Manipulating fluidized beds by using internals
Fluidization with baffles
Alex C. Hoffmann
Stratingh Institute for Chemistry and Chemical Engineering
University of Groningen
Nijenborgh 4
9747 AG Groningen, The Netherlands
a.c.hoffmann@chem.rug.nl
The fluidized bed is a process, which is difficult to manipulate: it either works or it doesn't. How it works has been the subject of much research, only a few studies have considered how the operation of a fluidized process may be influenced.
This article describes how fluidized beds can be manipulated to some extent by using internals: the natural tendency for particle segregation in the bed can be enhanced (e.g. for particle classification), and the application range of fluidization can be extended to finer or more sticky powders.
1. Product technology and fluidized beds.
Industrial products are increasingly engineered for very specific purposes. They have complex structures and tolerance limits are tightening. This is a strong trend for instance in the pharmaceutical, nutritional and coating industries. Producing such products often involves handling and processing components in a powder form. Powdered materials are also important in the recycling industry. Granulation, coating, mixing, classification, grinding and dedusting are processes of interest in this context. For some of these, fluidized beds are an attractive option.
It goes without saying that the more a process can be steered, the better. Different applications put different demands on the process characteristics. This is one of the weakness of fluidized beds: there is not much you can do with them in terms of manipulating the gas or particle flow in the bed. We mention some features of fluidized beds.
The aeration rate is limited to the range between the minimum required to fluidize the particles (Umf) and the rate at which the particles are blown out of the bed. Depending on the bed material, the bed will bubble at a given aeration rate, and a fraction of the gas will effectively by-pass the bed in the fluidization bubbles (see Figure 1). If there is particle in- and outflow ('continuous' beds), the bed will act as an ideal mixer unless the aeration rate is kept very low. If a mixture of particles is present, the heavier and/or larger ones will tend to sink. The severity of this effect depends on the particle properties and the aeration rate.
Whether a material can be fluidized at all is the question: if it is fine or sticky, the bed will be cohesive. It will then tend to form channels through which the aeration gas will escape rather than being dispersed through the interstices supporting the particles. In the other extreme: if the particles are too large and heavy the bed will not fluidize well either, but tend to be very turbulent and form a spout.
Figure 2 shows the well-known powder classification according to fluidization behaviour of Geldart1.
This article will focus on two aspects of fluidized bed operation:
It will be shown that introducing internals into the bed can influence its behaviour to some extent, and the mechanisms behind the effect of the internals will be brought to light.
2. Particle mixing and segregation
As mentioned, heavier and/or larger particles tend to sink (act as "jetsam") in fluidized beds. This can be a problem, since segregation can lead to defluidization in the bottom of the bed, causing costly process downtime.
On the other hand, segregation can be utilized if a particulate product can be brought to collect somewhere in the bed to be withdrawn preferentially. Another use of segregation is for classification of particles, for instance for recycling.
During a research project in Particle and Dispersed Phase Technology in Groningen, where particle dynamics in fluidized beds was investigated, it was found that the natural tendency for segregation could be enhanced by incorporating a series of sieve-like baffles in the bed2,3 (see Figure 1). The baffles had a large open area, and an aperture size much larger than the bed particles. Figure 3 shows the effect of baffles on the jetsam-profile in a fluidized bed containing a mixture of different-sized glass beads. The phenomenon of segregation is non-linear in nature: once it has started, it will lead to a decrease in bubble activity low in the bed, and this will in turn further decrease the mixing.
Initial measurements were made to probe the capability of the system to classify particles. Figure 4 shows that particle classification is possible in a bed consisting of two different types of plastic. For these measurements, two types of plastic (polypropylene en polyamide with densities of 903 kg/m3 1145 kg/m3, respectively) were ground and sieved between two neighbouring standard sieve sizes to obtain particles around 500 m m.
This shows that classification on basis of only small differences in particle density is feasible, even though the process was not optimized for this specific application.
2.1 The mechanism behind the effect of the baffles on segregation
The question arises: what is the effect of the baffles on the particle movement in the bed? Rowe and Partridge4 and Gibilaro and Rowe5 pioneered the idea, that particle motion could be attributed to the following processes:
The main mixing process is the first one. All the processes are dependent on the presence of fluidization bubbles. This is easy to understand for the three former ones, but why should the segregation also only occur in the presence of fluidization bubbles? This was explained by visual observation of the bed using a probe. It turned out that, even when the bed material is supported in the gas stream, the particles form lasting contacts, resulting in a grid-like structure upon which the larger/denser particles, which are not fully supported in the gas stream, can rest. Their migration toward the bottom is made possible by the shearing of this structure when fluidization bubbles pass6.
