I
J. C. BRESEEI and P. R. LARSON2 of Technology, O a k Ridge, Tenn.
Engineering Practice School, Massachusetts Institute
Continuous Cold Trap for Fluoride Volatility Processing of Uranium This continuous condenser, proposed for recovering uranium hexafluoride from inert diluent gases in fluoride volatility processes, is a novel and refreshing approach to a vexing problem
RECENTLY
LAWROSKI(6) described fractional distillation as a useful and probably economic method for chemical processing fully enriched uranium fuels. By using this method, certain types of alloy fuel elements may be dissolved in a fused fluoride salt bath in the presence of hydrogen fluoride gas a t temperatures between 500" and 700" C. Following dissolution, uranium may be separated from essentially all the fission products by stripping uranium hexafluoride from the salt bath with either fluorine or bromine pentafluoride. Overhead vapors may then be condensed and fractionally distilled to recover pure uranium hexafluoride from other volatile fluorides. The distillation portion of the process must be carried out a t pressures greater than 1.5 atm., the triple point pressure of uranium hexafluoride. At pressures less than 1.5 atm., uranium hexafluoride condenses directly from gas to a solid phase. Development of a low pressure volatility process a t Oak Ridge National Laboratory (2) was recently reported. Separation of uranium from fission products with volatile fluorides is accomplished by a sorption-desorption cycle on a sodium fluoride bed at less than atmospheric pressure. I t is often desirable to operate all portions of a radiochemical process under less than atmospheric gas pressures to minimize the possibility of gaseous fission products escaping. However, low-pressure fluoride volatility processing of uranium necessarily implies gas-solid condenser equipment or cold traps for final uranium recovery, As fluoride volatility processes developed thus far do not allow for plutonium recovery, only fully enriched fuels can be economically handled. Therefore, dimensions of all process equipment including cold traps are limited by criticality considerations. For example, if a tank with cooled walls were used for a Present address, Oak Ridge National Laboratory, Oak Ridge, Tenn. Present address, Standard Oil Co. of California, Richmond, Calif.
cold trap, diameter of the tank must necessarily be kept below 6 inches. Because a tank with such a small diameter would be relatively easy to plug with condensed solids, capacity of a single piece of batch equipment would be severely limited, regardless of size of the one or two noncritically safe dimensions. For a continuous process, may such batch units might be required for operation in parallel with many others in a standby status. There are potential advantages of continuous gas-solid condenser equipment, especially a type built with one or more dimensions critically safe. Advantages result from increased equipment capacity and decreased capital costs. One such device might be a shell-and-tube condenser with scraped tube walls. Another which might have advantages for enriched fuel processing is the fluidized condenser. Fluidized Condenser
Properties. One property of a fluidized bed which makes it attractive in many processes is even bed temperature distribution resulting from high rates of heat transfer caused by solids mixing throughout the bed. Heat transfer to surfaces in contact with a fluidized bed is rapid compared to that from a gas
F U E L ELEMENT
stream at the same superficial velocity as the fluidizing gas. The combination of these two properties suggests that a fluidized condenser would be efficient for condensing of gas to solid. A mixture of condensable and noncondensable gases would be continuously introduced to the bottom of a bed of fluidized solids. Because of the large area available for heat and mass transfer in the bed, the gas stream would quickly come to thermal equilibrium with the bed and the sublimable material would condense on the surface of the fluidized particles. For recovery of the condensed material, a stream of solids from the bed would be removed continuously. Under special conditions, sublimation condensation might take place in the vapor phase and produce a fog. I n a fluidized condenser, such fog particles would either vaporize and redeposit on nearby larger particles or be removed by the filter properties of the bed. Classification a n d Description of Condenser Types. Heat of condensation could be removed through the walls of the fluidized bed, or recirculated solids could be cooled externally. Condensation could occur either on crystals of the material to be condensed or on an inert carrier. Thus, there are four possible condenser types (Table I).
z
8
H F AND H2 WASTE
+ a
-
UF6 PRODUCT TO F2 COLD TRAPS
iT
FUSED FLUORIDE
HF
-
-
0 Lo
VOLATILE F P FLUORIDES
ti.
