- Email: [email protected]
The size distribution of gas bubbles leaving a three-phase fluidized bed
Powder Technology-Elsevier
Sequoia
SA., Lausanne-Printed
in the Netherlands
245
The Size Distribution of Gas Bubbles Leaving a Three-Phase Fluidized Bed R. E. PAGE* and D. HARRISON Department of Chemical Engineering, (Received
August
24, 1971;
University of Cambridge, Pembroke Street, Cambridge (Gt. Britain)
in revised form November
8, 1971)
Summary The size distribution of gas bubbles leaving a three-phase jluidized bed has been measured photographically, ana’ the influence of changes in liquid and gas jlow rates and in gas distributor design on the size of gas bubbles has been studied. It has been found that gas bubble size is reduced by increase in Iiquid velocity. At low liquid velocities the bubble size is reduced by an increase in gas jlow rate, but as the liquid velocity is increased the bubble size becomes independent of gas flow rate. Away from the gas distributor, the bubble size distribution is independent af the distributor design. The experimental results have been relatedqualitatively to the competing effects in a three-phase jluidized system of bubble splitting and bubble coalescence.
INTRODUCTION
In a three-phase fluidized system a bed of particles is fluidized co-currently by liquid and by gas; liquid makes up the continuous phase and the gas the discontinuous. As a process operation, three-phase fluidization is in competition industrially with other and perhaps more conventional contacting reactors, such as the agitated-slurry, the trickleflow, and the bubble-column slurry. Nevertheless it has been used, for example, for the catalytic hydrogenation of fatty oils and for the desulphurization by hydrogenation of liquid petroleum fractions. A review of three-phase processes has been given by (astergaard’.
The principal advantages of a three-phase fluidized system over its close rivals are (a) The easy equilibration of temperature within the particle bed. (b) A relatively low pressure drop across the bed l
Present
address
: Sandoz Ltd., Basle, Switzerland.
Powder Technol., 6 (1972)
which is insensitive to changes of liquid flow rate. (c) Operational flexibility : particles with differing physical and chemical properties can usually be handled successfully by modifying the process conditions. (d) The addition and withdrawal of particles from the process equipment is relatively easy (e.g. for catalyst regeneration). The rate of carry-over of particles from a threephase fluidized system can be substantial and in practice this is a serious disadvantage. The size of gas bubbles within the particle bed, and particularly the size distribution leaving the bed surface (where the bubbles erupt and transport particles into the freeboard), plays a major role in determining the amount of carry-over. The results of Massimilla et al.’ suggest that the bubble size distribution within the bed reaches an equilibrium after a bed depth of about 60 cm, and a similar result has been found for small particles by 0stergaard3 and for large particles by Lee4. Other work on bubble size in three-phase fluidized beds has been carried out by Sherrards, in which the bubble size was calculated from measurements of the projected area of the bubbles (measured by a light probe) and the gas hold-up (measured by a radioactive absorption technique). This paper describes an experimental investigationofthebubblesizedistributionleavingthesurface of a three-phase fluidized bed. The measurements have been carried out photographically, and the quantitative effects on gas bubble size brought about by changes in liquid and gas flow rate and in the design of the gas distributor have been studied.
