You are on page 1of 10

Powder Technology 350 (2019) 81–90

Contents lists available at ScienceDirect

Powder Technology

journal homepage: www.elsevier.com/locate/powtec

Determination of minimum fluidization velocity in fluidized bed at


elevated pressures and temperatures using CFD simulations
Yingjuan Shao a,b, Jinrao Gu a,b, Wenqi Zhong a,b,⁎, Aibing Yu b,c
a
Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education, School of Energy and Environment, Southeast University, Nanjing 210096, PR China
b
Center for Simulation and Modelling of Particulate Systems, Southeast University-Monash University Joint Research Institute, Suzhou, PR China
c
ARC Research Hub for Computational Particle Technology, Department of Chemical Engineering, Monash University, Clayton, Victoria 3800, Australia

a r t i c l e i n f o a b s t r a c t

Article history: The minimum fluidization velocity of particles is one of the most critical parameters for the design and operation
Received 1 January 2019 of fluidized beds. It is difficult and expensive to experimentally measure this parameter at elevated pressures and
Received in revised form 6 March 2019 temperatures. In this study, a three-dimensional model based on Eulerian−Lagrangian frames was used to pre-
Accepted 23 March 2019
dict the minimum fluidization velocity at elevated pressures and temperatures. The solid phase was simulated
Available online 25 March 2019
using the multi-phase particle-in-cell approach. The particle flow was described using the discrete particle ap-
Keywords:
proach, and a large eddy simulation was conducted to predict the turbulent gas motion. The simulation results
Computational fluid dynamics were compared with the experimental data reported in previous literatures. The bed temperature ranged from
Gas-solid flow 25 to 800 °C, and the operating pressure ranged from 0.1–4 MPa. In general, the minimum fluidization velocity
Minimum fluidization velocity decreased as the temperature and pressure increased. The variation trends were not significant at high temper-
Eulerian–Lagrangian approach atures and pressures (greater than 600 °C and 2 MP, respectively). With a decrease in the particle density, the
minimum fluidization velocity decreased. With an increase in the particle size, the minimum fluidization velocity
increased. Moreover, the influence of the particle size distribution on the minimum fluidization velocity was sys-
tematically evaluated. To precisely calculate the minimum fluidization velocity at elevated temperatures and
pressures, a correlation was proposed based on the simulation results.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction velocity (Umf), are significantly different from those at atmospheric


pressure and temperature.
Gas-solid fluidized beds, which exhibit excellent heat and mass The minimum fluidization velocity, as a transition velocity from the
transfer performances, have been widely used in various industrial ap- regime of a fixed bed to fluidized bed, is a critical parameter with re-
plications such as gasification and combustion for waste, biomass and spect to the design and operation of fluidized beds. It is defined as a su-
coal [1–3]. Due to the high heat–transfer rates, adequate control of the perficial gas velocity when the mass of the particles is in counterbalance
flow of solid particles, and the large thermal capacity of bed materials, with the drag force of the upward fluid. This fluidization parameter is
a fluidized bed reactor can demonstrate a high efficiency, operational significantly dependent on the of particle and fluid properties such as
flexibility, and output low–pollutant emissions. Moreover, the homoge- the particle size, density, pressure and temperature [4]. It is necessary
nous solid concentration and temperature in the fluidized bed are con- to understand the minimum fluidization velocity of the solids, to ensure
ducive to the scaled up operation. In recent years, pressurized oxy-fuel the operation of a fluidized bed in the right fluidization regime at ele-
combustion in fluidized beds has attracted significant research attention vated pressures and temperatures.
due to its lower pollutant emissions, high combustion efficiency and the The common method for the measurement of the minimum fluidi-
feasibility of efficient carbon capture. The hydrodynamics and gas-solid zation velocity under various conditions is the measurement of Umf
flow behaviors in the fluidized bed are influenced by several design and using experiments. Table 1 lists several experimental studies wherein
operational parameters such as the properties of the bed material and Umf was measured over a wide range of pressures, temperatures, and
the fluidization velocity. In general, the hydrodynamic parameters at el- particle sizes. Although the experimental method has been widely ap-
evated temperatures and pressures such as the minimum fluidization plied, it can be very costly, especially at elevated pressures and temper-
atures; which require the use of special equipment. Moreover, the
experimental method presents the potential safety hazard that results
from the high pressures and temperatures. In addition, the experimen-
⁎ Corresponding author at: Sipailou 2#, Nanjing 210096, Jiangsu, PRChina. tal method involves the use of correlations to predict the minimum flu-
E-mail address: wqzhong@seu.edu.cn (W. Zhong). idization velocity under various conditions, which includes those of the

https://doi.org/10.1016/j.powtec.2019.03.039
0032-5910/© 2019 Elsevier B.V. All rights reserved.
82 Y. Shao et al. / Powder Technology 350 (2019) 81–90

Table 1
Experiment studies of Umf at elevated pressure or temperature.

