Radial Dispersion and Flow Distribution of Gas in Magnetically

The radial dispersion was found to be signlflcantly lower in MSBs than in normal fluidized beds and higher than the values for packed beds. Velocity f...
6 downloads 0 Views 815KB Size
Ind. Eng. Chem. Process Des. Dev. 1982, 27, 135-141

135

Radial Dispersion and Flow Distribution of Gas in Magnetically Stabilized Beds Jeffrey H. Slegell Corporate Research Science Laboratories, EXXON Research & Engineering Company, Linden, New Jersey 07036

The effects of magnetic field, solids circulation, path to stabilization, and particle size on radial dispersion of gas were studied in magnetically stabiiized beds (MSB). Fluid flow distributions through MSBs have been obtained by point measurements along a diameter and circumference using hot wire anemometry. Point to point velocity fluctuations were used to calculate a mean value for each radial position. The radial dispersion was found to be signlflcantly lower in MSBs than in normal fluidized beds and higher than the values for packed beds. Velocity fluctuations in an MSB were found to be similar to those reported in the literature for packed beds. Mean flow distributions for the MSB changed from parabolic-like at low flow rates and magnetic fields to profiles with increasing flow near the tube walls at high flow rates and fields.

Introduction A number of workers have studied the influence of magnetization on the dynamics of fluidized solids. Early accounts of this phenomenon were reported by Filippov (1960,1961) and Kirko and Filippov (1960). Subsequent workers have reported on the influence that magnetization exerts on pulsations, heat transfer, structure, and other characteristics of magnetized and fluidized solids. A review of some of this work is given by Bologa and Syutkin (1977). Recently, Rosensweig (1979a, 1979b), Rosensweig et al. (19791, and Lucchesi et al. (1979) have reported on a number of features of flowable magnetically stabilized fluidized solids and a systematic interpretation of the phenomenon. They reported on the expanded fluidlike state of the magnetically stabilized fluidized bed (MSB) which results when a uniform magnetic field is applied to a bed of magnetizable solids in a direction collinear with the direction of the fluidizing gas flow. This magnetic stabilization produces an expanded flowable regime. Figure 1 shows the operating regimes for fluid/solid/ magnetic systems. Below the normal minimum fluidization velocity the bed is packed or nonfluidized. Above this velocity the bed may be in either the stable or unstable, bubbling, regime depending upon whether or not the transition velocity is exceeded. The stably fluidized solids resemble a liquid and as such enjoy the benefits that the solids are facilitated for transport. Since the bed is fluidized the pressure drop is independent of particle size or fluid throughput. In addition the bed does not show any gas or solids backmixing which normally occurs in bubbling fluidized beds. The radial dispersion and degree of flow uniformity in MSBs is important in determining their applicability as chemical reactors or contactors for physical processes. Bed nonuniformities may cause channeling and flow maldistribution which can be detrimental to the process. In order to study the above, radial dispersion data have been taken to compare MSBs to packed and bubbling fluidized beds. Experiments have also been made to measure the fluid flow profiles through an MSB. Equipment The equipment used in this study consisted of a 7.62 cm diameter Plexiglas vessel, a fluidizing gas control system, a tracer injection and measurement system, and a hot wire anemometer to measure the gas velocity. Surrounding the Plexiglas vessel were two solenoidal electromagnets used to produce an applied axial magnetic field. Each solenoid 0 196-43051821112 1-0 135$0 1.2510

