- Email: [email protected]
Powder segregation due to vibration
-
:
1.
_.~.
.P&der Technology. 16 < 1977) 51 - 57 @ Elsevier Sequoia S-A_. Lausanne - Printed in the Netherlands
Powder
COLIN
begation
51
Due to Vibration*
F. HARWOOD
IIT Research (Received
Institute.
10 West 35th Street.
Chicago, III. 606t6
(U.S.A.)
March 17.1976)
SUMMARY
rumble.
An experimental study on the segregation of particles when subjected to vibrational energy has been made. Radioactive tracer techniques were used to follow the movement of the powder_ The method did not disturb the bed and thus the segregation could be measured at various time intervals. It was found that the powder properties and the amplitude and frequency of the vibrations had a profound effect on the results obtained. In freeflowing powders, segregation was readily induced- Powder which demonstrated ‘ncipient cohesitin had pronounced segregation only at a critical vibrational energy. Truly cohesive powders exhibited very little segregation once the powder had packed down to its final bulk density.
INTRODUCTION
Powder beds are frequently sfibjecteci to vibrations_ The vibrations may be applied intentionally to consolidate a powder into a die or mold, or to aid the movement of a powder from a hopper or chute. Alternatively, the vibrations may occur inadvertently as a result of machinery rumble. In either case, the effect of i;he applied vibrations is a complex function of the amplitude and frequency of the vibrations and the properties of the powder bed. The objective of this study was to follow the movement of particles within a bed of similar particles which are typical of real @owder beds. The vibrations applied were within the range expected from heavy-machinery *Paper p&enti at the Secodd International Conference on the Compaction arid Consolidation of particulate Matter, Brighton, England. September 2 - 4. 1975.
As part
of the
study,
a radioactive
which ailowed the movement of the particles with time to be followed without disturbing the powder bed. Various powders were tested against a radioactive tracer sand and the relative segregation effects observed_ The materials were selected to be free flowing, incipiently cohesive, and cohesive. They had particle sizes greater and less than the marker sand and densities greater and less than the marker sand. Comparisons could then be made over a very wide range of properties. tracer
technique
was developed
BACKGROUND
Past studies on the segregation of powders in a vibrated bed have been extremely limited_ Brown [1] discussed, in general terms, the segregation of particles when they were vibrated and poured into heaps. Williams [2] was able to demonstrate the upward movement of a single large particle in a bed of free flowing powder when subjected to vibration. Oisen et al_ f3 - 51 studied the segregation of steel and glass spheres of 3/32 - l/4 in. diam. in a cylinder of 3/4 in. diam. and length 6 in. They were able to show that segregation proceeded to an equilibrium state and drew an analogy to fit-order reaction kinetics. They were able to make some observations on the effect of particle size and density of the spheres. The ratio of the particle size to vessel diameter was very low: in the worst case, only three spheres would fit into a cross-sectional planeWilliams and Shields [S] measured the segregation which took place as binary mixtures of fertilizer went down a channel. The degree of segregation was found to depend on the frequency and amplitude of the vibration as well as the direction in which it is applied.
52 The results were obtained using rounded free-
flowing granules of fertilizer of undisclosed chemical composition. More recently, Ahmad and Smalley [7] have made a detailed study on the rate at which a single 12,700 pm lead sphere rose through a bed of 500 - 600 .~rn sand particles. Accelerations of 1 - 10 g and frequencies of 50 - 150 cycles/set were applied to the system. They found that acceIeration was the most critical factor affecting the segregation. At constant frfzquency, segregation incxased with acceleration increase; however: with increasing frequencies, the s.®ation was reduced- It was also found that segregation increased with increase in size of the marker particle. An increase in density of the marker particle decreased the segregation, and particle shape had little effect. The above studies have used very large single particles to mark the progress of the segregation. In the present study, radioactive tracer techniques have made it possible to use many particles of a size comparable to the powder bed.
Fig_ 1. Schematic diagram of experimental setup for scanning the radioactive/non-radioactive powder SYS.tem
Vibmting table The vibratig
table used in this study is of the type used for testing metal parts for vibrational stress breakdown. It has an extremely wide range of frequency and amplitude positions and can easily simulate the vibrational values to be expected in industrial pkiIltS.
