Systematic Investigation of Particle Segregation in Binary Fluidized

Article pubs.acs.org/IECR

Systematic Investigation of Particle Segregation in Binary Fluidized Beds with and without Multilayer Horizontal Baffles Yongmin Zhang,†,* Haibo Wang,‡ Lili Chen,† and Chunxi Lu†,* †

State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing, 102249, P. R. China Wuhuan Engineering Corporation, Wuhan, Hubei Province, 430223, P. R. China

‡

ABSTRACT: Particle segregation was systematically studied in a binary fluidized bed of inner diameter 0.286 m with fine FCC and coarse millet particles as flotsam and jetsam, respectively. Segregation efficiencies of both flotsam and jetsam, axial and radial fraction profiles were studied and analyzed systematically. A new developed horizontal baffle was examined to see its effects on particles segregation. Experimental results showed that the baffle-free bed could only get high-purity flotsam particles at very low gas velocities. However, both nearly pure jetsam and flotsam particles were obtainable at much higher gas velocities after four layers of new baffle are installed. A wider operating range suitable for particle segregation and greater axial fraction gradients were found in the baffled bed, further proving the baffle’s effect on promoting particle segregation. This enables the baffled bed to be a promising continuous particle classifier in industry. Further analysis demonstrated that reduced bubble size, improve bubble flow distribution, and suppressed solid mixing are the intrinsic causes.

1. INTRODUCTION Fluidized beds are widely applied in many chemical and physical processes, attracting numerous research interests in both industrial and academic communities. In industrial processes, particles employed are usually not monodisperse particles. They either have multiple components with different densities or sizes, or have wider size distributions. In these fluidized beds, particle segregation and mixing as two competing mechanisms coexist. Sometimes, segregation is undesirable, for example, in some chemical reactors, where poor product yield and selectivity, or even reactor damage may result from uneven bed temperature distribution, defluidization or catalyst agglomeration due to particle segregation. However, segregation is sometimes welcome when the separation of particles of different properties is needed such as in ore beneficiation, removal of ash agglomerates from a fluidized-bed gasifier,1 particle classifier,2 etc. Because of its universality and practical importance, particle segregation in fluidized beds has attracted much research in recent decades. The research on fluidized-bed segregation was pioneered by Rowe and Nienow and their co-workers.3,4 The two terms now widely used in binary fluidized beds jetsam and flotsam were also first proposed by them. In a binary fluidized bed, particle component prone to sinking into the bed bottom is usually called jetsam, whereas component which tends to float toward the bed top is called flotsam. Jetsam is usually composed of denser or coarser particles, while lighter or finer particles usually belong to the flotsam fraction. Unlike other granular segregation mechanisms, for example, those due to vibration, percolation of fines, different repose angles or trajectories,5 segregation in fluidized beds is mainly due to the different drag/weight ratios of particles with different properties.6 In binary fluidized beds, jetsam has larger drag/weight ratios than flotsam, which is the intrinsic source of particle segregation. Bubble is another indispensable factor for particle segregation in gas fluidized beds. Before bubbling, particles in a bed are immobile. With the arrival of a bubble jetsam © 2012 American Chemical Society

collapses into it due to its larger drag/weight ratio. The gas velocity capable of moving flotsam is unable to move jetsam.3 On the other hand, bubble is also the essential factor that governs particle mixing in a fluidized bed. Rising bubbles carry particles upward in their wakes and suck a portion of particle beneath them, that is, the particle in the drift. The residual particles move downward to fill up the left voids, forming an internal particle circulation (usually referred as gulf streaming in literature) and causing strong particle mixing in a fluidized bed. It can be summarized with a parable that, for particle segregation in a fluidized bed, particle property difference is the dynamite, bubble is the fuse, and the extent of particle component classification is the result of a hand wrestling between bubble-induced mixing and segregation. Many factors were reported to influence particle segregation in fluidized beds. Differences in particle density and size are the two fundamental factors leading to particle segregation in fluidized beds. Density difference is much more prominent than size difference on particle segregation, as can be seen from the suggested correlation for mixing index by Rowe et al.4 ⎛ ρj ⎞2.5⎛ d j ⎞0.2 M= = f (u0)⎜⎜ ⎟⎟ ⎜ ⎟ x j,0 ⎝ ρf ⎠ ⎝ d f ⎠ x j,t

(1)

Here, M is defined as the ratio between jetsam fraction in bed top, xj,t, and the jetsam fraction under complete mixing, xj,0. Gas velocity is one of the most important factors that governs particle segregation. In most studies, segregation degree is reported to decreasing with increasing (u0 − umf).7−9 Rowe et al.7 reported that there exists a critical gas velocity uTO, beyond which particle mixing advantages over segregation. Received: Revised: Accepted: Published: 5022

October 25, 2011 February 19, 2012 March 17, 2012 March 17, 2012 dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Moreover, particle shape,10 pressure,11,12 bed height,8 and distributor13 were also reported to affect particle segregation. It was shown that a shallow bed, low pressure, and porous plate with even gas distribution are favorable to particle segregation. Baffle is an important issue in gas fluidization research. Different baffles are generally categorized into three types: horizontal baffle, vertical baffle, and fixed packing. Horizontal baffles include mesh grid, perforated plate, louver, and horizontal tube bundle. They are usually used in fluidized beds to break up the bubbles and stage the bed by suppressing axial particle mixing, which is especially favorable in fluidized chemical reactors. The effectiveness of suppressing particle mixing by horizontal baffles is mainly dependent on the baffles’ open area ratio. However, effectiveness of breaking bubbles is affected by both open area ratio and the geometry of their flow passages.14,15 Owing to the inclined passages, louver baffle and its other modified forms were proven to be more effective in breaking bubbles.15 Horizontal baffles are usually used in low-velocity fluidized beds. At high gas velocities, gas cushions may appear below the baffle layers, resulting in overexpansion and low volume utilization efficiency of the bed. Vertical baffles refer to heat exchange tubes, semicircular tubes, planar plated baffles, etc., mounted vertically in fluidized beds. They are mostly used to increase the surface effect by decreasing bed hydraulic diameter, thus making scale-up easier. Fixed packing in fluidized beds is similar to those used in a packed tower, but its cell dimension is much larger due to bad particle fundability. By appropriate design, fixed packing can have the advantages of both horizontal and vertical baffles. For related studies on fluidized bed baffles, the reviews by Harrison and Grace16 and Jin et al.17 can be referred. Recently, several studies have found that horizontal baffles can greatly enhance particle segregation.18−24 Gelperin et al.18 investigated the classification of coal particles with binary size distributions in a conical fluidized bed with perforated plates. At higher gas velocities, they could still get high-content fines in the bed top. Kawabata et al.19 employed a single horizontal perforated plate to promote segregation between sand and coal char in a fluidized gasifier. They found that the promotion of particle segregation was sensitive to decreasing open area, but less sensitive to decreasing hole diameter. Hartholt et al.20 employed Group B glass beads of different sizes as binary components in a fluidized bed to study the effect of perforated plates on particle segregation. Their study showed that increasing plate layer number, decreasing hole diameter, and decreasing distance between adjacent plates promoted particle segregation. Further, in Hoffmann’s analysis,21 he attributed the baffle’s promotion on particle segregation to the baffle’s solid backmixing suppression by eliminating the wake solids from bubbles crossing the baffle layer. In a later study in fluidized beds with horizontal sieve-like baffles, Bosma and Hoffmann22 further proved baffles’ promotion effect on particle segregation in a binary fluidized bed with two equal-density particle components. Otherwise, they proposed a mechanistic model to predict the component fractions and the segregation rate, with qualitative agreement between the modeled and experimental results achieved. Chyang et al.23 employed glass beads of different sizes in a continuous binary fluidized bed with several mesh grids to investigate the effects of gas velocity, baffle spacing, screen size, feeding location, feeding rate, and bed height on segregation. They found that segregation efficiency increased as gas velocity, baffle spacing, and screen hole diameter decreased. Because of the strong promotion of particle segregation by horizontal baffles, Bosma and Hoffmann24 proposed to employ a baffled fluidized

