The use of bulk density measurements as flowability indicators E.C. Abdullah ) , D. Geldart Department of Chemical Engineering, Uniersity of Bradford, Bradford, West Yorkshire BD7 1DP, UK Received 31 March 1998; accepted 28 August 1998 Abstract Two equipments for measuring the bulk density of powders have been used to assess the flowability of porous and non-porous powders in which particle size changes as a result of controlling the percentage of fine components in the mixtures. The Hosokawa Powder Tester, which measures the aerated and tapped bulk densities, gives accurate measurements of bulk density because a fixed volume of powder is used and the mass of the powder can be measured accurately. On the other hand, the Copley Tap Density Volumeter, which uses a fixed mass of powder, gives inaccurate measurements due to the difficulty of reading the volume from the graduated cylinder. Generally, the flowability of powder increases with the increase of particle size, and there appears to be a critical size range above which flowability does not show any improvement. The packing of particles in two conditions, i.e., aerated and tapped, is examined and compared with the ideal packing of binary mixtures spherical particles. q1999 Elsevier Science S.A. All rights reserved. Keywords: Bulk density; Flowability; Critical size range 1. Introduction The bulk density of a powder is its mass divided by the volume occupied by the powder. It is also referred to as the mass of powder that can be packed into a specific volume. The volume includes the spaces between particles as well as the envelope volumes of the particles them- selves. The bulk density of a powder changes tremen- dously depending on the way particles are packed, like compaction, consolidation, etc. Therefore there is no unique value for a given powder. There are two basic methods of evaluating the bulk density of a powder that provide a guide to its flow characteristics. . . 1 The aerated bulk density random loose packing of a powder is determined by allowing the dispersed powder to settle in a container under the influence of gravity. A powder with a strong structural strength will resist collapse when dispersed in a container and will have a low bulk density, while a structurally weak powder will collapse easily and have a high bulk density. High friction between the particles results in a low apparent bulk density. With decreasing friction, the bulk density increases. Svarovsky ) Corresponding author. Department of Chemical Engineering, Univer- sity of Malaya, 50603 Kuala Lumpur, Malaysia. Tel.: q60-3-7595206; Fax: q60-3-7595319 w x 1 defined aerated density as density measured when the particles are separated from each other by a film of air, and that, on the whole, they are not in direct contact with each other. However, most workers interpret it as the bulk density after the powder has been aerated and allowed to settle gently, which means the most loosely packed bulk density. . . 2 The tapped bulk density random dense packing is obtained by tapping the container holding the aerated sample. The structure of a cohesive powder will collapse significantly on tapping while the weak or free-flowing powder has little scope for further consolidation. The powder particles are forced to jump and to lose contact with each other for a moment while tapping. When there is reduced friction between the particles, the particles rear- range and thus tapping results in improved packing condi- tions. The ratio of the tapped bulk density to aerated bulk density is called the Hausner ratio, a term coined by Grey w x and Beddow 2 , which is a useful measure of cohesion. w x Dutta and Dullea 3 regarded a drop in the Hausner ratio as a decrease in cohesiveness of the powder. Generally, there are two types of packing, ordered and random. Ordered packing is packing where the particles occupy specific sites with respect to each other such that long-range order exists as in a crystal structure. On the other hand, random packing is packing of particles without 0032-5910r99r$ - see front matter q 1999 Elsevier Science S.A. All rights reserved. . PII: S0032- 5910 98 00208- 3 ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 152 order and can be subdivided into two types, based on the procedure used in assembling the packing, namely, ran- dom dense packing and random loose packing. Random dense packing is defined as a random packing where the particles have been agitated to attain the closest packing possible without introducing long-range order or deforma- tion, and is equivalent to the tapped bulk density. Random loose packing is defined as the packing of particles poured into a container without agitation or vibration, representing the lowest stable packing density without introducing long-range order or deformation, and is equivalent to the w x aerated bulk density 4 . 2. Literature survey The term Hausner ratio was taken from the work of w x Hausner 5 , who measured the aerated and tapped density of three copper powders of similar particle size distribution but having different shapes, i.e., spherical, irregular and flake. He found that the further the particle shape deviated from spherical the higher the ratio of the tapped to the aerated density for the bulk powder tested. Further tests with tungsten powders showed that this ratio decreased with increasing particle size. Friction conditions between powder particles can be altered by changes with respect to the particle surfaces, by additions to the powder, such as lubricants, and by the medium which surrounds the parti- cles. He also performed tests with two types of electrolytic . iron powder, i.e., one with an oxide film as is and the . other with clean surfaces as reduced . The results con- firmed that the powder with an oxide film had a higher aerated density, which indicates lower friction than is the case for powder with clean surfaces. The Hausner ratio is now widely used in industry as a w x characterisation property. Harnby et al. 6 mentioned that the significance of bulk density lies not so much with its absolute value, but more with the relative changes in bulk density which occur in response to slight