Effect of agglomerate properties on agglomerate stability in fluidized beds

Chemical Engineering Science 63 (2008) 4245 -- 4256

Contents lists available at ScienceDirect

Chemical Engineering Science journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / c e s

Effect of agglomerate properties on agglomerate stability in fluidized beds Sarah Weber a , Cedric Briens a,∗ , Franco Berruti a , Edward Chan b , Murray Gray c a Department

of Chemical and Biochemical Engineering, The University of Western Ontario, London, Ontario, Canada N6A 5B9 Canada Ltd., Research Centre, Edmonton, Alberta, Canada T6H 1H4 c Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada, T6G 2G6 b Syncrude

A R T I C L E

I N F O

Article history: Received 1 February 2008 Received in revised form 21 May 2008 Accepted 23 May 2008 Available online 11 July 2008 Keywords: Fluidization Agglomeration Attrition Particulate processes Wet agglomerate Agglomerate growth

A B S T R A C T

Fluidized bed agglomeration is used to stabilize particulate mixtures and reduce dust emissions. This technology is applied to a variety of production processes for the pharmaceutical, chemical, fertilizer and food industries. In most of these applications, agglomerate stability is an essential criterion. Agglomerates and granules that do not conform to size and shape specifications may create problems in downstream processes, such as tableting, thus compromising process efficiency and product quality. When an agglomerate is formed in a fluidized bed, it can grow by incorporating other bed particles, split into smaller fragments, or be eroded by fluidized bed solids. The objective of the present study is to determine the critical agglomerate liquid content at which the rates of agglomerate growth and shrinkage are balanced when artificial agglomerates made from glass beads and water are introduced into a fluidized bed. This study examined the effects of agglomerate size, agglomerate density, liquid viscosity, binder concentration, and fluidizing gas velocity on the critical initial liquid content. This study found that small agglomerates and low density agglomerates displayed higher critical initial moisture contents. When the viscosity was increased by using sugar solutions, agglomerates were very stable and had very low critical initial moisture contents. The study also found that as the superficial gas velocity increased, the agglomerates started to fragment, rather than erode. © 2008 Elsevier Ltd. All rights reserved.

1. Introduction Particulate operations play a very large role in many industries and their performance is essential to the success of many processes. In industries such as the pharmaceutical industry, agglomerates are intentionally produced (Simons, 1996). In other processes, such as fluid coking, however, agglomerates are not desired (House et al., 2004). Agglomerate properties can potentially be manipulated to ensure survivability, if they are desired, or to enhance destruction if they are undesired. The objective of this study is to investigate how factors such as agglomerate size, agglomerate density, agglomerating liquid properties, and superficial gas velocity affect the stability of agglomerates and, specifically, their critical initial moisture content. The critical agglomerate liquid content is the liquid content at which the rates of agglomerate growth and shrinkage are balanced when artificial agglomerates made from glass beads and water are introduced into a fluidized bed. Agglomerates containing a moisture level above this critical value will survive the fluidized bed conditions, whereas those

with moisture contents below this critical value will begin to be destroyed in the fluidized bed. Water and glass beads, a system which does not tend to form agglomerates because the contact angle between the liquid and solids is low, were used in this study (McDougall et al., 2005). Previous studies have looked at the effect of different parameters on agglomerate growth (Hemati et al., 2003) and on agglomerate growth and destruction (Weber et al., 2006). The present study differs from that of Hemati et al. (2003) because it controls initial agglomerate properties to examine stability, and it expands on our previous work (Weber et al., 2006) by investigating the effect of different parameters on agglomerate behavior. Agglomerates can be destroyed according to various mechanisms. These include external mechanical stresses caused by fluidized bed conditions, internal mechanical stresses caused by chemical reaction and production of vapor, and migration of liquid from the agglomerate to drier bed particles. This study investigated how agglomerate characteristics affected the mechanisms of agglomerate destruction in the absence of reaction or vapor evolution.

1.1. Agglomeration forces ∗

Corresponding author. Tel.: +1 519 661 2131; fax: +1 519 661 3498. E-mail address: [email protected] (C. Briens).

0009-2509/$ - see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2008.05.035

Agglomerates are formed by the aggregation of particulate solids that are held together by short-range physical or chemical forces

4246

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

1.2. Agglomerate growth

Fig. 1. A liquid bridge between two identical particles with diameter d. The liquid bridge here is attracting the particles together (adapted from Simons et al., 1994).

acting among particles, by chemical or physical modification of the particles triggered by specific process conditions, or by substances that act as binders by adhering physically or chemically to form material bridges among particles (Pietsch, 2003). One important binding mechanism for agglomeration requires liquid to be present. In agglomerates and granules, there are three situations where liquid can produce cohesive forces: mobile liquid bridges, adsorbed liquid layers on particles, and adhesive or viscous binders (Sherrington and Oliver, 1981). In the liquid–solid agglomerates studied in this paper, the cohesive forces were generated by liquid bridges. The force exerted by a liquid bridge is caused by the pressure difference across the liquid–gas interface. The force generated by liquid bridges may be repulsive or attractive (Rondeau et al., 2003). The Young–Laplace equation, defining the pressure difference across a liquid bridge, is a function of the surface tension and the curvature of the interface of the liquid bridge (Rondeau et al., 2003):

P = [1/R1 + 1/R2 ]

(1)