bed as a continuous high-efficiency particle classifier and proved the feasibility. Applications of these classifiers can be found in ore beneficiation (e.g., Sahan & Kozanoglu studied separation of coal and magnetite in a fluidized bed25), removal of ash from a gasifier,2 etc. Several years ago, Gao et al.26 proposed a new fluid catalytic cracking (FCC) process to adapt the increasing demand for updating FCC gasoline due to the more and more stringent emission codes in China. A prominent duty to updating FCC gasoline was to reduce its olefin content without octane number reduction. The new FCC process comprised two riser reactors, one for cracking heavy feedstock and the other for gasoline updating, and a coupled regenerator. Because of the different reaction mechanisms, two catalysts of different sizes were employed. The coupled regenerator was used to mix the two catalysts after reactions, burn off the coke in them, separate them, and then send them to the corresponding reactors. The key advantage of this new process is the capability of making the heat-residual FCC process and the heat-deficient gasoline updating process well coupled, thus making better product yields and heat utilization more efficient in a compact unit. A detailed description of this new FCC process can be referred in Lan et al.27 A key challenge of this process was the efficient continuous separation of the two mixed catalysts after their regenerations. To make this bicatalyst process economically advantageous, by a preliminary estimate, the purities of the two separated catalysts were required to exceed 90−95%. To solve the particle separation problem, we thought of baffled fluidized beds. This is the application background of this study. On the other hand, systematic studies on particle segregation in baffled fluidized beds are still few to date, with only relatively large Group B particles and few baffle types, for example, screens and perforated plates, involved. Fine Group A particles and other types of baffles (e.g., louver baffle) are still not evaluated on particle segregation. Otherwise, former studies usually focused on the effects on particle segregation efficiency and the axial distribution of component fraction. Few studies,28,29 especially in baffled fluidized beds, involved investigation of the radial distribution of the component fraction. Therefore, more systematic studies on the segregation process are still lacking. One aim of the study is to test the feasibility of separating FCC catalyst (Group A) and a larger-size gasoline-updating catalyst (Group B) in a baffled fluidized bed. The baffle employed was newly developed and proved to be very effective in enhancing gas−solids contact and suppressing particle backmixing. The other aim is to obtain more information on particle segregation in both baffle-free and baffled binary fluidized beds, including the segregation efficiencies of the jetsam and the flotsam as well as the axial and radial fraction profiles, by more systematic experimental investigations.

2. EXPERIMENTAL UNIT The schematic of the experiment unit in this study is as shown in Figure 1. The main body was a cylindrical fluidized bed of inner diameter 286 mm and height 4 m. For ease of observing the behavior of gas and solid flows in the bed, the bed walls consisted of four transparent plexiglas columns of height 1 m, connected by flanges. Compressed air supplied by a positivepressure Roots blower (rated flow rate and pressure: 8.33 m3/min and 49 kPa) and regulated by a rotameter, flowed into the bed through a sparger distributor. Two surge tanks were designed in the air transportation line to maintain steady gas flow. A cyclone of body diameter 134 mm was placed on the top of the bed column 5023

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

orientation are installed in every flow region, but their orientations in adjacent pair regions are opposite. With this configuration, flows of gas and solid through adjacent flow regions can form interlaced contact both above and below a baffle layer, which is used to generate stronger gas−solid flow turbulence, promote bubble splitting, and enhance gas−solids contact. Otherwise, multiple layers of the new baffles are arranged as shown in Figure 3b. The spacing between adjacent pairs of baffle layers is relatively small, usually several times of the baffle layer height. Axially, corresponding flow regions between each pair baffle layers are of similar geometry, but with opposite vane orientations, forming zigzag routes when gas and solid flow through them and further enhancing gas−solids contact. This new baffle has already proved its excellent performance in enhancing gas−solid contact and suppression of particle backmixing through extensive experiments in a large cylindrical cold column.31 Figure 4a shows the comparisons of the differential pressure fluctuations characterized by the average standard deviation of several differential pressure signals in a large fluidized bed of 0.8 m i.d. with and without three layers of the new baffles. FCC particles of particle density 1500 kg/m3 and mean diameter 81 μm were employed. The static bed height was 1.9 m. These differential pressure transducers were installed with small pitches between their two pressure tabs. Previous studies32,33 have demonstrated this configuration can lead to more compression waves that are irrelevant to bubble behavior being filtered out and more bubble-related pressure waves kept. Furthermore, the measured differential pressure signals were processed with a band-pass filter to keep the signals in the frequency range of 0.2−40 Hz to filter out signal components due to gas flow-rate fluctuations and noises. These pressure fluctuations after procession are more suitable to characterize the average bubble diameter or the gas−solid contact quality in the bed. As seen in Figure 4a, the processed pressure fluctuations decreased significantly after three layers of the new baffles were inserted in the bed, indicating a great enhancement in the gas−solid contact quality. Figure 4b shows the comparisons of axial gas dispersion coefficients measured in the same experimental unit. The axial gas dispersion coefficients were obtained by fitting the gas tracing data to a steady one-dimensional dispersion model. As seen in Figure 4b, the axial gas dispersion coefficient also decreased significantly in the baffled bed, indicating great suppression of axial gas backmixing. For fluidized beds of fine Group A particles, many previous studies have demonstrated that gas mixing is predominantly determined by the carrying of solids under internal circulation.34−36 Owing to the close relationship between gas backmixing and solid mixing, a strong suppression of solid backmixing can also be inferred in the baffled bed. On the basis of the understanding on particle segregation established in previous studies, we conjectured that this new baffle may promote particle segregation in binary fluidized beds and can be a better solution to the catalyst separation problem in the new FCC process proposed by Gao et al.26 It is also one of the aims in this study to verify this conjecture. Based on the above-mentioned design concept and experiences obtained in previous studies, the new baffle used in this study was designed as shown in Figure 5. The baffle is 30 mm in height. In Figure 5, each hatched rectangle represents an inclined vane, where the bold line represents its top edge and the opposite fine line represents its bottom edge. Vanes were installed in several auxiliary support sheets (also acting as partition plates) linked to two main support sheets. The main

Figure 1. Schematic of the experimental unit: (1) particle withdrawal tube; (2) sparger gas distributor; (3) horizontal baffles; (4) fastening shaft; (5) fluidized bed; (6) cyclone; (7) dipleg; (8) rotameter; (9) valve; (10) secondary surge tank; (11) primary surge tank; (12) Roots blower.