EXPERIMENTAL Procedure Apparatus
Powder samples were placed in a 250 ml graduated plastic cylinder of inside diameter 3-4 cm_ PIastic was preferred since it minimized the danger of the cylinder breaking and thus leaking the radioactive component. A metal plug was sealed into the base; the plug, which was 3 cm bigb, was there simply to raise the IeveI of the material in the cylinder so that the nucleonic probe could reach the side of the cylinder without obstruction from the cylinder base- It took no other part in the experiment. The Iengtt of the cylinder was scanned using the Nuclear Chicago DS-8 scintillation probe, which has been described more fully in a previous paper [S] _ In order to restrict the scanned volume, a collimator was fabricated by drilling a 0.75 cm diam. hole through a lead brick_ Because of the weight of the probe and lead brick, the cylinder and its contents were moved up and down by mounting it on a lab-jack. A diagram of the experimental arrangements is shown in Fig. 1.
Each of the materials were tested against the radioactive marker sand. The marker material was placed as a layer sandwiched between two layers of the test material. The procedure adopted was to place a 120 ml layer of material in the bottom of the cylinder, followed by a 20 ml layer of radioactive sand, and finally a top layer of 120 ml of test ma‘terial. The sandwich method of filling was chosen so that migration could be foIlowed in either an upward or downward direction. When the cylinder had been loaded in this way, it was then scanned with the probe to obtain the initial condition. The cylinder was then ti_ghtly corked and placed in a stout polyethylene ‘,ag to minimize the danger of a spill. It was placed on the vibrating table and vibrated at a known amplitude and frequency for a set time- After this time, it was then again checked with the probe and a new plot of count rate uersus vertical height obtained. This process was repeated at various time intervals up to a maximum of 60 minutes vibration time. A plot of count rate uersus
53
vertikl height was made.after each time in: terval. -_ _-- -._-I- -_ _ --_
Powder
.-
It was owed thk the h&k ofthe p9wi der bed was less after vibration. This was a& expect& reklt, &d-to allow for t&is, t+e data had to be normalized. Twelve scan points were t&en, all equidistant apart. The first-and the last scan points were always l-cm from the top and bottoti, respectively. Making the necessary assumption that an even packing had been induced, then the distance apart of the subsequent scan points ‘2a.s simply: (Height of column after vibrating -2)
materidg
In all, five powder materie
x
(Initial height of column -2)
have been used in the experiments. The tracer material was a free-flowing silica sand with a surface layer of iridium 192 baked bn the surface. Other mate. rials were two grades of silica sand, both free Bowing hut one larger in size than the tracer tid the other smaller in size; they had substantially the same density as the tracer sand. A pharmaceutical grade of lactose was used which was normally free flowing but became cohesive on compaction; it had a particle size much smaller than the tracer. Milled zircon was used which had a very small particle size and was cohesive in nature. The properties of these materials are given in Table 1.
Initial distance apart (cm)
TABLE1 Powderproperties
Property
Testpowders Tracersand
Sizedistribution.J.ml 420-600 300-420 210-300 150-210 105-150 75-105 50-75 30-50 10.30 5-10 3-5 <3 Density.g/cm3 *e,p Bu=, PB Tapped.
PT
Angleofrepose Drained Dynamic observedflow prOpertieS
Notmeasured. Quotedas 100%1805Ocun
Not
measured.
Assumesimilar tosand
Flintsand
2% 59 34 4 1
2-68 1.30 1.50
Banding sand
1% 3 33 37 20 20 6
2-68 1.30 1.50
Lactase
Milled ZircOll
10 17 58 12 3
-0.525 0.70 0.76
8 8 84
4.56 1.81 2.68
Notmeasured
32" 35"
34' 37"
26' 34"
46' 47"
Free
FLXS flowing
Free
Incipiently cchesive
Cohesive
flowing
flowing
It is suggested that in real powder systems where there are large numbers of particles of not too different size, the action of vibration is to induce a state of homogeneity into the powder bed by creating a mixing action. Another factor which can explain the results is the surface coating of iridium salts which impregnates the tracersand particles. While no layer is observkble under the microscope, it is well known that surface layers can lead to stray modifications of powder propertiesBanding
sand
The banding sand had a particle size closer to that of the tracer sand than did the flint sand. Vibration tests were conducted using identical conditions to those for the flint sand. The results obtained for zero time and after 30 min vibration at. 30 cyc!es/sec and 2 g are given in Fig. 3. By comparison with Fig_ 2, it is immediately obvious that the amount of segregation is
1 0
I
2
eorrom
3
4
5
6
sc.an
7 Pal”*
6
9
IO
II
I2 TOP
Fig. 2. Graph showing effect of vibrating at 2 g and 30 cps the flint siiica sandlradioactive sand system.