R1 and R2 are the principal radii of curvature of the liquid bridge as defined in Fig. 1. The contact angle of the liquid on the solid determines the radii of curvature. When the contact angle is 0◦ , the pressure difference is negative and there is an attractive force between the two particles (Rondeau et al., 2003). When the contact angle is greater than 90◦ , the force between the two particles becomes repulsive. For the intermediate case of partially wetting substances, the liquid bridge may begin in an equilibrium state, and forces may become attractive as the liquid evaporates from the agglomerate structure (Rondeau et al., 2003). Although researchers have developed criteria to indicate when a liquid bridge will rupture (Dai and Lu, 1998), these were not tested because the focus of this study was looking at macro-scale agglomerate failure. The amount of liquid in an agglomerate determines the state of liquid saturation in that agglomerate. In the pendular state of liquid saturation, liquid forms bridges at point contacts within the agglomerate structure. As the pore space of the agglomerate is filled with liquid, the agglomerate saturation state goes to the funicular and finally to the capillary states. In the pendular state, the agglomerate strength is caused by liquid bridges. In the capillary state, the agglomerate strength is caused by capillary pressure. The funicular state is a combination of the two forces (Sherrington and Oliver, 1981). As discussed above, an important aspect of the forces in liquid–solid agglomerates is the interaction of liquids with the solid particles, which can be characterized by the contact angle. Contact angle measurements can be used to assess the wettability of solids. Wetting behavior in agglomerates can lead to two different types of agglomerates. One occurs when the liquid covers the powder, producing a dense and less porous granule, and the other occurs when the powder covers the liquid, producing a more porous granule (Buckton, 1993). Spreading coefficients can be calculated and used to determine properties of granules such as density and size distribution (Buckton, 1993). While smaller contact angles provide better wetting, higher degrees of saturation are then required for successful agglomerate growth (McDougall et al., 2005).

Many studies have been done to examine the effect of different parameters on the growth of agglomerates in granulating processes. Several of these studies have attempted to model the growth of agglomerates by defining the different growth mechanisms (Hemati et al., 2003; Iveson and Litster, 1998). Traditionally, growth mechanisms were divided into categories such as layering, nucleation, coalescence, crushing and layering, and abrasion transfer. These classifications are somewhat arbitrary, therefore a more recent division of agglomerate growth mechanisms includes wetting and nucleation, consolidation and growth, and breakage and attrition (Iveson and Litster, 1998). Some work has been done to investigate growth regime maps for liquid-bound granules. Iveson and Litster (1998) constructed a regime map to qualitatively explain the variations observed in granulation behavior and were able to confirm the regime map experimentally. There are some limitations to these types of regime maps as the boundaries need to be better defined to be very effective (Iveson and Litster, 1998). Researchers have modeled the agglomeration process. Peglow et al. (2006) developed a model based on monodisperse nuclei formation, which allows the extraction of rate constants for agglomeration and growth processes. Saleh et al. (2003) examined fluidized bed coating and agglomeration using the concept of population balance to predict the time evolution of particle size distributions. Granule properties can be linked to granulation rates using multi-dimensional population balance equations that can be reduced to one-dimensional population balance equations by using simplifying assumptions. Biggs et al. (2003) were able to describe the results of high-shear granulation using this approach. Their work defined two model parameters: the aggregation rate constant and the critical binder volume fraction (Biggs et al., 2003). Attempts at modeling the agglomeration process are now considering simultaneous aggregation and breakage behavior. Tardos et al. (1997) introduced a general theoretical framework to look at granule formation, growth, and breakup. More success has been found modeling aggregation behavior rather than breakage kinetics (Tan et al., 2005). More work is required to be able to fully understand the fundamentals of the agglomeration process and manipulate it to increase product consistency. 1.3. Agglomerate destruction Controlling particle and agglomerate size is important for many industrial applications. To modify the size distribution, impact comminution is widely used. Difficulties can arise when unintentional attrition by impact occurs, causing problems in the process due to degradation of particles and granules. By understanding breakage in particles and granules, manufacturing efficiency and product quality can be improved (Salman et al., 2004). In a study by Salman et al. (2004), wet granules were studied using impact tests. They found that granule failure could be classified by the impact velocity. For low impact velocities, plastic deformation was observed in wet granules. At high impact velocities, wet granules were greatly reduced in size due to fragmentation. This study also found that small wet granules showed significant plastic deformation before failure. This differed from large wet granules, which exhibited localized debris formation and chipping (Salman et al., 2004). Salman et al. (2003) also found that increasing granule size caused agglomerate breakage to occur at lower impact velocities in impact studies of fertilizer granules. Breakage patterns and failure modes are dictated by the agglomerate structure and the velocity of the agglomerate impact (Salman et al., 2003, 2004; Subero and Ghadiri, 2001). The breakage pattern affects the size distribution of the impact product (Subero and Ghadiri, 2001). Researchers have

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

found that it is very difficult to model agglomerate failure because of the complex processes that are occurring, as well as the difficulty accounting for the structure of the agglomerate material (Subero and Ghadiri, 2001). The impact tests conducted by Subero and Ghadiri (2001) indicate that there are two main types of breakage. The first is localized damage and the second is distributed damage. Tardos et al. (1997) looked at the use of a critical Stokes number to define when granules will break when exposed to shearing in different types of granulators. Moreno-Atanasio and Ghadiri (2006) used a mechanistic model that relates the number of broken contacts in an agglomerate due to impact velocity, properties of the particles that form the granule, and the interparticle adhesion energy. This study found that the amount of damage was related to a dimensionless group, , defined by the following equation (Moreno-Atanasio and Ghadiri, 2006):

 = (D5/3 E2/3 V 2 )/ 5/3

(2)

Other studies have examined the effect of hydrodynamic forces on the dispersion of agglomerates. It has been found that there are two mechanisms of agglomerate dispersion; adhesive and cohesive failure. Adhesive failure occurs in agglomerates that are partially infiltrated with fluid. Cohesive failure occurs in agglomerates that have no infiltrating liquid or in fully infiltrated agglomerates. Erosion and fragmentation of agglomerates is caused by cohesive failure. The cohesivity of an agglomerate depends on the physical nature of the powder and its packing structure. Increasing the hydrodynamic forces causes agglomerate dispersion to shift from erosion to fragmentation (Boyle et al., 2005). Shamlou et al. (1990) studied the hydrodynamic influences on particle breakage in fluidized beds by studying a fluidized bed of granules made from soda glass beads and polymer. This study found that the size of the bed particles drops continuously until a plateau value is reached. This study also found that the mode of bed material breakage occurs predominantly by attrition rather than fragmentation and that this attrition is most likely caused by low energy impacts in the bed core (Shamlou et al., 1990). The complexity of both the agglomerate material and the forces exerted on the agglomerate by the fluidized bed environment require more study to fully understand agglomerate destruction mechanisms.