EXPERIMENTAL
Apparatus
The experiments here described were carried out in a circular glass column of 22.8 cm diam. The par-
246
R. E. PAGE,
title bed consisted of nearly spherical 500~pm sand particles of close size range (B.S.S. sieve range 25-30 ; U,r = 0.5 1 cm/s) fluidized by water and bubbles of air. The size distribution of the air bubbles was obtained photographically, but direct photographs of the column would have recorded considerable bubble distortion and therefore a technique similar to that of Stewart6 was used to minimize this effect. The upper part of the column was encased in a water-tight perspex box 60 cm high and with a base 27 cm square fitted with a drain cock. This box was open at the top, and the base had a hole large enough to take the glass column which was sealed to the box by an “0” ring. The box was filled with water. Although some previous workers have attempted to match the refractive indices of the liquids inside and outside the column with the material of construction of the column, it was not practicable to carry out the fairly large-scale experiments to be described using such materials. However, in practice it was found that the distortion of the bubbles due to differences of refractive index was negligible, and indeed may have been rather less than the distortion of the bubbles caused by irregularities in the glass of the column which can occur during manufacture. Some allowance could have been made for serious distortion by introducing into the analysis of results a scaling factor related to radial position, but this was not found to be necessary. Behind the bed, a white screen was illuminated uniformly by a quartz-iodine floodlamp, and the sides of the box which encased the column were darkened by black paper. This lighting gave good definition of the bubble edge, and it also introduced a useful three-dimensional effect : bubbles near the
I
-al
Camem
e,
I
/
Fig. 1. Illumination
Powder Tech&
e,
of hidized
6 (1972)
bed (plan view).
D. HARRISON
camera were illuminated by light incident from a smaller angle than those away from the camera (see Fig. 1). In this way near bubbles appeared very dark at the edges, with a small central light region where light passed straight through the bubble without deviation. Bubbles far from the camera, illuminated from a larger angle, appeared much lighter, and the dark rim around the bubble was much thinner. These differences were particularly helpful in the analysis of the films. Cine-photographs were taken of the column with a Mitchell HS16-F2 camera using a 73” shutter at 8 frames/s ; a 25-mm Dallmayer lens was used. Two types of gas distributor were used, giving large and small gas bubbles: (a) Large bubbles wereproduced from a0.635-cmdiam. single orifice pointing vertically upwards. The orifice was fed by a horizontal tube which entered the column about 15 cm above the liquid distributor. (b) Small bubbles were produced from a distributor with ten 0.154-cm-diam. holes pointing vertically upwards, arranged in a straight line across the diameter of the column. This distributor was also fed by a horizontal tube which entered the column about 15 cm above the liquid distributor. A wire gauze was placed over the top of the column to minimize the loss of sand particles. Method
The bed was fluidized uniformly, and particles were removed or added to bring the region under investigation-just above the bed-within the illuminated section. The bed depth was always greater than 120 cm to ensure that an equilibrium of bubble size distribution had been reached, and before experimental readings were taken the bed was allowed to settle by fluidizing for at least an hour. Photographs were taken of the region of the column just above the bed surfaceat 8 frames/s for approximately 40 s-for two gas distributors (single orifice, multiorifice), two liquid velocities (0.85, 1.43 cm/s), and three gas velocities (0.30,0.51,0.73 cm/s)-making twelve experimental runs altogether. With the column filled with water, a scale was photographed at several positions in the column to assess the extent of bubble distortion and the effect of perspective ; both these possible sources of errors were found to be of negligible influence. Analysis of results
The tine-photographs
enable measurements
of
SIZE
DISTRIBUTION
OF GAS
247
BUBBLES
bubble basal diameter D, (given by the bubble’s horizontal extremities) to be made, but a detailed analysis is extremely tedious without the use of an automatic or semi-automatic technique. An automatic analysis technique would need to be able to distinguish between bubbles and entrained particles, and, as no suitable equipment has yet been developed, a semi-automatic method was used. The 16-mm film was projected onto a film analyser which produced an image on a horizontal ground-glass screen. To obtain the bubble diameter and the bubble spatial distribution, equipment developed by Williams’ for measuring (logging) positions in one dimension was used. This equipment will be referred to as the x-logger. The x-logger shown diagramatically in Fig. 2 consists of a pointer which slides in one dimension,
Jpotentiwneter
Steel
OveroIl
dimensons
Length Width Height
55cI-n *cm 5cm
(opprox)
guided by 0.635-cm steel rails. Fixed to a pointer is a thin wire which at one end is wound round the axle of a helical potentiometer, and at the other end round a free wheel. A constant potential is applied to the potentiometer, and the voltage between two of its terminals is determined by the position of the pointer. This voltage can be fed into a digital voltmeter, and then to a data-logger. The results, in the form of punched tape, can be read directly to an IBM 1620 digital computer. The calibration of the x-logger over its whole range showed that the voltage output was a linear function of the distance from a fixed point. To measure a bubble basal diameter from a film, the x-logger was brought to the same vertical position as the bubble and the horizontal extremities of the bubble were recorded. By keeping the body of the x-logger in the same position with respect to the projected image of the column walls, the voltage then recorded could be used to give an accurate indication of the position of the bubble in the column. The bubble diameter was recorded as the difference of the two bubble-extremity co-ordinates, and the position of the bubble was recorded by the absolute voltage (see Fig. 3) A very large number of bubbles was recordedup to 3500 in a single run-and care had to be taken to avoid measuring the same bubble twice.
