215
F. Snieders and A.C. Hoffmann
Department of Chemical Engineering, University of Groningen, 9747
AG Groningen, The Netherlands
J.G. Yates and D. Cheesman
Department of Chemical and Biochemical Engineering, University
College London, Torrington Place, London WC1E, Great Britain.
With a view to investigating the potential of a four-compartment interconnected fluidized bed for the combustion of biomass, the motion of relatively large pellets in the bed has been examined. Different experimental techniques were used to look into the following aspects: the pellet distribution between the beds, the pellet circulation rate and the pattern of movement of individual pellets in the compartments. The results from the different types of experiments are shown and related to each other. Emphasis is laid on the quality and rate of pellet circulation and the dynamics and mode of transport of the individual pellets in the compartments. Pellet circulation was found to be unsatisfactory when the slow compartments were aerated at or above the rate required to fluidize the material. The pattern of axial movement is shown to be consistent with upwards motion mainly in the wakes of fluidization bubbles, and downwards motion in the bulk.Keywords: interconnected fluidized bed, coarse particle dynamics, particle circulation, positron emission particle tracking, residence time distribution.
Many processes involving fluidized beds feature continuous particle circulation between separate beds. Examples are FCC (fluidized catalytic cracking) and fluidized bed gasification or combustion. Different processes are carried out in the separate beds, and the quality and rate of solids circulation between beds is an important factor determining the efficiency.
The circulation is often brought about by a system of tubes connecting the beds. In risers the solids transport is pneumatic, by blowers, and in downcomers the driving force for solids flow can be gravity, often controlled by inclining the tubes to the vertical.
In the early 1980's, a compartmentalized fluidized bed with internal circulation ('interconnected fluidized bed') was proposed as an alternative process for gasification of solid wastes, coal and biomass 1. The configuration is sketched in Figure 1, and consists of two vigorously fluidized upflowing beds, and two downflowing bubble-free ones. Transport from fast to slow beds takes place over the separating weir, and, originally, from the slow to the fast underneath the separating walls. These walls extend to the roof of the vessel, separating the free-board above the compartments in two. Later 2 the feature was added of orifices low in the separating walls through which the solids could flow from slow to fast beds.
Since the initial concept was introduced, some work has been done
to characterise this sort of system, and investigate its potential
for different applications. A number of possibilities for controlling
the solids circulation have been indentified 3, the
aeration rates of the slow chambers appearing to be the most promising.
Lower aeration rate in the slow compartments gives more vertical
resistance to particle motion, providing a means of controlling
the circulation rate. The potential of an interconnected fluidized
bed for regenerative sulphurization has been investigated 4,
and in the process, the solids circulation was also considered.
Among other things, a radiotracer method was used to study the
particle circulation 5. The method used allowed proximity
to one or other of two detectors placed on opposite sides of the
bed vessel to be inferred, and could show which pair of compartments
the tracer particle was in, but not its detailed motion.
Figure 1 Sketch of the interconnected fluidized bed
In most gasification or combustion processes, the reacting species is in the form of large, relatively light particles (say biomass), fluidized in a medium of smaller, denser particles (perhaps sand). The dynamics and circulation of the reacting species is of paramount impor-tance for the potential of interconnected fluidized beds for the gasification of biomass. To date, no research work has been published exploring this, although it was suggested by the originators of the system 1.
The object of this work is to investigate the dynamics and circulation of large, relatively light, pellets in a fluidization medium consisting of denser, smaller particles.
The bed particles were glass ballotini with mean diameter, ds, 80 mm and particle density, rs, 2460 kg/m3. The minimum fluidization velocity, Umf, was determined experimentally to be 0.49 cm/s, a value close to the one predicted by the Wen and Yu Equation.
The dimensions of the bed vessel are indicated in Figure 1. In many of the experiments, baskets were used to capture the pellets. These baskets covered the cross-section of a compartment, and had as floor a mesh much coarser than the bed particles, but fine enough to capture the pellets.
Two sorts of pellets were used, one produced by extrusion and the other tabletted. Both types were cylindrical and consisted of alpha alumina with length and diameter around 3.0 mm. They had envelope densities, rt, of 1100 kg/m3 and 1300 kg/m3, respectively. After having used the extruded pellet for the initial experiments, it turned that it was unsuitable for positron emission particle tracking (PEPT) since it became brittle in the cyclotron.
Three experimental parameters were varied:
A series of experiments were performed, in which 50 pellets were added to a slow bed at a given time. The system was then allowed to run for varying intervals (multiples of 10 minutes), whereafter the baskets, which had been resting on the distributor plates, were raised to recover the pellets to ascertain their distribution over the compartments. In these experiments Uslow was kept constant at 1.0 times Umf, while Ufast and Mbed were varied.
This type of experiment should give information both about the quality of pellet circulation and a rough indication of the relative residence time in slow and fast beds.
At some combinations of Ufast and Mbed, the pellets never reached uniform distribution, many staying in the chamber to which they were added. Figure 2 shows two sets of results, one with inferior distribution, one where the distribution appears satisfactorily uniform. The results obtained are summarised in Table 1.