4247

bed will experience different forces and stresses within the fluidized bed than agglomerates that remain fluidized. 2. Materials and methods 2.1. Agglomerate preparation A syringe–piston system was used for agglomerate fabrication. For predetermined moisture contents, known masses of solids and liquid were combined in a container and thoroughly mixed. A constant mass of wet solids was placed in the syringe–piston system and compressed to a predefined length. To test the effect of agglomerate size on agglomerate stability, three cylindrical molds were used. The largest mold had a diameter of 0.0169 m and a height of 0.0169 m, the smallest mold had a diameter of 0.0067 m and a height of 0.0070 m, and the intermediate mold had a diameter of 0.0116 m and a height of 0.0120 m. One of these molds was placed inside the syringe and the piston was fully compressed. The mold was pushed out of the end of the syringe and removed, leaving an agglomerate sitting on the bench top. Fig. 2

1.4. Segregation in fluidized beds Agglomerates can tend to segregate in fluidized bed operations. In the polymer industry, the size distribution of particles taken from the bottom of a reactor can be very different from the size distribution of particles taken from the top of the reactor (Kim and Choi, 2001). The most influential parameter in fluidized bed segregation is particle density differences. Severe axial particle size distributions can also be observed in a fluidized bed of powder of a continuous size distribution at velocities that are much higher than the minimum fluidization velocity according to the study by Hoffmann and Romp (1991) (Kim and Choi, 2001). To model the segregation of particles in a gas phase fluidized bed, Kim and Choi (2001) postulated that the segregation of a particle mixture of equal density will occur because of the size-dependent transfer of particles between the wake and bulk phases as gas bubbles rise in the fluidized bed. In fluidized beds of group C powders, agglomerates are segregated by size (Xu and Zhu, 2005). Xu and Zhu (2005) found large differences in particle sizes depending on where in the fluidized bed the particles were removed. Large agglomerates were found at the bottom of the fluidized bed and small agglomerates were found at the top of the fluidized bed. Segregation can affect the behavior of agglomerates within the fluidized bed. Agglomerates that tend to settle at the bottom of the

Fig. 2. Examples of agglomerates used in this study after fabrication. (A) A large agglomerate (diameter = 0. 0169 m). (B) A small agglomerate (diameter = 0. 0067 m).

Table 1 Agglomerate properties before fluidization Agglomerate diameter (m)

Agglomerate height (m)

Average mass before fluidization (g)

Average agglomerate density (kg/m3 )

Porosity (dry basis)

0.0067 0.0067 0.0116 0.0169

0.0070 0.0070 0.120 0.0169

0. 37 ± 0. 01 0. 34 ± 0. 01 1. 93 ± 0. 02 5. 86 ± 0. 04

1484 ± 37. 8 1375 ± 43. 9 1524 ± 13. 3 1547 ± 9. 7

0.40 0.45 0.39 0.38

The intervals are the standard deviation of the measured value. The properties of the primary particles used to make the agglomerates are:dpsm =179 m, p =2500 kg/m3

4248

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

Fig. 3. Schematic diagram of experimental fluidized bed equipment.

shows an example of the largest and smallest agglomerates used in these experiments after fabrication. The reproducibility of the fabricated agglomerates is given in Table 1 based on agglomerate mass before fluidization. For this study, the agglomerates were made with glass beads and water or aqueous sugar solutions. The particle density of the spherical glass beads was 2500 kg/m3 and the Sauter-mean diameter was 179 m. On average, agglomerates were made with a density of 1520 kg/m3 . To observe the effect of agglomerate density, some agglomerates were also made with a density of 1375 kg/m3 . To observe the effect of the concentration of binder solutions on agglomerate stability, different aqueous solutions of white table sugar were used to form the agglomerates. The concentration of the sugar solutions ranged from 0 to 40 wt% sugar. 2.2. Critical desktop experiments Agglomerates were formed with different moisture contents to determine the minimum and maximum moisture contents that are required for an agglomerate to maintain its cylindrical shape when formed and placed on a flat horizontal surface. The effect of agglomerate size on the minimum and maximum values was investigated using the three agglomerate molds described above.

fluidization (m). If the agglomerates had fragmented during fluidization, the mass of all of the agglomerate fragments were weighed to determine their cumulative mass, mfragments . The number of agglomerate fragments was also estimated. When aqueous sugar solutions were used to form agglomerates, some fluidized bed experiments were done using humidified air with a 52–55% relative humidity. 2.4. Segregation experiments Agglomerates were made using both the large and small agglomerate molds. These agglomerates had an average density of approximately 1500 kg/m3 . Three large agglomerates or six small agglomerates were added at a time to the fluidized bed. The bed was fluidized at 0.17 m/s because this was the velocity used to investigate the effects of agglomerate size and liquid viscosity in this study. The fluidizing air was turned off 30 s after the introduction of agglomerates, causing the bed to quickly defluidize. The position of the agglomerates in the bed was determined by removing bed particles slowly using a vacuum cleaner. As each agglomerate was uncovered, its position was measured. 3. Results and discussion

2.3. Fluidized bed experiments

3.1. Critical desktop experiments

A bed formed with the same glass beads as the agglomerates was fluidized in a clear column with an internal diameter of 0.10 m. The static bed height was approximately 0.15 m. The bed was fluidized with compressed air through a distributor consisting of a polyethylene disk with 70 m pores. A large expansion section above the bed helped return entrained particles back to the bed. A schematic diagram of the fluidized bed equipment is shown in Fig. 3. Agglomerates were formed with a specified moisture content and weighed to determine their initial mass (m0 ). All agglomerate masses were determined using an accurate scale. They were then inserted into the bed at approximately minimum fluidization conditions. The fluidization air was then turned off and the expansion section of the fluidized bed column was attached. A second fluidization line was then opened to fluidize the bed at the desired superficial velocity for a predetermined time interval. The fluidization air was turned off, the expansion section of the column was removed, and the bed contents were emptied. The agglomerates, if they had survived fluidization, were recovered and weighed to determine their mass after

