ORNL 4434 UC 80 Reactor Technology £ e Low- PRESSURE DISTILLATION OF MOLTEN 1l e T _"_';"'FLUORIDE MIXTURES NONRADIOACTIVE TESTS | e e e FOR THE MSRE DISTILLATION EXPERIMENT | J R nghtower, J; AR 7_. , 4 1 L e g ZiOAK RIDGE NATIONAI. I.ABORA'I'ORY ' ~ 2 .. TR operate d by : | A T T e T T e UNION CARB!DE CORPORATION ST e for Ihe ' e u s ATOMIC ENERGY comwsstou ' BISTRIBUTION OF THIS DOCUMENT IS UNLIMITED - S Prmted in " the Umted States of 'America. Ava:!able from R : : ‘Nationa! Techmcal ‘Information Service U US Department of Commerce, Sprmgf:e!d Vlrgsma 22151 ' - “Price: Printed Copy $300 Microfiche $0.65 - reprESents that ItS ‘use would not mfr:nge prwately owned nghts _Thrs report was prepared as-an. account of work sponsored by the Umted‘ ‘States Government Neither - the Unlted States nor. the United States Atomic | Energy. Commlss:on nor any of their emplovees nor’ any of therr contractors, A subcontractors, or their emplovees makes any- warranty, express or implied, or | - assumes any legal liability or responsabmty for ‘the .accuracy, completeness orf' e "usefufness of any enformatlon apparatus, product or “ process d:scfosed or_ e 3 [ “haal W ORNL-4434 Contract No. W-7405-eng-26 ¥) CHEMICAL TECHNOLOGY DIVISION UNIT OPERATIONS SECTION LOW-PRESSURE DISTILLATION OF MOLTEN FLUORIDE MIXTURES: NONRADIOACTIVE TESTS FOR THE MSRE DISTILLATION EXPERIMENT J. R. Hightower, Jr. L. E. McNeese - —————————LEGAL NOTICE , This report was prepared as an account of work , sponsored by the United States Government. Neither the United States nor the United States Atomic Energy = Cor_nmission, nor any of their employees, nor any of . their contractors, subcontractors, or their employees, | makes any warranty, express of implied, or assumes any L legal liability or responsibility for the accuracy, com- | pleteness or usefulness of any information, apparatus, ! product or process disclosed, or represents that its use | 1 | would not infringe privately owned rights. JANUARY 1971 OAK RIDGE NATIONAL LABORATORY / Oak Ridge, Tennessee - operated by B UNION CARBIDE CORPORATION - for the - U. S. ATOMIC ENERGY COMMISSION DISTRIBUTION OF THIS DOCUMENT IS - - +) 3) (=)} . iii CONTENTS ABSTMCT - o - - . e - * - * * * » * * . » - . . . . . - INTRODUCTION « o . ¢« » ¢ o ® s 8 e « & s e * + o . . EXPERIMENTAL EQUIPMENT . . . « ¢ « o ¢ o o « & & 2.1 Process Equipment . . . « « ¢« &+ ¢ ¢ o s o o s o @ 2.2 Instrumentation . « ¢ « ¢ ¢ « o & o o .« .. 2.2.1 Measurement and Control of Temperature . 2.2.2 Measurement and Control of Pressure . . . 2.2.3 Measurement and Control of Liquid Level. . OPERATING PROCEDURE. . ¢ & & ¢ « o o o ¢ o » o o s & EXP ERIMNTAIJ RES ULT S - - 2 . a 2 [ ] - - * - - 2 . 4.1 Measurement of Distillation Rates . . + . « « « 4.2 Measurement of the Degree of Separation of NdF3 from LiF-Ber"ZI‘FZI_ Carrier Saltu e o 8 s o @ . . e 4 . 3 Difficulties - * » * - . . - . . L . - - » . . . . » CORROSION TESTS. & « v & & ¢ &+ o o o o o o o o o o o o POSTOPERATIONAL INSPECTION . . « « & & ¢ « o o o & CONCLUSIONS. « &+ o &+ o o o o o o o o ¢ o s s o ACKNOWLEDGMENTS. « « ¢ ¢ o« « o « o & o o o o & REFERENCES +« « ¢ ¢ ¢ ¢ ¢ ¢ ¢ o ¢ o s o s o s o o o A.PPENDIXES . . . . . . . . e . . e . . . . . . . . . . - 10.1 Appendix A. Derivation and Solution of Equations Describing Concentration Polarization. . . . . . . 10.2 Appendix B. Drawings Showing Postoperational Wall- Thickness and Dimensional Measurements . . . . . . oI N N 10 12 15 21 29 30 34 35 36 37 38 39 46 ) LOW-PRESSURE DISTILLATION OF MOLTEN FLUORIDE MIXTURES: NONRADIOACTIVE TESTS FOR THE MSRE DISTILLATION EXPERIMENT J. R. Hightower, Jr. L. E. McNeese ABSTRACT Equipment was designed and built to demonstrate the low-pressure distillation of a 48-liter batch of irra- diated fuel salt from the Molten Salt Reactor Experiment. The equipment consisted of a 48-liter feed tank, a 12- liter, one-stage still reservoir, a condenser, and a 48- liter condensate receiver. The equipment was tested by processing six 48-liter batches of nonradioactive LiF- BeFZ-ZrF4—NdF3 (65-30-5-0.3 mole %) at a temperature of 1000°cC. A distillation rate of 1.5 ft3 of salt per day per square foot of vaporization surface was achieved in the nonradioactive tests. Evidences of concentration polari- zation and/or entrainment were noted in some runs but not in others. Automatic operation was easily maintained in each run, although certain deficiencies in the liquid- level measuring devices were noted. Condensation of volatile salt components in the vacuum lines and metal deposition in the feed line to the still pot are problems needing further attention. Since a postoperational inspection of the equipment showed essentially no dimen- sional changes, the equipment was judged to be satis- factory for use with radioactive material. The results of these nonradioactive tests indicate that the application of distillation to MSBR fuel salt processing is feasible. | - 1. INTRODUCTION Low-pressure distillation hés_potential application in the proc- essing of salt from molten salt breeder reactors (MSBR's). —Ifi the single-~ fluid MSBR concept, disfiliation éouid be used to adjust the-composi— tion of the fuel salt for optimum removal of'thé'lanthanides by reduc- tive extraction or for partial fecovery of valuable COmpbnents from salt streams that are to be-discarded. In the two—fluid MSBR concept, distillation could be used to separate the slightly volatile lanthanide fluorides from the other'componéntS‘of the fuel carrier salt. A program to establish the feasibility of distillation‘of highly radiocactive salt mixtures from molten salt reactors has been under way for about three years. The work has included the measurement of relative volatilities, with respect to LiF, of a number of components of interest,l’2 as well as the operation of a relatively large, semicontinuous still with non- radioactive LiF—Ber-ZrFA-NdF3 (65-30-5-0.3 mole %). The results obtained during the nonradioactive testing of the still are presented in this report. The objectives of thé nonradioactive tests described in this report were: (1) to test the distillation equipment to determine whether it would be suitable for use with radioactive salt, (2) to gain ekperience in the operation of large, low-pressure, high~temperature stills and to uncover unexpected areas of difficulty, (3) to measure distillation rates attainable in large equipment, and (4) to determine the extent to which concentration polarization and entrainment occur in this type of operation. 2. EXPERIMENTAL EQUIPMENT 2.1 Process Equipment The equipment used in the nonradioactive tests included a 48-liter feed tank containing the salt to be distilled, a 12-liter still from which the salt was vaporized, a 10-in.-diam by 51-in.-long condenser, and a 48-liter condensate receiver. This equipment is described only briefly here; a complete description is available elsewhere.3 The feed tank, shown in Fig. 1, was a 15-1/2-in.-diam by 26-in.-tall right circular cylinder'made from 1/4-in.-thick Hastelloy N. It was designed to withstand,an external pressure of 15 psi at 600°C. The condensate receiver, shown in Fig. 2, was a l6-in.-~diam by 16-1/2-in.-tall right circular cylinder having sides of 1/4~in.-thick ORNL DWG. 88-10983 @—VENT 1* SCH. 40 PIPE Np INLET 4 0.D.x19GA. TUBE FEED INLET 26 §oo0.x150A. TUBE e e ——— T o — e ] 1 183 0.0, - . — — — —— A — o — . — — . ———— — — - o T o e ain m mw. e . m wm rrer emi oe , s S e AAR, S t t s s il ) h I I l i { i I | ! 15 1D, : | 'I 72° TYR —~i | 8e | I | | | I | | | ' l ! | b ! ! | I ! : i i I | : o i ! l | \ | i ! | | W\ ‘ = fe- % {- \» THERMOCOUPLE WELL || i 2 0.D.x19 GA. TUBE =S FEED OUTLET —— 30.0.x15 oA TUBE PLAN VIEW el Fig. 1. Molten-Salt Distillation Experiment: Schematic Diagram of the Feed Tank. Dimensions are given in inches. ORNL DWG. 66-10984 VACUUM LINE 1" SCH. 40 PIPE 3 -4 ol w x w - 4 =2 o o o = @ w X - © E THERMOCOUPLE WELL 0.0.x.042 WALL Tus 0.D.x.042 WALL TUBE THERMOCOUPLE WELL 30.0.£.042 WALL TUBE LEVEL PROBE TuBt w a & o < k< . © 0 (®) z a w B 2~ s 4 W o zZx iy g° o‘ ! a2 5 0O u. e ©.D. = Schematic Diagram 4 e g o X7 5 s - /.l/fil E imm— oo a ~ o :.fl...r IIIIIIIIIIIIIIIIII ————— ]| o e e e e = e e — — —— e = e = ——— i ——— - — - t v} v — - —— . 