ORNL-TM-3138.txt

2o
tr i g

TIN MARIET

i

RGY TEMS LIBRARIES

i

T

3 4456 0382732 1

 
T

 

 

 

This rehort 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,: rnor any of their cohtram‘.ors,
subcontractors, or their employees, makes any warranty, express or impiied, or
assumes any legal liability or responsibility for the accuracy, completensess or
usefu!neés of any information, apparatus, product or process disclosed, of
represents that its use wouldf: not infringe privately owned rights.

 

 
ORNL~TM~3138
Contract No, W-7405-eng-26

ENGINEERING DEVELOPMENT STUDIES FOR MOLTEN-SALT
BREEDER REACTOR PROCESSING NO. 3

L. E. McNeese

MAY 1971

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

WG

3 yysk 0382732 1L

 

 

 
ii

Reports previously issued in this series are as follows:

ORNL~4204
ORNL-4234
ORNL~4235
ORNL-4364
ORNL~4365
ORNL-4366
ORNL-TM~3053
ORNL-TM-3137

Period ending
Period ending
Period ending
Period ending
Period ending
Period ending
Period ending

Period ending

June 1967
September 1967
December 1967
March 1968
June 1968
September 1968
December 1968
March 1969
CONTENTS
Page
SIMARIES » - 2 & . L] . » - o - a2 o - * ® » - - . - . * . - + & » l
1. IRTRODUCTION & ¢ & v & 4 o o o s o o s s o o s 2 o 2 s o s 5
2. REMOVAL OF PROTACTINIUM FROM A SINGLE-FLUID MSBR . . . . . . 6
2.1 Steady-State Performance for the Case of an MSBR Fueled
with Uranium . . ¢ o ¢ & v 4 o o o o o & o o o o o o o 8
2.2 Steady-State Performance for the Case of an MSBR Fueled
with Plutonium .+ + & « ¢ ¢ 4 & ¢ o o 4 o o 2 « o o « « o 11
3. EFFECT OF CHEMICAL PROCESSING ON THE NUCLEAR PERFORMANCE OF AN
MSBR & v v ¢ o o o o o o o o 2 o o o s o s 4 s s e e s e . . 14
3.1 Computational Procedure . . ¢« & 4 ¢« « o s+ « o o « s « o 16
3.2 Chemical Behavior of Fission Products and Actinides During
Processing o ¢ v o o 6 o o 0 4 s 6 o 4 e s s s s e s e o« 17
3.3 Calculated ResultsS . « . 4 « ¢ o o o o o & s o o o+ « + « 18
3.4 Discussion of Results . . « & « v & o v v o o o o + « o 29
4., REDUCTIVE EXTRACTION EXPERIMENTS IN A MILD-STEEL FLOW-THROUGH
FACILITY & v v 4 o o o o 4 s o o = o o o s « s s o o « « » o« 30
4.1 Addition of Thorium, and Subsequent Transfers of Thorium-
Bismuth Solution . . . e & s 2 s 3 & & &+ ® 3 & 3 2 4 s » 31
4.2 Addition of Salt, and Hyvdrofluorination of Salt and Bis-
1115 o Y. 241
4,3 Hydrodynamic Experiments: Runs HR-1 and -2 . . . . . . 35
4.4 Hydrodynamic Run HR-3 , . . . . « « . o « « « + « « . . 38
5. DIGITAL SIMULATION OF THE FLOW CONTROL SYSTEMS FOR THE REDUC~
TIVE EXTRACTION FACILITY . |, . . ¢ & v v 4 & o o s o &« « « « 139
5.1 Salt and Metal Feed Systems To Be Studied . . . . . . . 41
5.2 Mathematical Analysis . . ¢ & v v o « o s o o o« « o & o 41
5.3 Solution of Equations and Calculated Results . . . . . . 45
6. ELECTROLYTIC CELL DEVELOPMENT . . . . + v & o v + o o o v v . 50

iid

6.1 Formation of Frozen Salt Layers in Regions of High Heat
Generation . . . . & 4 ¢« o v 4 4 4 e e s e 4 e e s« . . 50

6.2 Use of Beryllium Oxide as an Electrical Insulator . . . 54
10.

11.

iv

CONTENTS (Continued)

Page
ANALYSIS OF MASS TRANSFER IN FLECTROLYTIC CELLS . . . . . . . 58

7.1 General Mathematical Model . . & ¢« ¢ v &« « « « s =« « « 59

7.2 Assumptions and Simplifications . . . . . . « . + . . . 08

7.3 Cathode Current Efficiencies and Maximum Current Densities
with LiFwBerwBin MiXLUTES v v v v o ¢ o o &+ o o« « « o« 09

AXTAL MIXING IN PACKED COLUMNS WITH HIGH-DENSITY FLUIDS . . . 72

8.1 Mathematical Model . . . . & &« ¢ ¢ v ¢ ¢ &« « o o« o o » 13
8.2 Experimental Equipment . . : + « o v ¢« v o s o o « o o 14
8.3 Results . . « « v v« v ¢ o 5 o v e e e e e e e e .78

REMOVAL OF HEAT FROM PACEED-COLUMN CONTACTORS USIED FOR ISC “U NG
PROTACTINIUM . . . ¢ v« 4o v v v v &« & o s & & o s« o o« . 18

9.1 Maximum Column Temperatures, Assuming Only Convection of
Heat s 4 . e * 2 2 3 & o . . . . . " - - - . - . . . . 81

9.2 Maximum Column Temperatures, Assuming Only Heat Removal
Througll t—he CO]-ulnn wall L] - <+ s » * - » . . . . . . . . 82

9.3 Conclusion . & « 4 « ¢« & « o + s+ & s &+ & o 4 8 e s 2+ 83
MSRE DISTILLATION EXPERIMENT . . . . & ¢ v ¢ 4 o« & o « s « o« 84

10.1 Summary of Distillation Operation . . . . . . + « « . . 84
10.2 Summary of Available Experimental Data . . . . . « . . 85
10.3 Material Balance Calculations . + + ¢ v « &« o« + « « . « 86

10.ll' Results . . . . & - * > . - * - a * * s L] - & a . - + . 92

REFERENCES . . & & ¢ ¢t s o v 4« o o 4« s o o & s o s o o « « + 9p
SUMMARIES

REMOVAL OF PROTACTINIUM FROM A SINGLE-FLUID MSBR

System performance for the proposed flowsheet for isolating prot-
actinium from a single-fluid MSBR has been recalculated using current
data for reduction potentials and the solubility of thorium in bismath.
The results indicate that protactinium can be isolated satisfactorily

from reactors fueled with uranium or plutonium.

EFFECT OF CHEMICAL PROCESSING ON THE
NUCLEAR PERFORMANCE CF AN MSBR

A series of calculations hag been performed to investigate the
effects of individual fission product elements on the neutron poilsoning
of a molten-salt breeder reactor. The most important elements are Nd,

Sm, and Pm.

REDUCTIVE EXTRACTION EXPERIMENTS IN A MILD-STEEL
FLOW-THROUGH FACILITY

A bvismuth solution containing about 10m4 mole fraction thorium was
fed through the reductive extraction facility in order to complete the
removal of oxides from surfaces of the system. Analyses of samples of
bismuth taken from various vessels showed only small changes in the
thorium concentration. This indicates that the earlier hydrogen treat-

ment of the system had been quite effective in removing oxides.

Three experiments were performed in which bismuth and molten salt
(72-16~12 mole 7% LiF—BerwThFa) were fed to the packed extraction column.
In the first two runs,; the pressure drop across the column and associated
transfer lines was greater than expected due to iron deposits in the
bismuth exit line from the column. After the tubing had been replaced, a

third run was made at salt and bismuth flow rates of about 75 ml/min.
The flow rates were not steady, apparently due to bismuth that was
trapped in the salt overflow loop at the top of the column. The loop

was modified to correct this condition.

DIGITAL SIMULATION OF THE FLOW CONTROL SYSTEMS
FOR THE REDUCTIVE EXTRACTION FACTLITY

A digital simulation of the bismuth and salt flow control systems
for the reductive extraction facility was carried out to determine con~
troller constants that would result in satisfactory operation. We found
that: (1) the system is stable, (2) proportional control is acceptable,
and (3) the system can be brought to a steady-state condition in an

acceptably short time.

ELECTROLYTIC CELL DEVELOPMENT

The proposed reductive extraction processes for isolating prot-
actinium and removing rare earths are based on the use of electrolytic
cells for reducing lithium and thorium fluorides into a bismuth cathode.
These cells will require an electrically insulating material that can
withstand the corrosive conditions present at the cell anode. Frozen
layers of salt are being considered for this application; experiments
have shown that such salt layers can be maintained in a region of high
heat generation if sufficient cooling is provided. Beryllium oxide may
also be useful in cell construction since it is a good electrical in-
sulator, has a high thermal conductivity, and is relatively insoluble
in molten fluoride salts of interest., We are presently assembling the
equipment for an experiment in which an anode fabricated from Be0O will
be used. The portion of the BeO expected to contact molien salt will

normally be covered by a layer of frozen salt.
ANALYSTS OF MASS TRANSFER IN ELECTROLYTIC CELLS

A study of factors affecting mass transfer in electrolytic cells
has been initiated to identify important cell parameters and to aid in

understanding and interpreting experimental data.

A general mathematical model describing important changes in po-
tential and concentrations throughout a cell was developed. This model was
applied to the cathode region of an electrolytic cell containing an LiF-
BerwThF4 mixture of the composition expected in the electrolytic cell
for the rare-earth removal system. Calculated results (which are only
approximate) indicate that the maximum current density obtainable may

is

be about 0.16 amp/cmz; at this current density, reduction of BeF2

likely to begin.
AXTAL MIXING IN PACKED COLUMNS WITH HIGH-DENSITY FLUIDS

Axial diffusion may be important in determining the performance of
packed-column contactors in flowsheets proposed for MSBR processing.
Since relatively few data on axial dispersion are available, we have
initiated a study of axial dispersion in packed columns using mercury
and water to simulate bismuth and salt. Preliminary data using a steady-
state technique indicated that the dispersion coefficient in a 2~in.-diam
column packed with 3/8-in. Raschig rings is about 3.6 cmz/sec. No sig-
nificant dependence of the dispersion coefficient on mercury or water

flow rate has been observed thus far.

REMOVAL OF HEAT FROM PACKED~COLUMN CONTACTORS USED
FOR IS50LATING PROTACTINIUM

Egtimates have been made of heat transfer and temperature distribution
within packed columns used for isolating protactinium under various oper-

ating conditions. Tt was concluded that the maximum temperature difference
between the center line of the column and the column wall will be less
than 150°C if flow of either the salt or the bismuth can be maintained.
This value is acceptably low. The temperature difference between the

center line of the column and the wall is expected to be substantially

lower than 150°C during periods of normal operation.
MSRE DISTILLATION EXPERIMENT

The experiment to demonstrate high-temperature, low-pressure dis-
tillation of irradiated MSRE fuel carrier salt was successfully completed
in an uneventful 31-hr operation. Approximately 12 liters of salt was

distilled, and 11 condensate samples were collected.

Preliminary results indicate that the behavior of the major com-

ponents (LiF, Ber, and ZrFA) was as expected. The relative volatility

£ 144

0 CeF, was higher by about two orders of magnitude than the expected

3
value. An explanation of this discrepancy is not available; however, as

other data become available, it may be resolved.
1. INTRODUCTION

A molten-salt breeder reactor (MSBR) will be fueled with a molten
fluoride mixture that will circulate through the blanket and core regions
of the reactor and through the primary heat exchanger. We are developing
processing methods for use in a close-coupled facility for removing fis-
sion 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 section describe (1) calculated results
showing the steady-state performance of a protactinium isolation system
for reactors fueled with uranium or plutonium; (2) material-balance cal~
culations showing the effect of fission product removal times on reactor
performance for important fission products:; (3) experiments on reductive
extraction in a mild-steel flow-through facility; (4) a simulation of the
flow control system for the semicontinuous refiuctive extraction system;
(5) experiments related to the development of electrolytic cells for use
with molten salt and bismuth;‘(fi) an analysis of factors limiting the
rate of transfer of materials in electrolytic cells; (7) measurement of
axial dispersion in packed columns in which immiscible fluids having
large density differences are in. countercurrent flow; (8) calculated heat
generation rates and temperatures in an extraction column in which prot-
actinium is being extracted from molten salt; and (9) operation of equip~
ment at the Molten Salt Reactor Experiment (MSRE) for demonstration of
low-pressure distillation of molten salt using irradiated MSRE fuel
carrier salt. This work was carried out in the Chemical Technology

Division during the period April through June 1969.
2. REMOVAL OF PROTACTINIUM FROM A SINGLE-FLUID MSBR

L. E. McNeese M. E. Whatley

System performance for the proposed flowsheet for isolating prot-
actinium from a single-fluid MSBR (Fig. 1) has been recalculated using
current data for reduction potentials and the solubility of thorium in
bismuth. Calculations were made for an MSBR fueled with uranium, as
well as for the initial operation of an MSBR fueled with plutonium.
According to the flowsheet, fuel salt from the reactor enters the
bottom of the extraction column and flows countercurrent to a stream of
bismuth containing reduced metals. Ideally, the metal stream entering
the top of the column contains sufficient Th and Li to extract only the
U and Pu entering the system. The system exploits the fact that Pa is
less noble than U and Pu but more noble than Th. Both U and Pu are
preferentially extracted in the lower part of the column, while Pa
refluxes in the center. High protactinium concentrations are produced
in the salt and metal streams. Most of the protactinium in the system
can be isolated by diverting the salt stream through a tank of sufficient

size (i.e., with a volume of about 200 ft3).