RESULTS
AND DISCUSSION
Flint sand
The major difference between the flint sand and the iridium 192 tagged sand was that the flint sand had a larger particle size as can be seen from Table 1. A sandwich layer of the tagged sand gave the initial tracer response as seen in Fig. 2. The count level at the 12 equidistant reference points was recorded at time periods of 2,5,9,15,20,30 and 50 mix The vibration conditions were a frequency of 30 cycles/set and an amplitude of 2 g or 0.015 in. peak to peak. Representative results at time of 0, 5 and 30 min showing the diffusion of the sandwich layer mostly in an upward direction are given in Fig_ 2. This result is somewhat surprising because, fn other previous studies using a single large marker particle, the movement of the larger particle was always upward.
0
2
3
4
5
6
I
I
7
6
,
9
,
IO
,
If
Fig. 3. Graph showing the effect of vibrating the banding sand/radioactive sand system for 30 min at 30cpsand2g.
55
very much slower for the tracer in banding _ sandthanitwasinflintsand_Thereisasmall upward shift in the position and a general broadening of the peak. An interesting&zalxue is that there is more movement of the tracer in a downward direction than there &as with the flint sand. Lactose For the next. series of experiments, a sample of lactose was used as tbe test material_ Lactose was chosen because of its particle size; from the properties listed in TabIe 1, it can be seen that the average particle size is of ‘the order of 40 Ctm_At this size, cohesive forces are heginning to compete with the gravitational forces_ The cohesive properties of lactose become apparent when the powder is subjected to light pressure_ One could consider this powder to be at the mid-point between a free-flowizzg and a cohesive powder_ It is less dense than sand. Tbe results obtained when a sample was subjected to vibration at 30 cycles/set and
2gfor30minaregiveninFig_4_Itisseen that th&re is very little segregation (compare with flint br banding sand)_ It w&Id appear that there is sufficient cohesion within the s&em to prevent f&e movement within a bed and thus migration of the particles is inhibited. A further series of experiments was performed to test whether an increase in the applied energy of the vibration could overcome these cohesional forces. A cylinder filled with lactose was placed on the vibrating table. The values of &equency and amplitude were then varied until substantial movement was observed to take place with the powder - it had an agitated or seething appearance. It was found that the value was 100 cycles/set and 10 g. The experiment was now repeated using the new values of frequency and amplitude_ The results obtained after only 3 minutes of vibrating time are shown in Fig. 4, The experiment was stopped after this time since considerable movement within the system was obvious even by visual observation_ From Fig. 4, it is very obvious that substantial migration of the particles has taken place. It is also noticed that the migration has heen only in an upward direction_ It would seem that at a critical vibration condition, the powder bed can be induced to take on a semifluidized state. Under these conditions, comparatively
large
voids
are induced
in the
lactose bed, and the sand particles having a greater upward momentum are able to pass up tbrougb than
the
tbe
bed.
critical
At vibration value,
the
bed
conditions is not
other
fluidized
and the cohesive forces are sufficient to enable it to retain a fina structure.
scanmim mnun Fig. 4. Graph showing movement due to vibration lactose hydrous/radioactive sand system.
TOP
of
Milied zircon ExperimenB were conducted with milled zircon silicate. This material was chosen because it is truly cohesive_ The average particle size is substantially less than 1 pm and it has a density of 4.6 g/cm3_ The experiments were conducted as detailed previously using a sandwich layer_ Vibration at 30 cycles/s.ec and 2 g was applied to tha system for 30 min. Visually, it was observed that in thy initial few minutes a substantial rearrangement took place within the cylinderAfter this had occurred, no further movement was observed_ The experiment was continued for 30 min and the resultant scan is shown in
56
CONCLUSION This
experimental
study has revealed
interesting observations on vibrational
many segrega-
tion. These may be listed as follows. of
TOP
Fig_ 5. Graph showing zireonlradioactive sand sysiem before and after vibrating at 30 cps and 2 g.
