Vous êtes sur la page 1sur 4

Quantifying of Flow Regime Probability in a Gas-Solid Fluidized Bed

M. Aghabararnejad, R. Sotudeh-Gharebagh, N. Mostoufi*

Multiphase Systems Research Lab, School of Chemical Engineering, University of Tehran

Abstract
Generalized bubbling-turbulent fluidized bed model was quantified in this work. It has been proposed that
regime transition does not occur in a distinct gas velocity. This means that at a constant superficial gas velocity,
different regimes coexist in a fluidized bed. However, each regime of fluidization has a specific probability at each
gas velocity. At gas velocities lower than transition velocity from bubbling to turbulent (Uc) probability of being in
bubbling regime is higher than turbulent while at higher velocities turbulent regime is dominant. In the present
work, contribution of each flow regime was determined at different gas velocities. Experiments were carried out in a
0.15 m ID fluidized bed. Pressure fluctuations were measured at different gas velocities. Sand particles with average
size of 0.23 mm were used in the experiments. Data were analyzed by fast Fourier transform in order to search for
characteristics of bubbling and turbulent regimes of fluidization. The most noticeable character of bubbling is
bubble itself while it is void in turbulent regime. Probabilities of being in bubbling and turbulent regimes were
evaluated by such analysis and their profiles against gas velocity were obtained.

Introduction at a constant superficial gas velocity. At low gas


Fluidized beds have been used in a wide variety velocities, probability of being in bubbling regime is
of industrial processes such as catalytic cracking, high. By increasing the gas velocity, the probability
combustion and coal gasification. These reactors of bubbling regime decreases while the probability of
operate at different flow regimes, starting from being in turbulent regime increases. In the present
bubbling at low gas velocity to turbulent and then fast work, pressure fluctuation data were analyzed and
fluidization at high gas velocities. Understanding the probability of being in each regime was
regime transition is of great importance for design, experimentally determined at different superficial gas
operation, control and scale-up of these reactors. velocities.
There are different reactor models for each flow
regime. For example, two-phase model for bubbling Experiments
regime and single-phase axially dispersed plug flow The fluidized bed unit used in this work was a
model for turbulent regime. However, there are a few 150 mm ID and 2 m high Plexiglas column shown
models that are extended for a wide range of velocity. Fig. 1. Compressed air was used as the fluidizing gas.
Thompson et al. [1] presented a model valid for both Superficial gas velocity in the experiments varied
bubbling and turbulent fluidized bed reactors. Abba from 0-1.32 m/s. Table 1 shows the operational
et al. [2] extended the previous model to fast condition used in this study.
fluidization regime. In both these models,
probabilistic averaging concept was used for Table 1: operating conditions in experiments
predicting model parameters. Constantineaua et al. Property Units Value
[3] proposed a model that predicts probability of each
bubbling and slug flow regime at different bed Initial bed height (cm) 10
heights. They employed ratio of bubble diameter to Superficial gas velocity (m/s) 0-1.3184
column diameter for probability of slug regime. Pressure (bar) 1-1.5
Numerous researchers have used pressure Temperature (οC) 25-30
fluctuation measurements to characterize the Fluid density (Kg/m3) 1.151-1.756
hydrodynamics of fluidized beds. Fluctuations in
fluidized beds can be induces by various sources such At the bottom of the column, a bubble cap
as bubbles, bed-height oscillation and even the air distributor was used. Air leaving the top of the
supply [4, 5 and 12]. Fourier analysis has been used column was passing through a cyclone to separate the
frequently to obtain the dominant frequencies present entrained solid particles, returning them into the
in a fluidized bed [6, 7]* column while the gas leaves the system. Silica sand
The aim of this work was to quantify the regime particles with average size of 0.23 mm and density of
probabilities introduced by Thompson et al. [1]. It 2640 kg/m3 (Geldart group B) were used as solids.
was assumed that different flow regimes can co-exist Table 2 shows properties of solid particles used in
this work.
*
Corresponding author: mostoufi@ut.ac.ir
Proceedings of the European combustion 2009
Result and discussion
Data processing
There are many phenomena that cause pressure
Table 2: Properties of solids fluctuation in a fluidized bed. These phenomena are
Material Ar Umf (m/s) Uc (m/s) ρS(Kg/m3) movement of bubbles, coalescence and breakage of
Sand 1723.9 0.067 0.95 2640 bubbles, voids, clusters, interaction between particles
and wall and interaction between particles themselves
[5, 9, 15]. Usually, time series data are very
complicated and little useful information can be
achieved directly from them. In the present work, it
was tried to extract more information from pressure
fluctuation time series by applying different methods.