to recover the entrained particles in the outflow air and return them into the bed through a dipleg. A V-valve was installed at the end of the dipleg to seal the air flow upward from the dipleg and keep high separation efficiency and stable particle downward flow. The sparger gas distributor had six branch tubes with 36 holes of 6 mm i.d. evenly distributed along these tubes, which corresponded to an opening area ratio (the ratio between the sum of all holes and the inner column cross-sectional area) of 1.62%. A typical distributor pressure drop curve with increasing superficial gas velocity is plotted in Figure 2. A constant static

Figure 2. Typical pressure drop curve of the gas distributor.

bed height of 0.95 m was maintained throughout all the experiments. In the lowest plexiglas column, a square steel shaft with a 30 mm × 30 mm cross-section was installed along the column axis for fastening the baffles to avoid them from inclining, dropping down or shaking in operation. The design concept of the new horizontal baffle30 used in this study is shown in Figure 3. As seen in Figure 3a, there are multiple parallel partition plates in each baffle layer, separating the bed cross-section into multiple flow regions. Vanes of same 5024

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 3. Schematic structure of the new baffle.

hose outlet several seconds after taking away the clincher. This was to keep the residual particles in the sampling tube and silica gel hose out of the sample. Sampled particles were collected into a jar, with a weight of ca. 100 g and sieved by a sieve of 100 mesh into jetsam and flotsam. Finally, an electronic balance was used to measure their mass fractions. The sampling time was 5−10 s, short enough to avoid disturbance on gas−solids flow and local fraction change in the bed. In each test, discharged particles due to sampling were negligible in weight to the solid inventory in the bed, so the effect of the decreased solids inventory during sampling could be reasonably neglected. The above sampling method was in nature operated in a similar way as the “thief” probe tested by Wu and Baeyens.8 In studying particle segregation in fluidized beds, a more prevalent sampling method is the freezing-bed method, where a bed is fluidized at a given gas velocity for several minutes before shutting off the gas supply abruptly. The bed now at rest is then divided into different axial or lateral sections and analyzed (e.g., by suction, sieving and weighing) separately. In their comparison between the results of the two sampling methods, Wu and Baeyens found that the maximum relative error could be as large as 123%. They attributed the large errors to segregation during particle flow into the probe, the flow disturbance caused by the inserted probe, and the particle component change during sampling. The two methods were also compared in our unit, as seen in Figure 6 where comparison results at two typical gas velocities in the baffle-free bed are plotted. In general, the trends by two methods are agreeable. The relative errors at low gas velocities are larger than at high gas velocities. Most measured relative errors are within 15% and acceptable. The relative small errors in our study may due to the following reasons: (a) the probe is smaller causing weaker disturbance, (b) sampled

and auxiliary support sheets were steel sheets of 4 mm and 3 mm in thickness, respectively. Each vane was a 36.5 mm × 31 mm rectangular plate of thickness of 2 mm, which was welded to the support sheets with an inclination angle of 55°. At the center of the baffle, there was a 30 mm (width) × 30 mm (length) × 2 mm (thickness) square tube with screw holes used to fasten the baffle layer to the central shaft. In the experiment, the fluid medium was compressed air. The employed flotsam was FCC equilibrium catalyst with a particle density of 1500 kg/m3, a mean diameter of 70 μm, and a minimum fluidization velocity of 0.003 m/s. The employed jetsam was millet particles with a density of 1402 kg/m3, a mean diameter of 930 μm, and a minimum fluidization velocity of 0.491 m/s. The millet particles here were used as model particles for the gasoline-updating catalyst in the new FCC process. The similar density of the FCC particles was based on the assumption that zeolite and the matrix of amorphous alumina will still be the dominant components. A relatively larger size was to facilitate particle separation. According to the well-known particle classification method of Geldart,37 the binary mixture of particles in this study contained a Group A flotsam and a Group B jetsam. These two types of particles were mixed with a mass ratio of 1:1 and then fed into the bed. The superficial gas velocity range in this study was from 0.03 to 0.5 m/s. To obtain the fraction profiles of jetsam and flotsam, a quantity of particles was sampled from each sampling point. The sampling tube was made of a 250 mm long steel tube of 12 mm (o.d.) × 1 mm (thickness) connected by a section of silica gel hose. Usually, the silica gel hose was clamped by a clincher. When sampling, the clincher was taken away from the hose and particles flowed out due to the positive pressure in the bed. The sampled particles were the particles ejected from the 5025

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 4. Comparison of fluidization quality and gas mixing in a large fluidized bed with and without the new horizontal baffle (FCC particle: ρp = 1500 kg/m3, dp = 81 mm, dt = 0.8 m, H0 = 1.9 m.)31

Figure 5. The new baffles employed in this study.

and xj,0 are the initial flotsam and jetsam fractions, which are also equal to the average flotsam and jetsam fractions throughout the bed. Mf and Mj vary from 0 to 1. When they equal 1, it means that particles are mixed completely. When they equal 0, it means that flotsam (jetsam) is completely separated in the bed top (bottom). Accordingly, the smaller the Mf (Mj) is, the higher the segregation efficiency of flotsam (jetsam) can be obtained. The mixing number Mf is actually the same as the mixing number in eq 1. However, in many binary fluidized beds, the flotsam segregation efficiency in the bed top is different from the jetsam segregation efficiency in the bed bottom, which is the reason to introduce the other mixing number, Mj, in this study. Figure 7 shows the schematic of the positions of the sample tubes and baffles installed in the experimental unit. There are five axial sampling levels set in studying both the baffle-free and baffled fluidized beds. They were 100, 300, 500, 700, and 900 mm above the gas distributor, respectively. Additionally, at every axial level, there are five radial positions that are 0, 30, 60, 90, and 120 mm from the inner column wall. Due to the existence of the fastening shaft, there is no radial sampling position set in the column centerline. The average flotsam (jetsam) fraction at every axial level is calculated by area-weighted averaging; that is,

particles flow horizontally into the probe without turning, resulting in weaker segregation in sampling, (c) shorter sampling time. As seen in Figure 6, the flotsam fractions measured by the “thief” probe are mostly lower than those by the freezing bed method. This can be explained that the high-flotsam particles descend when the bed is freezing as bubbles leave the bed, causing a slight increase of the flotsam fraction. On the other hand, in our baffled fluidized bed, the bed is more expanded than in a baffle-free bed. When the bed is frozen by shutting the gas supply, the bed deaeration and the particle mixing induced by the crossing baffle vanes can result in considerable error to the sampling results. Therefore, the freezing-bed sampling is more unsuitable in baffled beds than in baffle-free beds. In this study, the degree of particle mixing was characterized by two mixing numbers, Mf and Mj, which were represented, respectively, by 1 − x f,t 1 − x f,0

(2)

1 − x j,b 1 − x j,0

(3)

∑

Herein, xf,t and xj,b are the flotsam and jetsam fractions measured at 900 mm and 100 mm above the distributor, respectively; xf,0

i=1

Mf =

and

Mj =

5

x f,ave = 5026

x f, iA i A

(4)

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 7. Schematic of the positions of baffles and sample tubes.