changes in the characteristics of bulk constituent particles within the pro- cess. Relative changes in bulk density can be very sensi- tive indicators of changes in the structural strength of a loosely compacted powder and hence of its flow character- istics in many process operations. w x Yu and Hall 7 used the funnel method as was also w x described in Ref. 1 to measure the aerated density, in which the powder sample was slowly poured into a stan- dard funnel with an opening of 5 mm in diameter, through which the powder flowed and filled a cylindrical container of height 60 mm and diameter 30 mm. The tapped density was measured by manual vertical tapping until there was no change in the height observed. They concluded that the Hausner ratio is a useful parameter reflecting the particle particle friction. They have also applied the Heckel equa- w x w x tion 8 and Kawakita equation 9 to predict the relation- ship of packing density and the number of taps. The Heckel equation and the simplified Kawakita equation are . . shown in Eqs. 1 and 2 , respectively. n y . T r sr y r yr e 1 . . n ` ` BA r BA r s 2 . n nb r yr . ` BA 1y 1qnb r . ` w x Velasco Antequera et al. 10 measured the Hausner ratio by placing a 100 g sample in a 250 ml graduated . cylinder and determined the occupied volume V . After o 10 and 500 taps, the occupied volumes are determined, V 10 and V , respectively. They calculated the Hausner ratio 500 w x. and compressibility index introduced by Carr 11 using the mean value of 10 measurements. In evaluating the packing characteristics of large non- cohesive mono-sized and non-spherical particles, Zou and w x Yu 12 found that the Hausner ratio decreased with the increase of sphericity, as given by HRs1.478=10 y0.136C , w x where C is the sphericity of the particle. Geldart et al. 13 found that the Hausner ratio enabled one to distinguish quantitatively between the free-flowing, easy-to-fluidise group A and the cohesive, difficult-to-fluidise group C powders. Essentially, it was found that cohesive group C powders exhibited a ratio larger than 1.4 while powders with a ratio less than 1.25 belonged to group A. Powders with Hausner ratios in the range of 1.25 to 1.4 are in the transitional group AC and have properties in common with both A and C type powders. It must be stressed that Hausner ratio reflects the friction conditions in a moing w x mass of powder rather than in a static situation 2 . w x Rastogi et al. 14 in studying dry ultrafine coals having mean diameters of 7 to 25 mm, used Hausner ratios to measure their cohesiveness. It was found that even with over 1000 taps this ratio did not reach a steady state indicating that further packing could be achieved. Using Geldarts criteria of Hausner ratio as mentioned above, they found that the Hausner ratio is consistent with other parameters measured, such as floodability, deaeration in fluidised bed and shear stress by the Jenike shear cell. w x Geldart et al. 13 , using irregular alumina, porous cracking catalysts and fillites, and iron oxide, found that the Hausner ratio was virtually independent of the relative humidity of the air surrounding the particles even though both the aerated and tapped bulk density varied as a function of the relative humidity. However, Harnby et al. w x 6 found the contrary. When working with different size fractions of glass ballotini they observed a marked increase in the Hausner ratio when a relative humidity of approxi- mately 60% was exceeded for the 4553, 5363 and 6375 mm size fraction. This could be due to the non-por- ous and smooth surface nature of ballotini where humidity has no place to hide. w x Orband and Geldart 15 added antistatic agent, Laro- stat w 519, to three samples of pure terephthalic acid which were known to have flow problems. With this antistatic ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 153 Fig. 1. The spherical shape of FCC particles. agent the Hausner ratio of the most difficult powder was lowered significantly from 1.66 to 1.27. 3. Materials used Two types of powders were used in this work, in which . porous Fluid Cracking Catalyst FCC and non-porous Fire . Retardant Filler FRF grade powders mixtures were tested. In each powder, a component of fine particles was added to the coarse particles in weight percentages of 1, 2, 5, 10, 20, 40, 50, 60, 70, 80, 90 and 100%. 3.1. Fluid cracking catalyst As the objective of this work was to study the effect of particle size on the flowability of powder, the spherical . FCC see Fig. 1 used, which originally had a Sauter diameter of 43.5 mm, was cut into two portions, i.e., fines and coarse. The fines portion should, ideally, contain particles with sizes less than 45 mm and the coarse portion should be made up of particles larger than 45 mm. The separation of this powder was performed in Hosokawa Micron in Runcorn, Cheshire, on a Tumbler Screening Machine ATS 600 with bouncing balls, which hit the screen of a 61 mm sieve. The coarse powder from this separation possessed all the flow properties of a free-flowing powder, but the fines
portions flow properties i.e., the Hausner ratios1.16
. and the angle of reposes29.28 differed little from the coarse, so that it was not suitable to be used as cohesive powder. A total of 30 kg of FCC was then sent to Hosokawa Micron for grinding and this gave a fines component with a mean Sauter diameter of about 7 mm which behaved cohesively. The flow properties of these two components of FCC is shown in Table 1. 