I
JAMES D. JACKSON,l HAROLD A. SORGENTIt2 GERALD A. WILCOX,* and ROBERT S. BRODKEY The Ohio State University, Columbus 10, Ohio
Nuclear Waste Disposal b y .
..
Fluidized Calcination of Simulated Aluminum-Type Wastes Fluid bed systems can be designed on the basis of known methods for heat balance and heat transfer
THE
FIRST LARGE scale fluid bed unit for calcining aluminum-type fuel wastes from test reactors is being built at Arco, Idaho (7). There, an aluminum nitrate solution, resulting from treating the fuel elements with nitric acid, is sprayed into a heated fluidized bed of aluminum oxide particles. Water and nitrogen oxides are removed, leaving the waste primarily as granular, free-flowing aluminum oxide. The feasibility of the process has been investigated (2, 5), but neither the rate nor the manner in which dehydration and decomposition occur was considered. In the work described here, these factors are investigated with both aluminum nitrate crystals and solutions. A thermogravimetric oven provided kinetic information on the rate of aluminum nitrate dehydration and decomposition in a fixed bed. Above 2OO0C., the chemical reaction is so fast that physical forces controlled the over-all reaction. The fluid bed gave much better conversions than predicted from the fixed bed data in the 300" to
Table 1. Heat of Reaction and Free Energy Are Functions of Temperature(4) [Al(NOa)a.9Hz0
(s)
A~(NO~)~.GHZO (s)
+ 3Hz0 (Q) ~ , cal./g.-mole A H z s s ~ ~ +43,310 .--t
=
+
ki.l(N03)3.GHzO ( s ) + 1/z A1203 3NOz a/40z 6HzO (Q) A H 2 8 8 ~ ~=c . 158,990 cal./g.-mole]
+
Temp., O
K.
298
350 400
450 500
+
400" C. temperature range; however, the over-all reaction is still physically controlled. Therefore, in designing calciners for aluminum-type nuclear wastes, the following points should be noted : first, that the fluidized bed provides an excellent system for distributing energy rapidly. The supply of energy is critical; consequently, internal as well as external sources of energy may be required. Second, that atomization of liquid feeds is important in preventing agglomeration of unreacted aluminum nitrate.
Preliminary investigations Early workers ( 6 ) reported that the dehydration and decomposition of aluminum nitrate crystals could be broken down into several independent steps. Limited thermodynamic data reported by Kelly ( 4 ) indicate that the reaction proceeds to completion at temperatures above 425' K. (Table I). To study the dehydration and decomposition steps, differential thermal analysis was undertaken. Its value is in the study of reactions that occur upon heating; heat evolved (or absorbed) by the material undergoing reaction is detected. Decomposition by Differential Thermal Analysis (DTA). Samples of hydrated crystalline aluminum nitrate and calcined product were heated side by
side in an electric furnace. The temperature difference between the sample and aluminum oxide was measured (chromel-Alumel thermocouple) and recorded. The DTA data (Figure 1) shows that an endothermic process occurs when crystalline aluminum nitrate, Al(NO3) 3 ' 9Hz0, is heated from 30" to 100°C. This process represents the combined effects of three physical changes: solution of crystalline nitrate in its own water of hydration, evaporation of free moisture, and partial dehydration of the nitrate. Near 100" C., the sample and the reference material slowly begin to equalize, indicating the continuation of an endothermic process, probably the continued dehydration of the nitrate. At about 145" C., a highly endothermic process, marked by the first evolution of nitrogen dioxide, takes place and continues until the temperature reaches 200' C. After this point, the sample and reference material temperatures begin to equalize as the dehydration and decomposition of the nitrate continues a t a slower rate. Decomposition by Thermogravimetric Analysis. Weight-loss data was obtained to study rates of reaction. Two to nine grams of sample were placed in a platinum crucible suspended in an electric furnace by a platinum wire attached to one arm of an analytical
+
Heat of Reaction, Cal./G.-Mole
Free Energy, Cal.