The minimum and maximum amount of moisture for an agglomerate to maintain its shape on a horizontal flat desktop was investigated. This was used to investigate the effect of the cohesive force on the agglomerate structure versus the stress due to gravity. The effect of the agglomerate size on the minimum and maximum moisture contents was studied. The results are shown in Fig. 4. The error bars in Fig. 4 represent the moisture contents that bracket the minimum and maximum moisture contents for each agglomerate size. The true minimum and maximum moisture contents lie between these two values. The minimum moisture content required to form stable agglomerates is very low (< 1 wt%), and Fig. 4 shows that agglomerates with a small agglomerate diameter require more moisture to form a stable agglomerate on a flat horizontal surface than agglomerates with larger diameters. This was a surprising result. An explanation for this behavior is the effect of the agglomerate size on the internal structure of that agglomerate. Agglomerates were made to have the same density, independent of the size of the agglomerate. It can be assumed that the packing and pore structure

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

4249

(U/Umf = 6. 5) was chosen because it was used in a previous study to observe the effects of different agglomerate properties on stability (Weber et al., 2006). The maximum fluidization time was selected to be 90 s because, in some reactors, it has been found that agglomerates may be stabilized by reaction products and survive for very long periods of time. The points bracketing the value m/m0 = 1 were used with linear interpolation to determine the critical initial moisture content as well as the 95% confidence interval for this value. During these fluidized bed experiments, some evaporation of the water used to form the agglomerates will occur, causing the moisture content of the agglomerate to change with time which might affect the survivability of the agglomerate in the fluidized bed. Some experiments were conducted using humidified air to greatly reduce evaporation and showed only small differences with the drier air case. This will be discussed further in Section 3.2.4.

Fig. 4. Relationship between the minimum and maximum moisture contents required to form a stable agglomerate on a flat horizontal surface and the agglomerate diameter. The error bars represent the moisture contents that bracket the true minimum and maximum values.

of the agglomerate is the same, independent of size. If the pores are the same size in all of the agglomerates, then a pore in the small agglomerate is a larger percentage of the structure than the same pore in the large agglomerate. This may cause the smaller agglomerate to be weaker than the larger agglomerate. If the small agglomerate is weaker, therefore, less stable, then more moisture is required to create a stable agglomerate for this agglomerate size than for the larger agglomerates. Without the influence of the fluidized bed, these experiments show that when agglomerates are formed with very low moisture contents, smaller agglomerates require more moisture to form a stable agglomerate. The maximum moisture content that can form a stable agglomerate on a flat horizontal surface is also shown in relation to the agglomerate diameter in Fig. 4. The maximum moisture content values are between 20 and 23 wt%. The agglomerates with a large diameter were able to maintain stable agglomerate structures with slightly higher moisture contents than the agglomerates with the smallest agglomerate diameter. Small agglomerates are unable to sustain extreme conditions in their structure, whether it is very little moisture or high moisture contents. Small agglomerates are more unstable. These values represent the extreme conditions that agglomerates can withstand without the interference of external forces other than gravity acting on their structures. These experiments help identify the extreme conditions that agglomerates can survive, especially for the fluidized bed as fluidization time goes to zero. As the time decreases, the amount of force that the agglomerate is exposed from the fluidized bed decreases. 3.2. Fluidized bed experiments Experiments were conducted to determine a critical initial moisture content where the mechanisms of agglomerate destruction and agglomerate growth are balanced and no net change in the agglomerate mass is observed. This critical concentration was determined by graphing the ratio m/m0 as a function of the initial moisture, as shown in Fig. 5. The superficial gas velocity of 0.17 m/s

3.2.1. Effect of agglomerate size—erosion regime The effect of agglomerate size on the critical initial moisture content was studied using three mold sizes. The agglomerates were fluidized at U = 0. 17 m/s for different time intervals. At this superficial gas velocity, the agglomerates were observed to erode rather than fragment. The results are shown in Fig. 6. Fig. 6 shows that for short residence times in the fluidized bed, the critical moisture content required for the agglomerate to maintain its initial mass decreases with increasing agglomerate size. The confidence intervals for these values are quite narrow. As the agglomerate residence time in the fluidized bed increases, the critical initial moisture content increases. When the residence time is 60 s, the increasing trend levels off and the critical initial moisture content becomes the same for both large and medium agglomerates. For fluidization times greater than 60 s, the medium and large agglomerates reach a dynamic equilibrium between the addition of solids and the erosion of solids, although they eventually slowly lose moisture. Therefore, the agglomerate size will then slowly decrease with time as erosion acts on the agglomerate structure. The critical initial moisture content for the smallest agglomerate size is different from the other two sizes and shows an increasing trend over the times studied. At the superficial gas velocity of 0.17 m/s (U/Umf = 6. 5), the main mechanism of agglomerate destruction was erosion. When the fluidized bed contents were screened after fluidization, a single large agglomerate core was observed with no other fragments. The data of Fig. 6 suggest that the smallest agglomerates, under these conditions, required more moisture in the agglomerate structure to overcome the erosion forces. This finding disagrees with the findings of Tardos et al. (1997), who found that a granule that is larger than a critical value will become unstable and fragment. However, erosion, not fragmentation, was the observed behavior under our experimental conditions, and the erosion mechanism may be affected by agglomerate size differently than the fragmentation process of agglomerate destruction. The ratio of surface area to volume of the large agglomerates is the smallest of the three agglomerate sizes. In the large agglomerates, it is likely that there is enough liquid in the agglomerate to provide sufficient moisture on the agglomerate surface to incorporate bed particles into the agglomerate structure and compensate for erosion forces acting on the agglomerate over a longer period of time than the smallest agglomerates. To investigate whether agglomerates fluidized differently depending on agglomerate size, segregation experiments were performed. The results are shown in Fig. 7. The segregation of agglomerates was investigated by fluidizing the largest and the smallest agglomerates for 30 s at a superficial gas velocity of 0.17 m/s to observe the two extremes. The experiments using large agglomerates were repeated six times and the experiments using small

4250

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

Fig. 5. Example of graphs used to determine the critical initial moisture content. This graph shows the behavior of an agglomerate made with the large mold, U = 0. 17 m/s. Fluidization time = 30 s (A) and 60 s (B).

Fig. 6. Effect of agglomerate size on critical initial moisture content at U = 0. 17 m/s. The error bars are 95% confidence intervals for each value.

Fig. 7. Agglomerate position within the fluidized bed. Agglomerates were fluidized for 30 s at U = 0. 17 m/s.

agglomerates were repeated three times. The results were pooled together to form one large population for each agglomerate size. The percentages presented in Fig. 7 were calculated from these larger populations. Segregation of agglomerates in fluidized beds can influence the behavior of agglomerates. Agglomerates may experience different forces and may be destroyed by different mechanisms depending on their position in the fluidized bed.