Figure 6. Comparison between the two sampling methods by “thief” probe and freezing bed.

Here, ⎡⎛ r + ri ⎞2 ⎛ ri − 1 + ri ⎞2 ⎤ ⎟ − ⎜ ⎟ ⎥ A i = π⎢⎜ i + 1 ⎠ ⎝ ⎠ ⎥⎦ ⎢⎣⎝ 2 2

(5)

i represents the number of the radial positions, ri is the distance between the centerline of the column section and the no. i radial position. Here, r6 = r5 = R and r0 is set to be 19 mm to exclude the influence of the fastening shaft. R is the inner diameter of the column; xi is the flotsam (jetsam) fraction at the no. i radial position. In the baffled fluidized bed, four layers of baffle were placed in the dense bed, which were 200, 400, 600, and 800 mm above the gas distributor (from the middle of the baffle layer), respectively. When starting up the experimental unit or shifting from one operating condition to another, the binary fluidized bed will undergo a transitional period before getting into an equilibrium state. To make sure that sampling was conducted in the equilibrium state, the dynamic sampling experiment was conducted to determine the transitional time during different conditions. Figure 8 shows two typical transitional processes, which indicate the transitional time is usually less than 10 min. Accordingly, all the sampling started 20 min after starting up the unit or condition changes in this experiment. Wu and Baeyens.8 also measured this transitional time in their binary fluidized beds with one Group B component and one Group D component. They found that the transitional time decreased with increasing gas velocity, from ∼30 min to less than 10 min. It seems that their transitional times are longer than those in our study. This may be attributed to the better fluidity of FCC particles. In most experiments, they sampled after 20 min to allow the bed to equilibrate, which is the same as in our study.

Figure 8. The change of flotsam fraction with time.

3. EXPERIMENTAL RESULTS 3.1. Effect on particle segregation efficiency. Figure 9 shows the change of the two mixing numbers, Mf and Mj, with increasing superficial gas velocity in the baffle-free bed. As seen in Figure 9, Mf reaches a value near zero when gas velocity is lower than 0.03 m/s, corresponding to a flotsam 5027

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

from the previous findings. To a large extent, this is due to the fine FCC particles used as jetsam in this binary fluidized bed, which have better fluidity and stronger mixing than the usual Group B particles in other studies. Except in this study, Upadhyay and Roy39 also observed similar phenomenon in binary fluidized beds with two particle components with different densities. When they used a radiation-based noninvasive technique, namely, radioactive particle tracking, to observe the particle movements, they also saw that heavier (jetsam) particles were fluidized by the action of the lighter (flotsam) particles even below their minimum fluidization velocity. The interaction between fine and coarse particle components can also be seen in other fluidization phenomena. For example, Geldart and Pope40 found in their particle entrainment experiments that, when mixed with fine particles, coarse particles of larger diameters could appear in the freeboard and the carryover flux of the coarse particles increased considerably. They attributed it to the collision and momentum exchange by fine particles. Different from Mf in Figure 9, Mj is always greater than 0.6 in the whole gas velocity range of this study. It is easily understandable that the violent bubble-induced particle mixing under high gas velocities, which entrains more flotsam particles into the bed bottom with well mixing with the jetsam particles, is the reason for the high Mj values at high gas velocities. At low gas velocities, the large jetsam particles are only partially fluidized or even defluidized, gas bypasses through the bed bottom, leaving some flotsam particles incapable of being entrained by gas flow. Thus, although the flotsam in the bed top is nearly pure, there are still some flotsam particles locked in the bed bottom, leading to a low jetsam fraction. Therefore, for such a baffle-free binary fluidized bed with Group A and B particles, the segregation efficiency of the jetsam is impossible to be high at any operating conditions. It is more suitable to acquire purified flotsam particles than to acquire purified jetsam particles in such a bed. When four layers of the new baffles were installed in the bed, different trends of Mf and Mj happened. Figure 11 compares

Figure 9. The mixing numbers under different gas velocities in the baffle-free bed.

fraction of 95% in the bed top. Mf increases from zero to nearly unity with increasing gas velocity, which means that particles at the top of the dense bed change from nearly pure fine FCC particles to binary mixtures with the same flotsam fraction as its initial fraction, xf,0. When gas velocity is higher than 0.08 m/s, Mf approaches unity and keeps constant with further increasing gas velocity, which corresponds to a complete mixing state for the binary particles. Mj follows a similar trend as Mf as gas velocity increases. When gas velocity is higher than 0.08 m/s, Mj also approaches unity and keeps constant with further increasing gas velocity, demonstrating that the particles in the bed bottom also reach a state of complete mixing. This critical gas velocity is called the “complete mixing velocity” and denoted as ucm in this paper. As all these data measured in dynamic equilibrium states, the state of complete mixing at u0 > ucm implies a fluidization state for both the flotsam and jetsam, inferring a completely fluidized bed. This fully fluidization state can also be proven by the change of the total bed pressure drop with increasing gas velocity, as seen in Figure 10. The umf

Figure 10. The determination of umf in the baffle-free fluidized bed.

Figure 11. Comparison of Mf in the baffle-free and baffled beds.

determined from Figure 10 is 0.075 m/s, almost equal to ucm in Figure 9. In previously published studies,38 a completely fluidized binary bed was often thought to have a higher superficial gas velocity larger than the umf of the jetsam particles. However, ucm in this study is far less than the minimum fluidization velocity of the jetsam particles (0.491 m/s). This means that the strong mixing of flotsam particles in the bed actually acts as a driving force to make the jetsam fluidized. In other words, both the gas and the fine flotsam can be treated as a fluidizing media. Owing to its larger apparent density or viscosity, it can make the jetsam particles fluidized earlier. This is clearly a different phenomenon

the changes of Mf with increasing gas velocity in the baffle-free and baffled beds. As seen in Figure 11, only when the gas velocity exceeds 0.35 m/s, does Mf reach a constant high value near unity, which means a postponed state of complete mixing in the baffled bed. Otherwise, Mf in the baffled bed is always less than that in the baffle-free bed, indicating that the baffles can hinder axial particle mixing and enhance segregation. Moreover, the lowest Mf (xf = 97%) in the baffled bed is also less than that (xf = 95%) in the baffle-free bed in this study, indicating purer flotsam particles obtainable in the baffled bed. 5028

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

The change of Mj in the baffled bed is also different from in the baffle-free bed as shown in Figure 12. With increasing gas

Figure 13. The axial profile of flotsam fraction in the baffle-free bed.

appearing near the middle of the bed, whose gradient is far greater than in other sections. In the study of Hartholt et al.,14 they also found in their binary fluidized beds with multiple layers of perforated plates that a bed section with distinct large fraction gradient appeared in the middle of the beds, agreeing with our results. Moreover, the level of the section with the largest xf,ave gradient descends with increasing gas velocity, as shown in Figure 14. When gas velocity reaches 0.3 m/s, the

Figure 12. Comparison of Mj in the baffle-free and baffled beds.