rails
RESULTS Fig. 2. X-logger.
Table 1 shows the conditions for different experimental runs using a sand/water/air system. TABLE
1
Experimental Run/tape number
0 10
6
6
Distance
4
Gas
--
---
Powder
0
column
2
4
centre
(cm)
flow-rate
x
0.30
cm/s
0
0.51
cm/s
Theoretical
Fig. 3. Spatial
2
from
line
distribution
Technol., 6 (1972)
for
uniform
distribution
of gas bubbles.
6
0
10
1 2 3 4 5 6 7 8 9 10 11 12
conditions Gas velocity
Liquid velocity
(cm/4
(cm/4
0.30 0.51 0.73 0.30 0.51 0.73 0.30 0.51 0.73 0.30 0.51 0.73
0.85 0.85 0.85 0.85 0.85 0.85 1.43 1.43 1.43 1.43 1.43 1.43
S = single orifice
; M = multi-orifice.
Gas distributor
Sample size
S S S M M M M M M S S S
900 1661 2700 853 2424 2019 2554 3525 3099 3232 2443 2505
R. E. PAGE,
248 1.0
i
I) i al
.;
I
0
x
I :, ,
,
,
,
2 3 4 5 6 Bubble diameter (cm)
7
8
9
i,
, 1
,
,
Run 9 (Multi-orifice 0
distributor)
complete shortened
Fig. 4. SizeXdistribution of gas bubbles.
Bubble
diameter
km)
D. HARRISON
A check was first made to ensure that large enough samples had been taken. To do this, one of the longer runs (tape 9) was divided into two sections, the shorter being approximately as long as the shortest run (tape 4). The results for the bubble size distribution-expressed as the fractionfof bubbles with basal diameter greater than a given size-for the two parts were then compared. Figure 4 shows them to be in reasonable agreement, and therefore it was assumed that the sample sizes given in Table 1 were sufficiently large. As has been explained, results recorded by the few points in Fig. 4 were obtained from measurements on a large number of individual bubbles, and similarly for the results given in Fig. 5. However, there are fewer individual measurements per point on the curves in Fig. 3 which shows spatial distribution of bubbles in the column, and for this parameter it is probable that the sample sizes were not large enough for accuracy, especially near the column walls. The theoretical line shown in Fig. 3 was computed on the assumption of a uniform supply of gas per unit area to all parts of the distributor. The fraction of bubbles with diameter greater than a given size can be represented quite reasonably by a straight line on a log-linear plot, as is shown in Fig. 5. The experimental curves deviate from linearity for high and low values of II,,. For large IIt,, Hinze’ and Sherrard’ have suggested that there is a maximum stable bubble size, and Fig. 5 indicates there is some evidence of this in the sharp fall inf: For small D,, the deviation from a linear plot is because there are proportionally more small bubbles than medium-sized bubbles in a given experimental run. At the higher liquid velocity (1.43 cm/s), the bubble size distribution was independent of the gas flow rate (see Fig. 5) and also of the design of the gas distributor. At the lower liquid velocity (0.85 cm/s) the gas distributor again exerted no influence, but Fig. 5 shows there is a marked decrease in the bubble size with an increase in gas flow rate. Nevertheless, Fig. 5 shows that the bubble size is always smaller for the high liquid velocity than for the low.
