Available online at www.sciencedirect.

com

Energy Procedia 17 (2012) 1535 – 1541

2012 International Conference on Future Electrical Power and Energy Systems

Numerical simulation of gas and solid flow behavior in the

pre-calciner with large eddy simulation approach
Zheng Jianxiang1, Yan Tingzhi1, Yang Jing2
1
School of Energy and Power Engineering, .Northeast Dianli University , Jilin, China
2
School of Energy and Power Engineering, Northeast Dianli University , Jilin, China

Abstract

Flow behavior of the gas and particle in the pre-calciner was simulated through using the large eddy simulations and
the kinetic theory of granular flow. The results of numerical calculation show the disciplines and features of the gases
movement and the particles concentration distribution in the pre-calciner. This verifies that the design of the spouted
pre-calciner is reasonable, and providing a theory basis for the further research about the heat transfer, mass transfer
and chemical reaction process.

© 2012
2011 Published
Publishedby
byElsevier
ElsevierLtd.
Ltd.Selection
Selectionand/or peer-review
and/or under
peer-review responsibility
under of Hainan
responsibility University.
of [name organizer]

Keywords:pre-calciner; large eddy simulation; kinetic theory of granular flow; numerical simulation

1. Introduction

The pre-calciner is the main thermotechnical equipment in the process of the NSP cement production.
It is not only a burning furnace, but also a chemical reaction unit. It has many various functions, such as
burning coal, changing heat between gas and solid and decompositing raw materials. So realizing these
functions need a certain environments and conditions. The gas and solid flow behavior has a great
influence on the uniformity of the material distribution and the staying time in the pre-calciner. Therefore,
the 2-d flow field of the spouted pre-calciner is studied primarily. The research methods of the model
experiments are adopted mostly in the traditional engineering research. But with the continuous
development of the computer technology, the calculation method, especially the computational fluid
dynamics, using the numerical simulation method can save manpower and resources and adapt various
factors in the research of the furnace process flow.
This paper used the corresponding numerical method simulate a spouting pre-calciner in a factory’s
2000t/d NSP production line, the equations based on the large eddy simulations for both gas phase and
particles, and the kinetic theory of granular flow for particle collisions are described, and then the internal
flow field is introducted and illustrated. The simulated results provided the reference for the optimization

1876-6102 © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Hainan University.
doi:10.1016/j.egypro.2012.02.278
1536 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541

design of the pre-calciner and the reconstruction of the existing pre-calciner for the further studies.
Besides the numerical simulation results also provide the idea for the further discussion of the subject.

2. Basic Principle Of The Numerical Simulation Mathematics model

2.1 Mathematics model

The gas and solid flow in the pre-calciner is the 3-d approximate axisymmetric turbulent flow with
strong backflow. In 2-d cartesian coordinate system, starting from the Navier-Stokes equations which
describe the instantaneous turbulence strictly, after Renault decomposition and homogenization, the gas
turbulent flow can convert into the control of differential equations which include the continuous
equations in the stable condition and the momentum conservation equations which are meaningful in the
engineering application. And these equations can be solved by using numerical calculation. After that we
can apply the large eddy simulation model which has developped maturly to calculate closed turbulent
control equations.
2.1.1 Continuity equations(k=g is gas phase, k=s is solid).
w (1)
H k U k   ’ ˜ (H k U k vk ) 0
wt

U k is the density of the k phase, Hk is the volume concentration of the k phase, vk is the velocity of
the k phase, t is the time.

2.1.2 Gas phase momentum conservation equations

w
H g Ug vg   ’˜ (H g Ug vg vg ) H g ’Pg H g Ug g ’˜W g  Evg  vs  (2)
wt

Pg is the pressure of the gas phase, g is the gravity acceleration , E is the drag coefficient between
the phase g and s, W g is the stress tensor of the gas phase.
2
Wg P g [’v g  (’v g )T ]  P g (’ ˜ v g )I (3)
3
Pg P g ,l  P t (4)

P g ,l is the laminar flow dynamic viscosity of the gas, Pt is the turbulence dynamic viscosity of the gas.
Assuming the gas is incompressible and the density is constant.
1
2
2
ª wv § wvi wv j ·º (5)
Pt c '  « i ¨
¨ wx
 ¸»
¸
¬« w x j © j wxi ¹ ¼»
ǻ (ǻxǻyǻz )1/3 (6)

2.1.3 Solid phase momentum conservation equations

w

H s Us vs   ’ ˜( H s Us vs vs ) H s’Pg  H s Us g  ’ ˜W s  E vg  vs  (7)
wt

Ps is the pressure of the solid phase, W s is the stress tensor of the solid phase.