consisted of approximately 1110 turns of insulated no. 12 copper wire, 15.25 cm inside diameter and 10 cm high. The resistance across each solenoid was about 3.78 ohms. They were connected in series and placed 10 cm apart, one above the other. Radial dispersion was measured using a point input of tracer and measuring the concentration profile downstream, as originally described by Towle and Sherwood (1939). The tracer injection and measuring systems are shown in Figure 2. A copper tube 0.3175 cm outside diameter was used to introduce tracer to the bed. The tube entered from the top surface along the vessel wall and traveled along the wall to a point below the injection height. It then followed a horizontal path to the center of the bed and turned upward as a vertical section. The inside diameter was then increased to 0.3175 cm and the tube was partially packed with cotton to produce an even flow. The tube mechanism was constructed so the distance between the injection point and the bed surface was variable. The tracer detector system consisted of a 0.3175 cm outside diameter copper sampling tube connected to an Ion Tract Instruments (Waltham, MA) Leakmeter. This copper sampling tube was mounted on a Velmex (East Bloomfield, NY) Unislide assembly for easy traversing across the bed’s top surface. The tracer gas used was a mixture of 8.2 ppm of sulfur hexafluoride (SFJ in nitrogen from Matheson Gas Products (East Rutherford, NJ). The fluidizing gas was ambient air, and the particles used in the tests were magnetite from Foote Mineral Co. (Exton, PA), and steel spheres. The hot wire anemometer was a ALNOR thermal anemometer type 8500. It used an approximately 0.32 cm diameter wire to measure the gas velocity, giving a “point” value. Experimental Method The solids were sieved to the desired size ranges and then placed in the Plexiglas vessel. Preliminary testing often included determination of transition to bubbling and pressure drop characteristics. Before each experiment the bed was fluidized vigorously for several minutes without a magnetic field. This limited the residual effects of past experiments a t high magnetic fields. In preliminary testing for the radial dispersion experiments, the final bed height was determined for each set of conditions so that the quantity of solids in the bed and the tracer injector probe height could be preset before the 0 1981 American Chemical Society

138

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 I

I

NON-FLUIDIZED

MAGNETICA,LLY STABI LEED

MAGNETIC FIELD INTENSITY

Figure 1. Operating regimes phase diagram for the fluid/solid/ magnetic system. TRACER GAS I N L E T

:IO PPM

SF6 IN

BED

Figure 3. Experimental setup for flow distribution measurements.

N21

/ , /

Table I. Summary of Test Conditions superficial gas applied magnetic field, H A , velocity, U,, Oe cm/s 12.7 35 14.5 53 16.4 56 18.2 70 87 22.0 82 23.9 120 50.0

TO SF6 DETECTOR POSITION S C A L E

---MAGNETICALLY STABILIZED BED

INJECTOR V I E W OF

*\mj-j)xpANDED

42 If 0

Figure 2. Experimental method for radial dispersion measurements.

experiment started. The tracer injector was set to the desired sampling height below the final surface of the bed, and the sampling probe was set to traverse along the top surface of the bed. The vertical distance through which the tracer gas could diffuse radially was the distance between the tracer injector and detector tubes. In the radial dispersion experiments five different distances were used for each set of conditions: 5 , 8 , 11, 15, and 20 cm. Since the calculated radial dispersion coefficient is not a function of sampling distance this provided a test of data consistency. Two different paths to stabilization were investigated: fluidizing the bed in a bubbling mode and then applying the magnetic field (“magnetized last”), and applying the field to a stationary packed bed and then increasing the gas flow (“magnetized fiist”). Concentration of tracer gas was measured as a function of radial position at several field values in both the bubbling and stabilized modes. The quantity of gas tracer injected depended on the gas velocity within the bed and voidage. It was desired to match the tracer injector velocity to the interstitial gas velocity in the bed. After this velocity was calculated from gas flowrate and voidage measurements, tracer gas was injected into the bed. Experiments to determine the radial dispersion coefficients for a packed bed required the installation of an 80 US.Sieve brass screen atop the bed to retain it. The effect of the presence of the screen on the dispersion coefficient was assumed to be negligible. For the experiments in a

circulating MSB, the bed was initially fluidized and circulating and then the magnetic field applied. A continuous circulating countercurrent MSB was used to determine the effect of solids circulation on the radial dispersion. Solids were withdrawn from an outlet near the bottom of the bed and then pneumatically conveyed back up to the top surface of the bed. The solids were found to move countercurrently to the gas in plug flow fashion. After the concentration of tracer gas was determined as a function of radial position the radial dispersion coefficient was obtained by application of the following equation from Anderson et al. (1966)

