4236
Ind. Eng. Chem. Res. 2007, 46, 4236-4244
Four Ways To Introduce Structure in Fluidized Bed Reactors J. Ruud van Ommen,* John Nijenhuis, Cor M. van den Bleek, and Marc-Olivier Coppens Delft UniVersity of Technology, DelftChemTech, Julianalaan 136, 2628 BL Delft, The Netherlands
Just like for fixed bed reactors, the rational structuring of fluidized beds is interesting from the point of view of process intensification. Structuring can facilitate scale-up and increase conversion and selectivity by controlling the size and the spatial distribution of the bubbles. We present four approaches to structure fluidized beds: oscillating the gas flow, imposing an electric field to induce interparticle forces, distributing the gas injection, and optimizing distributed particle properties such as particle size. Experimentally, we show that these methods can indeed lead to a drastic reduction of the bubble diameter in bubbling fluidized beds for a range of operating conditions. For two methodologies, imposing an electric field and distributing the gas injection, also a considerable increase in conversion is demonstrated. Introduction At present, many catalyzed gas-phase reactions in the chemical industry are carried out in packed beds. These reactors are relatively easy to design and operate (i.e., investment and operational costs are low), but disadvantages include maldistribution, a large pressure drop over the bed, and sensitivity to fouling by dust.1 A prominent tradeoff in packed beds is related to the particle size. Small particles are often desired for maximal catalyst effectiveness, which impacts activity, selectivity, and stability alike. However, smaller particles also lead to an increased pressure drop. At some point, the pressure drop simply becomes too high for practical purposes.2 It is expected that catalysts will become more efficient in the coming years, because of the new opportunities offered by molecular modeling, the use of high-throughput techniques in catalysis engineering, and advances in synthesis methods.3-5 This will lead to even more severe mass and heat transfer problems in packed beds. Therefore, a shift from beds of randomly packed particles to reactors with shorter diffusion lengths will be needed. The use of structured packings, such as monoliths, is one way to tackle these problems. Structuring reaction environments introduces extra degrees of freedom allowing decoupling of conflicting design objectives, such as high mass transfer versus low pressure drop. Over the past years, Jacob Moulijn and his co-workers have played a leading role in the research on structured packings and monoliths in particular.6-13 This paper pays homage to Professor Moulijn on the occasion of his retirement from Delft University of Technology. Fixed bed reactors are less suitable for processes in which regular catalyst replacement is needed, for example, because of catalyst deactivation. Moreover, it is not always possible to transfer enough heat to or from these reactors. In those cases, a mobile catalyst phase is desirable, and gas-solid fluidized beds are often a good choice. Normally, fluidized beds are chaotic systems, but we will show that, as for packed beds, it is possible to introduce a regular structure in fluidized beds as well. Despite their good heat transfer and short diffusion lengths, traditional (i.e., unstructured) fluidized beds have serious drawbacks, the two most important ones being their difficult scale-up14-16 and their inefficient use of reactant gas as a result of the formation of bubbles.17,18 The rather small particles * To whom correspondence should be addressed. Phone: +31 15 278 2133. Fax: +31 15 278 5006. E-mail:
[email protected].
(diameters on the order of 100 µm) used in fluidized beds almost always ensure the absence of mass transfer limitations inside the particles. At larger scales, however, mass transfer limitations are present in many fluidized bed processes, especially in the bubbling but also in the turbulent regime. Because of the mass transfer resistance between the dense and the dilute phase (the voids or “bubbles”), the reactant conversion in the dilute phase is much lower than in the dense phase (see Figure 1). This leads to a low overall conversion, because for fluidized beds of small particles the dilute phase conversion dictates the overall conversion. A reduction in bubble size and/or an increase in particle content of the bubbles will lead to a much smaller difference in conversion between the two phases and a much higher overall conversion, as is illustrated by calculations with the Kunii and Levenspiel19 model for a simple first-order reaction (Figure 1). Moreover, in case of a more complex reaction scheme (as in almost every realistic situation) the narrower residence time distribution resulting from a lower concentration difference between the phases typically leads to a higher selectivity toward the desired product. There is therefore plenty of motivation to manipulate the fluidized bed hydrodynamics to introduce more structure and reduce the bubble size. This simplifies scale-up and increases interphase mass transfer, typically leading to higher conversions and selectivity. This can be done either by modifying the gas supply or by interfering in the particle phase. In both cases, either the dynamics can be changed or the configuration can be altered. This yields a total of four different possibilities (see Figure 2): (1) oscillating the gas supply; (2) varying the interparticle forces, which is conveniently done using an alternating electric field; (3) distributing the gas supply over the height of the bed; and (4) varying the particle size distribution and other distributed particle properties. This paper briefly illustrates each of these four approaches. Dynamic Operation of the Gas Supply Conventionally, a fluidized bed is operated at constant gas flow. Over long times this value can be adapted, for example, to change the production level. Over very short times, fluctuations might take place due to variations upstream. Typically, varying the gas flow is not used to influence the hydrodynamics in a fluidized bed. However, in principle it would be possible to include the gas supply in a control loop acting on short time scales: on the basis of one or more measurements of the fluidized bed behavior, the gas flow rate is adapted to maintain
10.1021/ie061318o CCC: $37.00 © 2007 American Chemical Society Published on Web 03/20/2007
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4237
Figure 1. Predictions by the Kunii and Levenspiel19 model for the conversion (defined per unit of volume) in the dense phase (solid line), in the dilute phase (solid line), and the overall conversion (dots) as a function of the normalized residence time. The conversions are given for a fluidized bed of Geldart A particles. The graphs show the influence of the bubble diameter and the fraction of fines in the dilute phase on the conversions.
