g h t o f O u
t l e t ( m )
Air Velocity (ms)
R=09Free slip R=09S=0005 R=09S=001 R=09S=003 R=09S=005 R=09S=01 R=09S=02 R=09S=03
Experiments
Fig 4 Outlet air velocities for different specularity coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202198
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 69
zero from a height of 8 mm from the wall indicating the particle
velocities near the wall cannot be predicted by the simulation The
distance of 8 mm is about twice the particle diameter used in this
simulation From further inspection of Fig 5(a) results illustrate
that when the specularity coef 1047297cient values are smaller than 01
the simulated particle velocities greater than 40 mm above the
bottom surface of the chute have the same distribution This can
be explained by the simulated particle volume fractions shown in
Fig 5(b) indicating the volume fraction of particles greater than
40 mm above the bottom surface of the chute approach 0
Quantitatively the simulated velocities obtained from the larger
specularity coef 1047297cient values of 02 and 03 show better agreement
with the theoretical result in terms of the maximum velocity
42 In 1047298uence of the particlendash particle restitution coef 1047297cient
The particlendashparticle restitution coef 1047297cient describes the
amount of the energy dissipation due to collisions between solid
particles It in1047298uences the momentum conservation and granular
temperature conservation of the particle phase In transfer chute
analysis the concentration of the particles can vary considerably
throughout the transfer chute with inter-particle collisions being
subject to signi1047297cantly different energy losses Coef 1047297cient values of
090 075 060 045 and 030 were chosen to examine the
in1047298uence of particlendashparticle restitution in the CFD simulations
Furthermore it is noted that the free slip condition cannot
describe the real situation from the particlendashwall slip condition
study thus the particlendashparticle restitution coef 1047297cient study was
carried out using a specularity coef 1047297cient equal to 01
The results shown in Fig 6(a) illustrate the air velocities are notvery sensitive to variation in the particlendashparticle restitution
coef 1047297cient Air velocities greater than 70 mm above the bottom
surface of the chute are found to be quite similar for different
particlendashparticle restitution coef 1047297cients For the same specularity
coef 1047297cient values the overall trends of the air velocities are almost
the same for different particlendashparticle restitution coef 1047297cient
values with the only difference being the magnitude This 1047297nding
indicates that the velocity pro1047297le trend is more dependent on the
specularity coef 1047297cient rather than the particlendashparticle restitution
coef 1047297cient The smaller the particlendashparticle restitution coef 1047297cient
the lower the air velocity and particle velocity The percentage
decrease in velocity was not found to be proportional to the
decrease of the particlendashparticle restitution coef 1047297cient value with
the effect of particlendash
particle restitution coef 1047297cient on the velocity
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
00 10 20 30 40 50 00 01 02 03 04
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Volume fraction
Fig 5 Outlet particle velocity (a) and particle volume fraction (b) for different specularity coef 1047297cient value
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Experiments
00 10 20 30 40
00 05 10 15 20 25
000
002
004
006
008
010
012
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
R=09S=01 R=075S=01 R=06S=01 R=045S=01 R=03S=01 Calculated
Fig 6 Outlet air velocities (a) and particle velocity (b) for different particlendash
particle restitution coef 1047297cient values
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 199
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 79
becoming weaker with decreasing values Quantitatively the
simulated velocities obtained from the particlendashparticle restitution
coef 1047297cient value of 03 show the best agreement with the experi-
mental result in terms of the predicted air velocity pro1047297le while
the results obtained from the particlendashparticle restitution coef 1047297-
cient value of 06 show the best agreement with the theoretical
result in terms of the maximum particle velocity
43 In 1047298uence of fractional packing limit
In view of the potential high volume fraction for the solid phase
around the hood section the Schaeffer model was selected to take
into account the friction between particles The frictional stress
will be added to the stress predicted by the kinetic theory when
the solids volume fraction exceeds a critical value known as the
Fractional Packing Limit (FPL) The FPL represents a threshold
volume fraction at which the frictional regime becomes dominantIn order to investigate the effect of the FPL three fractional packing
limit values were investigated in this study including 061(default
value in FLUENT) 050and 040The specularity coef 1047297cient and
particlendashparticle restitution coef 1047297cient were set to 01 and
09 respectively The outlet air and particle velocities obtained from
the three different FPLs are shown in Fig 7(a) and (b) respectively
Fig 7(a) illustrates that a smaller FPL value generally leads to
lower air velocities while Fig 7(b) generally follows the same
trend with some discrepancy at higher particle velocities In terms
of the air velocity pro1047297le the results obtained from the FPL value
of 061 show better agreement with the experimental data While
the particle velocities which are obtained from the FPL values
equal to 04 are closer to the theoretical results given the purpose