158,990 158,290 157,610 156,250 156,250
47,360 28,030 9,470 - 9,010 - 27,410
1 Present address, Battelle Memorial Institute, Columbus, Ohio Present address, Atlantic Refining Co., Philadelphia, Pa. 3 Present address, 0. M. Scott and Sons, Co., Marysville, Ohio
Designers of Calciners for Aluminum-Type Nuclear Wastes Please Note! The fluidized bed provides an excellent system for distributing energy rapidly. The supply of energy is critical; consequently, internal as well as external sources of energy may be required 0 Atomization of liquid feeds is important in preventing agglomeration of unreacted aluminum nitrate
VOL. 52, NO. 9
SEPTEMBER 1960
795
5
6 Y)
z sso $ 2
3 c
p
Y ' 0
TIME, MINUTES
-eo 'U 0
XI
100
IXI
200
SAMPLE
1%
yx)
TEMPERATURE.
3%
400
4-
500
*C
Figure 2. Experimental data agreed well with calculated values for operalion at 400" C.
Figure 1. An endothermic reaction occurs when crystalline aluminum nitrate i s heated
ELECTROSTATIC PRECIPITATOR (3" DIAMETER 1/16" TYPE 304 S.S)
FEED
TUBE
1
THERMOCOUPLE
CYCLONE 1/2" 1.D.- 1/16" T Y P E 304 S . S . )
REACTOR FURNACE ( 1 / 8 " SHEET S T E E L )
balance. A chromel-Alumel thermocouple adjacent to the sample crucible measured temperature. The runs were made at IOO", 200°, 300°, 400', and 500" C. The order of the reaction and reaction rate constants (Table 11) were graphically determined for each run (7). Figure 2 compares an experimental with the corresponding calculated curve. At 100' C., the reaction required several days for any appreciable conversion, Doubtlessly this first-order reaction proceeds through some other mechanism than that which occurs at higher temperatures. The reaction was extremely rapid for the other runs, increasing with higher temperature; there was a brief constant rate period while the sample dissolved in its water of hydration. Kext, water vapor and nitrogen dioxide rapidly evolved, with no distinction between the dehydration and decomposition reactions. However, another observation, on a stainless-steel covered hot plate, showed that stepwise decomposition apparently does exist. The rate at 200' C. was zero order, and was probably reaction - controlled; above 200" C., the first order indicates that physical, rather than chemical forces controlled. Further confirmation was obtained from a plot of reaction rate constant us. the reciprocal of absolute temperature on semilogarithmic paper. An activation energy of 7800 calories per gram-mole was obtained for the 300' to 500" C. temperature range. An activation energy of less than 10,000
Table 11. Reaction Rate Constants Increased with Temperature
FLUID12 I N G 045 TUBE
Temp.,
Reaction Order
100 300
1 0 1
400 500
1 1
c.
200
Here are the structural details of the reactor itself. was the preferred material of construction
796
1NDUSTRlAL AND ENGINEERING CHEMISTRY
Type 304 stainless steel
Reaction Rate Constant, Min. -' 4.3
x
10-4 X 10-1 9 . 9 x 10-2 1.65 X 10-1 1.85 X 10-1
3.2
FLUIDIZED CALCINATION calories per gram-mole usually indicates that physical, rather then chemical, forces control the rate of reaction.
Table 111.