1 o i e e e — - 0 T T oL LT T > ord = e A oo b i | a - - AL S| o < 2% ) AN nkl | |||||||||||||||| — ———— — oa - m 2 T4 3 = 291 -— ¢ -— - ™ X - — - —_— e — - mnm —_— > o in inches. s are given in imension D of the Condensate Receiver. Hastelloy N and a bottom of 3/8-in.-thick Hastelloy N. It was designed to withstand an external pressure of 15 psi at 600°C. The still and condenser are shown in Fig. 3. The still pot con- sisted of an annular volume between the vapor line and the outer wall, and had a working volume of about 10 liters of salt. Both the still and the condenser were made of 3/8-in.-thick Hastelloy N and were designed for pressures as low as 0.05 to 1.5 térr*. The design tempera- ture for the still pot and for the condenser was 982°C. All valves and piping that did not contact the fluoride salts were made of stainless steel and were housed in a sealed steel cubicle which contained pressure transmitters and vacuum pumps. All other parts of the system were made of Hastelloy N. All-welded connections were used in the portion of the piping that was operated below atmospheric pressure. 2.2 Instrumentation Correct operation of the molten salt distillation equipment depended entirely on measurements of temperature, pressure, and liquid level. The instrumentation used in making these measurements is discussed below. 2.2.1 Measurement and Control of Temperature Temperatures were measured and controlled over two ranges: 500- 600°C for the feed tank and condensate receiver, and 800-1000°C for . the still and condenser. Platinum vs platinum-10% rhodium thermocouples were used for the high—temperature meaéurements,_whereas less expensive Chromel-Alumel thermocouples were used on the feéd tank, condensate receiver, and salt transfer lines. Each of the thermocofiples (total, 48) wés enclosed in a 1/8-in.-diam stainless steel sheath, and insulated junctions were used. Four 12-point recorders were available for readout: two fqr the Pt vs Pt-107% Rh thermotouples, and two for the Chromel- Alumel thermocouples. * 1 torr is 1/760 of a standard atmosphere. s3 14 0- 41 LEVEL PROBE 1} scu. 40 pipe HANGER BRACKET SAMPLE TuSE 1} scH. 40 PIPE Fig. 3. . l ¥ i SAMPLE ntszavom—/L_ 10 2.0 sy Molten-Salt Distillation Experiment: ORNL DWG. 88-10088 THERMOCOUPLE WELL \ 1/2°0.0. TUeE mug FEED LINE 3/4"00.x.072 WALL LEVEL PROBE I3 SCH. 40 PIPE DRAIN & SPECIMEN HOLDER THERMOCOUPLE WELL ,5%0 p TUBE x19 GA. 172"0.0. TUBExI® GA. VIEW AwA Schematic Diagram of the Vacuum Still and the Condenser. Dimensions are given in inches. There were nine individually heated zones on the feed tank, still, condenser, and receiver. Heaters on each of these zones were inde- pendently controlled by a Pyrovane "on-off" controller, and the voltage to heaters in each zone was controlled by Variacs. Heaters on seven lines were manually controlled by "on-off" switches and Variacs. 2.2.2 Measurement and Control of Pressure Pressure measurements over three ranges were required: 0-15 psia for monitoring the system pumpdown at the start of a run and for monitoring the system repressurization at the end of a run; 0-10 torr for suppressing vaporization while the salt was held at operating temperatures in the still; and 0-0.1 torr during distillation. Absolute-pressure transducers (Foxboro D/P cells with one ieg evacuated) covering the 0- to 15-psi range were used to measure the pressure in the feed tank and in the still-condenser-receiver complex. An MKS Baratron pressure measuring device with ranges of 0-0.003, 0-0.01, 0-0.03, 0-0.1, 0-0.3, 0-1, 0-3, and 0-10 torr was used to measure very low pressure in the condensate receiver. The system pressure was controlled in the 0.1-10 torr range by feeding argon to the inlet of the vacuum pump. The Baratron unit pro- duced the signal required for regulating the argon flow. Pressure was not controlled in the 0-0.1 torr rénge; insteéd, the argon flow to the vacuum pfimp inlet was stopped and the pump developed as low a pressure as possible (usually 0.05 to 0.1 torr). | It was neceséary to ensure that an excessive internal pressure did not develbp in the system since, at operating temperature, pressures in excess 6f'2‘a£m would have been unsafe. This was éécomplished by usifig an absolute4preésure transmitter in the condenser off-gas line to ‘monitor the system'pressure.‘ Wheh the pressure_éxéeeded 15 psia, the argon supply was shut off autdmatically. 2.2.3 Measurement and Control of Liquid Level The -pressure differential between the outlet of an argon-purged dip tube extending to the bottom of the vessel and the gas space above the salt was used to measure the salt level in the feed tank and in the condensate receiver. Two conductivity-type level probes were used in the still for measuring and controlling the liquid level. These probes essentially measured the total condfictance'between the metal probes (that extended into the molten salt) and the wall of the still; the total conductance was a function of the immersed surface area of the probe.4 The conductivity probes (see Fig. 4) were similar to the single- point level probes that were used in the MSRE drain tanks. Tests have shown that the range of fhis type of instrument is limited to approxi- matel& 30% of the length of the signal generating section because the signal, which is nonlinear, becomes extremely insensitive to changes'in molten-salt level outside this range. A 6-in. sensing probe was used to control the liquid level between points that were 1 in. and 3 in. beléw the still pot overflow; a longer sensing probe was used to measure very low liquid levels in the still pot. Metal disks were welded to the level probes to aid in their cali- bration. These disks provided abrupt changes in the immersed surfacé area of each probe at known liquid levels. 1In operation, the signal from a probe changed abruptly when the salt level reached one of the disks. The still-pot liquid-level controller was a Foxboro Dynalog circular chart recorder-controller, which consists of a 1-kHz ac bridge-type measuring device using variable capacitance for rebalance. The proper control action (see Sect. 3) was accomplished by héving a variable dead zone imposed on the set-point adjustment mechanism. With the controller set for the desired average liquid lefiél, the argon supply valve to the feed tank was opened when the level indicator dropped 3% below the set point and was closed when the level indicator rose 37 above the set point. EXCITATION SOURCE ORNL-DWG 67-4776R{ SIGNAL AMPLIFIER AND LEVEL INDICATOR SIGNAL LEADS HEAD COVER FOLDED EXCITATION SECTION CONTAINMENT VESSEL CUTAWAY SIDE VIEW 1 P K TS TR T T s ) (XY ‘;0‘.““\“‘\“““\“‘ P S A S s N N S S S N O S A S R LS T e T T T e S oo s S P /‘SiGNAL GENERATING SECTION DISKS TO AID CALIBRATION Fig. 4. 'Simplified Schematic of Conductivity-Type Liquid-Level Probe Used to Measure Salt Level in the Still Pot. 10 3. OPERATING PROCEDURE | &-J The following sequence of operations was followed during the dis- tillation of a 48-liter batch of salt. First, the salt was charged to the feed tank from a delivery vessel through a temporary heated line. | (This line was later disconnected, and the opening was sealed.) Then the still pot was heated to 900°C, and the system was evacuated to 5 torr. Subsequently, the valve between the feed tank and the vacuum pump was closed (see Fig. 5), and argon was introduced into the feed tank in order to increase the pressure to about 0.5 atm; this forced the salt to flow frofi the feed tank into the still pot. The condenser pressure was then reduced to 0.05-0.1 torr in order to initiate wvapori- zation at an appreciable rate. At this point, control of the liquid level in the still pot was switched to the automatic mode. In this mode, salt was fed to the still pot at a rate slightly greater than the vapori- zation rate. The argon feed valve to the feed tank remained open (forcing more salt into the still pot) until the liquid level in the still rose to a given point; the valve then closed and remained in this . position until the liquid level decreased to another set point. In this manner, the wvolume in the still pot was maintained at 8.5 to 9.5 liters. As the salt vapor flowed through the condenser, heat was removed from the vapor by conduction through the condenser walls and the insula- tion, and by convection to the air. The condenser was divided into three heated zones, the temperature of which could be controlled separately when condensation was not occurring. Sharp temperature increases above the set points near the condenser entrance, and gradual increases near the end of the condenser, accompanied the beginning of distillation as the condenser pressure was decreased. Operation of heaters to keep the temperature of the condenser above the condensate liquidus température was not necessary during condensation. The salt condensate drained through a sample cup at the end of the condenser (see Fig. 3) and flowed into the condensate receiver. Samples of condensate (10 g) were taken periodically by using a windlass - (sj i 11 ORNL DWG 70~20 LEVEL MEASUREMENT | r_-- T CONTROL | . | | | ];Ei ¢ STILL | POT : ! . CAPILLARY FLOW PRESSURE VACUUM CONTROL MEASUREMENT PUMP CONTROL CONDENSER B o FEED " ' CONDENSATE | ' ~ TRECEIVER Fig. 5. Flow Diagram of Molten-Salt Distillation-Experiment. 