DISCUSSION
fqg%Jq Fig. 5. Size distribution
Powder Tech&
of gas bubbles.
6 (1972)
The bubble size distribution to be found leaving a three-phase fluidized bed appears from observation to arise from the competing effects of bubble splitting and bubble coalescence. The physical me-
SIZE DISTRIBUTION
chanisms of these phenomena are not well understood and their detailed analysis is far from available. Nevertheless, in qualitative terms, we may list several related factors which are known to be important: (a) Hinze’ has shown that bubbles can be split by turbulent eddies, and therefore an increase in the overall agitation of a three-phase fluidized system -by increase of gas flow rate, for instance-would be expected to lead to some decrease in bubble size. (b) The rate of coalescence of bubbles is higher in liquids of higher viscosity’, and the apparent viscosity of a three-phase fluidized bed falls with increase of liquid velocity. (c) Bubbles can be observed to be split by single particles, and possibly by groups of particles. With this background, we may expect (and find in Fig. 5) that at the lower liquid velocity (and higher apparent viscosity) there is a fall in bubble size with the increased agitation of the system that is brought about by an increase in the gas flow rate. On the other hand, at the higher liquid velocity (and lower apparent viscosity), we may expect (and find in Fig. 5) the smaller bubbles which would be a consequence of smaller rates of bubble coalescence. It is clear, however, that particular situations could be influenced to an unknown extent towards smaller bubble sizes by the break-up of bubbles by the particles themselves. This may occur more readily at higher liquid velocities. If the bubble size distribution, particularly at the bed surface, is primarily a function of the competition between bubble splitting and bubble coalescence, then we would not expect it to be influenced by the form of the gas distributor. Experiment supports this conclusion for bubbles erupting from the bed. Bubbles leaving the surface of a three-phase fluidized bed carry particles into the freeboard, and experimental data on the bubble size distribution given in this paper have been used” to estimate the amount of particle entrainment from such a system.
Powder Tech&.,
6 (1972)
249
OF GAS BUBBLES CONCLUSIONS
It has been found experimentally for gas bubbles leaving the surface of a three-phase fluidized bed that (a) The bubble size is reduced by increase in liquid velocity. (b) At low liquid velocities the size of gas bubbles is reduced by an increase in gas flow rate, but as the liquid velocity is increased the gas bubble size becomes independent of gas flow rate. (c) Away from the gas distributor, the bubble size distribution is independent of the distributor design. ACKNOWLEDGEMENT
One of us (R.E.P.) would like to thank the Trustees of the Esso Fund for financial support during the course of this research work. LIST OF SYMBOLS
bubble basal diameter, cm fraction of bubbles with diameter greater than D, U,r minimum fluidizing (liquid) velocity, cm/s REFERENCES 1 K. P)stergaard, Advan. Chem. Eng., Vol. 7, Academic Press, New York, 1968; and Fluidization (eds. J. F. Davidson and
D. Harrison), Ch. 18, Academic Press, London, 1971. 2 L. Massimilla, A. Solimando and E. Squillace, Bit. Chem. Eng., 6 (1961) 232. 3 K. Bstergaard, Chem. Eng. Sci., 21 (1966) 470. 4 J. C. Lee, Discussion, Proc. 3rd European Symp. Chem. Reaction Eng., Amsterdam, 1965, Pergamon, Oxford, p. 211. 5 A. J. Sherratd, Three-phase fluidised beds, Ph. D. dissertation, Univ. of Swansea, 1966. 6 P. S. B. Stewart, Fluidization, Some hydrodynamic studies, Ph. D. dissertation, Univ. of Cambridge, 1965. 7 R S. Williams, Personal communication, 1970. 8 J. 0. Hinze, Am. Inst. Chem. Engrs. J., 1 (1955) 289. 9 P. H. Calderbank, M. B. Moo-Young and R. Bibby, Proc. 3rd European Symp. Chem. Reaction 1965, Pergamon, Oxford p. 91.