P LL
FUEL ELEMENT DISSOLUTION
URAhlUhl
-
F6
VO L AT I L I2AT I ON
0 _J 3
LL
2
2
n
w
F2 N2
-
.F2 STRIP
In this low pressure volatility process developed at O a k Ridge National Laboratory, enriched uranium is recovered from heterogeneous reactor fuel VOL. 49, NO. 9
SEPTEMBER 1957
1349
Figure 1
Two patents (7, 9 ) cover a fluidized condenser for possible use with phthalic anhydride, naphthalene, or other sublimable materials. However, essentially no information exists in the literature describing the performance of such pquipment . thesis was written on the design and operation of a fluidized condenser for zirconium tetrachloride (7). Because of experimental difficulties encountered at the high temperature of operation (approximately 400' C.), only one 40minute run was made. During this run, zirconium tetrachloride was removed continuously from a dilute mixture with nitrogen by condensation on externally cooled crystals of zirconium tetrachloride (condenser type 3, Table I). Earlier attempts to operate the fluidized condenser as a type 1 (inter-
Table 1.
Phase diagrams for uranium hexafluoride and iodine
nally cooled) condenser failed, because of extreme caking on heat transfer surfaces. Selection of Experimental System. To apply the concept of a fluidized condenser to uranium hexafluoride condensation, it was decided first that more stable condenser operation could be maintained by using an inert carrier. Although an inert carrier necessitates an external stripping operation, the condensed solids concentration in the condenser at any time will depend on the recycle rate of the carrier and may be varied for smoothness of operation. The result will be a more flexible piece of equipment. I t was decided further to attempt continuous condenser operation with internal heat transfer. With an inert carrier, the most obvious technique for
Classification of Fluidized Condenser Type
Type so.
Fluidized Solid
Heat Transfer
Remarks
1
Condensed material
Internal
No solid feed; product removed
2
3
Inert carrier Condensed material
Internal External
4
Inert carrier
External
1 350
INDUSTRIAL A N D ENGINEERING CHEMISTRY
directly External stripping required Product removed a s side stream from solids recycle External stripping and cooling required
external heat transfer is through use of" fluidization. External cooling of thc stripped carrier in a fluidized bed overcomes the possible difficulty of caking on internal heat transfer surfaces found in zirconium tetrachloride studies. However, it could be anticipated that an abrasive action by the fluidized inert carrier might keep internal surfaces clear FinaIIy, it was decided that operation of a continuous, type 2, fluidized condenser (Table I) should be investigated for a system similar to the uranium hexafluoride-fluorine system encountered in fluoride volatility but lacking the severe requirements for absence of moisture or most organic materials. The system chosen was iodine-nitrogen. Pressure-temperature phase diagrams for uranium hexafluoride and iodine are compared in Figure 1. Even though the triple points for the two materials are dissimilar (1.5 atm. a t 64' C. for uranium hexafluoride; 0.12 atm. at 114' C. for iodine) it was expected that performance of a fluidized condenser for each would be similar at comparable equilibrium vapor pressures. A cold trap for fully enriched uranium hexafluoride recovery would probably be operated at very low temperaturesperhaps -45' C., where vapor pressure of solid uranium hexafluoride is approximately 0.24 mm. of mercury. If a 50% mixture of uranium hexafluoride and
U R A N I U M PROCESSING fluorine a t 1 atm. were passed through a cold trap a t -45' C., more than 99.9% of the uranium hexafluoride could be recovered a t equilibrium. For a comparable recovery of iodine from an iodine-nitrogen mixture under the same conditions, the cold trap would need to be operated a t 21 O C. Thus, an experimental fluidized condenser for iodine recovery may be cooled conveniently with water, but an actual installation for uranium recovery would require elaborate refrigeration equipment.