Author(s) Materials Particle size Temperature Pressure


(mm) (K) (MPa)

Saxena et al. [5] Dolomite 1.41–1.488 291–700 0.179–0.834


Nakamura et al. [6] Glass beads 0.2, 4 285–800 0.1–4.9
Zhu et al. [7] Quartz sand, glass bead 0.3–2 Ambient, 1173 0.5–2
Ma et al. [8] Quartz sand 0.1–5 303–873 Atmospheric
Bottom ash 0.1–6
Singh et al. [9] Silica sand 0.2 288–973 Atmospheric
Goo et al. [10] Silica sand 0.27 298–1073 Atmospheric
Subramani et al. [11] Ilmenite 0.128, 0.163, 0.2 298–1073 Atmospheric
Sand 0.134, 0.163, 0.2
Limestone 0.134, 0.2
Girimonte et al. [12] FCC 0.076, 0.096 303–773 Atmospheric
Silica sand 0.079, 0.1
Corundum 0.067
Olowson et al. [13] Silica sand 0.31, 0.7, 0.98 Ambient 0.1–1.6
Kang et al. [14] Silica sand 0.18, 0.59 Ambient 0.1–0.8

elevated pressures and temperatures. However, the correlations are particles that, group the real particles with common properties, to re-
typically derived from the results of a limited number of experiments place individual particles. Hence, the computational cost is significantly
under specific conditions; they may not be valid under conditions dif- decreased.
ferent from those of the experiments. Over the past decade, more studies were conducted on the Umf
In recent years, computational fluid dynamics (CFD) has been using simulations than by experiments [25]. Several publications of
widely used as a simulation tool for the prediction of gas-fluid the simulation studies related to the Umf are listed in Table 2.
flows [15,16]. It provides a faster and simpler method for the accu- Gosavi et al. [24] focused on the influence of temperature on the
rate prediction of Umf without the use of expensive and time con- Umf of lithium titanate particles in a fluidized bed. The predicted
suming experiments. Furthermore, the CFD method is a reliable results had an accuracy of approximately 95% when compared
alternative for experiments under the dangerous operating condi- with the experimental data. Amarasinghe and Jayarathna et al.
tions of high pressures and temperatures. The Eulerian–Eulerian [22,23] validated the accuracy of predicting the minimum fluidiza-
method and the Eulerian–Lagrangian method are the two main tion velocity using MP-PIC approach for particles of Geldart A, B
types of CFD simulations. and D. However, most of the previous studies on Umf simulations
The implementation of the Eulerian–Eulerian method, which is one were not focused on the minimum fluidization velocity at elevated
of the most widely used multi-phase flow simulation method, is typi- pressures and temperatures.
cally cost-effective in pilots and industrial plants. In this method, the The aim of this study was to develop a CFD model in which the solid
fluid and solid phases are considered as interpenetrating continua. phase was simulated using the MP-PIC method, for the prediction of the
Therefore, the particle behaviors such as collision, drag and heteroge- Umf at elevated pressures and temperatures. The model was first vali-
neous reactions were solved in an Eulerian grid. The insight of the bed dated by comparing the simulated results with those from experiments,
interior is easily provided, and the computational cost is lower using as obtained from the literature. It was then used to predict the mini-
this method. However, the method does not consider the particle size mum fluidization velocity under different operating conditions of tem-
distribution and the actual interactions between particles. peratures and pressures. Moreover, the effects of the particle density
On the other hand, the Eulerian–Lagrangian methods track the mo- and particle size distribution on the minimum fluidization velocity
tion and collision for each particle, and then describe the process using were extensively evaluated in this study.
the Newtonian second motion equation. Therefore, the real particle
properties can be considered. There are three major methods based on
the Eulerian–Lagrangian framework: the discrete phase method 2. Mathematical model
(DPM), the discrete element method (DEM) and the multi-phase
particle-in-cell (MP-PIC) method. The high computational cost limit The gas phase was modelled as a continuum using the MP-PIC
the application of DPM and DEM in a gas-solid fluidized bed, which is method, whereas the solid phase was computed using the discrete par-
a dense particulate reaction system. The MP-PIC method is an appropri- ticle method, to track the motion and collision of computational parti-
ate approach to the simulation of gas and particle motion in dense par- cles. The drag model was used to describe the fluid drag force due to
ticulate reaction systems [15]. This approach employs computational the upward flow.

Table 2
Simulation studies of Umf with using method.

Author(s) Materials Method Temperature (K) Pressure (MPa)

Wang et al. [17] Geldart A particles E/E-TFM Ambient Atmospheric


E/L-DPM
Dong et al. [18] Magnetite powder E/E Ambient Atmospheric
Boyce et al. [19] Geldart B particles E/L-DEM Ambient Atmospheric
Bandara et al. [20] Geldart B particles E/L-MP-PIC Ambient Atmospheric
Xu et al. [21] Glass bead E/L-DEM Ambient Atmospheric
Amarasinghe et al. [22] Zirconia, bronze, steel E/L-MP-PIC Ambient Atmospheric
Jayarathna et al. [23] Zirconia, bronze, steel E/L-MP-PIC Ambient Atmospheric
Gosavi et al. [24] lithium titanate E/E 303–873 Atmospheric
Y. Shao et al. / Powder Technology 350 (2019) 81–90 83

Table 3
Governing equations of fluid and solid.

