ORNL-TM-3344.txt

 

ORNL-TM~334L4

Contract No. W-ThO5-eng-26

Reactor Division

N AN
EXPERIENCE WITH SODIUM FLUOROBORATE CIRCULATION I
MSRE-SCALE FACILITY

A. N. Smith

NOTICE
This report was prepared as an account of work
Sponsored by the United States Government, Neither
the United States nor the United States Atomic Energy
Commission, nor any of their employees, nor any of
their contractors, subcontractors, or their employees,

makes any warranty, express or implied, or assumes any
legal liability or responsibility for the accuracy, com-
Pleteness or usefulness of any information, apparatus,
product or process disclosed, or represents that its use

S E PTEMBER -l 972 would not infringe privately owned rights,

 

NOTICE This document contains informa?ion oflafiizeiimifizry
nature and was prepared primarily f?r 1nt§rn: e e
Oak Ridge National Laboratory. It 1s subjec f? oot
correction and therefore does not represent a fina

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
for the
U.S. ATOMIC ENERGY COMMISSION

T T QGO TMENT 1 1IN BATTIY
PISTRISUTVIN OF TEIS DGOITMENT 16 TN N

A
CONTENTS

BB R A T v ettt et ittt e sttt rannnensoeeessoesoetasensoeenseenananes
1 TN RODU CT IO 44t e tseteetnneenneruonsenesnsesnesoenonnanssnnnes
2 RS OB B T I S vttt ettt ernrsreeoenoetnaesennonenseenanennes
3 SUMMARY OF TEST RESULS vttt erenreatoresoonnsennsonsoesenss
S N €1 T - T
3.2 Pumping Characteristics ....viiiiriiiienieeneeeeneeennns
3.3 Control of the Salt Composition ...vvevreriereernnenens
3.4 Controlled Deposition of Corrosion Products ...........

3.5 Analysis and Correction of Restrictions on the
Off-Gas System tiveretniieinentnensoeeroeassenssennss
3.6 Miscellaneous ObServabions ...veeeeeeeeeeeenennnseneenns
L. DESCRIPTION OF TEST FACILITY v vvvenreneenennenneneennennenes
Bl GEMETAL vttt ittt et e i it e,
4.2 Salt Piping and COmMPONENtS ..vveeeererenenrnneneannnnns
4.3 Salt Sampling DEVICE +vuuvvrvrrrernnreneeneeoreneensnns
4.4 Hesting and Controlled Ventilation .....evveverenenenn.
L5 Gas System Desigh . ive e ineineneineeneeeneenoneeenenns
L6 BFE; DiSPOSAL «uvuutunninnnniiniieiiannaansnanannnennns
h,7 Instrument and CONtrols . uvueeeseeeeerereeeeeeneeeeoeees

iii

5. TEST PROCEDURES, OBSERVATIONS, AND CONCLUSIONS FCR THE

PRINCIPAL TESTS

5.1
5.2

5.3
5.4

Pumping Characteristics ......ciciiieiiiiierninennnnncens

Control of the Salt Composition .......cevivieiieenann.
5.2.1 Operation of the BF, partial pressure

control sysStem ..iveteneeienenestoenranneannnas
5.2.2 Methods for monitoring salt composition ........

5.2.3 Evaluation of the thermal conductivity method ..
Corrosion Product Deposition ......cievivieirveinnncnns
Off-Gas System Flow Restrictions ......cevvvevensnvonsn
5.4.1 Plugging €XperienCe .....civeeenenceerneronansss
5.4.2 Off-ga8 tES8TS vvrirrrerirorernrstenossnnnneasnss
5.4.3 Discussion and evaluation of test results ......

5.4.4 General conclusions from off-gas tests .........

-------------------------------------------

Page

O V1oV o oy o

~N N 1 O &

=P
o 0 o D

19
19
26

31
36
40
48
53
54
56
71
77
iv

6. CHRONOLOGY, PERTINENT OBSERVATIONS, AND OTHER TESTS ......... 79

£.1 ChYONOLOZY vt eteeeuuaeseeonnsososssnanaansssssnonanans 79

6.2 Analyses Of Salt SampPles ..veeerrerorooreanoscescnncens 83

6.3 Pretreatment and Transfer of Salt Charges .e.oeeeeseesras Bl

6.4 Freeze Valve Operation ..veeereeereereceesooaearoeassnns 88

6.5 Rubbler Tube Operation ....eeeieeseereeeesosssasassosns 90

6.6 Experience with Salt Ievel Instruments ....evvoevessosns 93

6.7 Back Diffusion at the Pump Shaft .vevevrvrerrorionannns 95

6.8 A Salt Leak to the AtmMOSPHEIrE ..vvvveerennrnnnennieansns 96

6.9 Experience with Handling BF, «.cevvcvivrerienonnnnonnn. 98

£.10 Corrosion EXpPeri€rCe ....eeeeervereereerasseneonnsosonss 99

6.11 Green S8l vuui it initereeniareatieroieiaaaaeaa 103

6.12 Valves in BF, Service ............ciieiiiienianianiann, 105

7. RECOMMENDATIONS FOR FURTHER DEVELOPMENT WORK ...vvvvnvennonas 106

7.1 Tmpurities in the Salt ..uuieereeereeneneroosossnsrsons 106

7.2 Corrosion Product Deposition ...eeeereeenrcrasosnsonnos 106

T.3 Salt Level Instruments .....iviirtieeneritiencnecnanannaas 107

7.4 Off-CGas System Restrictions vveeeivereruveceoeneneannns 107

7.5 Control of Salt Composition ...veieivirreeeerteesonsnnns 107

7.6 TIntermixing of Molten Salts vvvevvvrrrnonnenennanncenan 108

7.7 Solid Phase Transition ....euvveverenineneronnnennonss .. 108

7.8 BF, Recycle System ....vveeiivneiiiuneiierrrnrennenon, 108

ACKNOWLEDGMEITDS v s st e st et s s s evnnennnesessssseananenossassesooans 109

REFERENCES 2 et e vt etes et aenenenenneneesenensessessarensensanenes 110

BIBLIOGRAPHY ettt st et i ssnoseoananssraassasasassossossassssses 110
APPENDTX A. SELECTED PHYSICAL ANWD CHEMICAT, PROPERTIES OF

PROCESS MATERIALS .. ittt i insn i vennsonensannnanans 113

APPENDTX B. REFERENCE DRAWINGS .+ ieiiierenrnerernanssnosnensnnos 121

APPENDIX C. MATERTAL SPECIFICATTIONS . vvenrrvnocoareanooanonnssns 122
APPENDIX D. DERIVATION OF EQUATION FOR CALCULATION OF BE,

PARTTAL PRESSURE .. iiititenenreoenrsnssoensnsoenns 123

APPENDIX E. FREEZE-THAW STRESS TEST ....v0vvirirnerroreennnennns 125
EXPERTENCE WITH SODIUM FLUOROBORATE CIRCULATION IN AN
MSRE-SCALE FACILITY

A. N. Smith

ABSTRACT

A eutectic mixture of sodium fluoroborate and sodium fluoride was
circulated isothermally at a rate of about 800 gpm for 11,567 hr in s
h-in.-IPS Inconel test loop as part of the program to evaluate the
fluoroborate salt for use as a secondary coolant for the Molten-Salt
Breeder Reactor. Except for brief periods at 900, 1150, and 1275°F,
the bulk salt tempersgture was controlled at 1025°F. The obJjective of
the experiment was to obtain general experience in the handling and
circulation of the fluoroborate salts, with emphasis on the pumping
characteristics and on the design and operation of the gas system as
it related to handling BFé and controlling the salt composition.

The test results indicated {1) that water test dats may be used
to predict the performance of molten-salt pumps with the fluoroborate
salt, (2) that reliable performance may be obtained from systems
handling BF, gas if precautions are taken to exclude water and water-
related impurities, (3) that control of the salt composition should
not be a problem, and (4) that the thermal conductivity of the gas
phase above the salt surface may be used as an indicator for monitoring
the salt composition. Preliminary tests were made to examine the fea-
sibility of protecting reactor heat transfer surfaces by preferential
deposition of corrosion products in a cold trap. Further work is
recommended in this area. The test work indicated that flow restric-
tions in the off-gas line can be eliminated by pretreatment of the
salt to remove volgtile impurities and by the use of a hot-mist trap
and a cold filter in the off-gas line at the pump bowl outlet.

Additional work is needed to improve our understanding of the ef-
fects of cross-mixing between the fluoroborste salt and the reactor
fuel salt and the nature and properties of the acid impurities in the
fluoroborate salt.

Key words: reactors, secondary coolants, sodium fluoroborate,
fused salts, boron trifluoride, coolant loops, Molten-Salt Reactor
Experiment, molten-salt pumps.
1. INTRODUCTION

As currently conceived, molten-salt breeder reactor (MSBR) systems
require the circulation of a secondary coolant salt to transfer the nu-
clear heat from the fuel salt to the steam generator in the power con-
version (Rankine cycle) system (see Fig. 1). A mixture of lithium and
beryllium fluorides was used as the secondary coolant in the Molten-Salt
Reactor Experiment (MSRE), and the performance of this salt indicates
that it is suitable for MSBR use. The main disadvantages of its use are
high cost (gbout $12/1b) and relatively high melting point (850°F).
Another material, a sodium fluoroborgte-sodium fluoride eutectic mixture
[NaBF, -NaF (92-8 mole $)], has evoked interest because it costs less
(about 4% of the cost of the Li-RBe salt) and because its melting point
(725°F) is low enough to minimize the probability of salt freezing in
the steam generators.l

An extensive program has been under way at ORNL to qualify the fluo-
rotorgte salt for use as the secondary coolant for MSBR service. In
addition to studies of basic physical properties, engineering properties,
and materials compatitility, the program originally called for testing
the fluoroborate salt mixture in the MSRE coolant system under reactor
operating conditions. To help us determine design and operational changes
that would be needed at the MSRE for the coolant test, it was decided to
make g preliminary test in an existing isothermal pump test stand, which
was capable of operating at the flow rate (850 gpm) and temperature (1000
to 1200°F) of the MSRE. This report describes this preliminary test.
Conceptual work started in June 1967; initial loop operation started in
March 1968, and the test work extended through June 1970. After the
preliminary test was under way, a program change resulted in the cancel-
lation of plars to use the fluoroborate salt in the MSRE coolant system.
Consequently, the work reported here represents the current total expe-
rience with the ecirculation of fluoroborate salt in an MSRE-scale fa-

cility.
VENT

 
 
 
 
    

-COOLING

PRESSURE —== COOLING

COOLANT
SALT PUMP

 

 

   

FROM CHEMICAL

 

ORNL -DW(G 68-4492

 

PROCESSING

      
 

 

REACTOR

-
o1
-

 

 

 

 

 

 

       
 

STEAM SUPERHEATERS

  

STEAM SUPERHEATERS

L

'
i

|
|
\

{
i
1
|
J

 

   
  

 

 

 

 

 

 

 

     
 
 
 
  

CATCH
BASIN®

TO CHEMICAL
PROCESSING

 

PRESSURE _
AND VENT
T <
_/ HEAT REJECT
@ STACK
FUEL
DRAIN TANK

TO CHEMICAL

FREEZE VALVE PROCESSING
PF) PROPORTIONAL FLOW VALVE
GAS SEPARATOR
—— FUEL #»% STEAM GENERATING UNIT NO. 2
+ ———— COOLANT xxx STEAM GENERATING UNIT NO. 3
———w STEAM »x#% STEAM GENERATING UNIT NO. 4

Fig. 1.

STEAM GENERATING UNIT NO.1

. PRESSURE
AND VENT

 

 

 

COOLANT SALT

DRAIN TANK
TEMP TEMP
POINT °F FLOW PCINT °F

(¥; 1000 200 f1¥sec as 850
(Zy 1300 200 9 850
3 1300 50 43 600
(& 1150 75 (%) 1000
(55 850 75 3 1000
(6 1150 75 a8 700
&) 1150 0-20 ) 950
& 1150 16.2 8] 1250
25 850 16.2 (i9) 600
g 1150 10 &0 100

Flow diagram for 2000-MW(e) station.

TS
T2

FLOW

75 H3/sec

10

2.56 x 10% 1b/hr
2561 'nOS

1.26x 10°

1 26x10°%

4 ft¥sec

a4
420,000 ft ¥min
420,000
2. TEST OBJECTIVES

When the fluoroborate* salt is heated above its melting point, it
dissociates in accordance with the equation NaBF, < NaF + BF; . The de-
gree of dissociation and the resulting partial pressure of BF; over the
melt is a function of the temperature (see Appendix A, Sect. A.2). At
1025°F the equilibrium BF; decomposition pressure is 1.4 psia; at 1150°F,
which is the design temperature at the coolant pump inlet for the MSER,
the BF; decomposition pressure is 4.9 psia or one-third of atmosphere.

By way of contrast the Li-Be coolant salt at 1025°F has a vapor pressure
of gbout L X 10'6 psia. In a system such as the MSRE coolant system,
which operated at a total overpressure of 20 psia, the pump bowl gas
space and the associated off-gas stream would contain from 7 to 25% BF;,
depending on the temperature at the salt-gas interface. This rather
significant partial pressure of BFy suggested the possibility of problems
in the contrel of salt composition, in pump operation, and in the opera-
tion of the cover-gas system. In addition, so far as we know, this was
the first attempt to circulate molten fluoroborate salt in large-scale
equipment, and it appeared likely that some unforeseen problem might arise.
Therefore, the original objectives of the sodium fluoroborate circulating
test loop can be summarized as follows:

1. Examine the pumping characteristics. Compare head-flow data
with similar data for the Li-Be salt. Determine minimum overpressures
necessary to suppress cavitation.

2. Determiné what problems might be involved in monitoring and con-
trolling the composition of the salt.

3. Accumulate experience in the operation of a fluoroborate circu-
lation system. In particular, make observations on freeze valve opera-
tion, salt sampling, and the handling and control of gases containing
BF; . In general, make note of anything that might be useful in the de-

sign or operation of a fluoroborate system.

 

*Unless otherwise indicated, the terms fluorobtorate or sodium
fluoroborate will be used herein to designate the NaBF,-NaF (92-8
mole %) eutectic mixture.
After the test work was under way, two other items assumed sufficient
importance to warrant designation as a major objective.

L., Examine the possibility of preventing undesirable accumulations
of corrosion products (such as on heat exchanger surfaces) by providing
for preferential deposition in a cold trap.

5. Determine the nature of mists and vapors discharged from the
pump bowl vapor space into the off-gas line. Develop separators, traps,
filters, or other devices that will manage these materigls so as to mini-

‘mize flow restrictions and fouling of control wvalve trim in the off-gas

system.

3. SUMMARY OF TEST RESULTS
3.1 General

The test work produced no evidence of any engineering problem that
would preclude the use of NaBF;—NaF eutectic as a secondary coolant for

molten-salt reactor systems.

3.2 Pumping Characteristics

 

The results of hydraulic performance and cavitation tests indicate
that head-flow and cavitation characteristics using fluoroborate salt
should be predictable from water test data taken with similar pumps.
Cavitation inception pressures for a flushing batch of salt were 7 to
16% higher than similar data for a clean batch of salt. This effect was
attributed to an increase in BF, partial pressure resulting from contami-
nation of the flushing salt by the residual MSRE-type (Li-Be-~U-Th) salt

remaining in the loop from prior test fiork.

3.3 Control of the Salt Composition

 

Efforts to evaluate methods for composition control were hampered
by our inability to determine the composition of the salt with suffi-

cient accuracy and precision., However, after due consideration of the
test data, we believe that the composition of the salt remained essentially
constant over the 11,000-hr period of circulation, ahd we further conclude
from this that the use of a BF; overpressure system was successful as &
composition control method. Of techniques considered for monitoring the
salt composition, the test work indicated that the off-gas thermal conduc-
tivity method is feasible and that chemical analysis of salt samples by

use of currently available techniques is unsatisfactory.

3.4 Centreolied Deposition of Corrosion Products

Data were obtained on the relative size and chemical composition of
deposits formed on a 'cold finger'" which was inserted beneath the surface
of the salt pool in the pump bowl. The ultimste objective was to deter-
mine if cold trapping could be used in reactor secondary coolant systems
to control corrosion product concentrations and thus to inhibit the for-
mation of harmful deposits on the steam generator heat transfer surfaces.
However, the test results were too megger to permit any meaningful con-

clusions.

3.5 Analysis and Correction of Restrictions on the
Oif-Gas System

Initial operations were characterized Ty flow restrictions in the
off-gas line. The trouble was traced to a mixture of materials (salt
mists, acids, metal corrosion products) carried from the pump bowl by
a purge-gas stream and deposited in urdesirable places by condensation
and/er gravity. 'West results Indicate the problem can be controlled by
a properly designed system of traps and filters in the off-gas line at
the pump bowl ocutlet. Also the problem may be ameliorated by pretreat-
ment of the salt to minimize impurities and by designing the pump so as

to minimize formation of salt mist in the pump bowl gas space.

3.6 Miscellaneous Observations

Handling and circulation of the fluorotorate salt were accomplished

without difficulty using routine molten-sglt handling techniques. An
incident involving leaskage of salt to the atmosphere dramatized the
essential lack of secondary effects. The importance of providing a clean,
leak-~tight system for handling BF; was confirmed. The performance of the
salt freeze valve and of various instruments for measuring pressure and

flow in both the salt and gas systems appeared to be reliable and adequate.

L. DESCRIPTION OF TEST FACILITY

h.,1l General

The test work was done in an existing facllity that was modified to
meet the requirements of the fluoroborate test. The facility was con-
structed in 1956 and was operated for many thousands of hours circulating
NaK in order to obtain performance data on model PKP pumps for the Air-
craft Reactor Test (ART). In 1962 the facility was reassigned to the
Molten-Salt Reactor Program and between 1962 and 1966 was operated for
more than 17,000 hr in the circulation of molten-fluoride salts of the
type (Li-Be-U-Th) used in the MSRE. Changes made prior to the start of
the fluoroborate test included provisions to obtain salt samples, revi-
sions to The purge- and off-gas systems to provide proper equipment for
handling BF; gas, revision to the drain line freeze valve to better
simulate the MSRE installation, and revisions to the containment and
ventilation systems to insure proper containment and disposal of any
vapors that might accidentally leak from the loop. As a result of prob-
lems that arose during the course of the fluoroborate test work, the
BF; feed system was modified, and miscellaneous revisions were made in
the purge- and off-gas systems primarily to cope with flow restriction
problems. For design details of the complete facility see the drawings

listed in Appendix B.

4.2 Salt Piping and Components

 

The salt piping is shown in simplified outline in Fig. 2. The pump
and piping were fabricated from Inconel. The pipe size was 4 in. IPS

except for about 4 ft of 3 l/E-in. IPS at the pump discharge. The pump
l_;:j

ORNL DWG 72-2073

 

—200-hp MOTOR

{ ™

 

 

 

 

 

 

 

 

PKP PUMP

   
     
  
 
    

DISCHARGE
PRESSURE

_ ) PMD {09
VENTUR' METER.

SALT FLOW
THROTTLE _
VALVE — FREEZE VALVE

- — DRAIN LINE
VENTURI THROAT PMD'C48
VENTURI INLET PMC1044A - - DRAIN TANK
g 4>
_ e

 

2. Simplified schematic of fluoroborate circulation loop.
was located at the point of maximum elevation. From the pump discharge
the salt flowed in order through an ufiper horizontal section, a verti-
cally oriented 180° return bend, a lower horizontal section, a venturi
element, and a throttle valve. From the valve, which was at the point
of minimum elevation, the flow proceeded directly upward about 3 ft to
the pump suction. Each horizontal section was about 16 ft long and was
pitched slightly to facilitate draining of the loop. An 8-ft® Inconel
drain tank served as storage space for the salt inventory when the fa-
c¢ility was not in operation. A 3/h-in.-IPS drain line connected the

dip leg in the drain tank to the bottom of the throttle valve housing

in the loop. Transfer of salt was accomplished by means of gas pressure,
and a freeze valve (see Section 6.4) was provided to isolate the drain
tank from the loop. Table 1 lists principal data relating to loop geome-
try.

The pump, designated as model PKP, is a forerunner of the pumps used
in the MSRE fuel and coolant systems. It is a centrifugal sump pump with
integral pump tank and vertical shaft (Fig. 3). During the fluoroborate
test work the pump speed was 1800 rpm. The upper exterior surface of the
impeller is equipped with ribs which function to control the rate of leak-
age (fountain flow) from the volute into the pump tank by way of the upper
shaft seal.® Baffles and a thermal shield serve to protect the bearing
cavity from excess temperature. A forced-circulation oil system lubricates
the bearings and cools the thermal shield.

The shaft 0il seal is a metal-graphite rotating mechanical s§al. 0il
that leaks past the seal drains into a catch basin (Fig. L4). A top hat or
dam is provided to minimize the tendency for leskage oil to flow down the
pump shaft to the pump bowl. A continuous flow of inert gas (helium or
argon) serves as a shaft purge. At the shaft annulus the flow is split
into two streams. One stream, equal to about 90% of the total, flows down
the shaft into the pump bowl vapor space, serving to inhibit back diffusion
of pump bowl vapors. The remaining portion of the purge gas flows up the
shaft over the rotating seal and through the oil catch basin. The piping
is arranged so that most of the accumulated oil is forced out of the catch

basin with the gas stream and into an oil catch tank further downstream.
il
CATCH BASIN - ——-- = - - ‘Z?QQ-L
THERMAL BARRIER--._ | P90
T ‘z‘ P
BAFFLES—— T -
v T o
MAXIMUM < : :
PERMISSIBLE LEVEL —— /3 ,_fi_hrflfiw4l it i
i gr A
UPPER IMPELLER CASING A ‘
Hi | 5
NORMAL OPERATING LEVEL- —— t ;fi
N ~
_//i%§§§ o
MINIMUM PUMPING LEVEL - = " ‘\\“,’\
3 ¥
. N
l\‘\.
INLET

Fig.

10

 

 

 

 

 

 

 

 

 

 

 

 

 

ORNL-LR-DWG 243578

LOW TEMPERATURE
ROTARY ASSEMBLY

o 2 4 6
e

INCHES

 

 

 

 

 

 

 

 

 

 

 

"1 L1 - MECHANICAL
11 SHAFT SEAL

 

 

 

 

! > PUMP TANK

i DISCHARGE
T__ji__T PIPE
. A E
; ;
= /
_;f/,,” =
- <
\ VOLUTE
"IMPELLER

3. PKF purmp cross section.
11

QRNL DWG 72-2075

PUMP SHAFT

 

 

 

 

 

 

 

 

 

 

 

 

    
  
  

   
  

 

 

Z
L BEARING 7 %
N cavity P / N
) / %
N
ot g ot | SHAFT
TO OIL RETURN LINE fay 3~T0P % SEAL
T har / N2
}t./’a % ERTREEN
N / CATCH R
N / N BASIN
JiE 7~ s
— N
—— / N > LABYRINTHS
3 % b
N ;
FLOW CONTROLLER § % TO OFF-GAS
- N SYSTEM
N 7
olL l\\ 7 —
CATCH ASSITED /é RN
TANK % i3

 

SHAFT PURGE PUMP BOWL VAPOR SPACE

FLOW CONTROL

Fig. 4. Schematic diagram of shaft purge system on PKP pump.
12

The gas pressure in the bearing cavity is controlled at 1 -to 2 psi above
the pump bowl pressure so that oil seal leakage flows into the catch

basin.