3.2. Aluminium trihydroxide fire retardant fillers . Aluminium trihydroxide Fire Retardant Filler FRF is a white crystalline powder with an angular particle shape Table 1 Fines and coarse components of FCC powders Fines FCC Coarse FCC . Particle size, d mm 6.96 78.85 sv Hausner ratio 1.78 1.12 3 . Aerated density kgrm 448 763 3 . Tapped density kgrm 797 853 a . Angle of repose degree 43.9 27.3 Particle density 1242 . measured by gas flow method 3 . r kgrm p a Angle of repose above 408 is considered cohesive. . see Fig. 2 and can be represented by the chemical . formula Al OH , alternatively written as Al O P3H O. 3 2 3 2 The non-porous FRFs are general purpose flame retardant additives for use in a wide range of products based on polyester, epoxy and acrylic resins, PVC, cellulosic materi- w x als and many elastomeric compounds 16 . FRF grade powders have typical applications in glass reinforced plas- tics and encapsulation resins. They are also used in exte- rior cladding and tiles, synthetic marble, carpet and fabric backing, conveyor belting, cables, furniture foam, PVC and rubber goods, and adhesive, coatings and sealants. FRF 5 and FRF 40 were chosen to be the coarse and fine non-porous powders, respectively, in this study. The Sauter diameters of these two powders are 63.1 and 6.6 mm, respectively, as measured by the dry method in the Malvern Autosizer 2600C. As with FCC, the other proper- ties of FRF 5 and FRF 40 were also tested before the fine and coarse mixtures were made, to make sure that both did behave as free-flowing and cohesive powders, respec- tively. The results are shown in Table 2. 4. Experimental rig and procedure 4.1. Hosokawa Powder Tester w x Geldart and Wong 17 regarded the Powder Character- istics Tester, type PT-D, designed by Hosokawa Mi- Fig. 2. The angular shape of FRF particles. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 154 Table 2 FRF 40 and FRF 5 Powders FRF 40 FRF 5 . Particle size, d mm 6.6 63.1 sv Hausner ratio 1.86 1.23 3 . Aerated density kgrm 654 1201 3 . Tapped density kgrm 1217 1480 a . Angle of repose degree 54.4 29.1 Particle density 2449 . measured by gas flow method 3 . r kgrm p Helium Pycnometer 2447 2425 absolute density, 3 . r kgrm abs a Angle of repose above 408 is considered cohesive. . cromeritics Laboratory Japan , as a very useful apparatus for reproducibly measuring the physical characteristics of powders. The incorporation of automation has eliminated much of the human error typically present in such experi- ments. The aerated bulk density is measured by pouring pow- der through a vibrating sieve into a cup. Fig. 3 shows the equipment used, which is the assembly of a screen cover, screen, a spacer ring, a chute attached to a mains-operated vibrator of variable amplitude and a stationary chute aligned w x with the centre of a pre-weighed 100 ml cup 1 . A suitable sieve, through which most of the material would pass readily, has to be chosen. The combination of vibration amplitude together with sieve aperture size should be such that the time taken for the powder to fill the 100 cm 3 cup is not less than 30 s. The excess powder is scraped from the top of the cup using a ruler, without disturbing or compacting the loosely settled powder. . For the tapped bulk density Fig. 3b , a cup extension piece is fitted to the cup holding the aerated powder and extra powder added into it, and then this whole setup is subjected to tapping. A suitable number of taps needed to be found. By performing a few tapped density measure- ments at different number of taps, the number of taps required for tapped bulk density to reach its maximum could be determined. However, for this work a standard- ised number of 180 taps over a period of 6 min, was used in all cases. Care had to be taken during the tapping process that the level of powder did not fall below the top edge of the cup. When the tapping process finished, the extension piece was removed and the powder then scraped level with the top edge of the cup and the contents of the cup weighed to obtain the tapped density. These experi- ments were carried out at ambient conditions. Five runs were conducted on each sample for aerated density and tapped density. This leads to five independent estimates for the Hausner ratio for each sample. The average values were taken to be the aerated and tapped bulk density, respectively. 4.2. Copley Tap Density Volumeter Another tapping apparatus known as the Copley Tap Density Volumeter was also used in order to see the effect of a different tapping strength on the interparticle forces within powders. The Copley volumeter consists of a gradu- ated cylinder and tapping mechanism as shown in Fig. 4. Fig. 3. The Hosokawa Powder Tester for measuring the aerated and tapped bulk densities. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 155 Fig. 4. Copley Tap Density Volumeter. The advantage of the Copley over the Hosokawa Tester is that a fixed mass of powder is used, therefore there is no need to remove the container for weighing as the number of taps is progressively increased. The tapping action is provided by a rotating cam driven by a 60 W motor with an output speed of 250 rpm. The cam raises the cylinder platform and hence the cylinder itself through a distance of . 3 mm compared with 25 mm in the Hosokawa and then drops again, therefore producing the tapping motion re- quired. In this equipment, no specific method to measure the aerated density is recommended by the manufacturer, so the filling method used in the Hosokawa machine was used. About 100 ml of powder was first poured into the graduated cylinder through the sieve from the Hosokawa Powder Tester. Care has to be taken to switch off the vibrating sieve when about 100 ml of powder have been put into the volumeter. Normally, the powder does not form a horizontal surface in the cylinder, therefore the aerated density cannot be measured accurately, unlike the Hosokawa Powder Tester, in which accurate volume and powder weights are obtained when the powder is scraped level before weighing. Later, this cylinder together with the powder was put on the Copley volumeter with a rubber sleeve around its base and subjected to the number of taps desired. The tapped density is obtained by reading the volume of the powder from the scale, but this is difficult since the powder surface is seldom horizontal. In order to overcome the problem of inaccurate volume w x 3 reading Yusup 18 suggested using the 100 cm cup from the Hosokawa tester on the base of the Copley. As in the Hosokawa test, powder snowstorms into the 100 cm 3 cup and after obtaining its aerated density, an extension piece is placed on the cup and more powder is then added. The cup with the extension piece together with the aerated powder in it are then put on the Copley volumeter for tapping. When the desired number of taps is reached, the