R
ERELY 1213

MARTIN A B WY LISRARIES :

|ILTRRH TN

3 445k 0435287 9

 
 

 

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 -Commission, nor any of their employees, nor any of their contracuors,
subcontractors, or their employees, makes any warranty, express or implied, or
assumes: any fegal liability or responsibility for the aceuracy, mmpiéteness or
usefulness of any information, apparatus, product or process disclosed, or
represents that its use would not infringe privately owned rights, '

 

 
Contract No. W-TL05-eng-26

CHEMICAL TECHNOLOGY DIVISION

ENGINEERING DEVELOPMENT STUDIES FOR MOLTEN-SALT
BREEDER REACTOR PROCESSING NO. 5

L. E. McNeese

OCTOBER 1971

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee
operated by
UNION CARBIDE CORPORATION
for the
U.S, ATOMIC ENERGY COMMISSION

ORNL~TM-3140

s
f
}
!

[T

3 4456 0135287 9
ii

Reports previously issued in this series are as follows:

ORNL~4235
ORNL~L 364
ORNI~1365
ORNL~%4366
ORNL~-TM-3053
ORNL-TM-3137
ORNL-TM-3138
ORNL~TM—-3139

Period
Period
Periecd
Period
Period
Period
Period
Period

ending
ending
ending
ending
ending
ending
ending
ending

December 1967
March 1968
June 19068
September 1968
December 1968
March 1969
June 1969
September 1969
iii

CONTENTS
SU’M[MRI E S . . » . » " - . . " L . * » L] . - * . . » Ll - - . * . a -
1. INTRODUCTION v & v v 4 4 6 4 s e s o o o & o o o o o o« o o o
2. RARE~EARTH REMOVAL USING THE METAL-TRANSFER PROCESS . « « « . .
2.1 Equilibrium Data and Concentrations . . . ¢« + « +« + « « .

2.2 Removal of Rare Earths and Thorium from LiCl . . . . . . .
2.3 Conceptual Process Flowsheet and Mathematical Analysis . .
2.4 Calculated Rare-Farth Removal Times .+ + « o « o « o &+ + .

2.5 Effect of Rare~Earth Removal Time on Breeding Ratio . . .

2.6 Miscellaneous Points Related to the Metal-Transfer Process .

PROTACTINIUM ISOLATION USING FLUORINATION-~REDUCTIVE EXTRACTION .

3.1 Proposed Flowsheet for Isclating Protactinium by Fluorina-
tion--Reductive Extraction . « « + « ¢« « « ¢ 4 ¢ o o . .

3.2 Calculated System Performance . . . « ¢ v « o « o« « + o &
AXTAL DISPERSION IN OPEN BUBBLE COLUMNS + « v v 4 o v o o o o &

4.1 Experimental Technique and Equipment . . . . . . . . . .

4.2 Experimentally Determined Dispersion Coefficients . . . .

SEMICONTINUQUS REDUCTIVE EXTRACTION EXPERIMENTS IN A MILD-STEEL
FACTILITY o v v 4 v v 6 o o o s o & & s o s o & o & & o o o s o o

5 . l Hyarfldynmflic R'U.Il HR-‘6 ® . - - . . - » . . . . . » . . . . .
5.2 Hydrodynamic Runs HR-7 and HR-8; Tests for Obstructions -

SIMULATION OF THE FLOW CONTROL SYSTEMS FOR THE REDUCTIVE EXTRAC-
TION FAC ILITY - » . . » . a * . . . & 8 . . * » ? * » . * - a .

6.1 Analog Simulation of Jackleg and Extraction Column . .

6.2 Digital Simulation of Salt Flow Control System, Jackleg,
and Extraction Column . . + « « + o & & v ¢ o v o o v 4 .

CALIBRATION OF AN ORIFICE~-HEAD POT FLOWMETER WITH MOLTEN SALT

J.A.ND BI SM[JTH . . 2 . . . . - » [ ] . - - . . - . . L - . * - . L] *
T.1 Data from Steady-State Flow Experiments . . . . . . . . .
7.2 Data from Transient Flow Experiments . « ¢ + ¢« ¢« « + ¢« .+ .
ELECTROLYTIC CELL DEVELOPMENT: STATIC CELL EXPERIMENTS . . .

WO

O

12
12

15
15
17
22

26

30

30
32

36

37

L6

56

56
58

62
10.

11.

12.

13.

16.

iv

CONTENTS (Continued)

Page

ELECTROLYTIC CELL DEVELOPMENT: FORMATION OF FROZEN FITMS WITH
AQUEOUS ELECTROLYTES . ; 64
9.1 Mathematical AnalysSis . « v v v v v e e e e e e e 6L
9.2 Experimental Equipment . . . +v v v 4 o v 4 e v ou e e 69
9.3 Expected Mode of Operation . « v v v v v v v v 4 v e 0 T1
DEVELOPMENT OF A BISMUTH~~MNLTEN SALT INTERFACE DETECTOR . . . Th
10.1 Testing and Modification of Electronic Circuit Prior to

Installation . . v v v v v 4 v v e e e e e e o Th
10.2 Fabrication of Sight Tube . . + v v v 4 0 o o v o o 4 . 76
CONTINUOUS SALT PURIFICATION . v v v v v v ¢« v e o v v v v o T7
MSRE DISTILLATION EXPERIMENT . . . v v v & ¢ & v v o v v o o 79
RECOVERY OF 7Li FROM 7LimvBISMUTH~mRARE—EARTH SOLUTIONS BY
DISTILLATTON v « o o o« & 4 4 o o o o o o o o 2 o o o o o o o 88
13.1 Mathematical AnalySis v o ¢ v v 4o v 4 4 v e e e e e e 88
13.2 Calculated Resulbts . . & v v v v v v v e s e e e e e 03
PREVENTION QF AXTAL DISPERSION IN PACKED COLUMNS . . . . . . . 97

HYDRODYNAMICS OF PACKED COLUMN OPERATION WITH HIGH~-DENSITY FLUIDS 102

15.1 Experimental Technique and Results . . . . . . . . . . . 103
15.2 Pressure Drop Across the Column . « + v v « & & & + « . 104
15.3 Dispersed~Phase HOoLlduD . + « ¢ v « s o ¢« « o o & o o o 112
15.4 Correlation of Flooding Rates . . . . v v v v v « v + . 113

15.5 Prediction of Molten~Salt—Bismuth Flooding Rates in Packed
COLUMNS  + + v+ o o v o o o o o & o o v v e e e e e 116

REFERENCES + « v v v v v 6 v v v v v e v v e e s e e e e e 118
SUMMARTES
RARE~EARTH REMOVAL USING THE METAL-TRANSFER PROCESS

We have devised a new process, called the metal-transfer process,
for removing rare-earth fission products from the fuel salt of a single-
fluid MSBR. In the process, bismuth containing thorium and lithium is
used to transfer the rare-earth fission products from the reactor fuel

salt to an acceptor salt such as LiCl.

Both thorium and rare earths transfer to the bismuth; however,
because of favorable distribution coefficients, only a small fraction
of the thorium transfers with the rare earths from the bismuth to the
LiCl. The effective thoriumrare-earth separation factors for rare
earths for which data exist (i.e., Nd, La, and Eu) range from about lOu
to about 108. The final step of the process is removal of the rare
earths from the LiCl by extracticon with bismuth containing 0.05 to 0.5
mole fraction lithium. The new process does not require an slectrolytic
cell, This is an important advantage over the earlier reductive extraction

process, which also had the disadvantage of rare~earth—thorium separation

factors near unity.
PROTACTINIUM ISOLATION USING FLUORINATION--REDUCTIVE EXTRACTTION

We have recently developed a new process for removing rare earths
from a single-fluid MSBR. This process does not require an electrolytic
cell,which has led us to consider protactinium isolation methods that
do not require electrolyzers. One such method is fluorination~-reduc-
tive extraction, in which most of the uranium would be removed prior

to isolation of the protactinium by reductive extraction.

Calculated results indicated that an attractive processing system

based on fluorination--reductive extraction can be devised and that this

system will be quite stable with respect to variations as large as 207
vi

for most of the important operating parameters such as flow rates,

reductant concentrations, and number of extraction stages.

AXTAL DISPERSION IN OPEN BUBBLE COLUMNS

Axial dispersion coefficient measurements and gas holdup measure-
ments were carried out in a 2-in.-ID cpen column. Agueous solutions
of glycerol and n-butanol were used in order to determine the effect
of liguid-phase viscosity and surface tension on axial dispersion.

At low gas flow rates, increasing the liquid-phase viscosity from 1 cP
to 15 ¢P resulted in a 30% decrease in the dispersion coefficient. At
higher gas flow rates, little effect of viscosity on dispersion coef-
ficient was noted; however, oanly a limited range of gas flow rates was

examined.

SEMICONTINUQUS REDUCTIVE EXTRACTION EXPERIMENTS
IN A MILD-STEEL FACILITY

Hydrodynamic experiments using the 0.82-in.-diam column packed with
solid cylinders of molybdenum were continued. Only limited useful flooding
data were obtained because of deposition of iron in the column as well as
in the bismuth exit line from the column and the salt feed line to the col-
umn. The datas obtalined are in general agreement with predictions based on

data from a mercury-water system.

The column and the affected lines were replaced. The new column is
similar to the original column except that it is packed with 1/4-in.-diam

molybdenum Raschig rings.

SIMULATION OF THE FLOW CONTROL SYSTEMS
FFOR THE REDUCTIVE EXTRACTION FACILITY

A study of the salt and bismuth flow control systems for the
Reductive HExtraction Facility was made in order to determine the best

mode of operation of the existing equipment, to evaluate possible
vii

equipment changes, and to verify the validity of flooding data obtained
to date. One analog and three digital simulations were carried cut.
Fach simulation represented a different manner of controlling flow
through the column and allowed selection of optimum values for zystem

parameters.

All of the sygtems were found to be stable and to perform satis-

factorily; thus we believe that the actual system will function properly.

CALIBRATION OF AN ORIFICE~-HEAD POT IFLOWMETER WITH
MOLTEN SALT AND BISMUTH

Additional orifice calibration data, both steady-state and transient,
were obtained. ©Since these data suggest that the orifice drain chamber
may have been flooded at times, provisions were made for pressurizing the
zas volume above the fluid in the orifice--head pot and discharge chamber
in future experiments. This will ensure that the orifice drain chamber

does not fill with liquid.
ELECTROLYTIC CELL DEVELOPMENT: STATIC-CELL EXPERIMENTS

Two experiments related to cell development were made in a 4-in. -diam
quartz cell. In the first, current was passed between two molybdenum
electrodes to determine whether a dark material, observed in the salt in
previous tests, would be formed. Although the presence of a dark material

was noted, we were not able to identify the mechanism for its formation.

In the second experiment, we tested a porous carbon anode having a
much higher gas permeabllity than that of a graphite anode which had
previously been observed to polarize. As in the earlier test, low current

densities were observed.
viii

ELECTROLYTIC CELL DEVELOPMENT: FORMATION OF FROZEN
FILMS WITH AQUEOUS ELECTROLYTES

Formation of frozen salt films on structural surfaces exposed to the
molten-salt electrolyte in electrolytic cells is being considered as a
means for protecting these surfaces frowm corrosion by ]?a:"LF,3 produced at the
cell anode. A study of frozen film formation under condi£ions similar to
those expected in a cell is in progress. Part of the work will be carried
out using an aqueous electrolyte in order to avoid problems associated with
molten salt--bismuth systems. FEquipment has been installed for studying
the formation of ice films on surfaces of a simulated electrolytic cell
that uses an aqueous solution, rather than molten salt, for the electrolyte.
Alternating current will be used to minimize the effects of electrode reac-

tions and mass trancfer.
DEVELOPMENT OF A BISMUTH-MOLTEN~SALT INTERFACE DETECTOR

An interface detector based on the principle of eddy current gener-
ation in liquid metals is being developed. It is desired that all surfaces
in contact with bismuth be made of a refractory metal such as molybdenum,

which has a relatively low electriczl permeability.

Initial testing of the electronic circuit prior to installation
established the need for several modifications. Preliminary tests of

the system at temperatures above rocom temperature are in progress.
CONTINUCUS SALT PURIFICATION

A facility has been designed for the development of continuous
methods for removing contaminants from molten salts. The system con-
sists of a 1-1/4~in.-diam, 81l-in.~long column packed with 1/h~in.
Raschig rings, feed and receiver tanks, a salt filter, a flowing stream

salt sampler, and the necessary flow, temperature, and pressure instru-
ix

mentation. In-line instruments are provided for continuously measuring
the water and HF contents of the column exit gas stream. Systems have

also been designed to provide purified hydrogen, argon, and HF gas.

MSRE DISTILLATION EXPERIMENT

95 1hh 147

Analyses have been obtained for Li, Be, Zr, ar, Ce, Pm,

lSSEu, 91Y, QOSr, 898r, and 131

Cs in the 11 condensate samples taken
dAuring the MSRE Distilliation Experiment. Hffective relative volatil-
ities for each of these materials were calculated with respect to LiF
during the course of the experiment. The calculated values for Be, Zr,
and 952r are in agreement with previously reported values; however,
values calculated for the rare earths are significantly higher than

values measured in an equilibrium still.

RECOVERY OF TLiEfififlflYLiu-BISMUTmeRARE-EARTH SOLUTIONS BY DISTILLATION

Removal of divalent lanthanides from the LiCl stream in the metal-
transfer process produces a bismuth stream that contains lanthanides

T

and has a high concentration of 'Li {5 to 50 at. %). Because TLi is

expensive, its recovery may be economical.

The low vapor pressures of the lanthanide metals relative to lithium
and bismuth suggest that distillation might be used effectively to recover
both the lithium and bismuth. However, lithium and bimsuth form an iater-
metallic compound (LiBBi) having a high melting point (1145°C), which may
1imit the fraction of the lithium that could be recovered at still-pou
temperatures of 800 to 900°C. The fractional recovery of lithium was
calculated for a range of conditions that would allow reasonably high

recovery of the lithium.
PREVENTION OF AXTAL DISPERSION IN PACKED COLUMNS

Devices that will reduce axial dispersion in packed columns to
acceptable levels are being evaluated. Two dispersion preventers were
tested during the countercurrent flow of mercury and water in a 2-in.-
diam column packed with 3/8-in. Raschig rings. The preventers consist
of (1) a seal formed by a region in which the metsl phase accumulates
and prevents flow of the less-dense phase (salt or water), and (2)

restrictions through which the salt flows at an increased velocity.

The fraction of the less~dense phase that is recycled through a
dispersion preventer depends mainly on the water flow rate per restric—
tion and the diameter of the restriction. It appears that devices of
this type will reduce axial dispersion to desired levels; however, the

devices tested to date have reduced the column throughput at flooding.

HYDRODYNAMICS OF PACKED COLUMN OPERATION WITH HIGH-DENSITY FLUIDS

Measuremenls were made of pressure drop, holdup, and flooding during

the countercurrent flow of mercury and water in columns packed with 1/k-
in. solid cylinders, 3/8-in. Raschig rings, and 1/2-in. Raschig rings.
The data on holdup and flooding can be correlated in terms of a constant
superficial slip velocity, which is a function of packing size only. A
simple relation has not been found for correlating the pressure drop
data. An expression based on the mercury-water data is given for pre-
dicting flooding and dispersed-phase holdup in packed columns through

which salt and bismuth are in countercurrent flow.
1. INTRODUCTION

A molten-salt breeder reactor (MSBR) will be fueled with a molten
Tluoride mixture that will circulate through the blanket and core regions
of the reactor and through the primary heat exchangers. We are developing
processing methods for use in a close-coupled facility for removing fission
products, corrosion products, and fissile materials from the molten fluoride

mixture.

Several operations associated with MSBR processing are under study.
The remaining parts of this report describe: (1) a newly devised process
(called the metal-transfer process) for removing rare earths from an MSBR,
(2) a protactinium isolation process‘based on flucrination for uranium
removal, followed by reductive extraction, (3) measurements of axial dis-
persion coefficients in open bubble columns, (k) experiments carried out
in a mild-steel reductive extraction facility to study the hydrodynamics
of packed column operation during countercurrent flow of molten salt and
bismuth, (5) simulation of the flow control systems for the reductive
extraction facility, (6) calibration of an orifice--head pot flowmeter
with molten salt and bismuth, (7) experiments related to the development
of electrolytic cells for use with molten salt and bismuth, (8) design
and installation of a Tacility for studying the formation of frozen films
in electrolytic cells, using aqueous solutions to simulate molten salt,
(9) development of an induction~type bismuth-salt interface detector,
(10) design and installation of a facility for developing continuous
methods for the purification of molten salt, (11) calculation of
effective relatlive volatilities for various materials present in the
MSRE Distillation Experiment, (12) calculations concerning the reéovery
of 7Li from TLi—-—Bi-~~-rare-~earth gsolutions by low-pressure distillation,
and (13) measurement of axial dispersion coefficients in packed columns
in which immisecible fluids having large density differences are in counter-
current flow, and (14) experiments on the hydrodynamics of packed column
operation with liguids having high densities and a large density difference.
This work was carried out in the Chemical Technology Division during the

period October-December 1969.
2. RARE~-BEARTH ReEMOVAL USING THE METAI-TRANSFER PROCESS

L. E. McHNeese

We have devised a new process, called the metal-transfer process,
for removing rare-earth fission products from the fuel salt of a single-
fluid MSBR. In this process, bismuth containing thorium and lithium is
used to transfer the rare-earth fission products from the reactor fuel

salt to an acceptor salt such as LiCl.,

Both thorium and rare earths transfer to the bismuth; however, be-
cause of favorable distribution coefficients, only a small fraction of
the thorium transfers with the rare earths from the bismuth to the LiCl.
The effective thorium—rare-earth separation factors for rare earths for
wvhich data exist {(i.e., Nd, La, and Eu) range from sbout th to about 108
The final step of the process 1s removal of the rare earths from the LiCl
by extraction with bismuth containing 0.05 to 0.50 mole fraction lithium.
The new process does not require an electrolytic cell. This is an impor-
tant advantage over the earlier reductive extraction process, which also

had the disadvantage of rare-earth—thorium separation factors near unity.
2.1 Equilibrium Data and Concentrations
; . Lok . . . )
Ferris and Smith™ have measured the distribution of Th and the rare

earths La, Nd, and Eu between LiCl and bismuth containing a reductant

and have summarized their data as follows:

= L + 7.

log Dy log Cp. *+ 7 663,
log DLa = 3 log CLi + 2.025,
log DNd = 3 log C . + (n2.25),

log DEu = 2 log CLi ~ 1.699,
where
DA = digtribution ratio for material A,
= a(s1) *a(nic1)
KA(Bi) = mole fraction of A in Bi,
XA(LiCl) = mole fraction of A in LiCl,
CLi = lithium concentration in Bi, at. %.

The data for all materials except europium were obtained at 6L0°C; the
europium data were obtained at 630°C. Data for the distribution of the

materials being considered between 72-16-12 mole % LJ'_F-‘BeF2—’2[.‘hl§‘br and

bismuth containing a reductant have been summarized by Ferris as:

. , |
log DA =n log DLi + log K, ,

where
DA = distribution ratio of material A,
n = valence of A in salt,

log KA = modified equilibrium constant.

!
The variation of log KA with temperature is:

log K;h = -3.6386 + 11,115/T,

~2.005 + 7628/T,

log KLa

log KEu ~1.309 + 4366/T,

where T = temperature, °K, and Th, La, and Fu have valences of 4, 3, and
2 in the salt respectively. For Nd, the measured value of the Nd-Th

separation factor (3.0) was taken at Th saturation at 600°C, and the
1

!
same temperature dependence was assumed for log as that for log K. .
P ' d La

The derived expression is, then:

log KNd

= ~1,942 + 7628/T,

and Nd is assumed to be trivalent in the salt.
Using these data, the calculated equilibrium concentrations in Bi
and LiCl at 640°C are as shown in Table 1, where the rare-earth con-
centrations in the fuel salt are those resulting from a 50-day removal
time for the rare earths. The reductant concentration in the Bi was
chosen such that the Th concentration was approximately 95% of the Th
solubility in Bi at 640°C. No correction was made in the data for
europium to account for the fact that they are being used at 6L0°C
(although they were obtained at 630°C). All the concentrations are

given in mole fractions.