Figure 2. Overview of four approaches to introduce structure in gassolid fluidized beds.
a desired state. Some first steps in this direction were made,20-22 but it is very difficult to obtain instantaneous information about the size and position of tens to hundreds of bubbles in an industrial fluidized bed. It seems to be a long way before this kind of closed-loop of feedback control can be applied to industrial installations. An alternative to feedback control is to impose a continuous periodic variation on the gas flow rate. In this way, additional degrees of freedom are obtained: amplitude, frequency, and wave shape can now be chosen. It is known that pulsing the gas,23,24 as well as oscillating the distributor plate,25 can cause considerable changes in the fluid bed hydrodynamics and significantly improve reactor performance.26 Furthermore, chaotic and other strongly nonlinear systems may turn periodic by oscillating a characteristic (order) parameter. This explains ripples on sand beaches, and many other regular patterns seen in nature. Coppens et al.27 showed that by oscillating the gas flow introduced through the porous bottom distributor plate, bubble patterns in fluidized beds may indeed become ordered and periodic. Experiments were carried out in two quasi-two-dimensional beds. The first one is 15 cm wide, 1.0 cm thick, and 43 cm high; the second one is 40 cm wide, 1.5 cm thick, and 43 cm
high. A two-dimensional (2D) bed is convenient for the study of bubble dynamics, because bubbles are clearly visible. A light source was placed behind the 2D beds, to record clear images of the bubble pattern on video. The video recordings were carried out with a frame rate of 25 frames per second; recording times were typically about 2 min. For an air-sand system, regular bubble patterns were observed within a broad range of frequencies (from 2.5 to 7 Hz) and amplitudes (an oscillating component of 0.2 to 0.7 times the gas flow required for minimum fluidization). These patterns are hexagonal: bubbles rise in ordered rows with constant interbubble distance, with each row horizontally shifted with respect to the previous row by half the inter-bubble distance (Figure 3). Above a certain height, the regularity of the patterns is destroyed, as fluctuations in the system start to dominate the hydrodynamics. This is shown in Figure 4: the bubble pattern is regular to a height that is approximately equal to the bed width. The wider the bed, the less the left and right side walls influence the pattern, and the greater the height over which the pattern formation persists. The large front and back wall help in stabilizing the patterns in quasi-2D fluidized beds. Wang and Rhodes28 also observed regular bubble patterns in discrete particle simulations but not as clear as in the experimental systems. Coppens et al.27 observed regular bubble patterns in experiments in 3D cylindrical beds too, but only for bed heights of a few centimeters. These patterns are similar to the patterns seen in even more shallow, vibrated granular layers.29 Up to now, we have not yet succeeded to stabilize such patterns for higher beds. It is important to note that both for 2D and three-dimensional (3D) systems the waves are not a linear resonance phenomenon; the pattern wavelength is not inversely proportional to the driving frequency. In addition, the pattern is formed in a range of frequencies and not at specific frequencies. Also, the ordered patterns in fluidized beds are propagated upward via the rising gas flow, which differentiates these patterns from those observed in vibrated granular layers, where all the energy is transmitted to the particles via the moving bottom plate. As a result, dissipation is much stronger in vibrated granular matter than it is in gas-solid fluidized beds, where it is possible to influence the entire bed dynamics via a change in inlet gas dynamics. Further research is needed to apply this approach to deep 3D beds (i.e., beds considerably higher than a few centimeters) and to optimize the structuring such that a significant reduction in bubble size is reached.
4238
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007
Figure 3. Regular bubble patterns obtained in a quasi-2D bed of 40 cm wide and 43 cm high. The airflow is oscillated at a frequency f ) 3.5 Hz. The constant component of the gas flow is equal to the minimum fluidization gas flow; the amplitude of the oscillating component is half of that. The sequence shows four snapshots out of one period of the bubble pattern (frequency 3.5/2 ) 1.75 Hz).
Figure 4. Bubble patterns obtained in a quasi-2D bed of 15 cm wide and 40 cm high. The airflow is oscillated at a frequency f ) 4.5 Hz. The constant component of the gas flow is equal to 1.4 times the minimum fluidization gas flow; the amplitude of the oscillating component is 0.4 times the minimum fluidization velocity. The sequence corresponds to one period of the bubble pattern (with a frequency of 2.25 Hz) by plotting a dot for each bubble detected.