extension piece is then removed, the powder scraped level, and the contents weighed to obtain the tapped density. For the subsequent tapping, the extension piece is again placed on the cup and more powder is added. This differs from the use of the glass cylinder in which all powder experi- ences same number of taps, whereas in the new technique the recently added powder is only tapped for the subse- quent number of taps, as done in the Hosokawa tester. Using this modified technique, the tapped density can be compared more meaningfully to the Hosokawa tapped density as the error in reading the volume has been suffi- ciently eliminated. For convenience, the three methods used in tapping have been named as: Method 1: Hosokawa Powder Testers tapping as out- lined in Section 4.1. Method 2: Copley volumeters tapping where the gradu- ated cylinder was used. This is named as Copley filling and tapping in Figs. 1012. Method 3: Hosokawas 100 cm 3 cup is filled by the Hosokawa method and subjected to tapping in the Cop- ley volumeter. This is called Hosokawa filling and Copley tapping in Figs. 1012. 5. Results and discussion It should be mentioned that all the results discussed in Sections 5.1 and 5.2 were obtained using the Hosokawa ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 156 Fig. 5. Aerated density vs. mean particle size. Powder Tester while in Section 5.3 both the Hosokawa Powder Tester and the Copley Volumeter were used. 5.1. Hausner ratio, aerated and tapped densities Figs. 5 and 6 show the aerated and tapped densities of FRF and FCC mixtures. The densities of FRF grade mix- tures are always higher than the FCC mixtures in both aerated and tapped densities. The increase in particle size initially increases the aerated and tapped densities of both FRF and FCC mixtures. As particle size increases, the cohesivity of powder is expected to decrease. A more free-flowing powder should have a higher aerated density, as the interparticle forces between particles become weaker, therefore powder should pack in a denser condition. How- ever, beyond the size of about 25 mm, further increase in particle size shows no significant increase in the aerated densities for both powder mixtures. In the case of tapped density, initial increases in densi- ties were followed by gradual decreases beyond a mean particle size of about 15 mm for both powder mixtures. However, the Hausner ratios of the two materials do not vary very much from each other as shown in Fig. 7, where the Hausner ratio decreases with the increase in particle size. This confirms that the increase of particle size of a powder mixture is always accompanied by the decrease in cohesivity. When powder mixtures are dominated by fine particles, the interparticle forces, in this case the van der Waals forces, should be dominant in the packing of particles when they snowstorm down from the sieve into the 100 cm 3 cup. Due to this, low aerated densities are registered when the particle size is small. The increase in particle size slowly diminishes the influence of the van der Waals force, thus causing the aerated density to increase. When the van der Waals force becomes small or insignificant, . further increases in particle size beyond 25 mm do not have much effect on the aerated density as shown in Fig. 5. 5.2. Duration of tapping The tapped densities of both FRF and FCC mixtures drop as more fines components are added to them as shown in Fig. 6. This is particularly obvious with Sauter Fig. 6. Tapped density vs. mean particle size. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 157 Fig. 7. Hausner ration vs. Sauter diameter using tapped density measured after 180 taps. diameter less than 15 mm for both mixtures. As mentioned above, the interparticle forces are dominant due to the presence of more fine particles. By using mechanical force, i.e., tapping, it is expected that the density would eventually reach a constant value as most of the mixtures beyond mean size of 15 mm shown in Fig. 6. In order to investigate the reason for the low tapped density observed for mixtures less than 15 mm, three FRF powder mixtures with large percentage of fines, i.e., 70, 80 and 90% of FRF 40, were chosen for further tapping beyond the 180 taps used in the Hausner ratio test. These three powder mixtures were subjected to a pro- gressive increase in the number of taps up to a total of 1150 taps and their tapped densities resulting from these high numbers of taps are as shown in Fig. 8. It should be emphasised that in the process of increasing to higher numbers of taps the same powder in the container is used but with the more powder added. As experienced by w x Rastogi et al. 14 , after more than a thousand taps, all the three mixtures still show no sign of attaining their respec- tive constant densities. The tapped density increments due to tapping from 180 to 1150 taps are rather small, i.e., 6.0% for 70% FRF 40 mixture, 5.8% for 80% FRF 40 mixture and 7.3% for 90% FRF 40 mixture. This result indicates that although a high number of taps was used, i.e., the same repeated magnitude of mechanical force applied, it is still not enough to overcome the interparticle forces in the mixtures. The effect on the HR of increasing from 180 to 1150 taps is shown in Fig. 9. The HR increases by only about 5.8% for all the 70%, 80% and 90% FRF 40 mixtures in response to the increase of 970 taps for Method 1. The small increment in tapped density and HR observed in a way justifies the use of a standard- . ised number 180 of taps used in all other measurements. 5.3. Comparison of Hosokawa with Copley tapping In order to see if different tapping strengths could pack
the cohesive powder i.e., mixtures with more fine parti-
. cles better, the Copley Tap Density Volumeter was used Fig. 8. Tapped density vs. number of tapsMethod 1. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 158 Fig. 9. Hausner ratio vs. Sauter diameter using tapped density measured after 180 and 1150 taps for 70, 80 and 90% FRF 40. . see Section 4.2 . This drops the powder only 3 mm compared with 25 mm in the Hosokawa equipment. Figs. 1012 show that the Hosokawa tapping gives significantly higher densities than the other methods. In both Methods 2 and 3, after 1150 taps the tapped densities are still smaller than those of the 180 taps of Hosokawa tester for all the three mixtures. This is almost certainly because of the smaller dropping distance of the Copley