Table 1. Equilibrium Concentrations in Fuel Salt,
Bismuth, and LiCl at 6L40°C

 

 

Fuel Salt Bismuth LiCl
XLiF = Q.72 XLi = 0.00201
X = 0.16
BeF2
X = 0.12 = 0.0025 X = 0.33]1 x 10"’T
X = 10.8 x 10”6 X = 0.524 x 10“6 X = 0.607 x 10“6
LaF La, ‘ - LaCl
3 3
X = 26.2 x 10’6 X .= 1.h7 x 10“6 X = 1.02 x 10“6
NdF3 - Nd : . ‘NdCl?
-6 -8 _ ~6
- = 0.287 x 10 X, = 0.664 x 10 X = 8.22 x 10
XEUf2 Ei buCl?

 

The data in the table show that the rare earths are present in the LiCl
al low concentrations and are associated with only a small amount of
thorium. Clearly, if a practical means is available for removing the
rare earths and thorium from the LiCl in the relative concentration
ratios shown, one will have achieved the desired goal of removing the

rare earths with an acceptable thorium discard rate.
2.2 Removal of Rare Farths and Thorium from LiCl

The minimum rate at which the LiCl solution must be processed for
removal of the rare earths is given by the ratioc of the production rate
of a given rare earth to its equilibrium concentration in the LiCl.
These data, which are shown below, indicate that the minimum process-
ing rate, as set by Nd, is 1.29 x lO6 g-moles of LiCl per day. This
is equivalent to 1320 ft3 of LiCl per day, which appears to be g
higher processing rate than would be practical for low-pressure dis-

tillation.

 

Minimum

Production Concentration Processing

Rare Rate ‘ in LiCl Rate
Earth (moles/day) (mole fraction) {moles/day)
La 0.54 0.607 x 10“6 0.889 x 106
-6 6
Na 1.31 1.02 x 10 1.29 x 10
-6 3
Eu 0,014} 8§.22 x 10 1.75 x 10

 

Further consideration suggests that contact of the LiCl with a small
volume of bismuth containing a relatively high concentration of lithium
(5 to 50 at. %) would provide the desired removal of the rare earths
from the LiCl as well as the removal of the thorium associated with the

rare earths. This is the method selected for additional evaluation.
2.3 Conceptual Process Flowsheet and Mathematical Analysis

The conceptual process flowsheet consists of three contactors through
which the process fluids are circulated, as shown in Fig. 1. Fuel salt
from the Pa isolation syst
tfié rare earths at the reactor concentration, is countercurrently contacted
with Bi containing approximately 0.002 mole fraction Li in contactor 1.

The rare earths transfer, in part, to the downflowing metal stream and
are carried into contactor 2 by the bismuth stream. In contactor 2, the

bismuth stream is contacted countercurrently with LiCl, and fractions of
ORNL DWG 70-11037

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A Tf»:>t XM2
eqac OF“F;EETW“
(1)
Bi
FRS, XS XMl
Fuel Salt From FMI
Pa Isolation
XS3 §
[
(2)
LiCl
XS4
FCS L—-——»
| XM5
(3) Li-Bi
Li-Bi
i + Rare FEarths
FSM *+ Th

XM6& FM,XM§

Yig. 1. Metal-Transfer Process for Removing Rare Earths from a
Single~:1uid MSER.
the rare earths transfer to the LiCl. The resulting LiCl stream is
finally contacted with Bi containing & high concentration of Li (0.05
to 0.5 mole fraction) in contactor 3, where most of the rare earths and
thorium transfer from the LiCl to the Bi stream. A small stream of Bi
contalining Li is fed to the circulating Bi stream in contactor 3, and

an equal volume of bismuth is withdrawn.

From material balanc¢e and equilibrium considerations, the following
relations for a given transferring material can be written for the three

countercurrent contactors. The nomenclature of Fig. 1 is used.

 

 

 

N _+1
XSl - X82  _ U - K (1)
XS1 - XM2/Dy, AFNU+1 -1 F
( _l_)NLfl L
X1 - XM Mg o . (2)
ML - X8L D N_41 C
C <'1.) L
ol -1
c
N+l
xs3 - xsh __ _ g8 " (3)
XS3 - XMS/DS o Ngrl S
9
From material bvalance considerations,
X8l - XM2 = L/FRS (&)
XML - XM2 = L/FM1 (5)
X83 - X8 = L/FCS (6)
XME - XM5 = L/FSM (7)

XME = L/FM, (8)
X552

XM1

X873

Xsh

XM5

FRS
FM1
FCS

FM

N_
L

concentration of transferring
contactor 1, mole fraction,
concentration of transferring
contactor 1, mole fraction,
concentration of transferring
1, mole fraction,
concentration of transferring
2, mele fraction,
concentration of transferring
2, mole fraction,
concentratlon of transferring
2, mole fraction,
concentration of transferring
3, mole fraction,
concentration of transferring

3, mele fraction,

material

material

material

material

material

material

material

material

in

in

in

in

in

in

in

in

fuel salt entering

fuel salt leaving

Bi leaving contactor

Bi leaving contactor

LiCl leaving contactor

LiCl entering contactor

Bi entering contactor

Bi leaving contactor

distribution ratio between fuel salt and Bi in contactor 1,

distribution ratio between LiCl and Bi in contactor 2,

distribution ratio between LiCl and Bi in contactor 3,

FM1 - D

E1
FRS

ML - DC

FCS

FSM DS

rCB

fuel salt flow rate, moles/unit time,

metal flow rate in contactors 1 and 2, moles/unit tinme,

LiCl solution flow rate in contactors 2 and 3, moles/unit time,

metal addition and withdrawal rate, moles/unit time,

number of eguilibrium stages in contactor 1,

number of eguilibrium stages in contactor 2,

number of equilibrium stages in contactor 3,
KF = constant, defined above,
KC = constant, defined above,
KS = constant, defined above,
L = production rate or loss rate of transferring material.

The distribution ratios are assumed to be constant since there will be
a very small change in reductant concentration in the contactors; hence

the A's and K's defined above are constant.

The eight relations listed above define the state of the system
and can be solved simultaneously for the removal efficilency for a given

material in contactor 1 as:

 

A A

1_1 .1 .1 ¢ (1 € [ESM

Lek e b oS o e (B2 ), (9)
B K, Ay K, A "Kq Aphq \FM

where

B = fraction of material removed from fuel zalt in contactor 1.

The removal time for a given transferring material is, then:

v A A
R 1 1 1. c ;1 C FEM

e Bl Ly, Sl gy e (Emr L0 gy
RS KF AF KC AF KS AFAS M )

-

where

T removal times, days,

VR/FRS = processing cycle time, days.

2.4 (alculated Rare-~Farth Removal Times

There appears to be great latitude in choosing operating conditions
that will yield a desired rare-earth removal time. Conditions that give
rare~earth removal times equal to, or somewhat shorter than, removal times
presently assumed in reactor evaluation calculations will be given first.
Those calculations assume a 50-day removal time for all rare earths except
europium which is removed on a 225~day cycle. The table below shows the

removal times (in days) for the following conditions: FRS = 2.94% gpm
10

(3-day processing cycle for reactor); FML = FSM = 8.35 gpm (Bi rates in
contactors); FCS = 11.2 gpm (LiCl flow rate); FM = 22.7 liters/day (Bi
feed and withdrawal rate); Li concentration in Bi in contactors 1 and

2 = 0.00201 mole fraction; Li concentration in Bi in contactor 3 = 0.05

mole fraction; and number of stages in contactor 3 = 1.

 

 

 

Rare Vo, of Eguilibrium Stages in Contactors 1 and 2
Earth 1 2 3 L 5 6

La 52.1 38.0 34.0 32.1 31..0 30.2
ifs! 58.0 Lo, 7 38.4 36.5 35.4 3.8
Fu 222k 219.4 219.4 219.4  219.4  219.4

 

It is apparent that, with two or three stages in contactors 1 and
2 and with volumetric flow ratios near unity, one can obtain satisfactory
removal times with relatively small processing streams. If packed col-
unns were used as the contactors, the diameters would be approximately
6 in. Since the number of stages needed is low, the use of mixer-settlers
similar to those being used at Argonne National Laboratory should not be
dismissed, particularly for experiments almed at demonstrating the process

concept.

The rate at which TLi is removed from the system in the withdrawal
stream is 52.4 g-moles/day. The rate at which thorium is removed is
less than 0.069 g-mole/day. If the Bi withdrawal stream were hydro-
fluorinated or hydrochlorinated in the presence of a suitable waste salt

7

in order to recover only the Bi, the cost due to loss of Li would amcunt

to 0.0018 mill/kWhr.

It is apparent that one can obtain removal times much shorter than
those discussed in the previous section by increasing the bismuth and
LiCl flow rates and hence the size of the processing plant. In each case,
the reactor processing cycle time is 3 days (2.94 gpm). The removal
times resulting from increasing the bismuth and LiCl flow rates to ten
times their previous values are given in Table 2 for the following con-

ditions:
11

FRS = 2.94 gpm (3~-day cycle),
FM1 = FSM = 83.5 gpm (Bi flow rate in contactors),
FCS = 112 gpm (LiCl flow rate),

M = 16 ft3/day (Bi-ILi feed and withdrawal rate),

Li concentration in contactors 1 and 2 = 0.00201 mole fraction,

number of stages in contactor 3 = 1.

Table 2. Variation of Rare-Earth Removal Time (in days) with Lithium
Concentration in Contactor 3 and Number
of Stages in Contactors 1 and 22

 

Rare No. of Stages in Contactors 1 and 2

 

Barth 1 D 3 b 5 6

 

Li Concentration in Contactor 3 = 0.05 mole fraction

La 7.75 5.26 h.oh3 h.o3 3.79 3.6k
Nd 8.36 5.85 5,05 4,67 b, L6 L, 3k
Fu 16.6 1h, 7 1h.1 13.8 13.7 13.€

Li Concentration in Contactor 3 = 0.20 mole fraction

La 7.59 5,10 L, 28 3.87 3.63 5,08
Nd 8.02 5.71 h,91 4,54 L.33 4,20
Fu 8.71 6.81 6.23 5.98 5.86 5.79

Li Concentration in Contactor 3 = 0.50 mole fraction

La 7.59 5.10 L, 27 3.87 3.63 3.L8
Nd §.02 5,71 h.91 4,53 b.33 L. 20
Fu 8.n 6.37 5.79 5.54 542 5.35

 

aThe bismuth and LiCl flow rates in each case are assumed to be ten
times the values used to produce the data shown in the table on p. 10.

Thus, rare-earth removal times of 3.5 to about 15 days can be ob-
tained, depending on the concentration of lithium used in the final

contactor. It should be noted that one cannot afford to lose the Li by
12

a hydrofluorination operation as before, but would have to recover most
of the Li and Bi. A possible recovery method consists of distilling the
Li-Bi stream at the rate of about 16 ft3/day. The still would operate

at a temperature of about 800°C and a pressure of approximately 0.1 mm Hg,
and the rare earths would be remcved in a still-bhottoms volume of about

20 liters/day. (The bismuth could be recovered from this volume.) The

thorium would be discarded at the rate of about 1.4 g-moles/day.
2.5 Effect of Rare-Earth Removal Time on Breeding Ratio

The effect of rare-—earth removal time on breeding ratio has been
calculated by Be112 and is shown in Fig. 2. The changes are given with
respect to the base case, in which a removal time of 50 days is used
for all the rare earths except europium. (Europium has a 225~-day removal
time.) It should be noted that rare-earth removal times of 5 days., 3 days,
or 7 days would result in increases in the breeding ratio of 0.0126,

0.014, and 0.0115 respectively.
2.6 Miscellanesous Points Related to the Metal~-Transfer Process

The following observations relative to the proposed process have

been noted.

1. According to experimental data and calculations made by Ferris,
LiBr would probably be at least as satisfactory as LiCl as
the acceptor salt. This finding may be important in two
respects. It offers another material with which to work;
perhaps more importantly, however, the use of LiBr would
not introduce chloride into the reactor but, instead, would
simply increase the concentration of bromide already in the

system.

2. 'The feed stream exiting from the Pa isolation system will
have to be essentially free of U and Pa, since these mate-

rials would likely accumulate in the Bi circulating in
IN BREEDING RATIO

INCREASE

0.0i6

0.0t4

Qe

0 .010

0.008

Q0.006

0.004

0002

-0.002

-0.004 I

ORNL DWG. 69-12121 Rl

 

 

-

T T T T T ] r T T T T

 

REFERENCE CASE /O

225 DAY Eu REMOVAL

1 i i | i 1 1

] 1 ] i i

 

 

brd

i
3 4 5 e 7 B8 9 0 20 30 40 5 80 ™ 80

]

RARE EARTH REMOVAL TIME , DAYS

ffect of Rare Earth Removal Time on Breeding Ratio.

30 100

T
1k

contactors 1 and 2 and thus consume the Li and Th reductant
and prevent the transfer of rare earths. It will probably
be necessary to treat the salt stream leaving the Pa isola-

tion system for further removal of U and Pa.

Since virtually all of the partitioning of the rare earths
from Th is a result of the Bi~LiCl equilibrium, one does
not have to maintain the present free fluoride content of
the fuel salt and, hence, could alter the reactor fuel salt

composition.

While there is no chemical method for transferring chloride

ion to the reactor fuel salt, entrainment of LiCl in the bis-
muth leaving the second contactor would transfer chloride to
the reactor. Fortunately, there appears to be less likeli-
hcod of this type of entrainment than the entraimnment of the
bismuth in the salt. It may be necessary to treat the salt
returning to the reactor by hydrofluorination or fluorination
for removal of chloride. Another alternative would be to pass
the bismuth from contactor 2 through a captive fluoride salt
volume (such as T2-16-12 mole % LiFmBeF2~ThFu); this would
probably result in removal of the entrained chloride salt.
Although the inherent consumption of reductant for this process
is quite low (15 g-moles of lithium per day), the introduction
of oxidants (or even hydrogen) could increase the lithium

consumption significantly.

The proposed rare-earth removal process does not require an
electrolytic cell, and the operation of protactinium removal
systems not requiring a cell should be kept in mind as a
desired goal. An alternative to using a cell is fluorination
of the uranium on a 3-day cycle followed by reductive extrac-
tion processing for protactinium isolation. The cost of the
fluorine, assuming 100% fluorine utilization and a fluorine

price of $2/1b, would be 0.02 mill/kWhr.
15

T. Conditions in the circulating Bi loops are always reduclng
so that a corrosion inhibitor such as Zr would likely remain
in the Bi. This may allow Croloy to be used as the material
of construction, although the operating temperature is sig-
nificantly higher than the range in which the use of Croloy

with zirconium innibltion was successful.

3. PROTACTINIUM ISOLATION USING FLUORINATION--REDUCTIVE EXTRACTION

L. E. MclNeesge

We have recently developed a new process for removing rare earths
from a single-fluid MSBR. The advantages of this process, which does not
require an electrolytic cell, have led us to consider protactinium isola-
tion systems that do not require electrolyzers. One protactinium isola-
tion methed in which an electrolyzer is not required is fluorination--
reductive extraction. In this method, most of the uranium would be re-
moved from the fuel salt by fluorination priocr to isolation of the prot-

actinium by reductive extraction.

3.1 Proposed Flowsheet for Isolating Protactinium
by Fluorination--Reductive Extraction

The fluorination--reductive extraction system for isclating prot-
actinium is shown in its simplest form in Fig. 3. The salt stream from
the reactor first passes through a fluorinator, where most of the uranium
is removed by fluorination. Approximately 90% of the salt leaving the
fluorinator is fed to an extraction column, where it 1is countercurrently
contacted with a bismuth stream containing lithium and thorium. The
uranium ig preferentially removed from the salt in the lower extractor,
and the protactinium is removed by the upper contactor. A tank through
which the bismuth flows is provided for retaining most of the protactinium

in the systemn.
16

ORNL -DWG-69-13169

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

y
EXTRACTOR
2
o FPo
DECAY
: (8i)
REACTOR
EXTRACTOR
UFg
- 1
. FLUORINATOR
¥
HF —e OXIDIZER REDUCTANT
2 & ' Bi | ADDITION
SALT CONTAINING UF, AND Paof, ’
Li,Th

fig. 3. Isolatiocn of FProtactinium by Fluorination-Reductive
Extraction.
1t

The bismuth stream leaving the lower contactor contains some prot-
actinium, as well as the uranium that was not removed in the fluorinator
and the uranium that was produced by the decay of protactinium. This
stream is contacted with a HE—HF mixture in the presence of approximately
10% of the salt leaving the fluorinator in order to transfer the uranium
and the protactinium to the salt. The salt stream, containing UFM and
Pth, is then returned to & point upstream of the flucrinator, where
most of the uranium is remcved. The protactinium passes through the
fluorinator and 1s subsequently extracted into the bismuth. Reductant
(ILi and Th) is added to the Bi stream leaving the oxidizer, and the
resulting stream is returned to the upper contactor. The salt stream
leaving the upper contactor 1s essentially free of uranlum and prot-
actinium and would be processed for removal of rare earths before being

returned to the reactor.
3.2 Calculated System Performance

Calculations have shown that the system 1s quite stable with respect
to variations as large as 20% for most of the important parameters (i.e.,
flow rates, reductant concentrations, and number of extraction stages).
The required uranium removal efficiency in the fluorinator is less than
90%. The number of stages required in the extractors is relatively low,
and the metal-to-salt flow ratio (about 0.26) is in a range where the
effects of axial dispersion in packed column extractors will be negli-

gible. Since the protactinium removal efficiency is very high and the

231
system is guite stable, materials such as 3 Pa, Zr, and Pu should

233

accumulate with the Pa. These materials can be removed by hydroflu-

orination of a small fraction of the bismuth stream leaving the lower

extractor in the presence of salt which is then fluorinated for uranium

recovery. Sufficient decay time would be allowed for most of the aBBPa

233 233Pa and 233

to decay to U, and the U losses would be acceptably low.
18

Operating conditions that will yield a 10-day protactinium removal
time include a fuel salt flow rate of 0.88 gpm (10-day processing cycle),
a bismuth flow rate of 0.23 gpm, two stages in the lower contactor and
six to eight stages in the upper contactor, and a decay tank volume of
200 to 300 £t5, The required quantity of reductant is 340 to 430

equiv/day, which will cost 0.012 to 0.015 mill/kWhr if TLi is purchased.

The remainder of this section is devoted to a discussion of the
effects of important system parameters. Unless otherwise stated, the
operating conditions include an operating temperature of 640°C, a
processing cycle time of 10 days, a reductant addition rate of L29
equivalents/day, a protactinium decay tank volume of 300 ft3, a maximum
thorium concentration equivalent to 50% of the thorium solubility at

640°C, and a uranium removal efficiency during fluorination of 98%.

The effect of the number of equilibrium stages in the upper ex~-
tractor on protactinium removal time and uranium inventory in the decay
tank is shown in Fig. 4. Less than two stages are required in the lower
contactor; however, more stages can be used to advantage in the upper
contactor. Approximately five stages are adequate; little benefit is

obtained from using additional ones.

The effect of the fraction of uranium that is not removed during
fluorination is shown in Fig. 5. High uranium removal efficiencies
are not required since the retention of as much as 12% of the uranium

in the salt causes no harmful conseguences.