Fig_ 5. It can be seen that considerable segregation had occurred in an upward direction and to a less extent downward. This result was somewhat unexpected since the milled zircon, being cohesive, was expected to inhibit movement of the tracer. It was thought that the reason for this effect might
be that
the vibration
caused
a consider-
able quantity of air to be displaced from the system. As the air passed through the bed it would have the effect of fluidizing the bed_ Two observations supported this thought One was the visual observation of the re arrangement taking place .Ui"qg the early stages of the experiment. Secondly, the bulk density of the -Tstem, 1.8 g/cm=, was very much lower than the tapped density, 2.7 g/cm=, indicating that a considerabIe quantity of air was displaced by tappingAn experiment was then performed to test this idea, The experiment was repeated, but
this time the vibration was only applied for 5 min. It can be seen that apart from more irregular distribution, the same overall segregation had occurred. Thus, it was felt that the concept of segregation during fluidization due to the release of entrapped air had been substantiated. -Further, the limited movement of particles in a bed of cohesive material after thisinitialre arrangement has taken place is confirmed_
(1) When vibration is introduced at the base a vessel, then the migration of the particles
tends to be mainly in an upward direction even when the particles in the upper layer are larger in size than the marker particles. (2) In a binary system of freeflowing particles, where the marker particles and the other particles are similar in size and density, very slow migration takes place in an upward and downward direction. (3) In a binary system consisting of a freeflowing marker and a powder demonstrating incipient cohesion, segregation will normally be very limited. (4) At a particular level of vibrational energy, a system can be caused to exhibit high activity. At this point, the degree of segregation is very much increased. (5) Particle size has been confiied as the major controlling factor for segregation at a particular vibration condition. This has been shown for situations where the second powder has half the density and also twice the density of the marker particles. It is to be ezpected that an experimental study of limited scope, such as the one described here, will leave many questions unanswered. The study supports the view of previous authors, that the interactions between vibrational energy and particle properties are very complex. It is hoped that an extended study will follow in which the basic science may be developed by computer modelling supported by scientific experimentation. ACKNOWLEDGEMENTS
Grateful
thanks
are offered
to Kenneth
Walanski, who performed the laboratory tests, and to Richard Semmler, who advised on the use of radioactive tracer technique_ I would also like to thank Continental Can Company, E. R. Squibb & Sons Inc., Allegheny Ludlum Industries, United States Steel Corporation, Canadian Industries Ltd., _ Westinghouse Electric Corporation and Rank Xerox Corporation for their support of this . study-
-.. .-
__ __._
P&tide
__ L -.:57
-
-?
d+xzity, size ir&a~~~ona and v&ll effects, 53 (19ti) 1360.;. -_. - - J. Ii. Olsen. Ph_ D_ Diss.; Univ of Minntik 1964. J, C, Williams and &shields, stioq of gram&& in-vibrated beds. Powder Technol;, 1 (19673‘134 - 142. K. Abmad and I. J. Smalley, Observation of particle segregathn in vibrated granular systems, Powder Technol., 8 (1973) 69 - 75. C. F. Harwobd, K. Whnski and R. Semmler, A miniature scintillation probe for use in powder tracer studies, C&m. Instrum., accepted for publication, 6 (3) (1975).
.J.-Pbnn:Se&; R. L_ Brown, The fundamental pri&cipIes- of seg& gation;‘J_ Inst. Fuel; 13 (1939) 15 i 19. _ i_ C_V&iam& Fuel -!So&.‘J_, Ui&_ pf SheffieId[ 14 (1963)29. -J. L_ Olsen an< E G. Rippie. -ation kinetics of particulate soli+ systemS._L The influence’of_ particle size md particle size distribution, J. Pharmi sci, 53 (1964) 147.1 E. G. Rippie, J. L. Olsen z&d M. D. Faiman, Begregation kinetics of particulate solid systems. IL
-
_
:
1.