Fluid Bed Decomposition of AI u min u m Nitrate
Material Copper
Equipment. Both aluminum nitrate crystals and solutions were used in the fluid-bed studies. The basic equipment is illustrated. In both systems, air taken from the laboratory compressedair line is passed through an oil filter and then divided into two streams. The fluidizing air is dried and heated in a Chromalox preheater whose power is controlled manually by a powerstat; the outlet gas temperature is recorded by a Brown Electronik indicator-recorder. The reactor and most of the auxiliary equipment was fabricated from Type 304 stainless steel, shown to be preferable in corrosion tests (Table 111). The bottom of the 43/8-inch diameter reactor is conical, providing a more even distribution of the fluidizing air; however, in runs with shallow beds of solids, a porous stainless-steel distributor plate was used. The reactor head was equipped with a feed tube, thermowell, and gas exit line to which a cyclone separator and an electrostatic precipitator were attached. Liquid feed to the reactor was controlled by a Lapp Pulsafeeder. Solid feed was metered by a screw feeder
Corrosion Rates of Materials of Construction in Aluminum Nitrate" Show Type 304 SS to Be Preferable
Exposure, Corrosion Rate, Mils/Mo. Hr. Vapor Interface Liquid 1.4 3.9 13 Gain 878 0.77 1.6 1.4 2.4 Aluminum, 3 5 13 Gain 878 1.5 2.5 Mild steel 13 Gain 360 500 Type 304 SS 13 Gain Gain Gain Gain 878 Copperb 26.6 7.4 867 4.0 Type 304 SSc 26.6 0.11 867 0.001 Aluminum,. 3Sb 26.6 5.0 867 3.0 Type 304 SSd 26.6 0.16 0.016 867 a Room temperature; concentration, 2 . 4 M Al(NOs)a. Coupled t o Type 304 SS. Coupled Coupled to aluminum, 35. to copper.
... ...
... ... ... ... ... ...
... ... ... ... ... ... ... ...
... ... ...
driven by a '/r-hp. electric motor and controlled with a reversible, variable speed drive. The feeds were mixed with a stream of unheated air at the top of the reactor and then carried down the 8/s-inch feed tube to a point below the surface of the fluid bed. The air velocity in the feed tube was adjusted to prevent build-up of the reactant in the tube, and to atomize liquid feeds when used.
...
T h e reactor furnace was constructed of sheet steel and lined with insulating brick. Four glowbars, regulated by a 15 kv.-amp-55-ampere powerstat, provided heat for the reactor. Temperature was controlled by a Brown Pyrovane indicator-controller. Experimental Procedure. At the beginning of each solution feed run, aluminum oxide was charged to the 28 mesh) reactor: 1500grams (-10 to
+
ELECTROSTATIC PRECIPITATOR
COMPRESSED A I R SUPPLY
OIL FILTER
FRESSURE REGULATOR
MA N O M TER
PAN This schematic diagram gives the essential components of the system VOL. 52, NO. 9
SEPTEMBER 1960
797
0
z z
I
I
1 CWLETELY
(FIRST
w
+ 0.8 LL
O6
ORDER)
-
LIQUID
il
(ZERO
LL
z P
FEED
Figure 3. Short residence times are desirable for checking the zero-order curves
RUN
SOLID F E E D R U N
2 0 4 U 0
I
REACTOR
P I S T O N FLOW REACTOR (FIRST ORDER)
t a
I
MIXED
ORDER3
02-
e
TIME. MINUTES
for runs at 400’ C., where the bed was supported in the cone of the reactor; 400grams(-lOto +20meshat40O0C., and - 10 to +28 at 500 ’ C.) where the distributor plate was used. A preliminary heating period of 1 to 2 hours established a steady temperature state. Aluminum nitrate solution was fed to the reactor at a rate of 5 to 10 ml. per minute. The solution was 823 grams of Baker and Adamson reagent grade aluminum nitrate in 500 ml. of distilled water. Properties of aluminum nitrate solutions, such as density or volume us. concentration, have been reported (7) ’ When a run was completed, the reactor was disassembled and the bed product and cyclone dust were removed and weighed. The solids were washed and the resultant filtrate analyzed for residual nitrate by titration with sodium hydroxide. At solution feed rates above 10 ml. per minute, agglomeration of bed particles with partially reacted aluminum nitrate became a problem. Also, the technique of dismantling the reactor was not applicable to short residence times. A crystalline aluminum nitrate feeding technique was consequently developed. For the solid feed runs, approximarely 1500 grams of -14 to f 6 0 mesh aluminum oxide was charged to the reactor. After the preliminary heating period. the crystals were fed at approximately 20 grams per minute for 2 to 5 minutes. The bed was run for an additional 2 to 3 minutes and then rapidly dumped (less than 10 seconds) through the bottom of the reactor into a cooled collecting pan. This technique was employed to obtain short residence times, which were considered to be the sum of half the feed period plus all of any additional run time. The bed and cyclone products Lvere weighed and analyzed as in the solution runs.