The effect of reductant addition rate is shown in Fig. 6. 1Mo
effect of decreasing the addition rate is seen until one reaches a
value of 390 equiv/day, where a gradual but significant effect begins
to be observed. The changes shown are approximately those changes
vhich result from variations in the bismuth flow rate for a constant
inlet reductant concentration. Hence, there is little effect of

variations in the bismuth flow rate.
Po REMOVAL TIME (days)

19

ORNL DWG 69-13174 RI

 

 

16
15 —11.0
/ \\'
K —J0.8
TEMPERATURE, 640 °C
PROCESSING CYCLE, 10 DAYS
13 b Li REQUIREMENT, 429 MOLES Lisrpay —0.6
DECAY TANK VOLUME, 300 ft3
METAL TO SALT FLOW RATIO, 0.25
12 bee —0.4
... —0.2
10 | 0

 

 

 

 

 

|
3 4 o 6 7 8 9
NUMBER OF STAGES IN UPPER COLUMN

10

URANIUM INVENTORY IN DECAY TANK, % of Reactor Inventory

Fig. L. Variation of Protactinium Removal Time and Uranium Inventory

in Decay Tank with Number of Stages in Upper Column.
Pc REMOVAL TIME, doys

20

ORNL DWG 68-13179 RI

 

 

 

 

 

 

 

1 | | | l | | ] l | | | !
I7 — 1.4
16 — — L2
15— —11.0
'
14— —0.8
TEMPERATURE, 640°C
DECAY TANK VOLUME, 300 13 —
STAGES IN UPPER COLUMN, &
13— STAGES IN LOWER COLUMN, 2 -+0.6
METAL TO SALT FLOW RATIO, 0.26
REDUCTANT FEED RATE, 428.6 eq/day —
12 — —0.4
I —0.2
-

0 Lo e e b Ly
0 0.02 0.04 0.06 0.08 C.10 0.12 0.14
FRACTION OF U NOT REMOVED BY FLUORINATION

Fig. 5. Variation of Protactinium Removal Time and Uranium Inventory

in Decay Tank with ¥raction of Uranium Not Removed by Fluorination.

INVENTORY

iN DECAY TANK, % OF REACTOR

INVENTORY

URANIUM
ORNL DWG 69-13178

 

T T T T T T T T T

    

*
i

TEMPERATURE 640°C

DECAY TANK VOLUME 300 f1°
STAGES IN UPPER COLUMN 10
STAGES IN LOWER COLUMN 2
PROCESSING CYCLE 10 DAYS

Pa REMOVAL TIME ( DAYS )
w
l

   

 

|
o

|
o

l |
O O
B o0
URANIUM INVENTORY IN DECAY TANK
% OF REACTOR INVENTORY

i
o
~

 

 

ol L 1o Ly Lo
300 320 340 360 380 400 420
REDUCTANT FEED RATE {eq/DAY)

Fig. 6. Effect of Reductant Feed Rate cn Protactinium
znd Uranium Inventory in the Protactinium Decay Tank.

0
440

Removal Time
The effect of processing cycle time on protactinium removal time
and uranium inventory in the decay tank is shown in Fig. 7. The removal
time is equal to the processing cycle time (100% removal efficiency) for
processing cycle times greater than about 10 days. As the processing
cycle time is decreased below this point, however, the system slowly
loses efficiency. Figure 7 shows that the system is relatively insen-

sitive to minor variations in the salt feed rate.

The effect of the protactinium decay tank volume is shown in Fig.
8. No change 1s seen as the tank volume is decreased from the nominal
volume of 300 ft3 until a value of 280 ft3 is reached; at this point,
the system slowly loses efficiency. Smaller tank volumes could be used
without loss of removal efficiency by increasing the reductant concen-

tration in the bismuth fed to the extractors.

The effects of operating temperature and reductant addition rate
on protactinium removal time are shown in Fig. 9. The optimum operating
temperature is about 640°C, and the system is relatively insensitive to
minor temperature variations. The reductant costs associated with the
addition rates shown decrease in increments of 0.001 mill/kWhr from a

value of 0.015 mill/kWhr for an addition rate of 429 equiv/day.

L. AXTAL DISPERSION IN OPEN BUBBLE COLUMNS

J. S. Watson L. H. McNeese

Axial mixing or dispersion is likely to be important in continuous
fluorinator design and performance. Because molten salt saturated with
fluorine is corrosive, fluorinators will be simple., open vessels that
have a protective layer of frozen salt covering all exposed metal walls
and surfaces. The rising gas bubbles may cause appreciable axial mixing
in the salt. We have previously3 measured axial dispersion coefficients
in a 2-in.~diam column using air and water to simulate fluorine and salt.
Axial dispersion coefficients between 25 and 35 cmg/sec were observed for

gas flow rates between 3 and 30 cm3/sec and were independent of the liguid
23

ORNL DWG 69-13176

 

28— ~1.9 >
o

O

hns - ’._.

Z

e Ll

26— —48 >
z

24 b —.7
i TEMPERATURE 640° C ]
DECAY TANK VOLUME 300 ft°
a2l STAGES IN UPPER COLUMN 6 s
STAGES IN LOWER COLUMN 2
REDUCTANT FEED RATE 428.6 eq/day

Pa REMOVAL TIME (days)
5
|
|
w\
INVENTORY IN DECAY TANK, % OF REACTOR

 

 

 

 

61— -3
14 }— 2 Z
z
. | %
a
o
12 b— —.1
b LUy by T
O 2 4 6 8 i0 12 14

PROCESSING CYCLE TIME (doays)

Fig. 7. Effect of Processing Cycle Time on Protactinium Removal Time
and Uranium Inventory in Decay Tank.
24

ORNL DWG 69-13177

 

     
  
   

 

 

 

24

M

| o

<1 O

}....

<

22 ] L

>

<

[~ T

o

__ 20— —1.0 c&)

- w

o L n x

o u

" Q

— 5

_J 1 Z

;t TEMPERATURE 640 °C -

3 16l REDUCTANT FEED RATE 4286 eq/day g >

= STAGES IN UPPER COLUMN 10 ©

- STAGES IN LOWER COLUMN 2 2

> L PROCESSING CYCLE 10 days 1 a

o =z

Q' —

- -4 >

4 4z

o

-

- ~4 =

Lt

>

12— ] D z

=

2

| ™ =

=

|10 1 l l 1 l 1 l i ] l ] 0 -
180 200 220 240 260 280 300 320

Pa DECAY TANK VOLUME, ft3

rig. 8. Effect of the Protactinium Decay Tank Volume on Protactinium
Removal Time and Uranium Inventory in the Protactinium Decay Tank.
Pa REMOVAL TIME | days

20

19

 
  
 

18+

17

15

 

 

 

™D

ORNL DWC £9-13180 R

 

PROCESSING CYCLE, 10 DAYS
STAGES IN UPPER COLUMN, 10
STAGES IN LOWER COLUMN, 2
DECAY TANK VOLUME, 300 FT?

    

.......

343 eq REDUCTANT /DAY

  

 

 

 

14

43 |

12 t—

19 b

10l J N | |

600 610 620 630 640 650 660 570
TEMPERATURE (°(C)

Fig. 9. Variation of Protactinium Removal Time with Reductant

Addition Rate and Operating Temperature.
26

(water) flow rate. In this flow range, termed "bubbly flow," the gas
moves up the column in the form of discrete bubbles. When the gas flow
rate 1s increased to greater than 30 cm3/sec, "slugging flow'" occurs
and the gas bubbles coalesce and may fill the entire cross section of
the column. In this flow range, the diffusion coefficient is more
dependent on gas flow rate; it is approximately proportional to the

square root of the volumetric gas flow rate.

The air-water system differs from the salt-fluorine system in two
significant respects: (1) the viscosity of salt is 15 to 30 times higher
than that of water, and (2) the surface tension of water is about 75
dynes/cm as compared with about 200 dynes/cm for salt. OFf course, the
density of salt is about three times that of water, but this difference
is belleved to be of less importance. The present study, performed by
a group of MIT Practice Schocl studerfcs,!Jr was carried out to investigate

the effects of viscosity and surface tension on the dispersion coefficient.

k.1 FExperimental Technique and Equipment

The experimental technigue and equipment were the same as those
used earlier.3 The column was 2 in. in diameter and had a length of

6 ft. The gas inlet was 0.04 in. ID. A 0.5 M Cu(NO,), tracer was

3°2

continuously injected into the bottom of the column, and the steady-
state concentration of tracer was measured photometrically upstream
at several points. t each measuring point, a sample of soclution was
removed from the column, circulated through a photocell, and returned

to the colunn at the same elevation.

The effects of liquid viscosity and surface tension were studied
separately. The viscosity of the licquid fed to the column was increased
to 15 cP by adding glycerol. The surface tension of the liquid was
varied between 37.8 and 67.4 dynes/cm by adding n-butanol to water.

(The surface tension of water is 75 c¢P.) Addition of the glycerol
27

resulted in properties closer to the process conditions; however, the
decrease in surface tension resulted in properties farther removed from
process conditions. Nevertheless, data were obtained on the effects of

variations in surface tension and viscosity.

4.2 Experimentally Determined Dispersion Coefficients

Experimentally determined values for the dispersion coefficient are
summarized in Fig. 10. At low gas flow rates, the dispersion coefficients
for g glycerol-water mixture with a viscosity of 15 cP are about 30% below
those observed for water alone. The transition from bubbly to slugging
flow occurs at a gas flow rate of about 26 cm3/sec. At higher rates, the
digpersion coefficient values approach those obgserved with water. Unfor-
tunately, the data for glycerol in Fig. 10 do not extend to high air flow
rates; however, the data al these rates probably approximate those ob-

tained with water.

In the bubbly region, the gas bubbles were smaller than those ob-
served at the same gas flow rate with water, and the terminal rise
velocities were lower. However, the terminal velocity of "slugs" of
gas in the high gas flow rate region became independent of viscosity,
although the slugs were probably larger than with water. This is
because the gas holdup or void fraction increased with increasing

viscosity (Fig. 11).

The addition of n-butanol to water to decrease the surface tension
resulted in an increase in dispersion coefficient, as shown in Fig. 10.
These results, however, are not conclusive because of possible concen-
tration of the n-butancl at the gas-liquid interfaces. n Butanol and
other surface-active materials may have concentrated at the interface
so that the actual concentrations at the interface {and thus the surface
tension) would not be known. The interface could be more rigid than an

air-water interface, and this effect alone could increase the dispersion
28

ORNL-DWG-70-11038 R2

 

   

 

 

 

I T T 7171 1 | | I 1 P I i
200+ -
O - Butanol-Woter Sclution, o m 37.8 dynes/cm
A - Butanol-Water Solution, o = 53.2 dynes/cm
E QO - Butanol-Water Solution, o = 67.4 dynes/cm {single run)
» 1001~ —
o~
80H - - i = —
E B - Glycerol-Water Solution, u = [5¢p /
€ 60 -
2
2 SO - —
= R o e B
8 ‘o D "flgww;/’ 1
c . Dotq For \NO’E_!_,..,‘- '
S 30t Previon® s T ~
2 . j
; e — :
o a
2 '
B .—D’/‘D—”// )
10 1 b L | 1 ) )1 1 - !
I 2 3 4 5 6 8 10 20 30 40 5060 80 100 200

Gas Flow Rate {(cm? /sec)

Fig. 10. Effect of Liquid Surface Tension and Viscosity on Dispersion
in an Open Bubble Column.
26

ORNL DWG 70-11040R|

 

50— T

 

 

 

1 1 ¥ T Y T
a0k O Butonol-Water Solution,o = 37.8 dynes/cm !
A Butanol-Water Solution, o = 53.2 dynes/cm B
O Distilled Water U
30~ 0O Glycerol-Water Solution, u=15 cP & -
0
20 - -
™ (]
() f
2 10| =/ —
a ]
s 8 5 7
O
T
w s -
Q [
O - /D -
..
4 ~ ~
3“ —
er ® =
)
0
I ] L | 1 i 1 1 i ] l
5 6 8 10 20 30 40 50 60 80 100

Gas Flow Rate {cm3/sec)

Fig. 11. Effect of Gas Flow Rate and FProperties of Liguid FPhase on
Gas Holdup in an Open Bubble Column.
30

coefficient. The same problem could have existed in the glycerol-
water experiments; however, the glycercl concentration was sufficiently
high that the concentration of glycerol at the surface is not believed
to have been seriocus. Although it is desirable in experiments such as
these to use pure components, it is frequently, as in this case, unot

practical.

5. SEMICONTINUOUS REDUCTIVE EXTRACTION EXPERIMENTS
IN A MILD-STEEL FACILITY

B. A. Hannaford C. W. Kee
L. E. Mclisege

Hydrodynamic experiments with the 0.82-in.-diam column packed with
1/h=in. solid cylinders of molybdenum were continued. Only limited useful
flooding data were cobtained because of deposition of iron in the column as
well as in the bismuth exit line from the column and the salt feed line to
the column. The data obtained are in general agreement with predictions
based on data from a mercury-water system. The column and the affected

lines were removed for replacement.
5.1 Hydrodynamic Run HR-6

Only limited useful flooding data were obtained in run HR-6 despite
good control of the rates of flow of bismuth and salt from the respective
feed tanks. Subseguent examinalion of transfer lines attached to the ccl-
umn suggested thatl the unexpectedly high bismuth holdup may have been due

to a deposit of iron accumulating in the transfer lines.

The schematic diagram of the experimental equipment (Fig. 12) in-
dicates the method by which the pressure at the base of the column (a
measnre of the bismuth holdup) was deduced. As long as salt is entering
the column, the pressure at the base of the column is equal to the sum

of (1) the static column of salt between the column inlet and the jackleg
ORNL DWG 70-11039

Pressurizing

 

 

 

 

 

 

 

 

 

 

Arqgon
T6 Pressure Salt Level
Recorder _ Contraller - Recorder
Argon Bleed - >—— Argon Bubbler
N

Salt Jackleg
T6

          
      

 

   

 

 

 

 

 

 

 

smuth 747
7
724
g; Salt
_ \ 4] Feed
Entrainment %,
Detector A
5G-12 %5
A
M
1]
Flowing -
Bismuth
Sampler Salt-Bismuth
/ Interface
J (Constant)
- xx
Salt-Bismuth—" &z Xax X
Interface ¥ T—Dendritic
(Variable) X lvon
X Deposits
X
X
X
10 in.
5
\_ X
xxX
0

Pig. 12. Packed Column and Related Components Shown in True Ele-
vation; Schematic Diagram of Level Control System for Salt Jackleg.
32

bubbler, (2) the variable salt head in the jackleg, and (3) the vari-
able argon pressure above the salt in the jackleg. Pressure drop due
to salt flow in the line connecting the Jjackleg and the column is

negligible.

Run HR~-6 began with a salt flow to the column of about 75 ml/min;
however, when bismuth flow was started, the pressure at the bhase of
the column began to increase rapidly as reflected by a rapidly in-
creasing argon pressure in the Jackleg. By stepwise increase of the
level set-point in the Jackleg, it was possible to accommodate the in-
creasing pressure drop across the column. At a bismuth flow rate of
about 175 ml/min and a salt flow rate of 75 ml/min, the pressure at
the base of the column was about 250 in. HZO and was increasing slowly.
At this pressure, the bismuth-salt interface below the column would be
within a few inches of the bottom of the bismuth drain loop:; thus the
bismuth flow rate was reduced in order to prevent the salt-metal inter-
face from being forced through the drain loop. Despite a high bismuth
holdup (80%) inferred from the pressure at the base of the column, no
observable guantity of bismuth was carried out the top of the column
with the salt; that is, no variations were observed in the signal from
the bismuth entrainment detector (Fig. 12) that was being operated with

the freeze valve open.
5.2 Hydrodynamic Runs HR-7 and HR-8;, Tests for Obstructions

Run HR-T was begun at the lowest possible flow rates (w50 ml/min
for each phase); however, flow of salt through the column was obtained
only during a brief interval during which the salt Jackleg showed a
pressure equivalent to a static column of bismuth filling the extrac-
tion column. Bismuth flowed from the column only by way of the salt

overflow.

The presence of suspected plugs in lines attached to the base of

the column was verified by removing the lines and sectioning them for
33

examination after melting out the bismuth in an argon atmosphere. De-
posits of iron crystals, some of which were extensive enough to con-
stitute impermeable plugs (Fig. 13), were found in the locationsg in-

dicated in Fig. 12.

Salt and bismuth were drained from the column into a temporary
receiver tc permit testis to determine the condition of the molybdenum
packing. Radiographs of the column showed regions of trapped bismuth
which might indicate a plug. HMeasurements showed a pressure drop about
2.5 times that of a reference column packed with polyethylene cylinders;

however, this was not regarded as conclusive evidence of a restriction.

The column was reconnected to the system with 1/2-in.- and 3/b~in.-
0D tubing to reduce the likelihobd of vlugging the full cross section of
the transfer lines with iron deposits. In addition, a quantity of
Zircaloy~2 equivalent to 100 ppm of zirconium in bilsmuth was added to

the bismuth feed tank in an attempt to inhibit mass transfer of iron.

In the final hydrodynamic experiment (HR-8) in the original column,
the column flooded a® salt and bismuth flow rates of 75 ml/min. 8Sinee
the column had operated satisfactorily at these flov rates during run
Hk=5, 1t was apparent that the coundition of the columun had changed.

Therefore, the column was removed for examination and replacement,

The lower half of the column was filled with epcxy resin in order
to preserve the packing arrangement while the steel pipe was being
stripped away. Dendritic iron deposits were noted in the packing (Fig.
14) in an amount sufficient to increase the resistance to flow, that
is, to decrease the throughput at flocding. The iron deposits were not
uniformly distributed throughout the coluun, but were located primarily

in the lower section.

Air oxidation of the carbon-steel column was sgevere, amounting to
a loss of about 0.050 in. of wall thickness. The aluminum paint applied
initially, which was expected to withstand high temperatures, appeared

to have afforded little protection against oxidation.
3k

PHOTO 97581

 

 

?’ T T T,,,ET,,r__ g ,,,,‘r,,,T,N..,!
£y ¢ 2
g OAK HIDGE NATIONAL | ABORATORY 3

Fig. 13. Bectlion of Salt Transfer Line Between Salt Jackleg and
Column Inlet, Plugged with Dendritic Iron. Darkening of lower specimen
occurred during examination, when it was accidentally exposed to air
while hot.
33

PHOTO 98389

PHOTO 98387

 

Fig. 14. Section of 0.082-in.-ID Column Showing Dendritic Iron
Deposit (Upper Photograph). The lower half of the column was potted
in epoxy resin, and the steel pipe was stripped away.
36

6. SIMULATION OF THE FLOW CONTROL SYSTEMS FOR THE
REDUCTIVE EXTRACTION FACILITY

C. W. Kee L. E. McNeese

The successful operation of the Reductive Extraction Facility (see
Sect. 5) depends on a flow control system that will allow salt and bis-
muth flow rates in the facility to be quickly established at desired
values and to be maintained relatively constant during the course of
an experiment. Because of the importance of this aspect of operation,

a study of the flow control system is in progress.

It has previously been shown5 that steady flow from the salt and
bismuth feed tanks can be obtained if each phase discharges at a constant
pressure. This is essentially the case for the bismuth flow system since
the discharge point is at the top of the extraction column. In the case
of the salt, however, the discharge pressure is not constant since the
pressure in the salt Jackleg varies to match that at the bottom of the
column. At flow rates near flooding, changes in the salt discharge
pressure might be rapid since pressure surges could be transmitted
between the Jjackleg and column at the inherent resonant frequency;
these variations could affect the rates at which salt and bismuth flow

through the column.

It is thus desirable to simulate the transient behavior of the flow
control systems and equipment items such as the jackleg, column, and
transfer lines, which may affect coperation of the control system. This
study will (1) examine the stability of the present system, (2) examine
the merits of wvarious operating methods for the present system, and
(3) allow evaluation of proposed equipment modifications. Both analog
and digital simulations of the system have been carried out and are

described in the remainder of this section.
37
6.1 Analog Simulation of Jackleg and Extraction Column

For this simulation a stepwise change in bismuth flow rate was im-
posed on the simulated system, which was operating at steady state. The
rate of salt flow to the salt jackleg was assumed to remain constant
during the ensuing period in which the level in the salt jackleg and
the bismuth holdup in the column adjusted to the new steady-state

values.