Table 1. Descriptive data on fluoroborate circulation loop

 

Circulating Pump

 

1loop tank ~ tovel
Salt volume, ft° 4.1 0.5 4.6
Salt mass, 1b L76 58 534
Gas volume,a £t 0 1.2 1.2
Surface area of wetted metal, 2 - 50 L 5L
Free surface of salt, ft? 0 3 3
Length of piping, ft 4o Lo
Linear velocity of salt at 800 gpm, fps 20
Typical Reynolds number at 1025°F 7.7 x 10°

 

aAssuming salt level at 2 in. above volute midplane.

4.3 Salt Sampling Device

 

Salt samples were obtalned by dipping a copper bucket into the salt
in the pump bowl. Although the pump bowl inventory was not in the main
circulating stream, the leakage through the pump shaft labyrinth (foun-
tain flow) was estimated to be enough to cause a complete interchange
with the circulating inventory every 2 min, and so the salt samples were
assumed to be representative of the material in the circulating stream.
The sample tubes were hydrogen fired before use to remove oxide scagle,
and a special pipe housing was provided to prevent contact with air be-
fore and during the sampling process. The sampling procedure was as
follows: The sample device (Fig. 5) was screwed onto the pipe nipple
on the pump bowl sample access pipe. The sample unit ball valve was
opened, and the sample unit and nipple were purged and evacuated to re-

move wet air and then were pressurized to about 2 psi above the pump
13

QORNL DWG 72-2076

SAMPLE STICK ——
\ Ya-in.-0O0 Cu TUBE

50 in.
ENDS OF
TUBES
SEALED

OFF e
- 7/32-in. HOLE
, ,

 

SAMPLE VOLUME 3 ¢m®
SAMPLE WEIGHT 5.6g

 

e TEFLON SLIDING SEAL
PURGE

CONNECTION
———— PROTECTIVE HOUSING

 

SAMPLE UNIT BALL VALVE

fl\ PUMP TANK BALL VALVE

PUMP TANK

07 7777

 

 

 

 

Fig. 5. ©Salt sampling device.
1L

bowl pressure. The pump bowl ball vaglve was then opened and the sample
tube inserted until the bucket bottomed in the pump tank. After an in-
sertion pericd of 10 to 15 sec, the sample bucket was withdrawn into the
protective housing, and the pump tank ball valve was closed. After a
cooldown period of at least 15 min, the sample bucket was removed from
its protective pipe housing, cut loose from the l/h-in.VCOpper tube ex-
tension rod, and placed in a sample jar that had been flushed with argon

to remove excess wet air.

L.4 Heating and Controlled Ventilation

The loop was heated with Calrod heaters applied to bare pipe and
ceramic heaters installed on the pump tank, drain tank, and air cooling
shrouds. The heagter input was controlled by manually adjusted Variacs.
The circulating salt was cooled by using the suction of the ventilation
blower to draw a controlled flow of air through annuli formed on portions
of each horizontal section of the piping (Fig. 6). Except for the top
portion of the pump, the loop proper was completely enclosed in sheet
metal. The blower was used to maintain a slight negative pressure in
the enclosure, so that any gas leakage from the loop would be diluted
with air and discharged from the stack on the roof. The loop operating
temperature was maintained by balancing the power supplied by the pump
and by the resistance heaters against the power removed by loss to the

surroundings and by the cooling air.

L.5 GCas System Design

 

Gas input was constant and was made up of three separate streams
(Fig. 7): (1) the shaft purge, rnormally about 950 cnf /min of helium,
whose function was to inhibit diffusion of pump bowl vapors into the
bearing cavity; (2) the instrument line purge, about 230 crf /min of
helium, which maintained a continuous sweep of purge gas through the
line that was used to sense pump bowl pressure and also served as the

vapor phase tap for the salt level indicator; and (3) the mixed-gas
15

ORNL-DWG 72-2077

= SALT CIRCULATION PUMP
T0

s
7 / / SHEET METAL ENCLOSURE

 

 

~400,000 Btu/hr / STACK

PUMP ENERGY

 

 

 

 

    

CUBICLE
FOR
BFs GAS 2500 cfm
SYSTEM VENTILATION
CONTRQLS- FAN

 

a

 

——

 

Fig. 6. Loop ventilation and temperature control.
 

 

 

 

 

16

ORNL-DWG 72-2078

/SHAFT PURGE, 950cm3/min : ' TO
: LOWER SHAFT SEAL

 

 

<

o
'

  
 

PURGE FLOW, 100cm%/min

o7
A MAIN

 

TO PUMP BOWL

 

PRESSURE CONTROL VALVE

 

 

 

 

      
  

 

 

(D
"

 

 

 

 

40-psig
+HELIUM HEADER

 

 

 

 

 

     

  

 

 

 

 

 

 

@ 'I:UMP BOWL .
RESSURE TRANSMITTER :
* '%l OIL_CATCH o oS
D3—L1 - K - i
oo TAN 1400 cm>/min —]
~INSTRUMENT LINE PURGE : .
230 em¥/min  FROM PUMP BOWL  __ .
= } - PRESSURE TRANSMITTER 1
PUMP BOWL : i -
: SALT LEVEL : s " pUM
| (FR) INDICATOR A PORESSURE.
MIXED GAS HELIUM : : Y CONTROL
370 em3/min _..C_.)-Q—_— 1 . " VALVE
l N . - ‘—- !

 

SALT PUMP

 

@ CONSTANT AP FLOW CONTROLLER

B FuTeER

 

BF3. 50 cm¥/min (BASED ON 1025°F SALT TEMPERATURE)

s
LT

feed, about 370 cn?/min of helium plus BI,, which provided the necessary
BF; feed to the pump bowl and also served as the high side tap for the
salt level indicator. The BF; flow required for 1025°F operation was 50
cr® /min. When the loop was operated at other salt temperatures, the BF;
flow was varied accordingly, and the helium was adjusted to keep the
total mixed-gas flow at 370 cnf /min.

There were two effluent gas streams from the system. One was a
small fraction of the shaft purge, about 100 crf /min, which flowed up
the shaft and served to keep oll leakage swept out of the pump. This
flow was controlled at a constant value. The other effluent, the main
off-gas stream, consisted of the remainder of tThe shaft purge and the
other gas flows that entered the pump bowl vapor space. The flow rate
in the main off-gas stream was dependent on the pump bowl pressure con-
trol valve, which operated to keep the pump bowl gas overpressure gt the
desired set point. Since the other gas flows were constant, when the
system was at steady state the off-gas flow was constant and egual to
the total input minus the lower seal purge flow.

Ball valves with packed stem seals were used in the salt sample
access line to permit passage of the sample tubes and the cold-finger
assembly. The experimental traps and filters in the off-gas line also
were equipped with ball-type isolation valves because the gtraight-
through flow path minimized the tendency for salt particles and other
contaminants to collect in the valves. Otherwise, the gas system was
equipped with globe-type valves that had packed stem seals in existing
areas and with some few exceptions demountable bellows stem seals in
newly installed areas. Check valves and relief valves were of the
spring-loaded poppet type. Body material was brass and trim was brass
or Teflon, although in some cases stainless steel valves were substituted
because of availability. End connections were 1/4-in.-OD tubing or 1/k in.
IPS, except where larger sizes were dictated by specific operating require-
ments. All valves were selected from existing, commercially available

models.
L.6 BF; Disposal

The loop was operated for the most part with the salt temperature
at 1025°F and the total overpressure at 38.7 psia. At 1025°F, the BE,
partial pressure is 1.35 psi, and so the BF, concentration in the off-
gas stream was about 3.5% by volume. After passing through the loop
pressure control valve, the off-gas stream was passed through a mineral-
oil bubbler to inhlibit back diffusion of moisture. The gas stream was |
then vented into the 1l2-in. suction line to the stack blower. The off-
gas flow rate was 1.5 liters/min, and the blower flow was 2500 cfm; so
the concentration of BFy in the stack was (1.5/28.3)(0.035/2500), or 0.7
ppm. Dilution by the atmosphere probably reduced the concentration by a
Tfactor of 100, so this method of disposing of the BF; kept the atmospheric

concentrgtion well below the continuous exposure limit of 1 ppm.

4.7 Instrument and Controls

 

Approximately 32 sheathed Chromel-Alumel thermocouples were provided
on the salt system to monitor the salt temperature. One thermocouple well
was lmmersed in the sslt in the pump bowl and another was installed in the
drain tank; all other thermocouples were strapped to surfaces of piping
and components. Salt pressure was measured by pressure transmitters (PT),
which use a flexible thin-metal diaphragm and an NaK-filled transmission
line to relay pressure impulses to a remote receiver. 8Salt flow was indi-
cated by the AP across a full-flow venturi, and the AP was measured by the
difference 1n reading btetween a PT at the venturi inlet and another at the
venturl throst. Salt flow could be varied tetween L4OO and 1000 gpm by ad-
Justing the throttle valve. A metal tellows was used for the stem seal on
the salt valve, and an automatic gas-pressurizing system was connected to
a chamber surrounding the bellows to insure that the difference in pres-
sure across the bellows was always controlled within acceptable limits.
Spark plug probes were provided in the pump bowl and drain tank for single
point indication of salt level. A gas bubbler tube, mentioned earlier,
was provided in the pump bowl for continuous salt level indication over a
range of 10.75 in. A high-frequency conductance probe was used very

briefly for salt level indication (see Sect. 6.6).
19

Pressure gages with silver-scldered phosphor-bronze Bourdon tubes
were used for gas lines containing BF;. The pressure transmitters for
BF; flow and salt level had wetted parts of stainless steel or Teflon.
The O0- to 50-psig pump bowl pressure transmitter had a heliecal element.
Caplllaries were used for gas flow elements. The capillary for the BF,
flow element (FE-B9) was 0.031 in. in inside diameter by 45 in. long,
and the calibrated BF; flow rate at 40 psig metering pressure and 809
scale (16.in. H,0) was 330 cnf/min. The BF; calibrations were done
using a wet test meter filled with mineral oil. Constant AP flow con-
trollers were used for control of gas flow, with the input flows refer-
enced to the LO-psig supply pressure and the lower seal purge referenced
to atmosphere. Off-gas letdown was controlled by a pneumatic control
valve with stainless steel body and trim, gasketed bellows stem seal,
and a CV of 0.0L1.

The thermsl conductivity cell was constructed of tungsten-rhenium
elements, brass element block, and silver-soldered stainless steel tubing
leads. The cell temperature was controlled at 100°C, and sample and ref-
erence gas Tlows were 100 cn?/min. The off-gas sample take-off point was
Jjust upstream of the pressure control valve, so there was a time lag of
1 to 2 min between changes at the pump bowl outlet and cell response.

Figures 8 to 10 show the main control panel, the BI; supply cubilcle,

and the off-gas cubicle.

5. TEST PROCEDURES, OBSERVATIONS, AND CONCLUSIONS FOR
THE PRINCIPAL TESTS

5.1 Pumping Characteristics

 

The test was designed to obtain head-flow and cavitation inception
data and, by comparing these data with water and liquid-metal data for
the same pump, to lend assurance to the assumption that water test data
may be used to predict the performance of similar pumps when handling
fluoroborate salts.® A brief discussion of the tests is given below;

for further informstion see Refg. 3 and L.
 

control

20

panel, fluoroborate

 
 

oo

v}

2l

 

) Fig; 9. BF; supply cu'bicl»e‘..‘

PHOTO 75385

 
 

 

            

  
 
 

 

   

 

 

 
 

 

 
 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

. { ! . - C ' . -
\ ' ’ .
2 . »
-Q
—f
. o
8
=
o
. . 0
: @
Q Y
Gy
, o
_ ~
0 | ,
_ o
B ‘

 
   
 

 
23

The tests were made at a pump speed of 1800 rpm and at various salt
temperatures in the range 900 to 1300°F. TFor the head-flow tests the
procedure was to adjust the salt temperature to the desired polint and
then to vary the salt flow by adjusting the position of the loop throttle
valve. The procedure for the cavitation inception tests was to adjust
the salt temperature to the desired point and, with flow constant at 750
gpm, to reduce the gas overpressure in fixed decrements until the drop
in discharge head per unit decrease in pressure showed a disproportionate
change.

The hydraulic performance data are compared with HzO and NaK data in
Fig. 11. The excellent agreement between results from all sources creates
confidence that the head-flow characteristics‘of the PKP pump could well
have been adequately predicted, based on the results of available H,0 and
NaK tests. As a corollary, we can conclude that the performance of simi-
iar pumps, such as the MSRE coolant salt pump, could be adequately pre-
dicted based on water test results.

Cavitation data for the clean charge* of salt indicated that the
inception of cavitation for fluoroborate salt can be correlated on the
basis of NPSH (net positive suction head) and vapor pressure (see Fig.
12). Data for the flush charge of salt showed cavitation pressures from
7 to 16% higher than those for the clean charge. This effect 1s attri-
buted to the fact that the flushing charge contained the residue of a
molten salt (designated BULT-4) that had been used in the previous test
work in the facility. The total charge of flush salt into the test fa-
cility was 689 1b, and the estimated residue of fuel salt was 26 1b; so
that a 16% rise in partial pressure at 1150°F resulted from adding about
4% by weight of fuel to the coolant. Tables 2 and 3 present data on the

compositions of the salt charges.

 

¥The term "clean charge" is used to designate the mixture of salt
that resulted when the new batch of NaBF, was added to the heel (esti-
mated to be about 17 1b) of flush salt which remained in the system
after draining out the flushing charge.
2k

ORNL-DWG 68-13040
140 ?
i
O FLUOCROBORATE AT 900°F

® FLUOROBORATE AT {450 °F
120 b S

|

 

 

'

 

 

 

 

 

 

 

 

= //:/’NJK“ZOO F)
<OI -—'_._-’/—-v—-.. @
¥ 100 o e
a ; O
= H20// %
a

80 - -

SPEED 1795 rpm
|
60 l
400 600 800 1000

FLOW (gpm)

Fig. 11. Compariscon of PKP pump characteristics with sodium fluoro-
borate, NaK, and water.
25

ORNL-DWG-T72-2079

 

35
|

SALT FLOW: 750 gpm
PUMP SPEED: 1800 rpm /

30 /

 

 

 

25 — NPSH AT
: CAVITATION\-v/
INCEPTION NPSH AT

FLUSH CHARGE CAVITATION

 

 

INCEPTION
_ CLEAN CHARGE
° 20 ——— y
w
w /
@
o
B
&£ 15 /
Q.

* SEE APPENDIX D

10
CALCULATED /

VAPOR
PRESSURE *

900 10C0 1400 1200 4300
SALT TEMPERATURE (°F)

 

 

 

 

 

 

 

 

 

Fig. 1l2. Cavitation data for PKP pump with NaBF,-NaF eutectic salt.
(a) Pump NPSH requirements (Ref. 3); (b) vapor pressure of sodium fluoro-
borate (eutectic) (Ref. 5).
26

Table 2., Composition of salts circulated in PKP-1 loop

 

Composition (wt %)

 

Li Be U Th Na, B F

 

BULT-4 salt formerly in loop® 9.7 5.8 5.1  20.0 59.3
Fluoroborate salt flush chargeb 0.20 0.18 0.26 0.24 21.2 9.3 66.9
Fluoroborate salt clean chargeb 0.05 0.02 0.02 0.10 20.9 9.3 68.0

 

It

8Figures for BULT-4 ealt are calculated (LiF-BeF,-ThF, -UF,
65-30-4-1 mole %).

bFigures for fluoroborate salts are averages of all samples
analyzed during the entire test program.

Table 2. Relative amounts of fuel salt constituents in
fluoroborate charges

 

 

 

Method Moles per 1000 moles Na

Salt mixture of
determination Li Be U Th
Flush charge  Calculated™ sh 26 0.9 3.0
Flush charge Chemical analysis 32 21 1.0 1.0
Clean charge Chemical analysis 8 2 0.08 0.4

 

%Rased on 689 1v flush salt and an estimated BULT-L
heel of 26 1b.

5.2 Control of the Salt Composition

 

Tt is desirable to keep the salt at or near the eutectic (NaBF,-NaF
92-8 mole %) composition in order to take advantage of the minimum melt-
ing point* (Appendix A, Fig. A.1). The high BF, partial pressure, to-

gether with the continuous flow of purge gas through the pump bowl vapor

 

*Melting point as used herein refers to the appropriate point on
the liquidus curve.
27

space, provides a mechanism for loss of BF; and a change in the composition
of the salt. The objective of this phase of the test loop work was to de-
termine the problems that might be associated with composition control. A

preliminary survey revealed the following key points.

1. Loss of BF, would shift the composition in the direction of lower
NaBF, content. The resultant increase 1n melting point would increase the
probability of precipitating NaF in the cooler areas of the system, such
as in the steam generator tubes. The suggested solution was to feed BF,
gas into the pump bowl gas space at a rate such that the partial pressure
of BF,, based on BF, flow and total gas flow, 1s equal to the equilibrium
BF, pressure of the salt, based on the desired salt composition and on the

temperature of the salt at the salt-gas interface.

2. If a BF, feed system is used and the feed rate is too high, the
composition would be shifted in the direction of increased NaBF; content.
On this side of the eutectic, the rise in melting point i1s limited to a
moderate 27°F, but the increase in BF, partial pressure might be intolera-
ble. For example, at 1150°F the BF; partial pressure would increase from

5 to 30 psia if the NaBF, fraction increased from 92 to 98.5 mole %.

3. Rates of change in the salt composition are likely to be rela-
tively low, because the amount of Bl in the salt is very large compared
with the rate at which the BF; would normally be added or removed by the
gas stream. Thus, if BF; is removed at the maximum rate, that is, no
attempt is made to add BF; to the system (and ignoring the decrease in
partial pressure with composition), the time (in days) required to pro-
duce a change of 1% in mole fraction would be 2.L x 10-° (wo/F)(Pt/Pb)’
where W, is original weight of salt (1b) assuming a 92-8 mole % mix, F
is total gas flow (cfm), P, is partial pressure of BF, (psia), and P is
total gas pressure (psia). Assuming a BF; partial pressure of 5 psia
(salt temperature = 1150°F) and a total pressure of 25 psia, the above
expression was applied to three different systems with the results given

below.
28

Time required for

 

 

Inventory  Flow 1 mole % change
System (1v) (cfm) (days)
NaBF, test facility 700 0.05 1.7
MSRE coolant system 5,000 0.05 12
MSBR coolant system 500,000 2 30

We note that even for a relatively small system such as the test facility,
almost two days would be reguired to produce the indicated change. In the
MSBR system, even if the estimated inventory is high by a factor of 10,
the elapsed time would still be three days. We also note that the times
required to reduce the NaBF, mole fraction would actually be longer than
indicated since the BF; partlial pressure, and hence the rate of removal
of BF;, would decrease as the NaBF, fraction decreased. If we look at
the case where the salt composition is being shifted upward because of
the addition of an excess of BF,, we note that the rate of change of salt
composition is proportional to the difference between the BF; partial
pressure of the salt and the BI, partial pressure based on the gas flows.
In order to have a rate of increase of NaRF; fraction numerically equal
to the maximum rate of decrease postulated in the example above, it would
be necessary to assume a 100% error in the control of the BF; gas flow or
some appropriate combination of errors in the BF; and helium flows. We
consider errors of this magnitude to be very unlikely in a properly
instrumented system and hence conclude that rates of change of NaBF,
fraction in the upward direction will be no greater than, and are likely
to be somewhat less than, the maximum rates of change of NaBF, fraction

that could be expected in the downward direction.

4. Assuming steady-state conditions, the salt will eventually ad-
Just to a composition whose BF; partlal pressure is egual to the BIj
partial pressure based on the gas flows. Therefore, i1f we maintain proper
control of the salt temperature, then control of the salt composition is
limited only by the accuracy with which we can control the BF, partial
pressure of the cover gas. In an actual system, where there would be
finite limits of error in the control of salt temperatures and gas
flows, it would probably te best to control the BF; partial pressure
with a deliberate bias on the high side. The rationale is that it is un-

likely that we will be able to control precisely at the eutectic point;
29

therefore we should try to insure that the normal deviation is on the
side of increased NaBF, fraction in order to gain the advantage of the
lower incremental increase in melting point. By reference to Fig. A.l
in Appendix A, we note that the melting point increases an average of
23°F/mole % between 92 and 90 mole % NaBF,, whereas the increase between
92 and 100 mole % NaBF, is only 5.4°F/mole %. In order to examine the
feasibility of operating with a bias in BF; partial pressure, consider
a system in which the mole fraction of NaBF, is f and the temperature
of the salt at the salt-gas interface is 1150°F; a total gas flow of F
scfm is flowing past the salt-gas interface in the pump bowl; the gas
flow is made up of a mixture of Fh scfm of helium and Fb scfm of BF;;
the gas pressure in the pump bowl is maintained st a value of Pt psia;
the partial pressure of BF, is P psia; (Pb)g represents the partial
pressure of BF; based on gas flows,

(), = (B(E/F) = (B.R)/(F, +F) ;

g =
and (Pb)s represents the partial pressure of BF; based on the composition

and temperature of the salt (see Appendix D):

(P )S = (0.4e5%) /(1 i £) at 1150°F .

b
We would like to know how the salt composition is affected when we set
the gas flows so that they correspond to a value of (P.b)g that is
slightly higher than the ideal value for (Pb)s. We first set (P.b)g =
(Pb)s, then we solve for f in terms of F,, and finally we differentiate

to determine how f is affected by a change in Fb:

—
OJ"U
~—
1l

(7). ,

(o.h251)/(1 - £)

I

(PF)/ (B, + F)

from which

I%F%

f = .
(0.hos + Pt)(?£) + o.E25Fh
If we assume that P% and Fh are constant at 30 psia and 0.5 scfm, respec-

tively,¥* then
. 3OE£
- 30.h25Fb + 0.21

 

T

2

and

B 6.30
(35/3F )y = (30,0255, + 0.2LF

 

If we assume a value of 5 psi for the BF, partial pressure (Appendix
A, Fig. A.2, eutectic composition at 1150°F), we can estimate the ideal
value for ¥, by noting that the ratio of gas flows is equal to the ratio

b
of partial pressures:

(F./F) = (B)/(P, — B ,

(F /0.5) = (5)/(30 = 5) ,

from which

F =0.1
b

“and (Bf/aFé)T = 0.6 for Fb = 0.1. We conclude from this that a reasonable
biags in the BF; flow will result in an acceptable shift in the salt compo-
sition. For example, if the bias is 10% (ayé = 0.01), the composition
shift, 3f, is 0.006. 1In other words, if the BF; flow is adjusted to 10%
higher than the ideal value, assuming an initial salt composition of 92
mole % NaBF, , the composition will shift eventually to 92.6 mole % N=BF, ,
and the melting point at this composition is estimated to be about 3I°F
above the eutectlc melting point.