The dispersion coefficient was obtained from a plot of the natural log of the tracer concentration as a function of the radial distance squared. The slope of this line is equal to -Vi/ 403. In order to study the symmetry of gas flow and also the point to point velocity distribution, anemometry profiles were made along a circumference at several radial positions. Figure 3 shows the experimental setup. The series of results at each radial position were averaged to determine the flow symmetry. This allowed comparison with flow profiles, from the literature, of packed beds made using an averaging circular anemometer. The field strength and the gas velocity were determined from transition data. The field was set at a value just above that needed to stabilize the bed at the velocity for each run. Values of the applied magnetic field ranged from about 30 to 120 Oe and are summarized in Table I for the gas velocities tested. Experimental Results Effect of Magnetic Field on Radial Dispersion. Typical curves illustrating the change in the radial con-

Ind. Eng. Chem. Process Des. Dev., Vol.

21, No. 1, 1982

I

a a NUMBER

0.4

137

a MAGNETICALLY STABILIZED B E D

0.2

0

2

4

6 8 1 REYNOLDS NUMBER

0

1

2

Re

dpU,P/r

Figure 6. Radial dispersion of gas in packed and magnetically stabilized beds of -60 +80 U.S.Sieve magnetite.

RADIAL POSITION I”

Figure 4. Effect of magnetic field stabilization on radial dispersion of tracer. 0 BATCH BEDS

i

1 .6L

0 6

(CM~/SKI

L“IC!

04 1 I

.

I

,. I ,

55

60

,

~

,

0.3

-1

A 0.7

2 I“)

Figure 5. Calculation method.

centration profile of a gas tracer as a bubbling bed of -60 +80 US. Sieve magnetite was stabilized in response to increasing magnetic field are illustrated in Figure 4. Here the curves a t 0 and 23 Oe show a fairly large degree of tracer dispersion while the curves a t 35 and 47 Oe show much less dispersion of tracer. The bed became stabilized as the field was increased between 23 and 35 Oe. Thus the amount of radial dispersion is significantly reduced by stabilization. A sample of the data work up is shown in Figure 5. It was found that once the bubbling fluidized bed was stabilized the radial dispersion coefficient remained nearly constant, independent of magnetic field intensity. Comparison of Radial Dispersion in MSBs and Packed Beds. Radial dispersion experiments were made for a packed bed of -60 +80 US.Sieve magnetite retained by the 80 U.S.Sieve brass screen. A comparison between a “magnetized last” MSB and this experimental packed bed is made in Figure 6. Note that the Peclet number is nearly the same a t low Reynolds numbers near the minimum fluidization velocity. As the gas velocity is increased, a difference in the Peclet numbers is noted. The radial dispersion for the packed bed is slightly less than that measured for the MSB once the beds are above U, Further study is necessary to determine if this can be attributed to a difference in bed structure. It is possible that the higher radial dispersion in a MSB, as compared to a packed bed, is due to its higher void volume. This is elucidated in the Discussion.

0.2

50

65

70

75

APPLIED MAGNETIC FIELD Ha

IOERSTEOI

Figure 7. Effect of solids circulation on radial dispersion in magnetically stabilized beds of -60 +80 U.S.Sieve magnetite.

Effect of Solids Circulation on Radial Dispersion. The effect of solids circulation on radial dispersion is shown in Figure 7 for MSBs of the -60 +80 U.S. Sieve magnetite which were “magnetized last” after the bed circulation was started. Before data were taken, the bed had been circulating for a sufficient time to allow for all of the bed material to be removed and recirculated. Note that the radial dispersion of gas almost doubles when the solids are circulated. Since solids are introduced along the top surface of the bed, and this is where the tracer profile is made, the movement of solids across the surface could have caused some of this increase. The radial dispersion coefficient, with the exception of one or two data points, was found to be nearly constant with magnetic field in the stabilized regime. Effect of Path to Stabilization on Radial Dispersion. In achieving a fluidized stabilized state, one may increase the gas velocity and then the magnetic field (“magnetized last”) or proceed in the reverse order, field first then gas flow (“magnetized first”). These two paths have not been found to produce significant differences in transition velocity. However, it was found that the “magnetized last” path produces MSBs with about one quarter the amount of radial dispersion as for the “magnetized first” path. This is probably due to the formation of vertical channels in the bed caused by bubbles when a magnetic field is applied to a bubbling bed. Such channels are absent when the field is applied and then the gas velocity increased at the velocities and magnetizations studied. A comparison of the two paths is made in Figure 8 for the -60 +BO U S . Sieve magnetite.

