Abstract
A draft tube spouted bed was constructed with a screw conveyor attached at its base to feed particles into the column. Results on fluid dynamic characteristics and particle movement in this system are presented and discussed. Two methods of measuring the superficial air velocity in the annular region are compared. The particle velocity and recirculation rates have been determined in a half column with transparent walls. The effects of the particle feed rate, air flow rate and bed height on the spouted bed dynamics have been analysed and compared with those in the literature. Keywords: Spouted bed, continuous feed, dynamics
EXPERIMENTAL STUDY ON THE DYNAMICS OF A SPOUTED BED WITH PARTICLE FEED THROUGH THE BASE
L.A.P. Freitas 1 and J.T. Freire 2
1Faculdade de Ciências Farmacêuticas de Ribeirão Preto/ USP - Via do Café s/n - Ribeirão Preto - SP,
14040-903 Brazil. Fax: 55-16-633-1092 and E-mail: lapdfrei@usp.br
2Departamento de Engenharia Química/ UFSCar - Via Washington Luis, Km 235 - São Carlos - SP,
13565-905 Brazil.
(Received: March 3, 1997; Accepted: June 18, 1997)
Abstract: A draft tube spouted bed was constructed with a screw conveyor attached at its base to feed particles into the column. Results on fluid dynamic characteristics and particle movement in this system are presented and discussed. Two methods of measuring the superficial air velocity in the annular region are compared. The particle velocity and recirculation rates have been determined in a half column with transparent walls. The effects of the particle feed rate, air flow rate and bed height on the spouted bed dynamics have been analysed and compared with those in the literature.
Keywords: Spouted bed, continuous feed, dynamics.
INTRODUCTION
Spouted beds with a draft tube have been largely used for drying grains because of its feasibility in processing a continuous operation. From literature, particles are continuously feed into the column at the top of the equipment. Ponte and Freire (1989) have demonstrated that this type of particle feeding lead to variations of bed temperature and humidity in the radial and angular directions, making scale-up a difficult task. As an alternative Mannurung (1964) proposed the particle feeding through the column nozzle together with the air injection. However, this type of particle feeding design has been scarcely used, probably due to difficulties in setting up and operating the equipment.
Particles can be fed into the inlet air nozzle by gravity, as used by Jesus (1987) in a pneumatic transport, or by screw conveying, as proposed by Mannurung (1964). According to Gomide (1983) it is necessary to slope the screw conveyor to maintain a constant solids flow rate during the spouted bed operation. In this way, the conveyor operates at its full capacity and the solids flow rate becomes dependent on the conveyor speed only.
The goal of this work were to construct and analyse the dynamics of a draft tube cylindrical conical spouted bed with a screw conveyor to feed particles at the column bottom. In the experimental procedure, two different methods of measuring air velocities in the annular and in the spout regions have been developed. The first method consisted in calculating the annular air velocities by using the Pitot tube calibration curve obtained in a loose bed of the particles (Lim and Grace, 1987). The second method is an iterative one between the local dynamic and static pressures. The results obtained from these two methods have been compared with one to another and with those based on the air mass balance in the inlet and spout regions. In addition, the solids recirculation rate and velocity in the spout have been determined in a half spouted bed column. The effects of the inlet air flow rate, the solids feed rate and the bed height on air distribution and solids recirculation have been studied. Experimental data obtained here are also compared with those predicted by empirical equations from the literature.
MATERIALS AND METHODS
Materials
Table 1 shows the properties of the porous medium and the particles used in this work. The particle diameter was determined by sieving. The solid and bulk densities were calculated by picnometry. The static bed voidage was measured in a 2L beaker. The minimum fluidizing velocity was obtained in a loose bed of the glass particles, according to the classical procedure proposed by Davidson and Harrison (1971). Figure 1 presents the of the cylindrical column used in the experiments for calculating Umf, D P/L versus U curves and the Pitot tube calibration curves. The internal friction angle of the particles and the friction between the particles and the column walls
were determined in a Dawes cell, following the procedure proposed by Brown (1970). Table 2 presents a summary of the experimental conditions selected for this work.