12 to lower a small copper ladle (i.e., the type used for the MSRE) through a 1-1/2-in. pipe into the sample cup. Figure 6 shows the sampler assembly that was installed directly above the sample cup at the condenser outlet. During sampling, the sample ladle was lowered into the sample cup filled with molten salt, and retracted into position above the flange. After the sample had cooled, it was removed for analysis by closing the lower "valve and opening the flange. After operation had proceeded for approximately 4 hr at 900°C, the temperature of the still pot was increased to 1000°C. Then di#tillation was continued until about 10 liters of salt remained in the feed tank. At this}time, the condenser pressure was increased to 5 mm Hg, which reduced the distillation rate to a negligible value. The salt remaining in the feed tank was then transferred to the still pot, causing some of the material already present in the still pot to overflow into the con- denser. The resulting mixture in the still pot had a sufficiently low liquidus temperature that the still-pot temperature could be lowered to about 700°C without freezing the 4-liter salt 'heel" remaining in the still pot (after most of the still-pot contents had been transferred to the feed tank). Finally, the system pressure was increased to 1 atm. At this point, the run was complete. 4. EXPERIMENTAL RESULTS The distillation experiments were performed in Bldg. 3541, which is specially equipped for containing experiments that involve large ‘quantities of beryllium. 1Installation of the equipment was started on September 1, 1967, and the first 48-liter batch of salt was introduced into the.system on December 14, 1967. During the approximate 6-month period from the date of salt introduction until June 18, 1968, when the experimental work was complete, six batches of salt were proceSsed in the equipment. Results from the six runs are summarized in Table 1. Fig. 6. 13 Photograph of Condensate Sampler Assembly. Table 1. Summary of Nonradicactive Experiments with Molten-Salt Still Run Conditions Volume of Time Still Condenser Salt (Mggl) Dates Temp. Pressure Feed Mate:ial Distilled Re%;i§ed Purpose Remarks {°C) (torr) (liters) c-1 2/5/68- 990 0.06-0.5 LiF-BeF,-2rF, 35 83 To gain operating expe- 2/9/68 (65-30-5 mole %) rience and to determine effect of condenser pres- sure on vaporization rate. c-2 2/26/68- 1005 0.07 LiF-BeF,-ZrF, 32 40 Same as MSS-C-1. Metallic deposit 2/28/68 {65-30-5 mole %) restricted salt feed line. c-3 3/26/68- 1004 0.075 LiF-BeFy-ZrF,;-NdF, 26.4 31 To investigate polari- 3/27/68 (65-30-5-0.3 mole %) zation and entrainment. C-4 4/8/68- 980-1020 0.065 Distilled salt from 28 45 To determine the effect ZrF; condensa- 4/10/68 MSS-C-1 of temperature on vapor- tion in vacuum ization rate. line resulted in abnormally low ‘ rates. c-5 5/27/68~ 950-1025 0.06 LiF-BeFy-ZrF;-NdF3 32 41 To determine the effect 5/28/68 (65-30-5-0.3 mole %) of temperature on vapor- ization rate. c-6 6/11/68- 1000 0.07 LiF-BeFy-ZrF,~NdFq 32 53 To determine whether 6/13/68 {65-30-5-0.3 mole %) polarization becomes evident after long operating times. ZrFA condensation in vacuum line stopped distilla- tion. Heating the line to 950-1050°C removed the ob- struction. 71 15 During the period of operation, the feed tank, the receiver, and the lower zones of the condenser were maintained at temperatures of 550 to 600°C for 185 days. A summary of the times during which the still pot was maintained at various temperatures ranging from 25 to 1025°C is given below: 25°C 12 days 500-650°C 56 days 750-875°C 110 days 900-1000°C 100 hr 1000-1005°C 160 hr 1005-1025°C 45 hr In general, the performance of the equipment was satisfactory and all design criteria were met or surpassed. Distillation rates were higher than expected; automatic operation of the equipment was easily maintained in spite of certain difficulties with the still-pot level probes. These difficulties are discussed in Sect. 4.3. Air inleakage during normal operation was insignificant. All of the pneumatic components of the instrumentation system performed reliably, although the electronic components presented some problems. 4.1 Measurement of Distillation Rates Distillation rates were determined by observing the rate of rise of liquid level in the condensate receiver. The measured rates and the operating conditions under which they were observed are summarized in Table 2. The distillation rate is limited by friction losses in the passage- way bétween the vaporizétion and condensation'surfaces. The force that drives the vapor through this path is-thé-difference between the vapor pressure of the'liqdid'in the still pot and the pressure at the con- . denser outlet. 'Thus,‘the'distillation rate can be increased either by increasing the temperature of the still pot (which, in turn, increases the salt vapor pressure) or by decreasing the condenser pressure since, 16 Table 2. Summary of Distillation Rate Measurements Run - Still-Pot Condenser Distillation No. Temperature Pressure ' Rate (MSS-) (°C) (torr) (ft3/£t2- day) c-1 | 990 0.5 | 1.15 c-1 990 0.3 1.20 c~1 990 0.055 | 1.25 C~2 1005 0.07 1.50 c-3 1004 - 0.075 1.56 C-4 1020 0.065% 1.63 C-5 950 0.08 | 0.66 C-5 1000 0.08 S 1.21 Cc-5 1025 0.08 1.95 C-6 1000 0.08 1.40 3This may not be the actual condenser pressure since a ZrF4 Plug formed in the vacuum line during this run. in each case, the driving force for the vapor flow has been increased. If the condenser pressure is very low with-respect to the vapor pressure of the salt, the distillation rate should reflect the variation of salt vapor pressure with still-pot temperature. Figure 7 shows the effect of still-pot temperature on distillation rate for the runs in which the condenser pressure was.below 0.1 torr. The distillation rate is expressed as cubic feet of salt distilled per day per square foot of vaporization surface. A more useful correlation of distillation rate is one involving the condenser pressure and the vapor pressure of the salt, since this type of correlation could be used to estimate the performance of the still with other salt systems. Such a correlation is shown in Fig. 8. The vapor pressure of the salt (pl) was assumed to be the vapor pressure of 90-7.5- 2.5 mole Z LiF-BeF —ZrFa, which is given in ref. 5. A mixture of this 2 composition produces a vapor having a composition that is approximately 17 ORNL DWG 68-944| RI TEMPERATURE, °C 1040 1030 1020 1010 1000 990 980 970 960 950 4.0 —r 1. 11Tt vt 7T T 1" 3.0 — ’N ol nm g’f 2.0 \. - E o . ® . - 229 E . InR= |8.3336--T—,?K—°- « . o =z o 2 LOF - | = - = E 0.8 |- o - - @ 0.6 r - 0.4 : L ] I i | 7.6 7.7 7.8 7.9 8.0 8.1l 8.2 041K Fig. 7. Effect of Still-Pot Temperature on Distillation Rate for Runs in Which the Condenser Pressure Was Below 0.1 torr. VAPORIZATION RATE, ft° salt/day-ft2 18 ORNL DWG 70-21 03— 0.25 0.5 | 11 b b1l | 1 | Fig. Pressure 015 02 03 04 0506 08 10 1§ 2 25 (pf—pf),[torr ]2 8. Dependence of Distillation Rate on Estimated Still-Pot (pl) and Pressure at End of Condenser (pz). 