_.~.
.P&der Technology. 16 < 1977) 51 - 57 @ Elsevier Sequoia S-A_. Lausanne - Printed in the Netherlands
Powder
COLIN
begation
51
Due to Vibration*
F. HARWOOD
IIT Research (Received
Institute.
10 West 35th Street.
Chicago, III. 606t6
(U.S.A.)
March 17.1976)
SUMMARY
rumble.
An experimental study on the segregation of particles when subjected to vibrational energy has been made. Radioactive tracer techniques were used to follow the movement of the powder_ The method did not disturb the bed and thus the segregation could be measured at various time intervals. It was found that the powder properties and the amplitude and frequency of the vibrations had a profound effect on the results obtained. In freeflowing powders, segregation was readily induced- Powder which demonstrated ‘ncipient cohesitin had pronounced segregation only at a critical vibrational energy. Truly cohesive powders exhibited very little segregation once the powder had packed down to its final bulk density.
INTRODUCTION
Powder beds are frequently sfibjecteci to vibrations_ The vibrations may be applied intentionally to consolidate a powder into a die or mold, or to aid the movement of a powder from a hopper or chute. Alternatively, the vibrations may occur inadvertently as a result of machinery rumble. In either case, the effect of i;he applied vibrations is a complex function of the amplitude and frequency of the vibrations and the properties of the powder bed. The objective of this study was to follow the movement of particles within a bed of similar particles which are typical of real @owder beds. The vibrations applied were within the range expected from heavy-machinery *Paper p&enti at the Secodd International Conference on the Compaction arid Consolidation of particulate Matter, Brighton, England. September 2 - 4. 1975.
As part
of the
study,
a radioactive
which ailowed the movement of the particles with time to be followed without disturbing the powder bed. Various powders were tested against a radioactive tracer sand and the relative segregation effects observed_ The materials were selected to be free flowing, incipiently cohesive, and cohesive. They had particle sizes greater and less than the marker sand and densities greater and less than the marker sand. Comparisons could then be made over a very wide range of properties. tracer
technique
was developed
BACKGROUND
Past studies on the segregation of powders in a vibrated bed have been extremely limited_ Brown [1] discussed, in general terms, the segregation of particles when they were vibrated and poured into heaps. Williams [2] was able to demonstrate the upward movement of a single large particle in a bed of free flowing powder when subjected to vibration. Oisen et al_ f3 - 51 studied the segregation of steel and glass spheres of 3/32 - l/4 in. diam. in a cylinder of 3/4 in. diam. and length 6 in. They were able to show that segregation proceeded to an equilibrium state and drew an analogy to fit-order reaction kinetics. They were able to make some observations on the effect of particle size and density of the spheres. The ratio of the particle size to vessel diameter was very low: in the worst case, only three spheres would fit into a cross-sectional planeWilliams and Shields [S] measured the segregation which took place as binary mixtures of fertilizer went down a channel. The degree of segregation was found to depend on the frequency and amplitude of the vibration as well as the direction in which it is applied.
52 The results were obtained using rounded free-
flowing granules of fertilizer of undisclosed chemical composition. More recently, Ahmad and Smalley [7] have made a detailed study on the rate at which a single 12,700 pm lead sphere rose through a bed of 500 - 600 .~rn sand particles. Accelerations of 1 - 10 g and frequencies of 50 - 150 cycles/set were applied to the system. They found that acceIeration was the most critical factor affecting the segregation. At constant frfzquency, segregation incxased with acceleration increase; however: with increasing frequencies, the s.®ation was reduced- It was also found that segregation increased with increase in size of the marker particle. An increase in density of the marker particle decreased the segregation, and particle shape had little effect. The above studies have used very large single particles to mark the progress of the segregation. In the present study, radioactive tracer techniques have made it possible to use many particles of a size comparable to the powder bed.
Fig_ 1. Schematic diagram of experimental setup for scanning the radioactive/non-radioactive powder SYS.tem
Vibmting table The vibratig
table used in this study is of the type used for testing metal parts for vibrational stress breakdown. It has an extremely wide range of frequency and amplitude positions and can easily simulate the vibrational values to be expected in industrial pkiIltS.