The distribution of agglomerates in the fluidized bed was found to depend on the size of the agglomerate. The highest percentage of small agglomerates was found in the middle portion of the fluidized bed. The highest percentage of large agglomerates was found in the area above the distributor region. There were no large agglomerates in the middle portion of the fluidized bed, although 17% of large agglomerates made it to the top portion of the bed. After 30 s of

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

4251

the agglomerate to travel throughout the fluidized bed more than the more dense agglomerate, or to fall at a lower velocity through the emulsion. Increased motion in the fluidized bed may expose the agglomerates to more shear and erosion forces, requiring a higher liquid saturation to recruit and maintain particles in the agglomerate structure. Fig. 8 shows that higher moisture contents are required to balance the agglomerate growth and reduction processes in the fluidized bed for lower density agglomerates.

Fig. 8. Effect of agglomerate density on the critical initial moisture content. Agglomerates were made using the small mold (diameter = 0. 0067 m) and U = 0. 17 m/s. The error bars are 95% confidence intervals for each value.

fluidization with a superficial gas velocity of 0.17 m/s, only small portions of both sizes of agglomerates were found to be settled on the distributor. There are two possible explanations of why no large agglomerates were observed in the middle portion of the bed. Large agglomerates may not travel to the middle and top portions of the fluidized bed as frequently as smaller agglomerates or they may travel as frequently to these areas but settle more quickly through the emulsion phase than smaller agglomerates. Large agglomerates appear to have more of a tendency to segregate to the lower regions of the fluidized bed, however, not many were found to be directly on the distributor. The results shown in Fig. 7 indicate that the increased stability of large agglomerates observed in fluidized beds is not due to segregation alone. The behavior of agglomerates in the fluidized bed is a combination of many effects dictated by the size of agglomerates. 3.2.2. Effect of agglomerate density—erosion regime The effect of agglomerate density was investigated by making agglomerates using the small mold. Two agglomerate densities were tested and the results are shown in Fig. 8. The superficial gas velocity was 0.17 m/s and the dominant mechanism of agglomerate destruction at this velocity was erosion. Fig. 8 shows that the density of an agglomerate does play a role in the survival of agglomerates in a fluidized bed. Different agglomerate densities were achieved by changing the agglomerate porosity. Agglomerate density was increased by adding more agglomerate material to the mold and compressing it to the same height. When agglomerates were formed with higher densities, a lower critical initial moisture content was observed. This indicates that denser agglomerates are better able to withstand erosion. As the time that the agglomerate was exposed to the fluidized bed environment increased to 60 s, more variability was seen in the data, as shown by wider confidence intervals. Lower agglomerate densities are associated with more pore space in the agglomerate structure. This makes the agglomerates weaker and more susceptible to damage in the fluidized bed. This agrees with the findings of Subero and Ghadiri (2001) who found that agglomerate structure plays a very large role in the destruction of agglomerates. The lower density may also cause

3.2.3. Effect of superficial gas velocity To determine the effect of the superficial gas velocity on agglomerate stability, agglomerates were made using all three agglomerate molds. Three agglomerate moisture contents were tested: 9.1, 4.8, and 1 wt%. Agglomerates were then fluidized in a bed of small glass beads at different superficial gas velocities. These superficial gas velocities were 0.17 m/s (U/Umf =6. 5), 0.26 m/s (U/Umf =10. 0), 0.31 m/s (U/Umf =11. 9), 0.34 m/s (U/Umf =13. 0), and 0.40 m/s (U/Umf =15. 4). During the experiments, two different behaviors were observed. When the superficial gas velocity was low, the agglomerates were seen to erode if the moisture content was low, or gain mass if the moisture content was high. Agglomerates were recovered from the bed in one piece. When the velocity was high, however, agglomerates were seen to fracture into several smaller pieces. When this occurred, all of the agglomerate fragments were weighed to get their cumulative mass (mfragments ). The number of fragments was also estimated. When erosion was the dominant mechanism of agglomerate destruction, the number of fragments was equal to 1. The relationship between mfragments /m0 and superficial gas velocity for large agglomerates is shown in Fig. 9(A). The relationship between the number of agglomerate fragments and superficial gas velocity is shown in Fig. 9(B). Fig. 9(A) shows that at the lowest superficial gas velocity, agglomerates with 1 wt% moisture content lose mass. Agglomerates with higher initial moisture contents gain mass from the fluidized bed at this superficial gas velocity. Fig. 9(B) shows that at the lowest superficial gas velocity, all agglomerates remained in one piece. As the superficial gas velocity increases to approximately 0.26 m/s, Fig. 9(B) shows that the number of agglomerate fragments increases for agglomerates made with moisture contents of 1 and 4.8 wt%, while agglomerates with the highest moisture content showed no increase in the number of fragments. At this superficial gas velocity, agglomerates with a moisture content of 4.8 wt% have enough moisture at the surface of the fragments to continue to recruit more bed particles than are lost to erosion. This is shown by mfragments /m0 greater than 1 in Fig. 9(A). Agglomerates with a moisture content of 1 wt% do not have enough moisture at the surface of the fragments to recruit more bed particles than are lost to erosion and this is shown by mfragments /m0 less than 1. As the velocity increases further to approximately 0.31 m/s, the agglomerates with the highest moisture content begin to fragment as shown in Fig. 9(B). The fragmentation of these agglomerates causes the mfragments /m0 ratio to increase as shown in Fig. 9(A). This indicates that these agglomerates have enough moisture at the surface of the fragments to recruit more bed particles than agglomerates with lower moisture contents. Agglomerates made with the lowest moisture content continue to lose mass as erosion is dominant over particle recruitment. As the velocity continues to increase, agglomerates with 4.8 and 9.1 wt% moisture contents have increasing numbers of fragments. The number of fragments is very similar for these two groups of agglomerates. Agglomerates with the lowest moisture content continue to exhibit fragmentation behavior, however, the number of fragments is lower compared with the other agglomerates. The number of fragments also eventually starts to decrease. This behavior may be caused by the complete destruction of the smallest fragments of these agglomerates, decreasing the number of fragments observed.