In making the calculations, the following values were assumed: fuel

salt composition, 71.7~16-12-0.3 mole 7% LiFmBeFZ—ThF4"UF4;

1461 ft3; processing rate, 2.5 gpm (three-day cycle); operating tem-

reactor volume,

perature, 600°C; reactor power, 1000 Mw(electrical); and Pa decay tank
volume, 200 ft3. The extraction cdlumn consisted of three theoretical
stages above the decay tank and four stages below the decay tank. The
Th and Li concentrations in the Bi stream being fed to the column were
1.6 x lOm3 and 1.4 x 10_4 mole fraction respectively. In the calcula-

rions for an MSBR fueled with plutonium, the salt composition was assumed

to be 71.8~16-~12~0.2 mole 7% LiF-ReF —ThF4~PuF s

2 33 other values were the

same as those given above.
ORNL DWG 68-9438

 

  

 

 

 

 

 

UF, IN SALT
ELECTROLYTIC
OXIDIZER -
REDUCER
COLUMN
SINGLE
FLUID
REACTOR
COLUMN
UF, AND PaF, IN SALT U IN Bi

Fig. 1. Scheme for Isclating Protactinium in a Single-Fluid MSBR.
2.1 Steady-State Performance for the Case of an MSBR
Fueled with Uranium

Typical calculated concentration profiles in the extraction column
are shown in Fig. 2. The uranium concentration in the salt increases
from the inlet value of 0.003 mole fraction to approximately 0.004 mole
fraction in the first stage because of the reduction of U(1V) to U(I1II),
It then decreascs steadily to negligible values at the salt outlet. The
concentration of protactinium in the salt increases from the inlet value
of 1.39 x 10"5 mole fraction to a maximum of 0.0021 mole fraction, then
decreases to negligible values near the salt outlet. The cencentration
of Th in the Bi stream decreases from about 0.00132 wmeole fraction in the
upper part of the column to 7.9 x 10~8 mole fraction near thce 3i outlet.
The concentration of Li in the Bi decreases from about 0.00124 mole frac-
tion in the upper part of the column to about 0.00011 mole fraction at the

bottom of the column.

The concentration of uranium and protactinium in the salt entering
the decay tank are 1.25 x 10"5 and 1.325 x 10—3 mole fraction rcspectively.
The concentrations of uranium and protactinium in the decay tank are
2.63 x lf)m5 and 1.312 x 10_3 mole fraction respectively. Under ideal
steady~state operating conditions, approximately 93% of the protactinium
present in the reactor system would be held in the decay tank. !However,
it is likely that the actual amount of protactinium isolated from the

reactor will be somewhat below this value because of an inabilitv to

maintain optimum operating conditions.

The variation of the calculated steady-state protactinium concen-
tration in the reactor and in the decay tank with bismuth flow rate is
shown in Fig. 3. The minimum protactinium concentration in the reactor
is obtained when the bismuth flow rate is just sufficient to extract
the uranium entering the system. At slightly higher bismuth flow rates,
protactinium will also be extracted since it is the next componeat in
order of decreasing nobility. At bismuth flow rates slightly lower than
the optimum rate (about 5.3 gpm), some of the uranium will not bhe ex-

tracted; instead, it will displace protactinium from the decay tank,
ORNL-DWG 69-7640

 

 

 

 

 

 

 

 

 
  
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0
oLiF 10
Q'BQFZ
OThF4 10-—1
] 10“2
U IN SALT
.,...---—-—-T . Li AND Th
.-—-———-—*"Lw IN METAL
U IN: aonmeggel rovrrssmmremen(, 10-3
1074
| =
S
5
10-5 L
Q)
5
&
I 10~®
=
O
b Th IN 1077
i ]7 ME TAL
i....
=
S
= -8
Z 10
()
x Pa DECAY TANK
E | INLET CONC EXIT CONC 10~9
= Pa =1.325 x 1072 1.312 x 1073
* U =1.249 x 1075  2.628 x 1073
| | | | 10—40

1 2 3 4 3 6 7
STAGE NUMBER

Fig. 2. Calculated Concentration Profiles in Protactinium Isola-
tion Column for Reactor Fueled with Uranium.
Pa CONCENTRATION x 10% (mole fraction)

1400

1200 |

10C0

800

600

4C0

200

ORNL-DWG 69-7641

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T | T
| ; i CONDITIONS
5 E | REACTOR VOLUME: 1461 ft°
DECAY TANK VOLUME: 200 >
| SALT FLOW RATE: 2.5 gpm
i | |
. [ | L TOWER ABOVE TANK: 3 smeefl
| | TOWER BELOW TANK: 4 STAGES
; } +
| |
|
| | ]
| | -\ DECAY TANK
| | | |
REACTOR AND 5 \
_ DECAY TANK l REAC?R
. \ |
5.0 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9

8ISMUTH FLOW RATE (gpm)

Fig. 3. Variation of Calcuiated Steady-State Protactinium Concen-
ration in the Reactor and in the Decay Tank with Bismuth Flow Rate for
an MSBR Fueled with Uranium,

6.0

01
11

forcing the latter to flow out the top of the column if the flow rate
is not corrected. In either case, some protactinium would be allowed
to return to the reactor, and the effectiveness of the system would be

diminished.

The flowsheet has several very desirable characteristics, including
(1) a negligible holdup of fissile 233U in the isolation system, (2) an
almost immediate return of newly produced 233U to the reactor system,
and (3) a closed system that precludes loss of protactinium, 233U, or
other components of the fuel salt. Since the performance of the system
is sensitive to variations In operating conditions, attention has been
given to methods for controlling the system and for making the performance
less dependent on operating conditions. The removal of uranium from the
center of the columfi makes the system less sensitive to minor changes in
operating conditions (see Fig. 4). For example, removal of 2% of the
uranium in the salt entering the decay tank by fluorination results in
complete stabilization of the system with respect to bismuth flow rate
variations (below the optimum flow rate) as large as 1.03% of the optimum.
The uranium concentration in the salt, which is very sensitive to small’
changes in operating conditions near optimum conditions, increases by a

factor of 5000 for a decrease in bismuth flow rate of only 0.037%.

2.2 Steady~State Performance for the Case of an MSBR
Fueled with Plutonium

An MSBR may be fueled initially with plutonium, which would remain
233 233
Pa and U

inventories to build up. Since the value for the Pu-Pa separation

in the reactor system during the time required for the

factor is only about one-half that for the U-Pa separation factor, we
would expect the isoclation of Pa from a Pu-fueled system to be more
difficult than from a U~fueled system. Typical calculated concentration
profiles in the extraction column for a Pu-fueled system are shown in

Fig. 5. It was assumed that the Pa inventory in the salt was the steady-
X

Pa CONCENTRATION IN REACTOR

(mole fraction)

200

175

150

125

100

-~
w

o
O

N
w

o

ORNL-0OWG 69-7837

 

 

 

 

 

 

 

 

Pa IN REACTOR
2 % U REMOVAL

 

NO U REMOVALX
X A
: R

N T

Pa IN REACTOR!

 

 

 

 

U IN SALT

WITH NO U REMOVAL

' ENTERING DECAY TANK ~—

 

|

\

]

 

i
I
i

-~

 

\

 

 

 

 

 

 

 

E
‘ | /
VN ; ?
] \

 

 

 

5.0

Fig., 4.

5.15

5.20 5.25

5.3C

5.35 5.40 5.45

BISMUTH FLOW RATE (gpm)

Effects of the Uranium Removal and the Bismuth Flow Rate

on the Protactinium Concentration in the Reactor and on the Uranium Con-

centration in the Salt Entering the Decay Tank.

10°

107"

U IN SALT ENTERING DECAY TANK / MAX

U CONCENTRATION

¢l
13

ORNL-DWG 69-7638

 

 

 

 

0
oLiF 10
oBef. 2
R e — e 10~
e 4072
——

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0{/’,/’/b/r <
S
- S
| £
T o
Q
£
=
Q
}._,
et
'
b
<
O
—8
Z T Pa DECAY TANK 10
© INLET CONC EXIT CONC
S Pu=2.22 x107% 2.22 x1074 .
o - _ -3 ~3 — 10
Q Pa = 1.244 x 10 1.231 x 10 ;
L U =0 1.3 xA0
J J I J J 10"10
1 2 3 4 5 6 7

STAGE NUMBER

Fig. 5. Calculated Concentration Profiles in the Protactinium Tso-
lation Column for a Reactor Fueled with Plutonium.
14

state value and that a negligible quantity of U was present; although
this condition will not actually exist, the results for such a case
should indicate the relative ease or difficulty to be encountered in
isolating Pa from a system fueled with Pu. The plutonium concentration
in the salt decreases steadily from the inlet value of 0.0002 mole
fraction to negligible values at the salt outlet. The councentration of
protactinium in the salt increases from the inlet value of 2.5 x 10—5
mole fraction to a maximum of 1.244 x 10“3 mole fraction, then decreases
to negligible values. The concentration of Th in the Bi stream decreases
from 1.26 x 10“3 mole fraction in the upper part of the column to

5.6 x lO_6 mole fraction at the Bi outlet. The concentration of lithium
in the bismuth decreases from about 1.23 x 10*3 mole fraction in the

upper part of the column to about 3.2 x 10“4 mole fraction at the column

exit.

The concentrations of plutonium and protactinium in the salt entering
the decay tank are 2.22 x 10'"!‘r and 1.244 x 10“3 mole fraction respectively.
The concentration of protactinium in the decay tank is 1.231 x lOM3 mole
fraction. Under ideal steady—-state operating conditions, approximately
87% of the protactinium present in the reactor system would be held in

the decay tank.

The variation of the protactinium concentration in the reactor and in
the decay tank with bismuth flow rate is shown in Fig. 6. The results
indicate that the isclation of protactinium in a reactor fueled with

plutonium would be feasible.

3. EFFECT OF CHEMICAL PROCESSING ON THE NUCLEAR
PERFORMANCE OF AN MSBR

M. J. Bell L. E. McNeese

A series of calculations has been performed to investigate the im-

portance of individual fission product elements on the neutron poisoning
Pa CONCENTRATION x 10° (mole fraction)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ORNL-DWG 69-7642
i CONDITIONS
1200 { REACTOR VOLUME: 1481 2
DECAY TANK VOLUME: 200 ft2
SALT FLOW RATE: 2.5 gpm
1000 ]
\ TOWER ABOVE TANK: 3 STAGES
\ TOWER BELOW TANK: 4 STAGES
800 |
500
| \ DECAY TANK
400
500 REACTOR
TL//
0
2.5 2.6 27 2.8 2.9 2.0 3. 3.2 3.3 3.4 25

BISMUTH FLOW RATE {(gpm)

Fig. 6., Variation of Calculated Steady-State Protactinium Concen-
tration in the Decay Tank and in the Reactor with Bismuth Flow Rate for
an MSBR Fueled with Plutonium.

o1
16

of a molten salt reactor.

These calculations were made with a computer

code, MODROD, which is a combination of the ROD reactor optimization and

2
design code and the MATADOR steady-state material balance code.l’

MODROD allows an accurate computation

of the neutron production and loss

in the reactor core and of the material balances for 461 fission product

isotopes.

fluid MSBR fueled with 1680 ft3

71.7-16.0-12.0~0.3 mole % LiF-BeF.~-ThF,6 ~UF, .

2
reactor concept are given in ref.

The calculations were performed for a 2250-Mw(thermal) single-

of fuel salt of nominal composition

l 4 Further details of the

3.1 Computational Procedure

The MODROD computer code employs
reactor physics calculations that are
rates for MATADOR, which is used as a
energy~group diffusion calculation in
a two-dimensional flux synthesis. It

rial balance calculation for a number

the fuel salt, moderator, and reactor

ROD as a main program to perform
used to provide neutron absorption
subroutine. ROD performs a nine-
the radial and axial directions and
also performs a steady-state mate-
of important mneutron absorbers in

vessel. These calculations are

performed for a fixed reactor configuration (variable specified option)

for an assumed "lumped" fission product concentration in order to obtain

a set of concentrations and neutron absorption rates for the fuel material.

These data, along with the ratio of the resonance flux per unit lethargy

to the thermal flux and the spectrum-averaged neutron cross section for a

reference nuclide, are then used by the MATADOR subroutine.

The reference

nuclide has a cross section of 1.0 barn for 2200-m/sec neutrons, and this

. . . 1
cross section varies with neutron energy, E, as JE

MATADOR uses the

absorption rate, concentration, and average cross section of the reference

absorber to compute a mean thermal flux, assuming a value of 200 Mev for

the average thermal energy released in fission.

quantity f¢2, which has been defined by Alexander et al.4

to the thermal flux of Wescottfi5

This flux is equal to the

and is equivalent

It is used by MATADOR to compute neutron

absorption rates for the fission products and those materials not treated

explicitly in ROD:
17

ob = £6,, [Uizoo + RIFAC * RI], (1)
where
33 = gpectrum-averaged neutron absorption rate in given isotope,
f¢2 = thermal flux,
52200 = cross section for 2200-m/sec neutrons,
RIFAC = resonance integral factor,

RI = resonance integral for given isotope.

2
For the isotopes 23“Th, 233Pa, 233U, 234U, 235U, 236U, and 237Np, which

are treated separately in ROD, MATADOR uses the absorption rates cal-
culated by ROD directly. Using these neutron absorption rates, the
MATADOR subroutine computes steady-state inventories for all of the
materials in the fuel salt, as described in ref. 2, taking into account
diffusion of the noble gases into the graphite and mass transfer of the
noble gases and metals to the circulating helium bubbles. The steady-
state concentrations as determined by MATADOR are then used to compute
a new‘lumped fission product poisoning, which is used as input for ROD
to repeat the diffusion and neutron balance calculations. This cycle is
repeated until the change in the ratio of the lumped fission product

concentrations for two successive iterations is less than 17%.

3.2 Chenical Behavior of Fission Products and Actinides
During Processing

In the calculations that were performed, the effect of chemical
processing on the neutron poisoning by individual fission product elements
was investigated by wvarying the éfficiency with which these elements were
removed from the fuel salt about a reference value. The reference case
assumed that halogens and rare earths were removed on a 50-day cycle,
zirconium and seminoble metals on a 200-day cycle, protactinium on a

five-day cycle, and that salt was discarded on a 3000-day cycle to remove
18

the active metals and neptunium. In addition, the noble gases and the
noble metals were assumed to have a 50-sec residence time in the fuel
salt and a 110-sec residence time in the helium bubbles. The mass trans-~
fer coefficient for migration of the noble gases to the graphite was
selected to give a poisoning of 0.5%7 for 135Xe. The removal efficiencies
for the individual elements were varied from 22 to 2"6 for all elements

from Zn to Ho, except Kr and Xe. The removal efficiencies for krypton

and xenon were kept constant.

The base case used for these calculations differs slightly from that
used by Baumann to optimize the reactor design.3 Baumann assumed that Pa
was removed on a 3-day cycle; Eu was removed on a 500-day cycle; Zr, the
noble metals, and the noble gases were removed on a 20-sec cycle; and the
salt was discarded on a 5~year cycle. The effect of the different
assumptions on the reactor performance is illustrated in Table 1, which
compares the neutron balance for Baumann's system with the neutron balance
for the system described in this report. The removal times used in our
work give a lower breeding ratio than that for Baumann's Case CC-53
(1.0611 vs 1.0645). The corresponding fuel yield is decreased from

3.338% per annum to 3.214% per annum.