19 ./ 65-30-5 mole % LiF-BeF,~ZrF, and hence should approximate the composi- 2 tion of the material in the still pot at steady state. The correlation in Fig. 8 was thought to be applicable because a steady~-state mechanical energy balance for the isothermal flow of an ideal gas through a conduit of constant cross section shows that the flow of gas is a function of the difference between the squares of the upstream and downstream pressures. This can be seen from the following development. A steady-state mechanical energy balance for a flowing fluid yields: 2 v2f D vdv+-p]:dp+ az = 0 , (1) where v = velocity of fluid, = density, = pressure, Fanning friction factor, = diameter of duct, v N O M T T 0 = distance along duct. By assuming that the fluid is an ideal gas, that is, p = pM/RT , ' (2) where T = absolute temperature, M = molecular weight of the gas, R = gas constant, and by using the macroscopic mass balance equation which requires that ov = (p; v{) » (3) ‘where the subscript 1 refers to the entrance to the duct, Eq. (1) becomes: —%dp+%—1’——Ld 7 +————2fDdz= 0. (4) (py vy) 20 Since the friction factor, f, depends only upon the Reynolds number (which would be constant for an ideal gas at constant temperature), Eq. (4) can be integrated between the entrance to the duct (point 1) and the end of the duct (point 2) to yield: =0, () _}:_1_{ RT + 2fL My 2 D 1 where L is the length of duct. By solving Eq. (5) for vy and multiplying it by Py> the following expression for the mass flow rate is obtained: c = \/;1RTT V (plz - Pzz} /5 -l 1 (6) This expression shows the dependence of mass flow rate on the difference between the squares of the upstream and downstream pressures. Although the temperature was not constant and the cross section of flow was not uniform in the runs made to measure distillation rates (Table 2), a fair correlation of all the data was obtained (see Fig. 8). The calculated vapor pressure of the salt, Py> ranged from 0.70 to 1.28 torr, and the measured condenser pressure, Pys ranged from 0.055 to 0.5 torr. The results suggested that an exponent of 0.41 might fit the expression for mass flow rate somewhat better than the value of 0.5 suggested by Eq. (6); this discrepancy is proEably the result of not including the logarithmic term in the correlation. 21 4.2 Measurement of the Degree of Separation of NdF3 from LiF—Ber-ZrF4 Carrier Salt In three runs (MSS-C-3, -5, and -6), a number of condensate samples were taken and were analyzed for all salt components to determine the effectiveness of the still for separating LiF-Ber-ZrF4 carrier salt from NdF3, which is representative of the lanthanide fission products. The ease with which NdF3 can be separated from the carrier salt is conveniently expressed in terms of the relative volatility of NdF3 with respect to the least volatile carrier salt component, LiF. The relative volatility of NdF., with respect to LiF is defined as: 3 * / * INaF.. *NdF 3 3 ¢ = S (7) Yrir/ *LiF * * where YNdF and YiiF are the mole fractions of NdF, and LiF, in vapor which is in equilibrium with liquid containing x;d;3 and x;iF mole fractions of NdF3 and LiF, respectively. The asterisks emphasize that the concentrations are to be measured under equilibrium conditions. Relative volatilities for the other components of the system are defined similarly. Separation of a component from LiF by distillation is possi- ble if the relative volatility of the component with respect to LiF is not equal to 1l; the separation becomes easier as the deviation of the relative volatility from 1 increases. The relative volatility of NdF3 with respect to LiF has a value of 1.4 x 10_4,'which indicates that these -two components Eould be separated easily in a still that is equivalent'to a single equilibrium stége. In practice, a single physical stage (such as the still which Wasioperated in this study) may not be equivalent to an equilibrium stage because of entraimment, concentration polarization, or other factors. In assessing the effectiveness of a still, it is convenient to define an effective relative volatility that reflects nonequilibrium conditions present during the still operation,as follows: 22 yNdF3/deF3(avg) o = ’ (8) OBS yLiF/xLiF(avg) where the y's are mole fractions determined from condensate analyses and the x's are mole fractions in the still pot averaged over the entire still-pot volume. The performance of the still can be judged by the ratio of the effective relative volatility (aOBS) to the relative volatility (o), which will be denoted as R. Then the deviation of R from 1 is a measure of the deviation from equilibrium conditions in the still. The quantity R may deviate from 1 because of several reasons, including: (1) concentration gradients in the still-pot liquid (concen- tration polarization), (2) entrainment of droplets of still-pot liquid into the vapor leaving the still pot, or (3) contamination of the con- densate samples by small amounts of material having high NdF, concen- trations. These possibilities are discussed below. ’ Entrainment of small amounts of still-pot liquid into the wvapor leaving the still pot would cause the observed concentration of NdF3 in the vapor to be much higher than the equilibrium concentration. This, in turn, would cause the value of R to be greater than 1. In the absence of concentration polarization or other effects, a material balance gives the following relationship between the value of R and the fraction of the condensate that is entrained liquid: 1+.g(.jfi.§ R = = , (9 1+f(_r=i:r:) TLiF where f = moles of entrained liquid per mole of vaporized material, xLiF = mole fraction of LiF in the liquid, Yiip mole fraction of LiF in the vapor, R 0 relative volatility of NdFB, with respect to LiF, at equilibrium, as given by Eq. (7). 23 For the present system, the value of the X1 o ratio is about 1.6. /Yisp With this value for the ratio, entrainment of only 0.001 mole of liquid per mole of vapor would result in a value of about 12 for R. Concentration polarization would also cause R to have a value greater than 1. This can be explained as follows. As the more-volatile materials are vaporized from the surface, the NdF3, which is only slightly wvolatile, is left behind. Thus, the NdF3 will have a higher concentration at the surface than in the bulk of the liquid. In turn, the concentration of the slightly volatile NdF3 vaporization occurs from liquid with successively higher NdF will gradually increase in the vapor since further 3 concentra- tions. Hence, the concentration of NdF, in the vapor will be higher than 3 would be the case under equilibrium conditions, and R will have a value greater than 1. The extent to which R deviates from 1 because of concentration polari- zation depends on the dimensionless group D/vL, which qualitatively repre- sents the ratio of the rate of diffusion of slightly volatile NdF., away 3 from the liquid-vapor interface to the rate at which this material is transferred to the interface by convection. 1In the dimensionless group, D is the effective diffusivity of NdF, in the liquid and is a measure of the amount of mixing in the liquid, VBiS the average velocity of the liquid moving toward the interface, and L is the distance between the vaporization surface and the point at which feed is introduced. As the value of this group increases, the value of R will approach 1, as shown in Fig. 9. The method for calculating these curves is given in Appendix Al Contamination of the condensate samples by small amounts of material containing high concentrations of NdF3 could also result in R values greater than 1. Contamination of the samples during analysis is not considered likely. However, it is possible that salt‘having a high liquidus temperature and é high NdF3 concentration could have remained on the condenser wall after the still-pot flushing operation (see Sect. 3) at thé end of runs MSS-C-3 through MSS-C-6. If this had been the case, the material would have been washed from the condenser walls during the following run and would have contaminated the condensate samples. DISTILLATION RATE, liters/hr 24 ORNL-DWG-70- 4566R! 14 I T T | . T I 35 |lo 0.286 - v R’Qoas/fl | | | I | | o 05 1.0 15 20 25 30 V, NO. OF STILL POT VOLUMES OF CONDENSATE COLLECTED 35 Fig. 9. Distillation Rate and Ratio of the Observed Relative Volatility to the Actual Relative Volatility of NdF LiF, as Measured in Run MSS-C-3. 3 with Respect to 25 Experimental values for NdF3 concentrations in the condensate do not enable one to distinguish unambiguously between concentration polari- zation and entrainment; however, they do allow assessment of the importance of these effects if the level of condensate contamination is not too great. Figures 9 through 11 show calculated R values for runs MSS-C-3, -C-5, and ~C~6, respectively. 