Eng.,
Amsterdam,
10 R. E. Page, Some aspects of three-phase lluidization, Ph. D. dissertation, Univ. of Cambridge, 1970.
Sequoia
SA., Lausanne-Printed
in the Netherlands
245
The Size Distribution of Gas Bubbles Leaving a Three-Phase Fluidized Bed R. E. PAGE* and D. HARRISON Department of Chemical Engineering, (Received
August
24, 1971;
University of Cambridge, Pembroke Street, Cambridge (Gt. Britain)
in revised form November
8, 1971)
Summary The size distribution of gas bubbles leaving a three-phase jluidized bed has been measured photographically, ana’ the influence of changes in liquid and gas jlow rates and in gas distributor design on the size of gas bubbles has been studied. It has been found that gas bubble size is reduced by increase in Iiquid velocity. At low liquid velocities the bubble size is reduced by an increase in gas jlow rate, but as the liquid velocity is increased the bubble size becomes independent of gas flow rate. Away from the gas distributor, the bubble size distribution is independent af the distributor design. The experimental results have been relatedqualitatively to the competing effects in a three-phase jluidized system of bubble splitting and bubble coalescence.
INTRODUCTION
In a three-phase fluidized system a bed of particles is fluidized co-currently by liquid and by gas; liquid makes up the continuous phase and the gas the discontinuous. As a process operation, three-phase fluidization is in competition industrially with other and perhaps more conventional contacting reactors, such as the agitated-slurry, the trickleflow, and the bubble-column slurry. Nevertheless it has been used, for example, for the catalytic hydrogenation of fatty oils and for the desulphurization by hydrogenation of liquid petroleum fractions. A review of three-phase processes has been given by (astergaard’.
The principal advantages of a three-phase fluidized system over its close rivals are (a) The easy equilibration of temperature within the particle bed. (b) A relatively low pressure drop across the bed l
Present
address
: Sandoz Ltd., Basle, Switzerland.
Powder Technol., 6 (1972)
which is insensitive to changes of liquid flow rate. (c) Operational flexibility : particles with differing physical and chemical properties can usually be handled successfully by modifying the process conditions. (d) The addition and withdrawal of particles from the process equipment is relatively easy (e.g. for catalyst regeneration). The rate of carry-over of particles from a threephase fluidized system can be substantial and in practice this is a serious disadvantage. The size of gas bubbles within the particle bed, and particularly the size distribution leaving the bed surface (where the bubbles erupt and transport particles into the freeboard), plays a major role in determining the amount of carry-over. The results of Massimilla et al.’ suggest that the bubble size distribution within the bed reaches an equilibrium after a bed depth of about 60 cm, and a similar result has been found for small particles by 0stergaard3 and for large particles by Lee4. Other work on bubble size in three-phase fluidized beds has been carried out by Sherrards, in which the bubble size was calculated from measurements of the projected area of the bubbles (measured by a light probe) and the gas hold-up (measured by a radioactive absorption technique). This paper describes an experimental investigationofthebubblesizedistributionleavingthesurface of a three-phase fluidized bed. The measurements have been carried out photographically, and the quantitative effects on gas bubble size brought about by changes in liquid and gas flow rate and in the design of the gas distributor have been studied.
EXPERIMENTAL
Apparatus
The experiments here described were carried out in a circular glass column of 22.8 cm diam. The par-
246
R. E. PAGE,
title bed consisted of nearly spherical 500~pm sand particles of close size range (B.S.S. sieve range 25-30 ; U,r = 0.5 1 cm/s) fluidized by water and bubbles of air. The size distribution of the air bubbles was obtained photographically, but direct photographs of the column would have recorded considerable bubble distortion and therefore a technique similar to that of Stewart6 was used to minimize this effect. The upper part of the column was encased in a water-tight perspex box 60 cm high and with a base 27 cm square fitted with a drain cock. This box was open at the top, and the base had a hole large enough to take the glass column which was sealed to the box by an “0” ring. The box was filled with water. Although some previous workers have attempted to match the refractive indices of the liquids inside and outside the column with the material of construction of the column, it was not practicable to carry out the fairly large-scale experiments to be described using such materials. However, in practice it was found that the distortion of the bubbles due to differences of refractive index was negligible, and indeed may have been rather less than the distortion of the bubbles caused by irregularities in the glass of the column which can occur during manufacture. Some allowance could have been made for serious distortion by introducing into the analysis of results a scaling factor related to radial position, but this was not found to be necessary. Behind the bed, a white screen was illuminated uniformly by a quartz-iodine floodlamp, and the sides of the box which encased the column were darkened by black paper. This lighting gave good definition of the bubble edge, and it also introduced a useful three-dimensional effect : bubbles near the
I
-al
Camem
e,
I
/
Fig. 1. Illumination
Powder Tech&
e,
of hidized
6 (1972)
bed (plan view).