4252

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

Fig. 9. (A) Relationship between mfragments /m0 and the superficial gas velocity. Agglomerates were made using the largest mold (diameter = 0. 0169 m). Measurements were taken after 30 s of fluidization. (B) Effect of superficial gas velocity on the number of agglomerate fragments. Agglomerates were made using the largest mold (diameter = 0. 0169 m). Measurements were taken after 30 s of fluidization. Legend values are agglomerate initial moisture contents.

Fig. 10. (A) Relationship between mfragments /m0 and the superficial gas velocity. Agglomerates were made using the medium mold (diameter = 0. 0116 m). Measurements were taken after 30 s of fluidization. (B) Effect of superficial gas velocity on the number of agglomerate fragments. Agglomerates were made using the medium mold (diameter = 0. 0116 m). Measurements were taken after 30 s of fluidization.

The effect of the superficial gas velocity was also observed in agglomerates made with the medium mold (diameter = 0. 0116 m). The results are shown in Figs. 10(A) and (B).

As was observed in Fig. 9(A), medium sized agglomerates with low moisture contents had mfragments /m0 ratios less than 1 and the agglomerates with higher moisture contents had mfragments /m0

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

4253

Fig. 11. (A) Relationship between mfragments /m0 and the superficial gas velocity. Agglomerates were made using the smallest mold (diameter = 0. 0067 m). Measurements were taken after 30 s of fluidization. (B) Effect of superficial gas velocity on the number of agglomerate fragments. Agglomerates were made using the smallest mold (diameter = 0. 0067 m). Measurements were taken after 30 s of fluidization.

ratios greater than 1. One difference that was observed was that the mfragments /m0 ratio reached 0 at a lower superficial gas velocity for medium sized agglomerates with a moisture content of 1 wt% compared with the large agglomerates. Fig. 10(B) shows the number of fragments for medium sized agglomerates. As was seen in Fig. 9(B), the number of fragments increased as the superficial gas velocity increased for agglomerates with high moisture content. For drier agglomerates, there was an initial increase in the number of fragments with increasing superficial gas velocity. The number of fragments started to decrease as the superficial gas increased further, showing that some of the fragments were eroded away. One difference between Figs. 9(B) and 10(B) is the number of fragments that were observed. The medium sized agglomerates fragmented into fewer pieces than large agglomerates. Figs. 11(A) and (B) show the effects of superficial gas velocity on agglomerates made with the smallest mold (diameter = 0. 0067 m). As shown in Figs. 9(A) and 10(A), the effect of moisture content is generally very similar for all sizes of agglomerates. The effect of agglomerate size is seen in the magnitude of mfragments /m0 . Small agglomerates with a moisture content of 1 wt% were unable to incorporate enough bed materials into their structures to balance the erosion and fragmentation process as was seen for larger agglomerates with the same moisture content. The difference in the behavior observed in Fig. 11(A) is that the mfragments /m0 ratio is lower for the smallest agglomerates at this moisture content. Smaller agglomerates with higher moisture contents were able to recruit enough bed particles to maintain the mfragments /m0 ratio greater than 1. This ratio was not as high for the smallest agglomerates with the highest moisture content as larger agglomerates with the same moisture content. Fig. 11(B) shows that the smallest agglomerates had the lowest number of fragments of the three agglomerate sizes. The driest agglomerates had very few fragments survive at the highest superficial gas velocity.

Fig. 12 shows a large and a small agglomerate with a moisture content of 9.1 wt% after recovery from the fluidized bed after 30 s of fluidization. Fig. 12 shows the difference in the number of fragments as agglomerate size decreases when agglomerate behavior is in the fragmentation regime. Two destruction processes are occurring simultaneously in the fluidized bed. As fragmentation is occurring, the fragments are also undergoing erosion. This contributes to decreasing the mfragments /m0 ratio for the agglomerates with moisture contents of 1 and 4.8 wt% for all agglomerate sizes. Erosion has a large impact on the decrease of the mfragments /m0 ratio for the driest agglomerates at higher velocities, accounting for the low values of this ratio and the decreasing number of fragments at the highest superficial gas velocity. As the size of the agglomerates decreases, the number of fragments also decreases. Agglomerates fragment until the pieces reach a size that can survive in the fluidized bed. Smaller agglomerates do not have to fragment as much to reach the critical size to survive at a certain fluidized bed condition. That is why a decrease in the number of fragments was observed with decreasing size. The reduction in mfragments /m0 ratio and number of fragments observed as further erosion of the fragments occurs is in agreement with the findings by Shamlou et al. (1990). Several studies have used impact testing of agglomerates to examine breakage phenomena. Salman et al. (2004) examined the impact failure modes of spherical particles, including wet agglomerates. Although impact destruction in their study is different from the mechanism of destruction in the fluidized bed, the fragmentation of the agglomerates in the fluidized bed looked very similar to the high velocity failure of wet agglomerates in this study (Salman et al., 2004). Lower moisture contents in the fluidized bed study caused the agglomerates to be more susceptible to fragmentation because granule strength depends on local agglomerate structure (Subero and Ghadiri, 2001). With less moisture available in the agglomerates, fewer liquid contact points can be

4254

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

Fig. 12. Examples of large (0.0169 m) and small (0.0067 m) agglomerate behavior in fluidized beds. Moisture content was 9.1 wt%. Large agglomerate (A) before fluidization, (B) after fluidization at 0.17 m/s (growth), and (C) after fluidization at 0.40 m/s (fragmentation). Small agglomerate (D) before fluidization, (E) after fluidization at 0.17 m/s (growth), and (F) after fluidization at 0.40 m/s (fragmentation).