138

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982

,

,

I

,

---

28

t

i

*t

Figure 8. Effect of path to stabilization on radial dispersion in magnetically stabilized beds of magnetite.

U = 14 5 cm rec &=066

4-

0"'

I

60

"

120

I

I

180 I

4

240 I

"

300

1

1

I

360 J

C I R C U M F E R E N T I A L POSITION (DEGREES1 i

n i

\\\

Figure 10. Spatial velocity variations with circular position in a magnetically stabilized bed. 1

1

I

I

I

I

I

I

..

Figure 9. Effect of particle size on radial dispersion in magnetically stabilized beds.

5 1 0

16

J iy

In Figure 8 a comparison may also be made of the radial dispersion of gas in an MSB with that predicted in the literature for packed beds. The MSB data, which are for a Schmidt number of 1.5, are plotted with a curve from Himmelblau and Bischoff (1967) displaying the inverse of the Peclet number as a function of the Reynolds number. I t can be seen that for the case where the bed is bubbled first, the radial dispersion experimentally measured in an MSB is only slightly higher than that which would be expected from a packed bed. Effect of Particle Size on Radial Dispersion. The effects of particle size on radial dispersion in batch beds of magnetite and steel spheres that were "magnetized last" are presented in Figure 9. The results for the -120 +140 US.Sieve magnetite are in good agreement with the literature packed bed values; however, this result is based on very limited data taken with a tracer injector velocity ten times higher than the calculated interstitial velocity. Two size cuts of steel spheres were used, -45 +50 and -18 +20 U.S.Sieve. The results of these tests indicated radial dispersion data similar to that which would be expected in a packed bed for Reynolds numbers ranging from 0.5 to 200. Fluid Flow Distribution in MSBs. As previously described, in order to determine the symmetry of gas flow through the MSB, anemometry profiles were made along a circumference at several radial positions using MSBs of -60 +80 U.S.Sieve magnetite. All the fluid flow distribution runs were made for beds which were "magnetized first". Figure 10 presents typical data showing the spatial velocity variations as a function of circumferential position. Note that the flow is not uniform, but it shows substantial variations with position. It should be noted that if the experiment could be repeated with some of the top of the bed removed, the same degree of fluctuations would be expected but their positions would probably be altered. Later on, it will be shown that these profiles are typical of those reported in the literature for packed beds. Using a series of profiles at different radial positions, the flow pattern of gas through the MSB can be determined. Figure 11 shows such a profile, one of the data points of which was obtained from the average value of the

>

0 - 2 .

w U

0

0 0

4

1

U = 14 5 cm!rec

DISTANCE FROM T U B E W A L L (CMI

Figure 11. Gas flow pattern through a magnetically stabilized bed. I

1801

,

,

1 -4

U = 50 cm iec

1 R A D I A L POSiTIOh ICM'

Figure 12. Spatial velocity variations with radial position in a magnetically stabilized bed.

data in Figure 10. For the conditions of this test the gas flows uniformly through the bed over the area tested. No region was found which had a consistently higher throughput of gas than any other. Since the data plotted in Figure 11 are averages they show only the average flow of gas at a specific radial position. Making an anemometer traverse along a diameter reveals that, as expected, there are spatial variations in velocity similar to those in Figure 10. Figure 12 shows the profile that results from such a traverse. The change in gas flow pattern, through the MSB, as the gas velocity is increased is shown in Figure 13. Note

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 139

l

I

I

1

I

I

I

3 4 32

5

1

I

28

-

t

i.066

28-

W

i?

a

2.