Electric Fields To Induce Interparticle Forces Interparticle forces have a significant influence on the behavior of fluidized beds.30 Because the gravity and drag forces balance each other in a fluidized bed, interparticle forces, though they have a much smaller magnitude, can play a significant role.31 Manipulating these interparticle forces would provide a way to control the fluidized bed hydrodynamics. Pandit et al.32 recently showed via simulations that by introducing artificial interparticle forces, the behavior in a fluidized bed can be moved from the Geldart B regime to the Geldart A regime. Several researchers have investigated the control of fluidized beds using magnetic fields.33,34 As a result of the extremely high power needed in this approach, it does not seem to be viable for large-scale applications. The use of electric fields is more promising: it has been shown to be effective at a power consumption as low as 50 W/m3.35 By imposing an electric field on a fluidized bed of semi-insulating particles, the particles become polarized (i.e., dipoles are induced on the particles); the net electric charge on each particle remains zero.35,36 These dipoles lead to an interparticle force, which strongly depends on the particle separation distance and the relative orientation of the particles in the electric field, both in magnitude and in direction. Particles with the center-to-center axis aligned with the electric field will attract each other, while particles adjacent to each other in the field will repel one another (see Figure 5). Particles at an angle to each other will experience a torque that
Figure 5. Interparticle forces between polarized particles.
attempts to align them to the field. As a result of the electric field induced interparticle force, the particles in the bed will tend to form strings in the direction of the field. However, the electric field strength should be small enough to allow the free movement of particles; that is, the fluidity of the system must be preserved. By using an alternating current (AC) field instead of a constant (direct current, DC) electric field, this fluidity criterion is met. Experiments to study the effect of an electric field on the bubble size are carried out in a quasi-2D Plexiglas column, 20 × 1.5 × 80 cm. The column was operated within a cabinet controlled at 30 °C. The bed material consisted of monodisperse glass beads, with a diameter of 550 µm and a density of 2400
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4239
Figure 6. Ratio of bubble diameter with and without electric field, as determined by pressure fluctuation analysis at 20 cm above the distributor at 2% and 40% relative humidity, for five different gas velocities. Each subplot shows the bubble reduction ratio (cf. the color scale) as a function of field strength (linear scale) and frequency (logarithmic scale).
kg/m3 (Geldart B); the minimum fluidization velocity of these particles is 0.21 m/s. The settled bed height was 40 cm. The electric fields were introduced in the bed by stringing thin wires through the column, and alternately, both horizontally and vertically, driving these with an AC potential or grounding them (cf. Figure 2). This creates a strongly inhomogeneous field in both the horizontal and in the vertical directions in the column. When no field is applied, it was found that the influence of these wires on the bubble behavior is so small as not to be measurable. Oscillating (AC) electric fields with a frequency of 1-200 Hz and a strength of up to 8 kV/cm were applied. Pressure fluctuations were measured using piezo-electric pressure transducers, Kistler type 7261, at 20 cm above the sintered metal porous distributor and in the plenum. The sensors were connected to the column by 100 mm Teflon tubes (i.d. 4 mm), which were covered with 40 µm mesh wire gauze at the tips to prevent particles from entering.37 The probe tips were fitted flush with the sidewalls. The pressure fluctuations were measured with a frequency of 200 Hz. The bubble size was derived from the pressure fluctuations using a technique described by van der Schaaf et al.38 The incoherent standard deviation obtained in this way has been shown to be proportional to the average bubble diameter at a certain height in a fluidized bed, although a calibration of this value (for example, using video analysis or optical probes) is required to determine absolute bubble sizes.39 In this study, we are interested in the reduction of average bubble size and therefore only consider relative values of the incoherent standard deviation. Figure 6 shows the ratio of the bubble diameter as a function of field strength and frequency to a base measurement without the presence of an electric field. The bubble diameter is derived from pressure fluctuations as described in the previous section; the pressure fluctuations are measured at 20 cm above the distributor plate. The figure shows that a large decrease in bubble size is obtained at higher field strengths (around 5 kV/cm and up). It has been predicted that the interparticle force increases significantly as the humidity increases;35 we indeed see that for
the higher relative humidity the bubble diameter becomes smaller. Especially for the high humidity, lower frequencies lead to smaller bubbles. However, one should be careful: DC fields or very low-frequency AC fields may gridlock the particles and lead to defluidization. With increasing flow rate, the net effect of the electric fields becomes smaller. To demonstrate the influence of bubble diameter decrease on the conversion in electric field enhanced fluidized beds, we carried out experiments with ozone decomposition, a reaction with first-order kinetics. This is a fast reaction, to the degree that the overall reaction rate is limited by the transfer of ozone from the bubbles to the dense phase (emulsion phase). Experiments were conducted in a Plexiglas column with a circular cross section 82 mm in inner diameter and a settled bed height of 300 mm. Two compressed air feed lines were used, one enriched with ozone, the other enriched with water vapor to regulate the relative humidity of the system. Both the air flows and the column are thermostated to a set temperature of 50 °C. The O3 concentration passed over the bed ranged from 40 to 80 ppmv; the relative humidity in the bed was 2-10% (dew point temperature -15 to 17 °C), depending on the operating conditions. Ozone measurements of the reactor feed and product were conducted online with an INUSA IN-2000 UV-absorption analyzer (range 0-100 ppm, precision 0.1 ppm). Pressure fluctuations were measured in a similar way as in the quasi-2D bed, at 8 cm above the distributor plate and in the plenum. To impose the electric field, a continuous wire was strung each 12.5 mm in the vertical direction, creating a grating-like pattern of wire with a spacing of 10 mm. The electrode grating on the next level was rotated by 90°. Each electrode grating thus ran crosswise to its nearest neighbors. The total height of the electrode section was 300 mm. The “odd” electrodes (i.e., at 12.5, 37.5, 62.5 mm, etc.) were the live electrodes; the “even” ones were grounded. Optionally, only the wires in the lowest 10 cm of the bed could be electrified, with the wires in the top 20 cm being inactive (see Table 1). The particles used in the experiments were porous Puralox alumina particles, with a mean