Governing equations for the fluid

The continuity equation

∂ðεg ρg Þ
þ ∇  ðεg ρg ug Þ ¼ 0
∂t

The momentum equation

∂ðεg ρg ug Þ
þ ∇  ðε g ρg ug ug Þ ¼ −∇p þ F þ εg ρg g þ ∇  ðεg τ g Þ
∂t

The large eddy simulation (LES) model

∂ug;i ∂ug; j 2 ∂u
τ g ¼ μð þ Þ ¼ μδij k
∂x j ∂xi 3 ∂xk

μ = μl + μt
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 1 ∂u 1 ∂ug; j
2
2t g;i
μ t ¼ C s ρg Δ ð þ Þ
2 ∂x j 2 ∂xi

p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δ¼ 3
ΔxΔyΔz

The rate of momentum exchange between two phases

1
F ¼ ∬fmp ðDp ðug −up Þ− ∇pÞdmp dup
ρp

Governing equations for the solid phase

The particle distribution function

∂f dup
þ ∇  ðf up Þ þ ∇up  ð f Þ¼0 Fig. 1. Three-dimensional computational grid and geometric model.
∂t dt

The acceleration of particle


   
dup 1 1 180εp ρg ug −up 
¼ Dp ðug −up Þ− ∇p þ g− ∇  τp D2 ¼ þ 2 Cd ð3Þ
dt ρp εp ρp εg Re ρp dp
The solid volume fraction
3. Numerical conditions and solutions
εp = ∭ fVpdmpdupdVp

the inter-particle stress A simple cuboid fluidized bed with a height of 0.5 m and with an
inner length of 0.05 m was used to predict the Umf under different con-
P s ε βP
τp ¼ ditions. Based on the cuboid fluidized bed, a three-dimensional (3D)
max½εcp −εp ; αð1−εp Þ
Cartesian grid was manufactured, as shown in Fig. 1. The gas flow (Qg)
was inputted at the bottom to the fluidized bed, to promote the particle
flow. In addition, the gas outlet at the top was defined as a pressure
2.1. Governing equations boundary without a particle outlet. The gas flow was controlled by the
velocity boundary and each velocity was sustained for 5 s to reach
Table 3 lists the governing equations of the solid and fluid phases. steady–flow state. The initial atmosphere, bed martial and fluidization
gas were set as the same pressure and temperature. The bed martial
2.2. Drag model in this work was kept static under the initial conditions. The wall slip condition for
the gas phase was set as no-slip, and the wall slip condition for the par-
The gas-particle drag force is a critical factor of fluidized beds. The ticle phase was set as partial slip.
Wen–Yu–Ergun drag model, in which the solid concentration changes The MP-PIC method requires the minimum size of the grid to be
from 0.01 to 0.7 [26], conmbines both the functions of Wen–Yu [27] larger than the largest particle. The parcels (computational particles)
and Ergun [28]. In previous studies, the applicability of the Wen– are present in the cell; thus a suitable number and size of cells could en-
Yu–Ergun drag model was experimentally validated [20,22,23]. hance the simulation accuracy. An increase in the number of cells leads
Amarasinghe and Jayarathna et al. [22,23] successfully employed the to a higher prediction accuracy and increased computational costs. Four
Wen–Yu–Ergun drag model for the prediction of the minimum fluidiza- different computational domains of 1250, 10,000, 49,100 and 80,000
tion velocity for different particle sizes about Geldart A, B, and D. The were used to analyze the computational accuracy of the mesh under
drag model used in this study can be expressed as the same operating conditions. Fig. 2 reveals that the pressure decreased
across the fluidized bed along the height in various computational do-
8 mains. The predicted decrease in pressure for the 49,100 and 80,000
>
> D1   εp b0:75εcp
< εp −0:75εcp cells exhibited similar trends and were in a good agreement, whereas
Dp ¼ ðD2 −D1 Þ þ D1 0:75εcp ≥εp ≥0:85εcp ð1Þ the predicted decrease in pressure for the 1250 and 10,000 cells were
>
> 0:85εcp −0:75ε cp
:
D2 εp N0:85εcp significantly different. Considering the computational cost and accuracy
of the simulation, the mesh that consists of 49,100 cells was used in
  this study.
3 ρg ug −up 
D1 ¼ Cd ð2Þ In this study, the simulation was conducted on the commercial code
8 ρp rp
Barracuda virtual reactor (VR). In addition, the Courant–Friedrichs–
84 Y. Shao et al. / Powder Technology 350 (2019) 81–90