3.3 Calculated Results

The effect of chemical processing on the performance of the refer-
ence MSBR is illustrated in Figs. 7-15. The quantities plotted are the
change in fuel yield and the change in fuel cycle cost as a function of
removal time. The change in fuel cycle cost includes only the inventory
charges for the excess fissile material that would be required if an
element were removed less efficiently than in the base case for a fixed
processing scheme. The elements whose poisoning is most significantly
affected by chemical processing are the rare earths, especially Nd, Sm,
and Pm. Other elements which are important are Zr, Ba, and Sr. A number
of elements, primarily noble metals and seminoble metals, are not included
in the figures. No significant variation in the reactor performance was
observed for these materials over the range of processing efficiencies

investigated.
Table 1.

Comparison of Neutron Balances in Reference MSBR Design with Those
Calculated for Assumed Processing Conditions

 

Baumann Case CCnSSa

 

This Report

 

Abserption Capture Fission Absorption Capture Figsion
2324y 0.9889 0.9858 0.0031 0.9847 0.9816 0.0031
2335, 0.0017 0.0017 G.0000 0.0026 0.0026 0.0060
233, 0.9248 0.1001 0.8247 0.9230 0.0999 0.8230
234 0.0809 0.0805 0.0004 0.0826 0.0822 0.0004
335u 0.0752 0.0140 0.0612 6.0770 0.0143 0.0627
¢3fu 0.0085 0.0085 9.0000 0.0089 0.0089 0.0000
237 0.0059 0.0059 0.0000 0.0070 0.0070 0.0600
8o 8.0055 0.0009 0.0046 0.0055 0.0009 0.0046
TBe (m,) 0.0016 6.0016 0.0 0.0016 0.0016 0.0
i 0.0160 0.0160 4.0 0.0159 0.0159 0.0
¥ 0.0206 0.0206 0.0 0.0205 0.0205 0.0
611 5.0023 0.0023 0.0000 0.0021 0.0021 0.0000
Graphite 0.0522 0.0522 0.0 0.0520 0.0520 0.0
Fii?;iinii°iiiient 0.06000 0.0000 0.0 0.0000 0.0000 0.0
Leakage 0.0018 0.0018 0.0 0.0019 0.0019 0.0
Delayed neutrons 0.0032 0.0032 0.0 0.0032 0.0032 0.0
Pixed polson fraction 0.0050 0.0050 0.0000 0.0050 0.0050 0.0000
”§§§§§32851381°“ 0.0146 0.0146 0.0 0.0147 0.0147 6.0
Chromium 0.0014 0.0014 0.0000 0.0014 0,0014 0.0000
Iron 0.0007 0.0007 0.0000 0.0007 0. 0007 0.0000
Nickel 0.0183 0.0183 0.0000 0.0183 0.0183 00000
Molybdenun 0.0023 6.0023 0.0000 0.0023 0.0023 0.0000
ne 3.2316 2.2310

 

?MSR Program Semiann. Progr. Rept., Feb. 28, 1963, ORNL-4396, pp. 77-79.

6T
CHANGE IN FUEL YIELD (% per annum)

0.0

-0.5

-4.5

-2.0

-2.5

ORNL-DWG 70- 4509

 

II?IT

i d

 

1

-4 0.08

- 0.06

-4 0.04

40.02

CHANGE IN FUEL CYCLE COST (mili/kw hr)

40.00

 

i L e g d ] L b g ] nt

 

 

10

Fig.

7.

20 50 100 200 500 1000 2000 5000
REMOVAL TIME (doys)

Effect of Neodymium Removal Time on MSBR Performance.

0c
CHANGE IN FUEL YIELD (% per annum)

ORNL- DWG 70- 4510

 

 

 

 

=TT T Y T T T T =TT T T Y

0.00 —40.04
=
.
iy
~

-0.25 ~{0.03 €
-
o
O
O
wl

-0.50 | .

0.02 4

>
(&
-
w
e

.?

.78 — 0.0t z
w
O
=
<
r

-1.00 —40.00 ©

-14.25 it g § 1 i o b gl } 4 o log a5l 1 i

5 10 20 50 100 200 500 1000 2000 5000

Fig.

10.

REMOVAL TIME (days)

Effect of Zirconium Removal Time on MSBR Performance.

T¢
CHANGE IN FUEL YIELD (% per annum)

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

ORML— DWG 70- 454

 

 

 

 

 

1 1] T

B —0.03 __
£
z
-4
=

— —o.o02 E
o
0
o
o
W

- —~0.01 3
>
&
—
i
=
T

— —0.00 =
w
o
=
<
T
o

— —{-0.01

1 i 4 A l t 1 1 1 L 1 i 4 l 1 i 1 j. i 1 4 L 1 i i
5 10 20 80 100 200 500 1000 2000 5C00
REMOVAL TIME {days]
Fig. 9. Effect of Protactinium Removal Time on MSBR Performance.

¢C
CHANGE IN FUEL YIELD (% per annum)

ORNL- DWG 70- 4512

 

 

 

 

 

 

! T 1 i ' ’ 1 [} l T 1 ! TI l ¥ T ! ¥ 1 r i ' [ 1
0.0 = o003 _
£
z
4
~
=0.2 4002 E
-
v
O
o
L
-~0.4} —o.01 3
>
o
J
tw
D
il
-0.6 —10.00 Z
w
o
2
<
S
-0.8} —4-0.01
P, L} } 1 a1 o0 0] i e b o0y ] }
50 100 200 500 1000 2000 5000 104 2x104

REMOVAL TIME (days)

Fig. 8. Effect of Samarium Removal Time on MSBR Performance.

te
CHANGE IN FUEL YIELD (% PER ANNUM]

24

ORNL DWG T7i-2l
0.007.

 

0.05

-0.10

=015

-0.20

 

 

-1 0.006

-+ 0.004

-1 0.003

— 0.002

- 0.001

-0.001

 

| ool 1 1 oo b gl 1 el 1L L L LLI-0.002

 

-0.25 1
0

Fig. 11.

100 1000 10,000
REMOVAL TIME { DAYS)

Effect of Europium Removal Time on MSBR Performance.

CHANGE (N FUEL CYCLE COST {MILL/KWHR)
CHANGE IN FUEL YIELD(% PER ANNUM)

25

ORNL DWG 71.-90

 

 

 

 

0.10 0.008B
0.05 |~
-1 0.006
O
- -
- 0.004
“0.05}
~0.10 —
- 0.002
Q.15 -
- O
-0.20
-0.25 1 11 gl 1 1 L 1o t1 a0 1 1 Lol 111 )-0.002
10 100 000 10,600

Fig. 12.

REMOVAL TIME (DAYS)

Effect of Praseodymium Removal Time on MSBR Performance.

CHANGE IN FUEL CYCLE COST (MILL / KWHR)
CHANGE IN FUEL YIELD {% PER ANNUM)

26

ORNL DWG 71-92

 

 

 

 

 

 

 

0.02 0.008
0 e
-1 0.007
-0.02 %
- 0.006 =
~0.04 - ~
|
=
-0.06 ~ 0.005 =
a
~0.08 S
- 0.004 ©
Ll
-0.10 o
o>
| 0.003 ©
Q42 tfl
=
- 0.002
0.14 |- 002
~0.16 |- lc;é'
— 0.00i %
0.8 O
o
-0.20 -
_0-22 1 L 1 11§ [ 1 i } L 1 1.1 | 1 i L L L. 1L --O‘OO'
10 100 1000 10,000

Fig. 13.

REMOVAL TIME (DAYS)

Effect of Lanthanum Removal Time on MSBR Performance.
CHANGE IN FUEL YIELD {% PER ANMUM}

27

ORNL DWG 71-99 Ri

 

 

 

 

 

0.05 0.0I2

Or -4 0.010
~005 -1 0.008
~0.10 - - 0.006
‘0-|5— 4 0.004
-0.20+ - 0.002
~0.25+ 4 0

REMOVAL TIME { DAYS)
Fig. 14. Effect of Barium Removal Time on MSBR Performance.

CHANGE IN FUEL CYCLE COST (MILL/KWHR)
CHANGE IN FUEL YIELD (% PER ANNUM)

28

 

 

ORNL DWG 7I- IOOO

 

 

 

0.02 0040
- 00035
O Lo
- 0.0030
-0.02}
~ 00025
~004 - 0.0020
-0.06 |- - 0.0015
- 0.0010
-0.08 |-
-+ 0.0005
-QA0
=40
_O’os 1 111 1 i llllll[ } 1 Illllll i ) iillll__oflocos
102 103 10* Toke 0%

Fig.

15.

REMOVAL TIME {DAYS)

Effect of Strontium Removal Time on MSER Performance.

CHANGE IN FUEL CYCLE COST {MILL/KWHR)
29

3.4 Discussion of Results

We have investigated the effect of the removal efficiencies for
fission product elements on the performance of an MSBR. Experimental
data available for 4 number of elements indicate that the removal times
of these elements will differ significantly from the times set for the
relference caseg6 The present results allow estimates to be made of the
magnitude of the change in fuel yield and fuel cycle cost as a result
of such differences. The data are also useful for estimating the
change in power cost due to fissile inventory charges and changes in

breeding ratio for alternative processing schemes,

The present investigation shows no effect of processing of the noble
metals (Nb through Ag, Se, and Te) on the performance of the reactor,
This results from the assumed removal of these elements on a short (50~
sec) cycle as a smoke or mist. This treatment is consistent with re=-
ported noble-metal behavior in the MSRE,7 However, if conditions in an
MSBR are sufficientiy different from those in the MSRE (thereby resulting
in longer removal times), the chemical processing of these elements will

have an important effect.

A simplification made in this study was the treatment of the cross
sections of the isotopes of samarium in the same manner as those of the
other fission preoducts. However, in a typical ROD calculation, the
isotopes 1495m and 1518m are considered sufficiently important as neu-
tron absorbers to be included explicitly in the nuclear calculations.
Several calculations were performed using a more precise treatment of
gsamarium in order to determine the magnitude of the error introduced in
the present study. For samarium removal times of less than 100 days,
the agreement between the two treatments is quite satisfactory (i.e.,

+ 2% in the predicted changes). At longer times, the two-group treatment
of the cross section used here overestimates the samarium poisoning. The
poisoning is overestimated by 20% at 800 days, the longest removal time for

which the comparison was made. However, the operating conditions for the
30

MSBR anticipate that samarium will be removed efficiently at times

lesg than 800 days; the results are applicable in the expected range of
operating conditions. A few of the data points were also checked for
neodymium and promethium, using the complete ROD treatment of the fission
products. For these elements, the values for poisoning computed by both
ROD and MODROD showed excellent agreement over the entire range of removal

times.

4. REDUCTIVE EXTRACTION EXPERIMENTS IN A MILD-STEEL
FLOW-THROUGH FACILTITY

B. A. Hannaford H. D. Cochran
L. E. McNeesge C. W. Kee

Equipment8 for the study of reductive extraction in a mild-steel,
flow-through facility has been installed. Preliminary testing of this
equipment and the addition and purification of a bismuth charge (184 kg)
have been described.9 Following purification of the bismuth, sufficient
metallic thorium was added to the system to produce a thorium concen-

—1
tration of 10 & mole fraction.

The bismuth-thorium solution was fed through the system in order
to remove residual oxides from the surfaces of the system. The small
changes in the thorium concentration in the solution indicated that the
earlier hydrogen treatment of the system had thoroughly reduced oxides
present on the steel surfaces. A 52.8-~kg charge of salt (72-16-12 mole
7 LiF~BeF2—ThF4) was then transferred to the treatment vessel containing
the bismuth charge for a 30-hr treatment with HF*H2 to remove trace
quantities of oxygen and to oxidize the thorium metal to ThFé.

Three experiments were performed in which bismuth and salt were fed
to the column. In the first two runs, the pressure drop across the col-
umn and associated transfer lines was greater than expected; it prevented

the flow of bismuth from the column in the second run. The bismuth drain

line was found to be blocked by iron crystals, which resulted from the
31

mass transport of iron from the hotter regions of the system. After

the blocked tubing had been replaced, a third run was made at flow rates
of about 75 ml/min for each phase. However, flow rates were not steady,
apparently due to bismuth that was trapped in the salt overflow loop at
the top of the column. The loop was modified to allow bismuth to drain
from the loop to the line carrying bismuth from the bottom of the column.
This work is described in greater detail in the sections that follow. A
digital simulation of the automatic flow control systems for the salt and
bismuth was carried out to determine whether the control system was stahble
or whether it might introduce oscillations into the flow rates; this work

is discussed in Sect. 5.

4,1 Addition of Thorium, and Subsequent Transfers of
Thorium~Bismuth Solution

A total of 201.5 g of thorium metal in the form of irregularly
shaped slugs was added to the bismuth in the treatment vessel (Fig. 16)
in order to remove the oxide impurities that remained after hydrogen
treatment. Samples of bismuth were taken for thorium analysis by with-
drawing the metal into a sample capsule through a porous metal filter.
The thorium contents of these samples, shown in Fig. 17, are indicative
of the rate of dissolution of the thorium in bismuth. The dissclution
rate, which was lower than expected, was probably limited by the rate
at which the dissolved material diffused away from the surface of the

remaining thorium metal.

The bismuth~thorium solution, prepared as described above, was
transferred through a fritted-metal filter to the metal feed tank by
pressurizing the treatment vessel. The metal solution was then routed
through the column and the flowing -stream bismuth samplers to the bis-
muth catch tank by pressurizing the metal feed tank; this operation
served to remove residual oxide from the metal surfaces by reduction
with thorium. The pressure required to pass argon through the salt

feed line to the column indicated that about 12 in. of bismuth remained
ORNL-DWG 65-14347 A
{

. Y
Fyv-i FILTER
T-6

 

 

 

 

 

 

 

 

METAL FEED AND SALT FEED AND SALT AND METAL
COLLECTION TANK COLLECTION TANK TREATMENT VESSEL

Fig. 16. Simplified Salt and Metal Flowsheet.