1- Time domain analysis


Standard deviation is a classical measure and a
simple method to analyze time series data. For a
given discrete pressure fluctuation signal, x (t),
standard deviation is defined mathematically as:
1 N
σx = ∑ (x (t ) − x )2
N − 1 t =1
(1)

Where
N
1
x =
N
∑ x (t )
t =1
(2)

Fig. 3 shows variation of standard deviation with


superficial gas velocity. By increasing the gas
Figure 1. Schematics of a fluidized bed velocity, heterogeneity of the bed increases which
causes higher standard deviation until it reaches a
Pressure sensor is mounted on a 3 mm ID steel maximum value [11]. At higher gas velocities, bed
probe with a total length of about 20 cm. The data moves towards a more stable status, thus, standard
were recorded with a 12-bit data acquisition board deviation decreases. Maximum value in Fig. 3
assembled on a PC. The sampling frequency was 400 corresponds to Uc or transition velocity from
Hz and 72000 data are captured during 180 seconds bubbling to turbulent regime [10, 13]. In this case,
in each test [8, 14]. To ensure the validity and Uc=0.95 m/s.
reproducibility of the sampled signals, the
measurements were repeated three times at each 10000
Standard deviation

measuring position. Fig. 2 represents a sample time 8000


series of pressure data at U=0.442 m/s.
6000
4000
2000
0
0 0.5 1 1.5
U (m/s)
Figure 3. Standard deviation of pressure fluctuations

2- Frequency domain analysis


Using frequency analysis, more phenomena in the
fluidized bed can be identified. For a given discrete
pressure fluctuation signal, x (t), fast Fourier
transform (FFT) is defined as:

∫ x (t )e
−2π it ω
Figure 2. Time series of pressure fluctuations at F (ω ) = dt (3)
U=0.442 m/s −∞

2
As shown in Eq. (2), Fourier transform is defined As expected, the maximum value occurs at
on the total range of time. Fig. 4 shows the fast Uc=0.95 m/s which is the transition velocity from
Fourier transform of the signal x (t) at U=0.442 m/s. bubbling to turbulent.

2500
2000

Energy
1-2 Hz
1500 2-3 Hz
3-4 Hz
1000 4-5 Hz
5-6 Hz
500
Figure 4. Fast Fourier transform of pressure
fluctuations at U=0.442 m/s 0
0 0.5 1 1.5 2
Energy of a given signal, x (ω), between U (m/s)
frequencies i-j is defines as:
j Figure 6. Variation of energy of sessions with gas
E i − j = ∑ f (ω ) (4) velocity
ω =i
Fig. 6 illustrates the change of energy of each
Where f (ω) is amplitude of the signal x (ω) at frequency session against gas velocity. It can be seen
frequency ω. As mentioned earlier, there are many in this figure that energy of phenomena with
phenomena in a fluidized bed with different frequencies between 2-3 Hz decrease against gas
frequencies. For example, bubbles which are the most velocity and energy of phenomena with frequencies
noticeable characteristics of the bubbling regime and between 3-6 Hz increases against gas velocity. Other
voids for turbulent regime have their own frequency. frequency bonds show no significant change against
By increasing the gas velocity beyond minimum velocity. Therefore, frequencies in the range of 2-3
fluidization, excess gas enters the bubble phase and Hz can be attributed to bubbles while frequencies in
bubbles grow in size. Meanwhile, voids start to the range of 3-6 Hz can be contributed to voids.
occur. This means that with increasing the gas Consequently, probability of bubbling regime (PB)
velocity, energy of bubbles decreases while energy of was defined as the ratio of sessions with descending
voids increases. energy (EDS) to the total session energy (both
In addition to bubbles and voids, there exist other ascending and descending):
phenomena in the bed whose energy changes with
gas velocity. It is clear that with increasing the gas E Ds
velocity, probability of bubbling regime decreases PB = (5)
E A s + E Ds
and probability of turbulent regime increases.
Therefore, phenomena with decreasing energy (e.g.,
bubbles) represent the bubbling regime while E As
PT = (6)
phenomena with increasing energy (e.g., voids) E A s + E Ds
represent the turbulent regime.
In fluidized beds, dominant frequency of the
pressure fluctuations is normally below 10 Hz [6]. Fig. 7 shows probability of bubbling and turbulent
Considering the original signal up to 10 Hz it was regimes at different gas velocities.
broken it into 10 sessions (0-1, 1-2, 2-3, …, 9-10) and
the energy of each signal was calculated at different 0.8
gas velocities. Fig. 5 illustrates the variation of total
signal energy against gas velocity. 0.6
Probability