velocity, Mj first decreases gradually, then increases sharply after u0 > 0.3 m/s and keeps constant after u0 > 0.35 m/s. This is a different trend from Mf shown in Figure 11. We conjecture that this phenomenon is due to the promoted segregation effect in baffled beds, which results in more jetsam particles deposited in the bed bottom, leading to some defluidized zones in the bed bottom, resulting in gas bypassing and preventing some fine flotsam particles from being carried upward to the bed surface. Maybe, a better-designed distributor will help due to its better gas−solid contact quality and Mj at u0 < 0.3 m/s can be increased further. The lowest Mj appears when gas velocity is between 0.2 and 0.3 m/s, corresponding to a jetsam fraction higher than 99%. The gas velocity with the highest jetsam segregation efficiency in the baffle-free bed (u0 < 0.08 m/s) is different from that in the baffled bed (u0 = 0.2−0.3 m/s). Similarly, Mj in the baffled bed is also always less than in the baffle-free bed. This evidence further proves the new baffle’s excellent promotion effect on particle segregation in binary fluidized beds, which can help obtain purer flotsam in the bed top and purer jetsam in the bed bottom. This is particularly useful for binary fluidized beds that are utilized to classify particles. On the other hand, the new baffles also extend the gas velocity range suitable for particle classification, which, due to the higher operating gas velocity, is favorable to keeping good fluidity of large jetsam particles especially in continuously operated units. 3.2. Effect on Axial Fraction Profiles. Particle segregation efficiency only involves the fraction of jetsam or flotsam in the bed top or bottom, but they can not reflect the panoramic distributions of the binary particle components. Figures 13 and 14 show the axial profiles of the averaged flotsam fraction in the baffle-free and baffled beds at different gas velocities. In the baffle-free bed, the axial flotsam fraction usually increases with increasing bed height. However, it rises sharply in higher bed levels but much more gently in lower bed levels, as shown in Figure 13. This average gradient of the xf,ave curve diminishes with increasing gas velocity. The xf,ave curve becomes a flat line when the gas velocity exceeds 0.08 m/s, which corresponds to a state of complete mixing throughout the bed. However, the xf,ave curves show different trends after four layers of the new baffles were installed, as shown in Figure 14. The average gradients of all xf,ave curves at u0 < 0.3 m/s increase considerably compared to their counterparts in the bafflefree bed. For these curves, the gradients at different levels are not even. There is usually a section with a very large gradient

Figure 14. Axial profile of flotsam fraction in the baffled bed.

level of the section with the largest xf,ave gradient approaches the bed bottom. When gas velocity exceeds 0.35 m/s, the xf,ave curve becomes a nearly flat curve as happened in the baffle-free bed at u > 0.08 m/s. 3.3. Effect on Radial Fraction Profiles. Radial distributions of particle components, which were rarely studied previously,28,29 can further promote the understanding of the segregation process in binary fluidized beds. In this study, when gas velocity was lower than 0.08 m/s in the baffle-free bed, two different particle layers could be clearly observed in the dense bed from the transparent column walls. The top layer had more flotsam particles, while the bottom layer contained more jetsam particles. Figure 15 shows the radial flotsam profiles at different axial heights under three typical operating conditions. When u0 is less than 0.08 m/s, most of the radial flotsam profiles are flat except for the one near z = 700 mm, whose flotsam fraction varies considerably from the center to near the wall. There are more flotsam particles in the central zone and more jetsam particles near the wall around z = 700 mm. This location agrees with the height of the interface of the two particle layers observed in experiment. Furthermore, this special radial flotsam profile also infers a concavity interface surface (looking upward) around z = 700 mm, if an axially symmetrical distribution is presumed. The concave interface is believed to be related with the lateral uneven gas flow distribution in the bed. In such a 5029

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 16. Area-weighted deviation of the radial flotsam fraction in the baffle-free bed.

After installing the four pieces of the new baffle, the radial flotsam profiles become slightly different from those in the baffle-free bed, as shown in Figure 17. First, although the above-mentioned interface could also be observed in the baffled bed at u0 < 0.35 m/s, their levels are lower than in the bafflefree bed and decrease with increasing gas velocity. As seen in Figure 17 panels a−c, the interface level decreases from z = 500 mm at u0 = 0.06 m/s to z = 300 mm at u0 = 0.1 m/s. When gas velocity reaches 0.2 or 0.3 m/s, the interface descends to the bed bottom, as seen in Figures 17e,f. When u0 is between 0.1 and 0.3 m/s, the interface surface becomes a plate interface as indicated by the radial flotsam profiles shown in Figure 17d−f, although this interface could still be observed in the experiment. The interface level can also be observed in the σxf vs z curves of the baffled bed. As shown in Figure 18, the peak of σxf vs z curve can be clearly observed until u0 reaches 0.35 m/s. The height corresponding to this peak decreases with increasing gas velocity, from z = 500 mm at u0 = 0.06 m/s to z = 300 mm at u0 = 0.1 m/s, in agreement with the radial profiles of flotsam fraction shown in Figure 17 and the experimental observations. 3.4. Contours of Component Fraction in the Bafflefree and Baffled Beds. The above measured profiles of axial and radial flotsam fractions in the baffle-free and baffled beds enable, with the axial symmetry hypothesis of the component distribution, the panoramic distributions of the binary components shown more clearly in two-dimensional contours. Figures 19 and 20 show the flotsam contours of the baffle-free and baffled beds under different gas velocities, respectively. In these contours, different blocks with different packing shapes represent different values of flotsam fraction. As shown in Figure 19, when the gas velocity does not exceed 0.08 m/s, there are several regions with different flotsam fractions in the baffle-free bed. Although flotsam fraction increases with increasing bed height, the boundaries of these regions are not horizontal. In view of the possible error in experiment data, it can be seen that most interfaces are concave (looking upward), with more flotsam particles in the central zone than in the near-wall zone. The axial gradient of the flotsam fraction in the baffle-free bed decreases with increasing gas velocity, to nearly zero when u0 > 0.08 m/s. The largest fraction gradient appears above the middle of the dense bed, agreeing with the interface of the two particle layers observed in the experiment. The baffled bed has a larger flotsam fraction gradient than the baffle-free bed as shown in Figure 20, which enables it to produce both purer fine particles in the bed top and purer coarse particles in the bed bottom. As gas velocity increases, the

Figure 15. Radial profiles of flotsam fraction in the baffle-free bed.

small-scale fluidized bed, bubbles prefer flowing near the center of the bed, entraining more fine particles into the central zone of the bed due to their small drag/weight ratios. By defining an area-weighted deviation of the radial flotsam fraction σxf 5

∑ σxf =

A i (x f, i − x f,ave)2

1

A

(6)

the interface location can be more clearly observed. As seen in Figure 16, there is a distinct peak existing in the σxf vs z curve at u0 = 0.03 m/s or at u0= 0.08 m/s. When u0 exceeds 0.08 m/s, this special radial flotsam profile disappears and all the radial flotsam profiles become similar to Figure 15c, which corresponds to the disappearance of the two particle layers and a state of complete mixing in the bed. Otherwise, it can be seen that the σxf vs z curve becomes flatter and flatter with increasing gas velocity. 5030

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 17. Radial profiles of flotsam fraction in the baffled bed.