¢t
33

CRNL DWG 7I1-104

 

 

 

 

1500 I T I I
g o
Q.
o
- 1000 |~ 7
«I
o
}-..
=
Ll
Q
=
&
o
g 500 -
o
g o
l__
o
%
0 | | 1 i
0 5 10 15 20 25

TiME (hr)

Fig. 17. Dissolution of Thorium in Bismuth as a Function of Time.
Thorium metal (301.5 g) added to 200 kg of bismuth at 600°C; stirring
by argon sparge v 0.3 scfh.
34

in the column, while the lack of a static pressure required to pass
argon through the salt return line from the column indicated that this
line was free of bismuth. The remaining bismuth was transferred from
the column, the salt feed line, and the bismuth return line into the
bismuth catch tank by pressurizing the salt feed tank and the column.
The effluent bismuth was then transferred from the bismuth catch tank

to the treatment vessel.

Samples of bismuth were taken from the bismuth feed tank, the bis-
muth flowing-stream sampler, the bismuth catch tank, and the treatment
vessel. Analyses of these samples showed no appreciable change in
thorium content and indicated that most of the oxides that had formed
on the piping had been removed from these surfaces by hydrogen treat-
ment prior to the transfer of the bismuth-thorium solution through the

system,

4.2 Addition of Salt, and Hydrofluorination
of Salt and Bismuth

Sixteen liters of ealt (72-16-~12 mole % LiF—-BeF —ThFA) was trans-

2
ferred from a salt shipping vessel to the treatment vessel via a tem-
porary heated line. The weight of the salt added to the treatment
vessel (52.8 kg) was determined from the decrease in weight of the

shipping vessel,

The bismuth and salt in the treatment vessel were then sparged
with a H2~HF mixture in order to remove oxides from the salt and dis-
solved solutes (Th, Li, and Fe) from the bismuth. In order to prevent
excessive attack on the containers and piping, the HF concentration was
kept below 30 mole 7%, although an attempt was made to keep the concen-
tration as near 307 as possible in order to obtain the maximum oxide
removal rate. The nominal total flow rate was 33 scfh. The HF flow was

set by controlling the pressure drop across a capillary; the HZ flow was

set by use of a rotameter and was checked by a wet-test meter.
35

From the treatment vessel the H2~HF stream passed successively
through a NaF bed (to remove HF) and then through a wet-test meter (Lo

measure the total H. flow). Accumulation of HF in the NaF bed was fol-~

2
lowed by measuring the temperature along the length of the bed. The
weight of unreacted HF leaving the treatment vessel was determined by

changes in the weight of the bed.

The total weight of HF fed to the system was determined by weigh~-
ing the HF supply tank. The system also contained an aqueous scrubber
and a smaller (0.05 ft3/rev) wet-test meter in parallel with the NaF
trap for sampling the gas fed to the treatment vessel or leaving the
vessel. The concentration of HF in the gas stream was determined by
passing the gas through 250 ml of a 0.4 N NaOH =solution in the scrubber.
When 0.05 ft3 of H2 had passed through the wet-test meter, the gas flow
was stopped and the golution was removed for analysis. The HF concen-
tration in the gas was determined by titrating small samples of the
scrubbing solution with 0.1 N HCL. Both the feed and the discharge
streams were monitored in this manner. Utilization of the HF was cal-
culated from the feed and discharge concentrations. A plot of the
results (see Fig. 18) showed that the HF utilization quickly decreased

from an initial value of about 80% to about 15%, where it remained for

a significant period of time.

4.3 Hydrodynamic Experiments: Runs HR-1 and -2

In run HR-1, bismuth and meolten salt were fed countercurrently
through a packed column at a low flow rate (0.05 liter/min for each
phase). The purpose of this run was to demonstrate satisfactory op-
eration of the system in general and of the flow control system in

particular.

Run HR~1 was started by filling the column with molten salt.
However, when the bismuth flow to the column was initilated, salt was

forced out of the column toward the salt feed tank and resulted in a
36

ORNL DWG 71-102

 

 

 

 

 

 

100 | T I I T T T 20
A
& =118
A 4 A B Al L
80 A A Jie
. e
* A
- o
z a T
O 6D -2 ©
k- a
< w
"] & % HF in feed ~10 W
S ® % HF Utlization Z
4 - 8
g 1° %
&
-6
o e
20 -4
o o
"o ° efif“‘)\oi\\é
© 2
P %
o
0 | | | | ! ! 1 0
0 10C 200 300 400 500 600 T00 800
CUMULATIVE GAS FLOW{ft3) (Gas flow rate=s30scfh)
Fig. 18. Utilization of HF as a Function of Time. Hydrofluorina-

tion of 20 liters of bismuth and 16 liters of LiFmBervThFé (72-16-12

mole %) with HF-H

2

at approximately 600°C.
37

negative salt feed rate. During the remainder of the run, a kind of
quasi-countercurrent operation was maintained for only 30 min. In

this 30~min period, the salt flow reversed diréction three times,
corresponding to small surges in the bismuth flow rate. Approximately
1.5 of the 3 liters Qf bismuth fed to the column overflowed to the salt
catch tank. BSubsequent runs provided the explanation for this behavior,
nafiely, that the salt overflow loop was partially filled with bismuth
and the bismuth exit line at the base of the column was partially
plugged with iron deposits and/or frozen salt. It was concluded that
the flow metering system operated satisfactorily; molten salt was
metered to the column at a uniform rate for about 90 min, while bismuth

wae metered for a shorter time.

Run HR-2 was made to determine the cause of the low bismuth through-
put during the first run. 1In this run, the column was filled with salt,
and the salt overflow line from the column was sealed by freezing salt in
the exit freeze valve. When bismuth was then fed to the column, much
of the salt was displaced and forced back into the salt feed tank. During
the period in which bismuth was fed to the column at the rate of about
50 ml/min, bismuth flowed through the column, out the salt inlet line,
and into the salt feed catch tank. Very little bismuth flowed through
the bismuth outlet line to the bismuth catch tank. ¥Finally, the bismuth
flow was stopped, and the column was pressurized from the salt feed tahk.
Under a pressure of about 20 psi, only about 500 ml of bismuth flowed
through the bismuth outlet line in an hour; this made it obvious that
an obstruction existed in the line between the column and the bismuth

catch tank.

The portion of the bismuth line between the column and the flowing-
stream sampler was removed and cut into approximately 6-in.-long sections,
which were suspended in an argon atmosphere at about 600°C prior to
examination. The bulk of the bismuth and salt in the line drained from
the sections, leaving four partial obstructions at likely cold spots.

These deposits consisted of clusters of whisker-like crystals that were
38

shown to be primarily iron by spectrographic analysis. The line between
the flowing~stream sampler and thc bismuth receiver contained a single
deposit that completely filled the cross section of the 0.26~in.~ID
tubing at the location of a probable cold spot. Elsewhere along the
line, the steel tubing was heavily air-oxidized, confirming that these
portions of the line had been exposed to temperatures considerably in
excess of the 600-650°C range indicated by a few thermocouples. At one
point, the wall thickness had decreased from its initial value of 0.058

in. to 0.042 in.

The portion of the line that was replaced was painted with an
oxidation-retarding paint, and the nuwmber of thermocouples on the line
was increased in order to prevent localized overheating. More impor-
tantly, the temperature of the bismuth feed tank was reduced from 600 C
to 540°C in order to reduce the solubility of dissolved iron from about
70 ppm to about 25 ppm. Calculations indicated that the temperature
of the bismuth could be increased to 600°C during its passage through

atout 5 ft of heated line between the feed tank and the column.

4.4 Hydrodynamic Run HR-3

Run HR-3 was the third attempt to achieve countercurrent contact
of salt and bismuth in a packed column. During a period of about 2-1/2
hr, there was no evidence of incipient plugging in any of the lines;
however, the salt and bismuth flow rates were not constant. Although
these flow rates were programmed to be 90 ml/min and 75 ml/min, respec-—
tively, flow was intermitteunnr throughout most of the run. It appeared
that the salt overflow loop at the top of the column contained some bis-
muth — either from the initial transfer of bismuth through the system or
from entrainment or flooding at the top of the column. Subsequently, the
salt level in the column periodically increased sufficiently to overcome
the bismuth seal; this resulted in a sudden flow of salt from the top of
the column and, perhaps, replenishment of bismuth, which may have been

carried out of the loop.
39

The solution to this problem was to install a line to drain bismuth
from the low point in the salt exit line. For the purpose of observing
bismuth entraimment, a small chamber and a freeze valve were inserted
in the drain line (Fig. 19), With the freeze valve open, large quantities
of bismuth passing through the chamber would be detected by changes in
pressure between a pair of argon bubblers spaced about 3 in. apart (in
elevation). By freezing a seal in the drain line, the rate of bismuth
accumulation in the chamber could be measured over a period of time

and thus provide data on the rate of bismuth entrainment.

These initial experiments showed that it was sometimes necessary to
pressurize the salt jackleg (T-6) with argon in order to maintain the
salt flow to the column. Manual control of the argon pressurizing valve
and bleed valve proved to be difficult; therefore, we Installed an auto-
matic contrel system that maintained a constant level in T-6& by varying
the flow of argon as required. After the automatic controller had heen

added, operation of the jackleg became quite satisfactory.

5. DIGITAL SIMULATION OF THE FLOW CONTROL SYSTEMS FOR THE
REDUCTIVE EXTRACTION FACILITY

C. W. Kee L. E. McNeese

The rates at which the salt and metal flow out of the feed tarnks in
the semicontinuous reductive extraction facility (see Sect. 4) are set by
using a ramp generator to continuously change the get point of a conven-
tional level controller. The fact that flow from the tanks is propor-
tional to the rate of change of the level suggests that optimum controller
constants for this application may be quite different from those for simple
level control. In addition, the time lag in pressurization or depressur~
ization of the gas volume above the liquid suggests the possibility of an
unstable system. A digital simulation was carried out to investigate the
stability of the control system over various ranges of controller settings

and to determine the most desirable settings.
40

ORNL DWG 7i-103

 

SALT JACKLEG

EXTRACTION {T6}
COLUMNI(C?)

  
   

SALT
SAMPLER

   
 

 

§——BISMUTH IN

 

 

 

 

 

 

 

 

 

r‘——w~SALTIN
TO SALT RECEIVER &
| DETECT%R
TO BISMUTH BISMUTH (SG-12
RECEIVER ] SAMPLER }
COOLING |
AlR |
l
l
I
I
I
I
qq_mJ | 10
: ;
\. | ) - 4
h,inchas

Fig. 19. Extraction Column with Revised Salt Overflow. Vertical
dimensions are to indicated scale; horizontal dimensions are not to
scale. Packed section of column, 0.82%4 in. ID x 24 in. long, filled
with 1/4~in.~diam right circular cylinders of molybdenum.
41

5.1 Salt and Metal Feed Systems To Be Studied

The system that was investigated is shown schematically in Fig. 16.
It consists of a 20-liter tank filled initially with 15 liters of liquid
(salt or bismuth). During operation, the liquid flows ocut of the tank
through a dip tube and 10 ft of 3/8-in.-diam tubing to a discharge point
about 125 em above the entrance to the dip tube. The bismuth flows into
the top of the column at the off-~gas header pressure, while the salt is
routed to a tank in which the pressure at the salt inlet point is set by
the pressure drop through the column. The pressure drop of the salt
through the column is almost entirely dependent on the bismuth holdup in
the column and is approximately equal to the static head produced by the

bismuth.

The rate at which argon is fed to the vapor space above the liaquid
in the feed tanks is determined by a level controller. The set point of
the level controller is changed linearly with time and is proportional
to the output of a ramp generator whose ramp time is adjustable. Ideally,
the level controller maintains the level in the tank at a poini equal to
the set point. 1In this case, the flow rate out of the feed tanks is
constant and proportional to the output from the ramp generator. The
proportionality constant between the flow rate and the rate of change of

the ramp generator output is then the tank calibration constant.

5.2 Mathematical Analysis

Consider a feed tank of volume Vt, vhich consists of a gas volume,
Vg, and a liquid volume, VS. The rate of flow of liquid from the tank

is then related to the rate of change of the gas volume as follows:

dv
EEE = ¥, (2)

P = ==, (3)
42

where
P = pressure in the gas space of the vessel,
n = number of moles -0of gas in the vessel,
R = gas constant,

T = absclute temperature.

If it is assumed that the temperature is constant, P is a function of two

variables (n and Vg),Land-one‘can write

: dv

.
d° 3P dn 22 g )
dt an dt BVg dt

 

Also,
aP RT
m Ty )
g
1P nRT p )
-—-»-—8 = 22«“"‘}""‘, \6)
V)
g g g
dn _GP
at T RT )
and
dv dVS :
o a o (8)
where

G = volumetric flow rate of the gas into the vessel.

The rate at which gas is fed to the vessel was assumed to be propor-
tional to the difference between the volume cof liquid in the vessel, VS,
and the corresponding controller set point value, V e Thus

S,8€

G =K -v), (9)

3 (Vs,set s
43

where
K3 = controller constant.
As noted earlier, values of V‘3 sot VETe provided by a ramp generator and
~
result from the relation
=V ., . =-tF 0
ngset s,init t Yeet (10)
where
V ., .. = initigl liquid volume,
S,1nit

set flow rate.

#

set

Substituting Eqs. (5)-(8) into Eq. (4) yields

d¢ _ G _ FP (11)

Since Vg = Vt - Vs’ one can then write

e pe-0 (12)

dt -
_ Vt VS

The total pressure at the entrance to the dip tube in the feed tank is
the sum of the gas pressure above the liquid and the hydrostatic pressure
due to the weight of the liquid. Thus the pressure at the entrance to

the dip tube is

V
Pi = Pg -+ Kg' 0g, (13)
where
Pi = pressure at the entrance to the dip tube,
A = cross—~sectional area of the vessel,
p = density of liquid,
g = gravitational constant.
bl
If a tank calibration constant, Ko’ is defined as

K o=A
o Pg
Eq. (11) becomes

Pi = Pg + VS/KO. (14)

The resistance to the flow of liquid through the line is due to three
effects: (1) dnlet and exit frictional losses, (2) velocity losses

caused by bends in the line, and (3) viscous losses in the equivalent
length of straight line. For laminar flow, the functional form of all

these resistances is the same and is

F =K, /4P, (15)

=
H

a constant,

>
o
#

dynamic pressure drop through the line.