5. In order to design a reliable system for controlling the salt
composition by means of a BF; overpressure, one must have reliable in-
formgtion regarding the BF; partial pressure of the salt at concentrations
and temperatures of interest. Data have been published (see Appendices
A and D and Refs. 2 and 5) that provide mathematical statements relating

salt composition and temperature and BF; partial pressure, and these

 

*It is estimated that these values are typical of those that might
be used in an MSBR secondary coolant system.
relationships have been used in the remarks under items 1 through L above.
However, test work that was performed to investigate the pumping charac-
teristics of the salt (see Sect. 5.1) indicated that the BFy partial pres-
sure of the salt can be gltered significantly by the presence of contami-
nants. The results of those tests indicated that the pressure of incipient
cavitation was about 16% higher for the flush charge than for the clean
charge, and this effect was attributed to the residue, or heel, that re-
mained in the loop after draining out the old charge of BULT-4 salt. We
do not know that there is a one-to-one relationship between cavitation
pressure and the BF; partial pressure of the salt, but we assume that the
relationship is direct and therefore the BF; partial pressure of the flush
charge must have been higher than that of the clean charge. If such an
effect were present but not known, the BF; partial pressure of the gas
feed would be ilnadvertently adjusted to a value lower than required, and,
as indicated in the discussions above, the NaBF, fraction of the salt
would be reduced due %o transfer of BEF; out of the salt. Therefore, the
composition control system must be able to recognize the presence of con-
taminants and the nature and extent of their effect on the BF; partial
pressure of the salt.

In summary, the consequences of a change in salt composition can be
serious, but the kinetics of the system are such that rapid change is
highly improbable. We should be able 1o use an applied overpressure of
BF; to control the composition within desired limits, provided we have
available a sulitably accurate method for measuring the salt composition
and the necessary data relating the BY; partial pressure of the salt both
as & function of the ratio of the primary constituents, NaBF; and Naf',
and as a function of probable contaminants, such as fuel salt and corro-
sion products. The test work consisted in (1) obtaining operating
experience with the BF; partial pressure control system and (2) evalu-
ating several proposed methods for monitoring the composition of the

salt.

5.2.1 Operation of the BF; partial pressure control system

 

The data furnished by Cantor et al.® were used to make a plot of

BFy partial pressure vs salt temperature with salt composition as a
|
o

parameter (Appendix A, Fig. A.2, and Appendix D). The salt composition
was assumed to be at the eutectic point, NaBF,-NaF (92-8 mole %), and
the target value or control point for BF; partial pressure was obtained
from the plot by interpolafing at the specified salt temperature. The
target value for the BF; gas flow rate was then calculated by equating

the flow ratic to the pressure ratio:
Fo/Fy = B/ By 2

where F% is the required BF; gas flow, Ft is the total gas flow, P% is

the target value for BF; partial pressure, and Pt is the total gas pres-
sure at the surface of the salt.

A genergl description of the gas system is given in Sections 4.5
through L.7. Helium was fed into the pump bowl in three separate streams:
the shaft purge, which was held constant at 850 crf /min (the total flow
to the pump shaft was 950 an/min, but 100 an/min was split off as the
lower seal purge stream); the pressure transmitter purge, which was held
constant at 230 cn?/min; and the mixed-gas helium, which was mixed with
the BF; gas and fed into the pump towl by way of the salt level bubbler
tube. The total flow to the bubbler tube, that is, helium plus BF;, was
held constant at 370 cmF/min, so the total gas flow into the pump bowl,
Fy, was constant at 1450 cnf /min. The flow rate of helium to the bubbler
tube was obtained by subtracting the target value for the BF, gas flow
rete, Fb, from 370. The set point for the total pressure, P%, was some-
what arbitrary as long as a margin was provided atove the point of cavi-
tation inception in the pump. Test data (see Seect. 5.1) indicated that
at the maximum anticipated salt temperature of 1300°F the pressure of
incipient cavitation for the clean charge was about 19 psig, and 5 psi
was added to this figure to obtain a set point of 24 psig for P . This
setting provided a comfortable margin of 20 psi when operating at the
normal salt temperature of 1025°F. It might be noted that there is an
incentive to operate at a high total pressure due to the inverse effect
on the BF; gas flow rate. Also, the pressure was set at 24 psig, rather
than 23 or 25, because éh psig is almost exactly equal to 2000 mm Hg

absolute, and calculations involving metric units were simplified. At
33

1025°F the indicated value for P, is 1.35 psia. Using this value for P s
38.7 psia for P_, and 1450 cnf /min for F,, we obtain a value of 50 e /min
as the target value for the BF; gas flow. Except for brief periods (less
than 10% of total circulating time), when the salt was circulated at tem-
peratures other than 1025°F, and brief special tests that required varying
the BFs flow, the target value for control of the BF, gas flow was held
constant at 50 cmB /min.

We were unable to evaluate the effectiveness of the BF; partial pres-
sure control system on the basis of its ultimate objective, that is, con-
trol of the salt composition, because we did not have an adeguate method
for measuring the salt composition (see Sect. 5.2.2). Therefore, obser-
vations concerning partial pressure control will be limited to some com-
ments relative to the accuracy of the gas flows (Fb and Fh) and the system
overpressure (Pt) and to the mechanical religbility of the system. The
three gas streams that made up the total flow of helium into the system
were monitored either with capillary flow elements or rotameters. The
calibrations of the flow instrumenis were checked before, during, and
after the operation of the facility using a bubble meter and a wet test
meter. Since these test instruments are accurate to within 1%, we feel
that the limit of error in the helium flow rate was of the same order.

We used the wet test meter, filled with mineral oil instead of water, to
calibrate the BF; flow element using BF; gas. However, due to the limita-
tions of the wet test meter, we were unsgble to obtain readings below

200 crf /min, and we could not use the bubble meter for flows between O

and 200 cnfi/min because of the strong affinity between BF, and water. As
an alternative, we calibrated the BF, flow element with argon using the
bubble meter and then used the argon curve as a guide to extrapolate the
BE; flow curve into the low-flow region, attempting to maintain the same
relative separation between the curves (in the region between 200 and 40O
erf /min, the BF, flow was about 20% higher than the argon flow for the
same scale reading). Because of this difficulty with the calibration, the
BF; flow readings probably have a higher error limit than the helium read-
ings, but we feel that an estimate of the BF, flow error could conserva-
tively be placed at less than iEO%. An attempt was made to check the

accuracy of the BF; flow setting by comparison with BF, cylinder usage.
3h

The following data on BF, cylinder usage were extracted from the operating

 

 

log:
Salt Cylinder
Date circulating  pressure
Cylinder time drop
No. On line  Off line (hr) (psi)
2 1-6-69 4-28-69 2323 1450
3 L-28-69  8-15-69 2194 1300
L 8-15-69 1-27-70 2L63 1610

The flow setting for BF, was held constant at 50 cm®/min during the
periods covered by the above operating data. The BF; cylinder data were

used to estimate an average BF; flow rate as follows:

F, = (ap/at)(@/ap)(V,) ,

where
Fc = BF, flow rate,
dp/dt = gverage rate of drop in cylinder pressure,
dM/dp = estimated mass of BF; per psi of cylinder pressure,
V = specific volume of EF, gas.

The average value for dp/dt for all three cylinders was 0.0LOk psi/
min. The specific volume of BF, (Appendix A, Sect. A.L) is 5.6 fta/lbm.
The estimated mass of BF; per cylinder (the cylinder weights were not
checked) was assumed 10 be somewhere between 59 and 62 lbm/psi. The
lower figure was obtained by dividing the total mass of BFa, as stated
on the purchase order, by the total number of cylinders received, that
is, 1000/17. The higher value was the nominal mass per cylinder as
stated on the purchase order. The corresponding values for dM/dp, 0.038k
and 0.0365 lbm/psi, were obtained by dividing by 1600 psig, which was
the average pressure for a full cylinder. By inserting the indicated
values into the equation for F_, we get values of 60 and 63 cn® /min for

the BF, flow rate, or 20 and 26% above the target value of 50 cnf /min.
35

This relatively poor agreement between the BF; flow rate as indicated
by the flow element and the BF; flow rate as indicated by the cylinder
data may have been due to the cumulative effect of small errors in the
estimates for dp/dt and dM/dp and in the technique used for the cali-
bration of the BF; capillary flow element. If the BE; flow rate had
actually been 63 cnf /min instead of the intended 50 cm?/min, then we
could predict, using the method described in Section 5.2, that the salt
composition should have been shifted from 92 mole % NaBF, to 93.5 mole %
NaBF, and, other things being equal, this would have caused a 25% in-
crease in the BF; partial pressure. If the BF; partial pressure had
actually shifted this much, there should have been a significant change
in the reading on the thermal conductivity cell that was being used to
monitor the off-gas steam. Since no such change was observed, we con-
clude that the BF; flow as calculated from the cylinder usage data was
probably in error on the high side.

In addition to the control of the gas flows, performance of the BI;
partial pressure control system requires reliable control of the total
system overpressure (Pt). An error in the indication of tobtal pressure
willl cause an error in the calculation of the BF; gas flow setting, and
the salt composition would thus not be controlled at the desired point.
As an example, suppose that the salt composition and temperature are
such that the BF; partial pressure (Pb) is 5 psi, and 25 psia has been
selected as the set point for total overpressure (Pt)° Since Fb/Ft =
Pb/Pt = 5/25 = 0.2, the set point for BF; flow is 20% of the total gas
flow. Suppose, however, that the pressure indicator is in error, such
that the total overpressure is actually 20 psia. Since the partial
pressure of BF; is 5 psi, this means that the composition of the gas
phase, and likewise the composition of the off-gas stream, is 25% BF;.
As a result, we are feeding a 20% mixture and removing a 25% mixture,
and the salt will be depleted in NaBF, until the composition of the
off-gas stream is reduced to 20%. During the operation of the facility,
the pressure instruments were calibrated against a master gage such that
the error limit in total pressure indication could be conservatively

estimated to be less than *+0.5 psi. Using the normal operating conditions
36

of 1.35 psi for Pb and 38.7 psia for P,, the ratio of BE; flow to total
flow is calculated to be 1.35/38.7 or 3.49%. If we assume that the
total pressure was low by 0.5 psi, or 38.2 psia, we can calculate that
the gas phase, and the off-gas stream, was actually 1.35/38.2 or 3.5u4%,
and BF, would have been removed from the salt until the partial pressure
of BF; over the salt was reduced to (0.0349)(38.2) or 1.33 psi. This
reduction of 0.02 psi in BF; partial pressure would have corresponded
to a reduction of the NaBF, mole fraction of the salt from 92 to 91.9%.
We conclude from this that the total pressure indication during the op-
eration of the facility was accurate enough that a negligible reduction
in salt composition would have occurred even if the pressure indication
had been low by the maximum estimated limit of error.

From s mechanical standpoint, the operation of the RBF; partial
pressure control system was essentially trouble free. As one would ex-
pect, no problems at all were encountered with those portions bf the
system that contained only helium. The BF; flow control system operated
adequately, although there were occasional periods of nolse or hash on
the BF; flow recorder, indicating that the flow controller was responding
sluggishly. We attributed this trouble to accumulation in the controller
orifice of the typical fluid that results when moisture (or moist air)
is introduced into a system containing BF;. The pressure transmitter
that was used to indicate the total system overpressure performed quite
satisfactorily. As noted in the design description, a continuous helium
purge stream was maintained in the transmitter line in order to minimize
the concentration of BFy in the vicinity of the transmitter diaphragm.

No tests were made, however, to obtain a direct measure of the effective-

ness of this purge stream.

5.2.2 Methods for monitoring salt composition

 

Four methods were considered for monitoring the composition of the
salt.

1. Chemical analyses of salt samples. In this method the basic

 

constituents are determined from chemical analyses, and then the mole

fraction is inferred from stoichiometric ratios. A preliminary mathe-
metical analysis indicated that the limits of error of the chemical
37

analyses were too great to permit use of this method. This conclusion
was verified by a plot of the mole fraction as calculated from analyses
of a large number of salt samples taken during operation of the test
facility (see Fig. 13).

2. Change in heat transfer coefficient due to a salt deposit. A
heat transfer surface in the salt stream would be controlled at a tem-
perature slightly above the normal liguidus temperature. A change in
composition would result in a deposit on the cool surface that could be
detected by a change in heat transfer coefficient. It was concluded
that preferential deposition of corrosion products would make this
method difficult to interpret.

3, Differential thermal analysis. In this method the composition

 

is inferred from analysis of temperature profiles obtained by moving a
salt sample through carefully controlled freeze and thaw cycles. Al-
though this method may be applicable, it was not evaluated becguse it
required specialized sampling techniques that were outside the capability
of the test facility.

4. Thermal conductivity of off-gas stream. The off-gas is assumed
to contain the equilibrium concentration of BF; for the given salt tem-
perature and c:omposition.E’J6 The off-gas stream is monitored continuously
by a thermal conductivity cell (Fig. 14), which is calibrated in percent
BF; . The partial pressure of BF; is calculated using the thermal conduc-
tivity cell reading and the total overpressure as indicated by a pressure
transmitter. This value for BF; partial pressure is then compared with
the theoretical value for the eutectic composition as calculated from an

equation of the following form (Appendix D):

1n P = 16.837 — EEéEEQ , (1)

where P is partial pressure of BF, (psi) and T is temperature (°R).

The thermal conductivity cell incorporstes a Wheatstone bridge, with
one leg containing a conductivity element that 1s exposed to a reference
gas. The second leg has a similar element that is normally exposed to
the sample gas, and the output voltage is a function of the difference

in thermal conductivity between the reference and sample gases. A zero
38

ORNL-DWG 69-5491
CALCULATED FROM EQUATION

o RATIO OF B TO Na 2.126* (8/Na)
e RATIOOF F TO B 1/{[0.569% (F/B)]-3}

 

 

 

 

 

 

 

 

 

 

 

 

100 |l - o
g ¢l ol ° 5
S g6 , .
s 9 I
-~ olo- o cg e
% o o ¢ IOO
E 92 &o—-fl—--—'—-—'_oo o.—-—:L—-—B—P—-—-—-—-—t————-p— r—
I ° °
m | * .
2 « o . EUTECTIC
o 88 [« ———ef L% COMPOSITION -+ ———,
::EJ . . l .
3 -
S 84 _— 1
S : |*
"SALT ~— [—= CLEAN SALT
80 - L 1

 

0 20 40 20 40 60 80 100
CIRCULATING TIME (days)

Fig. 13. Mole fraction NaBF, calculated from salt sample analyses.
tion.

39

ORNL—DWG 72-2080
100 cm3/m|n o
N
REFERENCE GAS He - \ VENT

l 1 THERMAL
CONDUCTIVITY
CELL

  
 
     

B, GAS A '

CALIBRATION .,
SHAFT
PURGE He .~ —j{»q——v—-

   

 

   

N »
100 em®/min

 
 
 
 

 

T o OFF
s 6AS
: P Hr,C)

—_ ~— T = b — 5' s

P'Dw ’Df o BF3 —-Pf —P—¢
PUMP /[ - P, = BFy PARTIAL PRESSURE
BOWL R P, = TOTAL OVERPRESSURE
Jaron S T, = SALT TEMPERATURE
SPACE = ~1#:3 " AREAOF SALT g = SALT TEMPERATUR

SURFACE = ~ 32 Co = SALT COMPOSITION

Fig. 1k. Schematic disgram of monitoring system for salt composi-
4o

or background reading is obtained by temporarily exposing both legs to
the reference gas. In the test facility, the reference gas was helium
and the sample gas consisted of a mixture of BF; and helium, with the

BF; concentration normally between O and 5 vol %. The applicability of
the thermal conductivity method depends on there being a significant
difference between the thermal conductivities of the reference and sample
gaseés and on a predictable relationship between the thermsl conductivity

and composition of the sample gas.

5.2.3 Evaluation of the thermal conductivity method

 

In order to examine the applicability of the thermal conductivity
method for monitoring the salt composition, we must consider at least
four questions.

1. Is the BF; partial pressure a reliable indicator of the salt
composition?

2. Is the off'-gas stream in equilibrium with the salt?

3. Is the thermal conductivity of the gas a reliable measure of
the BFz partial pressuvre?

4. TIs the variation of BFy partial pressure with salt composition
reasonatle when compared with errors that might e expected in the meas-
urement of salt temperature, total overpressure, and BF; concentration?

Witk regard to the [irst question, experimental work® has teen per-
formed to determine thke varigtion between salt composition, salt tem-
perature, and B, pressure. These relationships may not te reliable,
Lowever, 1f contaminarts are present in the salt. We know from experi-
ence during the cavitation tests (see Sect. 5.1) that relatively small
amounts of fuel salt added to the fluorotorate salt can cause changes
which imply that the BF; partisl pressure has been significantly affected.
For example, the cavitation pressure of the contaminated salt at 1150°F
is the same as that of the clean salt at 1210°F. Unfortunately, thermal
conductivity cell data from the flusn salt operating period sre insuffi-
clent to determine whether there was a detectanle change in the thermal
conductivity of the off-gas stream. Additional experimental work is
needed to show the variation of EF; partial pressure as a function of the

concentration of any contaminant that might te present or that might be
L1

expected to leak into the fluorcborate salt. Depending on the results

of this work, we may need (1) to develop procedures for purifying the
fluoroborate salt; (2) to arrange the design to prevent entry of the
contaminants into the salt; (3) to provide a monitoring system whereby
we can monitor, and adjust for, the contaminant level; or (L) the effects
may be minor enough so that no action 1s necessary.

Question 2 has to do with the problem of assuring ourselves that the
gas sample we are looking at is truly representative of the equilibrium
partial pressure of BFy at the surface of the salt. In the test facility,
helium and BF; were added continuously to the system at controlled flow
rates; the shaft purge and pressure transmitter purge streams were vented
directly into the pump bowl vapor space, while the BF; and mixed-gas
helium were introduced a few inches below the surface of the salt in the
pump bowl. An equivalent amount of off-gas (1450 cnf /min) was vented
continuously by the pressure control valve so as to hold the system pres-
sure constant. The stream that was monitored by the thermal conductivity
cell was a lOO—cmP/min samplie flow taken from the off-gas line st o point
about 25 ft downstream from the pump bowl, that 1is, Just upstream of the
pressure control valve. If the partial pressure of the BF; in the gas
feed were equal to the equilibrium partial pressure of BFy over the salt,
then there should be no net transfer of BF; between the vapor space and
the salt, and the off-gas stream should represent the composition of the
salt. If the partial pressure of the BF; in the gas feed were not equal
to the equilibrium partial pressure of the BFz over the salt, then there
would tend to be a transfer of BF; between the salt and the vapor space.
The direction and rate of this ftransfer would depend on concentration
difference, surface area, temperature, gas residence time, etec., and it
would be theoretically possible for the off-gas stream to represent the
gas feed composition, the salt composition, or some intermediate value.
Conditions in the test facility were favorable to the ideal situation;
that is, it appeared probable that the gas space, and hence the off-gas
stream, would tend to be in equilibrium with the salt. For example, the
residence time for the gas was long (20 min), the contact surface was
relatively large (3 fta), and there was a good exchange of fluid (esti-

mated at 15 gpm or one loop volume every 2 min) between the main loop
Lo
£

and the pump towl salt pool. Several experiments were run in an attempt
to determine if the composition of the thermal conductivity cell sample
stream was representative of the gas composition that would be predicted
from the salt composition and temperature. These tests are described
below, and test results are summarized in Table 4. In all cases the test
period was short (<8 hr), and it was assumed that the salt composition
was at the eutectic and remained essentislly constant.

1. Change in salt temperature. With gas flow, total pressure, and

 

salt temperature constant, the BF; concentration of the off-gas stream
was determined from the thermal conductivity cell reading. The salt tem-
perature was then changed, and the concentration was again determined
from the thermal conductivity cell. The measured values were compared
with the calculated values. The test was performed over two temperature
intervals: 910 to 1025°F and 1025 to 1150°F. The agreement was only
falr in the first case and very good in the second case.

©. Change in total pressure. With gas feed (1450 crf /min) and

 

salt temperature (1025°F) constant, the total pressure was leveled out
at 30 psia, and a thermal conductivity cell reading was obtained at

this condition. The total pressure was then increased from 30 to L5
psia in about 12 mir by closing the pressure control valve. The thermal
conductivity cell reading was again recorded after the system had leveled
out at the higher pressure. Since nothing was done to change Pb’ the
ratio Pb/Pt’ and hence the RBRFy concentration fraction, would be expected
to decrease bty a factor of 2/3 in order to compensate for the factor of
3/2 by which the pressure increased. Also, if the thermal conductivity
cell were properly responsive to the change in BF; concentration, its
reading should show a corresponding decrease. The thermal conductivity
cell responded as expected, showing a drop from 0.9 to 0.6 MV .

3. Change in gas flow. If the BF; concentration in the vapor phase

 

over the salt were dependent on F%/F%, or 1if there were a time lag in
returning to equilibrium, then the thermal conductivity cell should show
a response to a change in the BF;/He flow ratio. With all conditions
constant, a reading of BF; concentration was obtained on the thermal
conductivity cell.‘ Sudden changes were then made in the gas flow ratio.

In one case the BF; gas flow was increased enough to triple the normal
Takle 4. Effect of changes in loop operating variables
on methods of estimating BF, partial pressure

 

 

 

 

 

Variable conditions Constant conditions Change in Change in BF; partial pressure (psi)
Date thermal
of Salt Total Gas flow conductivity o 13 Thermal G
Test Variable Start Pinish temperature pressure ratio cell reading thP:ir tfig conductivity ?i ‘f::ed
(°F) (psia) (F,/F,) (V) pere cell® °
8-30-68 Salt temperature, °F 910 1025 by .7 0.0L43 0.32 0.7 +1.0
9-26-68 Salt temperature, °F 1025 1150 L. 7 0.0k43 1.95 +3.8 +3.5
6-6-69 Total pressure, psia 29.7 Ly 7 1025 0.033 0.3
8-30-68 Gas Tlow ratio Q.04 0.17 910 L. 7 0 0 0 +5.8
8-30-68 Gas flow ratio 0.0h 0.0 1025 L7 0 0 C -1.8
10-15-68 Gas flow ratio with BF,
entering vapor space 0.04 0.10 1025 Ly 7 0 0 0 +3.7
10-16-68 Gas flow ratio with BF,
entering vapor space 0.0k 0.13 1025 Ly .7 0 O 0 +.0
#3alt temperature: B, = exp(16.83 — (24,540/T, °R) psi.
bGra.s composition and pressure: P = (thermal conductivity cell reading) (calitration factor) (total pressure}.

“Gas feed flow ratio: P = (totel pressure) (Fb/Ft)'

€n
Ly

concentration; in another test, the flow was cut off completely. The
flow variation tests were then repeated with the BF, flow‘admitted
directly to the pump bowl vapor phase rather than under the salt surface.
No response was noted on the thermal conductivity cell during any of the
flow variation tests.

It was concluded from these tests that the off-gas stream from the
test facility was a reliable indicator of the composition of the gas in
the pump bowl vapor space, and in cases where the BF; partial pressure
based on the composition of the gas feed was different from the BRs
partial pressure based on the salt composition and temperature, then the
vapor space composition tended to be that dictated by the salt composition
and temperature. Reactor systems using pumps of similar design and
similar operating conditions should perform in a similar manner. In using
a sample point that is remote from the pump bowl outlet, however, precau-
tions must be taken to insure that the composition of the gas is not
inadvertently changed during its passage from the pump bowl outlet to
the thermal conductivity cell.

Question 3 concerns a weakness in the use of the thermal conductivity
cell as an indicator of BF; concentration. If the thermal conductivity
cell is working properly, it will respond to any change in the thermal
conductivity of the gas whether that change has been caused by a change
in the concentration of the BF; or by the addition or removal of some
other materigl. In the'design and operation of the system, therefore,
precautions must be taken to exclude, or to detect and account for, any
materials that might give a false signal. Experimental work is needed
to investigate the variation in thermal conductivity cell reading as a
function of the concentration of probable contaminants.