A diagram of the spouted bed equipment used in this work can be seen in Figure 2. A draft tube with an internal diameter of 0.05 m was inserted into the column at 0.07 m from the air inlet. The length of this tube was 0.18 or 0.28 m. Figure 2 shows also the solids feed system consisting of an inclined conveyor screw attached to the bed air inlet tube. The screw conveyor was connected to an electric motor with adjustable speed, enabling variations in the particle flow rate. Two static pressure probes were installed along the conveyor system to obtain the relationship of pressure drop against the air flow rate in the screw conveyor. This relationship was obtained when the entrance of air in the column was closed by a baffle.
Figure 1: Diagram of the loose bed used.
Table 1: Characteristics of the porous medium used in this work and the constituent particles
During the spouted bed operation the particles were removed from the bed through four slits in the cylindrical column wall at positions diametrically opposed in the top of the annular region. The Pitot probe was composed by two tubes with an internal diameter of 0.183 cm and a length of 83 cm; the static pressure tube had a radial hole with a diameter of 0.15 cm.
The half spouted bed column was constructed with the same dimensions as the whole column, but with a cylindrical section height of 0.55 m. The flat wall was made of transparent acrylic. Two slits placed at the same height corresponding ones in the whole column drained the solids during operation
Methods
Spouted Bed Operation
After filling the column with particles up to a selected height, air was injected into the bed. By a control valve, the air flow rate was increased up to the previously established value. As soon as the bed started spouting, the screw conveyor was turned on. The solids feed rate was determined by measuring the mass of particles removed from the bed. The particles were collected in 2L plastic beakers during two minutes and then weighed. If necessary, the speed of the screw conveyor was changed and the procedure repeated until the desired flow rate was attained.
In the whole column, when the air outlet temperature was at steady state condition, the static and dynamic pressures along the bed were determined at every 0.05 m point in the axial direction and at every 0.02 m point in the radial direction. Inclined water column tube manometers with errors of 0.5 mm and 0.2 mm, were used.
In the half column, after stabilizing the bed temperature, the time of white tracers traveling a distance of 0.1 m (marked on the flat wall) in the annular cylindrical region was measured. For the solids flow rate measurements in the spout or inside the draft tube, a ruler was fixed on the flat wall and tracer particle movement in the spout was recorded with a JVC camcorder at a normal speed of 1/64 s.
Fluid Flow Rate
The Pitot tube calibration method consisted of determining the relationship Pd versus U in a loose bed of the particles (Figure 1). A Venturi flow meter was used in the air superficial velocity measurements. Assuming that the calibration curve was representative of the relationship Pd versus U for the annular region, the air superficial velocity can be calculated by the measured data of Pd .
The iterative method consisted of taking two initial estimations of the air velocity in the annular region, the first based on the direct Pitot tube measurement of dynamic pressures. The second initial estimation was based on the static pressure data and the DPe/L versus U curve obtained for the loose bed. The values of the velocities were compared. After, the velocity estimations were used for re-calculating the dynamic and static pressure reciprocally, i.e., the velocity determined with Pitot tube reading was used for calculating the static pressure. The re-calculated pressures were used for estimating the new velocities that were compared again. The procedure was repeated until the convergence of the two velocities.
Solids Movement
The tracer measurements obtained in the half column were used to calculate the particle velocity. The solids recirculation rate was then obtained by the following relationship:
Wr = r a. Va. Aa. (1-e a)
The recording of the particles movement in the spout was examined picture by picture in a video cassette. The particle velocity in the spout were determined by the distance traveled by the tracers from one picture to another and by the time interval between these pictures (1/30s).
RESULTS AND DISCUSSION
Calibration of the Pitot tube and D Pe x U
Data obtained of D Pe x U obtained in the loose bed were fitted to a quadratic equation with correlation coefficient R2 = 0.9998, as follows :
Data from the Pitot tube calibration test were fitted by the following equation:
with correlation coefficient R2 = 0.9969. The quadratic relationship of the fit between pressure and velocity provides an evidence that this method is suitable for calibrating the instrument. This result validates the use of this technique in further work.
In regard to flow rate versus pressure drop data obtained in the screw conveyor, it was shown that the amount of air deviated by the screw was negligible compared to the flow rate in the column. During the spouted bed operation, the pressure drop along the conveyor remained between 1.5 and 1.8 cm H2O, indicating flow rates around 0.02 m3/min.