volumeter. In Methods 2 and 3 using the Copley volume-
. ter , a constant value of the tapped density was reached after 640, 740 and 760 taps for 70, 80 and 90% of FRF 40 mixture, respectively; 560, 540 and 520 for 70, 80 and 90% of FRF 40 mixture, respectively. Unlike the Hosokawa . tester Method 1 the tapped density continued to increase slowly. Methods 1 and 3 differ only in the manner in which the powder is tapped. Both methods gave about the same aerated density before tapping. After the initial 180 taps, Method 1 registered tapped densities about 35% higher than the tapped densities measured by Method 3. These differences are reduced to about 19% at the end of 1150 taps. At this stage, tapped densities measured by Method 3 have already reached their respective constant values, but tapped densities measured in Method 1 are still slowly increasing, so the percentage difference could be higher if the mixtures continue to be tapped. It is clear that the Copley volumeter is unsuitable for assessing the flowabil- ity of powders because the tapping provided cannot over- come sufficiently the interparticle forces in cohesive pow- ders. Table 3 shows the aerated and tapped bulk densities as well as the Hausner ratio for the three mixtures tested, in the three different methods mentioned above. The tapped densities shown are those obtained after 180 taps for all the three mixtures. If the Hausner ratio of 1.4 is used as the criteria in distinguishing cohesive and non-cohesive powder, all the three mixtures were found to be cohesive by Method 1 as shown in Table 3. However, the Hausner ratios for all the three mixtures are less than 1.4 as measured by Methods 2 and 3. It was obvious from Fig. 10. Tapped densities for 70% FRF 40r30% FRF 5 obtained by three different methods. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 159 Fig. 11. Tapped densities for 80% FRF 40r20% FRF 5 obtained by three different methods. handling the three mixtures that they did possess strong cohesivity, but the use of the tapped densities from the . Copley volumeter Methods 2 and 3 indicated incorrectly that the mixtures were not cohesive. Another interesting point in Figs. 1012, is that there is always a crossover of tapped densities measured by Meth- ods 2 and 3. At low numbers of taps Method 3 registered higher tapped densities for all the three powder mixtures, however at higher number of taps Method 2 would pack the mixtures in a denser state. The crossover points are at 400, 580 and 600 number of taps for 70, 80 and 90% of FRF 40 mixture, respectively. There may be three effects at work here. In Method 3 because the steel 100 cm 3 cup is filled by the Hosokawa technique, the aerated density is already higher than in Method 2 which gives higher tapped . densities initially up to about 500 taps . Thereafter Method 2 leads to higher tapped densities because of the following. . a The different diameters of the container used, where 3 . the 100 cm cup IDs5 cm is larger than the graduated . cylinder IDs3.5 cm . It is relatively easier to fill the bigger diameter cup than the smaller graduated cylinder. These are shown in the higher aerated densities recorded in Method 3 compared to Method 2, where 10.0, 11.6 and 14.8% higher aerated densities are recorded for 70, 80 and 90% of FRF 40 mixture. As the mechanical force was applied, the smaller diameter cylinder experienced a bigger force per unit area so making the tapping energy more concentrated per surface area. Although Method 3 started with denser packing before tapping, as the number of taps is increased the smaller diameter cylinder received a larger cumulative force and eventually produced a denser pack- ing. The lower number of taps in Method 3 required to achieve constant densities for all the three mixtures also . implies that the lower force bigger diameter although . applied repeatedly higher number of taps could not over- come the interparticle forces better than in the smaller diameter cylinder Method 2. . b The Copley cylinder is glass which gives lower powderrwall friction than the large steel container and allows greater compaction. Fig. 12. Tapped densities for 90% FRF 40r0% FRF 5 obtained by three different methods. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 160 Table 3 Bulk densities measured by three different methods after 180 taps Method Mixture Aerated Tapped Hausner density density ratio 3 3 . . kgrm kgrm 70% FRF 40 808.1 1401.4 1.73 1 80% FRF 40 751.6 1338.0 1.78 90% FRF 40 691.9 1268.1 1.83 70% FRF 40 734.9 1006.7 1.37 2 80% FRF 40 673.6 910.3 1.35 90% FRF 40 602.7 861.0 1.43 70% FRF 40 800.4 1034.3 1.29 3 80% FRF 40 747.7 991.3 1.33 90% FRF 40 696.1 949.0 1.36 . c In Method 2, the entire powder has been subjected to the total number of taps whereas in Method 3, a significant amount of powder has far fewer taps. From this comparison of the results of these three methods, the Copley volumeter is considered to be unsuit- able for measuring the cohesivity of powder, due largely to the weaker tapping strength built in to the apparatus. Besides that, the other shortcomings of the Copley include the absence of a standard filling method for the measure- ment of aerated density and the inaccurate reading of the volume of powder before and after tapping. However, the Copley volumeter has the advantage of being mobile due its smaller size and weight. The Hosokawa Powder tester is a better apparatus as it has a good filling method, reliable measurement and good built-in tapping mecha- nism. 5.4. Effect of fewer fine particles in mixture In Fig. 6, the increase in mean particle size has a great influence on the arrangement of particles as they try to form the densest packing. It should be remembered that a large mean particle size corresponds to a small percentage of the fine components. In a segregating system, when powder is given mobility by tapping, the large and small particles are able to rearrange