It was assumed that the entrance and exit losses corresponded to 1 veloc-
ity head each and that the bends corresponded to 7.5 velocity heads. The
viscous losses were found to represent about 0.5 velocity head. Accord-
ingly, the total resistance to flow through the line was taken to be 10

velocity heads, which then sets the value of K The liquid was assumed

2.
to discharge from the line into a volume having an absolute pressure of

1 atm. UHence, the pressure drop across the line, AP, is given by:

s - - , 16
P =P, - P, - hog, (16)

where

H
i

discharge pressure (1 atm absolute),

h

#

discharge height above the inlet to the dip tube.
45

Hence, Eq. (15) becomes

SO
- Pe - hpg. (17)

2= .‘..:' ,
F=K,VP,

The final set of equations that was sclved consists of the following:

= -F (18)

p -

dt Vt -V
G = K3 ( s,set Vs) (20)
s,8et - Vs,init -t Fs@t (21)
i 7 g K, (22)
e K2 fPi - P, - e (23)

5.3 Solution of Equations and Calculated Results

The set of equations listed above was solved by numericsl integra-
tion, using a trapezoidal-rule approximation. Values for the constants
and initial conditions are given in Table 2. The values used for Vs,init
and Fset are believed to result in the least stable contrel action, that
is, a small gas volume and a high set point for the liquid flow rate.

The initial pressure in the vessel (Pinit) was varied over the ranges
2.0 to 3.5 psig and 6.0 to 9.0 psig for salt and bismuth flows, respec—
tively. The controller constant (KB) was varied over the ranges 0.04

to 0.2 sec"l for salt flow and 0.1 to 0.45 secml for bismuth flow,
46

Table 2. Constants and Initial Conditions Used
for Simulation of Flow Control Systems

 

 

Constant Value
Vt 20 liters
. 15 Iliters
s,init
I
FSet 150 cm™ /min
Ko 90.9 liters/atm for salt, or
31.5 liters/atm for bismuth
. , 1/2
K. 7.536 liters/min (atm)

 

One of the most difficult control situations is represented by
startup conditions in which the initial pressure in the tank is insuf-
ficient to overcome the static head of liquid in the exit line from the
tank. Since the available volumes of salt and bismuth are limited, we
are primarily interested in rapidly achieving steady flow rather than in
damping out disturbances after steady flow is achieved. The effect of
the controller constant on the approach to steady flow for the salt and
bismuth systems is shown in Figs. 20 and 21, respectively. Operation
of the salt system was investigated more extensively since disturbances

are more likely to occur in it.

As can be seen in Fig. 20, the critical value of the controller
constant appears to be about 0.1 secml for the salt phase. The nature
of the respounse changes rather slowly from an underdamped to a slightly
overdamped situation as the controller constant increases from 0.1 to
0.2 Secul. With higher controller constant values, there would be an
increased tendency for the flow to be completely stopped. This is
undesirable because bismuth could enter the salt feed line and further
upset the system. In addition, higher values for the controller constant

introduce instability and do not allow the system to be brought to steady-
47

ORNL DWG 71-106

 

 

 

  

 

 

 

0.5

0.1
600 _0.15
- TR\, 0.2
- 70.07
e ,lo.os '
S00 _10.05
e :F).O‘} Initial Tank Pressure=2 psig
= -
- 400
£ -
- u
}__ =
<L L
T 300
; e
Qo "
.
m %
I...
&J 200 Set Point
) B X
n = Y
- 0.07 “0.06 ~0.05
|00 Proportional Controller ——
B _ (Gas Fiow Rate -
i Constant= Salt Volume Error >C¢
0 - 14 i Lo e i o 11y L1
0 0.1 0.2 0.3 0.4
SALT TANK DEPLETION (liters)
Fig. 20. Calculated Salt-Phase Controller Response to a Stepwise

Disturbance.
tion.

Salrt flow rate is shown as a function of salt tank deple-
BISMUTH FLOW RATE {mi/min)

48

ORNL DWG 7i-105

 

 

 

 

600
B 0.2
5004 O.l \Y/O.zs—wo.ss
_ ~0.45
400 Initial Tonk Pressure=8psig
}.
300+
- 0,1
200+
B \< 02 /sm Point
l 4‘\~=o.25-»~w0.35
'00: 0.45
Oplllllj_llllll[lllllllllll
0 0.1 0.2 0.3 0.4 0.5

BISMUTH TANK DEPLETION {(liters)

Fig. 21. Calculated Bismuth-Phase Controller Response to a Step-
wise Disturbance. Bismuth flow rate is shown as a function of bismuth
tank depletion. Controller constants are given in sec™} units.
49

state operation with significantly less tank depletion. We thus con-
clude that values above 0.2 are to be avoided. On the other hand, con-
troller constants below 0.1 secwl provide slow response; therefore,
values below 0.05 sec:_-1 are also considered undesirable. As shown in
Fig. 21, satisfactory operation of the bismuth flow control system can

. . -1
be obtained with a controller constant of about 0.1 sec .

After steady-state operation has been achieved, the bismuth flow
rate is more likely to be stable than the salt flow rate since bismuth
discharges from the transfer line at a constant pressure. The discharge
pressure for the salt varies because of bismuth holdup in the column.
When bismuth flow is initiated, control of the salt flow rate will be
difficult (assuming that a salt flow rate has been established). The
associated change in discharge pressure is believed to be less severe
than a stepwise change of 30 cm of salt, which is equivalent to suddenly
’imposing a bismuth holdup of about 30% on the column. The response of
the salt flow control system to such a change was investigated by choos-

ing an initial tank pressure of 2 psig (see Fig. 20).

Several other factors associated with the control system were con-
sidered briefly. Results using proportional, integral control differed
little from those using proportional control alone, except that a slower
response was observed. The addition of integral control had ounly a
small effect since the system provides its own integral action; that is,

the integral of the flow rate, rather than the flow rate, is controlled.

Continuous removal of the pressurizing gas from the feed tank was
also considered. This provision was found to have no appreciable effect
on the system response as long as the regquired gas addition rate was well
below the difference between the flow capacity of the control valve and
the bleed rate. When the required gas flow approached the capacity of
the flow control valve, the use of the gas bleed decreased the rate of
response of the control system and hence increased the volume of liquid

required to achieve steady-state operation. However, a bleed stream is
50

desirable because it increases the response to a decrease in pressure
drop. We concluded that a small bleed stream, or no bleed stream at
all, should be used at the higher flow rates, while a progressively
larger bleed stream is desirable for the lower flow rates. From the
standpoint of practicality, the blead rate should be such that the

control valve operates at about a 30 to 407 open pesition.

In summary, we conclude that: (1) the system is stable, (2) pro-
portional control is adequate, and (3) the system can be brought to a

steady—-state condition in an acceptably short time.

6. ELECTROLYTIC CELL DEVELOPMENT

M. S. Lin J. R. Hightower, Jr.
L. F. McNeese

The proposed reductive extraction processes for protactinium iso-~
lation and rare—earthb removal require the use of electrolytic cells for
reducing lithium and thorium fluorides into a bismuth cathode to prepare
the metal streams fed to the extraction columns and for oxidizing extracted
components from the metal streams leaving the columns. The cells will re-
quire the use of an electrically insulating material that can withstand
the corrosive conditions present at the cell ancde. In the experiments
reported in the remainder of this section, frozen salt was used to
provide insulation as well as protection against corrosion. The use of

BeO, an insulator, is also under study.

6.1 Formation of Frozen S8alt Layers in Regions of High Heat Generation

An experiment for studying the formation of frozen salt in regions
of high heat generation was carried out previously in a cell comnstructed
from metal rather than quartz.9 Although viewing ports were provided,
it was very difficult to estimate the thickness of the frozem salt layer

formed around the anode, especially near the salt-gas interface. In order
51

to study the formation and maintenance of a frozen salt layer in a

region of high heat generation under conditions of improved visibility,

a quartz cell having a quartz divider was used. A U-shaped, 1/4-in.-diam
cooling tube was located about 1/4 in. above the divider, as shown in

Fig. 22, in order to form a frozen salt layer along the top of the divider.

Iin our first attempt to operate this type of cell, the bismuth was
hydrogen~treated at 700°C with the cooling tube in place. Subsequently,
after addition of the molten salt, nitrogen (coolant) was observed to be
leaking into the cell from the underside of the U-tube (Fig. 23).

We believe that the mild-steel tube was preferentially attacked by bis~-
muth, causing iron to be dissolved at points where bismuth was held
between the tube and the divider during the hydrogen treatment period.
The continual splashing of bismuth from the main bismuth pool against

the small quantity of bismuth in contact with the tube would have allowed
the dissolution of a sufficient quantity of iron to saturate all of the
bismuth at the operating temperature of 700°C. TIn the second trial, the
cooling tube was not lowered into position until after hydrogen treatment

was complete.

With no current flowing through the cell and with both electrodes
being sparged with argon at 0.1 to 0.2 scfh, a frozen salt film was
easily formed on the cooling tube; a mixture of N2 and water was used
as the coolant. The temperature of the salt (66~34 mole % LiFmBeFZ) in
the cell was about 510°C. As expected, the frozen salt film was thicker
at the two 90° bends in the tube than along the straight sections of the
tube. It was also thicker at the colder end of the horizontal tube than

at the warmer end (5/16 in. vs 3/16 in.).

When ac power was applied to the cell, the salt turned a brownish
color initially and then black (within a few minutes), as in previous
experiments. With a low power input to the cell (37.5 w), the thickness
of the frozen salt was practically unchanged (about 1/4 in.). With the
applied voltage maintained at 3 v, the current decreased from 12.5 amp to

11.4 amp in 4 min.
a3

PHOTO 95837

 

 

 

 

 

Fig. 22. Quartz Electrolytic Cell Vessel (4 in. OD) with 1/4-in.-
diam Cooling Tube Along Top of Quartz Electrode Divider.
53

PHOTO 95838

 

 

 

 

Fig. 23. Results of Bismuth Attack on the Underside of the 1/4-in.-
diam Tube Used for Forming Frozen Salt Film on Quartz Electrode Divider.
54

At higher currents, the salt became opaque; consequently, it was
not possible to observe the frozen salt layer on the tube above the
divider. With 4 v between the electrodes, the current was initially
17.8 amp and decreased to 8.8 amp over a period of 15 min. When the
voltage was increased to 6 v, the current increased slightly from about
13.8 amp to 14.2 amp (85 w) over a period of about 3 min even though the
flow rate of the cooling water was increased. At this time, both the
power to the cell and the coolant were turned off; subsequently, the
quartz cell cracked, allowing the salt to drain out of the cell. Later,
examination revealed that a rather uniform salt layer (about 1/2 in.
thick) remained on the cooling tube (Fig. 24). The observed thickness
is believed to be the actual thickness of the frozen salt layer at the
time that the power was turned off since the salt drained from the cell
in less than 30 sec. It was concluded that a frozen layer of salt can

be maintained in a region of high heat generation.

6.2 Use of Beryllium Oxide as an Electrical Insulator

Several of the properties of Be0 suggest that this material might
be useful in constructing an electrolytic cell. Beryllium oxide is a
good electrical insulator, has a high thermal conductivity, and is rel-
atively insoluble in molten fluoride salts of interest. However, since
BeO would tend to dissolve slowly as well as to be attacked by the BiF3
present in the salt, a layer of frozen salt would be used to protect it
from contact with the molten salt. In the event that the salt layer melted
during an upset in cell operation, a low rate of attack on the BeO would be
anticipated; thus sufficient time would be available to form a new frozen

salt layer. During this period, the BeO would serve as the electrical

insulator in the cell.

We have fabricated an anode in which a BeO cup is used as the elec-
trical insulator and serves to separate the two mild-steel liners of the

anode (Fig. 25). The assembled cell is shown in Fig. 26. The top edge

Rt bl o A e 2
55

PHOTO 96027

 

Fig. 24. Frozen Salt Film Remaining on Cooling Tube Above Quartz
Electrode Divider After the Cell Vessel Cracked.
56

PHOTO 96227

 

Fig. 25. Anode Cup Assembly Consisting of Imner and Outer Carbon
Steel Cladding Separated by BeO Insulator.
57

PHOTO 96226

g

 

 

 

    

”Ag

Fig. 26. Assembled Electrolytic Cell Incorporating the BeO-Insu-
lated Anode Cup Assembly.
58

of the inner liner is cooled by coolant that enters the annulus of the
two concentric tubes (3/4 x 3/8 in.) and exits through the center tube.
The BeO cup is attached to the inner cup by means of four pins on the
liner and four grooves inside the cup. The outer liner is supported

on the BeO cup which prevents the two metal liners from contacting each
other. The BeO will contact the bulk of the molten salt only before the
frozen salt layer is formed around the cooling tube. The anode will be
used to test the stability of a frozen salt layer that is in contact
with molten bismuth and that is subjected to localized heating which
results from high current densities between the two electrodes. Cups

of two different sizes (2-1/2 in. OD and 4 in. OD) have been fabricated.
The resulting gross anode areas are 13.8 cm2 and 49.5 cmz. The two

anodes will be used in 4~in.-OD and 7-in.-ID quartz cells, respectively.

7. ANALYSIS OF MASS TRANSFER IN ELECTROLYTIC CELLS

J. §. Watson L. E. McNeese

In the proposed reductive extraction processes for removing prot-
actinium and rare earths from molten-salt reactor fuels, an electrolytic
cell is required (1) to provide lithium and/or thorium metal reductant
for the liquid bismuth entering the extraction column, and (2) to oxidize
extracted materials from the bismuth after it leaves the column. A
study of factors affecting mass transfer in electrolytic cells has been
initiated to identify important cell parameters and to aid in under-

standing and interpreting experimental data.

In principle, one could use the bismuth stream leaving the bottom
of the column as the anode and selectively oxidize uranium and other
extracted materials from the metal. The bismuth leaving the anode would
be essentially pure bismuth, which could then be used as the cathode
from which the metal would flow to the top of the extraction column.

The salt stream flowing through the cell would serve as the cell elec-
trolyte. Since molten salts of interest are not good conductors, the

anode and cathode would have to be located relatively close to each
59

other in order to 1limit the rate of heat generation in the electrolyte.
If the extracted materials were selectively oxidized from the bismuth
stream at the anode, the current density obtainable at the anode surface
would be limited by the rate at which the extracted materials were trans-

ported to the anode.

Since the extracted materials are present at low concentrations, a
low ancde current density would be obtained; this would require the use
of a large anode area. This problem can be avoided by oxidizing bismuth
at the -anode and thereby generating BiF,, which would be soluble in the

3

salt. The BiFg-ceuld then be used to oxidize extracted materials from

the bismuth before it enters the electrolytic cell.