Fluid Dynamics of the Spouted Bed
Regarding the fluid dynamics of the spouted bed, the procedure for the analysis of data obtained in the central and annular regions was not the same, mainly due to the differences of bed porosity in these two regions. Thus, for the central region the air flow rates were directly calculated from the dynamic pressure measurements.
Figures 3 and 4 show the axial profiles for the static pressures in the annular region for bed heights of 35 and 25 cm, respectively. The concavity of these curves are opposed to those encountered in the literature for conventional spouted beds. However, this is in agreement with the results reported by Claflin and Fane (1983) for a draft tube spouted bed. One can imagine several consequences of this behavior, among which is the impossibility of using models and empirical correlations developed for conventional systems in the modelling of a draft tube possessing bed (Epstein et al., 1978). Pressure measurements taken in the cylindrical part of the annulus showed that the pressure is almost uniform in the radial direction and the pressure drop is constant in the axial direction. This confirmed the plug flow in this region, due to the presence of the draft tube.
The axial pressure gradient data obtained for the cylindrical part were used to calculate the superficial velocity by the iterative method. In the conical part, the pressure gradient is not constant, and its values were calculated from the fitted Pe versus Z curves. The annular air flow rates for a bed height of 25 cm, estimated by both the Pitot calibration and iterative methods, are presented in Table 3. Deviations between these two estimated values are no more than 10%. These differences should be attributed to the experimental pressure measurements errors rather than to the methods themselves.
On analyzing the iterative method, it can be observed that it consists in finding the root of a two equation system - D P x U relationships - given an initial range (values of experimental dynamic and static pressures). It is of course possible that there is more than one root in the given range. If there is more than one root, which will be encountered will depend exclusively on the convergence method used. In our case we chose the geometrical mean as a correction criterion of the pressures, based on the quadratic relationship between pressures and velocity.
Table 3 also shows the spout air superficial velocities obtained with a Pitot tube, as well as the comparisons between the sum of the air flow rates in the spout and in the annulus (for the two methods), and the real air flow rate entering into the bed measured in the Venturi flow meter. The highest value of deviation was of 6.7%, showing that the results obtained by the two methods employed are in agreement. Data in Table 3 shows that the solids feed rate do not affect the air flow distribution in the bed.
Figure 5 shows the percentage of air flow rate passing through the annulus as a function of bed height. The percentages were based on the mean of Qa, obtained for the different solid rates by the iterative method. The values of the percentage of air flow rate that pass through the annulus obtained in this work agree well with those reported by Claflin and Fane (1983) and Khoe and Van Brackel (1983), who encountered values of approximately 30%. The data in Figure 5 show that the amount of air flowing through the annulus tends to increase with the increase in bed height. Ijichi et al. (1994) found that the higher the load of particles in the bed, the lower the fraction of air passing through the annuls. However, the spouted bed system used by these authors differs significantly from the one used in this work, basically in the column and particle dimensions. The amounts of air passing through the annulus are below those usually encountered in the literature. They were calculated by mass balance between the spout air flow rate and the total flow air rate into the column. Determining the annular air flow rate by measurements of the local pressure gradient at the wall of the whole and half columns, Claflin and Fane (1983) obtained an opposite behavior of the air flow distribution in a bed with dimensions similar to the one used here. On studying the drying of rice, Khoe and Van Brackel (1983) found that increasing the load of particles does not alter the air distribution between the two regions. A reasonable explanation for the increase in annular flow rate with the increase in bed height is the increase in particle weight in the annulus. This results in a higher recirculation rate, which increases the particle load in the spout and hence increasing the resistance to air flow through this region. However, the conflicting results in the literature show that under different operational conditions the effect of a higher particle load in the spout may possibly be compensated by an increase in the resistance to air flow through the annulus, due to the augment of the travel distance through this porous medium.
The empirical parameters of the equations of Yokogawa et al. (1972) and Mamuro and Hattori (1968) were calculated with the aid of the experimental data of Umf, Hm and air superficial velocity for z = 0.2 m. On fitting the data to the Yokogawa et al. (1972) equation, a value of 2.7 cm-1 was obtained for the K.kp/Cf group; this was very different from that encountered by Epstein et al. (1978) on fitting this model to the data of Thorley et al. (1959), which was equal to 0.059 cm-1.
