Anda di halaman 1dari 13

This article was downloaded by: [Thapar University]

On: 14 July 2014, At: 14:23


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Particulate Science and Technology: An


International Journal
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/upst20

Bend Pressure Drop Predictions Using


the Euler-Euler Model in Dense Phase
Pneumatic Conveying
a a a a
D. Mcglinchey , A. Cowell , E. A. Knight , J. R. Pugh , A. Mason
a a
& B. Foster
a
School of Engineering, Science and Design, Glasgow Caledonian
University , Glasgow, United Kingdom
Published online: 14 Dec 2007.

To cite this article: D. Mcglinchey , A. Cowell , E. A. Knight , J. R. Pugh , A. Mason & B. Foster
(2007) Bend Pressure Drop Predictions Using the Euler-Euler Model in Dense Phase Pneumatic
Conveying, Particulate Science and Technology: An International Journal, 25:6, 495-506, DOI:
10.1080/02726350701492827

To link to this article: http://dx.doi.org/10.1080/02726350701492827

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
Content) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Particulate Science and Technology, 25: 495506, 2007
Copyright # Taylor & Francis Group, LLC
ISSN: 0272-6351 print/1548-0046 online
DOI: 10.1080/02726350701492827

Bend Pressure Drop Predictions Using the


Euler-Euler Model in Dense Phase
Pneumatic Conveying

D. MCGLINCHEY
A. COWELL
E. A. KNIGHT
J. R. PUGH
A. MASON
Downloaded by [Thapar University] at 14:23 14 July 2014

B. FOSTER
School of Engineering, Science and Design, Glasgow Caledonian
University, Glasgow, United Kingdom

Pneumatic conveying of powdered and granular materials is a very common


transport technology across a broad range of industries, for example, chemicals,
cosmetics, pharmaceuticals, and power generation. As the demands of these indus-
tries for greater efficiency increases and to comply with environmental regulations
there is a need for a more fundamental understanding of the behavior of materials
in pneumatic conveying systems. The approach presented in this article is to develop
a model of a section of pneumatic conveying line, a horizontal or vertical 90 bend, in
the commercial CFD software package FLUENT and to describe the multiphase
flow behavior by the mixture or Eulerian method. Models of this type have been used
in the past to show qualitative and quantitative agreement between model and
experiment. The model results presented were compared with experimental data
gathered from an industrial-scale pneumatic conveying test system. Broad qualitat-
ive agreement in trends and flow patterns were found. Quantitative comparisons
were less uniform, with predictions from around 10% to 90% different from experi-
mental results, depending on conveying conditions and bend orientation.

Keywords CFD, dense phase, experimental, FLUENT, pneumatic conveying

Introduction
Computational Fluid Dynamics (CFD) has become commonplace in the investi-
gation and design of single-phase fluid systems, in applications as diverse as
electronic circuits to aircraft wing design. The success of CFD technology in this area
has resulted in advances in economic performance across a broad spectrum of indus-
tries and driven theoretical advances, particularly in turbulence modeling. However,
the use of CFD for multiphase systems is still challenging and requires considerable
adaptation to be of the same level of industrial functionality.

Address correspondence to D. McGlinchey, School of Engineering, Science and Design,


Glasgow Caledonian University, Cowcaddens Rd., Glasgow G4 0BA, UK. E-mail:
d.mcglinchey@gcal.ac.uk

495
496 D. McGlinchey et al.

Figure 1. Modes of flow in a pneumatic conveying pipeline.


Downloaded by [Thapar University] at 14:23 14 July 2014

There are three basic modes of transport behavior in pneumatically conveying


bulk solids: the dilute phase where all the material is essentially in suspension flow,
a mode of flow available to most materials; dense phase plug flow, most often with
non-cohesive particles with high bulk permeability; and dense phase bed flow, for
materials with suitable aeration and deaeration characteristics. These modes have
been described elsewhere (Klinzing et al., 1997), but Figure 1 may aid understanding.