A second constraint on cell operation arises because of the rela-
tively low solubility of thorium in the bismuth. High cathode current
densities can cause thorium concentrations at the cathode surface to be
higher than the solubility of thorium in bismuth; this would result in

the formation of solid thorium metal.

7.1 General Mathematical Model

The geometry chosen for a simplified analysis of a molten-salt
electrolytic cell is shown im Fig. 27. In this figure, the bismuth anode
is on the left and the bismuth cathode is on the right. The anode
and the cathode are separated by a molten-salt mixture containing LiF,
Ber, BiF3,
enters the salt phase as BiF

Bismuth is oxidized at the anode and

h4+, and Bi3+

and possibly ThF4;
;- At the cathode, Li', T
are reduced to metals and enter the metal phase. Under some (undesir-
able) conditions, Be2+ can also be reduced at the cathode. Both the
szlt and the metal forming the electrcdes flow through the cell, and
all concentration changes are assumed to occur across a liquid film
that is present on both sides of the interface. Since the anode is a

single component {(i.e., bismuth), no concentration gradient is assumed

to exist in the anode metal.
60

ORNL DWG 717108

Bi
ANODE

B8i

SALT CATHODE

Csia Cgis

Cis Cksi[Crmi Cxm

CATHODE SALT FiLM
CATHODE METAL FILM

%’. ANODE SALT FILM
>
“n

Adg

L

O
©
n

 

 

 

 

AD, AP

Fig, 27. Schematic Representation of Model Used for Prediction of
Mass Transfer Rates in Electrolytic Cells.
61

Decreases in potential may occur across the anode surface (A¢A)’
the anode salt film (A¢AF), the bulk salt (A¢S), the cathode salt film
(A¢CF), and the cathode interface (A¢C). A description of each of
these regions and of the cathode metal film, along with equations
defining potential and concentration changes,;is given in the sections
that follow. Simplifying assumptions that will allow the problem to be

attacked at various levels of difficulty are also presented.

Tt should be noted that, in the salt regions, an electrical charge
must be assumed for each ion or speciesg. Lithium fluoride i1s probably
a strongly ionized salt, yielding only Li+ and F ions in solution.
However, thorium and beryllium appear to form fluoride complexes. The
behavior of bismuth fluoride is not well understood; thus the fluoride
complex could have a charge between 0 and 3+ in the salt phase. 1In the
following sections, transport in the salt regions will be described
assuming that, though LiF is completely ionized, BeFZ and ThF4 are com-
plexed and BiF3 is either ionized or complexed. The proposed models can
be applied only if these cations form single fluoride complexes; that is,
equilibria between several complexes may not be assumed. For example,
one can assume that Be, Th, and Bi are either completely ionized or com-—
pletely complexed; however, the assumption that these metals are partially

complexed to form several species is not possible.

Anode Interface. - Bismuth is oxidized to the trivalent state at the

 

anode interface. If bismuth fluoride is completely ionized, there will be
ne net flux of fluoride ions across the anode salt film; in this case, the
decrease in potential across the interface (AQA) can be approximated by the

potential at equilibrium:

 

_ RT BiA
A¢A = EBi + — In . (24)

Bi aBi sat
62

where

EBi = reduction potential of BiF3 (in volts),

Vpi T valence of bismuth (=-3),
F = Faraday constant,
R = gas constant,
T = temperature (absolute),

apin ~ activity of Bizi at the anode interface,

321 sar - activity of Bi~ in salt saturated with BiFS.

Replacement of the activity, a, with the product of mole fraction and

activity coefficient, vC, in Eq. (24) yields:

RT Ypia CBia

 

 

. C_.
Bi sat Bi sat

where
‘o . 3+ ,
Ypia = activity coefficient of Bi at the anode interface,
. . 3+, , -
YRi sat activity coefficient of Bi in salt saturated with 8153,
. .3+ . 8
CBiA = mole fraction of Bi at the anode interface,
C_. = golubility of Bi¥_ 1in salt, mole fraction.
Bi sat 3

If we assume that the activity coefficients are constant and note that

the solubility of BiF, in salt is a constant, we can rearrange this

3
equation as follows:

A
RT ] BiA RT
Ad, = EB' L In + T In C,., .

/ BiA
* Bi wBi sat CBi sat Bi *

 

 

Combining the first two terms on the right-hand side yields:

L RT
= | 25
Boy = Byy F Vi F In Cpine (25)

 
63

where
tf

EBi = constant defined above.

If BiF3 is not ionized, there will be a net flux of fluoride ions
at the anode salt film and the concentration of fluoride ions at the
anode surface will be dependent on current density and anode film thick-
ness. In this case, the decreasé in potential across the anode can be

written approximately as:

c
o RT BiA
b, = Eg, vy F In Cppl (26)

where CFA is the fluoride concentration at the anode surface and n is

the number of fluorine atoms in the complex.
Anode Salt Film. -~ If bismuth fluoride is ionized, the only ion
+
transferring across the anode salt film is Bi3 . This transfer results

from both concentration and potential gradients, as described by the

Nernst-~Planck equation:

3, = - ac +...%le V4 (27)

where

i

flux of ion k across anode salt film,

i

diffusivity of ion k in salt film,

o o o

e w oW m

concentration of ion k din salt film,

i

valence of ion k in salt film,

v

#

gradient of potential in salt f£ilm,

In determining the flux of a given component, one must actually solve

simultanecusly the set of equations consisting of expressions of the
64

form of Eq. (27) for all ionized species in the salt along with the
following relation, which states that at any point in the film the net

electric charge is negligible:

kack = 0, (28)

. . e 3+
If bismuth fluoride is ionized, the flux of only one species, Bi3 .

is nonzero.

Thus,
Jgi = T/vgeo
where
. 3+ .
JBi = flux of Bi across anode salt film,
I = current density in salt film,
Vi © valence of Bi in salt (= 3).

1f bismuth fluoride is mot ionized, the current across the anode

salt film is carried by fluoride ions. Thus,

where

flux of F across salt film,

4
=
# if

current density in salt film, equivalents per unit time per

unit area.

Also, for this case, Egs. (27) and (28) must be solved simultanecusly.
However, bismuth fluoride can be treated separately since it is an un-

charged component at low concentration. Thus,

C,.., = C_..
JB° : I/vB, _ DBi BiA BiB y (29)
* * CAF
65

where
JBi = flux of BiF3 across anode salt film,
I = current density in salt film,
DBi = diffusivity of BiF3 in salt film,
CBiA = concentration of BiF3 in salt at anode interface, mole fraction,
CBiB = concentration of BiF3 in bulk salt, mole fraction,
CAF = thickness of anode salt film.

Bulk Salt. — All concentrations within the bulk salt are assumed to
be uniform; however, a decrease in a potential will occur. This decrease

is assumed to be simply

Apg = IL/R, (30)

where
A¢S = decrease in potential across bulk salt,

1

L.

il

current density in bulk salt,

it

distance between electrodes,

R = resistivity of bulk salt.

Cathode Salt Film. — Transport across the cathode salt film is much

 

like transport across the anode film, except that the current is carried
by all reducible ions. If bismuth fluoride is partially ionized, or if
thorium and beryllium are complexed, there will be a flux of fluoride
ions from the cathode surface as well as a flux of reducible ions toward
the cathode. The flux of each ion is defined by Eq. (27). These trans-
port equations must be solved simultaneously with the condition of elec-
trical neutrality [Eq. (28)] and an equation expressing overall conser-

vation of charge,

zvkjk = 1, (31)

where

1}

valence of ion k,

L
i

flux of ion k.
66

Furthermore, conservation of each component requires that each Jk be

constant at all points within the film.

If bismuth fluoride is not ionized, its transport can be described

by the relation

JBI - DBi B1B5 BiC ’
CF
where

JBi = flux of BiF3 in cathode salt film,

DBi = diffusivity of BiF3 in cathode salt film,

C,., = concentration of BiF_ in bulk salt,

BiB 3
CBiC = concentration of BiF3 in salt at cathode interface,
SCF = thickness of cathode salt film.

If one assumes that the concentration of BiF3 in the salt at the cathode

interface is negligible, the above relation becomes

J = D C /6

Bi Bi "BiB (32)

CF*

In most cases, beryllium can be treated as if it were not reduced
at the cathode. When the potential drop across the cathode becomes
sufficiently large, beryllium will be veduced; however, it will not go
into solution in the biswmuth. Since this is probably an undesirable
situation, beryllium is treated as if it is not reduced. When the
decrease in poteutial at the cathode surface is calculated, one must note
whether beryllium would be reduced since we are interested in cell opera-

tion without beryllium reduction.

Cathode Surface. — A decrease in potential sufficient to reduce

 

lithium and thorium into bismuth will occur at the cathode surface. If

the decrease in potential at the cathode is sufficiently large, beryllium
67

will also be reduced to pure metal. The decrease in potential will be
a function of the lithium and thorium concentrations on both sides of

the salt-metal interface, C, , and C . .. If equilibrium exists at the
ksi kM.

cathode surface, the decrease in potential will be given by the relation

_RT KM{
boo =57 Img B
Ik ksi

(33)

where
A¢C = decrease in potential at the cathode surface,
v, = valence of ion k,
CkMi = concentration of component k in the metal at the cathode
surface,
Cksi = concentration of ion k in the salt at the cathode surface,
Ek = reduction potential for ion k required to transport component
k between salt and metal phases having the same concentration

of component k.

This equation, similar to Eq. (25), must be applied to both lithium and

thorium if both materials are reduced.

Cathode Metal Film. — A decrease in the concentration of all reduc-
ible components occurs across the metal film in the cathode. The dif-

fusion of lithium and thorium across the film is described by the rela-

 

tion
5 =D i~ S
C
\ kM1
o Dk.M % (34)
where

For the

lithium

68

= flux of component k in cathode metal film,

= diffusivity of component k in cathode metal film,

= concentration of component k in metal at cathode surface,
= concentration of component k in bulk metal in cathode,

= thickness of cathode metal film.

present case, we will assume that the concentrations of both

and thorium in the bulk metal are negligible. Then the above

relation becomes

J (35)

k= Pum Cemi/Sem

7.2 Assumptions and Simplifications

The model (as stated above) involves the following simplifications:

1.

3]

The geometry of the model is less complicated than that of

a real system.

A rather simple stagnant film theory is used. Actually, flow
near the interface causes the "fictive'" stagnant film thick-
ness to be a function of the diffusion coefficient or Schmidt

number.

The activity coefficients in both metal and salt are assumed
to be constant. This assumption is most important when the
decreases in potential drops across the salt-metal interfaces

are calculated.
Each important metal is assumed to form only one ion.

Changes in potential across interfaces are approximated from

equilibrium conditions.
69

These simplifications, as well as many.of those listed below, are
probably not important when we remember that many of the constants in
the model (especially diffusion coefficients) are not known and, therefore,
have been estimated. We believe that the model will gualitatively describe
the cell behavior and that the particular values assumed for the constants

will not affect the general behavior of the system.

In determining the solution to the set of equations developed ear-
lier, several constraints must be taken into account. First, the flux
of each component in the various regions and across the boundaries of
the regions must be constant. Also, the change in concentration across
boundaries must have prescribed wvalues; hence, a trial-and-error approach
must be used. Thorium saturation in the metal cathode at the salt-metal
interface must be checked. As with beryllium reduction, this will not
affect the nature of the zalculations, although it does place a limit on
the operating conditions if the production of thorium metal is to be

avoided.

To make the calculations practical, several additional simplifica-
tions will be made to provide models with varying degrees of complexity.
In general, we will look at the simplest models first. Then, after com-
parisons are made with experimental data, the more elaborate models can

be developed if required.

7.3 C(Cathode Current Efficiencies and Maximum Current Densities

with L}'.F*-BEFZ-"B]';FB Mixtures

The cathode region of an electrolytic cell containing a mixture of
LiF-BeF,~Bil

2 3
chogsen for the bulk salt were approximately those expected in the elec~

wae the first specific region studied. The compositions

trolytic cell for the rare-earth removal system. Most of the important
transport parameters for the system, such as diffusion coefficients

and the effective film (boundary layer) thickness, were not known.
Nevertheless, some important behavioral aspects of the system can be

predicted by using estimates for these parameters.
70

The values for the current efficiency and the maximum allowable
cathode current density are of greatest interest. The purpose of the
cathode portion of this cell is to reduce lithium into the bismuth
metal, Thus, any bismuth fluoride that is reduced at the cathode de-
creases the current efficiency based on LiF reduction. A maximum
allowable current density at the cathode is imposed if one does not
allow BeF2 to be reduced. Beryllium metal is insoluble in bismuth,
and the introduction of this third phase is considered undesirable.

Reduction of beryllium will not occur at low current densities and low

bismuth reductant loadings.

A BiF3—LiF~BeF2 salt composition of 20.8-62.5-16.7 mole % was

chosen as typical for the rare-earth cell. The diffusion coefficients
for lithium and bismuth in the salt were assumed to be 10“5 cm2/sec and
10‘“6 cmz/sec, respectively. These values were estimated from measure-
ments made with chloride salts. The molar volume of bismuth fluoride
was assumed to be 0.04 liter/g-mole.

The bismuth species which will be present in the salt are not

known. Two species were considered: BiB+ and BiFSZM. Equations (20),

(21), and (25) were integrated numerically for the four ions involved
+ 2o 3+ 2~

(Li , BeF4 , I sand either Bi3 or BiFSZ ). Calculated current den~

sivies were lower when B13+ rather than BiF5 " was assumed to be the
bismith species because the electric potential gradient across the
cathode salt film hinders the transport of negative ions and enhances
the transport of positive ions. The pradicted current efficiencies
with Big+ are shown in Fig. 28 for salt and metal film thicknesses

of 0.05 em (stagnant) each. With current densities of 0.1 amp/cm2 or
greater, the current efficiency is predicted ta be approximately 50%

or greater and would become 907 at 0.5 amp/cmz.

The conditions leading to the reduction of BeF2 were also estimated;

2 L
they are illustrated in Fig. 28. The point at 0.65 amp/cm” indicates the

current density and current efficiency at which the sali at the cathode
71

ORNL ~-DWG ©69-12633

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100 e
e
‘_// '
/
50 _ ///r
S
>
2 D, =103 c¢m?
Z 20 _ Li = cm</sec
L -
S Dg; =107° cm®/sec
W vg; = 0.04 liter/mole
u
b S =0.05 cm
>
= 10 ]
o
@
)
O
20.8 mole % BiFs
5
62.5 mole % LiF
16.7 mole % BeF2
2
0.05 0.1 0.2 0.5 1.0

CURRENT DENSITY {(amp/cm?)