In Figure 6, the annular air flow rates predicted by the empirical correlations of Yokogawa et al. (1972) and Mamuro and Hattori (1967) are compared with the experimental data of this work. As can be seen in this Figure, the annular air flow rates are underestimated by both equations. The predicted a value of 0.25 m3/min at the draft tube inlet (z = 0.07 m) is low compared with the experimental ones of 0.85 and 0.90 m3/min for bed heights of 0.25 and 0.35m, respectively.
Figure 3: Static pressure measured in the annulus as a function of z, for H = 35 cm and Qe = 2.1 m3/min.
Figure 4: Static pressure measured in the annulus as a function of z, for H = 25 cm and Qe = 2.1 m3/min.
Figure 5: Effect of the flow rate of the air entering the bed on the spout-annulus air distribution.
Figure 6: Predictions of literature correlations versus experimental data for Qe = 2.1 m3/min.
Recirculation Rates and Particle Velocity in the Annulus
Figures 7 and 8 show the effect of the air flow rate on the particle velocity in the cylindrical part of the annulus for the range of solids feed rate studied. As can be noted, the increase in the air flow rate results in a slight increase in the annular particle velocities. The same effect is observed with the increase in solids feed rate. These results agree with those reported by Thorley et al. (1959), Claflin and Fane (1983), Yang and Keairns (1983) and Silva (1987).
From Figures 7 and 8, it can be observed that the particle velocity tends to an asymptotic value, or maximum, as the air flow rate increases. This behavior was not observed by Yang and Keairns (1983), whose data presented a linear increase in particle recirculation rate with the increase in air flow rate. On the other hand, upon analyzing the effect of the air flow rate on the recirculation rates in the bed, Berruti et al. (1988) noted that their data showed a maximum in recirculation. These authors explained the existence of this maximum recirculation rate as a consequence of an alteration in the spout-annulus air flow distribution. A more viable explanation would be to attribute this effect to the porous medium interactions, i.e., the internal friction and friction with the walls.
Figures 7 and 8 shows that the increase in solid feed rate causes a decrease in particle velocity in the annulus. This could be explained by the additional resistance to the annulus-spout cross flow due to the increase in particle feed rate, i.e., the increase in the collisions of the particles which are accelerated from the screw conveyor with the spout-annulus interface represents an additional resistance to cross flow.
The annular particle velocity data were used to calculate the solids recirculation rate. The values of Wr obtained were fitted to an empirical equation as a function of the variables studied as follows:
Wr = -0.12 - 0.935.We + 0.692.Ho + 9.913.Qe
The correlation coefficient was R2 = 0.9538 and holds for 2.1 £ Qe £ 2.8 m3/min; 1100 £ We £ 2200 g/min; 25 £ Ho £ 35 cm.
Figure 9 shows the experimental data versus the predictions of the equation, showing that the equation fitted the data within a deviation always between ± 10%. In the fitted equation, Wr and We are in kg/s, Ho in m, and Qe in m3/s. Analysis of variance showed that 90% of the variation in Wr can be explained solely by the effect of the inlet air flow rate Qe.
Most models in the literature on the recirculation rate in spouted beds are based on the particle velocity near the walls. Although some authors state that for conventional spouted beds, the radial variation in solids velocity is significant, most empirical correlations take this effect into account by using the average value of the radial velocity for each axial position. In the spouted bed with a draft tube this effect can be neglected, since, as noted in this work, the solids movement in the annulus approaches a plug flow, with only little variation near the walls. Among the existing correlations, that of Thorley et al. (1959):
is noteworthy for conventional beds with Wr in kg/s, q = 60° , 30 < H < 183 cm, and where : K = 0.563 for dc = 61.0 cm, and K = 0.068 for dc = 15.2 cm, while that of Claflin and Fane (1984), which is the only one for draft tubed beds, is as follows:
Figure 10 shows a comparison between the data obtained in this work and the predictions of the correlation cited. The predictions of the correlations of Thorley et al. (1959) were compared with the experimental data solely on an illustrative basis, since this correlation has not been developed to predict recirculation in draft tubed beds. As can be seen in the figure, the predictions of both correlations gave values below those obtained in this work. The underestimated values of the Claflin and Fane (1984) equation can be explained by the fact that they used wheat in their experiments, as the bed dimensions used by them are very similar to ours. As the descending movement of the particles depends on the normal load on them, it is expected that for greater bed heights the recirculation rates are also higher, and that the same effect is caused by a higher particle density. The density of the glass used in this work is twice that of the wheat used by Claflin and Fane (1984).