The fourth question is concerned with probable errors in the meas-
urement of salt temperature and gas pressure and the effect such error
limits might have on the usefulness of the thermal conductivity cell
method. We start by referring to the general equation for BF; decompo-
sition pressure, which includes the effects of composition change as

well as temperature change (Appendix D):
L5

- (s 5 - 2429)] ©

where P, is partial pressure of BFg (psi), £ is mole fraction NaBF,, and
T is temperature (°R). Differentiating alternately with respect to compo-

sition and temperature, we get

(), - {exp[lu.(325_—f§22h,5¢0/T)3} psi/mole % , (3)

(%% £ - [ngf?] [25%259] {EXPE1A-395 — (Eh,5h0/T)]} psi/°R . (4)

Assuming values of 1025 and 1150°F for salt temperature and 0.91, 0.92,
and 0.93 fof mole fraction NaBF,, the calculated values for partial pres-
sure and for rate of change of partial pressure are as shown in Table 5.
At 1150°F & change in composition from 0.92 (the desired value) to 0.91
would cause a change in partial pressure of about 0.6 psi. The tempera-
- ture error required to cause an apparent change of the same magnitude
would be 0.60/0.047, or zbout 13°F. The expected accuracy of temperature
measurement is 0.5%, or about 6°F. From this we conclude that changes
gregter than 0.5 mole % should not be masked by temperature error.

The apparent rate of change of BF; partial pressure as a function
of total pressure is numerically equal to the concentration fraction

(% BF;/100), assuming constant gas flows:

P, = (B /P (E,)
from which
dp, /aP, = Fb/Ft = P /P, -
Assuming a total pressure of 24 psig, the concentration fraction at
1150°F (B, = 4.93 psi) would be 4.93/38.7, or 0.13. The error in total
pressure reading required to cause an apparent change of 1 mole % in
composition would therefore be:

de/df 067

5@;7&?; - 515 - 5.1 psi/mole % .
Table 5. Variation of BF; partial pressure with composition and temperature

 

Rate of change of

 

€m, Brror in temperature

 

T, Salt f, Mole Py gFé partial pressure required to produce an
. partial .
temperature fraction pressure apparent composition
(°F) NaBF, (psi) dP, /3f oP. /BT change of 1 mole %
P (psi/mole %) (psi/°R) (°F)
1025 0.93 1.58 0.2k 0.018 13
0.92 1.36 0.18 0.015 12
0.91 1.20 0.15 0.013 11.5
1150 0.93 5.69 0.87 0.05k4 16
0.92 4.93 0.67 0.0k7 1h
0.91 4.33, 0.53 0.0kl 13

 

 

ln P, = ln( L ) + 14.395 —

b i-7

(fl) _{exp[14.395 — (2k,540/T) 1}
of (L =T

(572)c - [Ee] [2481] {owstot. 5 - (ob,shormn} oms/ s

2k, 540
T

psi/mole %,

ot
k7

At 38.7 psia, we should be able to read pressure within 0.5 psi, so error
in measurement of total pressure should not be an important factor.

The accuracy of the thermal conductivity cell readout should not be
an important factor in the discussion because the reading is only rela-
tive, that is, the instrument compares the conductivity of the gas mixture
with that of a reference gas, and provision can be made for continual
calibration checks using mixtures of BF; and helium of known composition.
There should likewise be no problem in designing a thermal conductivity
system that will have the desired degree of sensitivity, that is, one
that will be gble to see small enough changes in salt composition. For
example, in the test facility, the sensitivity of the thermsl conductivity
cell was 4.3% BF; per millivolt, that is, for a given reading, R, in mV,
the concentration, C, in % BF,, was equal to 4.3R. If we let the concen-

tration fraction equal fc, and noting that fc = Pb/Pf’ then:

£, = C/100 = L.3R/100 = Pb/Pt

Solving for R in terms of Pb and differentiating, we obtaln the rate of

change of the thermal conductivity cell reading as a function of the BIj

partisl pressure:

v
I

= 23.25Pb/Pt ,

dR/de 23.25/£% mV/psi .

The rate of change of thermal conductivity cell reading as a funection of
salt composition can be obtained by multiplying by de/df, that is, the

rate of change of BF, partial pressure as a function of salt composition:
dr/df = (de/df)(dR/de) nv/mole % .

For the test facility, where P_ was 38.7 psia, the change in reading as
a function of BF; partial pressure was 23.25/38.7, or 0.6 mV/psi. From
Teble 5, we note that at 1025°F and f = 0.92, 4P /df has a value of 0.18;

so for this condition the value of dR/df = (0.6)(0.18) = 0.11 nV/mole %.
L8

The smallest change in thermal conductivity cell reading that could be
considered significant was about 0.05 mV, which was equivalent to a
change of about 0.5 mole %. The unit installed at the test facility,
however, was an old-fashioned unit, and improved cells are now available
with sensitivities that are higher by at least a factor of 10. It might
be noted that the rate of change in partial pressure as a function of
salt composition increases as the salt temperature increases, so that
the test facility unit would probably have been adequately sensitive for
an MSBR system. TFor example, the MSBR is expected to operate at a salt
temperature of 1150°F and a total overpressure of 30 psia. At these con-
ditions, the value of dR/df for the test facility unit would be 0.4 mv/
mole %, and a change in the salt composition of 0.13 mole % should have
been discernible. It might also be noted that the tesf facility unit was
monitoring a mixture of helium and BF,, and the sensitivity of the unit
was directly related to the difference in thermal conductivity cell
gignificantly, or it could rule out altogether the use of the thermal
conductivity method.

In summary, we conclude that for systems which are similar in de-
sign and operating conditions to the test facility, we should be able
to monitor the compogition of a circulating fluoroborate salt system by
using a thermal conductivity cell to monitor the gas stream that is

flowing from trhe vapor space which is in contact with the salt surface.

5.3 Corrosion Product Deposition

 

During extended periods of salt circulation, one or more of the
products of corrosion are likely to reach the saturation concentration
corresponding to the minimum temperature of the system. Continued op-
eration would then result in precipitation on the steam generator tubes
(the coolest surface in the intermediate system) with resultant loss in
heat transfer capability. It is proposed to protect vital heat transfer
surfaces from this fouling by providing for preferential deposition of

corrosion products in a cold trap whose temperature would be maintained
k9

a suitable amount below that of the heat exchanger. This section describes
sdme tests that were run to develop information on the nature and kinetics
of formation of cold surface deposits.

A cold finger was fabricated from a piece of nickel rod, 5/8 in. OD
by 1 1/2 in. long (Fig. 15). An attempt was made to enhance the heat
transfer properties by milling a series of circumferential grooves on the
exterior surface and by forming the interior space in five operations with
a 1/8-in.-diam drill. The operating temperature was sensed by 0.0LO-in.-
diam Inconel-sheathed thermocouples inserted in four 0.055-in.-diam verti-
cal, equally spaced holes located so that the outer edge of each hole fias
about 0.045 in. away from the surface of the test piece (Fig. 16). A
3/8-in.-OD nickel tube was welded to the test piece, and this, in combi-
nation with a central l/8-in.-OD tube, provided a reentrant-type arrange-
ment for introduction and removal of the coolant, a mixture of argon and
distilled water. Other tubes and fittings were provided so that the assem-
bled unit would be able to satisfy the containment and mobility criteria
(Fig. 17). A catch pan (small metal disk) was attached to the bottom of
the cold finger in the hope that it would retain pieces which might become
detached from the surface of the test piece. During a test, the unit was
attached to the salt sample access nozzle on the pump bowl, the sample
ball valve was opened, and the inner <ube was inserfied until the lower
hglf of the test piece was submerged under the salt surface. Movement of
the cold finger was through a Teflon seal above the ball valve. After the
test plece was in position, the coolant flow was started, and the total
and relative flows of argon and distilled water were adJjusted to obtain
the desired temperature. As a result of experience dfiring shakedown tests,
two revisions were made: (1) two of the four thermocouples were eliminated
because space limitations made operation with the larger number impossible,
and (2) the catch pan was eliminated because of clearance problems between
it and internal hardware in the pump bowl.

A total of five tests were run with the cold finger to investigate
the effect of wall temperature on the amount and type of deposit obtained.
Each test period was 6 hr. In addition, two tests were run with a 2-hr
test period to check the effect of time. The test results are summarized

in Tables 6 and 7. Some deposit was obtained in all the 6-hr tests; none
 

 

 

F

loop.

g

| 0.378-in. DIAM X Y-in. DEEP
L 7

NO. 54 {0.055in) DRILL
© % tYy-in. DEEP, 4 HOLES =~

ON 0.480-in. DIAM -

ORNL DWG T2-2081

Ya-in. DRILL X 1%-in. DEEP,
4 HOLES ON 0.200-in. DIAM

   
  
  
   

.NO. 30 {0.1285in) DRILL
X t¥y-in. DEEP, t HOLE .

 

 

 

 

 

¢
oo _ o
1 0.020-in. DEEP GROOVES, ,
. SPACED 0.020-in. APART '
} : o
\
1%in.
2in

 

15. Cold-finger detail, fluoroborate circulation test, PKP

0.032 in.

=Y
o e
Ya-in. omm—!

 

-

ot
 

 

 

3/4-in—0.D. SHIELD TUBE

i

 

BE COOLANT

snT

 

Fig. 16. Cold finger_showing'conhecting'tubing and thermocouples.

- . . .
.
- . .

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L A - S o

Flg 17. Cold-finger a.ssem'bly.-

o
53

was obtained in the 2-hr tests. All deposits were bright green, and the
amount varied from very thin isolated spots to a uniform coating over
the 1 1/2-in.? surface. WNo data were obtained on the weight of material
deposited; however, the total amount of material in the heaviest deposit
is estimated to have been <0.5 g. Chemical analyses indicated that as
the wall temperature of the cold finger was increased, the quantity of
chromium in the trapped sample increased relative to the other constitu-
ents. A rough atomic balance can be achieved if one assumes that the
deposits consisted of a small amount of Ngg CrF; mixed with large amounts
of NaBF, and NaF.

The data from the cold-finger tests are too meager to permit any
final conclusions. The results, however, are sufficiently promising to
indicate that removal of corrosion products by cold trapping is possible

and that further development effort is justified.

Table 6. Summary of cold-finger tests
Salt circulating temperature, 1025°F

 

 

Test Date Wall Tes? .
No. of temperature duration Deposit
test (°F) (nr)
1 L-23-69 750 6 Yes
2 h-29-69 850 6 Yes
3 5-7-69 350 6 Yes, two small areas
Y 5-9-69 890 6 Yes, thin
5 6-13-69 860 6 Yes, heavy
6 6-20-69 ohL 2 No
7 6-20-69 929 2 No

 

5.4 Off-Gas System Flow Restrictions

 

A profile of the off-gas line is shown in Fig. 18. The length and
layout of the line were fortuitous, and the profile is shown primarily

to illustrate the rather lengthy and tortuous flow path between the pump
5k

bowl outlet and the suction line of the ventilation blower, During most
of the loop operations the off-gas flow was 1500 cnP(STP)/min, and the

. pump bowl pressure was 24 psig. The gas temperature at the pump bowl
exit was close to the salt temperature (normally 1025°F), but the gas
gquickly cooled to room temperature in the unhegted off-gas line. Down-
stream of the pressure control valve, the pressure was essentially atmo-~

spheric.

Table 7. Analyses of deposits from cold-finger tests

 

 

Test wall
temperature Na B F Fe Cr Ni
No. o
(°F)
Weight %
1 750 25,1 5.07 59.0 1.54 2,76 0.031
2 850 ok 4.28 57.3 1.61 4.18 0.009
L 890 2L.0 5,26 46.5 1.52 8.38 0.090
Atomic sbundance relative to Fe
, Lo 17 113 1 2
o 36 17 12k 1 3
L 38 18 71 1 6

 

5.4.1 Plugging experience

 

Continuing difficulty was experienced with flow restrictions in the
off-gas system. The principal trouble spots were (1) just downstream of
the pump bowl, (2) the pressure control valve, and (3) the extreme end
of' the off-gas pipe at the point where the off-gas stream was vented into
the 12-in. suction line to the ventilation blower.

The restrictions at the pump bowl outlet were due to salt mist that
was generated in the pump bowl and carried out by the off-gas. When the
gas stream cooled below the freezing point of the salt (725°F), the mist
particles froze and some fraction of them agglomerated and settled out on

the pipe wall. Eventuslly the flow became blocked.
ELEVATION ABOVE FLOOR (ft}

55

* INDICATES 90° CHANGE IN HORIZONTAL DIRECTION,
TOTAL LENGTH OF OFF-GAS LINE FROM PUMP BOWL TO BLOWER
SUCTICN LINE = &7 ft

ORNL-DWG 72-2082

 

(1) PUMP BOWL OFF-GAS OUTLET

(@) YORK MESH FILTER

(® 50-x METALLIC FILTER

(@) PUMP BOWL PRESSURE CONTROL VALVE

 

(®) SUCK-BACK TRAP
(&) MINERAL OIL BUBBLER
(7) FORWARD BLOW TRAP

  

 

   

 

 

e gmeop v, YaminoD oy ¥gein 0D
TUBING 1 TUBING TUBING
3
|
STACK
e i 7
* * J
T T
VENTILATION
BLOWER
% 2500 cfm
™~~~ OOP ENCLOSURE
| | | | ] { | 1 1 |
5 10 15 20 25 30 35 40 45 50

HORIZONTAL DEVELOPED LENGTH (ft)

Fig. 18. Profile of fluoroborate (PKP) loop off-gas line.
56

Restrictions at the control valve were ascribed to two things: (1)
those mist particles that did not settle out in the pipe were carried
downstream and deposited in the valve by inertial force and (2) a vis-
cous fluld was carried out of the pump bowl by the off-gas stream and
condensed out in the control valve and other places in the off-gas lines.
During the initial operation of the loop with the flushing salt, the
emission rate of the fluid was very high, and an estimated 50 cn® was
drained from the line downstream of the control valve. With continued
operation, however, the rate dropped off considerably and leveled off
at about & % 3 mg/hr (equivalent to about 1 g per 15,000 liters of gas
or 0.07 mg/liter).

The restrictions that occurred at the extreme end of the off-gas
line were attrivuted to a slow buildup of materilals resulting from a re-
action bvetween BF; and water. The BF; was normally present in the off-
cas stream at about 3.5% by volume, and the wéter content of the building
atmosphere wag probably about 1% by volume. The plugs normally occurred
aoout 2 to U in. from the end of the 3/8-in. line (Fig. 19). We assume
tnat the moisture diffused up tre small line to this point, where it com-
tined with the BF; coming downstream and the reaction product was gradu-
ally deposited. A total of six plugs were formed. The time of formation
(salt circulation time) was not consistent, varying from six weeks to six
morths, as shown in Table &.

Tre identity of the plug material is subject to speculation since
cnemical analyses were not made. The principal reaction product¥* from
combingtion of EBF; and water is thought to be EF; .20, which presumably
would not be a solid at the conditions (atmospheric pressure and room

temperature) under which the plugs were formed.

5.4.2 0Off-gas tests

In December 1948, the off-gas line at the pump bowl outlet was modi-
fied to provide for the installation of experimental equipment to evaluate

methods of coping with salt mist and other contaminants in the off-gas

 

¥Private communications, R. F. Apple to A. N. Smith, Nov. 22, 1971-
 

 

-

»

 

 

OFF-GAS LINE

 

Fig. 19. Radiogreph of typical plug (formed 9-11-68) that formed at
‘extreme end of off-gas line, NaBF, circulation test, PKP loop. '

 

 
58

stream. The following is a description of the tests that were performed
from December 19, 1968, until the final shutdown of the facility in
April 1970. During the discussion, reference may be made to Figs. 20 to
23 for design details and other pertinent information relative to the
off-gas line and the various traps and filters used during the tests; to
Figs. 24 to 26 for the particular order in waich the various components
were arrayed; and to Table 9 for = chronological summary and for data on
collection rates for salt mists and condensed liquid. Connections be-
tween components were made with 3/8-in.-0OD (5/16-in.-ID) tubing, turns
were made using long radius bends, and the total length of interconnecting
tubing between the pump bowl nozzle and the final component in the test

section was never more than & 7t.

Table 8. Record of flow restrictions
at extreme end of coff-gas line

 

 

Salt Total
Dgte circulation elapsed

observed time time

() (br)
9-11-£8 1350 3200
12-16-68 1270 2300
1-29-69 980 1060
B-220.£9 L1£0 L8T70
11-25-559 1720 2030
2.18-70 1100 2000

 

For the initial series of tests, a Calrod heater was installed on
the off-gas line at the point of connection with the pump bowl (Fig. 20).
The arrangement was such that, with no power to the heater, and with the
system operating at normal steady-state condition (bulk salt temperature
of 1025°F and off-gas flow of 1.5 liters/min), the temperature of the
l/2-in.—ID off-gas line dropped from 755°F at the pump bowl to 4OO°F at
a point 3 1/2 in. away from the pump vowl, and it is estimated that the
59

ORNL-DWG 72-2083

ADAPTER, Y in.ID X %4gin.ID

  
  
  
 
 

(14"
- 430°F
Yo-in iD TUBE °
5% in 570°F
CALRCD 3L 755°F
HEATER o b J

 

2Y-in, DIAM X 3-in. HIGH
EXIT PLENUM
TYPICAL TEMPERATURE PROFILE

WITH CALROD HEATER TURNED OFF

DETAIL OF PUMP BOWL OFF~-GAS NOZZLE-TESTS [aT0O Ic

"

 

‘ L] —

I~in. PIPF—al @.fi-mve EACH 0.020-in. PLATES WITH FOUR EACH 3/4p- X3/35-in. NOTCHES

 

) R :B—D»OUT L1 _l
, i 2 /FOUR EACH 0.020-in. PLATES WITH FOUR EACH ‘4-in. DIAM HOLES
HEATER HOLE CENTERS TO BE 9,in. FROM PLATE CENTER
AND
13 in. ':NSULfiT{ON\‘ \
glbin, l 7

 

 

 

'/2 in.

> BAFFLE PLATE DETAIL-TEST Ilo ONLY

 

VERTICAL HOT-MIST TRAP DETAIL-TESTS LIaT0O llc

 

-~ 8 in——] NOTES
— ouT 1. COLD ELEMENT NOT USED IN ALL TESTS.
V0 J
. |
pall
1

kFROZEN SALT FILTER (COLD ELEMENT)

 

2. ELEMENT DIAM — t.8in.
3. ELEMENT AREA —0.018 ft2
4. FLOW RATE—3 scfm/f12 AT 4.5 liters/min (STP)

 

 

 

2-in.PIPE

   

MOLTEN SALT DEMISTER (HOT ELEMENT)
HEATER AND iNSULATION
HORIZONTAL HOT-MIST TRAP DETAIL—TESTS 1ila TO 111d

Fig. 20. Details of off-gas nozzle and hot-mist traps for NaBF, cir-
culating test, PKP loop.
CRNL-DWG 72-2084

S4g-in. 1D TUBING

 

GLASS PIPE
3in. 1D
BY
Z4-in. LONG—al

 

 

 

 

 

 

 

 

LINEAR VELOCITY OF GAS AT 1.5 liters/min (STP)
AND 24 psig WAS 0.0¢8 fps AND RESIDENCE
TIME WAS {10 sec

ASSUMING SPHERICAL PARTICLES OF sp gr = 2.0,
A SETTLING RATE OF 0.018 fps IN HELIUM
CORRESPONDS TO A PARTICLE SIZE OF Sp.
THEREFORE, PARTICLES = 10u SHOULD

REMAIN IN THE SETTLING TANK

DETAIL OF SETTLING TANK (ST)

 

 

 

 

 

 

 

 

 

 

 

 

 

% UNIT IN SERVICE "D L
-
I "“L"“““fl I-4 [1-43-69 TO 1-22-69 | *in. | 6in.
i i o I-2 | 1-22-69 TO2~-9~69 | 2in. | 12in.
|l DIAMETER D
IiLII EE
(J&j\
' 1
iMPACT
SURFACE :
\ 1
| 0400 in.—mm -  Ygin.
DETAIL OF IMPACTOR UNITS—TEST Ib
Fig., 21. Eettling tank and impactor units, tests Ia and Ib, NaBF,

circulation test, PKP loop.
61

ORNL-DWG 72-2085

/’ NO. ELEMENT

NG FILTER

// ELEMENT MODEL NO. ! NO. 2 NO. 3
NO

 

 

 

 

 

 

 

 

 

 

 

 

/' / | F—1o FM204 — —
f F—tb | NEVACLOG | FM225 | FM204
E [ ' F—1c FM225 FM 204 —_—
= by
F | '
b NOTE: ALL ELEMENTS WERE STAINLESS STEEL
11X 3in .
GLASS P!PEM FILTER DIAM—~1.75 in.
REDUCERS AREA Q. O17 ft2 FLOW = 3.4scfm/ft% AT 1.5 liters (STP)/min
NOTES

1. FILTER ELEMENTS TYPE FM ARE MADE BY HUYCK METALS, INC.
MFR'S REMOVAL RATINGS WHEN FILTERING GASES: FM-225 98% LESS THAN 1,4 1
00% LESS THAN 5 p
FM-204 100% LESS THAN 0.4

2. NEVA-CLOG FILTER MATERIAL 1S BY MULTI-METAL WIRE CLOTH, INC., AND CONSISTS OF
TWO PERFORATED SHEETS FASTENED TOGETHER WITH ABOUT A '44in. SPACING
BETWEEN THEM, AND ARRANGED SO THAT THE PERFORAT ONS IN ONE SHEET
ARE NOT IN LINE WITH THOSE OF THE OTHER. THE PERFQRATIONS
ARE 0.044 in. DIAM ON 043 in. CENTERS, EQUIVALENT TO 10% OPEN AREA

FILTER TYRPE F-1

  
 

L

 

 

 

 

 

T ' FILTER ELEMENT TYPE G FILTER ELEMENT TYPE 2232
—d L BY PALL TRINITY MICRO CORP., BY HOKE, INC., SINTERED
GLASS PIPE | T SINTERED STAINLESS STEEL, STAINLESS STEEL, FILTER
3in iD ! } 2.75 in. DIAM X 6in. LONG, FILTER AREA 0.006ft2 MFR'S
BY ; | AREA-0.36 12, MFR'S REMOVAL RATING 5-9 p
8in. LONG -] ! 1 RATING 98% 0.7x, 00% (.8 L-—S/:n—l
JI | 8"
i i FLOW = 0.15 scfm/ft2 FLOW = 8.8 scfm /f12
L-—-d AT 4.5 liters (STP)/min ‘ AT 1.5 liters (STP)/min
D — |
FILTER TYPE F-2 FILTER TYPE F-3

Fig. 22. Filters used during off-gas tests in NaBF, circulation test,
PKP loop.
62
~

ORNL-DWG 72-2086

3/-in. 0D TUBING

7
LINEAR VELOCITY OF GAS AT 1.5 liters/min (STP)

GLAS_S PIPE AND 24 psig WAS 0.076 fps AND RESIDENCE
1[_;'4,:(' 0 TIME IN {{in. COLD SECTION WAS 12 sec
18in. LONG —al

T

™

/WET-ECE BATH

1in.