of this study the accuracy of air velocity prediction is moreimportant and thus an FPL equal 061 will be used in subsequent
simulations
44 Combination of parameters
In order to 1047297nd the appropriate modelling parameters to
successfully simulate the Iron Ore and induced air 1047298ow in the
scale model transfer chutes different combinations of particlendash
particle restitution coef 1047297cient and specularity coef 1047297cient were
analysed Four different specularity coef 1047297cient values and 1047297ve
different particlendashparticle restitution coef 1047297cient values were com-
pared In total seven different combinations of specularity coef 1047297-
cient and restitution coef 1047297cient were investigated with the
selected parameters listed in Table 1
The air and particle velocity pro1047297les for different combinations
as well as the experimental air velocity pro1047297le and theoretically
calculated particle velocity pro1047297le are shown in Fig 8(a) and (b)
respectively
Fig 8 shows that up to a height of 50 mm greater particle ndash
particle restitution coef 1047297cients result in higher air and particle
velocities while higher specularity coef 1047297cients result in lower air
and particle velocities As noted previously specularity coef 1047297cients
greater than 01 show better correlation with the predicted air
velocities regardless of the particlendashparticle restitution coef 1047297cient
values
From the point of view of air velocity best overall agreement
between the model predictions and experimental data is obtained
using 045 as a particlendashparticle restitution coef 1047297cient and 02 as a
specularity coef 1047297cient This combination results in a maximum
variance between the measured and simulation values of 19 with
an average variance of 11 Although this combination gives the
closest results to the experimental data it is obvious that the
difference between the predicted particle velocities and theore-
tical result is large This combination resulted in a variance of 25
in terms of the maximum predicted particle velocity
From the point of view of particle velocity best overall agree-
ment between the model predictions and theory was obtained
using 06 as a particlendashparticle restitution coef 1047297cient and 01 as aspecularity coef 1047297cient with a variance of 6 in terms of the
maximum predicted particle velocity This combination resulted
in an average variance of 21between the measured air velocities
and the simulation results with a maximum variance of 51 Good
agreement with the measured and theoretical particle velocity is
also observed from the results obtained from the specularity
coef 1047297cient value equal to 02 and the particlendashparticle restitution
coef 1047297cient value of 075This combination resulted in an average
variance of 19 between the measured air velocities and simula-
tion results with a maximum variance of 41
Considering air and particle velocity prediction the seventh
combination using 045 as a particlendashparticle restitution coef 1047297-
cient and 02 as a specularity coef 1047297cient is considered the best
overall compromise
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25
000
002
004
006
008
010
012
H e i g h t
o f O u t l e t ( m )
Particle Velocity (ms)
Fig 7 Outlet air velocities (a) and particle velocity (b) for different fractional packing limit values
Table 1
Different combinations of specularity coef 1047297cient and restitution coef 1047297cient
Combinations 1 2 3 4 5 6 7
Restitution coef 1047297cient 045 090 030 060 075 060 045Specularity coef 1047297cient 003 005 005 010 0 20 020 020
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202200
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 89
45 Application of selected model parameters
To further evaluate the chosen modelling parameters Chute B
was investigated Due to the introduction of the restrictor plate
Chute B involved more particlendashwall and particlendashparticle inter-
actions than Chute A Based on the previous discussion the
combinations of modelling parameters labelled as 5 and 7 in
Table 1 were investigated
Fig 9 shows the simulated air velocity and particle velocity
distributions at the outlet of transfer Chute B using different
combinations of modelling parameters As expected the para-
meters have a similar performance on Chute B as Chute A In both
000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Air Velocity (ms)
00 10 20 30 40 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 8 Outlet air velocities (a) and particle velocities (b) for different modelling parameters
000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Air Velocity (ms)
00 05 10 15 20 25 30 35 00 05 10 15 20 25 30000
002
004
006
008
010
012
H e i g
h t o f O u t l e t ( m )
Particle Velocity (ms)
Fig 9 Outlet air velocities (a) and particle velocities (b) for different modelling parameters applied to Chute B
Fig 10 Air vectors for Chute B (a) from PIV experiments (b) from CFD analysis
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202 201
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202
7232019 Computational Fluid Dynamics (CFD) modelling of transfer chutes A study of the influence of model parameters
httpslidepdfcomreaderfullcomputational-fluid-dynamics-cfd-modelling-of-transfer-chutes-a-study-of 99
cases the particle velocities are underestimated with the primary
difference being that the air velocities are slightly underestimated
in Chute A while for Chute B the air velocities are slightly over
estimated Overall the trends of the air velocities agree well with
the experimental results This is best illustrated in Fig 10 where
the measured air vectors from the PIV experiments and the air
vectors predicted from the CFD analysis for Chute B are compared