24-

s9

20-

t

16-

A

A -

A

A

A

-

U

0

120

SUPERFICIAL GAS VELOCITY

8-

012.7 cmlrec 0145 A164 0182 m23.9

4-

0-

I

1

I

I

2

3

E ) ,

0

,

, , 60

, , ~,

, , , , , , , , ,

120

180

240

300

360

CIRCUMFERENTIAL POSITION IDEGREESl I

DISTANCE FROM TUBE WALL (CM1

Figure 13. Gas flow patterns through magnetically stabilized beds at different velocities.

that a t a low gas velocity, 12.7 cm/s, the profile appears to be parabolic-like, with a lower gas flow rate near the tube wall. At the intermediate flow rates of 14.5 and 16.4 cm/s the profile appears almost flat, with uniform flow through the bed. At the higher gas flow rates, the flow appears to become nonuniform with a region of high gas throughput near the wall. It will be shown later that this type of flow pattern, with higher flow in a region near the wall, is typical of packed bed behavior. The spatial velocity variations in the MSB have been found to be greater in magnitude for higher gas velocities. Figure 14 shows a comparison of the profiles for gas velocities of 12.7 and 22 cm/s. Note that for the higher velocity the spatial changes in gas flowrate seem to be both more numerous and of higher magnitude. It seems that the MSB structure becomes less uniform as the gas velocity and magnetic field is increased.

Discussion High rates of fluid radial dispersion are usually desirable to promote better fluid solid contacting and to provide a radially uniform reaction front passing through the vessel. Since MSBs have slightly more radial dispersion than packed beds, we can probably expect better contacting and more uniform flow of fluid in MSBs. This difference between the MSBs and packed beds is probably due to a difference in bed structure. The MSB is expanded and has a higher void volume than does a packed bed. As a result, the size of a fluid mixing cell is expected to be larger in an MSB than in a packed bed. One might expect more mixing of fluid for a circulating compared to a batch MSB because solids continuously being removed from the bottom of the bed and added to the top would allow a more random bed structure. This more random structure may produce larger mixing cells. One reason why applying the magnetic field before increasing the gas flow produces MSBs with about fourfold more radial dispersion than applying the magnetic field to a bubbling bed then increasing the gas flow may be that in the latter case some preferential flow paths may be formed by the passing bubbles as the field is applied. These preferential flow paths produce mixing cells with high longitudinal and low lateral dimensions and are un-

Figure 14. Spatial velocity variations in magnetically stabilized beds at different velocities.

desirable because they decrease the quantity of radial fluid dispersion. A quantitative description of the phenomenon is possible by considering the process to be one of "random walk". Baron (1952) has proposed a similar model which is applicable to beds in which the IateraI and longitudinal displacement of fluid are of a constant ratio. The radial dispersion coefficient is proportional to the mean - of the square of the lateral displacement of the fluid, x2, and inversely proportional to the time of mixing, r

Y2

D,

A

N

r

In a random walk model the lateral displacement of fluid is proportional to the lateral displacement in a mixing cell, W,, and the square root of the number of mixing cells, n

-

x2

N

n Wm2

(3)

The number of mixing cells is equal to the length of fluid travel, z, divided by the longitudinal mixing length, 1,

n = -z 1,

(4)

Substituting (4) into (3)

x2

N

z-

Wm2 1,

(5)

The mixing time is equal to the length of fluid travel, z , divided by the interstitial fluid velocity, Vi

Substituting (5) and (6) into (2) (7)

Equation 7 can be used to interpret the radial dispersion behavior reported here. Based on the larger voidage, the mixing cells are believed to be larger in an MSB than in a packed bed, and so both 1, and W , in an MSB are then larger than for a packed bed and from eq 7 above; the radial dispersion coefficient could well be higher in an MSB. Circulating MSBs are