Figure 7: Solids velocity in the annulus as a function of the inlet air flow rate (Ho = 25 cm).
Figure 8: Solids velocity in the annulus as a function of the inlet air flow rate (Ho = 35 cm).
Figure 9: Experimental and predicted values of the recirculation rate.
Figure 10: Recirculation rate. Comparison between experimental data and the predictions of equations from the literature.
Particle Velocity in the Spout
As described previously, the velocities were determined by recording at normal speed and accompanying the distance traveled between two pictures. It was observed that determination of the radial velocity distribution in this region was difficult by any method, for despite the predominant axial movement there is also a radial component in most cases. For example, a particle between two pictures travels approximately 5 cm in the axial direction, but may move up to 1 cm in the radial direction, which would already be significant if it were desirable to measure the radial velocity distribution. Despite this, the method was shown to be adequate in determining average velocities when a sufficient number of repetitions was used.
Figures 11 and 12 present the solid velocities in the central region for the bed heights of 35 and 25 cm, respectively. It is noted that the velocity increases with the total air flow rate in the bed and with the solids feed rate. The effect of the air flow rate is due to the higher drag between solids and gas, and the effect of the particle feed can be explained by the higher number of particles accelerated from the feed section, affecting the average velocity.
The nominal values obtained for Vj are in good agreement with those for conventional beds obtained by several authors, such as Morgan et al. (1983), Susciu and Patrascu (1977) and Waldie and Wilkinson (1986). Unfortunately none of the work encountered on the draft tubed bed dynamics gives information on the particle velocity in the spout. A comparison of the values obtained by these authors and by us can be seen in Table 4.
Figure 11: Solids velocity in the spout versus the air rate for different solids feed rates (Ho =35 cm).
Figure 12: Solids velocity in the spout versus the air rate for different solids feed rates (Ho =25 cm).
CONCLUSIONS
The results of this work show that the conical-cylindrical spouted bed with a draft tube and a particle feeding through the base is indicated for continuous processing. Preliminary studies confirmed that axial symmetry was maintained in the system, which is usefull for scaling up.
In fluid dynamics similar results have been obtained for the air flow distribution in the bed using three techniques: 1) mass balance with the spout air velocity measured by a Pitot tube, 2) Pitot calibration for the porous medium in a fixed bed to determine annulus air velocity, and 3) the iterative method of the local pressures to determine the annulus air velocity. Techniques 2 and 3 showed promise but need further investigation for their validation.
As to the air distribution in the bed, the values agree with those of other authors and a slight increase in flow rate through the annulus was observed with increase in bed height. Despite the fact that several authors have been using the equations of Mamuro and Hattori (1968) and Yokogawa et al. (1972) for their simulations of draft tubed beds, comparison with our data showed that these equations underestimate the flow rate through the annulus.
The recirculation rates obtained from the solids velocities in the cylindrical part of the annulus showed that the rate increases with the inlet air flow rate, but decreases with the particle feed rate through the base. The recirculation rate results are in agreement with data in the literature.
The technique used to determine the solids velocity in the spout is appropriate for obtaining approximate or average velocity values inside the draft tube. The values obtained showed that the velocity increases with both air inlet flow rate and particle feed rate. Comparatively the values obtained in this work agree with those obtained by other authors.
NOMENCLATURE
A Area, m2
D Diameter, m
E Distance between the draft tube and the air inlet, m
H Bed height, m
Le Draft tube position, m
M Weight of solids in the bed, kg
P Pressure, cm H2O
Q Volumetric flow rate, m3/min
U Superficial air rate, m/s
V Solids velocity, m/s
W Mass flow rate of the particles, g/min
Subscripts
a Annulus
ap Apparent
e Entrance
d Dynamic
f Fluid
j Spout
m Maximum of stable spout
mf Minimum fluidization
o Bed inlet
p Particle
r Recirculation
t Total
Greek Symbols
e Porosity
g Friction angle with the walls, degrees
r Specific mass, g/cm3
q Internal friction angle, degrees
REFERENCES
Berruti, F.; Muir, T.R. and Behie, L.A., Solids Circulation in a Spout-Fluid Bed with Draft Tube, Can. J. Chem. Eng., 66: 919-23 (1988).