400000 0.4
300000 PT
0.2
Energy

PB U c =0.95 m/s
200000
0
100000
0 0.5 1 1.5
0 U (m/s)
0 0.5 1 1.5 Figure 7. Probability of bubbling and turbulent
U (m/s)
regime with gas velocity
Figure 5. Variation of total signal energy with gas
velocity

3
As expected, two curves reach each other at Uc group B particles. Chemical Engineering
which means that in the transition velocity Uc, Research and Design. (2008) 1236-1248.
probabilities of bubbling and turbulent regime are [5] E. Piskova, L. Moral. Characterization
equal to each other. of spouted bed regimes using pressure
fluctuation signals. Chemical
Conclusion Engineering Science 63 (2008) 2307 -
In the present study, probabilities of bubbling and 2316.
turbulent fluidization were determined [6] F. Johnson, R.C. Zijerveld, J.C.
experimentally. Pressure fluctuation data were Schouten, C.M. van den Bleek, B.
measured at various gas velocities and analyzed using Leckner. Characterization of fluidization
fast Fourier transform to determine the contribution regime by time serious analysis of
of each fluidization regime at each velocity. It was pressure fluctuation. International
assumed that phenomena with decreasing energy of Journal of Multiphase Flow 26 (2000)
the signal against gas velocity (such as bubbles) 663-715.
represent the bubbling regime and phenomena with [7] T. U. Yang, L. P. Leu. Study of
increasing energy of the signal against gas velocity transition velocities from bubbling to
(such as voids) represent the turbulent regime of turbulent fluidization by statistic and
fluidization. Finally, probabilities of bubbling and wavelet multi-resolution analysis on
turbulent fluidization regimes were quantified and absolute pressure fluctuations. Chemical
plotted against superficial gas velocity. Engineering Science 63 (2008) 1950 -
1970.
Notation [8] J.S. Murguia, E. Campos-Canton.
Ar Archimedes number Wavelet analysis of chaotic time series.
E Energy of a signal Revisits Mexican of Physical 52 (2003)
EAs Energy of ascending signal 155-162.
EDS Energy of descending signal [9] F. Afsahi, R. Sotudeh-Gharebagh, N.
F(ω) Frequency signal Mostoufi. Cluster identification and
N Number of data characterization in a gas solid fluidized
pB Bubbling regime probability bed by wavelet analysis. Canadian
pT Turbulent regime probability Journal of Chemical Engineering (2009).
Uc Transition velocity from bubbling to [10] H. T. Bi, J. R. Grace, K. S. Lim.
turbulent fluidization (m/s) Mechanism of transition from bubbling
Umf Minimum fluidization velocity (m/s) to turbulent fluidization. Chemical
U Superficial gas velocity (m/s) Research and Development 34 (1995)
ω Frequency (Hz) 4003-4008.
x(t) Time series signal [11] S. W. Kim, G. Kirbas, H. T. Bi, C. Jim
Arithmetic average of pressure fluctuations Lim, J. R. Grace. Flow behavior and
x
(Pa) regime transition in a high-density
x(ω) Time series signal circulating fluidized bed riser. Chemical
Engineering Science 59 (2004) 3955-
References 3963.
[1] M. L. Thompson, H. T. Bi, J. R. Grace. [12] R. C. Brown. Pressure fluctuation as a
A generalized bubbling/turbulent-bed diagnostic tool for fluidized beds. (1996)
reactor model. Chemical Engineering Master Thesis.
Science 54 (1999) 2175-2185. [13] D. Falkowski, R. C. Brown. Analysis of
[2] I. A. Abba, J. R. Grace , H. Bi. Variable- pressure fluctuation in fluidized beds.
gas density fluidized bed reactor model Ind. Eng. Chem. Res. 43 (2004) 5721-
for catalytic processes. Chemical 5729.
Engineering Science (2002) 4797-4807. [14] S. Sasic, B. Leckner, F. Johnson. Time-
[3] J.P. Constantineaua, J. R. Grace, C.J. frequency investigation of different
Lima, G.G. Richards. Generalized models of bubble flow in a gas-solid
bubbling-slugging fluidized bed reactor fluidized bed. Chemical Engineering
model. Chemical Engineering Science Journal. 121 (2006) 27-35.
62 (2007) 70 - 81. [15] F. Johnson, G. Larsson, B. Leckner.
[4] C. Sobrino, S. Sanchez-Delgado, N. Pressure and flow fluctuation in a
Garcia-Hernando, M. de Vega. Standard fluidized bed-interaction with the air-
deviation of absolute and differential feed system. Chemical Engineering
pressure fluctuations in fluidized beds of Science 57 (2002) 1379-1392.

Vous aimerez peut-être aussi