 

 

 

 

2in.

 

 

 

 

 

)
WET-ICE COLD TRAP (CT-WI)

X .
2in. LINEAR VELOCITY OF GAS AT {.5liters/min {STP)
N AND 24 psig WAS 0.75fps AND RESIDENCE
TIME WAS 2.5 sec

L~ Y5-in.-0D SS TUBING

10 in.

l«—DRY-ICE-TRICHLOROETHYLENE BATH

 

 

 

 

 

 

 

 

)
DRY-ICE COLD TRAP (CT-DI)

Cold traps used during off-gas system tests, NaBF, circula-

Fig. 23.
tion test, PKP loop.
63

ORNL-DWG 72-2087

HEATER

_ FILTER
(F-1a)
TEST la
12-49-68 TO t-{3-69
PUMP BOWL
WET~ICE COLD TRAP
(CT—-wWD

SETTLING TANK (ST)—"

 

-

 

 

 

 
 
  

N

FILTER
(F-1b)

TEST Ib IMPACTOR I-f
1-13-69 TO {-22-69

 

  

PUMP BOWL
TEST Ib IMPACTOR I-2

IMPACTOR {-22-69 TO 2-9-69
A (I-1, 1-2)

 

 

 

FILTER
(F-1b)
HEATER

TEST Ic
2-9-69 TO 3-7-69
PUMP BOWL

Fig. 24. Arrangement of off-gas test equipment, tests Ia to Ic,
NaBFy, circulation test, PKP loop.

 
6k

ORNL-DWG 72-2088

TEST ILa
3-25-69 TO 5-13-69

 

NOTE:
HMT-V{ WAS EQUIPPED
WITH DISK AND DONUT
BAFFLE ARRANGEMENT

 

    

 

 

 

.. PUMP BOWL
~\, -=dd
\\\\ WET-ICE COLD TRAP
T~ (CT-WI)
TEST IOIb

5-26~-69 TO 6-26-69

NOTE:
HMT-v2 WAS EQUIPPED
WIiTH 14.3q9 WIRE MESH
(YORK MESH} COMPACTED
IN 2 LINEAR in. GIVING A
VO!D FRACTION OF 93 %

 

 

   
  
    
 

PUMP BOWL
™~ WET-ICE COLD TRAP
T (CT-WI)
HMT-V3 TEST Oc

6-26-69 TO 10-30-69

NOTE:
HMT-V3 WAS EQUIPPED
WITH A SINGLE 34in. DIAM
BAFFLE AT INLET END

 

PUMP BOWL

WET-ICE COLD TRAP
{(CT-WI)

Fig. 25. Arrangement of off-gas test equipment, tests Ila to Ilc,
NaBF, circulation test, PKP loop.
 

65

ORNL-~DWG 72-2089

-‘ABOVE SALT MELTING POINT J' BELOW SALT MELTING POINL

 

 

i

|
o
|
| EFILTER
[@ 7
|
L e—)—300°F
*

SHOT-MIST TRAP
MORIZONTAL MODEL

 

 

1000°F

 
 
  
  

 

1]

 

WET-ICE COLD TRAP

 

(CT-WI)
PUMP BOWL
TEST TIME PERIOD HOT MIST TRAP HOT ELEMENT COLD ELEMENT
IIlo {1-7-69 TO 11-24-69 HMT-HA1 FM=-225 - NONE USED
11Ib 12-5-69 TO 12-23-69 HMT-HZ2 NEVACLOG NONE USED
11I¢ {-20-70 TO 3-4-70 HMT-H3 Fm-225 FM-225
1I1le 3-24-70 TO 4-13-70 HMT-H3 FM-225 FM-225

Fig. 26. Arrangement of off-gas test equipment, tests IIIa to IIIc,
NaBF, circulation test, PKP loop.

 

 

 

 
 

Table 9. Chronological summary of tests and data collection rates for

salt mists and condensed liquid

 

Collection data

 

 

 

 

 

 

 

Cumulative
! Test Date clrculating Remarks Component Total time Total weight Average Pr
time onr stream collected rate ph::z?
(br) (br) (&) (/nr)
In 12-19-68 Test facility started up after modifying off-gas line to
: investigate off-gas emissions; collection equipment in-
cludes settling tank (ST), wet ice cold trap (CT-WI),
; and filter {Fl-a)
| 1-13-69 622 Settling tank, wet ice cold trap, and filter removed from ST 622 63 0.10 s
systen CT-WI 622 L 0.007 L
i Fl-a 622 -8 0.013 s
: Tt 11369 Thpactor umlt (I-1) and Til¥er (FI-0) installed
1-15-69 670 Pressure drop at pump bowl outlet is 6.5 psi; restriction
cleared by increasing temperature of off-gas line at
pump bowl outlet
1-22-69 830 Impactor unit, I-1 removed I-1 210 97 0.6 S
1-22-69 Impactor unit I-2 installed F1-b 210 1k 0.07 5
1-30-69 1020 Pressure drop at pump bowl outlet is k.5 psi; restriction
cleared by increasing temperature of off-gas line at
pump bowl outlet
Ic 2-9-69 1250 Impactor unit I-2 removed because of restriction in inlet I-2 430 5 0.012 s
line
3-6-69 1870 Pressure drop at pump bowl outlet is 8.0 psi; restriction
cleared by heating off-gas line
3-7-69 1900 Test facllity shut down to install hot-mist trap Fl-b 1070 y 0.004 s
IIa 3-25-69 Test faclility started up after installation of vertiesl
hot-mist traps with internal disk-donut baffles (HMT-V1);
collection equipment includes sintered metal filter PF-2,
absolute filter Fl-c, and wet ice cold trap CT-WI
4-21-69 650 Removed filter Fl-c Fl-c €50 0.02 0.0003 -
5-2-69 900 Replaced sintered metal filter F-2 with sintered metal F-2 900 T 0.008
filter F-3; reinstalled filter Fl-c
5-13-69 11ks Test facility shut down to change hot-mist trap HMT-V1 1145 9 0.008 S
F-3 2Ls5 5.1 0.02 8
Fl-c¢ 255 0 0 s
CT-WI 1145 T 0.006 L
I 5-26-69 Test facility started up with vertical hot-mist trap
with wire screen demister (EMT-V2); collection equip-
ment includes sintered metal filter F-3 and wet ice
cold trap CT-WI
6-26-69 1910 Test facility shut down to change hot-mist trap F-3 765 15.1 0.02 5
CT-WI ThO 6.4 0.009 L
1Ie 6-26-69 Test facility started up with vertical hot-mist trap
with single haffle plate at inlet; collection equip-
ment includes filter Fl-c¢ and wet ice cold trap
CT-WI
10-30-69 K610 Test facility shut down to install horizontal hot-mist Fl-c 2700 50 0.02 S
trap CT-WI 2700 12 0.005 L
IIIs 11-7-69 Test facility started up with Tirst horizontal hot-mist
trap (EMT-H1); collection equipment includes sintered
metal filter F-3 and wet ice cold trap CT-WI
11-24-69 k1o Test facility shut down to change hot-mist trap HMP-H1 k1o 5.3 0.013 s
- F-3 320 2.5 0.008 s
CT-WI ho 3.3 0.008 L
O
ITTv 12-5-69 Test facility started up with second horizontal hot-mist N
trap (HMT-H2); same collection equipment
12-23-69 8h0 Test facility shut down to change hot-mist trap HMT-H2 430 1.8 0.00k - s
F-3 k30 1.8 0.00k 5
CT-WI k30 2.5 0.006 L
II1c 1-20-70 Test facility started up with third horizontal hot-mist
trap (HMT-H3); same collection equipment
3-Lk-70 1840 Test facility shut down to change hot-mist trap and pre- HEMT-H3 1000 33 0.03 8
pare for water injection test F-3 1000 0.4 0.000L s
CT-WI 1000 5.6 0.006 L
3-24-70 Test facility started up to run water injection test; :
hot-mlst trap is same as in test IITe; collection equip-
ment includes wet ice cold trap CT-WI and dry ice cold
trap CT-DI
h-o-T0 2055 Check trap weights CT-WI 215 1.8 0.008 L
CT-DI 200 0.24 0.012 L
k-2.70 2056 Injected 10 g H,0 1nto loop CT-WI 265 2.7 0.01 L
4-13-70 2320 Test facility shut down CT-DI 265 ) 0.85 0.003 L
s

 

8,

S = salt; L = liquid.
67

line was at the salt melting point (725°F) at sbout 1/4 to 1/2 in. away
from the pump bowl. By applying power to the heater, the temperature
profile could be shifted so that the temperature of the line did not drop
below the salt melting point until a point 4 in. or more away from the
pump bowl. At 5 1/2 in. from the pump bowl, an adapter was installed
that converted the off-gas line from 1/2- to 5/16-in.-ID tube. About

30 in. downstream of the adapter, traps and filters were installed to
separate and collect the materials that were emitted with the off-gas
stream.

The facility was operated with the above arrangement for 1900 hr
(Table 9). The procedure was to operate without power to the off-gas
line heater until excessive pressure drop at the pump bowl outlet indi-
cated the formation of a salt plug. The heater was then turned on and
the temperature of the line increased until a sharp decrease in pressure
drop indicated that the plug had melted. Using this method, restrictions
at the pump bowl outlet were cleared sfter 670, 1020, and 1870 hr of cir-
culation (Fig. 27). At the start of the tests, the collection equipment
consisted of a 3-in.-ID glass settling tank (ST), a 1-in.-diam wet-ice
cold trap (CT-WI), and a porous metal filter (Fl-a). After 622 hr the
settling tank and cold trap were removed, and during the next 630 hr
two different models of impactor units were tested. After 1250 hr the
second impactor unit was removed, and the final 650 hr of operation
(test Ic) were completed with only the porous metel filter (Fl-b) in the
line. Table 9 summarizes solid and liquid collection rates for this test
period. After completion of test Ic, a salt plug was found in the irnlet
to the porous metal filter (Fig. 28).

Following completion of the first group of <ests, it was concluded
that the best approach to the salt mist problem was to use a hot-mist
trap, that is, one that is operated at a temperature above the salt
melting point, to collect a large fraction of the salt mist and return
it continuously to the salt system or store it until suitable disposition
could be made. The salt overflow, that is, the fraction not separated
out by the hot-mist trap, would be cooled below the salt melting point
and the frozen salt particles removed by a replaceable filter. The ob-
Jective of the second series of tests was to obtain performance data on

hot-mist traps using different internal arrangements for interception
68

ORNL -DWG 72-2090

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

=15

Q
u. S
Zw l
2 '
0 Bl HEAT APPLIED TO RELIEVE RESTRICTICON
w 2 l
<l !
ol i
e '
a -
23 ‘ A
[~ :
g3 s ¢ L
0 P
W ' !
05 i ! : \
& 5 | ‘_/T/\ //\

8 : L |

0
20 26 { 7 i3 1Q 25 Rl 6 i2 18 24 2 8
DEC. JAN. FEB. MAR.
1968 1969

Fig. 27. Pressure drop across pump bowl off-gas outlet nozzle, varia-
tion with time.
o4

-}

«)

 

 

. Fig. 28. Salt plug found in 5 1/16-in. line at filter inlet, Mar. 7,
1969. NaBF, circulation test, PKP.loop. - T -

     

 
70

of the salt mist. The trap housing consisted of a 1-in.-ID pipe, 13 in.
long (Fig. 20), vertically oriented. The lower part of the trap and the
off-gas line between the pump bowl and the trap were heated and insulated,
so the off-gas temperature was held above the salt melting point until
the gas had reached the upper part of the trap. Initially (test TIa),

the hot-mist trap was equipped with baffle plates drilled alternately at
the center and periphery to create continuously reversing cross flow be-
tween plates. A total of nine plates was used, with 1/4-in. spacing
between plates.

For test IIb, the baffles were removed, and a 2-in.-long bundle of
wire mesh (York mesh) weighing 14.3 g was installed. TFor the final test
(test IIc), the wire mesh was removed and the hot-mist trap was operated
without any internal hardware except for a 3/h-in.-diam deflector plate
at the inlet end. Three different types of porous metal filter were used
for solids ceollection during the series 11 tests. Operating data for
tests IIa, ITb, and TIlc are presented in Table 9. The tests were started
on March 25, 1968, and completed on October 30, 1969; total circulating
time was 4600 hr. Test IIc was interrupted twice because of salt plugs
at the hot-mist trap outlet and once because of a salt plug at the inlet
to the porous metal filter.

In the final series of tests (IITa to IIId), investigation of the
hot-mist trap idea was continued, but changes were made in the geometry,
orientation, and demister material. The size of the trap was increased
to 2 in. diam by 18 in. long, and it was turned so that the long axis
was horizontal (Fig. 20). In the first test (IIIa), the demister mate-
rial was porous metal type FM-225 (Huyck Metals, Inc.), while in the
second test (IIIb) it was a perforated metal sheet material called NEVA-
CLOG (Multi-Metal Wire Cloth, Inc.). For the third test (IIIc), the
IM-225 material was again used, and a second piece of FM-225 was installed
in the cold end of the mist trap to catch the frozen salt particles.
Descriptive data for the various demister materilials are given in Fig. 22.
In all cases, a porous metal filter (F-3) was used for solids collection
at the outlet of the hot-mist trap. After both tests IIla and IITh, each
of which lasted about 400 hr, the outlet line from the hot-mist trap was
71

found to be blocked with a black material, although only IITb showed a
significant increase in pressure drop across the hot-mist trap (2.3 psi
as compared with a clean AP of 0.4 psi). The black deposit was not
found after test 11Ic, and no appreciable rise in AP was noted during
the 1000 hr of operation. The final test (IIId) was a special test to
determine the effects of injecting water directly into the salt in the
pump ktowl. The hot-mist trap was the same design as that used for test
ITIc.

For both the series II and IIT tests, the collection equipment was
essentiglly the same, that is, a high-efficiency porous metal filter
followed by a wet-ice cold trap. The type of porous metal filter was
varied in some cases (see Figs. 25 and 26 and Table 9), and in the final
test (ITId), the porous metal filter was not used and the wet-ice cold
trap was followed by a dry-ice—trichloroethylene trap. Operating and

collection data for the series IIT tests are shown in Table 9.

5.4.3 Discussion and evaluation of test results

 

The off-gas system tests are subject to the valid criticism that
they were too broad and shallow. That is, a limited amount of experience
was obtained over an excessive variety of test conditions. This lack of
in-depth study argues for the use of caution in reaching conclusions hased
on the test results. With this in mind, we propose to examine the test
data first on a qualitative basis and then to draw whatever quantitative
conclusions seem permissible.

Tests Ia to Ic verified the fact that the plugs or flow restrictions
tend to form at the first place in the line where the temperature falls
below the melting point of the salt. The tests also showed that the
restrictions could be cleared by raising the temperature of the line.
Although we did not have a measure of the salt temperature at the exsct
points where the plugs had formed, we noted that the line temperature
did not exceed 1100°F during any of the times when heat was applied to
clear a restriction. We conclude from this that the plugs were probably

NaBF, -NaF eutectic, or some other mixture of NaBF, and NaF, rather
—J
e

than oxide, which would tend to have a melting point considerably higher
than 1100°F. We noted also that other tests (see Sect. 5.2.2) implied
that the off-gas stream tended to be in equilibrium with the bulk circu-
lating salt, and, since the off-gas would be in continuous contact with
deposits in the off-gas line, deposited mixtures of NaBF, and NaF would
tend to approach the bulk salt in composition. ZFinally, we noted that
chemical analyses of solids deposited in the off-gas line (sample PK-20,
11-4-68) corresponded closely to the NaBF,-NaF eutectic composition.

Tests Ila to Ilc and IITa to IIIc proved that restrictions could
be eliminated if the line were maintained at a temperature above the
salt melting point and 1f the line were pitched so as to prevent trapping
of the molten salt in low spots, as evidenced by the fact that no flow
restrictions were experienced in the heated line between the pump bowl
and the hot-mist traps. The occurrence of salt plugs at the hot-mist
trap ocutlet during test IIa indicated that the simple settling tank was
not as effective in removing salt mist as the baffles or the wire mesh.

Tests ITIa to IIIc seemed to indicate that either Huyck Felt Metal
or Multi-Metal NEVA-CLOG could be ugsed as an effective demisting agent.
This conclusion is btased on the lack of significant solids accumulations
in the porous metal filter (F-3) downstream of the hot-mist traps and,
in the case of test IIIc, on the filter element in the cool end of the
hot-mist trap. We noted, however, that the 5/16-in.-ID line at the out-
let of the hot-mist trap (upstream of filter F-3) was plugged with a
dark-solid material after both tests IIIa and IITb. No plug of dark
material was found after test IITc, which had a Huyck Felt Metal cold
element in the cool zone of the hot-mist trap. Also test IIlc was op-
erated more than twice as long as IITla and IITb without an appreciable
increase in pressure drop over the "clean" value. Thus the clean AP at
startup was 0.5 psi, while the AP after 1000 hr was 0.75 psi. The source
of the black deposits is not known, btut they possibly could have been
due to decomposition products of lubricating oil that inadvertently
leaked into the pump bowl; we have no positive evidence, however, that
such an oll leak occurred.

In order to make intelligent use of the quantitative data from the

off-gas system tests, it is necessary to give due consideration to the
T3

shortcomings that were present in the experimental design. First, the
procedure did not provide for periodic checks to insure that the mass
flow rate and physical properties of the salt mist in the off-gas stream
at the pump bowl outlet did not vary significantly with time. Also, it
seems logical to assume that the flow of salt mist into the test section
would need to be adjusted for holdup of salt in the line connecting the
pump bowl to the first component in the test section. The fraction held
up would drain back into the pump bowl or it would freeze and settle out
in the line, depending on whether or not the.line was heated. The flow
rate of salt mist into the first test component would be the flow rate
out of the pump bowl times the fraction not held up in the intervening
line. In comparing data taken over different time intervals, therefore,
we make the arbltrary assumption that for the periods of interest, the
flow rate of salt mist out of the pump bowl and the fraction of salt
mist held up in the line connecting the pump bowl to the first component
in the test section were essentially constant. In addition, we assume
that there were no major changes in the size range and density of the
mist particles.

Second, the collection rate determinations were susceptible to large
errors. This is because the quantity of material collected was usually
on the order of 1 to 10 g, while the weight of the collection component
was on the order of 100 to 5000 g. A 1% error in weighing, therefore,
would cause a 100% or larger error in the weight of material collected.
An attempt was made to alleviate this problem by making weighings on a
large analytical balance that had a sensitivity of 0.02 g with a 2000-g
load.

Third, there was some settling out or trapping of solids in the
5/16-in.-ID lines used to connect test components. The deposition was
greater at places where acceleration forces were present, such as at
90° bends. 1In cases where the performance of an upstream component was
based on data obtained from downstream componénts,»the figures may be
considered to be in error by some amount due to the trapping of material
in the intervening lines. We do not have an estimate of the size of the
error introduced by this effect; 1t would tend to be in direct proportion

to the solids flow rate. In the series IT tests, there was a tendency
7h

for solids to accumulate in the outlet line of the hot-mist trap upstream
of the filter. Where possible, this material was collected, and weights
were gdded to the filter weights in order to determine the filter collec-
tion rates. The tendency to settle out on the surfaces of interconnecting
lines may also have affected the liquids collection data. One would ex~
pect the error to be proportional to the amount of solids present since
this might greatly increase the available surface.

In the first series of tests (Table 9), the off-gas line between the
pump bowl and the first test component was normally cold and the solids
collection data are biased low becguse of the fraction retained in the
line. Of the total material collected in the settling tank and filter
during test Ia, about 90% was retained in the settling tank. We can
predict from Stokes law that particles larger than 10 u should have re-
mained in the settling tank (assuming spherical particles of density
2.0 g/cem® ). After the test, the solids were piled up under the inlet
nozzle,7 so 1f the assumptions regarding particle size and density are
reasonable, it appears that the average particle size may have been sig-
nificantly larger than 10 u or inertial forces due to the relatively high
inlet velocity (1.6 fps) may have had g pronounced effect. The solids
were tan and tended to "pall up" or stick together in clumps (rather than
the brilliant white free flowing appearance characteristic of NeBF, -NaF
powder). We attribute this uncharacteristic appearance to adsorption of
acld vapors and corrosion products that were emitted from the pump bowl
along with the salt mist.

In test It the collection rate for impactor I-1 was about five times
that for the settling tank in test Ia; the collection rate for impactor
I-2 was only about one-tenth that of the settling tank. In test Ic, when
only a filter was used, the collection rate was even lower than that of
impactor I-2. We consider the results of tests Ib and Ic to be completely
anomglous. There was no reason to expect any difference in performance
between the two impactors because they differed only in the diameter and
length of the housing. And the collection rate in test Ic should have
been the highest of all, because g high efficiency filter was used. We
attribute these strange results to holdup of salt mist at the pump bowl

off-gas outlet nozzle and in the intervening lines. If the cumulative
5

total of solids collected during the series I tests (191 g) is divided
by the total circulating time (1900 hr), a value of 0.1 g/hr is obtained
for the average collection rate. This should represent a lower limit for
the flow rate of salt mist into the test section. If the calculation is
repeated using only the numbers from tests Ia and Ib (182/830), we get

an average value of 0.2 g/hr, which we feel is probably the best estimate
we can obtain from the test data for the rate at which salt mist was
flowing from the pump bowl. In evaluating the series IT and III tests,
we assume that the flow rate of salt mist from the pump bowl was constant
at 0.2 g/hr and that the holdup in the line between the pump bowl and the
salt-mist trap was constant; thus the performance of the hot-mist traps
was directly proportional to the solids collection rate in the filters.

Table 9 shows that the solids collection rates in the filters during
the series IT tests were only 10 to 15% of the rate (0.2 g/hr) at which
the salt mist was estimated to be leaving the pump bowl. As noted pre-
viously, these filter collection rates include the material found in the
line between the hot-mist trap and the filter. Note, however, that in
test ITa, 9 g of solids were found in the cool upper part of the hot-mist
trap. If this material was added to the filter collection rate, the col-
lection rate for test Ila would be increased almost 50%. We conclude
from the series II tests that a significant reduction in the transfer of
salt mist to the filters was achieved in the case of all three models.
The fraction of salt mist that was transmitted, however, tended to collect
in the line upstream of the filter, and, in the case of test IIc, these
accumulations resulted in severe flow restrictions.

Table 9 also shows that the collection data for the series I[II tests
include quantities collected in the horizontal mist traps (IMT-H1, HMT-H2,
and HMT-H3). This was salt that passed through the hot demister element
but was retained in the hot-mist trap on the downstream side of the hot
demister element rather than being carried out with the off-gas stream.
For purposes of evaluating pérformance, this material must be added to
the materisl collected in the filters. A qualification is necessary in
the case of test IIIc, however, because the horizontal mist trap con-
tained a cool zone filter element that was not used in the IITa and IIIb

models. We were not able to measure the amount of material retained on
76

this cool zone element, but visual inspection indicated thét the quantity
was small. If the quantity had been 1 g or more, it should have been
highly visible, so we decided for comparison purposes to use a vglue of

1l g or 0.00L g/hr for the 1000-hr period. Considering these factors the
collection rates were 0.02, 0.0l (nearly), and 0.03 g/hr, respectively,
for tests ITIa, ITIb, and IIIc. We conclude that the efficiency of the
horizontal hot-mist traps was about the same as that of the vertical hot
mist. The horizontal hot-mist traps, however, gave less difficulty

with flow restrictions (due to salt), apparently because the salt mist
which passed through the hot demister element was retained in the mist
trap housing. In test IlIc, this material was found at the bottom of the
housing in the form of solidified droplets rather than crystals or powéer
which one would expect if frozen salt particles had settled out from the
gas stream. We conclude that this accumulation of material resulted from
the draining of droplets of salt from the downstream surface of the hot
demigter element. If this iIs true, then the efficiency of the horizontal
hot—mist'traps may have been higher than was indicated by the collection
data.