to form the densest packing. More large particles will provide more unoccupied voids as most of the small particles would migrate to the bottom of the cup to fill those voids in the lower region. This would leave most of the voids in the middle and upper region of the cup unoccupied. . In order to ascertain if tapping mechanical force does give the freedom of movement to the fine particles to reach to the bottom of the cup through the primary struc- ture formed by the large particle, the tapped powders compositions were checked. In this investigation, an FRF mixture with 2% FRF 40 and 98% FRF 5 was used, as the small percentage of fines available should make the segre- gation phenomenon mentioned above easily observable if it occurs. After the five measurements of aerated and Table 4 Composition of 2% FRF 40 and 98% FRF 5 mixture after tapping . . . Region Fines g Coarse g Total wt. g % Fines Upper 6.11 24.64 30.75 19.9 Middle 6.08 25.32 31.40 19.4 Lower 5.97 25.38 31.35 19.0 tapped densities, the last tapped powder in the 100 cm 3 cup was divided into three regions, i.e., upper, middle and lower, and about 30 g of samples were extracted from each region. The compositions of fines in each region were checked by separating fines from coarse using a 45 mm sieve. The results are as shown in Table 4. A consistent 19% of fines -45 mm is observed in the three regions, which firmly proves that it is not a segregating system. Therefore, the reason for the lower tapped density observed for mixtures with smaller percentage of fine particles is simply that there are not enough fine particles to fill the voids in between the large particles. It is believed that the fine particles in these mixtures coated the larger particles, and the repeated tapping simply could not dislodge the fine particles from the larger particles. The primary structure remains and there is no increase in tapped density as no rearrangement of particles occurs. Another FRF mixture with higher percentage of fine particles, i.e., 20% FRF 40 and 80% FRF 5 mixture, was also tested in the same manner, to check if the movement of fine particles could be observed more clearly. The results are shown in Table 5, where it is proven again that a consistent percentage of fine particles is observed throughout the three defined regions. However, no test was carried out on the FCC mixtures, as there were only limited amounts of fines and coarse components of FCC available, and at the time this test was performed most of the two components of FCC powders were used up for fluidisation tests. Nevertheless, we are quite sure that the same arguments used for explaining the lower tapped density observed in FRF mixtures are also valid for the FCC mixtures, as it has been proven convinc- ingly in another test, using the Warren Spring-Bradford Cohesion Tester, that the FCC mixtures are non-segregat- ing systems. 5.5. Packing efficiency Theoretically, it is possible to predict the maximum density a packing could achieve. It is assumed that only Table 5 Composition of 20% FRF 40 and 80% FRF 5 mixture after tapping . . . Region Fine g Coarse g Total wt. g % Fines Upper 11.85 24.78 36.63 32.4 Middle 11.46 24.18 35.64 32.2 Lower 10.78 23.73 34.51 31.2 ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 161 Fig. 13. Voidage of aerated and tapped FRF mixtures. the coarse particles, having a bulk density equal to their tapped density, are in a container, and form a primary structure. It is further assumed that when the fine particles are mixed in they occupy the spaces between the large particles at the same tapped density that they would have if they were in a container on their own. If the initial volume fraction of coarse and fine components, i.e., initial frac-
tional density of coarse and fine are, f and f 1y and
c f c . 1y , and the particle density of the coarse and fine f components, r and r , are known, the maximum packing c f density can then be written as: PD sf r q 1yf f r 3 . . max c c c f f where PD is the maximum packing density. The experi- max mental values of f and f are 0.61 and 0.5 for the coarse c f . and fine FRF powders see Fig. 13 . The particle density of fine and coarse components of FRF powder can be . found in Table 2. Inserting these values into Eq. 3 leads to a value of PD of 1956 kgrm 3 compared with the max 3 . experimental value of 1650 kgrm see Fig. 6 , i.e., the calculated value is 18.5% higher. The experimental values of f and f for FCC are 0.60 c f . and 0.56 Fig. 14 , respectively. The gas flow technique, as outlined in the Appendix A, was used to measure the particle density for the porous FCC, but because of effects such as channelling this method could not be used to measure the particle density of the fines component of FCC and there is no other known method to do this. Since it was impossible to measure the particle density of the fine FCC, and because it is known that at maximum . packing density Fig. 16 there is always more coarse material than fines, the particle density of FCC fines was taken to be the same as that of the coarse FCC, i.e., of 3 . 1242 kgrm . By substituting these values into Eq. 3 , the calculated PD of FCC is 1023 kgrm 3 , which is very max close to the experimental value of 1040 kgrm 3 . The reason for the closer match between the calculated and the experimental value of PD for FCC compared to max that for FRF powder could be due to the shape difference Fig. 14. Voidage of aerated and tapped FCC mixtures. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 162 Fig. 15. Density vs. percentage of FRF 40 in mixture. of these two powders. The fines component of the FCC particles could roll in between the larger spherical particles . see Fig. 1 thereby forming a dense packing which almost equals the theoretical estimation. However, the more angu- . lar fine FRF particles see Fig. 2 might not be able to roll and find their way as easily as the FCC, therefore leaving more voids in between the larger particles, and contribut- ing to an experimental PD lower than the theoretical max estimation. 