Modeling Pneumatic Conveying


There have been several CFD-based models reported in the literature for the three
modes of flow, for example, Mason et al. (1998) for dilute phase, Tsuji et al.
(1992), Xiang and McGlinchey (2004) for dense phase plug flow, and Mason and
Levy (2001) and Ratnayaka et al. (2004) for dense phase bed flow.
Ratnayaka et al. (2004) have reported results using the commercial code FLUENT
to model a horizontal to horizontal 90 bend based on the softwares inherent multi-
phase solvers with suitable input parameters. Their reported results are encouraging,
and they state that Fluent1 CFD software has a good potential to satisfactorily model
the gas-solid flow across a bend in dense phase pneumatic conveying.
With this inspiration the present authors investigated the ability of FLUENT to
predict the pressure drop across a 90 bend both in a horizontal plane and in a
vertical plane for an extended range of conveying conditions. The geometry of the
model was taken to match the experimental setup described, and model predictions
are compared with test data.

Experimental Tests and Results


Test data for comparison with model predictions were taken from previous
pneumatic conveying trials carried out by Mason (2001) and Foster (2005). Two
experimental test rigs were used to collect mass flow rate and pressure data. Both test
rigs control mass flow rate of air using venturi nozzles and control solids flow rate by
proportioning the air between the blow tank and the conveying line. Mass flow rate
of solids is determined from load cell data at the reception vessel. The important
Bend Pressure Drop Predictions Using the Euler-Euler Model 497
Downloaded by [Thapar University] at 14:23 14 July 2014

Figure 2. Bend geometry and pressure measurement locations for horizontal bend.

difference between the rigs germane to this investigation is the configuration of the
90 bend that was instrumented to measure pressure. The conveyed material was
cement with a mean particle size (d50) of approximately 25 micron (d10 6.5 micron,
d90 72.5 micron, approximately).
In Test Rig 1 the 90 bend of 52 mm internal diameter and D=d of 6 was config-
ured horizontally with single-ended and differential pressure transducers as shown in
Figure 2. Over 20 different conveying conditions were achieved with superficial air
velocities at the start of the conveying line, calculated from the measured gas press-
ure and temperature (inlet conditions) using the ideal gas law, in the range 1.60 to
33 m=s and solids loading ratios (SLR; the ratio of solids to air mass flow rates)
in the range of about 5 to 120. Five of these conditions are reported here in Table 1
for comparison with model predictions.
In Test Rig 2 the bend was configured vertically with single-ended and differen-
tial pressure transducers as shown in Figure 3. Nine different conveying conditions
were achieved with inlet air velocities in the range of approximately 7 to 15 m=s and
SLRs in the range of about 10 to 50. Three of these conditions are reported in Table 2
for comparison with model predictions. The estimated uncertainty in the measured

Table 1. Test parameters with horizontal bend


Test Airflow rate Solids flow SLR Pressure P1 Bend pressure
identifier [kg=s] rate [kg=s] [kg=kg] [kPa.abs] drop [kPa]

A 0.0562 1.00 18 126 8.2


B 0.0561 1.17 21 128 9.1
C 0.0571 2.61 46 146 15.6
D 0.0731 0.54 8 121 5.2
E 0.0110 1.30 119 132 10.3
498 D. McGlinchey et al.
Downloaded by [Thapar University] at 14:23 14 July 2014

Figure 3. Bend geometry and pressure measurement locations for vertical bend.

value is based on a combination of instrument accuracy and recorded standard


deviation and is given the value of 1000 Pa for all tests.
The two experimental instrumentation configurations used to supply data for
comparison with the model are unfortunately slightly different; the pressure
drop measured in the vertical bend extends over the sweep of the bend, whereas
the horizontal pressure measurement includes a post-bend acceleration length. A
new round of experiments is planned to provide data that are more directly
comparable.