Except for the latter part of the series I tests, the off-gas stream
was passed through a wet-ice cold trap {CT-WI) during the entire test pro-
gram, and a viscous dark-brown fluid was routinely accumulated. The final
test in the program (IIId) was designed to determine how the liguid col-
lection rate would be affected by a slug of water injected into the salt
pool in the pump bowl. For this test a dry-ice—trichloroethylene trap (CT-
DI) was used as a backup for the wet-ice trap. The liquid collection data
are summarized in Tabtle 9. It may be seen from the table that the liquid
collection rate varied randomly between 0.005 and 0.009 g/hr over the
7500 hr of operation, as though the emission was due to the continuous
introduction of a contaminant (such as water in the By or helium gas
supply). Thus, if the contaminant is considered to be water and the aver-
age collection is assumed to be 0.007 g/hr, then the required volumetric
input of water would be

(94581-5)(92.4 liters/18 g) ~ 0.0087 liters/hr .
T

If the gas flow is F liters/hr and the concentration of water i1s C ppm,

then the required concentration would be

(C x 10~° liter/liter)(F liter/nr) = 0.0087 ,

from which C = 8700/F. If we apply the normal gas flow rates to this

equation, the results are as follows:

 

Required
concentration of H,0
Normal flow in the gas
Gas (Liters/hr) (ppm)
BEF, 3 13900
Helium 87 100

If we assume that the trapped fluid is BFs .2H,0 rather than HyO, then

only one-fourth as much water would be required and the required impurity
concentration would be 475 ppm for BF; and 25 ppm for helium. Analysis

of the BFs; gas 1lndicated t@at the water concentration was less than 1 ppm.
The water content of the helium was supposed to be less than 1 ppm, but
samples taken from the helium supply header at the test facility frequently
indicated water levels of as much as 25 ppm. We concluded therefore that
the helium gas supply was most probably responsible for the continued low-
level emission of condensable fluid in the off-gas stream. The conclusion
was reinforced during the post-test examination of the pump, when evidence
of severe erosion was noted on the inner heat baffles near the point where

the shaft purge gas helium entered the pump (see Sect. 6.10).

5.4.4 General conclusions from off-gas tests

 

The following is a summary of the pertinent conclusions resulting
from evaluation of the off-gas system tests.
1. The prevention of flow restrictions due to salt-mist deposits
requires a twofold spproach. First, a hot-mist trap, that is, one that
operates above the salt melting point, would be installed at the pump

bowl outlet to remove as much of the salt mist as possible from the gas
78

stream. The fraction of salt mist removed would be either returned to the
pump bowl or stored until it could be returned to the system or discarded.
Second, the gas emerging from the hot-mist trap would be chilled to freeze
residual salt particles and then passed through a filter to remove the
frozen particles. The hot-mist trap would be designed for continuous op-~
eration. The particle filter would be arranged so that the filter unit

or the filtering material could te replaced when necessary. The pipeline
between the pump bowl outlet and the hot-mist trap would be operated at a
temperature above the egalt melting point and would have a continuous re-
verse pitch to permit deentrained salt to drain back into the pump bowl.

2. The salt-mist overflow, that is, the fraction (based on the flow
rate of salt mist at the pump bowl outlet) of salt mist not removed by the
hot-mist trap, was 15% or less for all the hot-mist trep models tested.
This would imply =85% removal efficiency, but it was not clear what part
of the removal was due to the demisting action of the hot-mist trap and
what part was due to gravitational settling in the pipe connecting the
pump bowl to the hot-mist trap. Of the different demister mechanisms
tested, the one using a porous metal screen (Felt Metal 225 by Huyck, Inc.)
appeared to have an advantage in that the pressure drop remained acceptably
low over 1000 hr of operation and most of the salt which passed through the
screen settled ouf as droplets in the trap.

3. The filter element for the particle filter should be fabricated
from a porous metal screen {such as Huyck Felt Metal) that has a removal
rating, in gas service, of not less than 95% of particles larger than 1 i
This material should be stainless steel.

4. In the operation of the particle filter, the frozen salt particles
may tend to settle out in the region upstream of the filter element. To
minimize flow restriction problems in this area, the distance between the
hot-mist trap and the particle filter should be kept as short as possible,
and the connecting pipe should be of uniform cross section and, if possible,
without bends. TIsolation valves should te btall or gate type.

5. 'The continuous low-level emissions of viscous brown fluid were
probably caused by moisture levels of gbout 25 ppm in the helium supply

gas. The presence of this viscous fluid in the off-gas stream at this
9

low level (0.007 g/hr or about 0.000l g/liter) did not cause an operating
problem (e.g., sticking pressure control valve) as long as there was no
salt mist. ©Since corrosion rates in fluorcborate systems are directly
proportional to the water level, the helium supply system should contain
a dryer designed to malntain the moisture content at a very low level,

with an upper limit of 1 ppm by volume.

6. CHRONOLOGY, PERTINENT OBSERVATIONS, AND OTHER TESTS

6.1 Chronology

Planning for the test program was started in mid-1967. By the end
of that year, a tentative program had been formulated, and work was in
progress on the necessary facility modifications. Figure 29 is a chrono-
logical summary of test activities starting with January 1968. Four
batches of salt, totaling 689 1b, were charged into the drain tank during
February. Initial filling and salt circulation occurred on March L, 1968.
Loop operation was divided into s flushing salt period, extending from
March through June 1968, and a main test period, extending from August
1968 through April 1970.

The flushing salt operation was originally scheduled to be completed
in about a week. However, a number of problems arose that required study
and resulted in severgl shutdowns for repairs and design changes, and
this phase of the test lasted about 115 days. The following is a brief
discussion of significant events.

1l. While running cavitation tests, a number of ingassing transients
occurred when the pump cavitated at unexpectedly high overpressures (see
Sect. 4.1 and Table 10). Impeller ingsssing is the pumping of gas from
the pump tank along the top side of the pump impeller and into the circu-
lation system salt. It is generally thought to occur whenever the head
produced by the working impeller is lower than that produced by the im-
peller top surface and is caused by pump operation either at excessive
flow rates or in deep cavitation. When an ingassing transient occurred,
there was a sudden increase in circulating volume, gnd this 1in turn
caused a rise in salt level in the pump tank. The pump was shut off

within a few seconds, which stopped the ingassing, but the transients
JAN
FEB
MAR
APR
MAY
JUN
1968

JUuL
AUG
SEP
ocT

NOV

DEC

TSI

i gy o T 0O D D A D M 0 G o o o e L D e D D AR R 0
o T I i 0 i e 1 o P 7 A o A D O A AP 0 A o ke o ¢

T
Y

2318
T3

T

3 FINISH UP SYSTEM REVISIONS
[ CHECK QUT SYSTEM
o0 TRANSFER SALT INTO DRAIN TANK

    

80

CAVITATION TESTS—FLUSHING SALT
PUMP INGASSING

GAS LINE PLUGGING PROBLEMS
INSPECT PUMP ROTARY ELEMENT
GREEN SALT

REVISE OFF-GAS VENTING SYSTEM
INSTALL MIXED-GAS BF3 FEED

 

CCT 973hr

REMOVE FLUSHING SALT
TRANSFER CLEAN SALT TO SYSTEM

COMPOSITION CONTROL TESTS

CCT 2787 hr

INSTALL OFF-GAS TEST SECTION
AT PUMP BOWL OUTLET

 

]CAVITATION TESTS—CLEAN SALT

 

JAN

196%
FEB

Fig. 29.

 

 

 

 

[ e - e - O O O
ety

=2

]OFF -GAS FHTER TESTS

 

CCT 4680nr

 

INITIAL CIRCULATION MAR. 4,1968
FINAL DRAIN APR 15,1970
TOTAL ELAPSED TIME 18,480hr

CCT = CUMULATIVE CIRCULATING TIME

Chronology — sodium fluoroborate

 

ORNL-DWG 72-209¢

}INSTALL HOT MIST TRAP

CCT 5825hr

 

PCOLD FINGER TESTS
MIST TRAP TESTS

CCT 8692 hr

 

 MIST TRAP TESTS

 

o

 

CCT14,089 br

14,
}WATERiNJECHON TEST

CIRCULATING TIME (hr)
FLUSH SALT 973
CLEAN SALT 10,594
TOTAL 11,567

circulation test.
81.

were sufficiently rapid and enough gas was ingested so that the pump tank
was filled with salt and salt was forced into the connecting unhested gas
lines. Each transient, therefore, resulted in a program delay for clean-

out or replacement of plugged gas lines.

Table 10. Ingassing transients for NaBF, circulation test,

 

 

 

PKP loop
Pump time (hr)
Date Ttem
Cumulative Net
3-7-68 23 23 Ingassing transient
3-13-68 43 20 Ingassing transient
L-9-68 187 Remove, inspect, and reinstall rotary
element; discovery of first deposit
of green salt
5-28-68 7o 285 Ingassing transient
6£-18-68 736 264 Ingassing transient
6-24-68 &72 136 Ingassing transient

 

2. Continuing difficulty was experienced with gas line restrictions.
The trouble was due in part to the ingassing transients, which forced salt
into the pump bowl gas connections, and in part to salt impurities and
salt mist which were being swept out of the pump bowl by the purge-gas
stream (see Sect. 5.4).

3. The pump rotary element was removed to check for an oil leak.
No leak was found, and the apparent loss of oil from the lube-oil reser-
voir was attributed to a shift of inventory from the reservoir to the
pump bearing cavity. When the pump rotary element was removed for inspec-
tion, several large chunks of green salt were found lying on top of the

impeller casing (see Sect. 6.11 for further details on this material).

L. BF, was normally fed to the system through a bubbler dip tube
inserted in the pump bowl. The back pressure in the tube was also used
a8 an indicator of pump bowl salt level. It was desired to keep the

bubbler flow uniform in order to avoid the need for level corrections
ge

due to flow variations and to keep the bubbler flow rate about the same

as that used in the MSRE coolant pump (370 cmf/min) for simulation. The
original calculations indicated that a BF; flow of about 300 cmP/min
would be required, and the system was designed on this basis. OSubse-
quently, an increase was made in the normal setting for the pump bowl
total overpressure, and revised BF; partial pressure data were received;
these changes reduced the normal BF, flow rate (at 1025°F) to 50 cmP/min.
In order to maintain the desired total flow, a mixed-gas system was in-
stalled in which the total flow was controlled at 370 cn?/min, and the
composition of the gas mixture (ratio of BFé/He) was adjusted as dictated
by the salt temperature, the pump bowl total overpressure, and the total
gas flow. The mixed-gas BF, feed system was installed during the shutdown
for inspection of the salt pump rotary element; Section 5.5 describes the
gas system as it appeared after this change was made. The use of a mixed-
gas BF, meant that the normal gas composition in the bubbler tube (at
1025°F) was about 12.5% EF, by volume, whereas the normal composition (at
1025°F and 2L psig) of the pump bowl atmosphere was about 3.5% by volume.
See Section 6.5 for a discussion of the effect of gas composition on
bubbler tube operation.

During July and August 1968, the flushing salt (702 1b) was trans-
ferred out of the system and replaced with a fresh charge of sodium
fluorcborate (765 1b). Pretreatment of the fresh charge was modified in
an attempt to reduce the water content (Sect. 6.3). TInitial circulation
with the clean charge of salt was on August 19, 1968. September and
October 1968 were spent in performing cavitation tests and in evaluating
the performance of the thermal conductivity cell to indicate salt compo-
sition. A severe restriction in the off-gas line at the pump bowl outlet
forced a shutdown of the loop on November 4, 1968. The off-gas piping
was revised to provide a test section for studying the problem of off-gas
restrictions, and operations were resumed on December 19, 1969. From that
point until the firal shutdown in April 1970, the principal activities
were evaluation of the cold zone deposition technique (Sect. 5.3) and
evaluation of various techniques for coping with the problem of off-gas
system restrictions (Sect. 5.4). The total salt circulation time was
11,567 hr, of which 973 hr were with the flushing salt. From March 1968

until November 1968, the set point for the circulating salt temperature
83

was varied frequently to accommodate cavitation inception tests and other
special tests; settings used during this time interval were 900, 1025,
1150, and 1275°F. After November 1968, the salt circulating temperature
was maintained at a constant setting of 1025°F. 1In September 1970, the
pump rotary element and the salt level bubbler tube were removed from the

system for inspection. Findings are summarized in Section 6.10.

6.2 Analyses of Salt Samples

 

Samples of salt were taken from the pump bowl using equipment as de-
scribed in Section 4.3. Thirteen samples were taken during the flushing
salt period, three in duplicate. Sixteen samples of the clean batch were
taken, all but one in duplicate. Routine salt samples were not taken
after June 1969. Normally the samples were transferred to the analytical
laboratory in a covered glass Jjar that had been flushed with argon. Sam-
ples taken during and after April 1969 were delivered to the analyst with-
out removing them from the protective pifie housing that is part of the
sampling unit. The purpose was to insure meximum protection against ex-
posure to moisture in the atmosphere. In addition to the routine samples,
a number of special samples were taken during the test program, mostly to
identify materials found in the off-gas system.

All pump bowl samples were analyzed for Na, B, and F and for the
common corrosion constituents Fe, Cr, Ni, and 0. A number of the samples
were analyzed for the constituents of the salt formerly used in the loop
(Ii, Be, U, and Th), and samples taken during and after April 1969 were
analyzed for HQO. Average analytical values for the basic constituents

are shown below.

Concentration (wt %)

 

Na, B F Ii Be U Th

 

 

Flushing salt 21.2 9.26 66.9 0.20 0.18 0.26 0.24
Clean salt 20.9 9.33 68.0 0.048 0.016 0.019 0.093
8l

Values for the impurities found in the salt samples are plotted in
Fig. 30. The oxygen results were quite scattered and did not seem to
correlate with anything in particular. Only a few samples were analyzed
for H,0, and the level seemed to increase with time. Ixcept for the
high-temperature operating period following addition of the clean salt
to the system in August 1968, the results for Fe, Cr, and Ni were repeti-
tive. The concentration of nickel was consistently very low. The concen-
tration of chromium in the clean salt started out about the same as the
concentration in the flush salt, but then leveled off at about twice this
value. The iron concentration in the clean salt was uniformly about

twice that in the flush salt.

6.3 Pretreatment and Transfer of Salt Charges

Pretreatment of the salt for removal of wvolatiles and melting, mix-
ing, and transfer of the salt were all done using a mobile unit that was
positioned next to the facility enclosure. The unit (Figs. 31 and 32)
contained a nickel-lined stainless steel pot of 2 ft® capacity housed in
a 23-kW furnace. Connections were provided at the top for charging of
salt and for the connection of vacuum and inert gas lines. The salt was
transferred through a 3/8-in.—OD tubing line that was preheated by passing
electric current through the walls of the tubing. The size of the system
was such that four batches were regquired to load the sump tank.

The procedure for pretreatment and transfer of the flushing charge
of salt was as follows. The calculated amounts of salt powders, based
on the eutectic mixture, were loaded into the melting pot through the
3-in. access pipe. The pot was sealed up, heated to 300°F, and held
under a vacuum (<1 psia) for at least 16 hr. The furnace temperature
was then raised to 900°F (melting point of salt = 725°F) and inert gas
blown through the dip leg for about 10 min to insure good mixing of the
charge. The transfer line was then connected to the dip leg and the salt
transferred into the sump tank by applying gas pressure to the melting
pot.

Prior to the addition of the flush charge of salt, a temporary dip

leg tube was inserted into the sump tank, and gas pressure was used to
CONCENTRATICON {ppm)}

ORNL-DWG 69-12589R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7000 : ; o ; : : - 1 e T
6000 . . — _l‘_. + [ I L ] + . . i e e ' "
i [ ‘ .
5000 i i , : S — OXYGEN
i ; 1 | i o
4000 J | o B — ;
3000 - 1 — ; ‘ . , .
2000 g - - : v
‘ ‘ | | ! Lo .
1000 g %z - — .‘%&I* . — . . ‘ o
ot . e 1 o2es | | |
2000 e - - . e
° | T ' | H,0 ! g ‘
1000 ! - —_— Tififiwfifi{ i TN [ -
0 ! | z ! ol j |
QOOO [ e - !‘._.. e i T
' g NICKEL
500 S e L - + ! + L + 1‘ A e — - . ce e e
) ee ® 8 000! ceeei se__sey . ’.6_2._- * ° . ® : - 3.l o o @ °
1000 : ;
f | CHROMIUM .
500 - £ 8 . : . . P
o &*° ° $ee %s e . | * | |
1000 | ' ! 5 : IRON |
500 ( ; < R '5- . . - . s j o ., . *» - - »
°® . l. Foe. e o0 ‘ | 1
0 | | ] oL J
1400 : , 1 : ‘
. ‘ | Jii ! ‘ PROFWEZOFSALT TEMPERATURE
1200 } - . . - - L . [
S Wl |
1000 I . - i . j"—l - i o i e e o !
800 I i J r J i
] - o _____
PROFHjZOF SALT(HRCULAflON INTERNAL VERTICAL LINES INDICATE BREF’SHUTDOWN
g Y] 1 (A O f,H_T T
MAR | APR | JUL. AUG SEPT | OCT ~ NOV | DEC JAN | FEB | MAR T APR MAY JUN
1968 1969

Fig. 30. Analyses of impurities in NaBF, salt samples, fluoroborate
circulation test, PKP loop.

58
86

ORNL—-DWG 72—2092

_~SALT PUMP
- MAIN LOOP ENCLOSURE

 
   

~OFF~GAS
4 LINE

      
 
 
  
  

X

L

.~ EQUALIZING
/ LINE

VACUUM
PUMP

-~ NICKEL-LINED
STAINLESS
STEEL
POT

 

 

 

 

 

 

"~ SALT FURNACE 23 kW

Fig. 31. Schematic diagram ot salt transfer system.
 

 

 

 

 

»)

»

87

 

32.

Salt trahsfer

 

 
88

blow the old BULT-L salt from the sump tank into a 55-gal steel drum. The
flush charge of salt was later removed the same way. The clean charge of
salt was processed essentially the same way as the flush charge, except
that the BF, partial pressure of the batch at 925°F was used as a criterion
for the effectiveness of the evacuation treatment. The limiting pressure
rise was arbitrarily selected as 10 psi. Three of the four batches that
showed pressure rises of U4 to 6 psi were transferred without further pro-
cessing. The remaining batch, which registered a 13-psi rise at 925°F,
was cooled down and reevacuated, remelted, and transferred. The pressure
rise during the second melt was only 1.5 psi.

A1l saglt batches were transferred without incident. There was inter-
mittent plugging of the dip leg during the process of mixing with the
helium stream (see Sect. £.5). The gquantities of material transferred are

given below.

Date Weight
of transferred
transfer (1b)
Flush charge added to sump tank Feb. 1968 687
Flush charge removed from sump tank July 19€8 702
Clean charge added to sump tank July 1968 765

6.4 Freeze Valve Operation

 

The freeze valve consisted of a 2-in.-long squeezed-down section at
the center of a 12-in.-long piece of 3/h-in.-IPS sched-L40 Inconel pipe.
Pipe wall spacing at the narrow part was l/h in. The gir tube was fabri-
cated from a piece of 1 1/2-in.-IPS sched-4O Inconel pipe. The conmtrol
system was minimal. There were thermocouples to monitor the temperature
at the center of the valve and at the shoulders and hand-operated Variacs
to adjust heater power and a hand-operated valve to adjust air flow. The
thermocouple and heater layout is shown in Fig. 33. Additional thermo-
couples were added for the fluoroborate test in order to obtain better

simulation of the MSRE freeze wvalves.
89

 

 

 

 

 

 

ORNL—-DWG 72-2093

HEATER 29
4 x 9',-in. FLAT CERAMIC

e HEATERS (2) 350 W 115V

 

 

 

 

' |
HEATER 28 : |
3 -
19/ 1D x 4 in LONG | ; .- HEATER 30
CEQ%SHEEF(E) | | /32-in CALRODS (2)
W V- /
- { 70 ; - fin. BETWEEN T%%ofiygafi?JTOM
1 ¢ 4 HEATERS
7T /j
78 1 s0 | : '
TO LOOP —a— o Ss  ges .71 E—--— SUMP
___: l i i
h&i___lji
FUEL LINE ~ l ; I .
; |
I
/ | o ox
s —in. T |TH
Y4 in. BETWEEN HEATERS -~ | | 27a X 37a X /p-in INSULATION W
I | SHIMSTOCK BETWEEN INSULATION
I | AND HEATERS
i {
AIR FLOW

Fig.

33.

NaBF, circulation test, PKP loop.

Heater and thermocouple layout drain line freeze valve,
90

The freeze valve was operated through six fill and drain cycles
during the flush salt program and ten fill and drain cycles during the
clean salt program. There was never any difficulty in establishing or
thawing the frozen salt plug in the valve. The temperature at the
center of the valve was maintained between 200 and 4OO°F while the valve
was frozen. The time required for draining varied between 3 min and 3 hr.
Although the longer drain times were attributed to variations in operating
techniques, an alternate possitility would be flow restrictions due to
ageccunmulations of corrosion products. The last six drains all required
7 to 10 min, which indicgtes that the situation improved with practice.

On March 21, 1968, the freeze valve was used to protect the drain
tank from excessive pressure while making a calibration check on the salt
pressure transmitters. No pressure rise was noted in the drain tank,
which was at O psig, during the 30-min periocd when the loop was pressur-

ized with helium at various levels between 50 and 100 psig.

6.5 Rubkler Tube Operation

 

During the preparation of the charge of flush salt in the salt fur-
nace (Sect. €.3), helium was bubbled through a tube submerged in the
molten mass to insure good mixing. It was observed that the gas-feed
pressure would build up and drop off intermittently, as though the tube
was alternately plugging and unpluggirg. The implication was that the
helium was stripping BF, away from the salt at the end of the tube,
legving a high-melting NaF-enriched phase that precipitated out and
cagused the flow restriction.

In September and October 1968, three tests were run in an attempt to
confirm this observation. In each case the BF; flow to the pump bowl bub-
bler tube (see Fig. 7) was reduced to zero and gllowed to remain there un-
til plugging of the tube was indicated, that is, until the indicated salt
level went off scale on the high side (the AP at full scale was 20 in.