Table 2 Properties of sucrose solutions (adapted from Reiser et al., 1995) Concentration of solution (wt%)

Viscosity at 20 ◦ C (cP)

Surface tension (mN/m)

0 30 40

1.005 3.187 6.167

72.68 75.89 77.08

maintained within their structure, causing them to be weaker than that of the agglomerates made with more liquid. Fragmentation can cause many problems in processes where agglomerates are not desired because it can potentially increase the amount of agglomerate material as more fragments are created and can recruit surrounding particles into the structure. If agglomerated material is not desired, then reducing the moisture content of the agglomerates can promote fragmentation and erosion, reducing the amount of agglomerate material left in the fluidized bed. If agglomerates are desired, the opposite is true. Decreasing the size of agglomerates will cause an agglomerate to lose mass more quickly, but will decrease the number of fragments. 3.2.4. Effect of binder solution concentration Aqueous sugar solutions were used to investigate the effect of binder concentration on the critical initial moisture content. They were also used to investigate the effect of liquid viscosity on the critical initial moisture content. These experiments were done using the largest agglomerate mold. The viscosity of the sugar solutions was measured and found to be very similar to literature values for aqueous sucrose solutions when adjusted for temperature. Properties of pure sucrose solutions at the same concentrations as those used for the experiments are found in Table 2. The sugar solutions have very different viscosities when compared with pure water, although the surface tension remains very similar. The critical initial moisture content was found for each sugar solution and they are shown in Fig. 13 in comparison with the distilled water case. The superficial gas velocity was 0.17 m/s and the dominant mechanism of agglomerate destruction at this velocity was erosion.

Fig. 13. Effect of sugar concentration on the critical initial moisture content. Agglomerates were made using the largest mold (diameter = 0. 0169 m) and U = 0. 17 m/s. The error bars are 95% confidence intervals for each value.

Fig. 13 shows that there is a dramatic difference in the stability of agglomerates made from sugar solutions when compared with agglomerates made with distilled water. Much less liquid is required to balance the agglomerate growth and reduction mechanisms when sugar solutions are used. Another difference between the two situations is that the critical initial moisture content is not greatly affected by the fluidization time and this is shown in an almost horizontal line for the agglomerates made with sugar solutions. Less solution

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

4255

increase is not very large. Fig. 14 also shows that when the agglomerate moisture content was higher, the humidification made the agglomerate less stable. It was expected that humidifying the fluidization air would prevent evaporation of water from he sugar solution used to form the agglomerate. This would prevent an increase in the sugar concentration, and its resulting increase in liquid viscosity. For this reason, the decreased stability observed in agglomerates with the highest moisture content using humidified air was an expected result. In general, Fig. 14 shows that humidification did not have a large effect on the agglomerate stability. The change in viscosity of the sugar solution due to the evaporation of water during fluidization was not occurring to a large extent during the previous experiments. 4. Conclusions Agglomerate behavior in fluidized beds is affected by many variables. The variables studied in this paper include fluidization velocity, agglomerate size, agglomerate density, and binder solution concentration. Critical desktop experiments were also done to observe the effect of agglomerate diameter on agglomerate stability without the influence of forces exerted by a fluidized bed. This study found that: Fig. 14. Effect of humidification of fluidization air. The liquid used to make the agglomerates was 30 wt% sugar solution. Agglomerates were made using the largest mold (diameter = 0. 0169 m), U = 0. 17 m/s, and fluidization time was 90 s.

is required to create a stable agglomerate structure when compared with the pure water case. The increase in the sugar concentration from 30 to 40 wt% causes the viscosity of the solution to approximately double. It was expected that this would have a large impact on the critical initial moisture content observed, but this was not the case. The critical initial moisture contents for both sugar solutions over time were very similar. In most cases, overlap was observed. The results from Fig. 13 show that the initial increase in liquid viscosity from 1 to 3.2 cP had a large impact on the critical initial moisture content. Beyond a liquid viscosity of 3.2 cP, in this agglomerate system, increasing the viscosity did not greatly impact the critical initial moisture content. This indicates that beyond a certain liquid viscosity, increasing the liquid viscosity further does not necessarily have a great impact on the stability of the agglomerate. This agrees with the findings of Hemati et al. (2003) and Pont et al. (2001) who found that the dominant forces in granulation are capillary forces and that the viscosity of the solution does not influence agglomerate growth as much as interfacial parameters. In this study, if the liquid viscosity was sufficiently low, the findings of Hemati et al. (2003) and Pont et al. (2001) did not apply. To determine whether the behavior in Fig. 13 occurred because of water evaporating from the solution during fluidization, causing the sugar concentration and the viscosity to increase, experiments were conducted using humidified air. Humidified air had a relative humidity that was between 52% and 55%. Normally, the air used to fluidize the bed had a low relative humidity, less than 10%. The superficial gas velocity was once again 0.17 m/s and the dominant mechanism of agglomerate destruction at this velocity was erosion. The results are shown in Fig. 14. It was observed that when the fluidization air had a relative humidity of 60%, the bed could not be fluidized because the particles had caked together. This can happen at a relative humidity less than 100% because of fine dust that may be stuck on the surface of the particles (Williams and Nosker, 1988). For this reason, the relative humidity of the fluidization air was kept between 52% and 55%. The bed appeared to be well fluidized at this level of humidification. Fig. 14 shows that humidifying the air increased the amount of particles incorporated into the agglomerate structure, although the

• Agglomerate diameter does affect the minimum and maximum moisture contents required to maintain its shape on a horizontal surface. Larger agglomerates can tolerate more extreme maximum and minimum moisture contents. • When erosion is the dominant mechanism of destruction, larger agglomerates are more stable in fluidized beds. This increased stability was not caused by complete segregation of the agglomerates to the bottom of the fluidized bed. • When erosion is dominant, agglomerates with higher densities were more stable and required less moisture in their structure to maintain their initial mass. • Increasing the superficial gas velocity caused agglomerates to fragment and secondary fragmentation and erosion destroyed fragments if the initial moisture content was low. Agglomerates with high initial moisture contents fragmented and recruited more bed particles than drier agglomerates. Decreasing agglomerate size caused the number of agglomerate fragments to decrease. • Increasing the liquid viscosity to a certain level, 3 cP in this study, caused the agglomerates to be much more stable. Further increases in the liquid viscosity did not make the agglomerates more stable, agreeing with literature findings. The results from this study show that more work must be done to understand agglomerate destruction in fluidized beds but provides directional information about agglomerate destruction. More research needs to be done to see how these principles apply to industrial agglomerating systems. Notation d D E m mfragments m0 P R1 R2 t

particle diameter, m primary agglomerate diameter, m elastic modulus, Pa mass of agglomerate after fluidization, g mass of fragments recovered from fluidized bed after fluidization, g mass of agglomerate after agglomerate formation, g pressure difference across a liquid bridge, Pa principal radius of curvature (as defined in Fig. 1) principal radius of curvature (as defined in Fig. 1) time