velocity, the height of the bed material was static. The particles were
gradually fluidized by gas, and the extent of expansion in the bed in-
creased after the superficial gas velocity exceeded the Umf. The bubbles
gradually appeared in the region of the bed material when the superfi-
cial gas velocity increased to 0.06 m/s, and then the bubbles rapidly in-
creased in size with an increase in the gas velocity.
For the validation of the model at elevated pressures and tempera-
tures, the minimum fluidization velocities obtained by the simulation
were compared with those obtained from the experiment [11,14]. The
minimum fluidization velocities at different pressures and tempera-
tures are presented in Fig. 5. The simulation results were in good agree-
ment with the significant decrease in the minimum fluidization velocity
due to the increase in the system pressures and temperatures. The max-
imum error of the results between the simulation and experiment were
approximately 8% smaller. Nevertheless, the calculated values and
Fig. 2. Simulated axial pressure distributions with different grid numbers. the general trends of the changes in the temperatures and pressures
were in good agreement with those the experimental data. For other
studies, as mentioned below, the minimum fluidization velocity was de-
Lewy (CFL) was set to automatically adjust the calculation time-steps, to termined using this simulation model.
maintain the stability and accuracy of the simulation. Moreover, the
graphic processing unit (GPU) accelerated code was used to decrease
the calculated time. The bed materials were filled up to a height of 4.2. Effect of temperature and pressure
200 mm. Different materials such as coal, sand, aluminum oxide
(Al2O3), nickel oxide (NiO) and ferric oxide (Fe2O3) were used to evalu- The properties of gas, such as density and viscosity, which have a sig-
ate the effect of density on Umf at different pressures and temperatures. nificant influence on the minimum fluidization velocity, are dependent
The other parameters used for the simulation are listed in Table 4. on the operating conditions. With an increase in temperature, the den-
sity of the gas phase gradually decreased, whereas the viscosity of the
4. Results and discussion gas increased. The density of the gas increased with an increase in pres-
sure. In this part, the different operating conditions were applied with
4.1. Minimum fluidization velocity estimation and verification the same particle diameters and mono-disperse size distributions, to
evaluate the variation in the minimum fluidization velocity.
After the injection of the upward flow, the gravitational force of the Fig. 6 presents the variation in Umf in accordance with an increase in
particle was offset by the drag force from the moving gas. The decrease temperature for various pressures. From the figure, it can be seen that
in pressure in the fluidized bed was proportional to gas velocity during the minimum fluidization velocity decreased with temperature. As the
the stage of the fixed bed. Once the bed reached the fluidized bed, the temperature increased, there was a gradual reduction in the decrease
pressure drop maintained the static bed pressure. The point, as an inter- rate of Umf, and Umf tend to constant. Moreover, the variation in Umf at
section of the fitted line of the pressure drop in the flow pattern of the different temperatures for higher pressures (2–4 MPa) was smoother
fixed bed and that of the static bed pressure, was determined as the than that for lower pressures. At a high pressure, there was a slight ini-
minimum fluidization velocity. Therefore, the Umf could be determined tial increase in the minimum fluidization velocity, and then a constant
by measuring the change in the pressure drop across the bed in accor- decline with an increase in temperature.
dance with a decrease in the gas flow rate [29]. The graph of the pres- The small particles (Geldart B) started fluidization in the viscous
sure drop across the bed with different gas flow velocities was flow regime at atmospheric temperature and pressure, and the gas vis-
employed to determine the Umf, as shown in Fig. 3. cosity has a significant influence on the fluidization in the bed [30].
Fig. 4 presents the change in the particle volume fraction with super- Therefore, the decrease in Umf was due to the increase in the gas viscos-
ficial gas velocity at a temperature of 373 K and pressure of 101,325 Pa. ity in accordance with an increase in temperature. However, the in-
Before the operating gas velocity exceeded the minimum fluidization crease in the gas density was not negligible at higher pressures. The
combined effects of a higher gas density and higher viscosity led to
Table 4
Simulation parameters.

Parameters Value

Operating conditions
Particle density, ρp (kg/m3) 1300, 2590, 3970, 5242, 6670
Temperature, T (K) 293–1073
Pressure, P (MPa) 0.1–4

Simulation parameters
Normal-to-wall retention coefficient 0.3
Tangent-to-wall retention coefficient 0.99
Maximum momentum redirection from collision 40%
Ps constant (Pa) 1
B constant 3
Eps constant 1 × 10−8
Max volume residual 1 × 10−7
Max pressure residual 1 × 10−6
Max velocity residual 1 × 10−7
Max energy residual 1 × 10−6
Drag model Wen-Yu-Ergun
Turbulence model LES
Fig. 3. Determination of the minimum fluidization velocity.
Y. Shao et al. / Powder Technology 350 (2019) 81–90 85

Fig. 4. Particle volume fraction at different superficial gas velocity (dp = 0.163 mm, ρp = 2820 kg/m3, T = 373 K, P = 101,325 Pa).

the slight initial increase in the minimum fluidization velocity, and thus, the change in pressure had a negligible influence on the minimum
followed by a constant decrease, as the temperature increased. fluidization velocity.
The minimum fluidization velocity gradually decreased with an in- As shown in Fig. 7, in general, the minimum fluidization velocity at
crease in pressure at the same temperature, as shown in Fig. 6. The de- higher pressures and temperatures was significantly lower than that
crease rate of the minimum fluidization velocity reduced with an at normal pressure and temperature. The minimum fluidization velocity
increase in pressure. At higher temperatures, an increase in the gas vis- gradually decreased with an increase in the temperature and pressure.
cosity and decrease in the gas density mitigated the effect of pressure;

4.3. Effect of particle density

Fluidized beds have been widely proposed as reactors for chemical


looping combustion (CLC), coal, and biomass combustion. There are
therefore many types of particles with different densities in the fluid-
ized bed, such as fuel, oxygen carriers, and sand. However, the solid den-
sity has a significant influence on the minimum fluidization velocity in
the fluidized bed. In this study, different particle densities with the
same particle diameter and initial height of bed material were used to
evaluate the variation in the minimum fluidization velocity. The particle
diameters were kept constant with the monodisperse size distribution.
Fig. 8 presents the pressure drop with the different gas flow veloci-
ties for different particle densities. The minimum fluidization velocity
increased with an increase in the particle density at atmospheric pres-
sure and temperature. Moreover, the pressure drop points followed
the same line during the decreasing gas rate for different particle densi-
ties. The phenomenon was due to the unchanged particle size and gas
properties. The increase in density led to an increase in the mass of
the particles; thus, the turning point was gradually shifted upward.