and the gas distribution plate. Both the carrier feed and withdrawal screws were powered by motors having variable-speed Graham transmissions manufactured by Doerr Electric Corp., Cedarburg, Wis. The gas distribution plate was a piece of brass plate with size 60 (0.04-inch diameter) drilled holes. A piece of fine-mesh Monel screen covered the plate to prevent the carrier from clogging the feed lines when the column was shut down. A heated borosilicate glass tee held the distribution plate in place and joined the column to the copper tubing. Iodine was vaporized in a copper pot, 4 inches in diameter by G inches long, submerged in a bath of molten wax. The wax bath was heated by means of Nichrome wire wrapped around the outside of the container-a large stainless steel pot. Nitrogen and iodine feed lines to the column were '/%-inch copper tubing wrapped with Nichrome heating wire and insulation. Temperature in these lines was maintained above 300' F. to prevent condensation of iodine. Concentration of iodine in the inlet gas stream to the column was determined by means of a calibrated Cenco Photelometer. The Photelometer cell was machined from a brass block and had borosilicate glass windows 11/2 inches in diameter and '/g inch thick with a 5/ginch gap between the windows. The
Experimental
Equipment. Performance of a fluidized condenser with the iodine-nitrogen system was studied ir, glass equipment, with auxiliary lines essentially of copper and brass (5). The condenser column consisted of a flanged borosilicate glass pipe 1.5 inches in inside diameter and 30 inches long, with a jacket 24 inches long. A 3-inch inside diameter borosilicate glass tee, connected to the top of the column by a reducer, was used as a calming section. The inert carrier, Scotchlite glass beads (No. 9-15 with a mean diameter of 0.00695 inch), manufactured by the Minnesota Mining and Manufacturing Co., St. Paul, Minn., was fed to the column through a plate covering the calming section and was withdrawn from a spout between the-bottom of the water jacket
cell was heated with Nichrome wire to approximately 350' F. Nitrogen was supplied to the system from two cylinders having a capacity of 220 cubic feet. Flow rate of nitrogen was controlled by needle valves, and upstream pressure was measured with a Bourdon-type gage and a 50-inch mercury manometer. Orifice pressure drops were indicated on two manometers; the dilution nitrogen-orifice manometer had mercury as the deflecting liquid, while mercury and Meriam red oil (0.827 specific gravity) were used alternately for the still pot nitrogen-orifice manometer. Temperature was measured with copper-constantan thermocouples and a Leeds & Northrup Speedomax temperature recorder. Where more precise temperatures were desired (&5' F. on the Speedomax recorder) the thermocouples were switched to a Lee& & Northrup potentiometer ( f 0 . 2 ' F.). Cooling water was supplied to the top of the condenser jacket after passing a n orifice used to regulate flow rate. The desired level was maintained by means of a variable liquid leg from the outlet of the condenser. Pressure drop across the fluidized bed, and absolute pressure at the column top and a t the still pot, were measured by Taylor pressure transmitters. Presssure lines, heated to prevent condensation of
PRESSURE L I N E S WET TEST
tv"# RESISTANCE HEATERS
@ -+"
THERMOCOUPLES THERMOMETERS
TO PRESSURE
--
GAGE
1
MANOMETERS
I / ! I ar-
I AIR
TO TRAPS (FOR PHOTELOMETER CALIBRATION )
--I
Performance of this fluidized condenser, proposed for adaption to uranium hexafluoride, was .studied using a nitrogen-iodine system VOL. 49, NO. 9
SEPTEMBER 1957
1351
Condenser column of the assembly paposed far adaption, consisted of a flanged borosilicate glass pipe with a jacket 24 inches long
This is the assembled fluidized condenser used for studying iodinenitrogen. Both the carrier feed and withdrawal screws were powered by motors having variable-speed Graham transmissions
iodine, were led to the transmitters mounted in a furnace maintained at 300" F. The transmitters, when supplied with control air having a 22-pound force per square inch gage give output pressures of 3 to 18 pounds per square inch gage which were read on Bourdon-type Duragages. Iodine vapor remained hot and did not condense or damage the output gages. Gas leaving the bed was analyzed for iodine concentration by passing a sample through 50 ml. of a O.OOILVsolution of sodium thiosulfate and a wet-test meter. An air ejector pulled the gas sample through the absorption train. The gas was dispersed in the thiosulfate solution with a fritted gas dispersion tube. Experimental Procedure. Iodine was heated in the copper still pot to a temperature below the boiling point of iodine (364" F.), and the vapors were forced from the pot by nitrogen gas. T h e iodine-nitrogen mixture was combined with the second nitrogen stream in a small jet mixer. The inlet iodine concentration in the combined stream was then measured colorimetrically with the Photelometer. The carrier was fed to and withdrawn from the condenser by screw mechanisms powered with variable-speed motors. The carrier feed rate was adjusted to maintain a n iodine concentration on the carrier of approximately 0.6 weight 70. A few batch runs indicated that approximately 0.6 weight % of iodine was the maximum allowable concentration if the bed were to remain
1352
well fluidized. A constant amount of carrier, about 0.78 pound, was maintained in the condenser, as indicated by a n approximately constant pressure drop across the bed. Iodine concentration in gas leaving the bed was determined by adsorption in sodium thiosulfate, and that on the carrier was determined by weighing the beads before and after the iodine was vaporized from the beads in a n oven. Water rate to the column was maintained at approximately 30 pounds per hour and the water level was adjusted to the average height of the bed.