4256

U Umf V

S. Weber et al. / Chemical Engineering Science 63 (2008) 4245 -- 4256

superficial gas velocity, m/s minimum fluidization velocity, m/s particle velocity, m/s

Greek letters

    p

liquid surface tension, mN/m interface energy, J/m2 dimensionless group for description of breakage agglomerate density, kg/m3 particle density, kg/m3

Acknowledgements The authors acknowledge the support of the Natural Sciences and Engineering Research Council of Canada and Syncrude Canada Ltd. for their funding and support for this study. The authors would also like to thank Ahmad Abu Romeh for his assistance with the experimental program. References Biggs, C.A., Sanders, C., Scott, A.C., Willemse, A.W., Hoffman, A.C., Instone, T., Salman, A.D., Hounslow, M.J., 2003. Coupling granule properties and granulation rates in high-shear granulation. Powder Technology 130, 162–168. Boyle, J.F., Manas-Zloczower, I., Feke, D.L., 2005. Hydrodynamic analysis of the mechanisms of agglomerate dispersion. Powder Technology 153, 127–133. Buckton, G., 1993. Assessment of the wettability of pharmaceutical powders. Journal of Adhesion Science and Technology 7, 205–219. Dai, Z., Lu, S., 1998. Liquid bridge rupture distance criterion between spheres. International Journal of Mineral Processing 53, 171–181. Hemati, M., Cherif, R., Saleh, K., Pont, V., 2003. Fluidized bed coating and granulation: influence of process-related variables and physiochemical properties on the growth kinetics. Powder Technology 130, 18–34. House, P.K., Saberian, M., Briens, C.L., Berruti, F., Chan, E., 2004. Injection of a liquid spray into a fluidized bed: particle-liquid mixing and impact on fluid coker yields. Industrial & Engineering Chemistry Research 43, 5663–5669. Iveson, S.M., Litster, J.D., 1998. Growth regime map for liquid-bound granules. A.I.Ch.E. Journal 44, 510–1518. Kim, J.Y., Choi, K.Y., 2001. Modeling of particle segregation phenomena in a gas phase fluidized bed olefin polymerization reactor. Chemical Engineering Science 56, 4069–4083.

McDougall, S., Saberian, M., Briens, C., Berruti, F., Chan, E., 2005. Effect of liquid properties on the agglomerating tendency of a wet gas–solid fluidized bed. Powder Technology 149, 61–67. Moreno-Atanasio, R., Ghadiri, M., 2006. Mechanistic analysis and computer simulation of impact breakage of agglomerates: effect of surface energy. Chemical Engineering Science 61, 2476–2481. ¨ Peglow, M., Kumar, J., Warnecke, G., Heinrich, S., Morl, L., 2006. A new technique to determine rate constants for growth and agglomeration with size- and timedependent nuclei formation. Chemical Engineering Science 61, 282–292. Pietsch, W., 2003. An interdisciplinary approach to size enlargement by agglomeration. Powder Technology 130, 8–13. Pont, V., Saleh, K., Steinmetz, D., Hémati, M., 2001. Influence of physicochemical properties on the growth of solid particles by granulation in fluidized bed. Powder Technology 120, 97–104. Reiser, P., Birch, G.G., Mathlouthi, M., 1995. Physical properties. In: Mathlouthi, M., Reiser, P. (Eds.), Sucrose. Blackie Academic & Professional, New York, p. 186. Rondeau, X., Affolter, C., Komunjer, L., Clausse, D., Guigon, P., 2003. Experimental determination of capillary forces by crushing strength measurements. Powder Technology 130, 124–131. Saleh, K., Steinmetz, D., Hemati, M., 2003. Experimental study and modeling of fluidized bed coating and agglomeration. Powder Technology 130, 116–123. Salman, A.D., Fu, J., Gorham, D.A., Hounslow, M.J., 2003. Impact breakage of fertiliser granules. Powder Technology 130, 359–366. Salman, A.D., Reynolds, G.K., Fu, J.S., Cheong, Y.S., Biggs, C.A., Adams, M.J., Gorham, D.A., Lukenics, J., Hounslow, M.J., 2004. Descriptive classification of the impact failure modes of spherical particles. Powder Technology 143–144, 19–30. Shamlou, P.A., Liu, Z., Yates, J.G., 1990. Hydrodynamic influences on particle breakage in fluidized beds. Chemical Engineering Science 45, 809–817. Sherrington, P.J., Oliver, R., 1981. Granulation. Heyden & Son, Ltd., Philadelphia. pp. 7–59, 153–165. Simons, S.J.R., 1996. Modelling of agglomerating systems: from spheres to fractals. Powder Technology 87, 29–41. Simons, S.J.R., Seville, J.P.K., Adams, M.J., 1994. An analysis of the rupture energy of pendular liquid bridges. Chemical Engineering Science 49 (14), 2331–2339. Subero, J., Ghadiri, M., 2001. Breakage patterns of agglomerates. Powder Technology 120, 232–243. Tan, H.S., Salman, A.D., Hounslow, M.J., 2005. Kinetics of fluidised bed melt granulation V: simultaneous modelling of aggregation and breakage. Chemical Engineering Science 60, 3847–3866. Tardos, G.I., Khan, M.I., Mort, P.R., 1997. Critical parameters and limiting conditions in binder granulation of fine powder. Powder Technology 94, 245–258. Weber, S., Briens, C., Berruti, F., Chan, E., Gray, M., 2006. Agglomerate stability in fluidized beds of glass beads and silica sand. Powder Technology 165, 115–127. Williams, R., Nosker, R.W., 1988. Strong adhesion of dust particles. In: Mittal, K.L. (Ed.), Particles on Surfaces, vol. 1. Plenum Press, New York, p. 193. Xu, C., Zhu, J., 2005. Experimental and theoretical study on the agglomeration arising from fluidization of cohesive particles—effects of mechanical vibration. Chemical Engineering Science 60, 6529–6541.