Comparison with Numerical Results


In the Euler-Euler approach, the different phases are treated mathematically as
interpenetrating continua. The volume fractions are assumed to be continuous func-
tions of space and time, and their sum is equal to one. Conservation equations for each

Table 2. Test parameters with vertical bend


Test Airflow rate Solids flow SLR Pressure P1 Bend pressure
identifier [kg=s] rate [kg=s] [kg=kg] [kPa.abs] drop [kPa]

F 0.0590 0.83 14 190 1.4


G 0.0606 3.22 53 315 5.3
H 0.0681 2.84 43 307 4.0
Bend Pressure Drop Predictions Using the Euler-Euler Model 499

phase are derived to obtain a set of equations that have similar structure for all phases.
These equations are closed by providing constitutive relations. In FLUENT, three
different Euler-Euler multiphase models are available: the volume of fluid (VOF)
model, the mixture model, and the Eulerian model (Fluent, 2003). The mixture
model is designed for two or more phases (fluid or particulate). Applications of the
mixture model include particle-laden flows with low loading.
The mixture model solves the continuity equation for the mixture, the momen-
tum equation for the mixture, the energy equation for the mixture, and the volume
fraction equation for the secondary phases, as well as algebraic expressions for the
relative velocities. The mixture model uses a single-fluid approach and allows the
phases to be interpenetrating. The volume fractions for a control volume can there-
fore be equal to any value between 0 and 1, depending on the space occupied by the
phases. The mixture model allows the phases to move at different velocities, using
the concept of slip velocities.
Continuity equation for the mixture:
Downloaded by [Thapar University] at 14:23 14 July 2014

@
q r:qm~
vm 0 1
@t m

Mass averaged velocity:


P
n
vk
ak qk~
k1
vm
~ 2
qm

Mixture density:
X
n
qm ak qk 3
k1

where ak is the volume fraction for phase k.


FLUENTs mixture model uses an algebraic slip equation of the following form
based on Manninen et al. (1996):

qp  qm dp2
vpq
~ a
~ 4
18lq fdrag

where dp is the particle diameter of the secondary phase p, and ~


a is the secondary
phase particles acceleration and is of the form

vm
@~
a ~
~ g  ~
vm  r~
vm  5
@t

The drag function is taken from Schiller and Naumann (1935) and is given as:


fdrag 1 0:15Re0:687 Re  1000 6
0:0183Re Re > 1000
500 D. McGlinchey et al.

The volume fraction equation for secondary phase p can be obtained from the con-
tinuity equation:
@
ap qp r  ap qp~
vm r  ap qp~
vdr; p 7
@t

The Eulerian model is the most complex of the multiphase models in FLUENT. It
solves a set of momentum and continuity equations for each phase. Coupling is
achieved through the pressure and interphase exchange coefficients (fluid-solid
and solid-solid). For granular flows, the properties are obtained from the application
of kinetic theory. Applications of the Eulerian multiphase model include bubble
columns, risers, particle suspension, and fluidized beds. The fluid-solid exchange
coefficient includes terms for solids volume fraction, solids density, solids particle
diameter, and a drag function based on Reynolds number. For example, the model
following Gidaspow et al. (1992), the fluid-solid exchange coefficient, Ksl, is of the
Downloaded by [Thapar University] at 14:23 14 July 2014

following forms.
When al > 0:8
3 as al ql j~
vs ~
vl j 2:65
Ksl CD al 8
4 ds
where
24
CD 1 0:15al Res 0:687  9
al Res
When al  0:8
as 1  al ll q as j~
vs ~
vl j
Ksl 150 1:75 l 10
al ds2 ds
Solids-solids exchange coefficient, Kls, based on Syamlal (1987) includes terms for
coefficient of restitution, coefficient of friction between solids, particle diameter,
volume fraction, and solids velocity and has the following form:
 
Cfr:ls p8 as qs al ql dl ds 2 g0;ls
2
p
31 els 2
Kls j~
vl ~
vs j 11
2pql dl3 qs ds3

where els is the coefficient of restitution, Cfr;ls is the coefficient of friction between the
lth and sth solid phase particles, dl is the diameter of the particles of solid phase l,
and g0;ls is a radial distribution coefficient.
For granular flows in the compressible regime (i.e., where the solids volume
fraction is less than its maximum allowed value), a solids pressure is calculated inde-
pendently and used for the pressure gradient term in the granular-phase momentum
equation. The solids pressure is composed of a kinetic term and a term due to par-
ticle collisions:
ps as qs Hs 2qs 1 ess a2s g0;ss Hs 12
where ess is the coefficient of restitution for particle collisions and has a default value
of 0.9, g0;ss is a radial distribution function, and H is the granular temperature.
Bend Pressure Drop Predictions Using the Euler-Euler Model 501