4240
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007
Table 1. Experimental Conditions for Ozone Conversion with Electric Field
experiment
bed height
energized electrode height
1.5umf (humidified) 1.5umf (dry) 2umf (humidified) 2umf (dry) 2umf, impregnated with Fe (humidified)
25 cm 25 cm 25 cm 25 cm 10 cm
10 cm 10 cm 30 cm 30 cm 10 cm
(sieved) particle size of 250-300 µm and a particle density of 900 kg/m3; the minimum fluidization velocity is 3.0 cm/s. For the experiment with iron-impregnated particles, 40 g of the particles were impregnated with 0.5 wt % Fe. Results for the characteristic bubble diameter derived from the pressure fluctuations are presented in Figure 7. When the electric field was applied, a reduction in bubble size was observed in all experiments. The reduction is stronger when the humidity of the system is slightly raised from the default dry condition. Raising the humidity slightly increases the conductivity, making the macroscopic electric polarization of the particles larger. Also, the bubble diameter reduction becomes larger when the electric field is utilized over the whole bed. Figure 7 also demonstrates the effect of the bubble diameter reduction on the O3 decomposition efficiency. In all the experiments performed, a decrease in bubble diameter led to a higher conversion. The optimal conditions are those in which the electric field efficiency is maximized (i.e., with a raised humidity). A greater gain can be achieved when the conversion rate is increased by impregnating the catalyst particles with 0.5 wt % iron. The absolute conversion increased from 11% to 25%, while the conversion gain by the electric field increased from 6% to 13%. This is consistent with modeling results,40 which show that a low reaction rate limits the positive effects of smaller bubbles. The reaction results presented here are just meant to be a proof of principle: in the future, a more extensive set of experiments with an optimized catalyst will be carried out. Moreover, further research will be aimed at increasing the effect at higher gas velocities and developing acceptable electrode configurations for industrial applications. Distributing the Gas Supply over the Height of the Bed In a fluidized bed, the gas is normally introduced via the bottom plate. However, distributing the gas supply over the height of the bed introduces additional degrees of freedom. Some examples of distributed gas supply already exist. In combustion it is not uncommon to apply staged injection of the air.41,42 Moreover, membranes may be utilized to supply (or remove) gas to (or from) a fluidized bed.43,44 To distribute the gas over the bed, Coppens45 proposed to connect all injection points by a hierarchical, tree-like fractal structure. Gas flows from the stem of this tree to all branch tips, spread out over the reactor at optimized locations, where it exits. Via the bottom plate, enough gas is fed to ensure at least minimum fluidization throughout the bed. One important reason to use a fractal design is its intrinsic scalability mimicking nature. When scaling up the tree-like injector, new branching generations are added to serve larger reactor volumes. The fractal injector branches in such a way that the length and diameter of branches of a given generation are the same. In this way, fluid leaves all outlets at the same flow rate, because the hydraulic path lengths or pressure drops from the inlet to all outlets are all the same. For outlets lying in the same horizontal plane, this avoids radial nonuniformity. In the vertical
Figure 7. Decrease in bubble diameter and increase in ozone conversion by imposing an electric field for five different experimental conditions.
Figure 8. Ratio of bubble diameter with and without secondary injection, as determined by pressure fluctuation analysis at 20 cm above the distributor for four different gas velocities.
direction, by spacing the outlets according to a designated pattern, one could compensate for the axial gradients in gas amounts and reactant concentrations. Moreover, the lower flow rate of gas via the distributor leads to smaller bubbles initially. Smaller bubbles are advantageous as they have a larger specific surface area and a lower rise velocity. A probable additional advantage of the secondary injections is that it tends to break up existing bubbles or blow particles apart, leading to an emulsion phase of higher void fraction. Unstable emulsions take time to break up into equilibrated phases, and it is this delay that is exploited.46 Experiments were carried out in a quasi-2D setup similar to the one used for the electric field experiments; the same particles are used. The fractal injector had 16 injection points (similar to the schematic in Figure 2), with the lowest row at 6 cm and the highest row at 14 cm above the bottom plate. Pressure fluctuations were measured at 20 cm above the sintered metal porous distributor and in the plenum, to obtain the relative bubble diameter. Figure 8 shows that for every total flow rate studied the bubble diameter is significantly reduced by feeding part of the gas (the secondary gas flow) via the fractal injector. The larger the fraction of gas that is introduced via the fractal injector, the smaller the bubble diameter. Further research aims at elucidating the mechanism behind the effects of distributed injection to allow a meaningful and directed approach to optimization. To show the influence of the injector on the conversion, experiments with ozone decomposition were conducted in the same quasi-2D column with the same fractal injector configu-
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4241
Figure 9. Decrease in bubble diameter and increase in ozone conversion by increasing the fraction of the flow through the fractal injector. The overall gas velocity is 4umf.