EXPERIMENTAL Procedure Apparatus
Powder samples were placed in a 250 ml graduated plastic cylinder of inside diameter 3-4 cm_ PIastic was preferred since it minimized the danger of the cylinder breaking and thus leaking the radioactive component. A metal plug was sealed into the base; the plug, which was 3 cm bigb, was there simply to raise the IeveI of the material in the cylinder so that the nucleonic probe could reach the side of the cylinder without obstruction from the cylinder base- It took no other part in the experiment. The Iengtt of the cylinder was scanned using the Nuclear Chicago DS-8 scintillation probe, which has been described more fully in a previous paper [S] _ In order to restrict the scanned volume, a collimator was fabricated by drilling a 0.75 cm diam. hole through a lead brick_ Because of the weight of the probe and lead brick, the cylinder and its contents were moved up and down by mounting it on a lab-jack. A diagram of the experimental arrangements is shown in Fig. 1.
Each of the materials were tested against the radioactive marker sand. The marker material was placed as a layer sandwiched between two layers of the test material. The procedure adopted was to place a 120 ml layer of material in the bottom of the cylinder, followed by a 20 ml layer of radioactive sand, and finally a top layer of 120 ml of test ma‘terial. The sandwich method of filling was chosen so that migration could be foIlowed in either an upward or downward direction. When the cylinder had been loaded in this way, it was then scanned with the probe to obtain the initial condition. The cylinder was then ti_ghtly corked and placed in a stout polyethylene ‘,ag to minimize the danger of a spill. It was placed on the vibrating table and vibrated at a known amplitude and frequency for a set time- After this time, it was then again checked with the probe and a new plot of count rate uersus vertical height obtained. This process was repeated at various time intervals up to a maximum of 60 minutes vibration time. A plot of count rate uersus
53
vertikl height was made.after each time in: terval. -_ _-- -._-I- -_ _ --_
Powder
.-
It was owed thk the h&k ofthe p9wi der bed was less after vibration. This was a& expect& reklt, &d-to allow for t&is, t+e data had to be normalized. Twelve scan points were t&en, all equidistant apart. The first-and the last scan points were always l-cm from the top and bottoti, respectively. Making the necessary assumption that an even packing had been induced, then the distance apart of the subsequent scan points ‘2a.s simply: (Height of column after vibrating -2)
materidg
In all, five powder materie
x
(Initial height of column -2)
have been used in the experiments. The tracer material was a free-flowing silica sand with a surface layer of iridium 192 baked bn the surface. Other mate. rials were two grades of silica sand, both free Bowing hut one larger in size than the tracer tid the other smaller in size; they had substantially the same density as the tracer sand. A pharmaceutical grade of lactose was used which was normally free flowing but became cohesive on compaction; it had a particle size much smaller than the tracer. Milled zircon was used which had a very small particle size and was cohesive in nature. The properties of these materials are given in Table 1.
Initial distance apart (cm)
TABLE1 Powderproperties
Property
Testpowders Tracersand
Sizedistribution.J.ml 420-600 300-420 210-300 150-210 105-150 75-105 50-75 30-50 10.30 5-10 3-5 <3 Density.g/cm3 *e,p Bu=, PB Tapped.
PT
Angleofrepose Drained Dynamic observedflow prOpertieS
Notmeasured. Quotedas 100%1805Ocun
Not
measured.
Assumesimilar tosand
Flintsand
2% 59 34 4 1
2-68 1.30 1.50
Banding sand
1% 3 33 37 20 20 6
2-68 1.30 1.50
Lactase
Milled ZircOll
10 17 58 12 3
-0.525 0.70 0.76
8 8 84
4.56 1.81 2.68
Notmeasured
32" 35"
34' 37"
26' 34"
46' 47"
Free
FLXS flowing
Free
Incipiently cchesive
Cohesive
flowing
flowing
It is suggested that in real powder systems where there are large numbers of particles of not too different size, the action of vibration is to induce a state of homogeneity into the powder bed by creating a mixing action. Another factor which can explain the results is the surface coating of iridium salts which impregnates the tracersand particles. While no layer is observkble under the microscope, it is well known that surface layers can lead to stray modifications of powder propertiesBanding
sand
The banding sand had a particle size closer to that of the tracer sand than did the flint sand. Vibration tests were conducted using identical conditions to those for the flint sand. The results obtained for zero time and after 30 min vibration at. 30 cyc!es/sec and 2 g are given in Fig. 3. By comparison with Fig_ 2, it is immediately obvious that the amount of segregation is
1 0
I
2
eorrom
3
4
5
6
sc.an
7 Pal”*
6
9
IO
II
I2 TOP
Fig. 2. Graph showing effect of vibrating at 2 g and 30 cps the flint siiica sandlradioactive sand system.