patterns of the flotsam particles under u0 < 0.35 m/s are similar to those in the baffle-free bed under u0 < 0.08 m/s. These boundaries of regions with different flotsam fractions are mostly concave surfaces (looking upward). 3.5. Discussion of the Experimental Results. The enhanced particle segregation in the baffled fluidized bed can be correlated with its effects on bed hydrodynamics and particle mixing. On the basis of the existing studies on particle segregation mechanism in fluidized beds,4,41,42 it can be concluded that, for a given binary fluidized bed (i.e., properties of particle and gas are given), the degree of particle segregation is to a large extent dependent on the properties of bubbles and bubble flow. On the one hand, smaller average bubble diameter and even lateral bubble flow distribution in a bed are favorable to particle segregation at a same gas velocity. This can be partially proved by the study of Nienow et al.13 who found significant improvement of mixing was obtained with perforated plate and standpipe distributors, compared to a porous distributor plate at a similar operating gas velocity. It is well-known that, compared to other distributors, a porous distributor has smaller initial bubbles and more even

Figure 18. Area-weighted deviation of the radial flotsam fraction in the baffled bed.

axial flotsam fraction gradient first increases and then decreases to near zero at u0 < 0.35−0.4 m/s. The location with the maximum axial gradient appears at the middle-top section of the dense bed under low gas velocities, descending to lower levels with increasing gas velocity. The radial distribution 5031

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 19. Contours of flotsam fraction in the baffle-free bed.

the comparison of pressure fluctuations in the baffle-free and baffled beds. It can be seen that the pressure fluctuation is significantly reduced in the baffled bed. Binary particles notwithstanding in this study, the baffle’s effect on splitting bubbles is similar to the finding in the single-component fluidized bed (see Figure 2a). If all bubble-irrelevant signals are assumed to be filtered out, the average bubble diameter ratio in the baffle-free and baffled beds should follow the following correlation,33

bubble flow distribution due to its high pressure drop and numerous small orifices. The study of Wu and Baeyens.8 provides another proof for this opinion. In their study, they found that increasing bed height favored particle mixing rather than segregation. It is well accepted that bubble diameter, especially in fluidized beds of Group B particles, increases with increasing height due to coalescence, resulting in an increased average bubble diameter and more uneven bubble flow distribution. On the other hand, particle mixing induced by bubbles counteract segregation in fluidized beds. Suppressing particle mixing will in turn promote particle segregation. On the contrary, to avoid particle segregation promoting mixing is a good choice. In the following text, we will show how bubble size, bubble flow, and particle mixing are affected by the inserted baffles of this study and in turn influence particle segregation. (a). Bubble Size. Measuring bubble size in gas fluidized beds is difficult, especially in beds of Group A particles where bubbles frequently break off and coalesce. Despite that there are many reported methods, no one is accepted widely by the fluidization community and has adequate accuracy. In this study, we resorted to pressure fluctuation measurement. Similar tap setting and data postprocession were adopted as in our previous study31 to keep more bubble-related signals and filter out more signals irrelevant to bubble behavior. Figure 21 shows

n d b1 ⎛ σdp1 ⎞ ⎟ = ⎜⎜ ⎟ d b2 ⎝ σdp2 ⎠

(7)

Here, σdp is the standard deviation of the processed pressure fluctuations. If σdp is proportional to bubble volume, n = 1/3; if proportional to bubble diameter, then n = 1. Due to the complex nature of pressure signal in fluidized beds, the actual value of n should be between 1/3 and 1. It can thus be estimated that the inserted baffles in this study can reduce the average bubble size by 13%∼72%, depending on different operating conditions and estimations of n. (b). Lateral Distribution of Bubble Flow. There is no direct proof for the baffle’s improvement on lateral distribution of bubble flow. However, it can be inferred from the steady gas 5032

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 20. Contours of flotsam fraction in the baffled bed.

baffles and gas samplers. The setting of the gas sampler was the same in the baffle-free bed to facilitate comparison. Figure 23 show the comparison of the radial distribution of tracer gas concentration in the baffle-free and baffled bed at a similar gas velocity to this study. Clearly, the radial distribution of tracer gas concentration in the baffled bed is much more even, indicating the improvement of lateral bubble flow distribution

tracing experiment in a large cylindrical cold model of diameter 0.8 m31 due to the close relationship between gas and solid mixing in fluidized beds of fine FCC particles. In this experiment, a tracing hydrogen flow was inserted downward into the bed center at the upper zone of the dense bed. Seven gas samplers were set radially for four different heights beneath the hydrogen injector. Figure 22 show the configuration of the new 5033

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Figure 21. Comparison of the pressure fluctuations in baffle-free and baffled beds.

Figure 23. Comparison of the radial distributions of tracer gas concentration in the baffle-free and baffled beds.

above the baffles and more jetsam particles to stay beneath the baffle layer due to their different drag/weight ratios. In fact, the baffle’s effect on bubble size and bubble flow distribution is closely correlated with its effect on solid mixing. According to the estimation of Geldart,44 the internal solid circulation flux in a fluidized bed can be estimated by

Figure 22. Configuration of the new baffles and tracer sampling tubes (length unit in mm).

by the baffles. Combined with the results obtained in other operating conditions, it was found that the baffles’ effect on improving lateral bubble flow distribution enhances with increasing gas velocity. (c). Solid Mixing. The suppression of solid mixing by this baffle can also be inferred by the results of gas tracing experiments. The significantly reduced axial gas dispersion coefficient in the baffled bed (see Figure 4b) can partially reflect the strong suppression of solid mixing by the baffles. The significantly reduced tracer gas concentration in the baffled bed (see Figure 22) is another proof of the baffle’s strong suppression on solid mixing. In studying solid movement in a baffled fluidized bed, van Dijk et al.43 found by an X-ray technique that the wake solids were detached when a bubble passed a layer of screens in a fluidized bed. This is also agreeable with our findings. Except for detaching wake solids in passing bubbles, the reduced flow area at the baffle layer is another cause of the suppressed solid mixing. As indicated in our previous study in a two-dimensional baffled bed with louver baffles,15 the reduced flow area across the baffle layer results in higher local gas velocity as well as reduced downward solid flux. To maintain a steady operation, a “gas cushion” appears beneath the baffle layer resulting in lower particle concentration and upward solid flux to accommodate the reduced downward solid flux. With retarded internal solid circulation, particle mixing is thereby suppressed. Otherwise, the local high gas across the baffle layer also helps particles segregate by enabling more flotsam particles to be entrained

J = ρp(1 − εmf )(u0 − u mf )Y (βw + 0.38βd)

(8)

Here, Y is a modified factor for the two-phase model, and βw and βd are the wake and drift fractions, which are closely related to bubble size and bubble flow distribution. In their X-ray imaging experiments, Hoffmann et al.45 measured wake angles and correlated them with respect to bubble diameter, θw = 160 − 160 exp( −60d b)

(9)

As θw monotonously increases with bubble diameter, βw also monotonously increases with bubble diameter. Therefore, as the average bubble diameter is reduced in the baffled bed, the internal solid circulation is also decreased. There is no direct study on the relationship between drift fraction and bubble flow distribution, but it can be inferred from other experimental facts that uneven bubble flow distribution causes larger drift fraction and stronger solid mixing. An example is in larger beds where May46 found that the axial solid diffusivity is an order of magnitude larger than that in a small laboratory unit. Similar results were also observed by other researchers.47−49 Clearly, this should not be the only contribution of enlarging bubble size. The increased drift fraction is another key contributor, which in nature originates from the uneven bubble flow distribution in larger beds. 5034