Figure 2 Two distribution experiments, one (left) where the quality
of circulation is inferior (many of the pellets stay in the first
and the second compartments), one where it is satisfactory (the
pellets are well distributed over the compartments already after
10 minutes). Uslow=1.0 Umf,
Mbed=11.2 kg, Ufast=8 Umf
(left) and 14 Umf (right).
Table 1 Summary of quality of distribution between the chambers
with Uslow=1.0 Umf. -: inferior,
a large fraction of pellets in the first bed; +: good distribution
between chambers.
| Mbed\(Ufast/Umf) | |||||||
| 10.4 kg | |||||||
| 10.8 kg | |||||||
| 11.2 kg | |||||||
| 11.6 kg |
It is clear from the table that the quality of pellet circulation improves with increasing Ufast and Mbed.
Another way of shedding light on pellet circulation was to add 50 pellets to the surface of a slow bed and catch them at the surface of the same bed having completed one circuit (called 'catch experiments' in what follows). Counting captured pellets every minute provides a cumulative curve of the pellet residence time distribution in the four compartments.
Also here inferior pellet circulation was noticed. Figure 3 shows an example curve under conditions where the circulation is unsatisfactory. Although a large fraction of the pellets complete a circuit fairly quickly, a number appear to 'get stuck' somewhere in the system. Table 2 summarises some results of a series of experiments carried out to find the operational window of satisfactory quality of pellet circulation.
Figure 3 Incomplete pellet circulation in a catch experiment.
Uslow=1.0 Umf, Ufast=12
Umf and Mbed=11.6 kg.
The table shows that the quality of pellet circulation improves
with decreasing Uslow and increasing Ufast.
When Uslow=1.0 Umf, Ufast
has to be increased to very high levels for satisfactory circulation
- so high that it is not viable to ensure full circulation in
that way. Comparing the last column of this table with the last
row of Table 1, it can be seen that the catch experiment is a
more stringent test for the quality of circulation than the distribution
experiment. The latter will also give information about the relative
pellet residence time in the fast and slow beds, if higher statistical
significance can be achieved.
Table 2 Results of exploratory catch experiments. Mbed=11.6 kg.
| Varying Uslow with Ufast=16 Umf | Varying Ufast with Uslow=1.0 Umf | ||
| Uslow/Umf (-) | Complete circulation? | Ufast/Umf (-) | Complete circulation? |
The results above show that the slow compartments have to be aerated below Umf of the bed material for the quality of pellet circulation to be satisfactory. The rest of the experiments were therefore concentrated on the region: Uslow<Umf, in particular: Uslow=0.75Umf.
Figure 4 shows a series of cumulative pellet circulation time curves, obtained by catch experiments, for various Ufast with Uslow=0.75 Umf and Mbed=11.6 kg. The effect of increasing Ufast in increasing the pellet circulation rate is clear.
Each catch experiment was performed three times, and the average results for the mean and spread of pellet circulation time with their standard deviations are shown in Table 3.
Earlier 'twin bed' experiments, in which only two beds were operated, material being added to the slow bed and captured from the fast one, showed the same trends for the circulation rate of the bed material itself. This stands to reason, since with a constant Uslow, increasing Ufast will make the fast beds more lean, increasing the pressure drop over the orifices, and the flow of bed material through them. Also the flow over the weir will increase due to the increased bubble activity. This will be discussed in more detail elsewhere 6.
Figure 4 Rate of pellet circulation at various Ufast.
Uslow=0.75 Umf, Mbed=11.6
kg.
Table 3 Mean and spread of pellet circulation time with their
respective standard deviations. Uslow=0.75Umf;
Mbed=11.6 kg.
| Mean (s) | Spread (s) | |
| 10 | 870±48 | 218±89 |
| 12 | 773±119 | 333±55 |
| 14 | 454±37 | 162±16 |
| 16 | 347±9.0 | 158±22 |
| 18 | 261±59 | 143±11 |
| 20 | 213±21 | 97±12 |
| 22 | 164±10 | 62±2.5 |
In order to determine the character of the movement of the pellets in the bed, Positron Emission Particle Tracking (PEPT) experiments were carried out. A pellet made radioactive in a cyclotron was entered into the system and tracked. The tracking is based on the detection and triangulation of pairs of g-rays. This method allows the position of the pellet to be determined as a function of time. The following operating conditions were used:
The duration of an experiment varied from one to two hours. In this time the pellet completed between 7 and 23 cycles. The total bed mass, Mbed, was always 11.6 kg.
Figure 5 shows the coordinate system used for the PEPT experiments. The y-coordinate is the height, the distributor plate being at y=149 mm, and the top of the weir at y=399 mm.
Figure 6 shows the x-, y- and z-coordinates of the pellet for a short period during the experiment carried out at Ufast=16 Umf and Uslow=0.75 Umf.