ration. The bed mass consisted of 40 g of 1 wt % iron impregnated porous Puralox alumina particles diluted with 855 g of the same non-impregnated particles to give a settled bed height of 40 cm. The particle size was 250-300 µm, and the particle density was approximately 900 kg/m3; the minimum fluidization velocity is 3.0 cm/s. The temperature of the feed gas and the cabinet in which the column was located were heated to 55 °C. The relative humidity stayed within the range of 1-2% over the course of these experiments. The ozone concentration in the feed was in the range of 60-70 ppmv. The ozone concentrations in the feed and outlet were analyzed with an INUSA 2000 O3 analyzer. Pressure fluctuations were measured just above the distributor plate and at 19 cm height to obtain the relative bubble diameter. The conversion results are presented in Figure 9 for a total flow rate of 4umf. The figure clearly shows that with increasing the part of the flow through the fractal injector, the bubble diameter decreases and the ozone conversion increases. This confirms our expectation that the conversion is dependent upon the bubble size. Further experiments will be carried out as it is expected that even better results can be obtained with an optimized design of the fractal injector. Optimizing the Distributed Particle Properties The particle properties have a strong influence on their fluidization behavior. Geldart’s47 famous regime mapping shows four types of fluidization behavior, depending on the size and density of the particles. Also other properties have a considerable influence on the fluidization behavior: Flemmer et al.48 experimentally showed the influence of particle shape, and Hoomans et al.49 used simulations to show the influence of the coefficient of restitution or elasticity. However, not only the average value of a given particle property matters, but also also its distribution influences the fluidization behavior matters. This is best known for the particle size distribution: the addition of
fines (particles with a diameter < 45 µm50) improves the fluidization behavior and leads to faster mass transfer. Sun and Grace51 showed experimentally that a broader particle size distribution leads to higher conversions for ozone decomposition, and they suggested that this is due to the disproportionate amount of fines in the dilute phase.52 However, the current practice is that fluidized bed particles used for catalyzed gasphase reactions are mainly optimized up to the scale of single particles. Most attention is given to their pore size distribution, such that a high surface area is achieved and that the active sites are easily accessible by the gaseous components. Little attention is paid to the earlier mentioned mass transfer from gas in the dilute phase to the particles in the dense phase, essential to practical fluid bed operation. We recently started research on optimizing the behavior of a gas-solid fluidized bed by tailoring the distributed properties of the particles, such as size, density, shape and elasticity. To be able to measure fluidization characteristic for a large number of mixtures, we developed an automated setup (see Figure 10). Such an approach, similar to the high-throughput screening that is frequently applied in, for example, catalysis and biochemical research, is novel for hydrodynamics research. Because of the rather expensive measurement and data acquisition equipment involved, we do not use multiple fluidized beds to run the experiments in parallel. In the tests, catalyst supports such as silica and alumina are used as particles. Experiments are carried out in two industrially relevant fluidization regimes: bubbling fluidization and turbulent fluidization. In the experiments, pressure measurements, optical probes, and video analysis are used to assess the hydrodynamics. Some preliminary experiments were carried out in a quasi2D Plexiglas column (30 × 2.0 × 80 cm), operated within a cabinet controlled at 25 °C. The bed material consisted of alumina Geldart A particles, the fines concentration of which was increased from 0 to 30%; the median particle diameter was 75 µm at 0% fines. The increase in fines resulted in a decreased minimum fluidization velocity from 4.3 to 2.9 mm/s and a decreased voidage at minimum fluidization due to a better fit of small particles between the larger ones. Figure 11 shows the results for fluidization at 4.3 cm/s. An increasing amount of fines leads to an increased bed expansion. Moreover, the addition of fines results in a decrease in bubble size and a smaller total amount of gas in the bubbles. This should enhance the mass transfer of components in the gas phase to the particle surface. Compared to the previous three approaches, our work on this fourth way of manipulating the fluidized bed hydrodynamics is still in the start-up phase. We will work on further development of the automated screening setup, production of tailored particle mixtures, and scanning of a large number of these mixtures. Moreover, computational fluid dynamics will be used to obtain more insight into the interplay between particles with different properties. In a later stage, experiments in larger scale
Figure 10. Schematic representation of the high-throughput screening procedure. A. Filling the fluidized-bed column. B. Carrying out the measurement program. C. Emptying the column. After C, the whole procedure is automatically repeated with the next batch of particles.
4242
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007
Figure 11. Influence of the fraction of fines in the bed on the bed expansion (compared to the bed height at minimum fluidization) and the decrease in bubble diameter (compared to 0% fines). The superficial gas velocity is 4.3 cm/s.
