(150g) Identification and Characterization of Meso-Scale Flow Structure in the Dense Gas-Solid Flow in a Fluidized Bed

Gas-solid fluidized bed reactors are of
great importance in the chemical and petrochemical processes, facilitated
by the advantages of simple structure, wide range of fluidized particle size
distribution and high mass and heat transfer efficiency. In many cases, the
gas-solid fluidized bed were not well designed, leading to inefficient
interphase contact and limited mass and hear transfer. One of the important
reasons is the lack of scientific and quantitative description of the meso-scale flow structure.

The description of dense gas-solid flows
in fluidized beds tends to follow the classical two-phase model. It assumes
that both meso-scale structures of bubble and
emulsion phase are uniform, which is not always consistent with experimental
results. Liu proposed that the meso-scale structure
of dense gas-solid flow can be described by a three flow structure model,
involving bubbles with particles inside, emulsion phase and particle
agglomerates, where agglomerate refers to the particle groups having a solid
holdup greater than the solid holdup at initial fluidization. Particles inside
the agglomerates are not fluidized, but as a group, they are fluidized with
emulsion phase.

In the present work, transient bed
density and velocity signals were measured by a specially designed optical
probe. The signals were decoupled by statistical analysis methods to identify
the meso-scale flow structures. The experimental
apparatus is shown schematically in Figure 1. The column was 286 mm in inner
diameter and 5460 mm in height. Air was humidified to avoid electrification in
the bed.. The bed material was a kind of mixed particles
(¦Ñ_p=1560
kg/m³, d_p,av.= 74 ¦Ìm)
of FCC catalyst (95 wt. %) and silica powder (5 wt. %). The solid holdup at the
minimum fluidization was 0.51, based on the plot of pressure drop versus air
velocity. The superficial gas velocity varied from 0.1 m/s to 0.5 m/s, among
which the superficial gas velocity from bubbling fluidization to turbulent
fluidization is 0.4 m/s. A PV-6D Particle Density and Velocity Analyzer,
consisting of a pair of optical probes of an interval of 3.88 mm, was used to measure the local bed density fluctuation and
velocity inside the bed. The relationship between the output signals and the
bed density is nonlinear and was fitted experimentally as

Fig.1
The schematic drawing of the experiment process

(Eq.1)

The
whole flow structures can be characterized by signal moment estimations, such
as mean solids holdup ¦Å_s,
standard deviation ¦Ò, skewness S,
kurtosis K. For analysis, the four
moments was directly calculated from the experimental series. Then, as shown in
Eq.s (2) to (5), the four moment values was used to
back calculate the three key parameters, namely the solid holdup of the dense
phase ¦Å_sd,
the solid holdup of the bubble phase ¦Å_sd and the fraction of the dense phase f_d.

(Eq.2)

(Eq.3)

(Eq.4)

(Eq.5)

Meanwhile,
by choosing a threshold of the bubble phase, the three key parameters (¦Å_sd¡¯, ¦Å_sb¡¯, f_d¡¯) can be statistically obtained from the
probability density distribution (PDD) curves of the solid holdup signals.
Based on traversing method, the precise threshold can be determined when the
following condition was satisfied.

(Eq.6)

Figure
2 illustrates a typical PDD curve which was characterized by a typical bimodal
distribution and divided by a threshold. The peak of low solids holdup
represents the bubble phase, which can be fitted by a log-normal distribution.
(Eq.7) The other peak of high solids holdup corresponds to the dense phase
which follows a Gaussian distribution. (Eq.8)

(Eq.7)

(Eq.8)

Fig.2 Probability density distribution of solids
holdup

The
PDD curve then was divided into three parts by the threshold and the initial
solids holdup ¦Å_s,mf. The three parts represent
three meso-scale flow structures as bubble phase,
emulsion phase and particle agglomerates, respectively. Figure 3 illustrates
the radial distribution of the volume fraction of the bubble phase, emulsion
phase and particle agglomerates at a bubbling bed (0.2 m/s) and a turbulent bed
(0.5 m/s) respectively. The sum volume fraction of the three parts is 1. The
volume fraction of bubble phase keeps stable at the center of the fluidized bed
and decreases at the wall. As the operating gas velocity increases, the volume
fraction of bubble phase increases while the particle agglomerates fraction
decreases.