The radial distribution function, g0, is a correction factor that modifies the prob-
ability of collisions between grains as the solids granular phase becomes dense. With
a single solids phase, FLUENT uses the formulation proposed by Ogawa et al.
(1980):
"  1=3 #1
as
g0 1  13
as;max

The granular temperature, derived from kinetic theory, is proportional to the


kinetic energy of the random motion of the particles, which from Ding and Gidas-
pow (1990) for the sth solids phase takes the form:
 
3 @  

vs Hs ps I s s : r~
qs as Hs r:qs as~ vs r:kHs rHs
2 @t
 fHs /ls 14
Downloaded by [Thapar University] at 14:23 14 July 2014

where
 

ps I s s : r~
vs
represents the generation of energy by the solid stress tensor, kHs rHs is the diffusion
of energy, kHs is the diffusion coefficient, fHs is the collisional dissipation of energy,
and /ls is the energy exchange between the lth fluid or solids phase and the sth solids
phase.
FLUENT uses an algebraic relation obtained by neglecting convection and dif-
fusion in the transport equation.
The solids stress tensor contains shear and bulk viscosities arising from particle
momentum exchange due to translation and collision. A frictional component of
viscosity can also be included to account for the viscous-plastic transition that occurs
when particles of a solid phase reach the maximum solid volume fraction.
The collisional, kinetic, and frictional parts are added to give a solids shear
viscosity:
ls ls;col ls;kin ls; fr 15

Based on work by Gidaspow et al. (1992) and Syamlal et al. (1993) the collisional
viscosity is given as:
 1=2
4 Hs
ls;col as qs ds g0;ss 1 ess 16
5 p
The default expression used in FLUENT for kinetic viscosity is again from Syamlal
et al. (1993) and given as:
p  
as ds qs Hs p 2
ls;kin 1 1 ess 3ess  1as g0;ss 17
63  ess 5
Using default conditions in FLUENT, the solids shear viscosity does not account
for the friction between the particles. The bulk viscosity is set to zero in the default
condition.
The modeling of turbulence in multiphase systems is an extremely interesting
and active research area with a large number of terms to be modeled. FLUENT uses
502 D. McGlinchey et al.

a default multiphase turbulence model called the mixture turbulence model, which
represents an extension of the single-phase k-e model and is applicable to stratified
multiphase flows.
The k and e equations describing this model are given as:
 
@ lt;m
qm k r  qm~
vm k r  rk Gk;m  qm e 18
@t qk
and
 
@ lt;m e
qm e r  qm~
vm e r  re C1e Gk;m  C2e qm e 19
@t re k
The turbulent viscosity is computed from:

k2
lt;m qm Cl 20
e
Downloaded by [Thapar University] at 14:23 14 July 2014

and the production of turbulent kinetic energy is computed from

Gk;m lt;m r~ vm T : r~
vm r~ vm 21
The bend geometry of Rigs 1 and 2 were modeled and meshed in GAMBIT,
FLUENTs geometry and mesh generation software. These geometries were used
in FLUENT with the following boundary conditions. At the inlet, velocity and con-
centrations of the phases were specified as estimated input parameters to match the
experimental conditions of the tests in Tables 1 and 2. The inlet pressure was not
specified and the outlet pressure was specified as an estimate of the experimental
condition. The gas phase was assumed to be incompressible as the pressure drop
is relatively small. At the wall, the tangential and normal gas velocities were set to
zero (no slip condition). The particle diameter and density were taken as a single
value of 25 micron and 2500 kg=m3 respectively, and there was no mass transfer
between the phases. All other parameters relating to turbulence and kinetic theory
were left as default values to provide a benchmark for subsequent analysis and
investigation. The FLUENT solver choice of mixture or Eulerian was made
principally on SLR (volume fraction) and inlet velocity; this is based on judgments
made on calculations of a Stokes number and interparticle spacing as recommended
in the Fluent Users Guide (2003) and from practical experience of conveying regimes.
The mixture solver was taken as steady and the Eulerian as unsteady.
An example of a pressure contour plot and volume fraction plot for the two
geometries are shown in Figures 4 and 5. The computation time was typically about
1 h on a standard desktop PC for each run.
Bend pressure drop predictions for the two bend configurations are presented in
Table 3 and are compared to equivalent experimental values in Figure 6 for horizon-
tal and vertical bends.