140

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982

believed to produce larger mixing cells than batch MSBs because the circulating MSBs are less structured due to solids motion. Therefore, by the same argument as above, the radial dispersion could be greater in a circulating than in a batch MSB. Visual observations of MSBs show that particles in beds which were “magnetized last” tend to orient along the magnetic field lines while beds which were “magnetized first” do not appear to structure at all. This type of structure may produce fluid mixing cells which have high longitudinal and low lateral dimensions. Such mixing cells have a small value of Wm2/1,. Therefore, from eq 7, it is predicted that there should be less radial dispersion of fluid in beds that were bubbling and then the field applied as compared to beds where the field was applied and then the gas flow rate increased. This prediction is confirmed by the data in Figure 8. In Figure 9 the radial dispersion seems to increase with Reynolds number. It is believed this is due to the increasing voidage and therefore larger mixing cells as the gas velocity is increased. The MSB is similar to a fluidized bed in pressure drop and flowability behavior, but the contacting characteristics are similar to a packed bed. A usual assumption is that flow in a packed bed two-phase contactor is close to being radially constant. This may be valid for small diameter reactors but is certainly not the case once they are scaled up. It is well known that radial nonuniformities exist in these contactors and can cause flow maldistributions which may alter the contacting. Data taken for the MSB, shown in Figures 10 and 12, are similar to resulta for packed beds. Morales et al. (1951) have reported packed bed anemometry profiles with point velocity variations of up to 200%. They also found large velocity variations with angular position. Although the experimental conditions are not identical in both cases, from comparison it appears that the MSB has a smoother profile, with less local velocity variations than a packed bed. Morales et al. attributed the above behavior to groups of pellets forming more and less dense packing arrangements thus causing some channeling of the fluid. At higher gas throughputs the MSB profiles, shown in Figure 13, are typical of those found for packed beds, with an increasing quantity of flow near the tube wall. The data at 18.2 and 23.9 cm/s seem to indicate a maximum velocity near the tube wall, Since the probe diameter was much larger than the particle diameter we were unable to determine if the maximum velocity in the MSB occurs at one diameter or so from the tube wall, as has been reported by Schwartz and Smith (1953) and Calderbank and Pogorski (1957) for isothermal packed beds, and for nonisothermal packed beds by Schertz and Bischoff (1969) where a higher temperature at the center of the bed seemed to augment the behavior. Morales et al. (1951) had earlier shown that the velocity was lowest at the center and near the tube wall. Similar profiles for noncircular packed beds have been reported by Newel1 and Standish (1973). At the lower gas velocities, the flow through the MSB was found to be more uniform than that reported in the packed bed literature. That the MSB behaves more like a packed bed at the higher gas velocities may be due to its decreased flowability at high magnetic fields. It is well known that for fluid flow through tubes, the flow is lower at the tube wall due to friction. For particulate beds it is also known that voidage is higher near the wall than at the bed center. Therefore, we would have expected a velocity profile that goes from zero at the tube wall, reaches a maximum, and then decreases toward the

bed center. Furnas (1929)was one of the first investigators to find that the void fraction was higher near the tube wall than in the remainder of the bed. I t was later shown by Roblee, Baird, and Tierney (1958) and Benenati and Brosilow (1962) that for packed beds of uniform particles, the bed voidage remained constant with radial position up to about three particle diameters from the wall. From the data in Figure 13 it is seen that the shape of the velocity profile changes with increasing fluid throughput. Thus, the bed structure or voidage distribution in an MSB must change with changes in gas velocity. This behavior is quite different from a packed bed which will retain its packed structure independent of the fluid flow. Calculations using the Ergun equation show the radial voidage distributions in the MSB to be similar in shape to the velocity distributions but with less variation with radial position.