columns and under reactive conditions will be carried out with the most promising particle mixtures. Application in Practice In the preceding sections, we have illustrated four ways to impose structure on a fluidized bed. In this section, we compare these four approaches and discuss some issues concerning their potential use in practice. When comparing the four ways of structuring presented in this paper, it appears that the type of structure obtained differs. A common fluidized bed has a heterogeneous nature, sometimes referred to as a heterogeneous structure,53,54 in the sense that there is a difference between the dense phase and the voids (bubbling and turbulent fluidization) or between the dilute phase and the clusters (fast fluidization). This heterogeneous nature can be made more homogeneous by, for example, bringing more particles into the voids. Moreover, it can be structured by taming bubble growth. The clearest structuring is obtained in pulsed fluidized beds: a clearly regular structure, namely, a hexagonal pattern, is obtained at the proper conditions in quasi-2D beds, and patterns are also formed in a 3D bed, although stabilization for deeper beds appears difficult, and the bubble size is not noticeably smaller. The obtained change in bubble behavior is likely to influence the solids motion: bubbles are known to play an important role in particle mixing in fluidized beds.55-57 Generally, it is desirable to maintain good particle mixing to ensure good heat and mass transfer and to have uniform bed characteristics. Qualitatively, the particles are observed to be well mixed for all four structuring approaches; however, the mixing has not yet been quantitatively studied. Especially for electric fields, it is important not to “freeze” the bed by applying a DC field or too high a field strength. Secondary injection of gas and the application of electric fields involve the use of internals in the bed. Even when they are inactive (i.e., no secondary injection and no electric field), internals can already lead to a reduced bubble size. However, Kleijn van Willigen et al.46 showed that the effect of inactive internals is small compared to the influence of the active internals. A problem with using internals in a fluidized bed might be their limited lifetime: a fluidized bed typically has a strongly eroding effect. Especially for the thin wires used in the electric field experiments, one would expect short life times. Nevertheless, the reduced bubble size reduces local shear rates, which should lead to less erosion. Furthermore, preliminary experiments show that thin wires live much longer than expected from typical erosion rates for tubes, even in the absence of a field. The wires, which have a thickness comparable to the particle diameter, probably suffer less from attrition because particles
move around them. In practice, the electric field may also be introduced using (adaptations of) existing internals (e.g., tubes for heat exchange) or electrodes that are more robust. The fourth approach presented, tailoring the particle properties, has the advantage that no modifications to the fluidized bed equipment are needed. However, changing the particle properties is not always desirable or possible. This method shows potential for fluidized bed reactors in which the particles catalyze the reaction of gaseous component(s) to gaseous product(s) but will be of less use in applications in which the particles are the desired product, such as drying, coating, and polyolefin production. When considering the use of a novel technique to increase conversion, it is important to compare it with other alternatives. Some of these alternatives, such as vibrated beds and magnetic fields, have already been briefly mentioned. Another option is to move to a different fluidization regime. For example, increasing the gas velocity to operate in the turbulent regime instead of the bubbling regime is also a way to increase mass transfer from voids to the dense phase. In fact, many industrial fluidized bed reactors are operated in the turbulent regime. Therefore, it will be useful to compare the improved operation of structured bubbling beds to unstructured turbulent beds. The structuring of turbulent fluidized beds is also a worthwhile topic of investigation. Finally, it may be noted that some of the methods discussed here could extend to other multiphase reactors, such as (slurry) bubble columns. While the physics are different, similar strategies might be used to achieve better reactor performance and easier scale-up. Conclusions Not only fixed beds but also fluidized beds may be structured, with the objective to improve their performance. We presented four approaches for introducing structure in gas-solid fluidized beds. These approaches should facilitate scale-up, which often is still troublesome for fluidized beds. Moreover, the bubble size can be decreased, which, for most catalytic fluidized bed processes, leads to increased conversion and selectivity, because mass transfer limitations between the bubble and the emulsion phase can be reduced or removed. Better hydrodynamic control is achieved by introducing additional degrees of freedom. This can be done either by manipulating the gas or the particles: 1. By varying the gas supply in time, we showed that an oscillating gas flow leads to regular bubble patterns. 2. Using an electric field, semi-insulating particles can be polarized. This induces interparticle forces, resulting in a significant decrease in bubble size. 3. To distributing the gas supply over space by using a fractal injector, a much more even distribution of the gas is obtained, leading to a strong reduction in bubble size. 4. By tailoring the size distribution and other distributed properties of the particles, the bubble size can be decreased and the bed voidage increased. Investigations to further develop and optimize these four approaches are ongoing. Acknowledgment Current and former group members are thanked for their contributions to this research: in particular, F. Kleijn van Willigen, D. O. Christensen, G. B. Schmit, S. Baltussen, Y. Cheng, and M. A. Regelink. Prof. J. van Turnhout is gratefully acknowledged for fruitful discussions concerning the effects of electric fields. Literature Cited (1) Cybulski, A., Moulijn, J. A., Eds. Structured catalysts and reactors, 2nd ed.; CRC Press: Boca Raton, FL, 2005.