5.6. Voidage profile Figs. 13 and 14 show the voidage of aerated and tapped powders for both mixtures. The voidage of aerated powder initially decreases with the increase in particle size and later stays virtually constant. However, the voidage for the tapped powder initially decreases, and after reaching its minimum values, increases again which is what was men- tioned above as there are not enough small particles to fill the voids in between the larger particles. These effects are observed in both powder mixtures. In order to calculate bed voidages, it is essential to . know the envelope or apparent density of the particles. This is also known as the particle density and is defined as the mass of the particle divided by its volume, where this density is different from absolute or skeletal density if the particle is porous, since the volume in question contains all w x pores, open or closed. It has been shown by Xie 19 that coarse FCC particles have a lower particle density than the fines. As we were unable to measure r for the fines in p order to calculate the r for the mixtures, r for the as p p . received FCC which contained a wide range of sizes . was measured by the gas flow technique see Appendix A and found to be 1418 kgrm 3 . This value has been used to calculate the voidage shown in Fig. 14. For the non-porous FRF mixtures, r was obtained using the helium pyc- p nometer. 5.7. Critical size w x w x w x Grey and Beddow 2 , Wong 20 and Orband 21 observed that Hausner ratios increase with the decrease of Fig. 16. Density vs. percentage of fines FCC in mixture. ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 163 w x particle size. Orband 21 discovered that there is a critical size above which Hausner ratio is virtually constant, which means to say that powders do not possess any cohesivity; it is only below this critical size that the Hausner ratio increases with the decrease of particle size. However, from the data of FRF and FCC mixtures as shown in Fig. 7, a critical size for either mixture cannot be categorically determined. Instead the Hausner ratios for both mixtures are proportional to 1rd 0.2 . If a critical size exists, then it is
believed that it should fall in range of 1.25-1.4 range
. represented by two dotted lines in Fig. 7 , which is also known as Geldarts group AC powder. The corresponding particle size in this region for FCC is 25-d -42 mm, sv where below this size range Hausner ratios increase with the decrease of particle size; and above this size range interparticle forces, especially the van der Waals forces, become less influential in the packing of powder. For the FRF mixture, the critical size is in the range of 30-d - sv 54 mm. 5.8. The packing of powder Although the particles used in this work do not have the same shape and are made up of the two different mean sizes mentioned earlier, they somehow do show a similar pattern of density change to each other with the increase in percentage of smaller particles. Figs. 15 and 16 show that . the tapped densities random dense packing of both pow- ders change in the same manner as in Fig. 17, which is the basic behaviour of random dense packing of binary mix- w x tures of spherical particles sketched by Furnas 22 . The tapped density curves for both powders also con- firmed that maximum densities for mixtures are achieved with more large particles in the mixture, as stated by w x Furnas 22 . For the FCC mixtures the maximum tapped density is achieved with about 38% of fines in the mixture. This matches closely the 35% of fines in mixtures pre- Fig. 17. The packing density variation with composition for a bimodal mixture. The maximum density for a homogeneous mixture f ) occurs at a composition X ) . Two-dimensional structures are shown for five possi- w x. ble compositions after German 25 . Fig. 18. The porosity vs. composition for bimodal mixtures of varying particles size ratios, assuming an inherent fractional packing density of w x. 0.5 after Furnas 22 . w x dicted by Furnas 23 for particle size ratios about 10:1 as shown in Fig. 18. This could mean that the many models developed for the prediction of packing of binary mixtures could also be applied to powders with distribution of sizes. The fines percentage which gives maximum tapped density . in FRF mixtures is less clearly seen Fig. 15 but is in the range of 2040%. 5.9. Effect of particle size ratio and shape The particle size ratio based on the Sauter mean diame- ters for FRF and FCC mixtures are 9.6:1 and 11.3:1, respectively. The minimum voidage of the tapped FCC mixtures is 0.27 at a mean particle size of 15 mm, and for the FRF mixtures is 0.32 at a mean particle size of about 23 mm. Despite the relatively small difference in the particle size ratio in this two mixtures, the bigger size ratio mixture still proved to have packed better, which is in line w x with that shown in Fig. 18 as stated by Furnas 22 . The more significant difference between these two pow- ders is their shape. The FCC particles are more spherical . Fig. 1 whereas the FRF particles are more angular Fig. . 2 . The reason why FCC mixtures packed better than FRF mixtures is influenced more by the shape of the particle rather than the small particle size ratio difference. This w x result agreed with that found by Yerazunis et al. 24 where the spherical glass mixtures they used showed better packing in comparison with the non-spherical glass mix- . tures crushed which have about the same particle size ratio. 