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
1"lE"-5
STEAM
-i
STR P P E A /
UF6 L I O J I D
4
UFg RECYCLE
'f'
-1YERT
GAS
L
'T
This i s the proposed uranium hexafluoride recovery system evolved from the fluidized condenser principle
U R A N I U M PROCESSING
4
Before operating the fluidized bed as a condenser, several batch runs were made to determine heat transfer coefficients from bed to wall. Water rate, amount of carrier in the column, and gas rate were varied for these runs. Nine batch runs were made using the apparatus as a fluidized condenser in order to determine the maximum iodine concentration allowable on the carrier. No other variables were investigated during these batch condenser runs. Before a continuous run was started, carrier rate, bed height, water rate, water level, and diluent nitrogen rate were set. When the carrier rate was low, heat transfer data were collected both before and during the introduction of iodine. The iodine valves were turned until iodine concentration as indicated by the Photelometer was approximately the desired value. Data were taken when the iodine inlet concentration was constant. Temperature, pressure, and flow-rate measurements were made over a 6- to 12-minute period depending on the carrier flow rate. Samples of withdrawn carrier were taken every 2 or 3 minutes. Variation in concentration of iodine on the carrier with time, determined after a run was completed, indicated whether steady state had been reached. At the end of a run, the still-pot nitrogen line was closed and the bed remaining in the column was flushed with fresh carrier and nitrogen. Steam was run through the water jacket to aid the flushing operation. For fluidized condensation runs, the independent variables were iodine inlet concencration (1.1 to 9.4 mole %) and the flow rate of gas in the condenser (0.21 to 0.60 foot per second). Dependent variables included gas temperature and concentration from the bed, heat transfer area, heat flow rate, and water outlet temperature. Results and Discussion. Results of this investigation may be divided into two parts-heat transfer (Figure 2) and mass transfer determinations. Figure 2 represents determinations of the inside heat transfer coefficient, h,, as a function of superficial gas velocity from 0.18 to 2.0 feet per second. Inside heat transfer coefficients were calculated from the over-all coefficient and previously measured water-side and wall coefficients. Runs made with and without iodine condensation in the bed are shown. As may be seen in Figure 2, no apparent difference was observed between heat transfer coefficients from the fluidized bed in the presence or absence of condensed iodine. The downward trend of coefficients with increasing gas velocity results from increased void volume in the bed at higher velocities. This result or
400
200
100 50
10
0 WITHOUT IODINE CONDENSAT 0 WITH IODINE CONDENSATION
5
2
1
05
02
01 01
02
20
05
20
SUPERFICIAL GAS VELOCITY ( f t /sec ) Figure 2.