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

Mf = mixing number of flotsam Mj = mixing number of jetsam r = radial position (m) R = inner radii of the column (m) u0 = superficial gas velocity (m/s) ucm = critical gas velocity for the bed to reach a completely mixing state (m/s) umf = minimum fluidization velocity (m/s) umf,f = minimum fluidization velocity of the flotsam particles (m/s) uTO = critical gas velocity between the predominance of segregation and mixing (m/s) Vb = bubble volume (m3) Vd = drift volume (m3) Vb = wake volume (m3) xf,ave = averaged flotsam fraction (m%) xf,0 = initial flotsam fraction (m%) xf = flotsam fraction (m%) xf,t = flotsam fraction in the bed top (m%) xj,0 = initial jetsam fraction in the bed bottom (m%) xj,b = jetsam fraction in the bed bottom (m%) xj,t = jetsam fraction in the bed top (m%) Y = correction factor for the two-phase model z = height above distributor (m) zs = particle sampling height above distributor (m)

On the application side, the baffled bed of this study is superior to a baffle-free bed as a continuous catalyst classifier such as in the new FCC process proposed by Gao et al.26 However, there is still unfinished work before industrial applications, such as geometry optimization, segregation kinetics, and equipment reliability evaluation, etc. On the other hand, a deeper understanding of the effect of horizontal baffles with inclined surfaces on particle segregation is another contribution of this study. However, there is still many fundamental phenomena remaining to be clarified and quantitatively determined in this area.

4. CONCLUSION In this study, particle segregation was systematically studied in a binary fluidized bed of inner diameter 0.286 m with and without horizontal baffles. Fine Group A FCC particles and coarse Group B millet particles were flotsam and jetsam components, respectively. Segregation efficiencies of both flotsam and jetsam as well as axial and radial fraction profiles were studied and analyzed with the following conclusions obtained: (1) In the baffle-free bed, binary particles approach a state of complete mixing at u0 = 0.08 m/s. The maximum segregation efficiencies of jetsam and flotsam are different, with only nearly pure flotsam obtainable in the bed top at very low gas velocities. The axial profile of the flotsam fraction is sharper in the upper section of the bed and much flatter in the lower section. With increasing gas velocity, the axial gradient of the flotsam fraction decreases and disappears at u0 > 0.08 m/s. (2) After four layers of the new baffles are installed, particle segregation is enhanced greatly. Higher maximum segregation efficiencies for both jetsam and flotsam particles were obtainable at much higher gas velocities. Both nearly pure jetsam and flotsam particles are obtainable at appropriate gas velocities. Binary particles reach a state of complete mixing at u0 = 0.35 m/s, much greater than in the baffled-free bed. The axial gradient of the flotsam fraction is much greater than in the baffled-free bed. (3) The promoted segregation in the baffled bed is related to the reduced bubble size, improved bubble flow distribution, and suppressed solid mixing by the new baffles, which makes it a promising continuous particle classifier within industry.

■

Greek Letters

■

AUTHOR INFORMATION

βd = drift fraction, Vd/Vb βw = wake fraction, Vw/Vb Δhj = distance from hydrogen injector (m) θw = wake angle (degree) ε = average bed voidage εmf = bed voidage under minimum fluidization ρf = particle density of the flotsam particles (kg/m3) ρj = particle density of the jetsam particles (kg/m3) ρp = particle density (kg/m3) σdp = standard deviation of the processed pressure fluctuation (kPa) σxf = area-weighted radial deviation of the flotsam fraction (m%)

REFERENCES

(1) Chen, J. L. P.; Keairns, D. L. Particle segregation in a fluidized bed. Can. J. Chem. Eng. 1975, 53, 395−402. (2) Shapiro, M.; Galperin, V. Air classification of solid particles: A review. Chem. Eng. Proc. 2005, 44, 279−285. (3) Rowe, P. N.; Nienow, A. W.; Agbim, A. J. A preliminary quantitative study of particle segregation in gas fluidized bedsbinary systems of near spherical particles. Trans. Inst. Chem. Eng. 1972, 50, 324−333. (4) Rowe, P. N.; Nienow, A. W.; Agbim, A. J. The mechanism by which particles segregate in gas fluidized beds: Binary system of nearspherical particles. Trans. Inst. Chem. Eng. 1972, 50, 310−323. (5) Thomson, F. M. Storage and flow of particulate solids. Handbook of Powder Science & Technology, 2nd ed.; Fayed, M. E., Otten, L, Eds.; Chapman & Hall: New York, 1997; Chapter 8, pp 446−451. (6) Baeyens, J. Geldart, D. Solids mixing. Gas Fluidization Technology; Geldart, D., Ed.; John Wiley & Sons Ltd.: New York, 1986; Chapter 5, pp 111. (7) Rowe, P. N.; Nienow, A. W. Particle mixing and segregation in gas fluidized beds. Powder Technol. 1976, 15, 141−147. (8) Wu, S. Y.; Baeyens, J. Segregation by size difference in gas fluidized beds. Powder Technol. 1998, 98, 139−150. (9) Olivieri, G.; Marzocchella, A; Salatino, P. Segregation of fluidized binary mixtures of granular solids. AIChE J. 2004, 50, 3095−3106.

Corresponding Author

*Tel.: +86-10-89731269. Fax: +86-10-89733803. E-mail: [email protected]; [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge the financial supports from the National Natural Science Foundation of China (Grant No. 20906101) and the Ministry of Science and Technology of China (Grant No. 2012CB215004).

■

NOMENCLATURE A = cross-sectional area the column (m2) Ai = sectioned area (m2) db = bubble diameter (m) df = average particle diameter of the flotsam particles (m) dj = average particle diameter of the jetsam particles (m) J = solids circulation flux in fluidized beds (kg/m2·s) M = mixing index 5035