6. Conclusions Although the FRF grade and FCC mixtures are not perfectly spherical particles, they both exhibit density pat- terns similar to those of an assembly of bimodal spherical particles. The maximum packing density always occurs when there is a smaller mass fraction of fine particles in a mixture compared to larger particles. The maximum pack- ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 164 ing densities for the FRF and FCC mixtures are observed at 2040% FRF 40 and about 38% fine FCC, respectively. These values are similar to those mentioned by Furnas w x 23 . The particle shape does play a role in making denser packing. The more spherical shaped FCC mixtures packed better than the more angular shaped FRF mixtures. It is well known that the cohesivity of powder decreases with the increase in particle size, and this was confirmed for both powder mixtures by means of the Hausner ratio. A critical mean size above which the increase in mean particle size would see no significant effect on the cohesiv- ity of the powder cannot be categorically seen from the data of the FRF and FCC mixtures. However, below a value of HR less than 1.25, interparticle forces become less significant with the increase of particle size; above the . HR range for group AC powders 1.251.4 , cohesivity increases rapidly with the decrease of particle size. There is evidence that in both FRF and FCC mixtures, when fine components are added to the coarse particles they adhere strongly to the large particles, preventing percolation of fines to the bottom of the mixtures. The Hosokawa Powder Tester when used to charac- terise flowability gives more reliable measurements than the Copley Volumeter, as accurate powder weights can be measured. 7. Symbols and abbreviations .. b a constant Eq. 2 d Sauter diameter sv f , f fractional density of coarse or fine particle c f fraction FCC Fluid Cracking Catalyst FRF Fire Retardant Fillers HR Hausner ratio K constant of CarmanKozeny equation L, L , L bed height, at aerated state and tapped state, A T n number of taps PD the maximum packing density max S , S slope of plot DP vs. U for aerated and tapped A T .. T compaction constant Eq. 1 U superficial velocity or empty tube velocity S r A BT . y s P S r T BA c sphericity of particle r , r bulk aerated or tapped densities BA BT r tapped density after n taps n r particle density p r maximum tapped density ` r , r particle density of coarse or fine particles c f D p pressure drop m viscosity of gas used in fluidisation , , voidage in aerated state and tapped state, A T respectively Appendix A An easy way to measure particle density of a porous powder is the gas flow technique, which is based on the Erguns equation. This technique is suitable for powders that have a significant difference between their aerated and tapped densities. As this method depends greatly on mea- suring the pressure drop across a fixed bed, it is not suitable for cohesive powders in which channelling is likely to occur. The primary task of this technique is to measure the pressure drop and bulk density of the bed at two voidages, i.e., the aerated and tapped beds. In the aerated state, the bed should be in its loosest form where it attains maximum voidage. The method to achieve this is by first fluidising the powder to mix it well and then let it settle by itself. The bed height at the aerated state is designated as L , and A pressure drop, D p, is measured for at least four superficial gas velocities. After that the bed is tapped to allow the powder bed to achieve a much lower voidage. The new bed height in the tapped state is designated as L , and T again D p is measured. The plot of DP against the superfi- cial velocity for both the aerated and tapped bed give two slopes S and S , respectively. By referring to the Ergun A T equation, the first term on the right-hand side of that equation which is called the CarmanKozeny equation can be written as: 2 Dp K 1y mU . s P . A1 . 3 2 L d sv . By rearranging the Eq. A1 and putting subscript A for the aerated experiment, the slope of the graph can be written as: 2 DP Km 1y . A S s s L . A2 . A A 2 3 / U d A sv A Similarly, the slope for tapped bed can be written as: 2 DP Km 1y . T S s s L . A3 . T T 2 3 / U d T sv T It is known that L r A BT s A4 . L r T BA and r BA 1y s A5 . A r p r BT 1y s A6 . T r p ( ) E.C. Abdullah, D. Geldart rPowder Technology 102 1999 151165 165 where r and r are the bulk densities for aerated and BA BT . . tapped beds, respectively. Dividing Eq. A5 by Eq. A6 and substituting for L , L , and A T A T S r r yr A BA p BT s . A7 . S r r yr T BT p BA S r A BT By assigning ys P A8 . / S r T BA particle density can be written as: r yy 1r3 Pr BT BA r s . A9 . p 1r3 1yy The most important point in using this method is to measure accurately the pressure drop across a fixed bed. References w x 1 L. Svarovsky, Powder Testing Guide, BMHB, Elsevier Applied Science Powder Testing Guide, BMHB, 1987, 146 pp. w x . 2 R.O. Grey, J.K. Beddow, Powder Technology 2 1968 323326. w x . . 3 A. Dutta, L.V. Dullea, AIChE Symposium Series 86 276 1990 2640. w x 4 R.M. German, Particle Packing Characteristics, Metal Powder Indus- . tries Federation 1989 443 pp. w x . 5 H.H. Hausner, International Journal of Powder Metallurgy 3 4 . 1967 713. w x 6 N. Harnby, A.E. Hawkins, D. Vandame, Chemical Engineering . . Science 42 4 1987 879888. w x . 7 A.B. Yu, J.S. Hall, Powder Technology 78 1994 247256. w x . 8 R.W. Heckel, Trans. AIME 221 1961 1001. w x . 9 K. Kawakita, K.-H. Lidde, Powder Technology 4 1970 6168. w x 10 M.V. Velasco Antequera, A.M. Ruiz, M.C.M. Perales, N.M. Munoz, M.R.J.C. Ballesteros, International Journal of Pharmaceutics 103 . 1994 155161. w x . 11 R.L. Carr, Chemical Engineering 1 1965 6972. w x . 12 R.P. Zou, A.B. Yu, Powder Technology 88 1996 7179. w x . 13 D. Geldart, N. Harnby, A.C. Wong, Powder Technology 37 1984 2537. w x 14 S. Rastogi, S.V. Dhodapkar, F. Cabrejos, M. Weintraub, G.E. . Klinzing, W.C. Yang, Powder Technology 74 1993 4759. w x 15 J.L.R. Orband, D. Geldart, Particle and Particle Systems Characteri- . zation 12 1995 204206. w x 16 Alcan Data Sheet 100, Baco Flame Retardant Fillers, Issue 3:2r93r100, Alcan Chemical Europe. w x . 17 D. Geldart, A.C.Y. Wong, Chemical Engineering Science 39 10 . 1984 14811488. w x . 18 S. Yusup, 1997 private communication . w x 19 H.Y. Xie, 1993, PhD Thesis, University of Bradford. w x 20 A.C.Y. Wong, 1983, PhD Thesis, University of Bradford. w x 21 J.L.R. Orband, 1994, PhD Thesis, University of Bradford. w x . . 22 C.C. Furnas, Industrial and Engineering Chemistry 23 9 1931 10521058. w x 23 C.C. Furnas, The Relations Between Specific Volume, Voids and Size Composition in Systems of Broken Solids of Mixed Sizes, Serial 2894, Report of Investigation, U.S. Bureau of Mines, North Central Experiment Station, Minneapolis, MN, October 1928. w x . 24 S. Yerazunis, J.W. Bartlett, A.H. Nissan, Nature 195 1962 3335. w x . 25 R.M. German, Advances in Powder Metallurgy 3 1992 115.