Fig. 5. Comparison of the results obtained from the simulations and experiments at
(a) different temperatures and (b) different pressures. Fig. 6. Minimum fluidization velocity at different temperatures for different pressures.
86 Y. Shao et al. / Powder Technology 350 (2019) 81–90

Fig. 7. Minimum fluidization velocity at different temperatures and pressures.

The decrease rate of the static-bed pressure gradually increased with an


increase in the particle density.
Fig. 9 presents the minimum fluidization velocities at different
temperatures and pressures. In general, the minimum fluidization ve-
locity exhibited a downward trend with an increase in the tempera-
ture and decrease in the solid density. The minimum fluidization
velocity with higher particle densities was greater at different pres-
sures. Moreover, the variation in the minimum fluidization velocity
gradually decreased at different pressures for particles with lower
densities. The change in the solid density resulted in the change in
the particle weight, and the minimum fluidization velocity was a
point at the equilibrium between the drag force and the mass of the
particles. Therefore, an increase in the particle density required the
greater drag force of the upward fluid. Consequently, the minimum
fluidization velocity increased.

4.4. Effect of particle size Fig. 9. Minimum fluidization velocity for different particle densities at (a) different
temperatures and (b) different pressures.

The particles with different sizes have different fluidized characteris-


tics in the fluidized bed, given that the surface area of the particles has a sizes, whereas the static bed pressure drop was different for different
significant influence on the skin friction. Therefore, the drag force of the particle densities, as mentioned in Section 4.3. This phenomenon was
upward fluid is enhanced with finer particles, which provide a larger due to the change in the drag force and constant mass of the bed. The
surface area; thus decreasing the pressure. The particles flow easier in higher drag force with small particles led to the bed pressure drop ear-
such a case. The sand particles with diameters of 0.18–1 mm with lier reach to static bed pressure drop and the smaller particle can begin
mono-disperse size distribution were employed to evaluate the effects to flow at a lower gas flow rate.
of the particle size on the Umf. As shown in Fig. 11, as the temperature and pressures increased, the
Fig. 10 presents the decrease in pressure with the different gas flow Umf gradually decreased. The minimum fluidization velocity increased
rates for different particle sizes. The decrease in the bed pressure exhib- with the particle size. Due to the lower drag force for large particles, a
ited a linear relationship with the gas velocity in the fixed bed. More- higher superficial gas velocity was required for flow to occur. In addi-
over, the static bed pressure was almost the same for all the particle tion, smaller particles had a smaller variation trend in accordance with

Fig. 8. Minimum fluidization velocity for different particle densities. Fig. 10. Minimum fluidization velocity for different particle sizes.
Y. Shao et al. / Powder Technology 350 (2019) 81–90 87

Fig. 11. Minimum fluidization velocity for different particle sizes at (a) different Fig. 13. Minimum fluidization velocity for different particle size distribution cases:
temperatures and (b) different pressures. (a) different temperatures and (b) different pressures.

the changes in temperature and pressure, which may be due to the large different-sized particles when compared with the mono-disperse
contact areas of small particles. size. The simulation can overcome this limitation by the addition of
different-sized particles to the bed. The different cases of particle
size distributions in this study are presented in Fig. 12. The different
4.5. Effect of particle size distribution
particle size distributions all had mean particle diameters of
0.59 mm. Cases 1 and 2 used the approximate Gaussian-type particle
The particles in pilot-scale and industry are generally not the size
size distribution. Moreover, Case 3 used the flat distribution. Gauthier
of mono-dispersed particles. Therefore, the particle size distribution
et al. [31] proposed that the Umf was used to describe the transition
can influence the minimum fluidization velocity. Moreover, it is
velocity from the regime of the fixed bed, due to the similar hydrody-
difficult to obtain the minimum fluidization velocity in a mixture of
namic behaviour with the particles of mono-disperse size. For the flat-
distribution particles, the complete fluidization velocity (Ufc) was
proposed, and the Ufc was higher than the Umf.
From Fig. 13, it can be seen that the minimum fluidization de-
creased with an increase in the pressure and temperature. It
should be noted that the mono-dispersed particles had the highest
minimum fluidization velocity when compared with the particles
with size distribution. Moreover, the minimum fluidization veloc-
ity in Case 2 was relatively close to that of the mono-dispersed
particles. The particles in Case 2 were of the narrow distribution,
and they were concentrated near the mean particle diameter of
0.59 mm. The particle with the wide distribution (Case 1) had a
significantly lower minimum fluidization velocity than those of
the size distribution. This may be due to the greater quantity of
smaller particles in the wide distribution. Bandara et al. [20] pro-
posed that the larger particles were forced by an extra drag force
from smaller particles that fluidized at a lower gas velocity. Rasteh
et al. [33] proposed that the lower particle segregation and the
Fig. 12. Different particle size distributions. small particles initially flow among the larger particles and collide
88 Y. Shao et al. / Powder Technology 350 (2019) 81–90

Fig. 14. Distribution of particles at different superficial gas velocity for Case 3 (ρp = 2820 kg/m3, T = 298 K, P = 101,325 Pa).