1In calculating R, the values of dypg Vere calcu- lated from Eq. (7), using analyses of the condensate samples. The values for y were measured values; the values for x were calculated from "NdF_ (avg) a material balance on NdF3 in the still pot in Wwhich it was assumed that a negligible amount of NdF3 was removed in the vaPor. The value of xLiF(avg) was estimated by calculating the liquid composition in equilibrium with the measured vapor compositions, assuming that the relative volatilities (with respect to LiF) of BeF2 and ZrF4 were 4.7 and 10.9, respectively. .The relative volatility for BeF2 was ob;ained from measurements made in small recirculating equilibrium stills. The relative volatility for ZrF4 was measured in run MSS-C-1 and is probably only wvalid for a still-pot composition of 65-30-5 mole % LiF—Ber-Zth. Also shown in these figures is the variation of distillation rate with time during each of the three runs. Values of the group D/vL which best represented the calculated R values from each of the runs (Figs. 9-11) were chosen by trial and error by assufiing that any deviation of R from 1 was caused entirely by concen- tration polarization. As seen in Fig. 9, a value of 0.286 for D/vl was found to generate a smooth curve that best represented the measured R values in run MSS-C-3. Contamination of the condensate was not possible in this run since no NdF3 had been used in the previous run. In run MSS-C~5, the increase in the R values when the distillation rate was suddenly increased was most closely represented by the assumption that D/vL changed from 0.227 to 0.0215 when the rate increased. The ldw initial R values indicate that contamination of the condensate from material on the condenser walls was not important during this rum. .In_run MSS-C-6, the last six'expefimental points correspond to a value of 0.051 for D/vL. There appears to be no straight-forward explanation for the high values of R observed in the first four samples taken during 26 ORNL DWG 70 4563RI| | | I T T T DISTILLATION RATE, Iiters/hr 90 80} 70l , - 60| . sol- — a0} - 30 20 H O O JODYWO L1 1] ] l o 1ol | ] 1 | 1 1 ' 05 10 15 20 25 30 35 V, NO. OF STILL-POT VOLUMES OF CONDENSATE COLLECTED Fig. 10. Distillation Rate and Ratio of the Observed.Relative Volatility to the Actual Relative Volatility of NdF3 with Respect to LiF, as Measured in Run MSS~C-5. 27 ORNL - DWG - 70 -4564RI | I I I I \iters / hr DISTILLATION RATE, 1000 * I TTir L1141l | 1 100 ® Lol T TUENT ® 1 1 D L + 0.051 I ® o 1 ob——— 1 1 L L Yo 10 20 30 a0 50 V, NO. OF STILL- POT VOLUMES OF CONDENSATE COLLECTED Fig. 11. Distillation Rate and Ratio of the Obsefved Re1ative Volatility to the Actual Relative Volatility Of'--NdF'3 with Respect to LiF, as Measured in Run MSS-C~-6. - 28 this run. The value of R was about 1000 in the first sample (0.8 still- pot volume distilled) and decreased thereafter. The vaporization rate was very high initially and may have caused both concentration polariza- tion and entrainment. If we assume that the high R value in the first sample was the result of concentration polarization only, the effective difquiVity in the still-pot liquid would be 3.4 x 10-6 cm2/sec, which is an order of magnitude lower than reported values6 of molecular diffusivities in molten salt at temperatures less than 750°C. Thus, concentration polarization alone cannot account for the high R values observed. Of course, it is possible that these high values resulted from contamination of the condensate. If salt with a high liquidus temperature and a high NdF3 concentration had remained on the condenser walls after the contents of the still pot were flushed out at the end of run MSS-C-5, this material would have been washed into the sample reservoir early in run MSS-C-6. Values were calculated for the effective diffusivity in the still- pot liquid by using values of the group D/vL given above and values for v and L. Values for the average velocity (v) were determined from dis- tillation rates, whereas those for the distance (L) between the feed inlet and the liquid surface in the still pot were known. The range of 4 -4 to 16 x 10 one to two orders of magnitude greater than reported values of molecular diffusivity values was 1.4 x 10 cmzlsec; these values are diffusivities in molten salts and are equivalent to those which would be expected if convective mixing were occurring in the still pot. It is considered likely that concentration polarization and/or entrainment effects were observed during the operation of the still, although contamination of the samples with NdF3 from the condenser walls prevents one from drawing firm conclusions. There were some periods of operation in the experiments described here in which these effects were within tolerable limits (in run MSS-C-3, R was never higher than 5, and. in the early part of the run MSS-C-5, R was also below 5); thus there is evidence that, by careful equipment design and proper choice of operating conditions, concentration polarization and entrainment can be held to acceptable levels. Further investigation would be required to determine the proper operating conditions and equipment design. 29 4.3 Difficulties Surprisingly few operational problems were experienced during the experimental program. However, certain significant difficulties were encountered, as discussed below. The condensation of ZrF, and unidentified molybdenum compounds obstructed the vacuum line oi two occasions. The first restriction occurred during the fourth run and was removed by cutting into the vacuum line; the second restriction occurred during the last run and was removed by heating the vacuum line to 950-1050°C, thereby redistrib- uting the material. Analysis of the material from the first deposit showed it to contain 39.4% zirconium and 11.6% molybdenum, with fluoride and oxide being the major anions. During the second run, the salt feed line to the still became obstructed. After the run had been completéd, the line was cut and a 5- to 10-g metallic deposit, consisting mainly of nickel and iron, was found at the point where the feed line entered the still. This line was replaced, and the still was operated for four additional runs. At the end of the series of experiments, the feed line was again removed and another metallic deposit was found at the same point. The composition and the appearénce of the second deposit were similar to those of the first; however, the open cross-sectional area at the point of the second deposit was still about 507 of that of the unobstructed tubing. The cause for the metal depbsition in the feed line is not known. Two possible sources of the deposited material are: (1) suspended metals and/or dissolved fluorides introduced with the feed salt, and (2) corrosion products. The possibility that corrosion of‘the system components may have been a factor is suggested by the composition of the deposits [approximately 0.9 wt % cobalt and 0.7 to 2 wt % molybdenum (both elements appear in 30 Hastelloy N)1, and of the plug in the vacuum line (high molybdenum con- tent). The extent of corrosion that would be necessary in order to pro- duce such deposits would not have been detected by wall thickness measure- ments (see Sect. 6) if the corrosion were general in nature. A possible method for reducing and depositing dissolved fluorides is based on the observation that higher-valence fluorides are, in general, more volatile than 1owgr—vélente fluorides of the same element. This condition could cause the still pot to be reducing with respect to the feed salt and, in turn, to.promote reduction and depositiom of\felatively noble metals at the entrance to the still pot. ‘Because the level probes were unexpectedly sensitive to changes in salt temperature and still-pot pressure, they could not be calibrated for exact still-pot conditions. To ensure that the still pot did not over- flow during the filling operation, we added salt until the level reached one of the calibration disks and a discontinuity in the recorder signal was noted. This provided a definite measurement of the saltllevel, al- though it was lower than the nominal operating level. Satisfactory automatic operation in the vicinity of this signal discontinuity was then possible. 