Mathematical Analysis.— The steady-state bismuth holdup in the col-

 

umn wasg assumed to be given by the relation

o= + -—
Hl kBVd kl (Vc Vref)’ (11)

where

H1 = fraction of column volume occupied by bismuth at steady

state (bismuth holdup),

Vd = dispersed-phase (bismuth) flow rate to column,

V, = continuous-phase (salt) flow rate to column,
Vref,kl,k_2 = constants.

The instantaneous bismuth holdup in the column was sssumed to change at
a rete proportional to the difference between the instantaneous holdup

and the steady-state holdup. Thus,

where
H = instantanecus bismuth holdup,
= time,
k3 = constant.

The pressure at the base of the column was assumed to result from the static
pressure of the dispersed-phase holdup, and from flow of the continuous and

dispersed phases; the dependence was assumed to be as follows:
38

P = ' + k
= KV, RS, (13)
where
P, = pressure at the base of the column, minus the static pressure

A
when the column is filled with salt,

kh’k5 = constants.

The rate of flow of salt between the jackleg and the column was as~
sumed to ccrrespond to laminar flow in the line connecting the two volumes

and is given as
V. o=k (P, - P,.), (14)
where

k' = constant,

pressure al outlet of jackleg, minus static pressure of the

i

salt in the column.

The change in pressure with time at the outlet of the jackleg is related

to the net rate of flow from the jackleg as

wnere

density of the salt,

©
H

a = cross—-sectional area of Jjackleg,

=
i

rate at which salt flows to Jackleg.

Scaled Fquations and Analog Circuit.— Eguations (11) through {15)

 

wvere scaled for solution on an analog computer by introduction of the

following variables:
where

Py

39

o= byl
Vc = VCVM’
H = hH,,
Fp = PpPy»
Fp = PpPye

scaled steady-state holdup,

maximum steady-state holdup (1.0),

scaled salt flow rate between the jackleg and the column,
maximum salt flow rate between the jackleg and the column
(10 ml/sec),

scaled instantaneous holdup,

maximum instantaneous holdup (1.0),

= scaled value of pressure at jackleg outlet minus static

pressure at base of the column when filled with salt,

maximum value of pressure at base of column, minus static
pressure at base of the column when filled with salt (200 in.
1,00,

scaled value of pressure at the base of the column, minus static

pressure at base of the column when filled with salt.

The resulting equations are:

o EVy By 5 Vier
MTTE YR, Ve TE, (16)
M M HM
dh

a—-{ = ks(hl - h)’ (17)
Lo

 

 

= khdMVM hv + kBHM h (18)
pA P o P ?
M
B v
“Py -2 o P M (19)
- 2
dt a PM a M c
and
k'P

schematic diagram of the analeg circuit used for solution of Equations

A
(16)~(20) is shown in Fig. 15.

Choice of Constants. — The value of the constant k2 (0.1 sec/ml) was

chosen such that the steady-state holdup had a value of 0.1 when Vref and

FO were each assumed to have a value of 1 ml/sec. The constant kl was

given the value of 0.01 sec/ml on the basis of observed holdup data from

the mercury-water system and from the salt-bismuth system (Sect. 15). The
value of the constant kS approximates the ratio of the bismuth flow rate
through the column to the bismuth volume held up in the column, and was
assigned values of 0.1 and 1 sec”l; the value of 0.1 sec_'-L is considered

to give a more accurate representation of the actual system. The constant
k5, which was assumed to be equal to the product of the difference in
densities of the salt and bismuth (6.4 g/cmB) and the column height (30.25
in.), had a value of 19L in. HQO. The constant kh was varied from 0.02

to 20 in. HEO + sec/ml. The constant k', which represents the salt flow
rate (in cm3/sec) obtained between the jackleg and the column for a pressure
difference of 1 in. HEO’ was varied from 0.2 to 10 cm3 salt/sec -+ in. HEO'
A value of 0.5 is considered to be most representative of the actual systen.

. . .2
The cross-sectional area of the jackleg was varied from 0.25 to L in.

Calculated Results. Results calculated for a range of operaling

 

conditions are shown in Figs. 16-19. The greatest change in system
 

 

 

 

 

 

GRML- DWE-TO-110486

 

 

 

 

 

 

 

{integrator used for Generation of Time Scale not Shown)

Fig. 15,

Diagram for Analog Simulation of
Column.

 

Jackleg and Extraction

o
ho

 

ORNL DWG 70- 11055 Re

 

 

3 [ | 1 T T T T T T T T T T T T T T I
o
o I
» o
: ; kgy=02cnd2
ok an et
o TN e ’/ kez0O.20and2
x . e
x ky=10
o
x k'=0.2 in.H20 sec/ml
L: a =Ares Of T-6 =1in?
° k3=0.l sec-!
w
ol 1 1 b b L L b ] L . _ ]
O 5 10 15 20 25 30 35 40 a5 50 55 20 65
Time (sec)
C.20 1 1 T S mi T T T T T T T
0.15
. Haldup Aimost Some
. ‘\__m_fi For Ronge Of kg,
j e — Considered
010 —
|
i
0.05
0 1 | el L . . | [ L e L
J 5 10 15 20 25 30 35 40 as 50 55 60 85
Time {sec:
Fig. 1¢. Variation of Bismuth Holdup in Column and I'low of Salt
Between Jackleg and Column with Time for a kB Value of 0.1 and a k' Value
of 0.2 in. 1.0 - sec/ml.

2

 

 
ORNL-DWG-70-11052 RI

 

     

 

 

 

 

 

 

 

4 v
— kg 20.2 T ' 1 ‘ T !
(2]
@
)
b
€
-2
2 —— 20<kq azlin2
o y . ‘ -
. i k'=0.2 in. H,0 -sec/mi
2
g 1 | ! i i | 1
0.20 T 1 T T | | 1
0.15 —
o
’.g Holdup Curves are same for above cases
T o0} -
MW
Q.05 —
0 | [ | | ] J 1
0 2.5 5 7.5 10 12.5 15 75

Time (sec)

Fig. 17. Variation of Bismuth Holdup in Column and Salt Flow Rate
Between Jackleg and Column with Time for kg BEqual to 1.0 and X' Equal to

0.2 in. HQO + sec/ml.
Flow Rate{ml/sec)

04

ORNL DWG 70— 1!

Ca4q Ry

 

 

I

 

 

 

k'=1.0in HpO sec/mi
a,Arag of T-6 =1 in2
k5 =0.0secl

 

 

! i { i : L i L 1 1 i
5 10 15 20 25 30 35 40 45 50 55
Yime { sec¢)
#ig. 18. Variation of Salt Flow Rate Between Jackleg and Column with
Time for a k. Value of 0.1 znd a k' Value of 1 in. H.O0 * sec/ml.

3

2

80

T
ORNL-DWG-T0-11041 R}

 

{mi/sec)

Rate

Flow

 

 

 

 

| T i ] i i [ I T 1 [
-
‘ \4 in.2
. 2in.2 ~
05in2 066in2 tin2
04in2 -
0 15 20 25 30 35 40 45 50 55 50
Time (sec)

Fig. 19. Variation of Salt Flow Rate Between Jackleg and Column with
Time for a Range of Jackleg Cross-Sectional Aress,
L6

response was produced by variation of k_, the rate constant for drainage

5
of bismuth from the column. Results caiculated for the most representa-
tive value of k3 (0.1 sec“l) and a range of values of kh are shown in
Fig. 16. The salt flow rate between the Jackleg and the column and the
bismuth holdup in the column are observed toc adjust smoothly to the
steady-state values after an imposed upset in conditions. Results are

shown in ¥ig. 17 for a k_, value of 1 secnl, which represents more rapid

drainage of bismuth from3the column. As expected, the system respconse

is faster and the steady-state values are obtained more guickly. As in
the former case, the system response was smcoth and no oscillatory be-
havior was observed. Changes in the value cof kh had little effect on the
transient behavior. As shown in Figs. 16-18, an increase in the value of
kh resulted in a small decrease in the maximum deviation of VC from the

steady-state value.

As expected, an increase in the jackleg diameter (which represents
a decrease in the resistance to flow between the jackleg and the column)
resulted in a higher rate of change of Vc, as shown in Fig. 19. IHowever,
the rate of change of the bismuth holdup in the column was not affected.
The system remained stable for flow resistance values much lower than

those present in the actual system.

Decreases in the cross-sectional area of the Jackleg resulted in a
more rapid return of the system to steady state, as shown in Fig. 19.
The system was observed to be stable over the range of cross-sectional

arcas of 0.25 to U in.2

6.2 Digital Simulation of Salt Flow Control System,
Jackleg, and Extraction Column

Although the analog simulation discussed previously indicated that
satisfactory flow control should be obtained with the existing equipment,
it was necessary to treat several aspects of the system in a simplified
manner. Therefore, a digital simulation was carried out in order to

obtain a more realistic representation of the system. Changes in sim-
L7

ulation of the system consisted of the following: (1) & material bal-
ance on salt and bismuth in the column was substituted for Eq. (2) used
previously to define the rate of drainage of bismuth from the column,

(2) the column was segmented into several sections to allow variation

of flow rates and holdup at points along the column; (3} more accurate
estimates were made of the frictlonal losses in the salt transfer lines
(between the salt feed tank and the Jackleg and between the jackleg and
the column), and the diameter of the lines was varied; (4) simulation

of the salt flow control system was included; and (5) the salt feed point
location in the Jjackleg relative to the salt level in the jackleg was

varied.

Three cases were considered. In Case I the salt inlet to the jack-
leg was located above the salt surface in the Jackleg. The gas pressure
above the salt in the Jackleg was held constant, and the salt level in

the Jackleg was allowed to vary.

In Case 11 the salt inlet to the jackleg was located below the salt
surface in the Jjackleg. The gas pressure in the Jackleg was held constant,
and the salt level in the Jackleg was allowed to vary. In Case 11l the salt
inlet to the Jackleg was located below the salt surface in the jackleg.

The gas pressure was varied in order to maintain a constant salt level in

the Jackleg.

Cases II and III represent two possible methods of operation for the
present experimental facility in which the salt inlet to the jackleg is

located below the salt surface in the Jackleg.

Mathematical Analyslis of Segmented Column. — In tréating the extrac-

 

tion column, the column was divided into a number of segments, each of
which was assumed to be at steady state during a given time interval.
For the 1th segment (numbered from the top of the column), Eq. (11) was

rearranged to vield the following relation:
48

 

 

Vil T E TR 21
d,i+1 ~ k, Tk, L 5 , (21)
where
a.i+1 = bismuth flow rate leaving the ith segment,
L1+l
Hi = fraction of column volume in ith segment occupied by
bismuth,
Vc 7 salt flow rate leaving ith segment,
>
kl’kE’Vref = constants.

This equation is similar to the relation used for the analog simulation,
except that the continuous~phase flow rate is important only for flow

rates near flooding.

The relation used for determining pressure within the column

segments, similar to BEg. (13), is:

i - P = 1 i, -
Pi+l 5 (ku hi Vc,i + k5 “{l)/N, (22)

where
Pi = presgsure at the top of ith segment minus pressure due to
static head of salt in column above ith segment,
N = number of segments in column,
kh,k5 = constants.

Values used for the constants kl, k2, kh’ k5, and Vref are the same as

used previously in the analog simulation.

A material balance on the dispersed phase in segment i yields the

relation

i Va1 7 Vaan (23)
g

where

H

t time,

C . = volume of column.
vol
If one assumes that the salt and bismuth phases are incompressible, one

obtains the relation

Ve,i4n = Veo1 Y Va,ia T VaLae (2k)

During a given time interval, Egs. (21) through (24) were solved by
trial and error to determine pressure, holdup, and flow rates throughout
the column in the following manner. A value was assumed for the rate of
salt flow from the top of the column (top segment); this value was used
with the known flow rate at which bismuth enters the top of the column
to calculate values for each column segment. These values include the
salt flow rate into the bottom of the column (bottom segment) as well
as the pressure at this location. The pressure value was then used to
calculate a value for the salt flow rate between the Jackleg and the
column which could be compared with the calculated rate of salt flow
into the bottom of the column. When the two values did not agree, the
assumed salt flow rate at the top of the column was adjusted and the

calculation was repeated to obtalin a better approximation.

Mathemgtical Description of Jackleg. — In Cases I and II the pressure
above the salt surface in the Jackleg was assumed to be constant so that
changes in pressure at the jackleg outlet were related only to changes
in the level of salt in the Jjackleg. Thus, the variation of pressure at

the Jackleg outlet with time is given by the relation

dap
B p ,
7 "o Yo Ty (25)
where
PB = pressure at Jackleg outlet, minus static head of salt in the

column,
50

t = tine,

p = density of the salt,

a = cross—sectional area of the jackleg,
F = salt flow rate to Jackleg,

0
v = salt flow rate between Jackleg and column.

N+1

In Case 111 the pressure above the salt surface in the jackleg was
controlled in a manner such that the salt level in the Jjackleg remained
constant. Hence, the pressure in the gas space above the salt surface
in the Jackleg was that regquired to obtain a salt flow rate between the
Jackleg and column vwhich was equal to the salt flow rate to the jackleg.
The resistance to flow between the jackleg and column wag due to fric-
tional losses in the line between the Jackleg and column; the manner of

calculating such losses is discussed in the following section.

Calculation of Frictional Losses in Salt Transfer Lines. — The
frictional losses in the salt transfer lines (one connecting the salt
feed tank to the jackleg, and one connecting the jackleg to the column)
were calculated from the Poiseuille equation for laminar flow and from
the Blasius equation for turbulent flow. These equations are given in

Eqs. (26) and (27) respectively:

L
o= HED (26)
where
Q = salt flow rate,
D = diameter of line,
p = viscosity of salt,
L = length of line,

AP = pressure difference, minus static head between ends of line;

and 1 7°D’AP _ 0.0791
32 2 1/h
LeQ NRe

 
oL

The transition between laminar and turbulent flow was assumed to occur

at a Reynolds number of 2200, where Reynolds number 1is defined as
N, = EEQ'. (28)

In Case I, salt was allowed to flow only from the salt feed tank to the
Jackleg since the salt inlet to the Jackleg was located above the salt
surface In the jackleg. For the remaining lines, salt was allowed to
flow in the direction indicated by the pressure difference across the

line in question.

Mathematical Description of BSalt Flow Control System. — A math-
ematical description of the salt flow control system has been given
px-eviously5 for a controller having only proportional action. In the
present simulation, both proportional and integral actions were assumed.

The following relation for the control action was used:

max

T =
G + B PR

 

where

G = rate at which argon flows to salt feed tank,

B! rate at which argon is bled from salt feed tank, plus one-

half of maximum argon feed rate,
ay © Maximum argon flow rate to feed tank (1 std £t°/ar),
PB = proportional band,
RT reset time, min,
t time,
Vs,set controller set point generated by ramp generator to represent

a linearly decreasing salt volume in feed tank,

Vv salt volume in feed tank during current time interval.
52

Calculated Results. — Results calculated for Case I for several jack-~

 

leg diameters and for two salt transfer line diameters are shown in Fig.
20, Tt is observed that an upset results in a relatively rapid response
in all cases. As expected, increases in the Jjackleg diameter result in
a slower approach to steady state. The approach to steady state is also
more rapid with 3/8-in.-diam salt transfer lines than with 1/4-in.-diam

lines.

Results calculated for Case II are shown in Fig. 21. The optimum
controller settings appear to be as follows: proportional band, 4.0;
and reset time, 0.5 min. The use of a gas bleed stream from the feed
tank is seen to be quite beneficial since it allows the feed tank
pressure Lo quickly decrease without the transfer of large volumes of

salt from the feed tank.

Results calculated for Case IIT are shown in Fig. 22. Here the
approach to be steady state appears to be slightly more sluggish than

in the previous cases.

Conclusions. — The performance of the flow control system appears
to be limited by the time required for pressurization and depressuriza-
tion of the gas space in the salt feed tank. The best mode of operation
appears to be with a constant gas pressure in the salt jackleg and with
the salt inlet to the jackleg located above the salt surface in the jack-
leg, as illustrated by Case I. However, the actual system will be more
sluggish than shown in Case I because considersable time will be required
for the feed tank outlet flow rate to become constant during start-up,
even though it is not affected by variations in the column and salt
Jackleg. It is believed that the actual system will function well since
the simulated system was stable and performed satisfactorily in all the

cases considered.
53

 

ORNL DWG 70O -1IOTI RI
e e e

[ T YT Y I T T TUUTTTTTYTTT T T

Jdackiag
Diamatar
~2.28in

Flow {mi/sec)

 

 

<

4
|
=4
-+
+
+
E

.15 »
3 \\M Jackleg Diomeler
x - M__‘%“_[—ZAZS in

 

05} -
0 1 1 b bbb [ - ]
G 5 10 15 20 25 30 35 40 45 50 54 8Q 65 70
Tima {sec)
S T St SR e o s oo

 

 

 

o
® {
: \\\
€ -N Jackleg Diamater
Z \\: ~2.25in
. ~— *“'--_g:__.__‘ -
o — T T e
2 T 20 — e
- e p AT e
_— — ‘-—-..____\ e LT -‘.___‘N
| = T—— S—— e ———
Q —t~ + + + + t 4 —+ e e e -+

 

 

G5 —J
Ol b
Q 9 1o 19 20 25 30 35 40 45 50 55 60 85 70

Tims (sac!

Fig. 20. Effect of Jackleg and Transfer Line Diameter on Flow Rate
to Column and Column Holdup. Upper curves are for g 1/k-in.-diam line,
and lower curves are for a 3/8-in.-diam line.
(mi / sec)

Flow

Flow (ml/sac)

QRMNL-DWG-T70-11058

 

  
 
   
 

i i 1 i i i !

1—No Bleed

! i

 

Bleed Rate=2.4 mi/sec

 

 

 

 

 

 

 

 

 

 
    

 

 

i i
(@} Flow out of Sait Jackieg Jackleg Area=2-1/4 in2
} % t 5 } } } Proportiona!l Bond = & -
FRese! Time = 0.5 min
r——NO 5=EEG T I
— ]
!
L Bleed Role=2.4 ml/sec _
{t) Flow out of Feed TYonk
} { | | 1 ] ] ] i
10 20 30 40 SO 60 70 80 90 100

Time {sec)

'ig. 21,

Variation of Salt “low Rate from “eed Tank and Jackleg

. el TT e T . ) - N
with Time Tor Cases With and Without Gas Bleed from Feed Tank (Case II;.
Flow (ml/sec)

Flow (ml/sec)

ORNL DWG 70-11O60R!

 

 

 

 

 

 

 

 

 

| i [ T I T I ] [
4 S TT——
Jackleg Area =2-i1/4 int
Bleed Rate = 2.4 mi/sec
2 Proportional Band =2 _
Reset Yime = 0.2 min
0 % | + % : | i | |
4 —
Jackieg Area =2-1/4in %
Bleed Rate = 2.4 ml /sec
Proportional Band =4
Reset Time =0.5 min
2 —
o i } i I 1 } I i }
0 0 20 30 490 50 60 70 80 90 100
Time {sec)
Fig., 22 Variation of Salt Ilow Rate from Jackleg with Time for
Case 111
56

T. CALIBRATION OF AN ORIFICE--HEAD PCT FLOWMETER WITH
MOLTEN SALT AND BISMUTH

C. W. Kee b, A. Hannaford

Use of an orifice-~head pot flowmeter, described previously,6 is
being considered for determining flow rates for molten salt and bismuth
streams in systems used in engineering development work. Since the flow-
meters under consideration are not of a standard design, experimental

determination of the orifice discharge coefficient is in progress.(

Additional calibration data were cobtained during both steady-state
and transient experiments with molten salt and bhismuth in a flowmeter
having a 0.118-in.-diam orifice. Results of these tests are given in
the remainder of this section. The data suggest that the orifice dis-
charge chamber was flooded part of the time; this would produce a sig-
nificant errcor in the measured pressure drop across the orifice. The
remaining data appear to have been obtained under conditions where the
discharge chamber was not flooded. In future experimeants, the orifice--
head pot and discharge chamber will be pressurized to ensure that the

orifice discharge chamber does not fill with liquid.