Fig.3
Radial distribution of the volume fraction of the bubble phase, emulaison phase and particle agglomerates at different flow
pattern

After
determining the threshold of bubbles and particle agglomerates, we need to
decouple the corresponding signals from the original optical signals. The
method is as follows. (Taking
the identification of bubbles as an example)

(1)     Find
the bubble peak along the time axis of the signals. When the value of the
solids holdup signal is less than the threshold value, consider this point A to
present the start of the bubble x,
and then search backward from this point to find the first point B reach the
threshold, if the following 20 consecutive points are all greater than the
threshold, point B is considered to be the end point of the bubble, and the
bubble peak number x, start and end
times T_x1 and T_x2 are recorded.

(2)     Find
the minimum position of the bubble x
in channel 1 and channel 2. n_x1 is the number of data points between the
lowest point of the bubble x and A in
the statistical channel 1, and n_x2
is the number of data points between the lowest point of the bubble x and B. Take n_x1 and n_x2
respectively on both sides of the lowest point of the bubble peak in channel 2.

(3)     When
bubbles move in the direction of the two fibers, two reflected signals with
similar shapes and a certain delay in time will be generated. For the same
bubble, the generated fluctuating signals should be similar. So we compared the
bubble peaks obtained from the two channels. It was found that for a
correlation coefficient greater than 0.7, the pair of bubbles from the two
channels correspond to the same bubble.

Figure
4 shows the flow structure of the bubble phase. The fraction of the bubble
phase increases with the increase of operating gas velocity. At the side wall,
as the number of particle agglomerates increases, it decreases due to the wall
effect. In the center of the fluidized bed, the distribution of solids holdup
seems flat, while it shows an increasing trend at the side wall. The average
solids holdup decreases as the operating gas velocity increases in a bubbling
bed. When the operating velocity continues to increase to
more than 0.3 m/s, the average solids holdup of changes little with the
increase of operating velocity. The standard deviation of solid holdup
of the bubble phase shows an increasing trend in radial direction at each gas
velocity. With the increase of operating gas velocity, it decreases at each
radial position. As the operating gas velocity increases, a large amount of gas
becomes bubbles and the bubble frequency increases. When the gas velocity is
greater than 0.4 m/s, that is, the turbulent bed, the frequency of bubbles
appears to increase significantly. At the side wall, the frequency of bubbles
is significantly reduced due to the side wall effect. As
the operating gas velocity increases, the bubble velocity also tends to
increase and sharply decreases at the side wall. The bubble chord varies from
10-43 mm and is less affected by changes in operating gas velocity.

Fig.4
Flow structure of dilute phase

Figure
5 shows flow structure of the dense phase. In a bubbling bed, the average
solids holdup is evenly distributed along the radial direction. As the
operating gas velocity increases, more gas enters the dense phase and the
average solids holdup of the dense phase decreases. After entering the
turbulent bed, the solids holdup does not change much with the gas velocity.
The standard deviation of the solids holdup shows a decreasing trend from the
center of the bed to the side wall. This is due to the fact that there are more
bubbles in the center than in the side walls, gas exchange is more frequent,
and fluctuations are greater. The fluctuation of the average solids holdup of
the particle agglomerates is very small, basically between 0.54 and 0.59. With the
increase of operating gas velocity, the frequency of agglomerates tends to
increase. In the center of the fluidized bed, it is relatively stable, but
increases at the side wall. The speed of particle agglomerates increases with
increasing operating gas velocity while it gradually decreases at the side
wall. The agglomerates size increases significantly for the gas velocity
greater than 0.3 m/s. The agglomerates chord length is 0.5-1.5 mm in a bubbling
bed and is 1.5-2.9 mm in a turbulent bed.

Fig.5
Flow structure of dense phase

A
new method of decoupling optical fiber fluctuation signals is presented in this
work, suggesting an accurate approach to identify the bubble and the dense
phases from original signals, and proposing the definition and the distinguish
basis of the agglomerates. By using the method, meso-scale
flow structures such as solids holdup, velocity, size and frequency are
quantitatively characterize.