5. CORROSION TESTS Corrosion specimens of Alloy 82, Hastelloy N, TZM, Haynes Alloy No. 25, and Grade AXF-5QBG graphite (see nominal compositions in Table 3), supplied by the Metals and Ceramics Division, were suspended in the wvapor and in the 1iquid in the still pot during the six runs. The specimens, 1/16 in. thick x 3/8 in. wide x 3/4 to 2-1/2 in. long, were arranged in two stringers mounted on a Hastelloy N support fixture that was tack- ‘welded to the Hastelloy N dip line of the still pot. The arrangement and position of each stringer were such that specimens of each material were eXpoéed to both vapor and liquid; a Hastelloy N specimen was centered at the vapor-liquid interface. Figure 12 shows the two stringers on removal from the still pot after the nonradioactive tests. An unexposed stringer is also shown for comparison. Table 3. Nominal Compositions of the Corrosion Specimens Exposed in the MSRE Vacuum Distillation Experiment , Element - Material Co Ni Mo Cr W Fe Ti Zr C Mn Alloy 82 - Bal. 18 0.05 - — - - - 0.2 Hastelloy N ‘Trace 72 16 7 - 5 - - 0.06 - TZM | - - Bal. - ~—— - 0.5 0.1 0.01 — Haynes Alloy No. 25 Bal. 10 — 20 15 3 — -- 0.10 1.5 Grade AXF-5QBG - - - - -— == - - 100 - isotropic graphite 1€ 32 - ORNL DWG 69-98 Botfom 1 Deswgn' Liquid ' Interfoce Legend A and B —Tested specimens as removed from the support. C—-Untested specimens lighted to accentucte the machmmg paiterns B2~ Alloy 82 (Ni-18 Mo-0.2 Mn-0.05 Cr) N- Hastelloy N (72 Ni-16 Mo-7 Cr-5 Fe~0.06 C) 25-Haynes Alloy No. 25 (Co-20 Cr-15 W-10 N|-3 Fe-1.5 Mn-0. 10 C) TZM-TZM (Mo-0.5 Ti-0.1 Zr- 0.01 C) - G -Grade AXF-5QBG graphite Fig. 12. Corrosion Specimens Removed from Still Pot in MSRE Distillation Experiment. Unexposed specimens are shown for comparison. -Vapor _ . op 33 At the conclusion of the nonradioactive tests, the specimens were returned to the Metals and Ceramics Division for examination.7 The resistance of the metals to attack in both the vapor and the liquid zones was found to be in the following order: Alloy 82 > Hastelloy N > TZM > Haynes Alloy No. 25 > Grade AXF-5QBG graphite. The specimens exposed in the liquid zone appeared to have been attacked more severely than those located in the vapor zone. The Alloy 82, Hastelloy N, and TZM specimens appeared to be essentially unchanged. These observations are only qualitative because (1) they are based only on visual examina- tion, and (2) oxygen in the system (introduced as a result of a heater failure described below) probably modified some of the corrosion results. The following observations were made regarding the materials located in the liquid zone. Some reaction between the salt and the Alloy 82 and Hastelloy N specimens is suggested because their surfaces were sufficiently etched to make the grain structure clearly visible. The Haynes Alloy No. 25 appeared to be severely cracked, and it broke easily. The graphite specimens in the liquid zone were missing. Crystal-like metallic deposits appeared to be clinging to the joining wires in this region. The presence of air, which was introduced into the system when a tubular heater failed (in such a manner that a hole was melted in an argon dip line while the system was at low pressure), accounts for the poor performance of the graphite. The air that was introduced while the system was pressurized remained in the system for about 500 hr; the still- pot temperature during this period was near 700°C. (No other air was introduced during this period.) The attack on the graphite specimens is probably the result of oxidation during this period. The loss of the graphite specimens in the liquid zone (but not in the vapor zone) is probably_due to a '"washing effect" of the molten salt on the damaged specimens as the still was filled and emptied. Although the Hastelloy N specimens at the design vapor-liquid interface had gray matte surfaces, no changes in the appearances of the specimens were clearly attributable to the presence of a wvapor-liquid interface. 34 6. POSTOPERATIONAL INSPECTION Since the equipment employed in the nonradioactive tests described in this report was to be used later with radioactive materials, it was inspected thoroughly after completion of the tests. Radiographic and ultrasonic measurements of wall thickness were made at 225 points on the vessel surface. In addition, length and diameter measurements were made between selected locations. (Center-punched tabs at these locations were provided for these measurements.) The postoperational measurements wére compared with similar measurements made on the equip- ment before operation. Measurements were concentrated in regions where the highest stresses were expected. Drawings M-12173-CD-019D, M-12173-Cp-020D, and M-12173-CD-021D (Appendix B) show the 1ocations,pf the 225 points where wall thickness measurements were made, as well as the locations of points between which length and diameter measurements were made. "Also shown are the pre— and postoperational measuiements. A comparison of the two sets of thickness measurements showed an average decrease of 1.6 mils in wall thickness, with both positive and negative deviations from the original thickness. The largest differences were +9 mils and -8 mils. The change in distance between two points about 50 in. apart was 0.026 in., which is not considered significant. There was some indication that the still pot had dropped to a slightly lower position, although the rotation of lines between points on opposite ends of the condenser was less than 0.5°. Visual inspection of the inside of the still pot showed the metal to be in good condition; the walls were shiny, and no pitting or cracking was evident. Radiography also showed no evidence of physical change. We concluded that the equipment was in satisfactory condition for use with radioactive materials. 35 7. CONCLUSIONS The following conclusions have been drawn as a result of the experi- mental work discussed above: (1) (2) (3) (4) (5) (6) A relatively large molten-salt distillation system has been operated successfully. Although some problems must be solved before a distillation unit can be incorporated into an MSBR processing plant, the distillation of irradiated mixtures of molten fluorides has been demonstrated to be feasible. The measured distillation rates are adequate to per- mit the use of distillation as a process step. For operation under conditions in which the vapor pressure of the still-pot material is 1 mm Hg or greater, distillation rates of at least 1.5 ft3 of salt per day per square foot of vaporization surface can be obtained. The distillation rate was limited by frictional losses in the vapor passageway; there- fore, higher rates might be attainable under the same operating conditions by careful design of the salt vapor flow path. Evidences of concentration polarization and/or entrainment were seen during some runs in these experiments. The fact that they were not seen in all runs indicates that further investigation could disclose the conditions under which a still could be operated with concentration polarization and entrainment held to acceptable levels. A postoperational inspection of the still showed only minor changes as the result of operation and indicated that the equipment was in satisfactory condition for use with radioactive materials. The cause of repeated metallic deposits in the salt feed line must be determined since such depositions would disrupt the long-term operation of distillation systems. : . The condensation of volatile salt components and corrosion products in the vacuum lines must be prevented if long-term operation is to be feasible. 36 (7) More-reliable level-measuring devices for controlling the still-pot liquid level should be provided. A method more desirable than the one used for this experiment would consist of several probes (of the type described in this report) used at various salt depths giving a control signal which would change stepwise rather than smoothly. A number of such probes, located at closely spaced intervals, would permit sufficiently accurate measurement and control of liquid level to make operation of a stil]l feasible. 