7.1 Data from Steady-State IFlow Experiments

Three calibration experiments were made in which a steady bismuth
flow was maintained through the mild-steel head pot, and one experiment
was carried out with a steady salt flow. Results from these experiments
are given in Tables 3 and 4. The data for salt flow appear to indicate
that higher orifice coefficients are assoclated with higher Reynolds
numbers. The values for the orifice coefficient also appear to approach
a constant value at high Reynolds numbers. The orifice coefficient is

defined as

Cp = e (30)
A 2pAP
1
-3

Table 3. Orifice Discharge Coefficient Data Obtained
with Balt Flow During Run 0OP-5

 

 

Flow Period of

Rate Reynolds Orifice Steasdy Flow
{(ml/min) Number Coefficient (min)

132 278 0.28 8.3

190k LOT 0.3k 5.3

25 515 0.36 4.3

296 623 0.37 3.7

 

a . ]
Based on orifice diameter.

Table 4. Orifice Discharge Coefficient Data Obtained with Bismuth Flow

 

 

Flow Pericd of

Rate Reynolds Orifice Steady Ilow
Run (ml/min) Numberd Coefficient {min)
IP-5A 185 12070 0.77 3.75
OP-6 127 8283 0.68 3.7

182 11840 0.76 .7

P56 166414 0.7k 1.7

200 18870 0,84 3

282 18363 0.81 8.2
OP-T7 175 11410 0.734 13

20D 1hhho 0.773 9

397 25830 1.11h 3

368 230940 0.773 1

 

a, . )
“Based on orifice diameter.
58

where
CD = orifice coefficient,
Q = liquid flow rate,
p = density of liquid,
A = cross-sectional area of the orifice,
AP = pressure drop across orifice.

Data from the experiment with bismuth flow do not show the trend
observed with salt flow; however, the Reynolds numbers during these ex-—
perinents are about two orders of magnitude higher than in the experiment
with salt. One of the calculated values for the orifice coefficient
(1.114) is much higher than expected. This higher value is believed to
be the result of the orifice discharge line being filled with bismuth,

which led to a decreased pressure below the orifice.

7.2 Data from Transient Flow Experiments

Results are shown in Fig. 23 for an experiment (OP-5) in which
drainage of salt from the orifice-~head pot was observed. As seen in
eagrlier experiments, the orifice coefficient appears to decrease as the
pressure drop across the orifice decreases. This is believed 1o result

from submergence of the orifice.

Some of the data obtained during drainage of bismuth from the orifice—-
head pot are shown in Fig. 2. These data indicate lower values than ex-
pected for the orifice coefficient. The lower values are probably caused

by the filling of the orifice discharge chamber.

The remainder of the data obtained during drainage cf bismuth from
the head pot is shown in Fig. 25. These data are believed to be more
reliable than those shown in Fig. 24, The values for the orifice dis-
charge coefficient are in gecod agreement with the expected values and

with each other.
y
Square Root of Pressure Drop Across OrHice(dwws/mnz)z

280

240

200

160

{20

w
Q

&
Q

ORNL-DWG-70-11065

 

 

 

 

¢ ! T T T ! Y I f
»_ —
"-:.\ —

"~
\t
\.\
.\ Cd =0.43
b - \.f —
\.\
"%m\
i \\\ - fcé =0.36 B
-\N\‘t::'-a.._“
.\-"‘4-.“___' fcd =O.25
B ~f \k\'\-_ - ]
\

} i 1 L 1 | 1 k i

0 0.5 1.0 1.5 20 2.3 3.0 3.5 4.0 4.5 5.0
Time {min)
Fig. 23. Data Obtained During Drainage of Salt from Orifice--Head

Pot Having an Orifice Diameter of G.118 in.

64
60

ORML-OWG-TO-11063

 

320 I 1 T T T T T T
£
. 280}~ _
~.
~.
\'\‘rcd =0.48
240 ~, B
\'
\-
™.
.
\l
~
200 " i

\Q\.\
~Cq =0.35
IGO.T\\\\\\\ N ]
\-
- \I

//
/

/

/

Squore Root of Pressure Drop Across Orifica (dynes/cm?2)

 

 

 

" \“-.
\.\\. \‘:\1
~.
80—
40— -
0 1 1 | ] ) | | ]
0 0.5 1.C -] 20 2.5 3.0 35 40 45
Time [min )
Fig. 24. Data Obtained During Drainage of Bismuth from Orifice--—

Head Pot Having an Orifice Diameter of 0.118 in.
61

b2

ORNL DWG 70-1i056

1

 

N
b
Q

—

200}

+\./Cd =0.65
160+ N\, .

N\ Cq = 0.59
120 + \(/ i

 

 

 

Square Root of Pressure Drop Across Orifice (dynes/cmzl

80} -
40}~ ~
0 i 1 i
0 0.5 .0 1.5 2.0
Time {min)

Pig. 25. Data Obtained During Drainage of Bismuth from Orifice==
Head Pot Having an Orifice Diameter of 0.118 in.
62

8. ELECTROLYTIC CELL DEVELOPMENT: STATIC CELL EXPERIMENTS

J. R. Hightower, Jr. M. S. Lin
L. E. McHeese

The proposed flowsheet for processing a molten~salt breseder reactor
requires the use of electrolytic cells for reducing lithiuwm and thorimm
fluorides at a bismuth cathode and for oxidizing materials from bismuth
solutions at a bismulth anode. Two experiments related to cell develop-
ment were made in a U-in.-diam quartz cell. In the first, current was
passed between two molybdenuin electrodes to determine whether the ususl
dark material would be formed when bismuth was not present in the systemn.
The second experiment was a test of a porous carbon ancde having a much
higher gas permeability than the graphite anode that had been used in an
earlier experiment in which poor results had been obtained. It was
believed that a higher gas permeability might allow removal of the CFh
formed a2t the anode (by flow through the graphite) and thereby prevent
the electrode surface from beling insulated by a layer of this material.
The salt (66-34 mole % LiFmBeFE) and bismuth were sparged with H, before
being transferred toc the quartz vessel.

For the first test, the depth of salt in the cell was 3 in.; the cell
contained no bismuth. The anode was a 1/4-in.-diam molybdenum rod immersed
to a depth of 1 in. at the center of the cell. The cathode was a 1/L-in.
molybdenum tube inserted to within about 1 in. of the bottom of the cell
and located about 1/2 in. from the cell wall. A gquartz tube surrounded
the cathode, and only about 1/L in. of the molybdenum tube was exposed
since it was to be used as the electrical lead in the second experiment.
Initially, an alternating current of about 2 A was passed between the
electrodes for approximately 2 min with a cell temperature of approximately
600°C. At the end of this time, a small amount of black material was ob~
served near the electrode having the quartz sheath around it. A direct
current of about 5 A was then passed between the electrodes. There was

a rapid formation of bhlack material in the vicinity of the anode, and the
63

salt became opaque within 1 min. After standing overnight, most of the
material had settled out, although a few large particles (~1/8 in. in
diameter) were still suspended in the salt phase. Current (5 A, dc)

was agaln passed, and the cell rapidly became opaque again; the black
material that formed at the anocde rose along the anode, spread out at

the salt surface, and dispersed throughout the salt from the surface.
Analysis of some of the solid material suspended in the salt indicated
that silicon might have been present. It was concluded from this experi-
ment that the presence of bismuth is not necessary for formation of the

black material.

After the test with the molybdenum electrodes, sufficient bismuth
was transferred to the cell to produce a 3-in. depth of this metal, and
the molybdenum rod at the center of the cell was replaced with a l-in.-
diam porous carbon electrode (density = 0.1 g/cc) immersed to a depth

of 1 in. in the salt.

With the cell at 600°C, a potential of 5 V was applied between the
bismuth cathode and the carbon anode. The initial current, approximately
4 A, decayed rapidly to a steady value of about 0.2 A (corresponding to
a current density of 0.0k A/cmg, based on the area of the end of the
carbon anode). The cell was then evacuated to an absolute pressure of
less than 29 in. Hg without affecting the current density. Oparging the
cell periodically produced short-term increases in current, but continuous
sparging did not help appreclably. If the CFh had escaped from the carbon
reaction surface at a rate limited only by resistance to gas flow through
the porous electrode, a current density of 12 to 29 A/cm? would have been
obtained. Apparently, an insulating film of gas was formed on the anode.
Alternatively, with such a high permeability, the salt may have penetrated
the electrode a short distance, thus isclating the reaction surface from
the salt-free porous carbon and nullifying the benefit of the low resist-~

ance to gas flow offered by this material.

We are continuing experiments in static cells 1n order to identify the
black material. We are installing an all-metal cell to determine whether

the material forms in the absence of silica.
6L

9. ELECTROLYTIC CELL DEVELOPMENT: FORMATION OF FROZEN
FILMS WITH AQUEOUS ELECTROLYTES

J. R. Hightower, Jr. C. W. Kee

Formation of frozen salt films on structural surfaces exposed 1o the
molten-salt electrolyte in cells is being considered as a means for pro-
tecting these surfaces from corrosion by BiF% produced at the cell anode.
A study of frozen film formation under condifions similar to those ex-
pected in a cell is in progress. Part of the work will be carried out
using an aqueous electrolyte in order to avoid problems associated with

molten salt--bismuth systems.

Equipment has been installed for studying the formation of ice films
on surfaces of a simulated electrolytic cell that uses an agueous solution
rather than molten salt as the electrolyte. Alternating current will be

used to minimize the effects of electrode reactions and mass transfer.

The remainder of this section will describe the equipment that has
been built, and outline a mathematical analysis for use in designing

cells and interpreting results.

9.1 Mathematical Analysis

Average Currvent Density and Power Dissipation. — One conceptual

 

design for an electrolytic cell consists of flowing molten bismuth

electrodes (alternately ancdic and cathodiec) in long, narrow trays

that are closely spaced and are submerged in the electrolyte. The

divider between the electrcdes mast be electrically insulating, and
the top edge of it must be protected from corrosion by a film of

frozen salt.

Figure 26 is a diagram of a half-anode-~-half-cathode combination
and shows a convenient idealization of an actual cell that will be used
in cell design. The current 1s assumed to be confined to an electrolyte

region bounded by the frozen film at a radius of r, and by a hypothetical

1
ORNL DWG 70-6245

|
HYPOTHETICAL ¢

INSULATING —
SURFACE
FROZEN |
FILM '/
COOLING {

 

 

 

TUBE

 

 

§=

ELECTRODE
DIVIDER

 

 

| CATHODE W ANODE

 

 

Fig., 26. Diagram of Idealized Model of Electrolytic Cell.

59
66

insulating surface at a radius of I, {which is half the width of an

electrode tray). A cooling tube of radius r. is positioned at the

| 0
top of the electrode divider and is covered with a frozen film of
thickness £y "~ Ty The cooling tube is assumed to be electrically
insulating.

The current density in the conduecting region is given by Ohm's

law as:
¥ >
J = gk, (31)
where
- . 2
J = vector current density, A/m”,
_).
E = vector electric field strength, V/m,
g = conductivity of the electrolyte, ohm”l m—l.

With no sources of emf in the conducting region, the field strength

ig derivable from a scalar potential:B

B = -grad O (32)

where
U = electric potentizl, V,

grad U = gradient of U.

In cylindrical polar coordinates, Eg. (32) becomes

(33)

—)' "> . - - - -
where &, and ae are unit vectors in the r and 8 directions respectively.

Substitution of Eg. (33) into Eq. (31) yields the relation

= -+ U > 1 93U
= e e o —)
J & (ar or ae r 88) (34)
67

In the conducting region, if one assumes that g is constant,8 the po-

tential U satisfies the differential equation

VU = 0, (35)

59Uy, 3%u _
rar (rgp) + 02 0. (36)

The boundary conditions assumed for the model shown in Fig. 26 are:

1. The radial current component at r = r. is 0, so that

 

1
ol _
3T ey = 0.
1
2. The radial current component r = r, is 0, so that

ol

- = Q.

r=r

 

2

3. The potential at the anode (8 = 0) ig V
at the cathode (6 = w) is O.

1° and the potential

Bquation (37),

- g
U=V (1 - =), (37)

is a solution to Eg. (36) in the region r, £r<r,and 0 <0 < m/2
zand satisfies the boundary conditions. Substitution of Eg. (37) into

Eq. (3L4) results in the following equation for current density:
=8, —. (38)

From Eq. (38) it can be seen that current flows only in the angular
direction; the radial component is everywhere zero. This relstion will
be used to calculate the average current density in the cell and to

calculate the power loss due to the resistance of the electrolyte.
68

The power loss can be calculated by integrating the specific power
dissipation over the volume of the cell. The specific power dissipation

is given by:

2 3

p==J , in W/m”, (39)

1
g

where J 1s the magnitude of the current density; and the total power dis-

P :f p dv, (h0)

where V 1s the volume of the region carrying current. The average current

sipation is given by

density is given by r

 

= = (41)

where Rln = (r2 - rl)/ln (rg/rl).

The power dissipated in the cell is calculated from Eq. (L0) and is given
by:

R
l 2 l *
. g avg‘fi—gw" in ¥, (h2)

avg

P

. _ 4 .
where Ravg l/E(r2 rl)

If a coolant at temperature TC (Tc < Tf, the freezing temperature of
the electrolyte) flows through the cooling tube along the electrode divider,
a frozen film will form and grow until the total resistance to heat flow
from the bulk of the electrolyte causes the rate at which heat flows into
the tube to be equal to that at which heat is carried away by the coolant.
At steady state, the heat carried away will be equal to that generated in
the hulk of the electrolyte, minus that lost through other surfaces (e.g.,

the top surface of the electrolyte). For these calculations, we will
69

assume that a known fraction of heat generated in the semiannular region
between rl and r2 (see Fig. 26) is removed through the cooling tube. We
will assume that all heat flow through the frozen film and tube wall isg
in the radial direction. We will also assume that the resistance to heat
flow through the tube wall and through the coolant is negligible when
compared to the resistance to heat flow through the frozen rilm on the
outside of the tube. Under these assumptions, the equation relating the
frozen film thickness to the heat generation is:

TLk f - T )

TO <, (43)

 

In

(T
T TN
{
\ 7

Q = heat transferred through the frozen film, W,

L = length of cooling tube, cm,

T, = freezing temperature of electrolyte, °C,

k = thermal conductivity of the frozen film, W/°C-cm,
Tc = temperature of coolant, °C,
rO = oputside radius of coolant tube, cm,

t = frozen film thickness, cm.

Equations (L1), (42), and (L43) are only approximate and thus can be
used only as guides in the design of equipment and for predicting its

performance.

0.2 Experimental Equipment

Coolant System. — A flow diagram of the coolant system is shown in
Fig. 27. This system consists of cooling tubes (that pass through the
electrolytic cell) for removing heat from the cell, a chiller for removing
this heat from the system and chilling the coclant (trichloroethylene)

below the freezing point of water, a pump to recirculate the coolant, and
ORNL DWG 70-6246

 

SURGE
TANK

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

COOLING TUBES
IN
PCREY X ELECTROLYTIC
BATH CELL
PUMP
s ROTAMETER ROTAMETER
0—10 GPM 0-20 GPM
>
ig. 27

Flow Diagram of Agueous Mock-up Coolant System.

0l
71

flownmeters and valves to control the flow. There are three coolant cir-
culating loops that allow the temperature of the coolant at the cell inlet
and the flow rate of the coolant through the cell to be zdjusted inde-

pendently.

The chiller consists of a 55-gal stainless steel drum containing a
coil made of 50 Tt of 3/h-in.-diam copper tubing. The tubing 1s immersed
in a trichloroethylene--dry ice bath, the temperature of which should be
at least as low as ~50°C. Dry ice will be added at a rate of 0.24 1b per
minute per kilowatt of heat generated by the cell and punp.

Ixperimental Cell Vessel.— The first cell vessel to be used in these

 

experiments is made of Plexiglas and contains two trays (1-7/8 in. wide x
12 in. long x 2 in. deep) separated by a 1/4-in.-thick divider; the trays
will contain mercury pools for electrodes. A diagram of the cross section
of the cell is shown in Fig. 28; a photograph of the cell vessel installed
in the cocling loop is shown in Fig. 29. The wider sections at each end
of the cell vessel accommodate installation of the cooling tubes. The
maximum heat generaticn rate of interest corresponds to an average current
density of 5 A/cm2 in molten salt having a resistivity of 0.65 ohm-cm (72~
16-12 mole % LiFmBeFQ-ThFu) at 600°C. TFor KBr or KI sclutions (27 and 33
vt %, respectively) a resistivity of about 8 ohm-cm can be cbtained with

a freezing point of -10°C. With these solutions, the heat generation rate
of interest could be oblalined with an average current density of about

1.5 A/cmg. Calculations show that, if this heat is removed through the
electrode divider cooling tube, an ice film having a thickness of only
avout 0.05 in. will form. Consequently, other cooling tubes were added
along the walls of the cell vessel to reduce the heat load on each tube.
In an actual electrolytic cell, these surfaces would require corrosion

protection also.

9.3 Expected Mode of Operation

Temperature measurements of the electrolyte, electrodes, and coolant

fluid, and measurements of current and voltage, will be made in order to
72

ORNL DWG 70-4541

AQUECUS
ELECTROLYTE
SOLUTION

 

 

 

 

COOLING
TUBES

ICE -
FILMS

LUCITE
CelLL
VESSEL

 

[

00 ©

|2

 

  

 

 

 

 

 

 

 

 

MERCURY
ELECTRODES

Tig. 28, Diagram of Cross Section of Electrclytic Cell to Be Used
with Agueous Electrolytes in Studying ormation of Ice Films.
13

PHOTO 97765

 

Fig. 29. Photograph of Cell Vessel for Studying Frozen Film
Formation with Aqueous Electrolyte Solutions.
Th

make energy balance calculations on the system. Measurements of the
amount of heat that is removed by each cooling tube should aid in

determining heat generation rates at various points in the cell.

The center tube will have the greatest heat removal rate and the
thinnest ice film. Observations of frozen film thickness, shape, and
stability will be made. In addition, qualitative observations will be
made of the behavior of the ice layer below the surface of the metal

electrodes.

10. DEVELOPMENT OF A BISMUTH--MOLTEN SALT INTERFACE DETECTOR

*
Jd. Roth L. BE. McNeese

An interface detector based on the principle of eddy current genera-
tion in liquid metals is being developed for use at molten salt--liguid
bismuth or gas--liguid bismuth interfaces. It is desired that all surfaces
contacting bismuth be made of a refractory metal such as molybdenum; however,
these materials have a relatively low electrical permeability, which makes
their use difficult. The device under development is a medification of a
liquid-level induction probe developed at Argonne National Laboratory9
and has been described previously.lo The design of the inductance coil
has been modified so that all lead wires are accessible from one end;

this simplifies the replacement of lead wires.

10.1 Testing and Modification of Electronic Circuit
Prior to Installation

Initial testing of the electronic circuit prior to installation
established the need for several modifications. An attempt was made to
establish component compatibility using the expected materials of con-
struction. Core materials tested in the inductance coil included aluminum,
Woods metal, and bismuth, in air, in a 304 stainless steel sight tube,and

in the required 347 stainless steel--molybdenum duplex sight tube that is

 

*
On loan from Combustion Engineering.
5

described below. The modifications involved changes in signal ampli-
fication, as well as stabilization of the signal received from the

detection coil.