to increase the drag force; thus, there were lower minimum veloc- the Umf. It was found that the constants C1 and C2 can be used to replace
ities for Gaussian-type particles. the coefficients of Remf; thus, the equation was expressed as follows:
For the particles with the wide particle distribution, the fine particles
could help the larger particles fluidize at lower gas velocities. However, qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
too many smaller particles may lead to the segregation of the fluidized μ
Umf ¼ C 21 þ C 2 Ar−C 1 ð7Þ
bed. Chiba et al. [32] proposed that three types of mixing occurred in dp ρg
the de-fluidization process with the wide particle-size distribution;
namely, complete mixing, complete segregation, and partial mixing.
Gauthier et al. [31] reported that segregation easily occurs in the cases The two constants C1 and C2 are 33.7 and 0.0408, respectively. Based
of binary and flat distributions. Fig. 14 presents the distribution of parti- on this correlation, P and T were used to modify it at elevated pressures
cles at different superficial gas velocities for Case 3, at atmospheric tem- and temperatures. Moreover, the Wen–Yu revision is expressed
perature and pressure. The smaller particles and larger particles had as follows:
significantly different fluidization conditions. The strong particle segre-
gation led to the significant fluidization of the smaller particles at the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  a  b
upper part, whereas the larger particles were static at the bottom. This μ P T0
Umf ¼ C 21 þ C 2 Ar−C 1 ð8Þ
was because the larger particles require higher gas velocities for fluidi- dp ρg P0 T
zation [33]. Moreover, this could lead to the deterioration of heat and
mass transfer in the bed. Hence, a suitable particle-size distribution is
essential for the operation of fluidized beds. Here, P0 and T0 are 101,325 Pa and 273.15 K respectively. The con-
stants a = 0.00913 and b = 0.10227 were obtained by fitting the simu-
4.6. The correlation for revised correlation lations results. The results of the predictions using the Wen–Yu revision
correlations compared with those of the numerical experimental data
The correlations of Umf were based on the decrease in pressure are presented in Fig. 15. It can be seen that the errors of predicted results
across a packed bed, as predicted by Ergun [28]. The Ergun semi- were less than ±12%. Furthermore, the Wen–Yu revision correlation
empirical equation, which has been widely used for the prediction of had a higher accuracy than the Wen–Yu correlation at elevated
the Umf, can be expressed as

1 1−ε
17:5 Remf 2 þ 150 2 mf Remf ¼ Ar ð4Þ
ϕεmf ϕ ε2 mf

where

dp Umf ρg
Remf ¼ ð5Þ
μ
 
3
dp ρg ρp −ρg g
Ar ¼ ð6Þ
μ2

However, it is inconvenient to determine the value of particle sphe-


ricity Φ and the voidage at minimum fluidization velocity εmf in practical
applications. Wen and Yu [27] proposed a modified Ergun equation, Fig. 15. Comparison of calculated and experimental numerical data using the Wen–Yu
which is one of the most frequently used method for the prediction of revision correlations.
Y. Shao et al. / Powder Technology 350 (2019) 81–90 89

pressure. At high pressures, there was a slight initial increase in


the minimum fluidization velocity, followed by a constant de-
crease, in accordance with an increase in temperature. At higher
temperatures, the change in pressure had a negligible influence
on the minimum fluidization.
(2) There was a gradual increase in the decrease rate of the static-
bed pressure as the particle density increased. With a decrease
in the particle density, the minimum fluidization velocity exhib-
ited a decrease trend. Moreover, the variation in the minimum
fluidization velocity gradually decreased with an increase in
pressure for low-density particles.
(3) With an increase in the particle size, the minimum fluidization
velocity increased. In addition, smaller particles exhibited a
smaller variation trend in accordance with changes in tempera-
ture and pressure.
(4) The particles with the wide size-distribution had a lower mini-
mum fluidization velocity. For the particles with the wide parti-
cle distribution, the fine particles could help the larger particles
fluidize at lower gas velocities. However, an excessively wide
particle distribution could lead to a decrease in the heat and
mass transfer in the fluidized bed.
(5) A revised correlation based on the Wen–Yu correlation was pro-
posed to predict Umf at elevated temperatures and pressures. The
correlation predicted Umf with an error of ±15% when compared
with the existing experimental data.

This model can be used to predict the minimum fluidization velocity


at elevated pressures and temperatures. In future work, several charac-
teristics of fluidized beds such as the bubble size, bubble velocity, and
segregation features can be systematically evaluated under various at-
mospheres and conditions.