The height oscillation of the pellet can be seen to have a relatively long period and a high amplitude, often spanning the entire bed height. The ascent is clearly faster than the descent, giving the impression that the peaks are 'leaning' to the left.
Figure 5 Coordinate system used for PEPT experiments
These features are consistent with the notion of ascent in bubble
wakes. The pellet is most likely caught at the distributor where
bubbles are formed, but may also be caught in the increasing wake
flow higher in the bed. Once caught, it will move upward with
the bubble velocity and often be brought all the way to the bed
surface. Descent is slower and occurs in the bulk, which moves
down compensating for the upwards transport in the wake phase.
Figure 6 x- y- and z-coordinates
of pellet as a function of time. Conditions: Ufast=16
Umf, Uslow=0.75 Umf
and Mbed=11.6 kg.
These ideas can further be tested and the processes quantified in the following way. The size of fluidization bubbles in the fast beds can be estimated from 7:
(1)
and their rise velocity from 8:
(2)
The velocity calculated from this equation, using the bubble size half-way up the bed, is indicated by a broken line in Figure 6. It can be seen to agree very well with the measured pellet ascent velocity.
In order to estimate the velocity of descent of the bulk phase we need to estimate the total flow in the wake phase. The total flow of empty bubble volume can be found from the so-called 'two-phase theory':
(3)
In these relatively long, narrow, beds, the two-phase theory should apply with reasonable approximation. The wake fraction of a spherical cap fluidization bubble (the fraction of the bubble-wake sphere taken up by wake material) can be calculated from the angle subtended from the centre by the wake, which can be estimated from 8:
(4)
Calculating the total flow in the wake phase in the axial middle of the bed from these equations, taking into account the fraction of bed cross-sectional area taken up by the bubble-wake phase, the descent velocity of the bulk can be calculated. This is also indicated in the figure. Agreement with experiment is reasonable, but the measured pellet descent is somewhat faster than the calculated one. This may be an indication that the pellet is segregating with respect to the bulk material, and thus travelling faster towards the bottom.
The point at which the pellet is thrown over the weir and entering the slow bed is also shown. In the slow bed, the pellet descends practically vertically through the bed with only little oscillatory movement, until it is accelerated towards an orifice and enters the subsequent fast bed.
During the distribution and the PEPT experiments the unsatisfactory pellet circulation at Uslow>=Umf was seen to be due to their segregating into less active regions low in the slow beds furthest from the orifice and the weir. This occurred in spite of the fact that the pellets and bed material should be compatible, the pellets neither segregating nor floating. When Uslow was < Umf, the circulation was satisfactory and appeared to be reliable. Under these conditions the less active regions in the bed are likely to be completely dead.
The analysis of the pellet motion in the PEPT experiments shown in Figure 6 has provided strong evidence that the upwards pellet motion mainly is in the wakes of fluidization bubbles, often bringing the pellets right to the surface of the bed, while the downwards movement occurs with the bulk material, probably with some pellet segregation.
The authors are indebted to Dipl.-Ing. M. Stein and Prof. J.P.K. Seville for having made the facilities for the PEPT experiments available, and helping with the experimentation and the interpretation of the data.
1. Kuramoto, M., Kunii, D. and Furusawa, T., 1986, Flow of dense fluidised particles through an opening in a circulation system, Powder Technol., 47: 141-149.
2. Yong, J., Zhanwen, W., Jingxu, Z. and Zhiqing, Y, A study on particle flow between fluidised beds, 1985, in: Fluidisation '85 Science and Technology, pp .172-183.
3. Fox, D., Molodtsof, Y. and Large, J.F., 1989, Control mechanisms of fluidised solids circulation between adjacent vessels, AIChE Journal, 35: 1933-1941.
4. Korbee, R., 1995, Regenerative desulfurisation in an interconnected fluidised bed system, Ph.D. dissertation, Delft University of Technology.
5. Abellon, R.D., Kolar, Z.I., den Hollander, W., de Goeij, J.J.M, Schouten, J.C. and van den Bleek, C.M., 1997, A single radiotracer particle method for the determination of solids circulation rate in interconnected fluidized beds, Powder Technol., 92: 52-60.
6. Snieders, F.F., Hoffmann, A.C., Cheesman D., Yates, J.G., Stein, M. and Seville, J.P.K., 1998, The Dynamics of Large Particles in a Four-Compartment Interconnected Fluidized Bed, Powder Technol. in print.
7. Geldart, D., 1972, The Effect of Particle Size and Size Distribution on the Behaviour of Gas-Fluidised Beds Powder Technol., 6: 201-215.
8. Hoffmann, A.C., 1983, Gas Fluidisation at Elevated Pressures Ph.D. Thesis, Ramsay Memorial Laboratory of Chemical Engineering, University College London.
A cross-sectional area of bed
d particle diameter
D (bubble) diameter (twice the radius of curvature of
a spherical cap bubble front)
f volume fraction
g gravitational acceleration
h height in bed from the distributor plate
M mass
Q volumetric flowrate
U velocity or superficial velocity
x, z lateral coordinates
y height coordinate