The electronic amplification factor was first increased by a factor
of 10, which then necessitated replacement of the dc¢ amplifier with one
generating a "cleaner" signal., After these modifications had been com-
pleted, it was found that, with the 347 stainless steel~-molybdenum sight
tube, insertion of a Woods metal rod into the coil caused a pen deflection
of 90% of full scale. Insertion of a bismuth rod into the coil caused a
pen deflection of only 24% of full scale; however, this deflection is
believed to be sufficiently high if the signal-to-noise ratio can be
increased. Toward this end, several additiocnal medifications were made.
All electrical lead wires were shielded. The detector coil was shielded
from the furnace element by a sched L0 type 347 stainless steel pipe, which
essentially eliminated induced signals generated by the Turnace element.
All segments of the electronic circuit were connected to a common ground,
which reduced the signal variations caused by differing ground potentials.
All electronic components, with the exception of the furnace element, were
connected to a l-kVa Sola constant-~voltage transformer. Final testing of
the system at room temperature revealed the presence of a 2-V impressed
signal that could not be eliminated electronically. This signal was found
to be dependent on lead wire location and was eliminated by proper place-
ment of the lead wires. Upon completion of this phase of the work, it was
found that noise level had been reduced from about 20% of full scale to

about 6% of full scale.

Testing of the system at temperatures above room temperature was
then started by first heating the coil slowly to about 250°C in an
attempt to dry out the detector coil and to determine the effects of both
heat and current flow through the furnace coils on the detector coil. The

following coneclusionsg were drawn as a result of these tests:

1. Loss of ground connection to the stainless steel furnace
shield caused a pen deflection of 9% of full scale. This
indicates the need for good ground connections on all

portions of the system.
76

2. The recorder responses resulting from full insertion of a
bismuth rod into the detector coil were 4% and 20% of full

scale at room temperature and at 220°C, respectively.

These data indicate that the sensitiviiy of the system increases with
increasing temperature and that a full-scale pen deflectioa might be
observed at the operating temperature of 600°C. However, the electrical
properties of bismulh exhibit a discontinuity at the melting point, which
mey decrease the sensitivity. Tt was decided to modify the electronic
circuit such that the output signal will be amplified by an additional

factor of 10.

The optimum coil operating frequency incresses with increasing tem-
perature; however, the corresponding output signal decreases with increas-
ing frequency. In addition, the electronic zero is also dependent on the
frequency. At room temperature, the optimum fregquency was 47 kHz, and an
output of 8 V was observed. At 150 to 500°C the optimum frequency was
59 kHz, with a corresponding output of 6 V. The electronic zero drifted
over a periocd of time. A Zener diode feedback circuit will be installed

in the system in an attempt to stabilize it.

10.2 Fabrication of Sight Tube

The seccnd attempt by the Metals and Ceramics Division to coat the
internal surface of a 0.500-in.~diam type 347 stainless steel tube with
0.005 to 0.010 in. of tungsten by chemical vapor deposition was success-
ful. However, repeated attempts to coat the external surface of the tube
failed because of an inability to produce a continuous bond between the
internal and external coatings. Faillure of this technique to establish
a suitable bond between the two surface coatings was attributed to the
high stress levels that are inherent with this type of joint. These
stress levels are caused partially by the differential thermal expansion
characteristics of tungsten and stainless steel and partially by the
failure of the vapor-deposited tungsten to bond to the stainless steel
substrate. No further attempts will be made to prepare sight tubes by
this method.
77

The sight tubes required for the test apparatus have been produced
by machining the external surfaces of several 36-in.-long pieces of
molybdenum tubing, which were originally 0.375 in. OD with a 0.0LO-in.
wall, to produce a wall thickness of 0.02 in. Although this thickness
is considerably greater than anticipated, tests indicate that satisfactory

results can still be obtained.

11l. CONTINUOUS SALT PURIFICATION

R. B. Lindauver E. L. Youngblood
L. E. MclNeese

Molten salt for the MSRE and for development work is presently puri-
fied from contaminants (mainly sulfides, oxides, and iron fluoride) by a
batch process.ll The total cost of salt (natural lithium) thus produced
has amounted to about $l600/ft3, of which less than 40% is material cost.
Tn April 1968, a2 commercial vendor bid $266O/ft3 on a 280~ffi3 quantity.
It is believed that the labor costs could be reduced considerably by use
of a continuous process for the most time-consuming operation (hydrogen
reduction of iron fluoride). Although the removal of sulfides and oxides
by the batch process 1s fairly rapid, additional advantages would probably

be gained by performing this operation continuously.

Preliminary experiments on the reduction of ilron flucride in molten
salt were performed by MIT Practice School students in 1968 with encouraging
results. A substantial reduction in the concentration of iron fluoride was
obtained, although most of the operation was carried out at less than 2% of
the calculated flooding rate. For the present studies, the column diameter
has been reduced from 3.375 in. to 1.38 in. (1.25-in., sched L0 pipe) to
allow operation nearer the flooding point. The column is 81 in. long and
is packed with 1/4-in. Raschig rings. Equipment has been designed to
supply purified hydrogen, argon, and HF gas in sufficient gquantities
to operate at the higher rates. Figure 30 is a simplified flowsheet of
the system. Molten salt is charged toc the receiver tank and then trans-

ferred via pressurization to the feed tank, which is elevated about 5-1/2
 

@ ORNL-DWG-59-14298

r_1
-———*(\_J BACK PRESSURE
— CONTROL VALVE Y ¥

 

 

 

 

 

 

 

 

 

WATER FLUORIDE
ANALYZER ANALYZER

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ARGON § St)g
l Naf
| TRAP
E Pd T COLUMN
3
M VENT
; = —= OUTSIDE
SALT BLDG.
FEED | ABSOLUTE FLAME
TANK | | FILTER ARRESTER
. 7
HYDROGEN {& HF)}
SAMPLER

 

 

FILTER

 

 

 

 

RECEIVER

 

 

 

-

Fig. 30. Flow Diagram of Eguipment for Study of Coatinuous Salt
Purification.

gl
9

ft to reduce the argon pressure required to feed salt to the colwumn.

The salt feed rate is contrelled by adjusting the argon flow rate to

the feed tank. In order to limit the stress on the feed tank, the tank
is maintained at 650°C; additional heat is supplied to the feed line to
increasze the temperature of the salt to the column operating temperature
of TO0®C. The hydrogen is also preheabed to 700°C before it enters the
bottom of the column. Hydrogen leaving the top of the column flows
through & pressure contrel valve In order to maintain the sslt-gas inter-
face at the bottom of the column below the gas inlet. Balt leaving the
bottom of the column passes, first, through a liguid seal lovp and, then,
through & Tibrous metal filter Lo remove particulate iron., The filtered
sa’t passes through a Tlowing stream sampler and iz subsequently routed
to the receiver. Hydrogen Tfluoride that is produced during reduction is

removed by a Nal trap.

Instrumentation is provided to measure the differential oressure
across the column, the pressure at the top of the column, and the salt
head above the salt filter. An analysis for water in the system off-~
gag 1s made during oxide removal to determine the rate at which oxide
is removed, and an analysis for fluoride in this stream 1s made during
the iron reduction step to provide a check on the rate of reductlon as
indicated by the analysis of filtered salt samples. Periodic examination
of the filter (and possibly of the column packing) will provide informa-

tion on the particle size, filterability, and disposition of the iron.

12, MSRE DISTILLATION EXPERIMENT

J. R. Hightower, Jr. L. E. McNeese

95

Results of final analyses have been obtalined for Li, Be, 2r, 21,

! 5S¢
147Pm lSSEu, 91Y, 908r 09 137

1h!
”h4Ce ” Sr, and Ceg in the 11 condensate

> 2

samples taken during the MSRE Distillation Experiment. These results
(shown in Table 5) have been used to calculate the effective relative
volatility of each component, with respect to LiF, during the course

of the experiment by metheds described previously.ld
41

 

 

 

 

Table 5. Anslyses of Samples from the MSRE Distillstion EX
T T Comoonent - - 5 Salt \{olmnes Aésociated
o 9szr o te e Tk ’ Y ’ S ’ . Tot:itfiiizsgemaz:tz:spésiuensate
(wt %) (wt. %) (d¢is min~lg~1) {daie min~T1g™t) (dis minTdz™Ll) {dis min~lg~1) (dis min~lg-1) (dis min-lg-1) {dis min—lg-1) {dig min g ) Fed Coliected

Tuel storage tank-1 — — 1.79 x 109 314 x 1010 3,48 x 109 — 3,11 % 107 —
{Date analyzed) (b/2L/69) {(4/25/6G) {4/24/69) L/2l/e9)

Fuel storage tank-2 0.6 13,39 1.62 x 109 3,20 x 1010 3.75 x 107 — 3.35 x 209 —
{Date analyzedi) (5/1/69) (s/1/69) (4/29/€9) (5/2/69) {1/30/69) (4/29/69)

Condensate samples 2,88 15,66 .52 x .07 6.83 x 10°  <2.6 x 10° <8 x 103 6.5L x 2.1 x 105 2.10 x 10" 9. 7.9 0.42
-2 4.7 12,05 1.39 x 107 1.21 % 109 1.53 x 107 “1.b % 107 1,18 x 5.2 x 10% 5.15 x 107 6. 10.0 1.93
-3 6.86 10.0L 1.1 x 107 2.39 x 20° 2.78 x 167 n3.0 & 10° Lk x 8.5h x 10° 1.61 x 107 3.8 i1z 5.96
-4 7.0k 10.09 Lo x 107 1.97 x 10° 2.23 x 107 we.l x 167 1.32 x 5,76 x 105 .3y x 107 3. 12.6 3.81
-5 8.15 10,48 1.oh ¢ ig? B.09 x 207 8.85 x 1° “1.0 x 107 6.1k x 2.5 x 208 2.49 x 107 3. 3.4 .35
-6 7.05 9.1 10,43 1.22 x 10° 2.89 x 10° 3.20 x 107 v3.4 x 107 1.81 x 8.93 x 108 1.08 x 107 o. 13.6 k.95
-7 - .82 3. 10.53 1.16 x 109 1.99 x 1P 216 x 107 V2.5 x 107 1.15 x .57 x 10° i.90 x 107 1. 13.8 5.80
-8 8.77 8.1 9.83 11T x 107 3.0L x 10° 3,08 x L0 v3.3 x 107 60 x 8.75 x 10° 3.38 % 107 5, 13.8 6.34
-9 9.80 7. 8.71 .78 x 20° .63 x 108 5.98 x 107 n5,5 x 107 29 x 1.56 x 107 3.50 x 107 L, 13.8 6.89
-10 .02 5.1 8.26 g. bk x e .75 x 105 6.27 x 107 w6.6 x 107 .01 x 2.37 x 107 1.69 % 107 2. 13.8 7.49
-1 .28 5.7 8.1l 9.09 x 100 Lol x _of £.90 x 100 n8.1 x 10° L0 x 3.0h x 107 b.z x 100 o, 13.5 7.85
(Date anaiyzed}  {5/21/€9) {5/23/69) (4/23/69) (5/2./69) (7/21/499) (7/2/69) (10/8/69) 9/30/69) (7/8/69) {7/6/69)

 

 

aDuplicate samples differed by a Factor of 4 or more.

08
81

The effective relative volatility of each component, with respect
to LiF, is shown in Figs. 31-3L4 as a function of the volume of collected
condensate. The most self-consistent values were obtained when the car-
rier salt composition in the feed was assumed to be 65-30-5 mole % LiF-
BeFQ-Zth rather than a slightly different composition as indicated by
the analysis of the salt from the fuel storage tank. The difference be-
tween the analytical wvalues and the values used is probably due to the
presence of zirconium metal in the fuel storage tank. The zirconium was
filtered cut when the salt was transferred to the still feed tank.

The effective relative volatilities of the major salt components

BeF2 and. ZrFu and of the fission product 95

The effective relative volatility of BeFB, which was essentially constant

Zth are shown in Fig. 31.

during the run, had an average value of 3.8. This value agrees favorably
with the value of 3.9 reported by Smith, Ferris, and Thompson,l3 but is
lower than the value of 4.7 measured by Hightower and McNeese.lh Slight
inaccuracies in calculations (especially the material balance calculations
described earlier) and analyses probably account for these differences.

95

The effective relative volatilities of fission product ZrFu and of
nztural ZrFu are in agreement. When the analysis of the fuel storage tank
salt was used in the relative volatility calculation, the resulting relative
volatilities of the natural Zth were about a factor of 2 lower than the

values for QBZth. Figure 31 shows that o decreagsed from an intial

LrFy~LiF
value near 4 at the start of the run to about 1 at the end of the run.
These values bracket the value of 2.2 measured by Smith and co-workers in
salt mixtures having a ZrFu concentration about 2% of that used in our

system.

The effective relative volatilities for the lanthanide fission products,

. 1h7PmF3’
tive relative volatility of lb’L‘LCeF3 increased sharply from 6.1 x J.O“l‘L at the

luhCeF

and 155EuF2, are shown in Fig. 32. The value of the effec-
time of the first sample to about 1.0 x 10_2, where it remained for nearly
all the subsequent samples. The fifth sample, however, was lower than 1.0

x 1072 by a factor of 3. The low initial value was 1.5 to 3.l times the
value measured in an egquilibrium St111512 the higher values were 24 to 56

times the value measured in the equilibrium still.
RELATIVE VOLATILITY OF BeF,

WITH RESPECT TO L.iF

o

H

ORNL-DWG 70 - 4516

 

 

 

 

 

 

i 1T T T T T T T T390 4
dg &
— o3
—+5 N
0 4 &
A >
—
0 1 =
0 o o a
A O A 42 Z
A A g
O zZrF, 8 %2 o
G >
=
A A 4 Gj
— o
® ® 7
® ® - ® o BeFp |
&
@ $
#® Bef,
& ZrF,
O %87,
| | i | i i L | 3 | 1 |
2 3 4 5 6 T 8
LITERS COMDENSATE COLLECTED
Effective Helative Volatilities of ZrF), 95Zth, and

Ber',, for the MERE Distillation Experiment.
o

WITH RESPECT TO LiF

cg
83

ORNL- DWG 70— 4515

 

10~ T

 

 

 

 

 

 

T 7 | ' [ T I ! I 1 I 1 ‘ i
e
@ . 144
1072 ® Ce
- ® ® ® o
o - . ~
I?:- — -
3 ] A ]
P u A A R
3 A AA
g " A ' A HTqu
W A
b L iy
-
L
@x
1073 4 —
i ® 14co ]
- [® A "Tpm .
.. O 158 Eu
O
. 0 o O o 4
O
i o o 158
0o
<1.4 x 10°° 0.86 x 107*
10-4 __{ 1 b J I 1 1 | 'K 1 | ] | ] i
o 1 2 3 4 O 5 6 7 8
LITERS CONDENSATE COLLECTED
: : : e L4 L
Fig. 32. Effective Relative Volatilities of 1 Cel's, 1 7PmF3, and

155E}uf’2 in the MSRE Distillation Experiment.
8l

ORNL DWG 70-4514

 

 

 

 

 

 

 

'0“1, T ] | ] T ] T ] T T ( | T T T
2
= ‘ e
&
-—Q ‘
L. . o1
= & Y
1072} ]
e " . :
- i ]
5 i
P— ot
& A & A
- N fi c
t &\ 905r
Z . A i
> A
}.—
a
S| A d
L~ 4
-
O
>
5
-3 4 —
2 10731 A5 1
J - i
o - ® 'y -
8 & %g; |
10-4 i | j ] s L 1 1 | 1 1 I ] 1
1 2 3 L 5 6 T a8
LITERS CONDENSATE COLLECTED
91, .. Q0

rig. 33. Effective Relative Volatilities of Yh3 and SrF2 in the

MERE Distillation Experiment.
RELATIVE VOLATILITY WIiTH RESPECT TO LiF

85

ORNL DWG 70-4513

 

100 .

T

|

A

 

10.0

1.0

 

f=4.02 x10~® mole fraction & A

 

  

 

 

04 1 [ 1 | 1 ! 1 1 L 1 L 1 ] [ 1
0 f 2 3 4 5 6 T 8
LITERS CONDENSATE COLLECTED
Fig. 34. Effective Relative Volatility of lSTCsF in MSRE Distillation

Experiment, Showing Effect of Variation of Assumed Feed Concentration.
1k

The effective relative volatility for YPmF is based on a computed

3
feed concentration, which is believed to be a more accurate value than
the measured feed concentration (since measured concentrations of other
lanthanlide fission products were in good agreement with computed concen-

trations and since the measured concentration is in doubt). As in the

L) 7
case of 1‘¥CeF3, the relative volatility of lu(PJflle..S at the time of the

first sample was low-~less than 7.8 x lomh; it rose sharply to about

- 3l
3.4 x 10 3 for the other samples. Again, as in the case of lHCeF the

3:
relative volatility of the fifth sample was low. There were no previously

measured values of the relative volatility of Pmi', with which to compare

3
these results. It is interesting to note that when the effective rela-

: o 14
tive volatility of 7PmF was calculated based on the measured concen-

LT 3

tration of Pm in the fuel storage tank (instead of a calculated con-

centration), the relative volatility value for each sample except the

1hh

first nearly coincided with the resgpective value for Cel',.
3
The variation of the relative volatility of 155EuF during the run

1k 2

closely paralleled the variations for CeF3 and L 7PmF3. The value

for the first sample was low (less than 1.5 x lOmb), but the values
for all other samples except for the fifth sample were a great deal
higher (about 2.2 x lO“h). The value for the fifth sample was lower by

a factor of 2.6. These calculated effective relative volatilities (which

l55Eu)

were based on a computed feed concentration of were lower than

the value of 1.1 x lOM3 measured in recirculating equilibrium stills.

However, the values for lDSEu in the condensate are suspected of being

inaccurate since difficulty was experienced in the analyses; therefore,

11 155

a Eu analyses for the condensate samples were reported as approxi-

mate values (see Table 5). On the other hand, it is significant that

l55Eu during the run closely

147

W
paralleled the relative volatilities of lHCeF3 and 'PuF, (snd, as
91 90

seen later, YFB and Srb2).

Figure 33 shows the effective relative volatilities of

91

the variation in the relative volatility of

91YF,3 and.

90

Ser. The effective relative volatility of YF3 had an average value

-2
during the run of 1.4 x 10 ~, about 410 times the value measured in
91YF relative

3

recirculating eguilibrium stills. The variation of the
87

volatility during the run resembled variations in the relative volatili-
ties of the lanthanides; the low value for the fifth sample was the most

noticeable similarity.

QOSrFQ (based

on the measured concentration in the feed) had an average value of 4.1

During the run, the effective relative volatility of

X 10*3, which is about 84 times the value measured in recirculating

equilibrium stills. Although not shown 1in Fig. 33, the average value

89

of the relative volatility of SrF2 (based on a computed concentration

in the feed) was 0.193, about 3900 times the value measured in equilib-

89 90

rium stills; and the ratio of the Sr activity to the Sr activity in

the condensate samples, which should have been about 0.002 for each

sample, varied from 0.22 to greater than 10.

137

The concentration of Cs in the feed salt was not measured; instead,

calceulation of the effective relative volatilily was based on an esti-

137

mated feed concentration. Since
'(137

Cs has a fairly long~lived gaseous

precursor Xe, half-life of 3.9 min), 1ts actual concentration in the

salt will be lower than that calculated if we assume that all the pre-
cursors remained in the salt. Alsoc, because the actual relstive vola-
tility of CsF is fairly high, the results of the calculaticns of the

effective relative volatility are sensitive to the assumed feed concen-

137

tration of Cs. Figure 34 shows calculated relative volatilities of

137

CsF for two assumed feed concentrations. The points around the lower
curve result from the feed concentretion that would arise as s conse-
quence of all precursors staylng in the salt during MSRE operation and
represent lower limits for the relative volatility; the points around

the upper line result from a feed concentration just high enough to

137

prevent the computed concentration of Cs in the still pot liguid from

falling to zero, and represent upper limits for the effective relative

137 137

volatility of Cs¥. The highest effective relative volatility of Csl

seen in these calculations was only about 20% of the value measured in

LiF—BeFP mixtures by Smith and co-workers.