Nomenclature
dp Diameter of particle (μm)
Ar Archimedes number
U Velocity
g Gravity (m/s2)
f Distribution function for particle
Fig. 16. Comparison of calculated and existing literature data using the Wen–Yu revision
correlations at (a) elevated temperatures and (b) elevated pressures. Dp Aerodynamic drag function
n Total number of particles
x Location in space (m)
pressures and temperatures. The modified method is applicable for the
V Volume (m3)
calculation of Umf at elevated pressures and temperatures.
Ucf Complete fluidization velocity
Fig. 16 presents the comparison between the calculated and existing
Q Fluidizing gas flow rate
literature data at elevated temperatures and pressures. The relative er-
P Pressure (Pa)
rors between the calculated and various literature data were approxi-
T Temperature (K)
mately ±15%. The revised correlation based on the Wen–Yu
t Time (s)
correlation adequately predicted the Umf at elevated temperatures and
u Velocity (m/s)
pressures.
H Height of bed (m)
Re Reynolds number
5. Conclusions
Greek letters
A CFD model based on the MP-PIC approach was employed to pre-
ε Volume fraction
dict the minimum fluidization velocity at elevated pressures and tem-
εmf Voidage at minimum fluidization
peratures. The simulation results of the minimum fluidization velocity
Φ Particle sphericity
for different temperatures and pressures were validated by the experi-
τg Fluid stress tensor
mental data. Hence, CFD simulations can help predict the Umf under
ρ Density (kg/m3)
complex conditions. Moreover, they are more low-cost, reliable, and
μ Viscosity (m2/s)
less time-consuming than experiments. Besides, the effects of the oper-
μt Turbulent viscosity (m2/s)
ating conditions such as particle size, density, and size distribution on
μl Laminar viscosity (m2/s)
Umf were also analysed using this model. Some of the findings are as
follows:
Subscripts
(1) The variations in the gas density and viscosity play a major role in g Gas
determining the minimum fluidization velocity at super-ambient p Particle
temperatures and pressures. The minimum fluidization velocity cp Close pack
gradually decreased with an increase in the temperature and mf Minimum fluidization
90 Y. Shao et al. / Powder Technology 350 (2019) 81–90