The particle segregation caused by one fluidization bubble was quantified empirically as a 'segregation distance', made dimensionless with the radius of the fluidization bubble, by Tanimoto et al.7 Their expression was later modified slightly8 to make the segregation distance zero for particles identical to the bulk particles:
(1)
Where r and d are the particle density and size, and subscripts j and b refer to jetsam and bulk particles, respectively. This shows that a difference in density will have a stronger effect than a difference in size.
We determined the effect of the baffles by X-ray photography of the bubbling bed. Fluidization bubbles were made to move from a layer opaque to X-rays into a layer more transparent. Figure 5a shows a series of pictures obtained. The dark wake behind the bubble is clearly distinguishable. In Figure 5b, a baffle has been introduced. It can be seen how the wake material is left underneath the baffle.
Thus the baffles effectively eliminate the most important mixing process in the bed, while the segregation can continue more or less unhindered.
2.2 Considerations in applying the principle, and comparison with other equipment
An obvious application of the baffles is in fluidized processes where it is advantageous to keep some particles in certain regions of the bed, perhaps for continuous withdrawal. One can think of fluidized bed granulation, coating or catalyst regeneration.
The process can also be used as a particle classifier since it will classify particles according to density and/or size. The advantage compared with processes such as windsifting and zig-zag sifting is that the bed of particles is dense, so that much less gas is required per kg of classified material. A disadvantage may be that the classification is less easy to control. Some types of classification now done in liquids, such as plastic classification by flotation in water, may more advantageously be carried out in the baffled fluidized bed.
The relative importance of particle density and size as a driving force for classification is different in the windsifter and the fluidized bed. We can see this in the following way: If Stokes law applies, the terminal particle velocity in the sifter (if it works in a gravitational field) is:
(2)
If two particles (denoted by subscript j and b, respectively) have the same terminal velocity in the windsifter, we get by dividing their terminal velocities and simplifying:
(3)
Thus if the density of particle j is a factor 1.2 higher, the diameter of particle b must be a factor 
higher for the two to behave the same in the windsifter. In the fluidized bed, on the other hand, if the segregation distance in Equation (1) is zero:
(4)
showing that the diameter of particle b will have to be a factor 1.23 higher to compensate for the same ratio in the density. The density is therefore a more dominant driving force in the fluidized bed.
Nevertheless, the size classification may interfere with a density classification, and this has to be carefully thought about when considering a particular application. The classification in Figure 3 between two materials with close densities was obtained after sieving the bed material between two neighbouring standard sieve sizes. Also particle shape may interfere with the classification, an effect that may be utilized for some applications.
When a baffled fluidized bed is used as a particle classifier, the optimal aeration rate is where fluidization bubbles can just be seen in the bottom of the bed. For other processes the fluidization rate may be higher, dictated by other considerations. Then the baffles have to be made open enough for sufficient down-flow through them. If not, an empty region will appear under each baffle, and the bed will loose its coherence and take on the nature of multiple fluidized beds in series.
One other consideration is that the particles have to be fluidizable. As a rule of thumb this means that the particles are limited in size to between around 40 m m to a couple of mm (see Figure 2 for more detail). It is, however, possible to reduce the lower limit by vibrating the baffle module, and this brings us to the second topic of this article.
3. Fluidizability
Outside the particle size range mentioned above, particles are normally not fluidizable. Bringing fine or sticky powders into a fluidized state with a gas is a well known problem. The interparticle cohesion is so strong in such powders that a bed will form channels through which the fluidizing gas will escape, rather than the gas supporting the particles.
Fluidization quality is often characterized by a fluidization index, FI. If the particles are fully supported in the gas stream, the bed pressure drop equals the bed weight (W) divided by the cross-sectional area (A). FI is defined as the actual pressure drop, Dp divided by this:
(5)
If the bed channels, FI will be low, while it will be close to unity if fluidization is ideal. Note, however, that a cohesive bed can rise as a piston in the containing vessel if the aeration is increased from zero. In this case FI can exceed unity for bad fluidization. It is best to test the fluidization quality by FI by decreasing the aeration rate from a high value, so that possible channels have a chance to form initially.
The two techniques used most for breaking interparticle bonds and improving the powder fluidizability are stirring and vibration.
Stirring the bed can break the channels on a large scale, but in some cases the bed material tends still to be agglomerated on a smaller scale, so that gas-solid contacting can remain inferior.