Comparing Predicted and Measured Bend Pressure Drop for the


Horizontal Bend
A comparison on the horizontal bend shows relatively good agreement between pre-
dicted and measured bend pressure drop for Test E, within 10%, and would seem
to confirm to the results of Ratnayaka et al. (2004), which show the potential of
Bend Pressure Drop Predictions Using the Euler-Euler Model 503

Figure 4. Example of pressure contour plot.


Downloaded by [Thapar University] at 14:23 14 July 2014

FLUENT to model flows of high solids loading ratios (120) using the Euler-Euler
solver and transient analysis. The ability to model lower solids loading ratios
( < 50) using the mixture solver and steady-state analysis was investigated in Tests
A, B, C, and D. Although the overall trend was correct, the predicted bend pressure
drop for these conditions was a significant underestimation of the measured test
result by a factor of at least three. Interestingly the predictions from both the mixture
and Eulerian models of test C were close to each other at 4.0 kPa and 3.8 kPa respect-
ively, although the pressure distributions across the pipe were qualitatively different.

Comparing Predicted and Measured Bend Pressure Drop for the


Vertical Bend
The prediction of bend pressure drop for conditions of Test F and H is within
50% of the measured values, whereas the prediction for conditions of Test G is an

Figure 5. Example of volume fraction plot.


504 D. McGlinchey et al.

Table 3. Predicted and measured bend pressure drop


Test Predicted pressure FLUENT Measured pressure
Bend identifier drop [kPa] solver drop [kPa]

Horizontal A 2.4 Mixture 8.2


B 2.8 Mixture 9.1
C 4.0 Mixture 15.6
C a 3.8 Eulerian 15.6
D 1.2 Mixture 5.2
E 9.6 Eulerian 10.3
Vertical F 2.1 Mixture 1.4
G 2.8 Eulerian 5.3
H 3.0 Eulerian 4.0
a
Experimental data from test C has been predicted using both the mixture model (C) and
Downloaded by [Thapar University] at 14:23 14 July 2014

the Eulerian model (C ).

Figure 6. Comparison of predicted and measured pressured drops.

underestimation by about 90% of the measured value. This is interesting as Test F


was modeled using the mixture solver and steady-state analysis, which proved
much less successful in the case of the horizontal bend.

Conclusions
The ability of the CFD software FLUENT to reliably to predict the pressure drop
across a 90 bend in a pneumatic conveying line transporting cement using standard
input parameters was investigated. Overall, there was broad qualitative agreement in
trends and flow patterns, however, there was large quantitative disparity. The results
suggest that at high solids loading ratio (120) the Eulerian solver and transient analy-
sis was reasonably effective with the bend in the horizontal plane. The effectiveness
of this approach was not as good when the bend was in the vertical plane for the
Bend Pressure Drop Predictions Using the Euler-Euler Model 505

conditions tested, with one point within 50% of the measured value and the other
about a 90% underestimation of the measured value. At lower solids loading ratios
the mixture model and steady-state analysis were considered more appropriate;
however, this method led to underestimations by a factor of around three for the
bend in the horizontal configuration. The mixture method proved better for the case
described with the vertically oriented bend, with a prediction within 50% of the
measured value. This may be related to the change in distribution of material in
the pipe as the bend changes from horizontal to vertical. There is a clear need to
study the modes of flow, how they are modeled, and the potential for using user-
defined functions to improve the predictions and to incorporate distinguishing
material characteristics. There is still a great deal of scope for further investigation
of the parameters relating to the kinetic granular theory embedded in the Eulerian
model, and this work is ongoing.
Downloaded by [Thapar University] at 14:23 14 July 2014