HoO or 0.72 psi). While the BF, flow was at zero, the helium flow was
increased so as to maintain the total bubbler flow at 370 cnf /min. The
salt temperature was mostly 900°F during two of the tests and mostly

1025°F for the third test. Some variations in salt temperature were
91

caused by other test work that was being carried out at the same time.
The BF; flow was restored to normal immediately after the indicated
level went off scale. The results (see Fig. 34) show that where the
average temperature was 900°F, the bubbler tube plugged in about 140 hr,
and where the salt temperature was 1025°F, plugging occurred in about
54 hr. 1In each case the salt level indication returned to normal within
an hour after the BF; flow was restored. The time required to develop a
plug was less at higher salt temperature, presumably because the higher
BFy partial pressure permitted more rapid removal of the BFy. It is
interesting to note that the plugging time required during the salt pre-
paration operation was lower by a factor of more than 100 than the
plugging time required during the bubbler tube test. A probable expla-
nation is that there was much better mixing in the pump bowl due to the
relatively high rate of exchange of salt (estimated to be one loop vol-
ume every 2 min) between the pump bowl and the bulk salt circulation
system. Other contributing factors may have been the difference in tube
geometry (the salt transfer tube was 1/4 in. ID, whereas the bubbler tube
was 1/2 in. ID) or the gas flow in the salt preparation tank (a flowmeter
was not used and so the actual flow was not known) may have been signifi-
cantly higher than the bubbler tube flow.

Although tThe above tests indicate that a continuous bubbler flow
cannot be maintained using pure helium, it must be concluded that the
use of a pure BFs flow might also cause plugging problems of a different
nature. As an example, assume operating conditions of 1025°F (BFz partial
pressure = 1.35 psia), 24 psig, and a 100% BFy feed to the bubbler. Now,
assume the BF; flow is inadvertently cut off at a valve in the BF; header.
The BF; downstream of the valve will react with the salt, and the header
pressure will approach the equilibrium value of 1.35 psia. As the header
pressure drops, salt will be driven back through the bubbler tube, and
the unheated portion of the BF; feed header will be plugged with frozen
salt.
92

ORNL-DWG 72—20‘94

 

100

50

INDICATED
SALT LEVEL
(%)

600 ‘
|
400 HELIUM FLOW

148 hr
200

GAS FLOW
(cm%finm)

BF FLOW

1200

1000

(°F)

800

SALT
TEMPERATURE

600
6 20 22 24 26

SEPT. 1968 OCT. 1968 0CT.1968

Fig. 34. Effect of operating liquid level bubbler tube without
BF3; flow, NaBF, circulation test, PKP loop.
93

6.6 Experience with Salt Level Instruments

Prior to the fluoroborate test program, the loop was equipped with
five single-point, or spark plug, salt level indicators, one each for
maximum, minimum, and normel pump bowl level and one each for maximum
and minimum drain tank level. The operating potential was 110 V, and
when salt contacted the probe, the circuit was closed, causing an indi-
cator bulb to light up. The probe material was changed from Inconel to
Hastelloy N for the fluoroborate service. The normal and minimum level
probes in the pump bowl were removed from service early in the program
in order to provide access for other tests. The three remaining level
probes gave excellent service for the duration of the test program. The
only trouble was in April 1968, when the drain tank maximum probe shorted
out. The failure was traced to a leaky tube fitting. Wet gir diffused
in through the leak, and the resulting accumulation of corrosion products
bridged the gap between the probe and the nearby vessel wall.

During normal operation, a continuous indicgtion of the pump bowl
salt level was provided by monitoring the back-pressure in the BF, feed
line (see Fig. 7). The BF, bubbler feed tube was 5/8-in.-OD by 0.065-in.-
wall Inconel tubing welded into the nozzle formerly occupied by the pump
bowl minimum level probe. The bottom of the bubbler tube was 12 1/2 in.
below the top of the nozzle; a study of the pump drawings indicates that
this should have placed the tube end about 3/h in. below the volute center
line. The end of the tube had & single 45° notch approximately 1/4 in.
wide by 1/8 in. deep. At the normal operating temperature of 1025°F, the
salt density was 116 lbm/fta and the instrument range was 10.8 in. salt
at full scale. During the early phases of the test program, the salt
level was adjusted to 25 to 35% of the level scale; later on, the level
was adjusted to L5 to 50% to insure that the pump bowl thermocouple well
would be adequately covered with salt. In normal operatlion, an upper
limit of 54% (6 in. salt) was maintained on the salt level. A backflow
preventer, a piece of 1 l/2-in.-IPS, 9-in.-long stainless steel pipe, was
installed just upstream of the bubbler tube to minimize plugging of the BF;

feed line when pressure level changes (e.g., during pump ingassing) caused
oM
backflow of salt from the pump bowl. A Calrod heater was provided so that
any salt in the backflow preventer could be melted and allowed to drain
back into the pump bowl. Some difficulty was experienced with plugging
of the B¥; feed line during the early part of the loop operation, but
the plugs were believed to be associated primarily with the ingassing
transients that occurred during cavitation tests. 1t was also noted that
plugging difficulties would occur if the proper gas feed composition was
not maintained. Just prior to the final shutdown, an apparent rise in
salt level indicated that there might be a flow restriction in the bubbler
tube line. Then at the shutdown the restriction became worse, so that
there was an apparent salt level even after the loop had been drained.
For the bubbler flow conditioms (0.37 liter/min), an orifice of about
0.030 in. diameter (0.0036% of the original cross-sectional area) would
be required to cause a detectable change (21% scale) on the salt level
instrument.

After shutdown, the bubbler tube was removed from the pump bowl.

The end of the tube was found to be completely filled with a hard, grainy,
metallic-appearing magnetic deposit (see Sect. 6.11). ILesser amounts of
the deposit were found inside the tube as far as 5 1/2 in. from the end.
We conclude that the performance of the btubbler tube level indicator was
reliable and appeared to bhe reasonably accurate, although serious plugging
was noted at the final inspection. Further study should indicate whether
the plugging problem can be moderated by design or operational changes.

At worst, we should be able to accommodate tube replacement on a two-year
cycle.

In June 1968, in response to a suggestion by J. W. Krewson, ORNL
Instrumentation and Control Division, a brief test was made using a
resistance probe for indicating salt level in the pump bowl. The input
signal was 1000 Hz at low voltage, and the output signal was a straight-
line function of salt level over the lower third of the probe. Calibration
to a depth of 2 in. produced a reasonably straight line with a slope of
75 mV/in. The test was terminated after one week due to catastrophic cor-
rosion of the stainless steel probe by the fluorcborate environment.
Subsequently, corrosion tests were run using probes made from Hastelloy N

material, and the results indicated that the severe corrosion of the
95

stainless steel probe was normal for operation in fluoroborate salt at
1025°F, that acceptable corrosion rates could be achlieved by using an
appropriate material such as Hastelloy N, and that the use of a high-
frequency electrical current did no% appear to contribute to the corro-
gsion. We did not have an opportunity to repeat the salt level calibration

using the Hastelloy N probe.

6.7 Back Diffusion at the Pump Shaft

 

The purge gas that entered the pump shaft annulus was split into
two streams; the larger stream flowed down the shaft and into the pump
bowl vapor space, and the other stream, 100 cnf /min or about 11% of the
total flow, flowed up the shaft, past the shaft seal, through the oil
catch tank, and finally was vented into the loop enclosure. The end of
this vent line, a l/h-in. tube, would gradually plug over a period of
several months of operation, indicating that the gas contained some BFj.
The composition of the plug material was not determined.

Several tests were made in which the seal purge stream was passed
through the thermal conductivity cell when the shaft purge was at the
normal setting of 850 cnf /min and the salt temperature was at 1025°F.

In none of the tests was there any discernible change in signal when com-
pared with pure helium sample gas. The thermal conductivity cell had a
calibration slope of L4.7% BFs per mV, and the smallest chart change that
could be detected with confidence was 0.02 mV, which would be equivalent
to O.l% or 1000 ppm. It was concluded from this that the BF; content of
the seal purge stream was less than 1000 ppm and that the concentration
ratio between the purge stream and the pump bowl gas was not more than
0.03 (0.1%/3.5%). This result is in agreement with back-diffusion tests
run on this pump® using 8° Kr, where a concentration ratio of 0.0l was
observed for the same purge flow. In another test, with the system at
steady state, we reduced the shaft purge rate by a factor of 2 and did
not get any response from the thermal conductivity cell. Data from

Ref. 8 indicate that a factor of 2 in total purge would change the con-

centration ratio by an order of magnitude. From this we might infer that
96

at the normal shaft purge rate of 950 cm® /min the BF, content of the lower

purge stream could have been as low as 100 ppm.

6.8 A Salt Leak to the Atmosphere

 

On the morning of September 23, 1969, after 8700 hr of operation with
the fluoroborate salt, a salt leak was discovered. The initial indications
of trouble were a decrease in pump bowl liquid level of 8% (~0.8 in. salt),
an apparent decrease in the temperature of circulating salt, and irregular
temperatures at the two venturi pressure transmitters. A visual inspection
inside the loop enclosure revegled a stalactite of frozen salt hanging from
the insulation surrounding both venturi pressure transmitters and & small
gquantity of salt in the floor pan beneath them. Using normal shutdown pro-
cedures, the pump was stopped and the salt was drained into the sump tank.
The leak was found to be at the venturi inlet pressure transmitter (PT
1044) in the weld that joins a 1/2-in.-IPS pipe to the boss on the dia-
phragm Tlange (Fig. 35). This is a stagnant salt zone serving only to
sense venturi inlet pressure, which is about 100 psig with the pump on.

The defective pressure transmitter was removed from the system, the
pipe nozzle to which it had been attached was capped off, and operation
of the loop was resumed.

Metallurgical examination of the pressure transmitier indicated that
the weld (made in 1956) had been defective when made by the vendor, and
that the defect had been repaired by applying braze metal to the interior
surface. It is estimated that the pressure transmitter had been in service
in the loop for at least eight years and, prior to the current fluoroborate
service, was operated for three years with a salt similar to the MSRE fuel
salt.

The conclusions resulting from the investigation of the leak were as
follows:

1. The salt leak was attributed to the failure of a welded closure
that was defective when originally made and which had been repaired by a
technique, which based on current technology, could not be recommended
for molten-salt service at 1000°F. The fact that the leak occurred during

NaBF, service was colincidental.
o

ORNL DWG 72-2095

  
  

5 PIECE DIAPHRAGM

NoK SIDE

 

 

7
—F—-..Hh—-'//
QL L —3
-
~
™~

 

 

 

 

 

 

 

Fig. 35. Location of leak on venturl inlet pressure transmitter
(PT 10LA), NaBF, circulation test, PKP loop.
98

o, Corrosion in the stagnant leg between the pressure transmitter
and the loop was relatively mild and was similar to that experienced with
Inconel in other NaBF, loops.

3. The consequences of the leak can be characterized as rather mild.
If there were any visible fumes, they were carried off to the stack by the
controlled ventilation system. There was no indication of fire or sudden
energy release. The condition of the containment system metal in the
vieinity of the hole seemed to indicate that the leak was not self-
aggravating, that is, continued legkage did not result in an increase in

the rate of leakage.

6.9 Experience with Handling BF,

 

Properties of BF, and suggestions for handling and storage have been
excerpted from manufacturers' literature and included in Appendix A. In
general, our experience during the test program was in agreement with the
literature, and the following comments are included for emphasis or to
present observations which should be of special interest.

1. There was no evidence of corrosive attack in tubing, pipe, and
vessels where the gas was dry, that is, excluded from contact with moist
alr. Conversely, where there was opportunity for contact with moist air,
the resulting combination was extremely corrosive. Handling problems
will be minimized if extra precautions are taken to insure a dry, leak-
tight system before admitting 35, .

2. A flow restriction developed periodicglly at the pcint where
the off-gas stream (3 1/2% BF, in helium) was vented into the stack
suction line (see Sect. 5.4).

3. In September 1967, seventeen 1800-psig cylinders of BF, (nominally
60 1bm per cylinder) were purchased to provide for expected usage during
the test program. The use rate was lower than originally predicted, so
that most of the cylinders were still on hand by the summer of 1968. In
May, June, and July 1968, five of the stored cylinders developed leaks to
the atmosphere. One of the leaks was due to defective packing, but the
other leaks resulted from the fact that the cylinder valves were equipped

with improperly rated safety devices. The leaky cylinders were
99

vented to the atmosphere at a remote locatlion, and eight of the remaining
cylinders were returned to the vendor.

4. Just before venting to the stack suction, the off-gas stream was
passed through a mineral-oil trap to inhibit back diffusion of moist air.
There were no visual 111 effects on the mineral oil during the test period.
The effectiveness of this device in preventing back diffusion of moisture
wvas not measured guantitatively, but two observations indicated that it
was satlsfactory. First, there were no accumulations of moisture in the
mineral oil bubbler tank. Second, there was no evidence of corrosive
attack in the tubing on the upstream side of the bubbler, indicating the

absence of significant moisture levels.

6.10 Corrosion Experience

 

The development and evaluation of container msteriagls for fluoro-
borate service is being done by others in other test equipment.® How-
ever, there were certaln pieces of hardware installed new in the fluoro-
borate system that were exposed to service conditions of interest. A
brief history and general description are presented below.

The pump tank salt level bubbler tube installation is described in
Section 6.6. The tube, which was made from a 5/8-in.-0D, 0.065-in.-wall
Inconel tube, was installed before the loop startup, and 1t was removed
on October 23, 1970. The post-test exterior appearance is shown in
FMg. 3¢. The tube was cut into four pieces by making transverse cuts at
1, k 1/2, and 7 1/2 in. from the lower end. The three lower pieces were
then slit axiglly. Figure 37 shows the interior appearance of the tube.
The bubbler tube was examined chemically and metallurgically, and the re-
sults are presented in Refs. 10 and 11.

The pressure control valve was used throughout the test program.
The valve operated at room temperature and absorbed the full pressure
drop of 24 psi. It had to be opened and cleaned a number of times to
remove gccumulations of salt and liquid impurities. The stainless steel
valve trim (Fig. 38) appeared to be essentially free of corrosive attack

when examined visually on October 26, 1970.
 

 

 

 

 

 

 

 

100

 

 

 

 

 

Post-test appearance of salt level bubbler tube exterior.

PHOTO 78236

 
 

 

 

 

 

101

 

PHOTO 78258

 

 

 

Fig. 37. Post-test appearance of salt level bubbler tube interior.

 
 

102

 

 

 

 

 

 

 

 

 

 

 

 

 

Post~test appearance of pressure control vglve trim,

38

g

Fi

 

 

 
During the final examination of the pump rotary element, pieces of
gray~btlack, magnetic, metallic-appearing material, shaped somewhat like
gutomobile tire weights, were found lying in the annulus next to the im-
peller hub. The chemical analysis [Na-5, B-3, Fe-2.5, Cr-0.2, and Ni-66.3
(wt %)] indicated that the material was elemental nickel mixed with some
NaBF, . We speculate that the material was formed near the upper end of
the impeller hub by mass transfer of nickel from the salt that flowed
through the impeller-shaft labyrinth seal (fountain flow). Evidence of
severe corrosion was found on the inner heat baffles,'® and this condition
was ascribed to continuous low levels of moisture (estimated to be about
25 ppm) in the shaft purge helium gas. The mechanism is believed to have
been as follows: residual droplets of fluoroborate salt were left on the
heat baffle surfaces as a result of the ingassing transients which occurred
during the cavitation inception tests and the moisture in the incoming
purge gas combined with these salt droplets to form corrosive reaction

products.

6£.11 Creen Salt

When the pump rotary element was removed for inspection in May 1968,
several large chunks of green salt were found lying on top of the upper
impeller casing. It was concluded that this material was a relatively
insoluble phase that resulted from mixing the fluoroborate flushing salt
with the residue of MSRE-type salt left over from the previous test pro-
gram. The color was similar to, but not quite as dark as, the green de-
posits (NEBCrPg) obtalned during the cold-finger test work. In September
1970, when the rotary element was again removed from the pump, a similar
deposit of green salt was found on the upper impeller casing. In both
cases, 1t was obvious that the deposits had been formed on the under side
of the outer thermal baffles and had been dislodged during removal bf the
rotary element. This thermsl baffle surface was probably somewhat colder
than the bulk salt temperature, which would account for precipitation of
the less-soluble phase; but the surface was about U in. above-the maximum

normal pump bowl salt level, and it 1s not clear how the salt came into

contact with this surface. There are two possibilities: first, splashing
10k

or spraying of the salt due to impeller fountain flow, and second, the
violent but momentary expansions of the salt that occurred during the
ingassing transients (see Sect. 6.1). The first mechanism seems more
probable, since the ingassing transients were so brief that it is diffi-
cult to see how there could have been sufficient contact time to allow
transfer of the material. On the other hand, the first deposit was found
after only 187 hr of salt circulation, which implies a rather rapid trans-
fer rate even for the spray theory. Table 11 compares the chemical analy-
ses of the green salts with the various charges of circulating salt, and
Table 12 compares molecular compositions of the circulating salts with
possible molecular compositions for the green salts.

The discovery of the deposits of green salt suggests the possibility
of similar deposits in an MSBR secondary coolant system 1f there were
leakage of the fuel salt into the coolant salt. The possible ramifications
of such an incident need to be thoroughly considered to insure that there

would e no intolerasble side effects.

Takle 11. Comparison of green salt with present and
former salts for NaBF, circulation test, PKP-1 loop

 

 

 

 

BULT-L salt NaBF, salt {wt %) Green salt (wt %)
forme?iz %? HooF Flush charge® Clean batch ¥3g8 i;?g‘
i 9.72 0.21 0.048 0.21 0.17
Be 5.82 0.17 0.016 0.0k 0.10
U 5.12 0.25 0.019 12.2 0.35
Th 20.0 0.25 0.093 26.5 13.9
Na 21.6 20.9 10.9 17.8
Q.45 9.3 3.91 2.6
59.34 67.7 68.0 43.5 45.3
Fe 0.02 0.06 1.48 9.45
Cr 0.01 0.04 0.27 8.9
i 0.04 0.005 0.03 0.08

 

aAfter mixing with heel of salt left in drain tank from previous
test program.
105

Table 12. Comparison of molecular compositionsa of
circulating salts and green salts

 

Mluoroborate salt Green salt

 

BULT-4 salt

formerly in loop May Sept.

Flush charge Clean charge

 

1968 1970
NaBF, 88 93 53 32
NaF 6 5 13 8
LiF 65 3 0.1 L
BeF, 30 2 0.2 0.7
UF, 1 0.1 0.01 T 0.2
ThF, L 0.1 0.04 17
Neg CrFs 0.02 0.08 0.7 23
FeF 0.0k 0.1 L 23

2

 

®Inferred from chemical anslyses; all values in mole %.

6.12 Valves in BF, Service

 

In general, the performance of in-line valves, whether packed stem
seal or bellows stem seal, was very satisfactory. We attribute success
in this area to the use of resistant materials (fluorinated polymers)
for gaskets and packing and to care in leak testing to insure a tight
system. In a few instances a valve stuck in the closed position. This
difficulty was attributed to accumulation of fluid or solids and to the
fact that the valves opened on spring action, that is, the stems were
not directly connected to the valve wheel.

In cases where a valve was used to separate the BF; system from the
atmosphere, the experience was not so good. One example was the 1l-in.
ball valve that was inétalled on the salt sample access line on the pump
bowl. This valve had a stainless steel body and Teflon trim, but we had
continual trouble with seat leskage. We attributed the trouble to attack
by corrosive materials produced by the reaction of BF; with atmospheric
molsture. Another example was the safety relief valve that was in a line

connected to the BF; feed header. This valve had a spring-loaded brass
poppet and a synthetic-rubber elastomer (Buna N) seat. A check after

the final shutdown of the loop showed that the poppet opened at about

the right cracking pressure but failed to reseat when the pressure was
lowered. We believe the failure to reseat was due to BFz -H,0 reaction
products which caused a change in the properties of the elastomer. A
third example was the check valve in the off-gas line just before the
stack suction. The original valve, which had a spring-loaded brass
poppet with a Buna N seat, stuck closed after 766 hr of operation. Fail-
ure was attribvuted to deterioration of the elastomer seat. We changed
the rubber ring to a Poly-TFE ring and had no further difficulty.

In summary, the valve applications that require close attention are
in locations where there is an opportunity for the BF, to come into con-
tact with atmospheric moisture. Good performance can be obtained by the
use of well-designed valves having proper materials of construction.

For infrequently used lines, it may be advisable to provide for purgling

and capping off the line on the atmospheric side of the valve,

7. RECOMMENDATIONS FOR FURTHER DEVELOPMENT WORK

7.1 Impurities in the Salt

 

Certain impurities, variously identified as water, hydroxyfluoroboric
acid, BFs -H;0 reaction products, etc., cause increased corrosion ratest?
in the circulation system and flow restrictions in the off-gas system.
Work is needed in the following areas: (1) identity and pertinent physi-
cal properties (e.g., vapor pressure) of the impurities, (2) maximum
concentration of the impurities which can be tolerated in the salt, and

(3) procedure for pretreating salt to remove impurities.

7.2 Corrosion Product Deposition

 

Additional work is needed to verify the suitability of a cold trap-
ping technique to inhibit the deposition of corrosion products on steam

generator tubes and other functional surfaces where fouling is unwanted.
;.w
H
-]

7.3 Salt Level Instruments

We need a rellable, accurate method for continuous indication of
salt level. Two devices were tested. The test of the resistivity probe
was cursory, and additional work is needed to provide g reliable evalu-
ation of it. Additional work is needed with the bubble-tube indicator
to insure that we have a suitable construction material and to develop
a design that will not be susceptible to plugging. The accuracy and
stability of both types need to be established.

7.4 Off-Gas System Restrictions

 

We have proposed the use of a hot-mist trap and a replaceable filter
to help eliminate flow restrictions in the off-gas system. The word
"replaceable" implies a parallel filter installation with provisions for
isolating and servicing the off-stream unit. The problem is to design
an arrangement that does not tend to plug at the isolation valve between
the hot-mist trap and the filter. Our test work showed that globe valves
would be unsuited for this application. Ball valves, which have g
straight-through flow pattern, might be satisfactory from the standpoint
of solids accumulstion, but their packed-stem seal design might be un-
accéptable for a nuclear reactor installation. One possible approach
would be to dodge the problem altogether by eliminating the isolation
valves and sizing the filter so that replacement would be necessary only
during scheduled reactor shutdowns. We need to examine the various
alternatives and to define and perform tests that will demonstrate that

we have a satisfactory solution to this problem.
7.5 Control of Salt Composition

In connection with the use of the thermal conductivity cell to moni-
tor the composition of the salt two tests are needed: (1) the instrument
should be used to check a salt that is known to be noneutectic, and (2)
tests should be run to determine the variation.of BF; partial pressure
as a function of the quantity of fuel salt that is mixed with the fluoro-

borate.
108

7.6 Intermixing of Molten Salts

 

Additional study is needed to examine the relgtive miscibility of
vagrious mixtures of the fluoroborate salt with the reactor fuel salt.
The purpose is to insure that cross mixing of salt between the fuel and
coolant systems would not cause any insoluble safety or operating prob-
lems. This study can probably be carried out in conjunction with the

work noted in Section 7.5 above.

7.7 Solid Phase Transition

 

Studies should be made to determine 1f special designs or operating
technigues are needed to prevent damaging stresses when salt-filled com-
ponents are heagted upward through the solid phase transition temperature
(LAE9°F). A brief, simple test was run (see Appendix E) that indicates
that proper adjustment of heater temperatures is probably an important
factor. Additional study is needed, however, to insure that we can cope
with this problem in complex circulating systems.