As seen from the previous results, all components except BeF2 and

Zth had effective relative volatilities that differed significantly
88

(in some cases, drastically) from values predicted from work with equi-
librium systems. Possible explanations for these discrepancieg are

being examined.

13. RECOVERY OF TLi FROM TLiwaISMUTHm—RAREmEARTH
SOLUTIONS BY DISTILLATION

J. R. Hightower, Jr. L. E. McNeese

Removal of divalent lanthanides from the LiCl stream in the metal-
transfer process produces a bismuth stream that contains lanthanides,
along with a high concentration of TLi (on the order of 10 to 50 at.

%). Because of the high cost of TLi, its recovery may be economical.

The low vapor pressures of the lanthanide metals, as compared with
those of Li and Bi, suggest that distillation might be used effectively
to recover Li and Bi. However, since Li and Bi form an intermetallic
compound (Li3Bi) having & high melting point (1145°C), the Li concen-
tration in the still could become high enough in the course of dis-
tillation that solid LigBi would be formed in the still pot. Calcu-
lated results that indicate the expected performance of a one-stage
continuous equilibrium still sre discussed in this section. The
fractional recovery of lithium is calculated for a set of conditions
that allow a reasonably high recovery. The extent of separation of

the recovered lithium from the lanthanides was assumed {0 be adequate

and was not considered.
13.1 Mathematical Analysis

Material Balance Around a Continuous Still. — Consider a single-

 

stage equilibrium still to which a bismuth stream containing lithium

is fed (the lanthanide fission products are assumed to be nonvolatile
and will be neglected) and from which an overhead stream and s bottoms
stream are drawn (each stream contains bismuth and lithium). A steady-

state material balance around the still yields the following equstion:
89

7 = (1 ~ f)X + fY , (4h)
where
2 = atom Traction of Li in the feed strean,
¥ = atom fraction of Li in the still bottoms stream,
Y = gtom fraction of Li in the overhead stream from the still,
f = fraction of the feed stream vaporized in the still, moles

vapor per mole of liquid fed.

The vapor phase and the liquid phase leaving the still are assumed to be
in equilibrium; the concentrations of these phases are related through
the relative volatility of lithium with respect to bismuth, wvhich is

defined as:

_Y(1 —X) _
“Li-Bi © X(1 = Y) ° (45)
where o is a function of system temperature, pressure, and compo-

Li-Bi

sition. (The evaluation of o will be discussed in a later section.)

Li-Bi
Substitution of Egq. (45) into Eg. (44) yields the following eguation,
which relates the still-pot composition to the fraction of material dis-

tilled:

7 - [(1 —a )2 £ (1 — ) + o Flx + (1 —a Y1 — £)%° = 0 . (u6)

Li~Bi Li-Bi Li-Bi

Because A B depends on X, this equation must be solved iteratively.

The relative volatility of lithium with respect to bismuth is related

to the vapor pressures of the two components through the relation

S P
_ Li Li (LLT)

here “Li-Bi - vpPpy
Y14 and Vg 8re the liquid-phase activity coefficients of Li and
Bi, respectively, at the temperature, pressure, and
composition of the liguid in the still;
PLi and PBi are the vapor pressures of Li and Bi, respectively,
at the temperature of the still.
90

Evaluation of the activity coefficients for use in Eq. (I7) is

discussed in the following section.

Evalnation of Activity Coefficients for the Li~Bi System. — The

 

15

activity coefficient for lithium in Li-Bi solutions has been measured’

and is expressed by the following equation:

 

 

vy = [Ag) " B}(&?)] *o (48)

wnere

f

9397 + 18.16T "'O.OlO9T2 cal/mole,

7103 — 19.4h7 + 0.0068T2 cal/mole,

gas constant = 1.987 cal mole—l OKml,

|

Ho e
il

H

absolute temperature, °K.
This equation is valid for the composition range 0.05 < X < 0.60, in the
temperature range T75°K < T < 1100°K and at a pressure of 1 atm.

The activity coefficient of bismuth is obtained from Eq. (48) through

the Gipbs-Duhem equation for a binary system at constant temperature,

 

in the following form:l6
_a In g, X 3 In vy, . (v — XV, .~ (1 MX)VB],_] ap o)
39X T 1 X X RT(1 ~ X) ax °
where
v = volume of one mole of Li~Bi alloy,
fi}i = molar volume of pure Li,
V%i = molar volume of pure Bi,
P = total pressure of system.

When Eq. (49) is integrated with respect to X from X = 0 to the desired

value, the following equation is obtained:
91

 

 

X

3 1ln Y1,
W yp, = | —3x

0
X | Pat X =X
- L 8 1ln YLidX+ [v—-XvLi-(l-X)VBi]dP (50)
1 — X 3% RT (1 — X) A
0 Pat X =20
Y

The last term will be neglected since its value is less than 6 x 10 ,
which is negligible as compared with the value of the next-to-last term
(i.e., > 0.1). Thus, the following relation, valid for low pressures,

will be used to evaluate YBi:

3 1In y..

N Li .o
In Ygi = T =%) Y ax. . (51)

 

0 1ln Y14

3 X
this quantity into Eg. (51) and integrating the resulting expression

The quantity can be evaluated from Eq. (L48); substituting

results in:

_ B(T)
n yp; = g7

[X + 1n (1 —X)] . (52)
Equations (52) and (48) can be combined with Eq. (47) to give the following
relation for the relative volatility of lithium, with respect to bismuth:

B(T)

_ _Li _A(T)] _ 2B(T) _ RT
i pi = P exp [ J exp [ = ,X] (1 ~ X) . (53)

 

 

The vapor pressure of lithium is given by the following equation, which

was derived from data given in ref. 17:
92

. _ 16746
in PLi 16.813 T , {5u)
where
PLi = vapor pressure of Li, mm Hg,
T = absolute temperature, °K.

The equation for the vapor pressure of bismuth used in these calculations

was obtained from ref. 18 and is as follows:

. 9150.3 y
log,y Pp; = [.6327 =, (55)
where
PBi = vapor pressure of Bi, mm Hg,
T = gbsolute temperature, °K.

Method for Solving Equations. - The method for solving the equation

 

for the still-pot composition (assumed to be the same as the still bottom

stream) consisted of the following steps:

1. A still-pot temperature was chosen; this allowed the
evaluation of the following terms in Eq. (53): P

“Li,
Poyo A(T)/RT, B(T)/RT.

2. The above terms were substituted into Eg. (53) to

give an expression for o which was dependent

Li-Bi
only on X.
3. The expression for 0. gy W8S substituted into Eq. (L46).
L. Equation (46) was solved using the Newton-Raphson algo-
rithml9 for a range of values of f and Z.

5. Once the still-pot composition was determined, the
liquidus temperature for this composition could be
determined from the Li-Bi phase diagram.go The
fractional recovery of lithium in the overhead stream

was calculated from the following equation:

R=1-(1—0)3, (56)
93

where R is the number of moles of lithium in the
overhead stream per mole of lithium in the feed

strean.
13.2 Calculated Results

Still-pot compositions, liguid temperatures, and fractional lithium
recoveries were calculated for still-pot temperatures of 800°C and 900°C,
feed compositions of 5, 10, and 20 at. % lithium, and values of f from
0.70 to 0.992. Figure 35 shows the liquidus temperature of the still-pot
liquid for an operating temperature of 800°C, the three feed composi-
tions, and vaporization of up to 95% of the incoming Bi-Li stream. For
a feed composition of 10 or 20 at. % lithium, the maximum lithium recovery
would be less than 36%. For a feed composition of 5 at. % lithium, a
lithium recovery of greater than 50% could be achieved before solid LiBBi
would be formed in the still pot.

Figure 36 shows the liquidus temperature of the still-pot liquid
for an operating temperature of 900°C, the three feed compositions, and
vaporization of up to 95% of the incoming Bi~Li stream. For a given
fraction vaporized and a given feed composition, distillation at 900°C
preduces a still-pot composition with a slightly lower liquidus tempera-
ture. The lower liguidus temperature, combined with the higher still-
pot temperature, allows higher recovery of lithium than distillation at
800°C does. With a feed composition of 10 at. % lithium, a maximum
lithium recovery of 66% can be obtained. With a feed composition of S
at. % lithium, very high lithium recoveries can be achieved, as shown
in Fig. 37. About 90% of the lithium could be recovered by vaporizing
99% of the incoming Bi-Li stream; the liguidus temperature of the still-

pot material in this case would be only about 630°C.

The following difficulties with recovery of TLi from bismuth by
distillation became apparent as a result of these calculations: (1) Tt
will be difficult to find materials of construction that can withstand
the high températures necessary to achieve high lithium recoveries.

(2) The vapor pressures of the Bi-Li mixtures are low even at tempera-

tures around 900°C (for example, for recovery of 90% of the lithium, a
Pot Liquidus,°C

Still

ORNL-DWG-70-6249

 

$00

BOO

700

600}—

500+

 

 

400 :_—__/——-—-—-" /

300k

 

 

f. Fraction Distilled

Fig. 39. Licguidus Temperature of Still-~Pot Materizl in Conitiruous
: . - o 4 . o o sy .
Bi-Li St1i11 Operating at 800°C. Effects of feed composition and fraction
of feed vaporized.

095
Pot Liquidus, °C

Still

ORNL DWG 70-6248

 

1000

900

800

700

600

300

400

 

 

300

o

 

] | i

 

075 0.8 0.85 09 095
¥, Fraction Distiled

Fig. 36. Liguidus Temperature of Still-Pot Material in Continucus
Bi-Li Still Operating at 90C°C. Effects of feed composition and fraction

of feed vaporized.

G6
Stitl Pot Material Liquidus,®C

ORNL DWG 70-5247

 

700

 

600 [—
500 -
400 |

300 —

 

i i i ] { | 1 |

 

 

200
020

0.9 092 093 094 095 096 0.97 098 099
¢, Fraction Distitied

Fig. 37. Liguidus Temperature of Still-Pot Material and Fractional
Li Recovery for Large Fractions of Feed Vaporized. Operating temperature
of stiil is 900°C, and feed composition is 5 at. % Li.

100

20

BO

-~
Q

@
o

%o 7Li Recovered

50

30

96
o1

vapor pressure of only 0.2 to 0.3 mm Hg was calculated for the still-
pot material); the low vapor pressures would result in low vaporizaticn
fluxes. Consequently, large vaporization areas would be needed. (3)
Because the lithium concentration in the feed must be low to prevent the
formation of Li3Bi in the still pot, additional processing complexities
must be added to dilute the Bi-~Li stream coming from the extraction
column. Although these difficulties are not insurmountable, they pro-
vide incentive to search Tor alternative methods for recovering the TLi-.

14%. PREVENTION OF AXIAL DISPERSION IN PACKED COLUMNS

J. S. Watson L. E. MclNeese

The use of packed column contactors in MSER fuel processing systems
is being considered. As shown previously,gl.axial dispersion can
significantly reduce the performance of this type of contactor under
operating conditions of Interest. As part of our contactor development
program, we are evaluating modifications to packed columns that will
reduce the effect of axial dispersion to an acceptable level. The pro-
posed modifications consist of inserting devices at points along the
column to reduce dispersion across the column at these points. I the
devices are separated by a column length equivalent to one theoretical
stage (an extent of separation that can be achieved even if the fluids
between the devices are completely mixed), the stage efficiency of the
column segment will be greater than T5% if 15% or less of the salt

flowing through the segment is recycled to the previous segment.

The type of device under investigation is shown in Fig. 38. The
metal stream flows down an annular section in the upper portion of the
device into the portion that is effectively an inverted annular bubble
cap. The accumulated metal forms a seal and forces the less-dense phase
to flow upward through restrictions in g sieve plate. A significantly
higher salt velocity is obtained through the restrictions, and a con-
gsiderably diminished recycle of salt should occur across the device.

We have tested two devices of this type, using mercury and water to
98

ORNL DWG 70-4506

 

 

 

73g. 38. OSchematic Diagram of an Axial Dispersion Preventer.
99

simulate bismuth and salt. One or more dispersion preventers were
inserted near the middle of a 2-in.-diam column that was packed with

3/8~in. Raschig rings.

The experimental technique consisted of establishing a counter-
current flow of mercury and water through the column and injecting a
steady flow of tracer solution (cupric nitrate) at a point near the
top of the column. The tracer concentration was determined at several
points along the column. When the logarithm of the tracer concentration
was plotted as a function of distance along the column, a discontinuity
in tracer concentration was observed at the preventer location, as shown
in Fig. 39. The ratio of the concentrations at the discontinuity is a
measure of the fraction of the water that was recycled through the pre-

venter.

The fraction of the water recycled through a preventer appears to
depend mainly on the water flow rate per sieve opening and the diameter
of the sieve opening. No dependence on the mercury flow rate was evi-
dent, although the small number of data vpoints and the usual data scatter
make this conclusion tentative. The effect of metal flow rate is
expected to become quite large, however, for metal flow rates suffi-

ciently high to prevent complete metal coalescence in the downcomer.

The first preventer to undergo testing consisted of a 1/2-in.-deep
bubble cap and a 1/8-in.-thick sieve plate containing four 3/32-in.-diam
holes. As shown in Fig. 40, only three data points were obtalined with
this device; however, the fraction of water that is recycled is observed
to decrease with increasing water flow rate. Less than 15% of the water
was recycled for a water flow rate greater than 90 ml/min (22 ml per
minute per hole). The openings were then drilled out to a diameter of
1/4% in. As shown in Fig. 40, similar results were cbtained, although
the fraction of water recycled did not reach an acceptably low level in
any of these experiments. By extrapolating the curve, one can estimate
that the fraction of water recycled would be less than 15% at a water
flow rate between 150 and 200 ml/min (approximately 50 m1 per minute per
hole).
100

\ ORNL DWG 70-4507

 

fos b b1

1
3

o
o
!
i

i

04 -

!

 

0.3

C/CREFERENCE

 

 

 

 

0 i L | | l 1
-5 0 5 10 15 20 25

DISTANCE FROM TRACER INLET (IN.)

'ig. 39. Concentration FProfile Using a Backflow Preventer with a
Single 3/8-in. Sieve Opening. Mercury flow rate, L0O0 ml/min; water flow
rate, 115 ml/min.
101

ORNL DWG 70~-11043

 

 
    
  
  
 
     
       

 

 

 

  

 

1.0 T - 1 T T 1 T
Preventer Number of Hole Diam.
Holes {in)
A ® i 4 3/32
0 i 4 /4
2 | /4
.

8 -
0-6_' r—y
=
x
E
4
o
n
m
=
9
3
0
h &t a—

2 —

0 ] ! { 1 l

0 20 40 60 80 100 120 {40

Woter Raote, mi/min

7ig. 40. Fraction Backmixing Across Two Axial Dispersion Preventers
Having Various Orifice Diameters.
102

A second dispersion preventer, which had a 2-in.-deep bubble cap
and only a single 1/16~in.-diam opening in the sieve plate, was also
tested. No recycle could be detected at the lowest measurable water
rate (23.5 ml/min). The diameter of the opening was increased to 1/8
ia., and again no recycle was detected. When the orifice diameter was
increased to 1/h in., the results shown in Fig. 40 were obtained. The
fraction of water recycled iz less than 15% if the water flow rate is
greater than 50 ml/min. Note that this value agrees with the results
estimated by extrapolating results from the first preventer with four
1/4=in. holes. The orifice diameter was further increased to 3/8 in.
to obtain the results shown in Fig. 40. The fraction of water recycled
was increased considerably by this change in diameter; again, howsver,
the fraction of water recycled could be reduced to any desired value

at sufficiently high water flow rates.

An undesirable feature of dispersion preventers of the type dis-
cussed above is that the column capacity or flooding rate is reduced.
This is to be expected, and it is unlikely that a device which will
reduce the column capacity by less than 50% will be developed. The
devices tested have operated satisfactorily with mercury flow rates as
high as 15% of the flooding rate for a column packed with 3/8-in.

Raschig rings.

15. HYDRODYNAMICS OF PACKED COLUMN OPERATION
WITH HIGH-DENSITY FLUIDS

J. S. Watson L. . McNeese

The hydrodynamics cf packed column operation with fluids having high
densities and a large density difference is being studied in order to
evaluate and design countercurrent contactors for use in MSBR processing
systems based on reductive extraction. Mercury and water are being used
to simulate bismuth and molten salt in these studies. FEarlier experi-
ments,22 carried out with 1/8-in. and 1/4-in. solid cylinders and with
3/16~in. and 1/4-in. Raschig rings in a 1-in.-ID column, demonstrated

that a transition in mode of flow of the dispersed phase occurs between
103

the packing sizes of 3/16 in. and 1/4 in. The transition appeared to

be a function of the packing size only and was not related to packing
shape (solid cylinders or Raschig rings). With the larger racking, the
mercury was dispersed into small droplets, which produced a large inter-~
faclal area. With the smaller packing, the mercury flowed down the column

in continuous channels and resulted in a much smaller interfacial area.

A large interfacial area (and hence large packing) is desired in
order to obtalin high rates of mass transfer between the salt and metal
phases. Since only one packing diameter above the critical size had
been studied, it was considered desirable to study several larger packing
materials. This required two modifications to the experimental system:
(1) installation of a2 column larger than the l-in.-diam unit used
previously, and {2) installation of a mercury pump having a higher
pamping capacity than that used previously. A 2-in.~ID Lucite column
having a length of 24 in. was constructed, and a Lapp* CP5~3 pump was
installed. The system was used by a group of MIT Practice BSchool
student523 for measuring flooding rates, dispersed-phase holdup, and
pressure drop with 1/4~in. solid cylinders and with 3/8-in. Raschig
rings. We later obtained data with 1/2~in. Raschig rings. Results of
these studies and development of a relation for predicting packed
column performance during countercurrent flow of molten salt and bismuth

are discussed in the remainder of this section.
15.1 Ixperimental Technique and Results

The experimental technigue is the same as that used previously22
during studles with the l~in.-diam column. Pressure drop through the
packed column was determined from displacement of the mercury-water
interface at the bottom of the column. Dispersed-phase heldup was
measured by simultaneously closing two ball valves, thereby isclating
a section of the column and allowing determination of the trapped

mercury volume.

 

*Lapp Process Equipment, Interpace Corp., North St., Le Roy, Wew York
14482.
10k

Experimental data showing the variation of column pressure drop
with continuocus-phase (water) superficial velocity for several dispersed-
phase superficial velocities are shown in Figs. 41 and L2 for 1/L-in.
solid cylinders and 3/8-in. Raschig rings respectively. Data on the
variation of dispersed-phase holdup with continuous~phase superficial
velocity for several dispersed-phase velocities are shown in Figs. W3-
45 for three packing materials studied. Flooding data are shown in
Figs. 46 and 47 for 1/b-in. solid cylinders and 3/8-in. Raschig rings

respectively.

15.2 Pressure Drop Across the Column

Surprisingly, the pressure drop across the column was observed to
decrease with increases in water flow rate at high mercury flow rates
and low water flow rates. At high water flow rates, the pressure drop
increased with increases in water rate as expected. Thig effect was
most apparent with 1/2-in. Raschig rings, although the same behavior
is also observed to some extent with 3/8-in. Raschig rings. For
example, with 1/2-in. Raschig rings and a superficial mercury velocity
of 380 ft/hr, the column pressure drop decreased from 1.1 psi to 0.27
psi as the water velocity was increased from 0 to 12 ft/hr. This
phenomenon was checked carefully and was found to be reprcducible,

The present data for pressure drop with 1/2-in. Raschig rings are con-
fusing and difficult to represent. It is interesting to note (Fig. 45)
that there was no significant decrease in dispersed-phase holdup when

water flow was initiated.