Fig. 28. Calculated Current Efficiency with an L:i.l“wl?»er-~B:LF3 Salt.
72

interface becomes (under the simplifications of these calculations) pure

BeF Beryllium will certainly be reduced with current densities higher

9
than this. The point at 0.16 amp/cm2 represents the current density at
which the reduction of beryllium is more likely to begin; this value was
estimated by assuming that the salt and metal are in equilibrium at the
cathode interface and that the diffusion coefficients and boundary-layer

thicknesses are the same in the metal and the salt phases.

Experimental measurements with static electrolytic cells have in-
dicated that considerably higher current densities (5 amp/cmz) can be
achieved without reduction of beryllium. Some solid material has been
observed in the static cells, but this is not believed to have resulted
from reduction of beryllium. However, one should keep the limitations
of these calculations in mind. Although all of the assumed parameters
will affect the results, the most important (and probably the most un-
certain) parameter is the filwm thickness since this value determioes
the horizontal scale on Fig. 28. Thus, if a smaller film thickness is
assumed, a higher allowable current density will be estimated. The
general shape of the curve in Fig. 28 is probably more reliable than
particular numerical values, especially estimates of current densities.
At present, no experimental data on current efficiency are available.
When data on current efficiency become available, Fig. 28 may be used
to determine how efficiency will vary with current density and when

reduction of beryllium is likely to occur.

8. AXTAL MIXING IN PACKED COLUMNS WITH HIGH-~-DENSITY FLUIDS

J. S. Watson .. BE. McNeege

Axial diffusion may be important in determining the performance of
reductive extraction contactors in flowsheets proposed for processing a
molten-salt breeder reactor (MSBR). This is especially important in the

proposed rare-earth removal system, in which high wetal-to-salt vol.-
73

umetric flow ratios are required. It is likely that packed-column
contactors will be used and that salt will be the continuous phase.
Under these conditions, the salt will flow through the column with a
low velocity, and axial mixing is likely to severely decrease the per-
formance of the column. Data on axial dispersion are needed in order

to estimate the importance of this effect.

Relatively few data on axial dispersion coefficients are available;
no data are available for high~density, non-wetting fluids. To supply
the needed information, a study of axial diffusion coefficients has

been initiated using mercury and water to simulate bismuth and salt.

8.1 Mathematical Model

A steady-state technique was used for the axial diffusivity measure-
ments; its theoretical development is as follows. Comnsider a column of
constant cross section in which a fluid moves with constant superficial
velocity, v, If a . tracer material is introduced near the column exit,
the tracer will tend to diffuse upstream, and a concentration profile
will be established. At steady state, the flux of the tracer due to

axial diffusion is equal to the convective flux; that is,

ac _ . 5
-B a7 = wC, (36)

where
E = axial diffusivity,
C = tracer concentration at positicn Z,
v = superficial fluid velccity,
Z = position along the column.
74

Integration of this relation, assuming that the concentration of tracer

at point Z. is C

1 1° yields the relation

c. _ N .
Ing-=-2 (2 -2), (37)

which indicates that a semilogarithmic plot of C/Cl vs Z should yield

a straight line of slope -vy/E.

8.2 Experimental Equipment

The experimental equipment for the study is shown schematically in
Fig. 29. While the packed column is operating with mercury and water in
countercurrent flow, a tracer solution of cupric nitrate is introduced
near the top of the column at a constant rate. The steady-state concen-
tration pfofile of tracer is then measured photometrically at several
points upstream (i.e., down the column) in the aqueous phase. At each
measuring point, a small stream of water (approximately 1 ml/min) is
removed from the column by a small centrifugal pump and circulated
through a photocell and returned to the column. Only three sample loops
are shown in Fig. 29; however, as many as ten loops have been used in

the actual systen,

A photograph of a photocell is in Fig. 30. The cells are made of
Lucite. Light from a small incandescent lamp (shown on the left) passes
through the cell to a Clairex Model CL707L photoresistor (shown on the
right). No filters are used to restrict the spectrum of light passing
through the cell or into the photoresistor; however, the sensitivity of
the photoresistors reaches a maximum at a satisfactory wavelength for

cupric nitrate.

A typical circulation pump is shown in Fig. 31. This pump is also

made of Lucite; however, the inlet and ocutlet parts are made of stainless
75

ORNL DWG 69-10321

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

WATER Hg
ouUT IN
<— TRACER IN
. PUM_F’ g“\‘\\ |
PHOTOCELL o= ] =
, 5 R
- PUMP T~
PHOTOCELL S -
- _
PUMP ~<|
PHOTOCELL ~— | -
WATER Hg
IN ouT

Fig. 29. Schematic of Axial Diffusion Column.
76

. PHOTO 97025

 

Fig. 30. Photocell Used in Measurements of Axial Dispersion in
Packed Columns.
dd

PHOTO 97027

 

Fig. 31. Circulation Pump Used for Measurements of Axial Dispersion
in Packed Columns.
78

steel for extra strength. The impeller is a Teflon-coated magnetic

stirrer driven by a rotating magnet located outside the pump.

The assembled column is shown in Fig. 32. The circulation pumps
are mounted on the rack shown on the right. The driving magnets are
rotated by a single motor. The column, made of Lucite, has an inside

diameter of 2 in. and a height of 4 ft.

8.3 Results

Typical data from the system are shown in Fig. 33, where the log-
arithm of the tracer concentration is plotted as a function of distance
down the column (Z) from the tracer inlet. The resulting curve, obtained
by using 3/8-in. Raschig rings, a mercury flow rate of 29.1 ft/hr (super-
ficial velocity), and a water flow rate of 9.1 ft/hr, is essentially
linear. This indicates that the axial diffusion is independent of the
distance down the column and suggests that the extent of mercury disper-
sion has reached a steady-state condition in this section of the column.
The dispersion coefficient measured in this experiment was 3.4 cmz/sec.
No significant variation in this coefficient was detected when several
other mercury and water flow rates were used; the average value was
approximately 3.6 cmz/sec. However, these flow rates did not include
rates close to flooding, where changes in flow patterns are known to

occur.

9. REMOVAL OF HEAT FROM PACKED-COLUMN CONTACTORS
USED FOR ISOLATING PROTACTINIUM

J. S. Watson L. E. McNeese

Packed-column contactors are being considered for use in reductive
extraction processing systems. Heat transfer and temperature distribu-

tion within these columns are important since significant quantities of
79

PHOTO 97029

 

Fig. 32. Experimental Equipment Used for Study of Axial Dispersion
in Packed Columns.
80

ORNL DWG 69-10320
I ! | !

 

29.1 ft/hr MERCURY
20 ~ 9.1 ft/hr WtTER 7

c3.4 £M
E= 3.4 sec

1.0
09 -
08 -

0.7
0.6

05 -

CREF

04 |-

0.3 |~

0.2

 

 

 

 

0 10 20 30 40 50

Z,in,

Fig. 33. Axial Diffusion Measurements Made with 3/8-in. Raschig
Rings.
81

heat will be generated as a result of radiocactive decay. 1In the remain-
der of this section, estimates are made of temperatures in columns used
for isolating protactinium under various operating conditions.

The generation of heat in the protactinium isolation system results
primarily from the decay of 233Pa, although significant quantities of heat
are produced by the decay of fission products. Under normal operating con-
ditions, the concentration of protactinium in the salt reaches a maximum
value midway up the column; at this point, the heat resulting from the
decay of both fission products and protactinium is generated at the rate
of about 37 kw per cubic foot of salt. As the salt flows through the col-
umn, the rate at which heat is generated from the decay of fission preducts
in the salt decreases from an inlet value of 16 kw per cubic foot of salt to
an exit value of about 4 kw per cubic foot of salt. A significant quantity
of salt having a high protactinium concentration will be held up in the
protactinium decay tank; the total generation rate in this salt is about

29 kw per cubic foot of salt.

The most difficult heat-transfer situation likely to be encountered
is believed to be represented by conditions in which the extraction column
contains salt having a heat generation rate equal to that in the decay
tank (29 kw/ft3}. Two simplifiea heat-transfer situations related te this
condition were examined. In the first, it was assumed that no heat was
removed through the walls of the column, and the resulting exit salt and
metal temperatures were calculated. In the second case, 1t was assumed
that all of the heat was removed through the walls of the column, and the
difference between the temperature of the column wall and the temperature

of the center line of the column was calculated.

9.1 Maximum Column Temperatures, Assuming Only Convection of Heat

The rate at which salt flows through the column will be 7.1 x 105

g-moles/day, or 525 g/sec. The heat capacity of salt is 0.33 cal g~1

- 3
°C 1.10 The metal flow rate is 1.4 x 106 g~moles/day, or 3.39 x 10
82

g/sec. The heat capacity of liquid bismuth is 0.035 cal gml °CM1_ 1t

will be assumed that the extraction column is 8 in. in diametrer and 20

ft long. If packed with Raschig rings, it would have a void fraction of
approximately 0.75 and the total heat generation rate would be 3.08 x lO4
cal/sec. 1f there is no heat loss through the wall and if the salt and
metal streams enter the column at the same temperature, the temperature
increase during their passage through the columns will be 106°C. If
metal alone was flowing through the column, the increase in temperature

would amount to 260°C; if salt alone was flowing through the column, the

increase would be 178°C.

9.2 Maximum Column Temperatures, Assuming Only Heat Removal
Through the Column Wall

Heat can be transferred radially in the column, and hence through the
column wall, by the combined effects of conduction and convection. No data
were found on effective thermal conductivities in packed beds during two-
phase countercurrent flow. In the absence of such data, effective thermal
conductivities in packed beds with single-phase flow were used. We feel
that this approach probably gives a counservative estimate of the conduc-
tivity, that is, a value lower than would actually be observed. Data of
Bernard and Wilhelmll were considered most relevant because these authors
studied mass transfer (diffusion) through packed beds, a situation in which
no credit could be taken for conduction through the packing. Most data
related to conduction through packed beds were obtained by using gases for
the fluids; in such casesg, the conductivity of the packing may be important.

Bernard and Wilhelm reported that

N, = Dp U/E % 10, (38)
83

=
B

Peclet number,

D = packing diameter,

o
i

superficial fluid velocity,

il

effective diffusion coefficient (of heat or mass).

With 3/8-in. packing and a salt velocity of 62.7 ft/hr, the effective
diffusivity given by this relation is 0.20 ftzlhr, which yields an

effective thermal conductivity value of 7.75 Btu hr“l ft-—1 °F_l. The
difference between the temperature of the wall and that of the center
line of a cylinder of infinite length having a volumetric heat source

of uniform strength is:

_ L oad
Tc Twall ! (39)

where
= yolumetric heat generation rate,
d = diameter of the cylinder,
ke = gffective conductivity of the cylinder.

The calculated difference between the center-line and the wall temperatures
is 147°C.

9.3 Conclusion

Since heat will be removed from the column by conduction through
the wall, as well as by the materials flowing through the column, the
maximum temperature rise in the colummns should be less than 150°C if
salt from the decay tank fills the reductive extraction column. However,
in order to hold the temperature increases to this value, it will be nec-
essary to maintain at least the normal metal flow rate. Actually, the
increase in temperature will be tolerable if the normal flow rate of

either one of the phaseg is maintained.
84

10. MSRE DISTILLATION EXPERIMENT

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

Equipment for the MSRE Distillation Experiment and its installation
at the MSRE site have been described earlier.9 This section summarizes
operation of the equipment, describes the data obtained, and presents

the results of some of the performance calculations.

10.1 Summary of Distillation Operation

The experiment to demonstrate the high-temperature, low-pressure
distillation of irradiated MSRE fuel carrier salt has been successfully
completed. Although the quantity of salt processed was only 12 liters
(instead of the anticipated 48 liters), all objectives of the experiment

were met in an uneventful 31-hr operation.

On May 1, 1969, we attempted to transfer 48 liters of irradiated
MSRE fuel salt (containing no uranium) from the fuel storage tank (FST)
to the still feed tank. The transfer was initiated by evacuating the
feed tank, which contained about 2 liters of uunirradiated salt, to about
1.5 psia. After 12 liters of salt had been transferred from FST, the
pressure in the feed. tank rose quickly (over a period of 2 min) from
about 4 psig to atmospheric pressure, which indicated that gas was flow-
ing from the FST to the feed tank. Although the bubbler in the FST
indicated that more salt was present, we made only one additional attempt
to transfer salt because of the potential danger of removing the salt
from the freeze valve between the two vessels. After it was established
that a tight seal could be made in the Ireeze valve, we decided to pro-

ceed with the experiment using the salt already transferred.

At the start of the distillation, 7 liters of salt was transferred
from the feed tank to the still pot. The still pot was heated to 900°C,

and distillation was initiated by slowly decreasing the condenser pressure
85

to 0.2 mm Hg. The remaining salt in the feed tank was fed to the still
pot with the control system maintaining a near-constant salt level in
the still pot. During this part of the run, the still-pot temperature
was slowly increased since the concentrations of volatile BeF2 and ZrF4
were still fairly high; also, it appeared that the capacity of the
condenser would be exceeded since the temperature at the end of the
condenser was becoming abnormally high. We reduced the distillation
rate to a point where the condenser temperature remained at a constant

and acceptable level (about 700°C) by increasing the pressure of the

condenser to 0.8 mm Hg.

When the remaining salt in the feed tank (7 liters) had been dis-
tilled, salt in the feed line to the still pot was frozen; then 4 of the 7
liters of salt in the still pot was distilled by batch distillation. . The
still-pot temperature during this part of the run was 980°C; the condenser

pressure was about 0.1 mm Hg.

During the first part of the experiment, the average distillation
rate was 0.71 liter/hr; during the last part, it was about 0.30 liter/hr.
Eleven condensate samples were taken at approximately 90-min intervals
during the run. When these samples were removed at the end of the experi-
ment, the radiation readings ranged from 4 r/hr at contact for the first
sample to 500 mr/hr at contact for the last sample. The predominant

activity in all samples was 137Cs

-

10.2 Summary of Available Experimental Data

The following quantities were measured during this experiment: the
concentration of each component in the feed salt to the still and in the
cendensate (several times), the volume of liquid fed to the still pot, and
the volume of liquid collected in the receiver. All concentrations were
calculated as of May 7, 1969. The volumes of salt in the feed tank, still

pot, and condensate receiver were determined by measuring the weight of
86

liquid over the end of a bubbler, dividing by a salt density of 2.2
g/cm3 to obtain the depth of liquid, and multiplying this value by the
cross~sectional area of the particular vessel to obtain the volume. We
assumed that the mass density of the liquid was independent of composi-
tion and that the volume of liquid in the still pot could be calculated
by subtracting the volume of condensate collected from the total volume
of salt fed to the still from the feed tank. Assuming molar volumes
can be added (which is a fairly good assumption for fluoride salts),

we calculated densities for liquids in the concentration range seen in
this experiment and found only a 57 variation in mass density; hence,
our assumption of constant density is acceptable. Table 3 summarizes
the data collected during the distillation. Complete analyses for all
fission products have not been received; thus analyses are given only

for l440e and 95Zr.