Ws  p s  H s[ s ’ ˜ v s I  2H s P s S s (8)
 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541 1537

Ss
2
> 3
@
1 ’ v s  (’ v s ) T  1 ’ v g (9)

Ps is the shearing viscosity of the particles, [s is the dynamic viscosity of the particles.
ps H s U sT >1  2( 1  e )g 0H s @ (10)
4 2 T 10U d ST 4
Ps Hs Usdgo(1e)  s [1 g H (1e)]2 (11)
5 S 96(1e)Hs go 5 o s
1
4 §T ·2 (12)
[s H s U s d s g 0 1  e ¨ ¸
5 ©S ¹
The radial distribution function g 0 is expressed as follows:

1 1
ª º
«1  §¨ H s ·¸ »
3
(13)
g0
« ¨© H s max ¸¹ »
«¬ »¼
The final polynomial of the solid phase momentum conservation equations show the force between the
gas phase and the solid phase. Trag coefficient£ is expressed as follows:

3 H sH g U g v g  v s 2.65
E CD Hg (14)
4 ds
24
CD
Re p
>
1  0.15Re p 
0.687
@ Re p  1000 (15)

CD 0.44 Re p t 1000 (16)

İ g ȡ g v g  vs
Re p ds (17)
ȝg

2.1.4 Turbulent kinetic energy Equation of the solid phase

3ª w
H s UsT   ’ ˜ H s UsTvs º» ’ ˜ N s’T   J s  3ET (18)
2 «¬ wt ¼
J s is the energy dissipation rate because of the collisions between particles, k is the turbulent kinetic
energy transfer coefficient.

ª 1 º
4 §T ·2 (19)
Js 31  e  H s U s g 0T ««
2 2 »
¨ ¸  ’ ˜ vs »
ds © S ¹
«¬ »¼
1

k 2H s2 U s d s g 0 1  e§¨ T ·¸ 2 (20)
©S ¹
The control equations are expressed the form of the partial differential equations, the purpose of the
numerical calculation is to take the place of the derivative of this set of partial differential equations by
1538 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541

using the difference method. It can achieve the purpose of solving by translating the differential equations
into algebraic equations furtherly. The process is called the discretion of the control equations.

2.2 Numerical solution

The above differential equations are established in the control volume by using the staggered grids
and the mixed format integration, getting the universal form of the differential equations. The difference
equations can be solved by using SIMPLE algorithm and under-relaxation algorithm, TDMA iteration by
line from upstream to downstream along the axial direction. The convergence criteria take the relative
error less than 10-3 between two iterations of the dependent variable.

2.3 Boundary conditions

(1) For the gas-phase, a no-slip boundary condition is used at the wall. For the solid phases, either a
no-slip condition or a partial slip wall boundary condition. In the partial slip wall boundary condition, the
shear stress in the particle phase (on approaching the wall) balances the transfer rate of momentum to the
wall by particles colliding with it, i.e.
Hs ws
Ww g 0.w U s S 3 T s us (21)
H s. max 6
(2) The flux of the turbulent kinetic energy at the wall is given as the generation of turbulence
kinetic energy by slip and dissipation by inelasticity of collisions between particles and the wall, i.e.
Hs ªw 1 º
qw  
g 0SU s 3 « s T s u s2  1  e w2 T s T s » (22)
H s. max ¬6 4 ¼
where ew is the coefficient of restitution for particle-wall collisions.

3. Discussion And Analysis Of The Calculating Results

The simulation cases were based on the actual size of a spouted pre-calciner which is running in the
2000t/d production line in a factory. The figure 1 shows its actual size. The pre-calciner’s high is 20m,
effective diameter is 4.54m, the inner diameter of the entrance which spraied the primary air is 1.7m. The
eight high-speed jet nozzle whose diameter is 40mm are added around the bottom of the pre-calciner,
distributing at 0.3457 in the vertical wall evenly. Assuming the primary air and particles are injeted from
the bottom of the furnace, the secondary air are injeted from the nozzle at a high speed. The speed of the
primary air is 25m/s, the speed of the secondary air is 190m/s, the high-speed air from the jets and the
mainstream are cold, and the temperature is 25ć.
 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541 1539