Brown, R.L. and Richards, J.C., Principles of Powder Mechanics. Pergamon Press, 221pp. (1970).
Claflin, J.K. and Fane, A.G., Spouting with a Porous Draft Tube. Can. J. Chem. Eng., 61: 356-63 (1983).
Claflin, J.K. and Fane, A.G., Gas Distribution and Heat Transfer in a Draft Tube Spouted bed. AIChE Symp. Ser., 80 (241): 17-23 (1984).
Davidson, J.F. and Harrison, D., Fluidization. Academic Press, London (1971).
Epstein, N.; Lim, C.J. and Mathur, K.B., Data and Models for Flow Distribution and Pressure Drop in Spouted Beds. Can. J. Chem. Eng., 56: 436-47 (1978).
Gomide, R., Operações Unitárias. Vol. 1 - Operações com Sólidos Granulares. 293pp. Author edition (1983).
Ijichi, K.; Tanaka, Y.; Uemura, Y.; Hatate, Y. and Yoshida, K., Solids Circulation Rate and Holdup in the Draft Tube of a Spouted Bed. Int. Chem. Eng., 34 (3): 370-76 (1994).
Jesus, J.M., Transferência de Calor Gás-Sólido no Transporte Pneumático. MSc Thesis, Universidade Federal de São Carlos, São Carlos, Brazil, 85pp. (1987).
Khoe, G.K. and Van Brackel, J., Drying Characteristics of a Draft Tube Spouted Bed. Can J. Chem Eng., 61: 411-18 (1983).
Lim, C.J. and Grace, J.R., Spouted Bed Hydrodynamics in a 0.91 m Diameter Vessel. Can. J. Chem. Eng. , 65: 366-72 (1987).
Mamuro, T. and Hattori, H., Flow Pattern of Fluid in Spouted Beds. J. Chem. Eng. Jap, 1(1): 1-5 (1968).
Mannurung, F., Studies in the Spouted Bed Technique with particular reference to low Temperature Coal Carbonization. PhD Thesis. Univ. of New South Wales, Kensington, Australia (1964).
Morgan, M.H.III; Day, J.Y. and Littman, H., Spout Voidage Distribution, Stability and Particle Circulation Rates in Spouted Beds of Coarse Particles I: Theory. Chem. Eng. Sci., 40(8): 1367-77 (1983).
Ponte, M.J.J.S. and Freire, J.T., Determinação Experimental da Area de Troca Térmica Fluido-Partícula em Leito de Jorro com Alimentação Contínua de Sólidos. Anais do XVII ENEMP, 1: 149-62 (1989).
Silva, E.L., Transferência de Calor Parede-Leito em Sistemas de Jorro Convencional e com Tubo Interno. MSc Thesis. Universidade Federal de São Carlos, São Carlos, Brazil (1987).
Stoker, R.K.; Eng, J.H. and Behie, L.A., Hydrodynamics and Thermal Modelling of a High Temperature Spouted Bed Reactor with Draft Tube. Can. J. Chem. Eng., 68: 302-11 (1990).
Susciu, G.C. and Patrascu, M.H., Phase Distribution and Residence Time in a Spouted Bed. AIChE J., 23(3): 312-18 (1977).
Thorley, B.; Saudby, J.B.; Mathur, K.B. and Osberg, G.L., An Analysis of Air and Solid Flow in a Spouted Wheat Bed. Can. J. Chem. Eng., 37: 184-92 (1959).
Yang, W.C. and Keairns, D.L., Studies on the Solid Circulation Rate and Gas Bypassing in Spouted Fluid Bed with Draft Tube, Can. J. Chem. Eng., 61: 349-55 (1983).
Yokogawa, A.; Ogino, E. and Yoshi, N., Flow Pattern of Fluid in the Annulus of the Spouted Bed. Trans. Japan Soc. Mech. Eng., 38: 148-57 (1972).
Waldie, B. and Wilkinson, D., Measurement of Particle Movement in a Spouted Bed using a new Microprocessor based Technique. Can. J. Chem. Eng., 64: 944-49 (1986).
Publication Dates
-
Publication in this collection
09 Oct 1998 -
Date of issue
Sept 1997
History
-
Accepted
18 June 1997 -
Received
03 Mar 1997