D. HARRISON
camera were illuminated by light incident from a smaller angle than those away from the camera (see Fig. 1). In this way near bubbles appeared very dark at the edges, with a small central light region where light passed straight through the bubble without deviation. Bubbles far from the camera, illuminated from a larger angle, appeared much lighter, and the dark rim around the bubble was much thinner. These differences were particularly helpful in the analysis of the films. Cine-photographs were taken of the column with a Mitchell HS16-F2 camera using a 73” shutter at 8 frames/s ; a 25-mm Dallmayer lens was used. Two types of gas distributor were used, giving large and small gas bubbles: (a) Large bubbles wereproduced from a0.635-cmdiam. single orifice pointing vertically upwards. The orifice was fed by a horizontal tube which entered the column about 15 cm above the liquid distributor. (b) Small bubbles were produced from a distributor with ten 0.154-cm-diam. holes pointing vertically upwards, arranged in a straight line across the diameter of the column. This distributor was also fed by a horizontal tube which entered the column about 15 cm above the liquid distributor. A wire gauze was placed over the top of the column to minimize the loss of sand particles. Method
The bed was fluidized uniformly, and particles were removed or added to bring the region under investigation-just above the bed-within the illuminated section. The bed depth was always greater than 120 cm to ensure that an equilibrium of bubble size distribution had been reached, and before experimental readings were taken the bed was allowed to settle by fluidizing for at least an hour. Photographs were taken of the region of the column just above the bed surfaceat 8 frames/s for approximately 40 s-for two gas distributors (single orifice, multiorifice), two liquid velocities (0.85, 1.43 cm/s), and three gas velocities (0.30,0.51,0.73 cm/s)-making twelve experimental runs altogether. With the column filled with water, a scale was photographed at several positions in the column to assess the extent of bubble distortion and the effect of perspective ; both these possible sources of errors were found to be of negligible influence. Analysis of results
The tine-photographs
enable measurements
of
SIZE
DISTRIBUTION
OF GAS
247
BUBBLES
bubble basal diameter D, (given by the bubble’s horizontal extremities) to be made, but a detailed analysis is extremely tedious without the use of an automatic or semi-automatic technique. An automatic analysis technique would need to be able to distinguish between bubbles and entrained particles, and, as no suitable equipment has yet been developed, a semi-automatic method was used. The 16-mm film was projected onto a film analyser which produced an image on a horizontal ground-glass screen. To obtain the bubble diameter and the bubble spatial distribution, equipment developed by Williams’ for measuring (logging) positions in one dimension was used. This equipment will be referred to as the x-logger. The x-logger shown diagramatically in Fig. 2 consists of a pointer which slides in one dimension,
Jpotentiwneter
Steel
OveroIl
dimensons
Length Width Height
55cI-n *cm 5cm
(opprox)
guided by 0.635-cm steel rails. Fixed to a pointer is a thin wire which at one end is wound round the axle of a helical potentiometer, and at the other end round a free wheel. A constant potential is applied to the potentiometer, and the voltage between two of its terminals is determined by the position of the pointer. This voltage can be fed into a digital voltmeter, and then to a data-logger. The results, in the form of punched tape, can be read directly to an IBM 1620 digital computer. The calibration of the x-logger over its whole range showed that the voltage output was a linear function of the distance from a fixed point. To measure a bubble basal diameter from a film, the x-logger was brought to the same vertical position as the bubble and the horizontal extremities of the bubble were recorded. By keeping the body of the x-logger in the same position with respect to the projected image of the column walls, the voltage then recorded could be used to give an accurate indication of the position of the bubble in the column. The bubble diameter was recorded as the difference of the two bubble-extremity co-ordinates, and the position of the bubble was recorded by the absolute voltage (see Fig. 3) A very large number of bubbles was recordedup to 3500 in a single run-and care had to be taken to avoid measuring the same bubble twice.