8. ACKNOWLEDGMENTS The authors gratefully acknowledge the assistance of the following individuals, whose help in operation of the equipment and interpretation of the observations were indispensable: B. G. Eads and B. C. Duggins, Instrumentation and Controls Division, for consultation about and repair of liquid-level and pressure instrumentation; Anna M. Yoakum, Analytical Chemistry Division, for analysis of the condensate samples; W. H. Cook, Metals and Ceramics Division, for examination of corrosion specimens and interpretation of results; B. A. Hannaford for advice and assistance in photographing the results of the experiment; H. D. Cochran for assist- ance in calculations and interpretation of measurements; V. L. Fowler, R. 0. Payne, and J. Beams, technicians in the Unit Operations Section of the Chemical Technology Division, for operation of the distillation equipment; D. M. Haseltine, co-op student from the University of Missouri, for assistance during equipment installation; and J. L. Wade, pipefitter assigned to Bldg. 3541, for his diligence and ingenuity in keeping the equipment in excellent operating condition. « 1. 10. 37 9. REFERENCES J. R. Hightower, Jr., and L. E. McNeese, Measurement of the Relative Volatilities of Fluorides of Ce, La, Pr, Nd, Sm, Eu, Ba, Sr, Y, and Zr in Mixtures of LiF and BeFZ, ORNL-TM-2058 (January 1968). F. J. Smith, L. M. Ferris, and C. T. Thompson, Liquid-Vapor Equilibria in LiF-BeF, and LiF-BeF,~ThF, Systems, ORNL-4415 (June 1969). W. L. Carter, R. B. Lindauer, and L. E. McNeese, Design of an Engineering—-Scale, Vacuum Distillation Experiment for Molten Salt Reactor Fuel, 0RNL-TM—2213 (November 1968). M. W. Rosenthal, MSR Program Semiann. Progr. Rept. Feb. 28, 1967, ORNL_ll'llg, p- 76' R. B. Briggs, MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, p. 126. Milton Blander (ed.), Molten Salt Chemistry, p. 592, Wiley, New York, 1964, M. W. Rosenthal, MSR Program Semiann. Progr. Rept. Aug. 31, 1968, ORNL-4344, p. 282. R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena, 1st ed., p. 502, Wiley, New York, 1960. R. V. Churchill, Operational Mathematics, 2nd ed., p; 183, McGraw- Hill, New York, 1950. L. A. Schmittroth, Communications of the ACM 3, 171 (March 1960). 38 APPENDIXES 10. i 39 10.1 Appendix A. Derivation and Solution of Equations Describing Concentration Polarization As a salt mixture is vaporized from the surface of the still-pot liquid, the concentration of the less-volatile component (e.g. NdF3) at the interface will increase above its average concentration in the still pot. Therefore, the effectiveness of a still for separating NdF3 from a feed salt will gradually decrease, since the concentration of NdF3 in the vapor will increase as the concentration of-NdF3 at the surface increases. Relationships defining the extent of separation to be expected with concentration polarization have been derived, and a method for calculating R, the ratio of the concentration of NdF, in the vapor 3 to that in the liquid, is explained. We will assume that the salt mixture in the still pot is composed of only LiF and NdF3. about 90% LiF, with the remainder being BeFZ, ZrF4, and NdF3.) Calcula- (Actually, the steady-state composition will be tions are simplified considerably for this assumed binary mixture. Figure A-1 is a schematic diagram of the model used for estimating the effect of concentration polarization. At any level, z, in the still pot, the concentration of NdF_ is determined by the following equation: 3 . dc oN _R _ __Rz , (10) ot dz where g = molar concentra;ion of NdF3: N, = molar flux of NdF, in the z direction, t = time. | The_flux of NdFB, NRz’ is related to the concentration of NdF3 by the following equation: 9X R Np, = XR(NRz + NLz) cD == (11) 40 ORNL DWG 67-11699 Vopor removal stream . moles Xg (1,0) =mole froctio! F —sec ' of rare-ecorth fluorid ot Xr (1,8) =mole fraction of at liquid surface rare - earth fluoride _\ ) in vapor , AAAAAAALLA z=L, g:l L 2+ § - 20, §=0 X (0,06) =mole fraction oi’X rare-earth fluoride moles F sec Xj=mole fraction of rare~- ecrth flvoride in feed siream Fig. A-1. Diagram of Model Used for Estimating Effect of Concen- tration Polarization. , & 41 where NLz = molar flux of LiF in the z direction, Xp = mole fraction of NdF3, ¢ = molar density of mixture, D = effective diffusivity of NdF, in an LiF-NdF, mixture. 3 3 Substituting Eq. (11) into Eq. (10) and dividing by the molar density, ¢, which is assumed to be constant, yields: ox X 3 x R _ _ . _R R 3t = v 3z + D 2 ’ (12) 3z NRz + NRL : where v = — < the velocity of salt mixture in the still pot. Equation (12) must be solved with the following boundary conditions: (1) At t =20, Xp = X = the initial NdF3 concentration, which is equal to the feed composition. (2) At z = 0, an NdF, balance over the boundary between the _ 3 feed stream and the still pot gives: ax 32 |,e0 v = 2 [x (0, ©) —x,1 . (13) (3) At z = L, an NdF, balance over the vapor-liquid interface 3 gives: ' v wl_, T A ¥ x (L, t) , (14) where a = the relative volatility of NdF3 with respect to LiF. In deriving the third boundary condition, the following approximation is used: 42 where YR is the mole fraction of NdF3 in the vapor phase. This approximation is valid for a binary system in which the low- volatility component is present at low concentrations. Equation (12) and its associated boundary conditions can be put in dimensionless form by making the following substitutions: | Xp T X, o = — = dimensionless NdF3 concentration, i g = g%- = dimensionless time, which is also equal to the number of volumes processed by the still in time t (to see this, multiply numerator and denominator by Ac’ the cross section of the still), Z . . ; £ = I = dimensionless distance. With these substitutions, the differential equation and its boundary conditions become: oq d g o0 aa - 8 7o T TF (15) ] 352 £ 6 =0: o=20, | (15a) £ = 0: —3—% = bo (0, 6) , (15b) leco a0 £ = 1: e = ¢[o(l, 8) + 1] , - (15¢) g=1 where a = ‘é% b = 1/a, c = (1—o)b. ¢ * 43 With the substitutions shown above, the quantity R described in Sect. 4.2 is given by: OBS® "ACTUAL yR/XR(L, t) XR(avg) 1 L = x (z, t) dz L R 0 or . xi[o(l, 8) + 1] [o(L, 8) + 1] R=—"n =) (16) X, I [o(g, 6) + 11dt avg 0 where 1 e = - cavg( ) Jo o(g, 6) d§ Method of Solving Equations. — For the solution of Eq. (15), with Egqs. [15(a)-15(c)] as boundary conditions, the parameters a, b, and c are assumed to be constant. By taking the Lapléce transform of Eqs. (15), and Egqs. [15(a)-15(c)], the following ordinary differential equation and boundary conditions are obtained: 2— - d c(&é s) __ddég, s) —'s 0(E, ) =0 (17 dg %2' = b 5(0, s) (17a) | - do T a—g—_- = C O'(l, VS) + s °* (17b) / 44 where o (£, s) = f;’ o(E, e)e'Se and s = the Laplace transform variable. Equation (17) has the solution: do, the Laplace transform of o(, 8), oc(E, s) = Ae + Be , (18) where _ 1+ Y1l + 4as 51 2a » _ 1 - V1 + 4as Yo 2a * When Eq. (18) is substituted into Eqs. [l17(a)] and [17(b)], the constants A and B can be determined; then EYE, s) is found to be: Yo8 Y,& el(vy; —ble > = (v, —b)e = ] o(E, s) = — . a9 Yo 1 S[(Yl - b)(wr2 —c)e " — (YZ —b)(v; — cle 7] Since we also desire oavg (8) = 4} oc(t, 0) dg, we can find the Laplace transform of this quantity by integrating Eq. (19) with respect to E. This integration yields: Y Y 2 21 = (1,2 = bre = D) Yo Yy . (20) SYlYZ[(Yl —b)(y, — ce " — (v, — b)(wr1 — c)e 7] C[(le — by,)(e E;vg(s) = The quantity cavg(fl) is obtained from Eq. (20) by performing the following integration in the complex plane: 1 et+ie 8 Uavg(e) = "2—“'{[ o (s)e avg ds , (21) €= " 45 where i = v¥~1, and € > 0 and constant. Equation (21) can be written in the following equivalent form by recog- nizing that s = ¢ + iw and ds = idw: ed [ | oavg(e) 35;-I‘m Oan(E + iw)[cos (w8) + i sin (wbd)] dw . (22) Because the integral in Eq. (22) converges to zero when 6 < 0 (ref. 9), 10 it can be shown™ that cavg(e) is given by a pair of equations: 0 _ 2e — . cavg(e) == I: —Im[cavg(e + iw)] sin (w8) dw (23a) 9 £0 o _ oavg(e) = i_ IO Re[oavg(s + iw)] cos (wh) dw , (23b) where Im[gévg(e + iw)] = the imaginary part of the function o (e + iw), Re[E- (e + iw)] = the real part of the function o (e + iw). avg avg Equation [23(a)] was chosen for evaluating cavg(e). The integral in this equation was computed numerically by using the CDC 1604—A computer to evaluate.Im[Eévg(e + iw)]. The details of the numerical integration can be found in ref. 10. The quantitj R is calculated by finding o(l, 6) and cavg(e) for the ~same value of 06 and substituting into Eq. (16). The quantity o(l, 8) is found by inverting the transform givén by Eq. (19) with £ = 1, and the quantity o avg (6) is found by inverting the transform given by Eq. (20). The described inversion technique is requlred for both transforms. 