10.3 Material Balance Calculations

One of the most concise ways to express the separation performance
of the distillation equipment is to convert the condensate analyses to
effective relative volatilities with respect to LiF. The effective
relative volatility of component i with respect to LiF is defined as

X,
y1/_1

= et (40)

Of, [ 1 3
i-LiF yL]’.F/

*LiF

where

i

mole fraction in the condensate,

mole fraction in the still pot.

Although the composition of each component in the still pot was not meas-

ured directly during the run, the data allow the composition of the still
Table 3. Summary of Data from Distillation Experiment

 

 

 

Volume of
Salt Fed Volume of
from Condensate Condensate Analvses (mole fraction)
Feed Tank Collected 95 144
{(liters) (liters) LiF BeF2 ZrF4 7r Ce
Start of
run 7.75 0 - - — - —
FDE-1 7.9 0.42 0.284  0.629  0.0873  2.03 x 100 < 1.80 x 1070
FDE-2 10.0 1.93 0.385  0.539  0.0756 2.07 x 10°° 2.03 x 1078
FDE—3 11.2 2.96 0.439  0.512  0.0489 1.31 % 1078 1.14 x 1078
FDE-4 12.6 3,81 0.466 0.484 0.0494 1.36 x 1070 2.49 x 107°
FDE-5 13.4 4.35 0.489  0.463  0.0479 1.35 x 1078 -
FDE-6 13.6 4.95 0.455  0.494  0.0515 1.43 x 1072 1.25 x 1072
Start of batch
distillation 13.8 5.07 - - - - -
FDE-7 13.8 5.80 0.497 0.452 0.0509 1.35 x 107° 9.21 x 1077
FDE-8 13.8 6.34 0.546  0.407  0.0466 1.33 x 1078 1.41 x 1078
FDE~9 13.8 6.89 0.599  0.361  0.0405 1.09 x 107° 2.32 % 1070
FDE-10 13.8 7.49 0.740  0.224  0.0357 9.83 x 1077 1.79 x 1070
FDE-11 13.8 7.85 0.741  0.224  0.0344 9.47 x 10~° 2.51 x 1072
Feed _8 -
composition -~ -~ 0.663 G.273  0.0637 1.24 x 10 1.16 x 10

 

L8
88

pot to be estimated from a material balance for each component. In the
remainder of this section, we derive the material balance equations and

outline the calculational procedure,

For the semicontinuous mode of operation, a differential mole bal-

ance for component i1 gives:

£ - = dM,,
£.dI - y.do i (41)

where
J = total moles fed to the still pot,

O = total moles removed from the still pot,

Mi = moles of component i in the still pot,
fi = mole fraction of component i in the feed,
vy = mole fraction of component i in the condensate.

Since volume was the measured quantity, we can make the following sub-

stitutions:

dVi
dI = — i ,
£V,
=1 J 1
("2
dv
out
do = N ,
v,V
j=1 ]
where
vin = yolume of salt fed to the still pot, liters,
vout = volume of condensate collected, liters,
vy o= molar volume of component j, liters/mole.

 

- .
Assumed to be independent of liquid concentration.
89

When we substitute the quantities in Eq. (42) into Eq. (41), we obtain

 

£, v,
— JY, | = e ¥ = dM (43)
N in N out i’
| f.v Y,V
Integration of Eq. (43) yields
£ Vout
Ly, - S N (44)
N in N out i’
- fLv, ‘ YV,
j=1 11 o =1 31
where the conditions Vin = Vout = Mi = (0 at the start of the experiment

were used. The composition in the still pot can be determined by solving
for M, for each component. This equation is valid up to the point where

Vout = 5,07 liters (the end of the semicontinuous distillation).

During the batch distillation, if Mi moles of component i are present
in the still pot and dMi moles of component i are vaporized, the mole

fraction of component i in the vapor is given by

v, = 5 (45)

V., = W . (46)
90

From Eqs. (45) and (46), we obtain the expression

dMi de
i3 (47
or
Vi
dM, = — dM, . (48)
i 3 j

Assuming that the molar volumes v, are independent of composition (as
we did before), we multiply both sides of Eq. (48) by v and sum both

sides over all i to obtain the following equation:

| 3
Vit Vg T Oy Ty (49)

where VS = volume of salt in the still pot, liters.

till

Solving Eq. (49) for de yields:

Y3
s e . 0
de N dvstill (50)
. yvi
o
Integrating Eq. (50) yields:
still~y
] )
M= M+ -—~—J-———"N av_, .11 (51)
y. vV,
v ii
91

where
Mjo = moles of component j present in the still pot at the start
of the batch distillation,
VO = volume of salt in still pot at start of batch distillation.

The quantity Mjo in Eq. (51) is calculated by evaluating BEg. (44) using

the values of vin and Vou corresponding to the end of the semicontinuous

t
distillation. As previously, the composition of the still pot is deter-

mined by evaluating Eq. (51) for each component present,

The integrals in Eqs. (44) and (5]1) were evaluated by plotting

N .
yi/ ijj vs the volume of condensate collected in the receiver during
j=1 |

the semicontinuous distillation and yi/N y, vV, vs the still-pot volume

(calculated as explained in the previous section) during the batch dis-
tillation. The evaluation was made by fitting the data to simple empirical
functions of the appropriate values and then integrating these equations.
For any component, a fairly wide range of equation parameters fit the
scattered data equally well; however, the values of the integrals were not

significantly sensitive to these variations In the parameters. The sums
N N

v v , i
i=1 Yi'1 and j=1 1 i were adequately represented by considering only the

major salt components LiF, Ber,.and ZrF4 since the mole fractions of the

fission products were negligible as compared with the mole fractions of

these components.

Using Eqs. (44) and (51), we calculated the number of moles of each
component present in the still pot at the time each condensate sample
was taken. The mole fraction of each component in the still pot at that

time was calculated using the following equation:
92

The calculated X, values are summarized in Table 4. The measured values
of ' from Table 3 and the calculated values of x, were used in Eq. (40)
to calculate the effective velative volatility of each component, with

respect to LiF, for each sample.

10.4 Results

Calculations, as described in the previous section, have bheen

carried out for the major salt components LiF, BaFZ, and ZrF4, and for
95. 144 ;
the fission products 7r and ~ Ce. As shown in Fig. 34, the effective

relative volatility of BeF, remained constant at a value of 5.4 during

2
the semicontinuous distillation, but increased gradually to 8.4 during

the batch distillation. The value measured during the semicontinuous
portion of the run is near the value (4.7) that was measured in recir-

2
culating equilibrium stills in LiF-BeF systems.l The increase during

2
the batch distillation is in qualitative agreement with measurements

. B RaR
made in LiPF Be_2 ThF4

mole fraction of Ber in the liquid became very small, the relative

systems. These measurements showed that, as the

volatility of BeF2 increased.

The relative volatility of ZrF4 (both stable zirconium and fission
produci 95Zr), which was approximately 4 at the start of the experiment,
decreased, and then became constant at a value near 2 about halfway through
the semicontinuous mode of operation. It then decreased further during
the batch distillation to about 1. These values are consistent with

values measured under equilibrium conditions.

As shown in Fig. 35, the effective relative volatility of 144CeF3

had an initial value of about 3.5 x 10W2 and decreased steadily until it
reached a value of about 1 x 10—2 at the end of the experiment. These
values are about two orders of magnitude higher than the value measured
in small recirculating equilibrium stills in LiF—BeF2 systems. Although
an explanation of this discrepancy is not presently available, consistent
explanations may emerge as the analyses for the other fission products

become available.
Table 4. Calculated Still-Pot Compositions for the MSRE Distillation Experiment

 

 

 

Sample Concentration (mole fraction)
(ggé—) Li¥F Ber ZrF4 952r 1&40e
1 0.684 0.255 0.0617 1164 % 1070 1.22 x 1070
2 0.738 0.202 0.0593 1.039 x 1070 1.41 = 1070
3 0.766 0.175 0.0597 1.002 x 10°° 1.54 x 1070
4 0.785 0.162 0.0606 0.966 x 107° 1.61 = 1070
5 0.785 0.154 0.0612 0.996 x 1075 1.66 x 1070
6 0.800 0.138 0.0618 0.974 x 1070 1.76 = 1070
7 0.828 0.109 0.0626 0.93 x 10°° 1.91 x 1670
8 0.843 0.0877 0.0635 0.904 x 1070 2.07 x 1070
9 0.868 0.0667 0.0650 0.881 % 1070 2.19 % 10°°
10 0.888 0.0444 0.0675 0.862 x 1070 2,46 % 1070
-8 6

11 0.898 0.0323 0.06%906 0.855 = 10

b
Ln
e
"

1,..{
[

 

£6
94

ORNL — DWG 69— 12636

LATIVE VOLATILITY

C

Ti_Lirr R

BATCH
DISTIL.LLATION
O 1 2 3 4 5 o 7 8
CONDENSATE COLILECTED (liters)

 

Fig. 34, Effective Relative Volatilities of Major Salt Constituents
and 957r,
EFFECTIVE RELATIVE VOLATILIiTY OF “*Ce

95

ORNL DWG 717107

 

 

 

 

107 -
..
.
®
® - .
21 *
10 - . \.
- fi
}..
BATCH
DISTILLATION
lo“’# i | 1 A ! ] ]
| .2 3 4 5 & 7 8
LITERS CONDENSATE COI.__LECTED
144

Fig. 35. Effective Relative Volatility of Ce.
8@

9.

10.

11.

12.

13.

11. REFERENCES
MSR Program Semiann. Progr. Rept. Aug. 31, 1968, ORNL-4344,
pp. 68-70.

MSR Program Semiann. Progr. Rept. Feb. 28, 1969, ORNL-4396,
pp. 275~78.

 

MSR Program Semiann. Progr. Rept. Feb., 28, 1969, ORNL-4396,
pp. 77-79.

 

L. G. Alexander et al., Molten Salt Converter Reactor Design
Study and Power Cost Estimates for a 1000 MWe Station, ORNL-
TM-1060 (September 1965), pp. 209-20.

 

 

R. W. Stoughton and J. Halperin, Nucl. Sci. Eng. 6, 100-118
(1959).

 

MSR Program Semiann. Progr. Rept. Feb. 28, 1969, ORNL-4396,
pp. 284-85. '

MSR Program Semiann. Progr. Rept. Aug. 31, 1968, ORNL-4344,
pp. 115-41.

 

L. E. McNeese, Engineering Development Studies for Molten—Salt
Breeder Reactor Processing No. 1, ORNL-TM-3053 (November 1970).

 

 

L. E. McNeese, Engineering Development Studies for Molten~-Salt
Breeder Reactor Processing No. 2, ORNL-TM-3137 (February 1971).

 

 

S. Cantor, Physical Properties of Molten Salt Reactor Fuel,
Coolant, and Flush Salts, ORNL~TM-2316 (August 1968).

 

R. A. Bernard and B. H. Wilhelm, Chem. Eng. Progr. 46, 233-44
(1.950).

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

 

 

 

F. J. Smith, L. M. Ferris, and C. T. Thompson, Liquid-Vapor
Equilibria in LiF~-BeFo-ThF, Systems, ORNL-4415 (June 1969).

 
WL WO PN

et et
= O
- o

12,
13.
14,
15.
16,
17.
18,
19.
20,
21,
22.
23.
24,
25.
26,
27.
28.
29-39,
40.
41,

/8.
79.
80.
81.
82.
83-84.
85,
86.
87-88.

f / 97

INTERNAL DISTRIBUTION

C. F. Baes 42. E. L. Nicholson
H. F. Baumann 43, J. H. Pashley (K-25)
S. E. Beall 44. A. M. Perry
M. J. Bell 45-46. M. W. Rosenthal
R. E. Blanco 47. J. Roth
F. F. Blankenship 48. A. D. Ryon
G. E. Boyd 49. W. F. Scanaffer, Jr.
R. B. Briggs 50. Dunla~ Scott
R. E. Brooksbank 51. J. H. Shaffer
K. B. Brown 52. M. J. Skinner
W. L. Carter 53, F. J. Smith
H, D. Cochran, Jr. 54, Martha Stewart
F, L., Cullex 55. R. E. Thoma
J. R, Distefano 56. D. B. Trauger
W. P. Eatherly 57. Chia~Pao Tung
D. E. Ferguson 58. W. E. Unger
L. M. Ferris 59. C. D, Watson
J. H. Frye 60. J. S. Watson
W. R. Grimes 61. H. 0. Weeren
A. G. Grindell - 62, A. M. Weinberg
P. A. Haas 63. J. R. Weir
B. A. Hannaford 64. M. E. Whatley
P. N. Haubenreich 65. J. C. White
J. R. Hightower, Jr. 66. R. G. Wymer
C. W. Kee 67. E. L. Youngblood
R. B. Lindauer 68-69. Central Research Library
J. C. Mailen 70-71. Document Reference Section
H. E. McCoy 72-74. laboratory Records
L. E. McNeese 75. Laboratory Records, RC
. M. Moulton 76. Y-12 Document Reference Section
J. P. Nichols | 77. ORNL Patent Office
EXTERNAL DISTRIBUTION
D. F. Cope, Atomic Energy Commission, RDT Site Office (ORNL)
A, R. DeGrazia, USAEC, DRDT, Washington, D.C. 20545
D. FElias, RDT, USAEC, Washington, D.C. 20545
Norton Haberman, RDT, USAEC, Washington, D.C. 20545
Kermit Laughon, Atomic Energy Commission, RDT Site Office (ORNL)
T. W. McIntosh, Atomic Energy Commission, Washington, D.C. 20545
M, Shaw, Atomic Energy Commisgion, Washington, D.C. 20545
Laboratory and University Division, ORO
Divigion of Technical Information Extension, ORO