Conclusions The effects of magnetic field, solids circulation, path to stabilization, and particle size on the radial dispersion of gas were studied in MSBs. I t was found that the radial dispersion of gas is significantly lower in MSBs than in normal fluidized beds. The radial dispersion coefficients for MSBs where the magnetic field was applied to a bubbling bed were found to be higher than the literature values for packed beds as well as the experimental values for packed beds determined in this study. The radial dispersion coefficients for MSBs that were bubbling and then the field applied were found to double when the beds were circulated as compared to batch beds. MSBs in which the magnetic field was applied after the gas flow were found to have about fourfold less radial dispersion as for beds where the field was applied f i t and then the gas velocity increased. In general, the flow distribution through the MSB was found to be similar although more uniform than those reported in the literature for packed beds. Large spatial variations with radial and angular position were found, but these are similar and even less severe than those reported in the literature for packed beds. These velocity variations increased in spatial frequency and magnitude as the fluid throughput was increased. At low gas velocities the MSB flow pattern is parabolic-like, with flow decreasing near the wall. At intermediate throughputs the flow is nearly uniform, much more uniform than a packed bed. At high gas throughputs the velocity profiles begin to resemble those shown in the literature for packed beds, with increasing flow near the tube wall. The velocity profile in the MSB is useful in determining the fluid flow behavior of this new contacting device. Maldistributions or preferential flow is usually caused by nonuniform bed structure or voidage. Since the MSB flow patterns change with increasing gas throughput, the structure of the bed must also be changing. Based on the interpretations of velocity distributions in the packed bed literature it is concluded that at low gas velocities, the voidage in an MSB is higher in the center than near the wall: thus the parabolic-like profile. The intermediate velocities produce MSBs with the most radially uniform porosity. At high gas throughput, the MSB begins to behave like a packed bed, with higher flow near the wall indicating a region of high voidage near the wall. Acknowledgment The author wishes to express his appreciation to Mr. J. E. Hankins for his assistance in performing the experiments and to Dr. R. E. Rosensweig for his suggestions on

Ind. Eng. Chem. Process Des. Dev. 1982, 21, 141-149

the theoretical interpretation of the radial dispersion results.

ir

= 3.14159

p = gas density p = gas viscosity

Nomenclature C = concentration of tracer gas d = particle diameter $, = gas radial dispersion coefficient G , = solids mass circulation rate Ha= applied magnetic field ,I = longitudinal displacement of gas in a mixing cell n = number of mixing cells P e = Peclet number, Uid,/D, r = radial distance from vessel center and point of tracer injection R = bed radius Re = Reynolds number, d,Uip/M Sc = Schmidt number, fi/pDr Vi = interstitial fluid velocity U , = minimum fluidization velocity Uo= superficial fluid velocity UT = transition to bubbling velocity W, = lateral displacement of gas in a mixing cell W = mass rate of tracer injection = mean radial displacement of fluid 2 = axial distance from tracer injection point, longitudinal movement of fluid in time 7