rails
RESULTS Fig. 2. X-logger.
Table 1 shows the conditions for different experimental runs using a sand/water/air system. TABLE
1
Experimental Run/tape number
0 10
6
6
Distance
4
Gas
--
---
Powder
0
column
2
4
centre
(cm)
flow-rate
x
0.30
cm/s
0
0.51
cm/s
Theoretical
Fig. 3. Spatial
2
from
line
distribution
Technol., 6 (1972)
for
uniform
distribution
of gas bubbles.
6
0
10
1 2 3 4 5 6 7 8 9 10 11 12
conditions Gas velocity
Liquid velocity
(cm/4
(cm/4
0.30 0.51 0.73 0.30 0.51 0.73 0.30 0.51 0.73 0.30 0.51 0.73
0.85 0.85 0.85 0.85 0.85 0.85 1.43 1.43 1.43 1.43 1.43 1.43
S = single orifice
; M = multi-orifice.
Gas distributor
Sample size
S S S M M M M M M S S S
900 1661 2700 853 2424 2019 2554 3525 3099 3232 2443 2505
R. E. PAGE,
248 1.0
i
I) i al
.;
I
0
x
I :, ,
,
,
,
2 3 4 5 6 Bubble diameter (cm)
7
8
9
i,
, 1
,
,
Run 9 (Multi-orifice 0
distributor)
complete shortened
Fig. 4. SizeXdistribution of gas bubbles.
Bubble
diameter
km)
D. HARRISON
A check was first made to ensure that large enough samples had been taken. To do this, one of the longer runs (tape 9) was divided into two sections, the shorter being approximately as long as the shortest run (tape 4). The results for the bubble size distribution-expressed as the fractionfof bubbles with basal diameter greater than a given size-for the two parts were then compared. Figure 4 shows them to be in reasonable agreement, and therefore it was assumed that the sample sizes given in Table 1 were sufficiently large. As has been explained, results recorded by the few points in Fig. 4 were obtained from measurements on a large number of individual bubbles, and similarly for the results given in Fig. 5. However, there are fewer individual measurements per point on the curves in Fig. 3 which shows spatial distribution of bubbles in the column, and for this parameter it is probable that the sample sizes were not large enough for accuracy, especially near the column walls. The theoretical line shown in Fig. 3 was computed on the assumption of a uniform supply of gas per unit area to all parts of the distributor. The fraction of bubbles with diameter greater than a given size can be represented quite reasonably by a straight line on a log-linear plot, as is shown in Fig. 5. The experimental curves deviate from linearity for high and low values of II,,. For large IIt,, Hinze’ and Sherrard’ have suggested that there is a maximum stable bubble size, and Fig. 5 indicates there is some evidence of this in the sharp fall inf: For small D,, the deviation from a linear plot is because there are proportionally more small bubbles than medium-sized bubbles in a given experimental run. At the higher liquid velocity (1.43 cm/s), the bubble size distribution was independent of the gas flow rate (see Fig. 5) and also of the design of the gas distributor. At the lower liquid velocity (0.85 cm/s) the gas distributor again exerted no influence, but Fig. 5 shows there is a marked decrease in the bubble size with an increase in gas flow rate. Nevertheless, Fig. 5 shows that the bubble size is always smaller for the high liquid velocity than for the low.
DISCUSSION
fqg%Jq Fig. 5. Size distribution
Powder Tech&
of gas bubbles.