Results of heat transfer using a fluidized condenser
trend has been found by other investigators of heat transfer in fluidized beds (4, 8). A rapid decrease in the inside coefficient would be noticed a t lower gas velocities not sufficient to fluidize the bed. The same plot gives calculated curves based on correlations of Colburn and Hougen ( 3 ) for inside heat transfer coefficients in the absence of the fluidized bed. At very low superficial velocities, the coefficient is essentially independent of superficial velocity because of natural convection. Natural convection heat transfer requires that the coefficient (at constant gas pressure) depend upon the difference between gas and wall temperatures. This effect is shown by the two lower curves. The results of five runs with a mean
temperature difference of 140' F. are compared with the upper natural convection curve. The agreement is within the precision of the data. T h e AT,,, of 140" F. is typical of the difference between the inlet gas temperature to the fluidized bed and the jacket wall temperature. However, temperature of gas in the bed (and fluidized solids) typically was only I O o F. above wall temperature. Under such conditions, natural convection heat transfer would be further reduced to the lower line of Figure 2. Inside heat transfer coefficients a t low superficial velocities were increased b y the presence of a fluidized bed. The increase ranged from 25-fold to almost a hundredfold. Fifty tests were made on efficiency of iodine removal by the fluidized bed conVOL. 49, NO. 9
*
SEPTEMBER 1957
1353
denser. During 47 runs the recoverv was greater than 99% (lowest was 98.3%) and during 22 runs the recovery exceeded 99.9%. Exit concentration of iodine ranged from 0.02 to as low as 0.0002 mole 70, and was relatively independent of inlet gas rate or composition. Thus, the per cent iodine recovery tended to increase with increasing feed concentration. In all cases the exit concentration IvaS less than equilibrium concentration or iodine-i.e., vapor pressure of iodine at the temperature of the bed. The carrier was dried in an oven at 150’ to 200”C. and then cooled in a gas-tight container to room temperature before being fed to the column. I t is probable that iodine molecules were adsorbed on the carrier and held by forces stronger than those from condensation. As an indication of adsorptive characteristics for oven-dried microspheres, the carrier adsorbed moisture if exposed to room air to such an extent that fluidization was severely impaired. Adsorption took place even though moisture content of the air was well below saturation at the temperature of the carrier. Without knowledge of the adsorption isotherms for small quantities of iodine adsorbed on microspheres, it was impossible to calculate mass transfer coefficients which could be applied to other chemical systems. However, ease of operating the fluidized bed during iodine removal runs made its application to uranium hexafluoride recovery attractive. The heat transfer surface remained clear during condensation runs, because of the scouring action of the fluidized bed. Only when gas velocity was insufficient for turbulent bed mixing was there partial plugging of the bed near the inlet plate. Application to Uranium Hexafluoride Condenser Design
Proposed System. The uranium hexafluoride recovery system consists of a condenser column, a stripping column, and a Ietdown and liquefaction unit. The gaseous mixture of uranium hexafluoride and inerts (fluorine, nitrogen, plus trace fission products) enters the bottom of the condenser column, where it is quenched to bed temperature. The inerts pass out of the bed through a cyclone separator and are vented to a hot off-gas system. The carrier (probably a corrosionresistant metal powder) containing condensed uranium hexafluoride is withdrawn continuously and fed to a steam-heated stripping column. The vaporized uranium hexafluoride is recycled through the column to maintain fluidization, and a stream is withdrawn through the letdown and liquefaction
1 354