Applying vibration to the bed is known to improve fluidizability and breaking interparticle bonds even on a small scale, but the problem is that the shock waves from the vibration only travel some centimetres into the bed before they are attenuated, leaving much of the bed unimproved.
3.1 Vibrating internals
The problem of bringing the vibration to the heart of the fluidized bed can be overcome by vibrating internals spanning the bed, rather than applying vibration to the bed-containing vessel or the gas distributor plate9. We used a baffle-module similar to the one shown above to cause segregation, but any structure can be used.
We tested this principle on two powders, which would not fluidize in a conventional bed.
The chalk powder is extremely fine, and falls to the left of the chart in Figure 2. The wheat starch powder also falls in the region of C powders. Moreover, starch powders are extra cohesive, so that even when falling in the A region on the chart, they will tend to exhibit group C behaviour.
Curves of FI against superficial fluidization velocity are shown in Figures 6 and 7
In a conventional bed the chalk powder did not fluidize at all, as expected. In fact it could not be aerated at more than about 0.2 mm/s. With vibrating internals, the bed expanded, and the FI rose to a high value, although it remained below unity. This powder is extremely fine and it is likely that it was not broken up into the primary particles even when using the vibrated baffles. Nevertheless, the experiments showed that the bed could be brought into a fluidized-like state using this principle.
A similar plot is seen in Figure 7 for the wheat starch powder. This powder could not be brought to fluidize in a conventional bed either, but it was possible to bring FI to rise toward unity when aerating at a very high velocity. Using the vibrated internals the fluidization quality was good, with an FI of around unity right from Umf, which was calculated to be 0.17 mm/s using the equation of Wen and Yu10.
4. Conclusions
It is possible to manipulate the working of fluidized beds to some extent by introducing internals into the bed.
The particle segregation can be enhanced by reducing axial mixing. This is useful for particle classification, or for fluidized processes where it is desirable to manipulate the particle distribution in the bed, for instance to keep a specific sort of particle in (or out of) a particular zone of the bed.
Vibration can be brought into the heart of the bed by applying the vibration to internals spanning the bed vessel. Powders which are not fluidizable in a conventional bed can be fluidized In this way. The internals can be optimised to a particular application by design.
Applications for fluidization with baffles
Applications for the segregation effect
The applications for the segregation effect of the baffles can be divided in two groups:
a) To control the axial particle transport in fluidized bed processes,
In continuous beds the baffles can reduce the axial mixing and change the particle residence time distribution, for instance in catalyst regenerators or in fluidized bed dryers. If the product particles in a continuous fluidized bed have different physical properties to the feed particles, the baffles can be used to concentrate the product in one section of the bed, for easier and more controllable discharge.
The baffles can also be used to concentrate a certain sort of particles in a section of the bed. For instance in fluidized bed granulation and coating, and in processes where particles are formed, the product uniformity can be improved if the already finished particles are prevented from re-entering the active zone (e.g. the spray zone).
b) As a dry particle classification process.
The baffled fluidized bed will classify particles according to density, size or shape. There are many applications for this in the mineral, agricultural, food, pharmaceutical and recycling industries. Metal powders, polymers, starch and ground waste materials are examples of powders which can be classified for purification or to improve their composition. Advantages compared to wet classification are obvious: energy intensive drying steps and problems with material wettability and solubility are avoided. Compared to windsifting, the air requirement per kg classified material can be orders of magnitude smaller. As shown in the main text, the relative importance of density and size as driving force is different in the fluidized bed, which might be advantageous.
Applications for the improved fluidizability
In practice the main factor limiting the use of fluidization technology is the material fluidizability. For group C behaviour, bad fluidizability is due to powder cohesiveness caused by either:
- a small particle size, for instance for some metal powders and carbon black, or
- particle stickiness.
Stickiness may be an inherent property of the powder, for instance starch or other organic powders, or it may be a feature of the process, such as in fluidized bed polymerization, granulation, coating, or drying.
In all these systems, the principle of vibrating internals is likely to improve the fluidization quality, and extend the range where fluidized bed technology can be used.
Vibrating the internals will break interparticle bonds and may in this way enhance particle segregation in a fluidized bed. When looking at the segregating effects of vibration and aeration separately, they have synergistic effects when the driving force for segregation is a difference in particle density: both will cause heavier particles to move to the bottom. When the driving force is a difference in size, however, they have opposing effects: vibration causes larger particles to move to the top, while aeration causes them to move to the bottom. Judiciously combining aeration and vibration can help in optimising the segregation for a given application.
References