On the average, the mercury is not being accelerated while flowing
through the column. Thus, the weight of the mercury must be supported
by interaction with the continuous phase and with the packing. If
interaction with the packing were negligible, the pressure drop in
the continuous phase would be equal to the sum of the force on the
column wall and the quantity {(Ap)X, where Ap is the difference in the
densities of the two fluids and X ig the fraction of the columnn void

volume that 1s occupied by the dispersed phase. The observed pressure
105

ORNL DWG 70-1{0O53 Rl

‘III[IIl 1 |

 

  

20

 

DISPERSED-PHASE
SUPERFICIAL VELOCITY

Column Pressure Drop [cm (Hg~-H20)]

 
  
    
 

 

 

 

 

}INDICATES FLOODING (fE’TL

— A

+ -3

102 —

139 —

0.1 L. ] 1 | | , |
175 20 30 50 100 200 300 500 1000

Continuous- Phase Superficial Velocity (ft/hr)

Fig. 41. Variation of Column Pressure Drop with Superficial Water
Velocity and Dispersed-Phase Superficial Velocity for 1L/b-in. Cylindrical
Packing in a 2-in.~diam Column.
106

ORNL. DWG 70-2830

100 TlTllT]ll’r1[ T I T 1Tlll'|'[

 

o
O
!

g
o
!

PRESSURE DROP, c¢cm of Hg MINUS H,0

 

 

 

 

10+
0.5+
i +
A
— O
102 O
i {39 v
172 O
Oi J 1 A Jod 1) i el l 1 i 1 L ] b 1|
20 100 600

SUPERFICIAL WATER VELOCITY, V¢ (ft/hr)

Fig. 42. Variation of Column Pressure Drop with Superficial Water
Velocity and Dispersed-Phase Superficial Velocity for 3/8-in. Raschig
Ring Packing in a 2-in.~diam Column.
107

o.8 ORNL DWG 70-11050
' l I | l | |
DISPERSED - PHASE SUPERFICIAL VELOCITY

 

®-36ft/hr
O~ 68 ft/hr
A -i02 tt/hy
A-139 ft/hr

Ve = 315.4 ft/hr

Dispersed —Phase Holdup

 

 

 

. 1 | 1 1 1 1
0 20 40 60 80 100 120 140

Continuous~Phase Superficial Velocity (ft/hr)

 

Fig. 43. Variation of Dispersed-Phase Holdup with Continucus- and
Dispersed-Phase Superficial Velocities for 1/4-in. Solid Cylindrical
Packxing in a 2-in.-diam Column.
108

ORNL DWG 70-~-2005R!t
04 p . g e R Z

 

SUPERFICIAL SLIP VELOCITY: 819 fi/hr |
PACKING : ¥g—in RASCHIG RINGS DISPERSED PHASE
. SUPERFICIAL VELOCITY
®— 36 ft/hr
O -G8 ft/hr
A—102 ft/hr
A—439 ft/hr

B—172 ®tinr
O3 e e s fi,

 

 

s
H//f’/’/’ A
A

DISPERSED PHASE HOLOUP

 

 

 

 

100 200 300
CONTINUCUS PHASE SUPERFICIAL VELQCITY [ tt/ )

O

Yig. 4k, Variation of Dispersed-Phase Holdup with Centinuocus- and
Dispersed-Phase Superficial Velocities for 3/8-in. Raschig Rings in & 2~
in.-dlam Column.
109

ORNL DWG 70-1105]
T l I

DISPERSED ~ PHASE SUPERFICIAL VELOCITY
® - 93 tt/hr
O -150 ft/hr
W -i77 tt/br
0O ~246 ft/hr
0.4+ A ~-300 ft/nr
A ~390 ft/hr

 

Ve * 1530 ft/hr A

o
("
]

Dispersed - Phase Holdup

 

 

 

o | | 1 |
Q 100 200 300

Continuous- Phase Superficial Velocity (ft/hr)

Fig. k5. Variation of Dispersed-Phase Holdup with Continuous-
Dispersed~Fhase Superficial Velocities for 1/2-in. Raschig Rings in
2-in.-diam Column.,

and
172

Square Root Of Dispersed—Phase Superficiol Velocity(fi/hr)

110

ORNL DWG 70-11064

 

20 T T T T T T 77T 77T 7

 

l b | 1 1 |

 

 

0
0 2 4 6 8 1o 12 4 &6 I8 20 22 24 26

Square Root Of Continuous - Phase Superficial Velocity (H’/hr)'/2

Fig. 46. Predicted and Measured Flooding Rates with 1/4-in. Solid
Cylindrical Packing.

28
111

ORNL DWG 70-9004

 

28

24

      

20 2

# in. RASCHIG RINGS

 

 

 

-'-|N
< 16
<
"‘"N -
o
T
8_._..
4 p—
0 | | I |
0 4 8 12 16 20 24 28

1 |
Vo2 (ft/hr)2

Tig. 7. Predicted and Measured Flooding Rates in a Packed Column
During Countercurrent Flow of Mercury and Water.
112

drop is less than the quantity (Ap)X, which indicates that a substantial
fraction of the mercury is supported by the packing. A decrease in
pressure drop without a corresponding decrease in dispersed-phase holdup
indicates that there is an increase in the interaction between the mercury

and the packing.

One can qualitatively visualize the mercury as flowing through all
of the channels between the packing when there is no water flow. When
a water flow is started, it is possible that the mercury will flow
preferentially through only part of the channels. In such a case,
higher mercury velocities would be obtained and increased interaction
of the dispersed phase with the packing would occur. A guantitative

treatment of this phenomenon has not been devaloped.
15.3 Dispersed-Phase Holdup

Several methods, including an expression ussd by Pratt2h and a
relation developed earlier by the authors,22 have been suggested for
correlating holdup data. The present datsa indicate that holdup can
be correlated in terms of a constant slip velocity, Vs’ which is

defined as:

<
<

c ., .4
5 1 —X X 2

 

where VS = guperfiecial slip velocity,
VC = continuous~-phase superficial veloccity,
Vd = dispersed-phase superficial velocity,
X = dispersed-phase holdup.

The curves drawn through the holdup data in Figs. 43, 4b4, and u5
correspond to a comnstant slip velocity for each packing material. For
a given packing, the slip velocity was obtained by averaging slip
velocity values calculated for each data point from Eq. (57). The
standard deviations of the average slip veloecity thus calculated were

10 to 15%, which is estimated to be the experimental error in the
113
.

holdup data. There are significant deviations between the calculated
and measured holdup data for holdup values below 10%. Otherwise, the
data for 1/U-in. and 3/8-in. packing show no trend away from the pre-
dicted holdup values. The predicted holdup values for 1/2-in. Raschig
rings are higher than the measured values at low mercury superficilal
velocities. This trend is not considered to be significant since 1t

ig not confirmed by data from other packing materials.

Average values were also calculated for the characteristic veloecity,

VD, suggested by JP}_"attglL and defined as

Vc Vd
o~ (1 _x)?2 KI=X)° (58)

 

The standard deviation resulting from assumption of a constant charac-
teristic velocity was larger than that resulting from assumption of a
constant slip veloclity, and a dependence of VO on holdup (other than

that shown above) was noted.

The holdup data for 1/4—in. Raschig rings and 1/b-in. solid cylinders

in 1- and 2-in.-diam columns suggest that the superficial slip velocity

is proportional to the packing void fraction, as shown in Fig. 48. The
superficial slip velocity is also expected to be dependent on packing
size. The slip velocity for 3/8~in. Raschig rings (819 ft/hr) is not
significantly higher than that for 1/4-in. packing when a correction

iz made for the difference in packing void fraction. However, the slip
velocity for 1/2-in. Raschig rings (1530 ft/hr) is considerably higher

than that for the smaller packing sizes.
15.4 Correlation of Flooding Rates

The observation that the dispersed-phase holdup data can be corre-
lated on the basis of a constant superficial slip velocity is especially
important since this suggests a method for correlating flooding data.

At flooding, it is assumed that the following conditions exist:
Superficial Siip Velocity {(ft/nr)

11k

ORNL DWG 70-11057

 

 

 

 

l | | I l 1
700 [—
600 — B
500 -
400 —
® SOLID
300 — CYLINDERS ]
200 S
100 (— ~
o L l 1 I l 1
0 Ol 0.2 0.3 0.4 0.5 0.6 0.7

Packing Void Fraction

7ig. 48. Variation of Superficial Slip Veloecity with Packing Veid
ion for 1/4-iu. Packing in l-in. and 2-in.-diam Columns.
-3 =0 - (59)

Partial differentiation of Eq. (57) with respect to X, the dispersed~
phase holdup, yields the relation

Vc 1 BVC

d
4 + ——

<3

3V
-3--9—= 0. (60)

 

i

If the conditions stated by Eg. (59) are met at flooding, Eq. (60) reduces
to

Vc,flood : Vd,flood (61)

(1 -—X)2 X2

Multiplying Eq. (57) by 1/(1 — X) and combining the resulting equation
with Eq. (61) yields the following relation:

Vdgflood + Vdlflood - Vs (62)
52 (1 ~%) 1-%x° '

~ ’v
¥ = dzglOOd ) (63)

S

which reduces to

Multiplying Eq. (57) by 1/X and combining the resulting equation with
Eg. (A1) yields the relation

Vc,flood Vc,flood - Eg_ (64)
(1 —x)X (1 —x)2 X

 

which reduces to
116

th ;
1 - X = czélood . (65)

S

Elimination of X from FEqs. (63) and (65) yields the final relation

12, 1/2 _ ., 1/2
(Vc,flood) - d,flood) = Vg > (66)
where Vc,flood = contlinuous~phase superficial velocity at flooding,
Vd,flood = dispersed-phase superficial velocity at flooding.

Thus, if the superficial slip velocity remains constant up to flocoding,

1/2 1/2
c,flood) Vs (Vd,flood)

having a slope of =l1. BSimilar flooding curves with slopes near —1 have

25

a plot of (V should produce a straight line

been observed in the literature, and our earlier data for 1/4~ and
3/8-in. Raschig rings and 1/Lk-in. solid cylinders were correlated in
this manner. Recent results obtained with 1/k-in. solid cylinders and
3/8~1in. Raschig rings, shown in Figs. 46 and b7, are also represented

well by this correlaticn.

15.5 Prediction of Molten~Balt--Bismuth Flooding Rates
in Packed Columns

We have attempted to extend the correlations developed for the
mercury-water system in order to obtain relations that are applicable
to molten-salt—--bismuth systems. The physical properties of the fluids
were not varied in these studies. Significant differences in properties
exist between the mercury-water system and molten-salt--~bismuth systems;
potentially important differences include the viscosity of the continuous
phase, the difference in the densitles of the phases, and the interfacial
tension. The correction asscciated with differences in interfacial
tension is believed tc be small since the interfacial tension for the
mercury-vater system is approximately equal to that for a molten-salt—-—

bismath system.
117

As the dispersed phase flows down the column, it interacts with
both the packing and the continuocus phase. Interaction with the con-
tinuous phase is inertial in nature, and the continuous-phase viscosity
iz expected to have only a slight effect. If the principal interaction
were between the two flulds, the slip velocity would be proportional
to the square voolt of the difference in densities of the fluids. As
noted earlier, however, the metal phase is more nearly supported by
interaction with the packing; for this condition, it is estimated
that the slip velocity is directly proportional to the difference in

the densitles of the phases.

On the basis of these considerations, the superficial slip veloc-
ity was assumed to be proportional to both the packing void fraction
and the difference in densities of the phases. The resulting relation

for the flooding rates in a salt~bismuth system 1s, then:

 

1/2 1/2 Ap E .
v o M, My [ (61)
c,flood d,flood s,Hg—HQO ApHg~H20 € ot
Where Vv = continuous-phase superficial velocity at flooding,
c,flood
Vd,flood = dispersed~phase superficial velocity at flooding,
v = guperficial slip velocity for mercury-water system
s,Hg~H20
with the type packing to be used,
Ap = difference in densities of the phases,
Ap = difference in densities of mercury and water,
Hg—H20

e = packing void fraction,

e = void fraction of packing for which v

ref s,Hg~H20 was

determined.
2.

10.

11.

12.

118

16. REFERENCES

L. M. Ferris, ORNL, personal communication.
M. J. Bell, ORNL, personal communication.
M. S. Bautista and L. E. McNeese, "Axial Mixing in an Open Bubble

Column," Engineering Development Studies for Molten-Salt Breeder
Reactor Processing No. I, ORNL-TM-3139 (in press).

 

 

A. M. Skeikh and J. D. Dearth, Axial Mixing in an Open Bubble
Column, MIT-CEPS~X-91 (December 1969).

 

C. W. Kee and L. k. McNeese, Engineering Develovment Studies for
Molten-Salt Breeder Reactor Processing No. 3, ORNL-TM-3138, up.
39-50.

 

 

J. R. Hightower, Jr., st al., "Design and Installation of the Flow
Electrolytic Cell Facility," Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. h, ORNL-TM-3139 (in
press).

 

 

C. W. Kee and B. A. Hannaford, "Calibration of an Orifice--Head
Pot Flowmeter," Engineering Development Studies for Molten-Salt
Breeder Reactor Processing No. 4, ORNL-TM-3139 {in press).

 

J. R. Reitz and F. J. Milford, Foundations of Electromagnetic
Theory, 2d ed., p. 136, Addison-Wesley, Reading, Mass., 1967.

T. R. Johnson, F. G. Teats, and R. D. Price, An Induction Probe
for Measuring Liquid Levels in Liquid Metals, ANL-7153 (February
1966).

 

J. Roth and L. E. McNeese, "Bismith-Salt Interface Detector,”
Engineering Development Studies for Molten~Salt Breeder Reactor
Processing No. 4, ORNL-TM~3139 (in press).

 

J. H. Shaffer, MSR Program Semiann. Progr. Rept. July 31, 1964,
ORNL-3708, pp. 288-303.

J. R. Hightower, Jr., and L. E. McNeese, Engineering Development
Studies for Molten-3Salt Breeder Reactor Processing No. 3, ORNL~
TM-3138, pp. 8h-95.

 

F. J. Smith, L. M. Ferris, and C. T. Thompson, Liguid-Vapor Equilibria

 

in LiF-BeF, and LiF-BeF ~ThF) Systems, ORNL~4N15 (June 1969).
2=

J. R. Hightower, Jr., and L. E. McNeese, Measurement of the Relative
Volatilities of Fluorides of Ce, La, Pr, Nd, Sm, Fu, Ba, Sr, Y,
and Zr in Mixtures of LiF-BeF_,, ORNL~-TM-2058 (January 1968).

 
15,

16.

1T.

18.

19.

20.

21.

22.

119

M. 5. Foster, 8. E. Wood, and C. E. Crouthamel, Inorg. Chem. 3,
1428 (1964).

N. V. Ibl and B. F. Dodge, Chem. Eng. Sci. 2, 120 (1953).

Handbook of Chemistry and Physics, 4lst ed., p. 2343, Chemical
Rubber Publishing Co., Cleveland, Ohio, 1960.

A. K. Fischer, J. Chem. Phys. 45, 375 (1966).

Leon lLapidus, Digital Computation for Chemical Engineers, p. 288,
MeGraw-Hill, New York, 1962.

M. Hansen, Constitution of Binary Alloys, 24 ed., p. 316, McGraw-
Hill, New York, 1958.

J. 8. Watson and H. D. Cochran, Jr., "Effect of Axial Mixing in
Packed Column Contactors Used for MSBR Processing,” Engineering
Development Studies for Molten-Salt Breeder Reactor Processing
No. 4, ORNL-TM-3139 (in press).

J. 8. Watson and L. E. McNeese, Unit Operations Section Quarterly

 

Progress Report, July~September 1968, ORNL-4366, pp. 57-98.

A. J. Frederiksen, J. J. Protulpac, and S. C. Trindade, Hydro~
dynamics of a Mercury~Water Packed Column, MIT-CEPS-X-88 (1969).

R. Gayler and H. R. C. Pratt, "Holdup and Pressure Drop in Packed
Columns," Trans. Inst. Chem. Eng. 29, 110 (1951).

F. R. Dell and H. R. C. Pratt, Trans. Inst. Chem. Eng. 29, 89-109
(1951).
25
26
27
28-38

O o W o

o

2

@

@

2

&

39.
Lo.

7
78

79
80

81.
82.

83.
8L.

°

-

121

INTERNAL DISTRIBUTION

¢. F. Baes . L1. E. L. Nicholson

H. F. Bauman Le. J. H. Pashley (K-25)
S. E. Beall L3, A, M. Perry

M. J. Bell Lh-hs. M. W. Rosenthal

M. R. Bennett L6, A, D. Ryon

R. E. Blanco 47. W. F. Schaffer, Jr.

F. F. Blankenship 48. Dunlap Scott

G. E. Boyd L9, J. H. Shaffer

R. B. Briggs - 50. M. J. Skinner

R. E. Brooksbank 5L. F. J. Smith

K. B. Brown 52. D. D. Sood

W. L. Carter 53. Martha Stewart

H. D. Cochran, Jr. 54L. 0. K. Tallent

¥. L. Culler 55. R, E. Thoma

J. R. Distefano 56. D. B. Trauger

W. P. Batherly 5T. W. E. Unger

D. E. Ferguson 58. C. D. Watson

L. M. Ferris - 59, J. 5. Watson

J. H. Frye 60. A. M. Weinberg

W. R. Grimes ‘ 6l1. J. R. Weir

A. G. Grindell 62. M. E. Whatley

P. A. Haas 63. J. C. White

B. A. Hannaford 6bh. W, M. Woods

J. R. Hightower, Jr. 65. R. G. Wymer

C. W. Kee 66. E. L. Youngblood

R. B. Lindauer _ 67-68. Central Research Library
H. E. McCoy - 69-T70. Document Reference Section
L. E. McHeese 71-73. Laboratory Records

D. M. Moulton Th. Laboratory Records, RC
J -

P, Nichols Tfi. Y=-12 Document Reference Section
. 76, ORNL Patent Office

EXTERNAL DISTRIBUTION

J. A. Accairri, Continental 0il Co., Ponca City, Oklahoma Th601
R. M. Bushong, UCC, Carbon Products Division, 12900 Snow Road,
Parma, Ohic 4h130

D. F. Cope, Atomic Energy Commission, RDT Site Office (ORNL)

C. B. Deering, Black & Veach, P. O. Box 8405, Kansas City,
Missouri 6hkl1k

A. R. DeGrazia, USAEC, DRDT, Washington, D.C. 20545

Delonde R. deBoisblanc, Ebasco Services, Inc., 2 Rector Street,
New York, N.Y. 10006 '

D. Elias, RDT, USAEC, Washington, D.C. 20545

Norton Haberman, RDT, USAEC, Washington, D.C. 20545
85.
86.
87-88.
89.
90.
91.
92.
93.
9l ,
95.
96.

oT.
98-99.

122

'
N

£
——

EXTERNAL DISTRIBUTION (continued)

T. R. Johnson, Argonne National Laboratory, 9700 S. Cass Avenue,
Argonne, Illinois 60439

Kermit Laughon, Atomic Energy Commission, RDT Site Office (ORNL)
T. W. McIntosh, Atomic Energy Commission, Washington, D.C. 20545
E. H. Okrent, Jersey Nuclear Co., Bellevue, Washington 98004

R. D. Pierce, Argonne National Laboratory, 9700 S. Cass Avenue,
Argonne, Illinois 60439

J. Roth, Combustion FEngineering, Windsor, Connecticut 06095

M. Shaw, Atomic Energy Commission, Washington, D.C. 20545

N. Srinivasan, Head, Fuel Reprocessing Division, Bhabha Atomic
Research Center, Trombay, Bombay T4, India

C. L. Storrs, Combustion Engineering Inc., Prospect Hill Road,
Windsor, Connecticut 06095

B. L. Tarmy, Esso Research and Fngr. Co., P. 0. Box 101, Florhamn
Park, N.J. 07932

J. R. Trinko, Ebasco Services, Inc., 2 Rector Btreet, New York,
N.Y. 10006

Laboratory and University Division, ORO

Division of Technical Information Extension, ORO, AEC