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036

Industrial & Engineering Chemistry Research

Article

(10) Nienow, A. W.; Cheesman, D. J. The effect of shape on the mixing and segregation of large particles in gas-fluidized bed of small ones. In Fluidization; Grace, J. R., Matsen, J. M., Eds.; Plenum Press: New York, 1980; pp 373−380. (11) Chen, J. L. P.; Keairns, D. L. Particle segregation in a fluidized bed. Can. J. Chem. Eng. 1975, 53, 395−402. (12) Chiba, S.; Kawabata, J.; Yumiyama, M.; Tazaki, Y.; Honma, S.; Kitano, K.; Kobayashi, H.; Chiba, T. Pressure effects on solid mixing and segregation in gas-fluidized beds of binary solid mixtures. In FluidizationScience and Technology; Kwauk, M., Kunii, D., Eds.; Gordon and Breach: New York, 1982; pp 69−78. (13) Nienow, A. W.; Naimer, N. S.; Chiba, T. Study of segregation/ mixing in fluidized beds of different size particles. Chem. Eng. Commun. 1987, 62, 53−66. (14) Dutta, S.; Suciu, G. D. An experimental study of the effectiveness of baffles and internals in breaking bubbles in fluid beds. J. Chem. Eng. Jpn. 1992, 25, 345−348. (15) Zhang, Y.; Grace, J. R.; Bi, X.; Lu, C.; Shi, M. Effect of louver baffles on hydrodynamics and gas mixing in a fluidized bed of FCC particles. Chem. Eng. Sci. 2009, 64, 3270−3281. (16) Harrison, D.; Grace, J. R. Fluidized beds with internal baffles. In Fluidization; Davidson, J. F., Harrison, D., Eds.; Academic Press: London, 1971; pp 599−626. (17) Jin, Y.; Wei, F.; Wang, Y. Effects of internal tubes and baffles. Handbook of Fluidization and Fluid Particle Systems; Yang, W. C., Ed.; Marcel Dekker Inc.: New York, 2003; Chapter 7, pp 171−199. (18) Gelperin, N. I.; Einstein, V. G.; Kwasha, V. B.; Kogan, A. S.; Vilnits, S. A. Apparatus for classification of free-flowing materials in a fluidized bed. Int. Chem. Eng. 1964, 4, 198−203. (19) Kawabata, J.; Tazaki, Y.; Chiba, T.; Yoshida, K. Effect of perforated plate baffles on particle segregation in gas fluidized beds of solid mixtures. J. Chem. Eng. Jpn. 1981, 14, 246−249. (20) Hartholt, G. P.; la Rivière, R.; Hoffmann, A. C.; Janssen, L. P. B. M. The influence of perforated baffles on the mixing and segregation of a binary Group B mixture in a gas−solid fluidized bed. Powder Technol. 1997, 93, 185−188. (21) Hoffmann, A. C. Manipulating fluidized beds by using internals: Fluidization with baffles. NPT Procestechnol. 2000, 7 (2), 20−24. (22) Bosma, J. C.; Hoffmann, A. C. On the capacity of continuous powder classification in a gas-fluidized bed with horizontal sieve-like baffles. Powder Technol. 2003, 134, 1−15. (23) Chyang, C. S.; Wu, K. T.; Ma, T. T. Particle segregation in a screen baffle packed fluidized bed. Powder Technol. 2002, 126, 59−64. (24) Bosma, J. C.; Hoffmann, A. C. Feasibility study of particle classification in fluidized beds with internal baffles. In Proceedings WCPT4, the Fourth World Congress on Particle Technology, Sydney, 2002. (25) Sahan, R. A.; Kozanoglu, B. Use of an air fluidized bed separator in a dry coal cleaning process. Energy Convers. Manage. 1997, 38, 269−286. (26) Gao, J.; Xu, C.; Cao, B.; Bai, Y. A new fluidized catalytic cracking process with double reaction zones and a coupled regeneration zone. Chin. Patent, ZL 200310121341.6 (in Chinese). (27) Lan, X.; Xu, C.; Wang, G.; Chang, J.; Lu, C.; Gao, J. Heat coupling of gasoline upgrading and fluid catalytic cracking processes. Int. J. Chem. React. Eng. 2010, 8, A27−42. (28) Joseph, G. G.; Leboreiro, J.; Hrenya, C. M.; Stevens, A. R. Experimental segregation profiles in bubbling gas-fluidized beds. AIChE J. 2007, 53, 2804−2813. (29) Gilbertson, M. A.; Eames, I. Segregation patterns in gas-fluidized systems. J. Fluid Mech. 2001, 433, 347−356. (30) Lu, C.; Zhang, Y. A new internal for gas-solid fluidized beds. Chin. Patent, ZL 200610114153.4, 2009 (in Chinese). (31) Zhang, Y. Hydrodynamic and mixing properties of a novel baffled fluidized bed. Doctoral Dissertation, China University of Petroleum, Beijing, China, 2008. (32) Bi, H. T. Flow regime transitions in gas−solids fluidization and transport. Doctoral dissertation, University of British Columbia, Vancouver, Canada, 1994.

(33) Zhang, Y.; Bi, H. T.; Grace, J. R.; Lu, C. Comparison of decoupling methods for analyzing pressure fluctuations in gas fluidized beds. AIChE J. 2010, 56, 869−877. (34) Du, B.; Fan, L. S.; Wei, F.; Warsito, W. Gas and solids mixing in a turbulent fluidized bed. AIChE J. 2002, 48, 1896−1909. (35) Lee, G. S.; Kim, S. D. Axial mixing of solids in turbulent fluidized beds. Chem. Eng. J. 1990, 44, 1−9. (36) Deshmukh, S. A. R. K.; van Sint Annaland, M.; Kuipers, J. A. M. Gas back-mixing studies in membrane-assisted bubbling fluidized beds. Chem. Eng. Sci. 2007, 62, 4095−4111. (37) Geldart, D. Types of Gas Fluidization. Powder Technol. 1973, 7, 285−292. (38) Chiba, S.; Chiba, T.; Nienow, A. W.; Kobayashi, H. The minimum fluidisation velocity, bed expansion, and pressure-drop profile of binary particle mixtures. Powder Technol. 1979, 22, 255−269. (39) Upadhyay, R. K.; Roy, S. Investigation of hydrodynamics of binary fluidized beds via radioactive particle tracking and dual-source densitometry. Can. J. Chem. Eng. 2010, 88, 601−610. (40) Geldart, D.; Pope, D. J. Interaction of fine and coarse particle in the freeboard of a fluidized bed. Powder Technol. 1983, 34, 95−97. (41) Beeckmans, J. M.; Bergström, L.; Large, J. F. Segregation mechanisms in gas fluidized beds. Chem. Eng. J. 1984, 28, 1−11. (42) Nienow, A. W.; Chiba, T. Fluidization of dissimilar materials. In Fluidization, 2nd ed.; Davidson, J. F., Clift, R., Harrison, D., Eds.; Academic Press: London, 1985; Chapter 10, pp 357−382. (43) van Dijk, J. J.; Hoffmann, A. C.; Cheesman, D.; Yates, J. G. The influence of horizontal internal baffles on the flow pattern in dense fluidized beds by X-ray investigation. Powder Technol. 1998, 98, 273−278. (44) Geldart, D. Gas Fluidization Technology. John Wiley: Chichester, 1986; p 107. (45) Hoffmann, A. C.; Janssen, L. P. B. M.; Prins, J. Particle segregation in fluidized binary mixtures. Chem. Eng. Sci. 1993, 48, 1583−1592. (46) May, W. G. Fluidized-bed reactor studies. Chem. Eng. Prog. 1959, 55, 49−56. (47) DeGroot, J. H. Scaling-up of gas-fluidized bed reactors. Proceeding of the International Symposium on Fluidization; Drinkenburg, A. A. H., Ed.; Netherlands University Press: Amsterdam, The Netherlands, 1967. (48) Thiel, W. J.; Potter, O. E. The mixing of solids in slugging gas fluidized beds. AIChE J. 1978, 24, 561−569. (49) Yerushalmi, Y.; Avidan, A. High Velocity Fluidization. In Fluidization; 2nd ed.; Davidson, J. F., Clift, R., Harrison, D. Eds.; Academic Press: New York, 1985.

5036

dx.doi.org/10.1021/ie202451d | Ind. Eng. Chem. Res. 2012, 51, 5022−5036