RESULTS
AND DISCUSSION
Flint sand
The major difference between the flint sand and the iridium 192 tagged sand was that the flint sand had a larger particle size as can be seen from Table 1. A sandwich layer of the tagged sand gave the initial tracer response as seen in Fig. 2. The count level at the 12 equidistant reference points was recorded at time periods of 2,5,9,15,20,30 and 50 mix The vibration conditions were a frequency of 30 cycles/set and an amplitude of 2 g or 0.015 in. peak to peak. Representative results at time of 0, 5 and 30 min showing the diffusion of the sandwich layer mostly in an upward direction are given in Fig_ 2. This result is somewhat surprising because, fn other previous studies using a single large marker particle, the movement of the larger particle was always upward.
0
2
3
4
5
6
I
I
7
6
,
9
,
IO
,
If
Fig. 3. Graph showing the effect of vibrating the banding sand/radioactive sand system for 30 min at 30cpsand2g.
55
very much slower for the tracer in banding _ sandthanitwasinflintsand_Thereisasmall upward shift in the position and a general broadening of the peak. An interesting&zalxue is that there is more movement of the tracer in a downward direction than there &as with the flint sand. Lactose For the next. series of experiments, a sample of lactose was used as tbe test material_ Lactose was chosen because of its particle size; from the properties listed in TabIe 1, it can be seen that the average particle size is of ‘the order of 40 Ctm_At this size, cohesive forces are heginning to compete with the gravitational forces_ The cohesive properties of lactose become apparent when the powder is subjected to light pressure_ One could consider this powder to be at the mid-point between a free-flowizzg and a cohesive powder_ It is less dense than sand. Tbe results obtained when a sample was subjected to vibration at 30 cycles/set and
2gfor30minaregiveninFig_4_Itisseen that th&re is very little segregation (compare with flint br banding sand)_ It w&Id appear that there is sufficient cohesion within the s&em to prevent f&e movement within a bed and thus migration of the particles is inhibited. A further series of experiments was performed to test whether an increase in the applied energy of the vibration could overcome these cohesional forces. A cylinder filled with lactose was placed on the vibrating table. The values of &equency and amplitude were then varied until substantial movement was observed to take place with the powder - it had an agitated or seething appearance. It was found that the value was 100 cycles/set and 10 g. The experiment was now repeated using the new values of frequency and amplitude_ The results obtained after only 3 minutes of vibrating time are shown in Fig. 4, The experiment was stopped after this time since considerable movement within the system was obvious even by visual observation_ From Fig. 4, it is very obvious that substantial migration of the particles has taken place. It is also noticed that the migration has heen only in an upward direction_ It would seem that at a critical vibration condition, the powder bed can be induced to take on a semifluidized state. Under these conditions, comparatively
large
voids
are induced
in the
lactose bed, and the sand particles having a greater upward momentum are able to pass up tbrougb than
the
tbe
bed.
critical
At vibration value,
the
bed
conditions is not
other
fluidized
and the cohesive forces are sufficient to enable it to retain a fina structure.
scanmim mnun Fig. 4. Graph showing movement due to vibration lactose hydrous/radioactive sand system.
TOP
of
Milied zircon ExperimenB were conducted with milled zircon silicate. This material was chosen because it is truly cohesive_ The average particle size is substantially less than 1 pm and it has a density of 4.6 g/cm3_ The experiments were conducted as detailed previously using a sandwich layer_ Vibration at 30 cycles/s.ec and 2 g was applied to tha system for 30 min. Visually, it was observed that in thy initial few minutes a substantial rearrangement took place within the cylinderAfter this had occurred, no further movement was observed_ The experiment was continued for 30 min and the resultant scan is shown in
56
CONCLUSION This
experimental
study has revealed
interesting observations on vibrational
many segrega-
tion. These may be listed as follows. of
TOP
Fig_ 5. Graph showing zireonlradioactive sand sysiem before and after vibrating at 30 cps and 2 g.