completely mixed and piston flow reactors (Figure 3). The 99+% experimental conversions at 15 and 30 minute residence times are much better than predicted. I t was suspected that the zero-order reaction (chemical) might be the controlling factor in the fluid bed, rather than the first-order reaction (physical). A crude estimate of the zero-order conversion was made (Figure 3). T h e desirability of short residence time runs to check the zero-order curve is apparent. However, the short residence time runs, with crystalline feed at temperatures less than 4OO0C., were again much better than predicted from the thermogravimetric data (either first or zero order). Assuming that the 2-minute conversicns are either chemically or physically controlled, one can obtain the zero- or first-order reaction rate constants, respectively (Table IV). From chese constants. an activation energy of approximately 6000 calories per gram-mole was calculated (8). This falls well within the range of physically controlled reactions (first order). Therefore, the fluid bed is seemingly controlled by either heat or mass transfer to the particles. Since the technique for obtaining fluid bed conversions in this study was admittedly not very sophisticated, an error analysis was made (8). I n the range of the residual nitrate concentrations, the water leaching, sodium hydroxide titration assay gave results approximately 59;b low for synthetic beds. Other errors, such as slow cooling
Table IV.
First-Order Zero-Order Rate Rate Constant, Constant. Temp., O C . 313 367
Results and Qiscussion
With the first-order reaction constant at 400’ C. (thermogravimetric), conversion predictions were made for
798
Rate Constants and Activation Energies
Activation energy, cal./g.-mole
INDUSTRIAL AND ENGINEERING CHEMISTRY
Min. -1
1.09 1.66 5,840
Min-1
0.443 0.482 1,180
of the reacting bed, were estimated at the maximum conceivable limits. The modified activation energies were calculated, and in the worst case, they increased the activation energy only 16%. This is still less than 10,000 calories per gram-mole. Heat transfer was limited in the fluid bed. By switching from solution to crystalline feed, the effective aluminum nitrate feed rate could be increased 23%. The solid feed did not hinder operation, because the nitrate first dissolved in its water of hydration and became a concentrated .liquid feed. A wider particle distribution used in the solid feed runs significantly improved the degree of fluidization and, subsequently, bed efficiency. \\Then narrower particle distributions were used, severe agglomeration occurred and lower conversions were obtained. I n conjunction with the solution feed runs, particle mechanics studies were undertaken ( 3 ) . Considered were particle structure, changes in structure, effect of space velocity, attrition, and carryover. The system was not at a true steady state and the feed spray was not representative of a typical commercial installation; consequently, these results are not reported in this work. Acknowledgment
The authors gratefully acknowledge the aid of Battelle Memorial Institute, through its educational program, and the advice and assistance of the members of its staff. The Aluminum Co. of America provided the aluminum oxide used as initial bed material. Literature Cited (1‘) Chem. Eng. 65, 51 (August 28, 1958!. (21 Grimmett, E. S., “Calcination of
Aluminum-Type Reactor Fuel Wastes in a Fluidized Bed,” Idaho Operations Office, U. S. Atomic Energy Commission, Phillips Petroleum Co. Rept. I D 0 14416, August 1, 1937. (31 Jackson, J. D., “Particle Mechanics of Calcination of Aluminum Nitrate Solutions in a Fluid Bed,” M.S. thesis, The Ohio State University, 1958. (4) Kelly, I