cal Calculated data [15] W. Zhong, A. Yu, G. Zhou, J. Xie, H. Zhang, CFD simulation of dense particulate reac-
tion system: approaches, recent advances and applications, Chem. Eng. Sci. 140
exp Experimental data (2016) 16–43, https://doi.org/10.1016/j.ces.2015.09.035.
[16] H. Xu, W. Zhong, Z. Yuan, A.B. Yu, CFD-DEM study on cohesive particles in a spouted
bed, Powder Technol. 314 (2017) 377–386, https://doi.org/10.1016/j.powtec.2016.
09.006.
Acknowledgments [17] J. Wang, M.A. van der Hoef, J.A.M. Kuipers, CFD study of the minimum bubbling ve-
locity of Geldart a particles in gas-fluidized beds, Chem. Eng. Sci. 65 (2010)
This work was supported by the National Natural Science Founda- 3772–3785, https://doi.org/10.1016/j.ces.2010.03.023.
[18] L. Dong, Y. Zhao, Z. Luo, C. Duan, Y. Wang, X. Yang, B. Zhang, A model for predicting
tion of China (project number: 51876037 and 51576045). ABY is also bubble rise velocity in a pulsed gas solid fluidized bed, Int. J. Min. Sci. Technol. 23
grateful to the Australian Research Council for the partial financial sup- (2013) 227–230, https://doi.org/10.1016/j.ijmst.2013.04.015.
port of this work. [19] C.M. Boyce, A. Ozel, J. Kolehmainen, S. Sundaresan, Analysis of the effect of small
amounts of liquid on gas-solid fluidization using CFD-DEM simulations, AICHE J.
63 (2017) 5290–5302, https://doi.org/10.1002/aic.15819.
References [20] J. Chandimal Bandara, M. Sørflaten Eikeland, B.M.E. Moldestad, Analyzing the effects
of particle density, size, size distribution and shape for minimum fluidization veloc-
[1] A.J. Minchener, Coal gasification for advanced power generation, Fuel 84 (2005) ity with Eulerian-Lagrangian CFD simulation, Proceedings of the 58th Conference on
2222–2235, https://doi.org/10.1016/j.fuel.2005.08.035. Simulation and Modelling 2017, pp. 60–65, https://doi.org/10.3384/ecp1713860.
[2] J. Xie, W. Zhong, Y. Shao, Q. Liu, L. Liu, G. Liu, Simulation of combustion of municipal [21] Y. Xu, T. Li, J. Musser, X. Liu, G. Xu, W.A. Rogers, CFD-DEM modeling the effect of col-
solid waste and coal in an industrial-scale circulating fluidized bed boiler, Energy umn size and bed height on minimum fluidization velocity in micro fluidized beds
Fuel 31 (2017) 14248–14261, https://doi.org/10.1021/acs.energyfuels.7b02693. with Geldart B particles, Powder Technol. 318 (2017) 321–328, https://doi.org/10.
[3] W. Zhong, A. Yu, G. Zhou, J. Xie, H. Zhang, CFD simulation of dense particulate reac- 1016/j.powtec.2017.06.020.
tion system: approaches, recent advances and applications, Chem. Eng. Sci. 140 [22] W. S. Amarasinghe, C. K. Jayarathna, B. S. Ahangama, B. M. E. Moldestad, L.-A.
(2016) 16–43, https://doi.org/10.1016/j.ces.2015.09.035. Tokheim, Experimental study and CFD modelling of minimum fluidization velocity
[4] R.R. Pattipati, C.Y. Wen, Minimum fluidization velocity at high temperatures, Ind. for Geldart a, B and D particles, Int. J. Model. Optim. 7 (2017) 152–156, https://doi.
Eng. Chem. 20 (1981) 705–708, https://doi.org/10.1016/j.ces.2015.09.035. org/10.7763/IJMO.2017.V7.575.
[5] S.C. Saxena, G.J. Vogel, Measurement of incipient fluidisation velocities in a bed of [23] C. Jayarathna, B. Moldestad, L.-A. Tokheima, Validation of results from barracuda
coarse dolomite at temperature and pressure, Trans. Inst. Chem. Eng. 55 (1977) CFD modelling to predict the minimum fluidization velocity and the pressure
184–189. drop of Geldart a particles, Proceedings of the 58th Conference on Simulation and
[6] M. Nakamura, Y. Hamada, S. Toyama, A.E. Fouda, C.E. Capes, An experimental inves- Modelling 2017, pp. 76–82, https://doi.org/10.3384/ecp1713876.
tigation of minimum fluidization velocity at elevated temperatures and pressures, [24] S. Gosavi, N. Kulkarni, C.S. Mathpati, D. Mandal, CFD modeling to determine the min-
Can. J. Chem. Eng. 63 (1985) 8–13, https://doi.org/10.1002/cjce.5450630103. imum fluidization velocity of particles in gas-solid fluidized bed at different temper-
[7] Z. Zhiping, N. Yongjie, L. Qinggang, Effect of pressure on minimum fluidization ve- atures, Powder Technol. 327 (2018) 109–119, https://doi.org/10.1016/j.powtec.
locity, J. Therm. Sci. 16 (2007) 264–269, https://doi.org/10.1007/s11630-007- 2017.12.026.
0264-2. [25] A. Anantharaman, R.A. Cocco, J.W. Chew, Evaluation of correlations for minimum
[8] M. Jiliang, C. Xiaoping, L. Daoyin, Minimum fluidization velocity of particles with fluidization velocity (U) in gas-solid fluidization, Powder Technol. 323 (2018)
wide size distribution at high temperatures, Powder Technol. 235 (2013) 454–485, https://doi.org/10.1016/j.powtec.2017.10.016.
271–278, https://doi.org/10.1016/j.powtec.2012.10.016. [26] D. Gidaspow, Multiphase Flow and Fluidization, Academic Press, 1994http://gen.lib.
[9] B. Singh, G.R. Rigby, T.G. Callcott, Measurement of minimum fluidisation velocities at rus.ec/book/index.php?md5=5ccad0ef05050a782c1bd3f130cf02f2, Accessed date:
elevated temperatures, Trans. Inst. Chem. Eng. 51 (1973) 93–96. 4 June 2018.
[10] J.H. Goo, M.W. Seo, S.D. Kim, B.H. Song, Effects of temperature and particle size on [27] C.-Y. WEN, Mechanics of fluidization, Chem. Eng. Prog. Symp. Ser. 62 (1966)
minimum fluidization and transport velocities in a dual fluidized bed, in: G. Yue, 100–111.
H. Zhang, C. Zhao, Z. Luo (Eds.), Proceedings of the 20th International Conference [28] S. ERGUN, Fluid flow through packed columns, Chem. Eng. Prog. 48 (1952) 89–94.
on Fluidized Bed Combustion, Springer Berlin Heidelberg, Berlin, Heidelberg 2009, [29] D. Kunii, O. Levenspiel, H. Brenner, Fluidization Engineering, 2nd ed., 1991.
pp. 305–310, https://doi.org/10.1007/978-3-642-02682-9_43. [30] B. Formisani, R. Girimonte, L. Mancuso, Analysis of the fluidization process of parti-
[11] H.J. Subramani, M.B. Mothivel Balaiyya, L.R. Miranda, Minimum fluidization velocity cle beds at high temperature, Chem. Eng. Sci. 53 (1998) 951–961, https://doi.org/10.
at elevated temperatures for Geldart's group-B powders, Exp. Thermal Fluid Sci. 32 1016/S0009-2509(97)00370-9.
(2007) 166–173, https://doi.org/10.1016/j.expthermflusci.2007.03.003. [31] D. Gauthier, S. Zerguerras, G. Flamant, Influence of the particle size distribution of
[12] R. Girimonte, B. Formisani, The minimum bubbling velocity of fluidized beds oper- powders on the velocities of minimum and complete fluidization, Chem. Eng. J. 74
ating at high temperature, Powder Technol. 189 (2009) 74–81, https://doi.org/10. (1999) 181–196, https://doi.org/10.1016/S1385-8947(99)00075-3.
1016/j.powtec.2008.06.006. [32] S. Chiba, T. Chiba, A.W. Nienow, H. Kobayashi, The minimum fluidisation velocity,
[13] P.A. Olowson, A.E. Almstedt, Influence of pressure on the minimum fluidization ve- bed expansion and pressure-drop profile of binary particle mixtures, Powder
locity, Chem. Eng. Sci. 46 (1991) 637–640, https://doi.org/10.1016/0009-2509(91) Technol. 22 (1979) 255–269, https://doi.org/10.1016/0032-5910(79)80031-5.
80023-R. [33] M. Rasteh, F. Farhadi, G. Ahmadi, Empirical models for minimum fluidization veloc-
[14] S.H. Kang, Y. Kang, K.H. Han, G.T. Jin, Effects of pressure on the minimum fluidization ity of particles with different size distribution in tapered fluidized beds, Powder
velocity and bubble properties in a gas-solid fluidized bed, J. Ind. Eng. Chem. 10 Technol. 338 (2018) 563–575, https://doi.org/10.1016/j.powtec.2018.07.077.
(2004) 330–336.

You might also like