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4243 (2) Michiel, T.; Kreutzer, M. T.; Kapteijn, F.; Moulijn, J. A. Shouldn’t catalysts shape up?: Structured reactors in general and gas-liquid monolith reactors in particular. Catal. Today 2006, 111, 111-118. (3) Broadbelt, L. J.; Snurr, R. Q. Applications of molecular modeling in heterogeneous catalysis research. Appl. Catal., A 2000, 200, 23-46. (4) Senkan, S. Combinatorial Heterogeneous Catalysis - A New Path in an Old Field. Angew. Chem., Int. Ed. 2001, 40, 312-329. (5) Ying, J. Y.; Mehnert, C. P.; Wong, M. S. Synthesis and applications of supramolecular-templated mesoporous materials. Angew. Chem., Int. Ed. 1999, 38, 56-77. (6) Cybulski, A.; Moulijn, J. A. Modeling of heat-transfer in metallic monoliths consisting of sinusoidal cells. Chem. Eng. Sci. 1994, 49, 1927. (7) Smits, H. A.; Stankiewicz, A.; Glasz, W. C.; Fogl, T. H. A.; Moulijn, J. A. Selective three-phase hydrogenation of unsaturated hydrocarbons in a monolithic reactor. Chem. Eng. Sci. 1996, 51, 3019-3025. (8) Lebens, P. J. M.; vanderMeijden, R.; Edvinsson, R. K.; Kapteijn, F.; Sie, S. T.; Moulijn, J. A. Hydrodynamics of gas-liquid countercurrent flow in internally finned monolithic structures. Chem. Eng. Sci. 1997, 52, 3893-3899. (9) Cybulski, A.; Stankiewicz, A.; Albers, R. K. E.; Moulijn, J. A. Monolithic reactors for fine chemicals industries: A comparative analysis of a monolithic reactor and a mechanically agitated slurry reactor. Chem. Eng. Sci. 1999, 54, 2351-2358. (10) Heiszwolf, J. J.; Engelvaart, L. B.; van den Eijnden, M. G.; Kreutzer, M. T.; Kapteijn, F.; Moulijn, J. A. Hydrodynamic aspects of the monolith loop reactor. Chem. Eng. Sci. 2001, 56, 805-812. (11) Nijhuis, T. A.; Beers, A. E. W.; Vergunst, T.; Hoek, I.; Kapteijn, F.; Moulijn, J. A. Preparation of monolithic catalysts. Cat. ReV.-Sci. Eng. 2001, 43, 345-380. (12) Nijhuis, T. A.; Dautzenberg, F. M.; Moulijn, J. A. Modeling of monolithic and trickle-bed reactors for the hydrogenation of styrene. Chem. Eng. Sci. 2003, 58, 1113-1124. (13) Kreutzer, M. T.; Bakker, J. J. W.; Kapteijn, F.; Moulijn, J. A.; Verheijen, P. J. T. Scaling-up multiphase monolith reactors: Linking residence time distribution and feed maldistribution. Ind. Eng. Chem. Res. 2005, 44, 4898-4913. (14) Knowlton, T. M.; Karri, S. B. R.; Issangya, A. Scale-up of fluidizedbed hydrodynamics. Powder Technol. 2005, 150, 72-77. (15) Sanderson, J.; Rhodes, M. Bubbling fluidized bed scaling laws: Evaluation at large scales. AIChE J. 2005, 51, 2686-2694. (16) van Ommen, J. R.; Teuling, M.; Nijenhuis, J.; van Wachem, B. G. M. Computational validation of the scaling rules for fluidized beds. Powder Technol. 2006, 163, 32-40. (17) Chavarie, C.; Grace, J. R. Interphase mass transfer in a gas fluidized bed. Chem. Eng. Sci. 1976, 31, 41-749. (18) Kai, T.; Tsutsui, T.; Furusaki, S. Features of fluidized catalyst beds for proper design and operation of catalytic reactions. Ind. Eng. Chem. Res. 2004, 43, 5474-5482. (19) Kunii, D.; Levenspiel, O. Fluidization Engineering, 2nd ed.; Butterworth-Heinemann: Boston, 1991. (20) Kaart, S.; Schouten, J. C.; van den Bleek, C. M. Improving conversion and selectivity of catalytic reactions in bubbling gas-solid fluidized bed reactors by control of the nonlinear bubble dynamics. Catal. Today 1999, 48, 185-194. (21) de Korte, R. J.; Schouten, J. C.; van den Bleek, C. M. Controlling bubble coalescence in a fluidized-bed model using bubble injection. AIChE J. 2001, 47, 851-860. (22) Harrison, A. J. L.; Lim, C. N.; Gilbertson, M. A. Modelling and control of the dynamics of bubbling fluidized beds. Proc. Inst. Mech. Eng., Part E: J. Process Mechanical Engineering 2006, 220, 43-53. (23) Massimilla, L.; Volpicelli, G.; Raso, G. A study on pulsating gas fluidization of beds of particles. Chem. Eng. Prog., Symp. Ser. 1966, 62, 63-70. (24) Wong, H. W.; Baird, M. H. I. Fluidization in a pulsed gas flow. Chem. Eng. J. 1971, 2, 104-113. (25) Ko¨ksal, M.; Vural, H. Bubble size control in a two-dimensional fluidized bed using a moving double plate distributor. Powder Technol. 1998, 95, 205-213. (26) Pence, D. V.; Beasley, D. E. Chaos suppression in gas-solid fluidization. Chaos 1998, 8, 514-519. (27) Coppens, M.-O.; Regelink, M. A.; van den Bleek, C. M. Pulsation induced transition from chaos to periodically ordered patterns in fluidised beds. Proceedings of the 4th World Congress on Particle Technology (CDROM); Sydney, Australia, July 21-25, 2002; Paper 355. (28) Wang, X. S.; Rhodes, M. J. Pulsed fluidization - a DEM study of a fascinating phenomenon. Powder Technol. 2005, 159, 142-149.