unit to maintain constant pressure in the recycle stream. Stripped carrier is withdrawn from the heated column and raised by a gas-lift to the condenser column. The gas used in the lift could be precooled to return the carrier a t any desired temperature. Rate of circulation for the carrier through the entire system can be varied to change the condensed uranium hexafluoride concentration in thr fluidized condenser bed. Sample Design Calculations. Approximate dimensions and performance characteristics of a fluidized condenser for uranium hexafluoride recovery are illustrated by sample design calculations. Design Criteria. Capacity is 1 pound of uranium hexafluoride per minute; feed gas composition, 10% by volume or 51y0 by weight of uranium hexafluoride, and 90% by volume or 49% by weight of fluorine. Recovery is 9 9.9%. Assumptions. Superficial velocity for fluidization is 1foot per second; uranium hexafluoride on carrier in bed, 0.5% by weight; and over-all heat transfer coefficient is 50 B.t.u./hr./sq. foot/’ F. Temperature for wall is - 60 ’; for bed, -50’; and for inlet gas 70’ F. Heat Load on Condenser. Total heat load is 7080 B.t.u. per hour-2770 from specific heat of fluorine and 590 from uranium hexafluoride; also 3720 from heat of sublimation for uranium hexafluoride. Heat Transfer Area Requirements. Q - 7080 Area = -- = 14.2 sq. feet UAT 500 Condenser Dimensions. Inside diameter is D and inlet gas rate is 10.2 cubic feet (STP) per minute. At the fluidization velocity of 1 foot per second, inside diameter D is 5.6 inches. Length, L (assuming a simple jacketed column), is 14.2 sq. feet X 12/5.67r = 9.7 feet. (The cooled column length could be reduced by an internal cooling coil.) Carrier Recycle Rate. One pound of uranium hexafluoride per minute and 0.005 pound per pound of carrier equals 200 pounds of carrier per minute. Heat Transfer Area Comparison with Ratch Cold Trap. Batch cold trap temperature assumptions for wall is --GO”, inlet gas, 70”, and exit gas, -50” F. Over-all heat transfer coefficient is 1 B.t.u./hr./sq. foot/’ F. Log mean batch temperature difference,
Summary
A fluoride volatility process has been described recently for chemically processing enriched uranium fuels. A fuel element is dissolved in a mixed-fluoride molten salt in the presence of hydrogen fluoride gas, and uranium is subsequently separated from the bulk of fission products by fluorination. Two alternative schemes have been proposed for final purification and recovery of uranium hexafluoride. The first consists of fractional distillation a t pressures greater than 1.5 atm. The second involves sorption-desorption and may be operated a t less than atmospheric pressure. Under low pressure conditions, uranium hexafluoride is recovered by solids condensation in batch cold traps. A continuous cold trap based on the fluidized condenser principle has been proposed. The design would consist of an internally cooled column containing a fluidized inert carrier which would be removed continuously, stripped by superheated uranium hexafluoride vapor, and recycled to the column. Tests with the system, iodine-nitrogen, show this cold trap design is feasible and that compared with a batch cold trap of comparable capacity, the fluidized condenser principle may reduce the heattransfer area requirements by more than a factor of IO. Literature Cited (1) Beck, C. B., Canby, T. D., Zonis, I.,
(2)
(3) (41
(5)
(6)
- 10 ATL..,~.is 130 -__ - 47.3”F. In 13 Area ratio is calculated by
- 50 X 10 _ ____ 1 x 47.3
_AB __AF.B.
10.5
Results of Design Calculations. Fluidized bed dimensions for a jacketed column are 5.5 inches in outside diameter and 10 feet long; condenser capacity is 7100 B.t.u. per hour, and area ratio reduction by fluidization is 10.5.
INDUSTRIAL AND ENGINEERING CHEMISTRY
(7)
(8) (9)
“Fluidized Condenser,” S. M. thesis, Rept. KT-123, MIT Engineering Practice School, Oak Ridge, Tenn., 1952. Cathers, G. I., Leuze, R. E., “Volatilization Process for Uranium Rccoverv.” Nuclear Ene. and Sci. Congiess, Cleveland, Ghio, December 1955. Colburn,-A. P., Hougen, 0. A., Iiw. ENG.CHEM. 22, 522 (1930). Fairbanks, D. F., “Hrat Transfer to Fluidized Beds,” Sc.D. thesis, Massachusetts Institute of Technology, Cambridge, Mass., 1953. Larson, Paul R., “Fludized Condensation of Iodine,” S.M. thesis, Massachusetts Institute of Technology, Cambridge, Mass., 1954. I,awroski, S., “Survey of Separation Processes Other than Solvent Extraction Recovery,” Inter. Conf. on Peaceful Uses of Atomic Energy, U.N.-A/Conf. 8/P823, Geneva, Switzerland, August 1955. Lewis, W. K. (to Standard Oil Development Go.). U. S. Patent 2,607,440 (Aug. 18, 1952). Mickley. H. S., Trilling, C. A., IND. ENG.CHEM. 41,1135 (1949). Patterson, J. A. (to Standard Oil Development Co.), U. S. Patent 2,583,013 (Jan. 22, 1952). KECEIVED for review June 29, 1956 ACCEPTED March 22, 1957
American Nuclear Society, Chicago. Ill., .June 1 9 5 6