Nomenclature
a
~ acceleration
d particle diameter
e coefficient of restitution
fdrag drag function
g
~ gravitational acceleration
g0 radial distribution function
k turbulent kinetic energy per unit mass
p pressure
t time
v
~ Velocity
C constant
CD drag coefficient
Cfr coefficient of friction
G production of turbulent kinetic energy
K exchange coefficient
Re Reynolds number

Greek letters
a volume fraction
e turbulent kinetic energy dissipation rate
l viscosity
q density
s

solids stress tensor
s
/ energy exchange term
H granular temperature
Subscripts
col collisional
D drag
fr friction
kin kinetic
506 D. McGlinchey et al.

max maximum
k, l, m, p, q, s sphase identifiers

References
Ding, J. & D. Gidaspow. 1990. A bubbling fluidization model using kinetic theory of granular
flow. AIChE J. 36 (4): 523538.
Fluent Inc. 2003. Fluent 6.1 Users Guide. Lebanon, NH: Fluent.
Foster, B. 2005. BEng Honours Project, School of Engineering, Science and Design, Glasgow
Caledonian University.
Gidaspow, D., R. Bezburuah, & J. Ding. 1992. Hydrodynamics of circulating fluidized beds,
kinetic theory approach. In Fluidization VII: Proceedings of the 7th Engineering
Foundation Conference on Fluidization. New York: Engineering Foundation. pp. 7582.
Klinzing, G. E., R. D. Marcus, F. Rizk, & L. S. Leung. 1997. Pneumatic Conveying of Solids:
A Theoretical and Practical Approach. 2nd ed. London: Chapman and Hall.
Downloaded by [Thapar University] at 14:23 14 July 2014

Manninen, M., V. Taivassalo, & S. Kallio. 1996. On the mixture model for multiphase flow.
VTT Publications 288. Espoo: Technical Research Centre of Finland.
Mason, A. 2001. Bend effects in pneumatic conveying, MSc Project Report, Glasgow
Caledonian University.
Mason, D. J. & A. Levy. 2001. A model for non-suspension gas-solids flow of fine powders in
pipes. Int. J. Multiph. Flow 27: 415435.
Mason, D. J., P. Marjanovic, & A. Levy. 1998. A simulation system for pneumatic conveying
systems. Powder Technol. 95: 714.
Ogawa, S., A. Umemura, & N. Oshima. 1980. On the equation of fully fluidized granular
materials. J. Appl. Math. Phys. 31: 483.
Ratnayaka, C., M. C. Melaaen, & B. K. Datta. 2004. An experimental investigation and a
CFD modelling of gas-solid flow across a bend in a dense phase pneumatic conveying
system. In Proceedings of 8th International Conference on Bulk Materials Storage, Hand-
ling and Transportation, University of Wollongong, Wollongong, Australia. pp. 388393.
Schiller L. & A. Naumann. 1935. Z. Ver. Deutsch. Ing. 77: 318.
Syamlal, M. 1987. The particle-particle drag term in a multiparticle model of fluidization.
Springfield, Va.: National Technical Information Service. DOE/MC/21353-2373,
NTIS/DE87006500.
Syamlal, M., W. Rogers, & T. J. OBrien. 1993. MFIX Documentation: Volume 1, Theory
Guide. Springfield, Va.: National Technical Information Service. DOE/
METC-9411004, NTIS/DE9400087.
Tsuji, Y., T. Tanaka, & T. Ishida. 1992. Lagrangian numerical simulation of plug flow of
cohesionless particles in a horizontal pipe. Powder Technol. 71: 239250.
Xiang, J. & D. McGlinchey. 2004. Numerical simulation of particle motion in dense phase
pneumatic conveying. Granular Matter 6: 167172.

Anda mungkin juga menyukai