(29) Melo, F.; Umbanhowar, P.; Swinney, H. L. Transition to parametric wave patterns in a vertically oscillated granular layer. Phys. ReV. Lett. 1994, 72, 172-175. (30) Rietema, K.; Piepers, H. W. The effect of interparticle forces on the stability of gas-fluidized bedssI. Experimental evidence. Chem. Eng. Sci., 1990, 45, 1627-1639. (31) Seville, J. P. K.; Willett, C. D.; Knight, P. C. Inter-particle forces in fluidisation: a review. Powder Technol. 2000, 113, 261-268. (32) Pandit, Jai Kant; Wang, X. S.; Rhodes, M. J. On Geldart Group A behaviour in fluidized beds with and without cohesive interparticle forces: A DEM study. Powder Technol. 2006, 164, 130-138. (33) Rosensweig, R. E. Process concepts using field-stabilized two-phase flow. J. Electrost. 1995, 34, 163-187. (34) Hristov, J. Magnetic field assisted fluidization - A unified approach - Part 1. Fundamentals and relevant hydrodynamics of gas-fluidized beds (batch solids mode). ReV. Chem. Eng. 2002, 18, 295-509. (35) Kleijn van Willigen, F.; van Ommen, J. R.; van Turnhout, J.; van den Bleek, C. M. Bubble size reduction in electric-field-enhanced fluidized beds. J. Electrostat. 2005, 63, 943-948. (36) Colver, G. M. An interparticle force model for ac-dc electric fields in powders. Powder Technol. 2000, 112, 126-136. (37) van Ommen, J. R.; Schouten, J. C.; vander Stappen, M. L. M.; van den Bleek, C. M. Response Characteristics of Probe-Transducer Systems for Pressure Measurements in Gas-Solid Fluidized Beds: How to Prevent Pitfalls in Dynamic Pressure Measurements. Powder Technol. 1999, 106, 199-218. Erratum: Powder Technol. 2000, 113, 217. (38) van der Schaaf, J.; Schouten, J. C.; Johnson, F.; van den Bleek, C. M. Non-intrusive determination of bubble and slug length scales in fluidized beds by decomposition of the power spectral density of pressure time series. Int. J. Multiphase Flow 2002, 28, 865-880. (39) Kleijn van Willigen, F.; van Ommen, J. R.; van Turnhout, J.; van den Bleek, C. M. Bubble size reduction in a fluidized bed by electric fields. Int. J. Chem. React. Eng. 2003, 1, paper A21 (http://www.bepress.com/ ijcre/vol1/A21). (40) Kleijn van Willigen, F.; van Ommen, J. R.; van Turnhout, J.; van den Bleek, C. M. Conversion Increase in Electric Field Enhanced Fluidized Beds. 2006, in preparation. (41) Jensen, A.; Johnsson, J. E.; Andries, J.; Laughlin, K.; Read, G.; Mayer, M.; Baumann, H.; Bonn, B. Formation and reduction of NOx in pressurized fluidized-bed combustion of coal. Fuel 1995, 74, 15551569. (42) Lyngfelt, A.; Amand, L. E.; Gustavsson, L.; Leckner, B. Methods for reducing the emission of nitrous oxide from fluidized bed combustion. Energy ConVers. Manage. 1996, 37, 1297-1302. (43) Adris, A. E. M.; Grace, J. R. Characteristics of fluidized-bed membrane reactors: Scale-up and practical issues. Ind. Eng. Chem. Res. 1997, 36, 4549-4556. (44) Patil, C. S.; van Sint Annaland, M.; Kuipers, J. A. M. Experimental study of a membrane assisted fluidized bed reactor for H2 production by steam reforming of CH4. Chem. Eng. Res. Des. 2006, 84, 399404. (45) Coppens, M.-O. Structuring fluidized bed operation in a nature inspired way. In Fluidization XI; Arena, U., Chirone, R., Miccio, M., Salatino, P., Eds.; Engineering Conferences International: New York, 2004; pp 83-90. (46) Kleijn van Willigen, F.; Christensen, D.; van Ommen, J. R.; Coppens, M.-O. Imposing dynamic structures on fluidised beds. Catal. Today 2005, 105, 560-568. (47) Geldart, D. Types of gas fluidization. Powder Technol. 1973, 7, 285-292. (48) Flemmer, R. L. C.; Pickett, J.; Clark, N. N. An experimental study on the effect of particle shape on fluidization behavior. Powder Technol. 1993, 77, 123-133. (49) Hoomans, B. P. B.; Kuipers, J. A. M.; van Swaaij, W. P. M. Granular dynamics simulation of segregation phenomena in bubbling gasfluidized beds. Powder Technol. 2000, 109, 41-48. (50) Yates, J. G.; Newton, D. Fine particle effects in a fluidized-bed reactor. Chem. Eng. Sci. 1986, 41, 801-806. (51) Sun, G.; Grace, J. R. The effect of particle size distribution on the performance of a catalytic fluidized bed reactor. Chem. Eng. Sci. 1990, 45, 2187-2194. (52) Sun, G.; Grace, J. R. Experimental determination of particle dispersion in voids in a fluidized bed. Powder Technol. 1994, 80, 2934. (53) Yerushalmi, J.; Cankurt, N. T. Further studies of the regimes of fluidization. Powder Technol. 1979, 24, 187-205.
4244
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007
(54) Jinghai, L.; Cheng, C.; Zhang, Z.; Yuan, J.; Nemet, A.; Fett, F. N. The EMMS model - its application, development and updated concepts. Chem. Eng. Sci. 1999, 54, 5409-5425. (55) Rowe, P.; Partridge, B.; Cheney, A.; Henwood, G.; Lyall, E. The mechanics of solids mixing in fluidised beds. Trans. Inst. Chem. Eng. 1965, 43, 271-286. (56) Hoffmann, A. C.; Janssen, L. P. B. M.; Prins, J. Particle segregation in fluidized binary mixtures. Chem. Eng. Sci. 1993, 48, 1583-1592.
(57) Eames, I.; Gilbertson, M. A. Mixing and drift in gas-fluidised beds. Powder Technol. 2005, 154, 195-203.
ReceiVed for reView October 14, 2006 ReVised manuscript receiVed February 10, 2007 Accepted February 13, 2007 IE061318O