141

7

= mixing time

Literature Cited Anderson, K. L.; Stokie, 0. M.; Qiibert, R. E. Ind. Eng. Chem. Fund. 1888, 5 , 430. Baron, T. Chem. Eng. frog. 1852, 48. 118. Benenati, R. F.; Brosilow, C. B. A I C M J . 1882, 8, 359. Bologa, M. K.; Syutkin, S. V. Electron. Obrab. Meter. 1877, 1 , 37. Calderbank, P. H.; Pogorski, L. A. Trans. Inst. Chem. Eng. 1857, 3 5 , 195. Fiiippov, M. V. rr. Inst. F k . Akad. Mauk Let. SSR 1880, 12, 215. Filippov, M. V. Izv. Akad. Mauk Lat. SSR 1881, 72, 47. Furnas, C. C. U.S. Bureau of Mines, Washington, D.C., Bulletin 307. 1929. Himmeibiau, D. M.; Bischoff, K. B. “Process Analysis and Simulation”; Wiley: New York, 1967; p 319. Kirko, I . M.; Fiiippov, M. V. Zh. Tekh. F k . 1880, 30, 1081. Lucchesi, P. J.; Hatch, W. H.; Mayer, F. X.; Rosensweig, R. E. “Proceedings of the 10th World Petroleum Congress, Bucharest”; Heyden 8. Sons: Phil., Voi. 4, 1979; p 419. Moraies, M.; Spinn, C. W.; Smith, J. M. Ind. Eng. Chem. 1851, 4 3 , 225. Newell, R.; Standish, N. Met. Trans. 1973, 4 , 1851. Roblee, L. H. S.; Baird, R. M.;Tlerney, J. W. AIChE J. 1858, 4 , 460. Rosensweig. R. E. Sclence 1878a, 204, 57. Rosensweig. R. E. Ind. Eng. Chem. Fundem. 1878b, 78, 260. Rosensweig, R. E.; Siegeil, J. H.; Lee, W. K.; Mikus, T. “Magnetically Stabilized Fluidized Solids,” paper presented at the AIChE 72nd Annual Meeting, San Francisco, Calif., Nov 1979, Schertz. W. W.; Bischoff. K. 8. A I C M J . 1888, 15, 597. Schwartz, C. E.; Smith, J. M. Ind. Eng. Chem. 1853, 45. 1209. Towie, W. L.; Sherwood, T. K. Ind. Eng. Chem. 1838, 31, 457.

Received for review January 23, 1981 Revised manuscript received July 2, 1981 Accepted July 25, 1981

Greek Letters t = bed void fraction

Thermogravimetric Analysis of Swedish Shale Char. Kinetics of the Steam-Carbon and Carbon Dioxide-Carbon Reactions Ingemar BJerle,Hans Eklund;

Marlta Llnn6, and Owe Svensson

Department of Chemical Engineering, Lund Institute of Technoicgy, P.O. Box 740, 5220 07 Lund 7, Sweden

The kinetics of gasification of shale char from the Swedish Ranstad deposit have been investigated at atmospheric conditions in the temperature range 800-1000 O C by thermogravimetric anaiysis. The reaction rate of fixed carbon can be described by modified Langmuir-Hinsheiwood expressions. Similar to coal, strong retarding effects of hydrogen and carbon monoxide were found. The catalytic effect of impregnation with potassium and calcium salts was found to be of minor importance in the temperature range 850-940 OC.

Introduction The Swedish shale deposits, exceeding 50 X 109 tons,and being one of the largest uranium reserves in Europe, are interesting for a commercial exploitation. In connection with the uranium recovery, a simultaneous utilization of the fossil energy in the kerogen is under consideration. A gasification process seems most suitable to fulfil the requirementa on a thermal process for the shale from the Ranstad deposit. A low oil yield of 1.5 wt % (5 gal/ton) excludes a low-temperature pyrolysis process. The direct combustion is less attractive from a high sulfur content of 7 % , even if the environmental aspects could be handled in a technically feasible way. Research and development in the field of gasification has so far been conducted in fluidized bed reactors of bench scale and small pilot plant size. Results from these investigations are reported elsewhere, i.e., Bjerle et al. (1980) and Berggren et al. (1980). The tests in the bench and small pilot reactors have been useful in the assessment of the gasification process and in evaluating the feasibility 0196-4305l82l112 1-0141$01.25lO

of using fluidized bed reactors for this process. For more specific studies of the fiied carbon gasification, thermogravimetry is an applicable method. In this article, the kinetic conditions for the shale char gasification with steam and carbon dioxide in the hydrogen and carbon monoxide atmospheres have been examined. Additionally, the catalytic effects of potassium and calcium salts were studied. Experimental Section The thermogravimetric equipment (TG) used is schematically shown in Figure 1. The unit consists of a standard Cahn 2000 microbalance, together with a furnace section, specially designed for operation in steam atmosphere. The microbalance is built into a stainless steel shell to allow operation at moderate elevated pressures. Noncondensable reactant gases are measured with rotameters and preheated before entering the furnace. When operating with steam, the water flow is controlled by a stepmotor driven syringe. To prevent gas leakage and to avoid 0

1981 American Chemlcai Society