Fig_ 5. It can be seen that considerable segregation had occurred in an upward direction and to a less extent downward. This result was somewhat unexpected since the milled zircon, being cohesive, was expected to inhibit movement of the tracer. It was thought that the reason for this effect might
be that
the vibration
caused
a consider-
able quantity of air to be displaced from the system. As the air passed through the bed it would have the effect of fluidizing the bed_ Two observations supported this thought One was the visual observation of the re arrangement taking place .Ui"qg the early stages of the experiment. Secondly, the bulk density of the -Tstem, 1.8 g/cm=, was very much lower than the tapped density, 2.7 g/cm=, indicating that a considerabIe quantity of air was displaced by tappingAn experiment was then performed to test this idea, The experiment was repeated, but
this time the vibration was only applied for 5 min. It can be seen that apart from more irregular distribution, the same overall segregation had occurred. Thus, it was felt that the concept of segregation during fluidization due to the release of entrapped air had been substantiated. -Further, the limited movement of particles in a bed of cohesive material after thisinitialre arrangement has taken place is confirmed_
(1) When vibration is introduced at the base a vessel, then the migration of the particles
tends to be mainly in an upward direction even when the particles in the upper layer are larger in size than the marker particles. (2) In a binary system of freeflowing particles, where the marker particles and the other particles are similar in size and density, very slow migration takes place in an upward and downward direction. (3) In a binary system consisting of a freeflowing marker and a powder demonstrating incipient cohesion, segregation will normally be very limited. (4) At a particular level of vibrational energy, a system can be caused to exhibit high activity. At this point, the degree of segregation is very much increased. (5) Particle size has been confiied as the major controlling factor for segregation at a particular vibration condition. This has been shown for situations where the second powder has half the density and also twice the density of the marker particles. It is to be ezpected that an experimental study of limited scope, such as the one described here, will leave many questions unanswered. The study supports the view of previous authors, that the interactions between vibrational energy and particle properties are very complex. It is hoped that an extended study will follow in which the basic science may be developed by computer modelling supported by scientific experimentation. ACKNOWLEDGEMENTS
Grateful
thanks
are offered
to Kenneth
Walanski, who performed the laboratory tests, and to Richard Semmler, who advised on the use of radioactive tracer technique_ I would also like to thank Continental Can Company, E. R. Squibb & Sons Inc., Allegheny Ludlum Industries, United States Steel Corporation, Canadian Industries Ltd., _ Westinghouse Electric Corporation and Rank Xerox Corporation for their support of this . study-
-.. .-
__ __._
P&tide
__ L -.:57
-
-?
d+xzity, size ir&a~~~ona and v&ll effects, 53 (19ti) 1360.;. -_. - - J. Ii. Olsen. Ph_ D_ Diss.; Univ of Minntik 1964. J, C, Williams and &shields, stioq of gram&& in-vibrated beds. Powder Technol;, 1 (19673‘134 - 142. K. Abmad and I. J. Smalley, Observation of particle segregathn in vibrated granular systems, Powder Technol., 8 (1973) 69 - 75. C. F. Harwobd, K. Whnski and R. Semmler, A miniature scintillation probe for use in powder tracer studies, C&m. Instrum., accepted for publication, 6 (3) (1975).
.J.-Pbnn:Se&; R. L_ Brown, The fundamental pri&cipIes- of seg& gation;‘J_ Inst. Fuel; 13 (1939) 15 i 19. _ i_ C_V&iam& Fuel -!So&.‘J_, Ui&_ pf SheffieId[ 14 (1963)29. -J. L_ Olsen an< E G. Rippie. -ation kinetics of particulate soli+ systemS._L The influence’of_ particle size md particle size distribution, J. Pharmi sci, 53 (1964) 147.1 E. G. Rippie, J. L. Olsen z&d M. D. Faiman, Begregation kinetics of particulate solid systems. IL
-
_