Figure 1 the dimension of the pre-calciner

Figure 2 shows the instantaneous concentration distribution of the particles. From the figure we can
see that particles are blown into pre-calciner evenly from the bottom with the primary air. The secondary
air is blown from the jet at a high speed. When particles are blowed into the furnace, the particle
concentration is high in the center of the furnace and low on both sides. Besides the local particle
concentration is very high in the area between the primary and secondary air. There are two low
concentration vortex areas near the air vents of the bottom. Because of the secondary air which is jetted
into the pre-calciner at a high speed, the distribution of particle concentration presents core-annulus fluid.
When t=3s, the particle movement is chaos. When t=5s, the movement trend tends to gently. When t=7s,
as can be seen from the firgure, particles fill the whole pre-calciner equably, the flow field has smoothed
gradually.

t=3s t=5s t=7s

Figure2 Instantaneous concentration distribution of the particles

Figure 3 shows the average concentration of the particles at different heights. From the figure we
can see that the average particle concentration is high in the center of the pre-calciner and low on both
sides. There are two particle concentration peaks, the first peak appeared in the center of the furnace as
the function of the primary air. The particles do the return movement between the area of the primary and
secondary air because of the gravity, so there is a low. It appears a second peak in the near wall because
of the high-speed jet of the secondary air that gives an upward momentum to the particles. In the near
wall, the particle concentration is not very high in the absence of high-speed jet of secondary air. Figure 4
shows that the gas average velocity at different heights. From the figure we can see that the gas velocity is
high in the center of the furnace and low on both sides. The primary air is transported into the furnace at
the speed of 25m/s, so there is a peak at the center of the furnace. It appears a secondary peak in the near
wall due to the secondary air which is injected into the furnace at the speed of 190m/s. Between the area
of the primary and secondary air, the gas velocity appears negative value because the gas do the back
movement.
1540 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541

0.3
y=1.89
y=12.42
y=16.58

Particle concentration
y=7.3
0.2

0.1

-2 -1 0 1 2
x(m)

Figure 3 Average concentration distribution of the particles

Figure 4 Average velocity distribution of the gas

Figure5 Average velocity distribution of the particles

 Zheng Jianxiang et al. / Energy Procedia 17 (2012) 1535 – 1541 1541

Figure 5 shows the particles velocity at different height. It is somewhat similar to the gas velocity
distribution. The primary air with particles is transported into the pre-calciner from the bottom of the pre-
calciner at the speed of 25m/s, appearing the first peak in the center of the furnace. The secondary air
which played the role of spouting the material is jetted into the pre-calciner at the 190m/s. The particles
do the chaotic motion under the action of the secondary air, thus creating anther peak. The particles fall
between the area of the primary and secondary air, producing a low speed, even the values of the speed is
negative.

4. Conclusion

The gas and particle turbulent flow in the pre-calciner is simulated by using the large eddy simulation
and the kinetic theory of granular flow model, the structure of it is confirmed.
The simulation results show that there is a jet core in the entrance of the turbulent flow field, there is a
circumfluence area in the near wall. The strongest turbulent kinetic energy appears at the bottom of the
pre-calciner, so we can launch materials in here where it can scatter fully.
The flow in the pre-calciner is decided by the inlet parameters, the boundary condition, etc. So these
parameters must be setted very detailed in the process of the simulation.

Acknowledgment

The authors would like to thank the Scientific Research Foundation for Doctor of Northeast Dianli
University (BSJXM-200806), for their financial support to the project.

References

[1] Zhao weilin, Li zhaofeng. The numerical simulation of the FLS two-dimensional flow field,The industrial furnace. 2002,
24(2): 45~47
[2] Ye xuchu, Hudaohe. The numerical simulation of the burning and decomposition in the SLC pre-calciner. The silicate
Journals. 1994ˈ22(4): 325~330
[3] Wang chaoqun. A new kind of spouting pre-calciner, The cement technology, 2000, (4): 10~14
[4] Li changyong, Jin chunqiang, Hu daohe. The research of the gas-solid two-phase movement pattern in the NKSV pre-
calciner. The silicate Journals. 2000. 28(6): 529~533
[5] Chen yiliang.The more common of the K - İ turbulence model. The Engineering thermal eurphys. 1990ˈ11(5): 238
[6] C. charles k.S. Xianliang. Orthonormal Wavelets and Tight Frames with Arbitrary Real Dilations. Applied and
Computational Harmonic Analysis, 2000, 9(3): 243~264
[7] D. Gidaspow. Multiphase Flow and Fluidization: Continuum and Kinetic Theory Description. Academic Press, San
Diego. 1994.
[8] J.J. Mc Guirk, J. M. LM Palma. The Efficiency of Alternative Pressible Turbulence Flow Problems. Comput Fluids.
1993, 22(1): 77~87