directly
Fig 10 illustrates that the predicted 1047298ow patterns agree wellwith the measured results Note there is no experimental data
available beneath the black line in Fig 10 (a) due to the steel
gussets in the corner of the chute The turbulence is captured quite
well in the PIV results and can be clearly seen in the CFD results
The experimental results are not as smooth as the simulation
results but this is most likely due to the uneven distribution of
seeding particles Quantitatively the CFD analysis generally over-
estimates the air velocities when compared to the velocities
measured by the PIV system but given the nature of the experi-
mental work satisfactory conclusions can still be drawn
5 Conclusion
A two-1047298uid model (TFM) approach using the commercial CFDsoftware FLUENT has been used to simulate the granular and
entrained air 1047298ow in a transfer chute The solid particle behaviour
near the wall was investigated using different slip conditions the
in1047298uence of the particlendashparticle restitution coef 1047297cient studied
and the effect of frictional viscosity investigated by altering the
maximum packing limit Several cases with different specularity
coef 1047297cients (S frac140 0005 003 005 01 and 02) particlendashparticle
restitution coef 1047297cients (Rfrac1409 075 06 045 and 03) and max-
imum fraction packing limit (FPL frac14061 05 04) were simulated to
evaluate their impact on the 1047298ow behaviour Corresponding
experimental studies were conducted to evaluate the multiphase
CFD model The air and particle velocities at the transfer chute
outlet were compared and discussed with the following conclu-
sions drawn from the study
(1) Generally smaller specularity coef 1047297cient values result in
larger velocities however the change of air and particle
velocities is not proportional to the value of the specularity
coef 1047297cient Air velocities are very sensitive to the specularity
coef 1047297cient values less than 01
(2) Varying the coef 1047297cient of particle restitution did not affect air
velocity predictions for the top section of outlet where the air
volume fraction is 1 In the bottom section of chute outlet
higher particlendashparticle restitution values result in higher air
and particle velocities
(3) Decreasing the maximum fraction packing limit decreases the
air and particle velocities
(4) In the zone close to the base of the chute particle velocities
cannot be predicted by any of the combinations of parameters
investigated
(5) For the range of parameters investigated the combination
labelled 7 (Rfrac14045 S frac1402) proved to be the best overall
combination of parameters for this particular application
with good overall agreement between the calculated and
measured air velocities
References
Almuttahar A Taghipour F 2008 Computational 1047298uid dynamics of high densitycirculating 1047298uidized bed riser study of modeling parameters Powder Technol185 (1) 11ndash23
Ansart R Ryck Ad Dodds JA 2009 Dust emission in powder handling freefalling particle plume characterisation Chem Eng J 152 (2ndash3) 415ndash420
Ansart R Letourneau J-J de Ryck A Dodds JA 2011 Dust emission by powderhandling in1047298uence of the hopper outlet on the dust plume Powder Technol212 (3) 418ndash424
Chen XL Wheeler CA Donohue TJ McLean R Roberts AW 2012 Evaluationof dust emissions from conveyor transfer chutes using experimental and CFDsimulation Int J Miner Process 110ndash111 (0) 101ndash108
Donohue TJ Roberts AW Wheeler CA McBride W 2009 Analysis of transfer
chute performance focusing on reduced dust emissions In Eighth WorldCongress of Chemical Engineering 2009 Montreal Quebec
Du W Bao X Xu J Wei W 2006 Computational 1047298uid dynamics (CFD) modelingof spouted bed in1047298uence of frictional stress maximum packing limit andcoef 1047297cient of restitution of particles Chem Eng Sci 61 (14) 4558 ndash4570
Gidaspow D Bezburuah RDing J 1992 Hydrodynamics of circulating 1047298uidizedbeds Kinetic theory approach In Conference Seventh International Confer-ence on Fluidization Gold Coast Australia 75ndash82
Goniva C Kloss C Chen X Donohue TJ Katterfeld A 2012 Prediction of DustEmissions in Transfer Chutes by Multiphase CFD and Coupled DEM-CFDSimulations In BulkSolids Europe 2012 Berlin Germany
Lan X Xu C Gao J Al-Dahhan M 2012 In1047298uence of solid-phase wall boundarycondition on CFD simulation of spouted beds Chem Eng Sc 69 (1) 419ndash430
Lun CKK Savage SB Jeffrey DJ Chepurniy N 1984 Kinetic theories forgranular 1047298ow inelastic particles in Couette 1047298ow and slightly inelastic particlesin a general 1047298ow1047297eld J Fluid Mech 140 223ndash256
McIlvenna P Mossad DR 2003 Two dimensional transfer chute analysis using acontinuum method In Third International Conference on Computational FluidDynamics in the Minerals and Process Industries (CFD2003) MelbourneAustralia 547ndash551
Meinhart CD T WS Gray MHB 2000 Volume illumination for two-dimensional particle image velocimetry Meas Sci Technol 11 809ndash814
Melling A 1997 Tracer particles and seeding for particle image velocimetry MeasSci Technol 8 (12) 1406
Roberts AW 2003 Transfer chute performance and design for rapid 1047298owcondition Chem Eng Technol 26 (2) 163ndash170
Schaeffer DG 1987 Instability in the evolution equations describing incompres-sible granular 1047298ow J Differential Equations 66 (1) 19ndash50
Schmitt T Koster JN Hamacher H 1995 Particle design for displacementtracking velocimetry Meas Sci Technol 6 (6) 682ndash689
Witt PJ Carey KGNguyen TV 1999 Prediction of dust loss from conveyor usingCFD modelling In Second International Conference on CFD in the Minerals andProcess Industries CSIRO Melbourne Australia 281ndash286
X Chen C Wheeler Chemical Engineering Science 95 (2013) 194ndash 202202