6 (1972)
The bubble size distribution to be found leaving a three-phase fluidized bed appears from observation to arise from the competing effects of bubble splitting and bubble coalescence. The physical me-
SIZE DISTRIBUTION
chanisms of these phenomena are not well understood and their detailed analysis is far from available. Nevertheless, in qualitative terms, we may list several related factors which are known to be important: (a) Hinze’ has shown that bubbles can be split by turbulent eddies, and therefore an increase in the overall agitation of a three-phase fluidized system -by increase of gas flow rate, for instance-would be expected to lead to some decrease in bubble size. (b) The rate of coalescence of bubbles is higher in liquids of higher viscosity’, and the apparent viscosity of a three-phase fluidized bed falls with increase of liquid velocity. (c) Bubbles can be observed to be split by single particles, and possibly by groups of particles. With this background, we may expect (and find in Fig. 5) that at the lower liquid velocity (and higher apparent viscosity) there is a fall in bubble size with the increased agitation of the system that is brought about by an increase in the gas flow rate. On the other hand, at the higher liquid velocity (and lower apparent viscosity), we may expect (and find in Fig. 5) the smaller bubbles which would be a consequence of smaller rates of bubble coalescence. It is clear, however, that particular situations could be influenced to an unknown extent towards smaller bubble sizes by the break-up of bubbles by the particles themselves. This may occur more readily at higher liquid velocities. If the bubble size distribution, particularly at the bed surface, is primarily a function of the competition between bubble splitting and bubble coalescence, then we would not expect it to be influenced by the form of the gas distributor. Experiment supports this conclusion for bubbles erupting from the bed. Bubbles leaving the surface of a three-phase fluidized bed carry particles into the freeboard, and experimental data on the bubble size distribution given in this paper have been used” to estimate the amount of particle entrainment from such a system.
Powder Tech&.,
6 (1972)
249
OF GAS BUBBLES CONCLUSIONS
It has been found experimentally for gas bubbles leaving the surface of a three-phase fluidized bed that (a) The bubble size is reduced by increase in liquid velocity. (b) At low liquid velocities the size of gas bubbles is reduced by an increase in gas flow rate, but as the liquid velocity is increased the gas bubble size becomes independent of gas flow rate. (c) Away from the gas distributor, the bubble size distribution is independent of the distributor design. ACKNOWLEDGEMENT
One of us (R.E.P.) would like to thank the Trustees of the Esso Fund for financial support during the course of this research work. LIST OF SYMBOLS
bubble basal diameter, cm fraction of bubbles with diameter greater than D, U,r minimum fluidizing (liquid) velocity, cm/s REFERENCES 1 K. P)stergaard, Advan. Chem. Eng., Vol. 7, Academic Press, New York, 1968; and Fluidization (eds. J. F. Davidson and
D. Harrison), Ch. 18, Academic Press, London, 1971. 2 L. Massimilla, A. Solimando and E. Squillace, Bit. Chem. Eng., 6 (1961) 232. 3 K. Bstergaard, Chem. Eng. Sci., 21 (1966) 470. 4 J. C. Lee, Discussion, Proc. 3rd European Symp. Chem. Reaction Eng., Amsterdam, 1965, Pergamon, Oxford, p. 211. 5 A. J. Sherratd, Three-phase fluidised beds, Ph. D. dissertation, Univ. of Swansea, 1966. 6 P. S. B. Stewart, Fluidization, Some hydrodynamic studies, Ph. D. dissertation, Univ. of Cambridge, 1965. 7 R S. Williams, Personal communication, 1970. 8 J. 0. Hinze, Am. Inst. Chem. Engrs. J., 1 (1955) 289. 9 P. H. Calderbank, M. B. Moo-Young and R. Bibby, Proc. 3rd European Symp. Chem. Reaction 1965, Pergamon, Oxford p. 91.
Eng.,
Amsterdam,
10 R. E. Page, Some aspects of three-phase lluidization, Ph. D. dissertation, Univ. of Cambridge, 1970.