46 10.2 Appendix B. Drawings Showing Poétoperational Wall-Thickness and Dimensional Measurements The drawings included in this appendix contain a complete tabulation of the wall-thickness and dimensional measurements made immediately after the still assembly was constructed and after the nonradioactive testing of the equipment.was concluded. 47 : SRaw 'D‘,- t-.u- e ) PL‘\N RU\NS " “l'.n P-.-Q—llnl-.-so &-T. PLAN S.. . ~ Ny ax +— $us % 7\/\'!.1& \xu / o~ 30° g ot 4 <3°' ?‘E}_t/’( w \ Yoo "< PLAN Row " *d3 Ry ~ ]\‘ / F K e ‘ML T N Lo L Loy + o AW s . : ‘ + \Kl o Ma K2 l _ PLAN . ROW “BbAS Row “A”" & Paiars & 45° I Awamy On 2" Rap,Miasumen Hosutouyauyt Frea Couan Of Huan Rew “B" 12 Powrs @ %0° APARY Om 4" RaD, MirtoniG Horizoutarwy Fasw Coutea OF vdas Row "¢~ & Points Asg Srswn @AB® Ol &7 Rmo, i Mueasunin HorizonTauly Faema : Cantan’ O Hano ! PLANM OF STiILL HEAD A < X — _'_x e Rl PLA ROW “y " _ +x1"f 30* s d ot \l\( Ax 3° 9 N . \/ X3 8~—+ X6 PLAN ROWS “v" B "Z" YA ’J\wl 10°* ;'m WA " wfl i F o~ n wy _4$ h ' o~ Axs 4 ~ 1% Pamore Eauauiy Seazed | 1 $ ~ ALong Cirtumrgnanct OF b ~ Rivipsa,, 4 T 3 ™~ Cimcum * 33,933 10, ve V1 r ~ Drvisens 5 Z.B2TS . cq_) (uy L 24 3 ~ \ . z1 2% B \ N\ ~ ~ N \ » Arter, FasmicaTion |u Commite \ \.\/+ ,_9\ [RAabiogrAPH Tris Seomam| Or ST . X ok Waw. ANt HEAD Wirn Sursicieuy \ / \:/ Exrosvnge To Aotguatary MaP Cuting 5‘ . Suaeace, Burwuen Weao Aun Seyvem . Or Comican Biction, Access Far Rapio- \ / AETIVE Sounce Can Be Maoe Tueu Newa'fF’. \ ) . AT S O NeuRACIOACTIVE TAPERIMEWTS o4 X - RapeaT Tue Racwcharuy Discrinso Ated. ““--.___ml_ \ / / . \ Xa, NOTES 5 .’,\ / 1. Dumise Fasmeantion Mawg Dm“Lva‘KN‘“ MEASUME M ENT S AN - X\, A% Powte Denanaten By "X In Uvrem Lgry Connen, OF . ) '}i\_._q-," \ 2 ‘;D\QIZ'(F\M\TM\." C.:‘\.-UMHN.\ o N N RN ABRICAT IO AWK EAGUREMENT & TRASouICALLY T . Foates OXsiznaTan |u Thg TAmuwatens On Dwes. MI2LT73-£D-026D PLAN Rowd "' e, “y* Ao MILITHE0-0TI0 Cravrert Trose Mamxad ''Na Myasoagraants Recomns RosuuTs ln ArPpetriats Llowvmn, 3. Pomts AT WwicH MuAsuasmuni (s Ard Mave ARC IDENTICED Bv A Latren (A0, -, 2) To Iumeate Tan Rew . Ann A Nowaan Tambriation - Deimoons, G Rame sasruions Meawammats MIZITI-CO~ OO (\ T,o-2 2 Te Snow Positions On Thig Roiv, L - N Tans o - o« o 4. bewiielcations ta Pamiwtuesis [ eq. (P12), C@%); Ere. J Inoicate scavien -Luenr Merwaenants f Tamnamies Oam Tuar Pewt |s Dimgeyuy Crrosite Tua Duspiaves &mnr. REFERENCE DRAWINGS NO. Deverer UvtnAsorne Tuicuness Measvmameny Procgours During FAmmicaTs Wiien Wice Protues T Samd Thickuits MeasomiwiaTs 0AK RIDGE NATIONAL LABORATORY Ae Tuose Feuwt Dimdctuy AT PoinTs DengnATEd , NorTk | Amevt, OPERATER BY Recwono Proctoure Precisauy. €. Avtei NenmADIOACTIVE EXPERMENTS, Maxe UnTangems Thickuass UNION CARBIDE NUCLEAR COMPANY MuArume merats As tn Kot 2 Ameva Aun Rzcons Resucts in, DIVISION OF UNION CARBIDE CORPORATION Avrrorratre Cowumn, . OAK IDGE, TENNESSEE T STATE Twicknois Moadunrusuts n Diomacs . STATe Aw OTnem MEAWREMEITS W lucnuss Anos Fracmiows, . agal 8 Artcr FApmcames, Raowosmatn Whiw Surmciguy Exnpsgurxs To l#l“m;‘e DIMENSIONS UNLESS | jqmE DASTILLATION EXPERIMENT ufl%‘ & 71509 CarmapileTuy MAR 'V:fl-.\-é Wate Batwadn W=\£ \.\utc‘\‘\ Serress SPECIFED: . Or Lnigaw SEctien OF Tt Aun Rew " G ConbEneaN. FRACTIONS + Locationss OF Poisxs Fom Duisuiiaiar Aus Apqar MonmaowAGIVE Sxraruaints, Rinrtar Tar Racwocmacuy, _— Raoiocmaprie Medtwnamants, On, St Awe DECIMALS £ e Covnansenm Astemeiy ANGLES £ | e . LAPF ScuLE: Nons MIZI73 APRTYEL EQR CONSTRGETON ' 48 WALL THICKNESS AND ORSERVATIONS ‘ WALL MEASUREMENTS AND ‘ WALL THICKNESS MEASUREMENTS AND OBSERVATIONS . LOr | ma | nmac Peanct | MTCAND s o O™ | INaL | iimac | orerations | Cmaiae AND REMARKS OO N | et | maL]on Do L | RADIOGEARNY, QMEAVATIONS, MEASUREMENT | (inches) (Inches) Gnches) MEASLIREMENT | (lnchei) | (nches) |~ (nches) | (inches ~ i . -, - NR - Al G . . —_ A G38 P2 — At - ‘ P2 - B 10 P 8 1l Ps B3 2 P& B4 - al . ‘ Pt 8§ Ht P8 : H3 : : 7o B? . 4 ‘ ' Pl 8 us ; Hé H1 He HS OAK RIDGE NATIONAL LABORATORY OPERATED &Y UnioN CARBIDE NUCLEAR COMPANY DIVISION OF UNION CARBIDE CORPORATION OAK TENNESSEE BLDG. MSIRE DISTILLATION EXPERIMEWT 0. B Yoo Tagueafiow OF Data Frev Dimtwsaoad AN Rasioararmi, MeAsuRemiEnts ON- ST Anp ConbBEnscr Asttmiuy : . LILLTS FMENSIONS MTIAL DIMENSIONS MIZI73 020 - g ey i e g, '“] LOCATION OF 13 DIRECY HNITIAL CHANGE WALL THICKNESS MEASUREMENTS AND OBSERVATIONS RADIOGRAPHY, OBSERVATIONS, ~ AND REMARKS LOCATION OF YIZ Tl e i3 T5 6 WALL THICKNESS MEASUREMENTS ‘DIRECT INITIAL | INITIAL | OPERATION DIMENSIONAL CHANGE (Inches) 1 1 OBSERVATIONS RADIOGRAPHY, AND RE| 4 Byrjens @ Ea. Peritien Seazad 90° Awany OBSERVATIONS, MARKS LENGTH AND "MEASUREMENTS Caimeas, | post o | ASBUILT | OPERATION LENGTH | LENGTH REMARKS Firk Cmowns Fuat Te APPROKMATIN " D, Awn CouterPuncw A Maxe Bugyews By FOSIYING (Wu.puug) Hast eaney X Weup Matar Ow Surraces, FAyues 28 Wao R 14 Aw AcsiPganLc SUBSTITVTE . »Dimengova. & . “ Measuagrad iy > " = pleD NO. REFERENCE DRAWINGS y OAX RIDGE NATIONAL LABORATORY TED BY UNION CARBIDE NUCLEAR COMPANY DIVISION OF UNION CARBIDE CORPORATION 0AK TENNESSEE ) 1541 MIRE DHATILATION EXPRERIMENT o %1504 LacaTiem OF Porys Fom LimZAMm MEASURIMEINTS Ane Tagunatiers Or Dagxa Fom Stw Conosustf= Recewna Assumuny 12\ ) - 51 ORNL-4434 UC-80 - Reactor Technology INTERNAL DISTRIBUTION 1-3., Central Research Library 83. R. H. Guymon 4. ORNL — Y-12 Technical Library 84. B. A. Hannaford Document Reference Section 85. P. H. Harley 5-39. Laboratory Records Department 86. P. N. Haubenreich 40. Laboratory Records, ORNL R.C. 87. R. F. Hibbs 41-42, MSRP Director's Office 88-89. J. R. Hightower Bldg. 9201-3 Rm. 109 90. H. W. Hoffman 43, R. K. Adams 91. R. W. Horton 44, G. M. Adamson 92. W. H. Jordan 45. J. L. Anderson 93. P. R. Kasten 46. C. F. Baes 94. C. W. Kee 47. C. E. Bamberger 95. M. J. Kelly 48. C. J. Barton 96. S. S. Kirslis 49. H. F. Bauman 97. J. W. Koger 50. S. E. Beall 98. R. B. Korsmeyer 51. M. J. Bell 99. A. I. Krakoviak 52. E. S. Bettis 100. T. S. Kress 53. R. E. Blanco 101. C. E. Lamb 54. F. F. Blankenship 102, J. A. Lane 55. J. 0. Blomeke : 103. J. J. Lawrence 56. R. Blumberg 104. M. S. Lin 57. E. G. Bohlmann 105. R. B. Lindauer 58. G. E. Boyd 106. A. P. Litman 59. J. Braunstein 107. M. I. Lundin 60. M. A, Bredig 108. H. G. MacPherson 61. R. B. Briggs 109. R. E. MacPherson 62. S. Cantor 110. J. C. Mailen 63. W. L. Carter 111. H. E. McCoy 64. G. I. Cathers 112, L. E. McNeese 65. J. M. Chandler 113. A. S. Meyer 66. H. D. Cochran _ 114. R. L. Moore 67. E. L. Compere 115. D. M. Moulton 68. W. H. Cook 116. J. P. Nichols 69. B. Cox 117. E. L. Nicholson 70. C. W. Craven 118. A. M. Perry 71. F. L. Culler ‘ 119, J. L. Redford 72. J. H. DeVan 120. G. D. Robbins 73. S. J. Ditto 121. K. A. Romberger 74. W, P. Eatherly 122, W. F. Schaffer 75. J. R. Engel 123, C. E. Schilling 76. D. E. Ferguson 124, Dunlap Scott 77. L. M. Ferris 125. J. H. Shaffer 78. A. P. Fraas 126. M. J. Skinner 79. H. A, Friedman 127. A. N. Smith 80. J. H. Gibbons 128, F. J. Smith 8l. W. R. Grimes 129. D. A. Sundberg 82. A. G. Grindell 130. R. E. Thoma 131. 132. 133. 134, 135. 136. 137. 138, 146. 147, 148. 149-150. 151. 152, 153. 154, 155-374. C. 2HRARZ> J. 52 -~ P. Tung 139. Gale Young E. Unger 140. E. L. Youngblood M. Watson 141. P. H. Emmett (consultant) S. Watson 142, J. J. Katz (consultant) M. Weinberg 143. J. L. Margrave (consultant) R. Weir 144, E. A. Mason (consultant) E. Whatley 145. R. B. Richards (consultant) C. White EXTERNAL DISTRIBUTION Giambusso, Atomic Energy Commission, Washington, D.C. Haberman, Atomic Energy Commission, Washington, D.C. 0. Laughon, RDT Site Office, Oak Ridge National Laboratory W. McIntosh, Atomic Energy Commission, Washington, D.C. Shaw, Atomic Energy Commission, Washington, D.C. A. Swartout, Union Carbide Corporation, New York 10017 Patent Office, Atomic Energy Commission, ORO Laboratory and University Division, AEC, ORO Given distribution as shown in TID-4500 under Reactor Technology category (25 copies — NTIS)