The solid phase transition might also be expected to cause problems
with the operation of freeze valves, although this hypothesis was not
supported by our experience (see Sect. 6.4) during the test program. We
recommend a critical review of the design and operation of fluoroborate
freeze valves followed by the execution of any experimental work that

might be indicated by this review.

7.8 BF, Recycle System

 

The BF, usage rate 1is directly proportional to the BF, partial pres-
sure. In the MSER concept, the BF, partial pressure will be relatively
high (about one-third of an atmosphere) and the usage rate for a once-
through BF, feed system would be on the order of 100 ft2/day or 20 lbm/day
(based on a toctal pressure of 2 stm and total flow of 12 liters/min).

The problem of BE, supply and disposal would be minimized if a system
were provided to recycle a large fraction of the BF,;, and a study should
be made to examine the feasivility and desgirability of such a recycle

system.
109

ACKNOWLEDGMENTS

The work described herein required and received constructive inputs
from many persons. The major contributors are listed below.

Dunlap Scott, guldance in formulation of test program; R. B. Gallaher,
P. G. Smith, and H. C. Young, execution and analysis of tests; W. H. Duck-
worth, F. E. Lynch, and Doyle Scott, operation of test equipment; Ben
Squires, design of control system; C. A. Brandon, design of cold finger;
R. F. Apple, S. Cantor, A. S. Meyer, Jr., and J. A. Shaffer, chemistry
support; Dunlap Scott, A. G. Grindell, R. E. MacPherson, and R. B. Briggs,
constructive review of report drafts.

We are also indebted to V. B. Maggart for typing of the report and
to M. R. Sheldon for assistance with the final editing. All contributions

are hereby gratefully acknowledged.
10.
11.

12.

13.

20
REFERENCES

R. C. Robertson et al., Two-Fluid Molten-Salt Breeder Reactor Design
Study (Status as of January 1, 1968), ORNL-4523, pp. 9-11 (August
1970).

Reactor Handbook, vol. IV, 24 ed., p. 832, Interscience, New York,
196k,

 

P. G. Smith, "Hydraulic Performance and Cavitation Inception Charac-
teristics with Fluorcborate — PKP Salt Pump (MSR-69-3)," unpublished
internal memorandum {Jan. 10, 1969).

A. N. Smith et al., MSRP Semiannu. Progr. Rep. February 1969, ORNL-
4396, p. 100.

S. Cantor et al., Physical Properties of Molten-Salt Reactor Fuel,
Coolant, and Flush Salts, ORNL-TM-2316, pp. 33-35 (August 1963).

 

S. Cantor, MSRP Semiannu. Progr. Rep. August 1967, ORNL-4191,
pp. 159—61.

MSRP Semiannu. Progr. Rep. February 1969, ORNL-4396, p. 105.

"Rotating Machinery for Gas-Cooled Reactor Application," Proceedings

of Meeting at Oak Ridge National Laboratory, April o—%, 1962,
TID-7631.

J. W. Koger and A. P. Litman, Compatibility of Fused Sodium Fluoro-
borates and BF; Gas with Hastelloy N Alloys, ORNL-TM-2978 (June 1970).

S. Cantor, MSRP Semiannu. Progr. Rep. Feb. 1971, ORNL-LGT6, pp. 92-93.
J. W. Xoger, ibid., pp. 211 ff.
MSRP Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 31 ff.

J. W. Koger, MSRP Semiannu Progr. Rep. August 1971, ORNL-4622,
pp. 168-69.

BIBLIOGRAPHY
H. 3. Booth and D. R. Martin, Boron Trifluoride and Its Derivatives,
Wiley, New York, 1949.

Matheson Gas Data Book, The Matheson Co., Inc., East Rutherford,
N.J., 1961.

 
111

APPENDICES
113

Appendix A

SELECTED PHYSICAL AND CHEMICAL PROPERTIES OF
PROCESS MATERIALS

A.1 NaBF, and NaF

Name Sodium fluoroborate Sodium fluoride
Formula NaBF, NaF
Molecular weight 109.8 Lo
Density, g/cnf, 20°C 2,47 2.79
Melting point, °F 765 1818
Solubility, g/100 g H,O

At 25°C 108 I

At 100°C 210

A.2 The System NaBF, -NalF

 

Phase diagram See Fig. A.1l
BF, decomposition See Fig. A.2
pressure

. . wt % boron
Mole fraction NaBF, (2'126)(wt T sodium)

wt % fluorine)

1
(0.569R — 3)” , where R = (wt % voron .

A.3 DNaBF, -NaF Eutectic

 

Comgosition

 

NaBF, -NaF NaBF, NaF Na B F
Moles 1 .92 0.08 1 0.92 3.76
Grams 10k, 39 101.3 3.36 23 9.95  71.k4k

Weight % 100 96.77 3.22 22.02 9.53 68.44
1ik

Properties

(Note: The following properties are for the liquid unless other-
wise indicated.)

Conductivity
1 3 :
Electrical, (Q-cm) w=2.7+7.2x10 (T °F - 932)
_1 _1 _1
Thermal, Btu hr  ft (°F) K = 0.3
Density
_4
g/cnf p =2.27 —4.1x10 T °F
_<
lbm/ft3 p = 1k2.5 —2.56 Xx 10 T °F
1025 °F (551.7°C) p =1.86 g/ler® = 116 1bm/ft3
1150°F (621.1°C) p =1.81 g/er® = 113 1bm/ft3
Density of solid, g/cnf Solid phase transition at 469.4°F
(2k3°C)
5
<UE9°F: o = 2.52 — 6.67 X 10 T °F
S
2U69°F: o = 2.15 — 6.67 x 10~
(T °F — 469.4)
Expansion
_1 _4%
Thermal, {°F) o =1.1 x10
3
Solid @ =6.7 x 10~
Heat capacity
R
Btu 1b (°F) cP = 0.36
L4
Solid, 77 — L69°F Cp = 0.22 +3.2x10 T °F
W69 — 718°F cp = 0.3k
Heat of fusion
Btu/lbm NH = 55.8
Solid transition, L4L69.L°F AT = 26.1

Melting point, °F 723
5
Solubility of inert gases, 10~ 1b/mole
of gas per Tt® melt per atm

Surface tension, dynes/cm

Decomposition pressure

Viscosity, cP

1025 °F

AL BE;*

Name

Formula

Molecular weight

Specific volume, ft2/1b_

RBoiling point, °F

Freezing point, °F

Specific gravity, 50°F (air = 1)

Critical temperature, °F

Critical pressure, psia

Specific heat of gas at —HL°F,
Btu 1o~ mole” (°F)”

He — 4.7
Kr - 9.2
Xe — 8.5

o =121.3 — 0.0417T °F

See Fig. A.2
M= 0.04 exp(%gg%>
M=1.52

Boron trifluoride (also boron
fluoride)

BF,
67.82
5.6
—148.5
-196.8
2.37
10

732

10

Description. Boron triflucride is a colorless gas that fumes in

moist alr and has a pungent, suffocating odor. It is nonflammable and

does not support combustion. It is normally packaged in cylinders as

a nonliquefied gas at a pressure of 2000 psig at 70°F. It is very solu-

ble in water with decomposition (forming fluoroboric and boric acids)

and igs hegvier than sir.

 

*"Boron Trifluoride," p. 33 ff. in Matheson Gas Data Book, The
Matheson Co., Inc., East Rutherford, N.J., 1961.

 
b

Toxicity. Boron trifluoride is very irritating to the respiratory
tract. Exposure of the skin or eyes or the breathing of boron tri-
fluoride should be avcoided. Although the relative toxicity of the gas
to humans has not been established, no medical evidence of chronic ef-
fects has been found among workmen who have fregquently been exposed to
small amounts for periods up to seven years. At high concentrations
boron trifluoride will cause burns on the skin similar to, but not as
penetrating as, hydrogen fluoride.

First-aid suggestions. Observe procedures specified by the Indus-

 

trial Hygiene Department. If official procedures are not available,
treat irritstion or burns of the eyes or skin with copious amounts of
water and obtain services of a physician or trained clinician as soon
as possible. The treatment of BF; burns is normally the same as that
used for anhydrous hydrogen fluoride burns.

Precautions in handling and storage. The following rules should be

 

followed in the handling and storage of boron trifluoride.

1. Cylinders should be stored in & dry, cool, well-ventilated area.
Cylinders may be stored in the open, but in such cases should be pro-
tected against extreme weather and from the dampness of the ground to
prevent rusting.

2. Since boron trifluoride is reactive with water, alcohol, ether,
and other compounds, introduction of the gas below the surface of a
liquid may create a hazard owing to the possibility of suckback into the
cylinder. This should be guarded against by the use of traps or check
valves.

3. Equipment exposed to boron trifluoride should not be used with
other gases, particularly oxygen, since the gas may have oll vapors that
will coat out on equipment and may cause fires when combined with oxygen
under pressure.

legk detection. Small leaks may be detected visually by checking
for an accumulation of fluid (product of reaction between BF; and atmo-
spheric moisture) or with the aid of an aqueous ammonia squeeze bottle
(formation of white fumes). If the lesk is large enough, reaction with

atmospheric moisture will produce visible "smoke."
1177

Materials of construction. Dry boron trifluoride does not react

 

with the common metals of construction, but if moisture is present, the
hydrate aclds identified gbove can corrode all common metals rapidly.
In consequence, lines and pressure reducing valves in boron trifluoride
service must be well protected from moist air. Cast iron must not be
used becaguse active fluoride attacks it structure. If steel piping is
used for boron trifluoride, forged steel fittings must be used with it
instead of cast iron fittings. Materials recommended for the handling
of dry boron trifluoride are steel tubing or pipe, stainless steel,
copper, nickel, Monel, brass, and aluminum, and the more noble metals.
These metals will stand up adequately to at least 200°C. Pyrex glass
is also suitable up to about 200°C at low pressures. For moist gas:
copper, Saran tubing, hard rubber, paraffin wax, and Pyrex glass show
fair resistance; plastic materials, such as Teflon, Epons, polyethylene,
and pure polyvinyl chloride are not attacked at 80°C; rubber tubing,
phenclic resins, nylon, cellulose, and commercial polyvinyl chloride are
readily attacked.

Chemical properties. (a) With elements: Alkali and alkaline-earth

 

metals reduce boron trifluoride to elemental boron and the metal fluoride.
Gaseous or liquid boron trifluoride does not react with mercury or
chromium, even at high pressures for long periods. Red-hot iron is not
attacked by boron trifluoride.

(b) With oxides: When boron trifluoride is allowed to react with
slaked lime, calcium borate and fluoroborate are formed with evolution
of heat. With anhydrous calcium oxide or magnesium oxide, the metal
fluoride and the volatile boron oxyfluoride are formed.

(c) With halides: BCl, and BF,; do not react when heated to 500°C.
Aluminum chloride or aluminum bromide react with boron trifluoride when
gently heated to give the corresponding boron halide and aluminum fluo-
ride. Boron trifluoride forms no coordination compounds when passed at
1 atm over the solid chlorides of copper, silver, or potassium at tem-
peratures from =75 to 530°C.

(d) As a catalyst: Boron trifluoride acts as an acid catalyst. It
catalyzes numerous types of reactions, namely, esterification, nitrations,

oxidations, reductions, halogenations, etc.
A.5 TInconel and Hastelloy N

 

Nominal composition (wt %)

 

 

 

 

C Mn Si Cr Ni Fe Mo
Inconel 0.0k 0.35 0.20 15 78 7
Hastelloy N 0.06 0.50 0.50 7 70 L 17
A.6 Helium and Argon
Helium Argon
Molecular weight L 40
Specific volume, fta/lbm 90 9
=3 1 21
Specific heat, Btu 1b~ mole” (°F) 5.00 5.00
21 21 21
Thermal conductivity, Btu hr~ £t~ (°F) 0.082 0.0094%

1 21 _5 _5
Viscosity, 1b rt sec , 32°F, 1 atm 1.25 x 10 1.43 x 10
TEMPERATURE (°C)

119

ORNL- DWG 67-9423AR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1000
\i.\ l. LIQUIDUS |
g995°C T & SOLIDUS |
900 {—- 7% ~ CRYSTAL INVERSION =~ |
E ~e | |
| \ [
800 | *~\ L LIQUID -
206 NoF +LIQUID N ! L
600 | o
NaBF, N
(HIGH- TEMPERATURE FORM) 1\ 408
500 | A +LIQUID ———— e
| S
400 +384°C - R
[ ]
| __Naf + NoBF, (HIGH-TEMPERATURE FORM)__
300 b I (‘ ) l |
243°C i 5 l 5 | | 5
00 " NaoF + NaBF, (LOW-TEMPERATURE FORM)
NaF 20 40 60 80 NaBF,,

NaBF, (mole %)

Fig. A.1l. Phase diagram for the system NaF-NaBF,.
120

ORNL-DWG 72~ 2096
98 —— 1

 

100

 

 

 

 

 

 

 

 

S REFERENCE:
SEE APPENDIX

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VAPOR PRESSURE (psio)}

 

 

 

 

 

 

 

 

 

 

» o i |
8CQ aGo 1200 1100 1200 1300 1400
TEMPERATURE (°F)

Fig. A.2. BFj3 partial pressure vs temperature for NaF-NaBF, mixtures.
ORNL Dwg.
No.
D-h89L2
D-48okkL

D-2-02-054-9779
D-2-02-054-9780
D-2-02-054-9783
D-2-02-054-978k
D-2-02-054-9785
D-HH-7-41778
Q-1512A~1RD
SK-JWK-11-20-68
B-2-02-054-1926
B-2-02-05L4-1927
D-2-02-054-7440

121

Appendix B

REFERENCE DRAWINGS*

Title

Flow Diagram for Fluoroborate Circulation Test

Elementary Flectric Schematic for Fluoroborate
Circulation Test

Tlow Diagram Be Molten Salt PKP-1 Pump Test Loop
Alarm Schematic

Electric Heater Layout

Thermocouple Layoutl

Gas Control Cabinet

Capillary Restrictor

Thermal Conductivity Cell Power Supply
Resistance Type Level Indicator

Fluid Line for Freegze Valve

Air Tube for Freeze Valve

Sample Device

 

*Partial listing.
122

Appendix C

MATERTAT: SPECIFICATIONS

C.1 NaBF,

 

The sodium fluoroborate used in the PKP fluoroborate circulation
test was received under Order No. 33Y-63903, April 10, 1967, from the
Harshaw Chemical Company. The vendor gives {memo of 5-12-67) the chemi-

cal analysis as follows (wt %):

NaBF, 99.08 Ca 0.01
Og 0.025 Fe 0.023
Pb 0.00k4 Water insoluble <0.01
Si 0.01 Hg O 0.01

C.2 NaF

The sodium fluoride used was cp grade obtained from Laboratory
Stores.
C.3 Bfs

The BF; was purchased from J. T. Baker Chemical Company under Order
No. 79S8-1266, August 28, 1967. Our specification was as follows:

BFy minimum, 99.7% v/v

Maximum impurities, % v/v

Air or noncondensasbles 0.65
S0, 0.001
S04 0.001
SiFy, 0.02

We did not analyze for impurities at ORNL. We did, however, make
a test at the facility in an attempt to determine if there was any
water or hydroxyl compounds in the BF; . The test consisted in passing
several liters of the BF, through a dry-ice trap and then analyzing the
trap contents using the Karl Fischer method. No evidence of water was

found.
123

Appendix D

DERIVATION OF EQUATION FOR CALCULATION OF BFs PARTIAL PRESSURE

From Ref. 6:

 

1~ f
)

Q, = (B ) (==

where f = mole fraction NaBF,, Qp = equilibrium quotient, and Pb =

partial pressure of BFy. Rearranging:

f
Py = (r=9)(Q,) -

 

Also, from Ref. 5, Qp varies with temperature:

B
QP = exp(C - T) .
Then
P = (i—é}gd [exp(C — %)] :

For a 92-8 mix:

f _,0.92y
(l — f) - (6f5g) =11.5 ,
and
P = 11.5 [exp(c - é)J
also
In P =2.4lp +C -5 .
b T
From Ref. 5:
lOg(P%, mm Heg) = 9.024 — ;922 ’
but
In P =2.303 log P, = 20.782 — l%zé%i - 20.780 — 2;25uo ;
also

1n(P, , mm Hg) = ln(Pb, psi) — 1n 51.7 = ln(Pb, psi) — 3.9L455
1000

Then

1n(P, , psi) = 16.837 -3%12%9 . (11)

Comparing Egs. (7) and (11):

2,442 + C = 16.837 and C = 14.395; B = 2k,540 .

Substituting in Eq. (4):

2&,5&0)] _

Pb’ psi = (j_""if“_—}') [exp(1k.395 — T R

(12)
Appendix B

FREEZE-THAW STRESS TEST

As the salt temperature is being increased, there is an abrupt solid-
state expansion (density decrease) of about 15% when the temperature
reaches 469°F. This phenomenon introduces the threat of damaging stresses
in salt-filled components if the salt is cooled below the transition tem-
perature and then reheazted.

A bench test was run to study the effect of varistions in heating
technique. A 2-in.-ID by 18-in.-long Pyrex glass tube was filled with
the fluoroborate eutectic mixture to a depth of about 8 in. Heat was
furnished by a Calrod helix whose inside diameter was about 1/L in.
larger than the outside diameter of the glass tube, so that heat trans-
mission was largely by radiation. An aluminum foil reflector surrounded
the assembly. Thermocouples were used to indicate the surface temperature
of the heater and the salt temperature on the tube center line about 2 in.
from the bottom. The test assembly was then operated through three ther-
mal cycles during which the salt was melted and then refrozen. In all
three heatups, the heater temperature (Th in Fig. E.1) was controlled at
from 75 to 300°F above the salt melting temperature, and in all three
cases the melting was accomplished without difficulty. Then a fourth
heatup was made wherein the heater temperature was controlled below the
melting point of the salt. In this case, when the salt temperature TS
registered about h55°F, the Pyrex tube suddenly shattered.

The conclusion is tThat the poor thermal conductivity of the frozen
salt causes a steep temperéture gradient, and, where the heater tempera-
ture is above the melting point, a film of liquid salt forms at the salt
container interface before the bulk salt temperature reaches the solid
phase transition point. The liquid film serves to provide an expansion
space to protect the container during the phase transition of the bulk
salt. The results of this test suggest that, whenever the salt tempera-
ture is raised through the solid phase transition point, mechanical
stresses can be minimized by proper control of heater temperature. It

will be necessary to consider other factors, however, such as system
126

geometry and the probability and possible effects of shifting and packing
of salt crystals, before one can properly assess the solid phase transi-

tion problem.
127

ORNL- DWG 72-2097

 

2000 T T l
—o~ SALT TEMPERATURE, T i
1500 |~e~ HEATER TEMPERATURE, 7, 4 . ._.__|
| | |
1000 |—— g S
MELT CYCLE NO. 4 PYREX GLASS TUBE
—.——¢—o—y—t"* *e CRACKED AT THIS POINT

T_:—T‘
*

500

2 in. DIAM

 

 

 

 

  

500 fozpfi ol —— SOLID PHASE TRANSITION

- 0

& PYREX GLASS TUuBEL
: 1500 T T

% . MELT CYCLE NO. 3 !

E 1000 “/::""'!‘f‘f - o { HEATER COIL -—
m !

wl

a

=

w

—

 

 

 

1500 TEST APPARATUS
1000

500

 

 

 

TIME (hr)

Fig. E.1l. Data from filuoroborate freeze-thaw stress test.
O O-10vWi Fwno -

1

T
!
o

MR
m—] O\ =

19-23.

WM N
O\ Co—~1 O\t

Lo
RN

78.
79-80.

8L.

82.
83-8k,

85.

Tnternal Distribution

129

 

¥. Apple

. B. Beall
Bender

S. Bettis

. G. Bohlmann
B. Briggs

. Cantor

. Compere
. Cottrell
. Crowley
. Culler

. Engel

. Fraas

. Gallaher
. Grimes

. Grindell
. Guymon
Haubenreich
Helms
Hoffman
Huntley
Jordan
Kasten
Koger
Krakoviak
Krewson
Lundin
Lyon
MacPherson
MceCoy

. McCurdy

HE R RGP LT RERS I P H P LA NS ER I EEE 0

C)@EHEZffl?j?i?{?tfltUEjEiZiflGifijw'fitflt*t*Uifi

39.
Lo,

HEpEIEEEOD

FRPURHSEHEEEY

H.
R.
J.
N.

flfiipiggficflUltiw =

ORNI,-TM- 334

McNeese
McWherter

. Metz
. Meyer

Miller
Moore
Perry
Robertson
Rosenthal

Shgffer
Sheldon
Skinner
Smith

Spiewak

<3 Ffi?fi?ihlfiib J

. Stulting

Sundberg
Taylor
Thoma,
Trauger
Weinberg
Whitman
Wilson

Gale Young

H. C. Young

Central Resegrch Library

Y-12 Document Reference Section
Laboratory Records Department
Laboratory Records Department (RC)

External Distribution

 

A. Agostinelli, Worthington Corporation, Harrison, N.J. 07029

C. E. Anthony, Electro-Mechanical Division, Westinghouse Electric
Corporation, Box 217, Cheswick, Pa. 1502k

F. F. Antunes, Ingersoll-Rand Company, Cameron Engineering
Division, Phillipsburg, N.J. 08865
R. G. Barrett, Foster Wheeler, Livingston, N.J. 07039

R. N. Bowman, Pump Division, Bingham-Willamette Company, 2800
N.W. Front Avenue, Portland, Ore. 97210

A. H. Church, Mechanical Engineering Department, New York
University, Bronx, N.Y. 10400
99.

100.
101.

102.
103.

104—105.
106.

107.
108.
109.
110,
111.

112.

- 113,
11k,

115-132,
133-134,
135-137,

138-139.

130

Gary Clasby, Byron Jackson Pumps, Inc., P.0. Box 20177, Terminal
Annex, Los Angeles, Calif. 9005k

D. F. Cope, RDT, SSR, AEC, ORNL

V. R. Degner, Rocketdyne Division, North American Aviation,
Canoga Park, Calif. 91303

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

D. Elias, RDT, USAEC, Washington, D.C. 205L5

J. R. Fox, USAEC, Washington, D.C. 20000

A. Giambusso, USAEC, Washington, D.C. 20000

F. C. Gilman, Pump and Heat Transfer Division, Worthington
Corporation, Harrison, N.J. 07029

R. Gordon, Aerojet-General, Azusa, Calif. 91702

C. W. Grennan, Colt Industrles, Chandler Evans Control Systems
Division, West Hartford, Conn. 06100

Norton Haberman, USAEC, Washington, D.C. 20000

F. G. Hammitt, The University of Michigan, Ann Arbor, Mich.
48103

M. J. Hartmann, NASA-Iewis Research Center, Cleveland, Ohio
44100

C. H. Hauser, NASA-Lewis Research Center, Cleveland, Chio 44100
J. W. Holl, Pennsylvania State University, Garfield Thomas Water
Tunnel, Ordnance Research Laboratory, University Park, Pa. 16802
Kermit Laughon, USAEC, OSR, ORNL

Liquid Metal Englneerlng Center, c/o Atomics Internatlonal,
P.0. Box 309, Canoga Park, Calif. 91303 (Attention: R. W.
Dickinson)

T. W. MecIntosh, USAEC, Washington, D.C. 20000

D. C. Reemsnyder, NASA-Iewis Research Center, Cleveland, Chio
L4100

M. A. Rosen, DRDT, USAEC, Washington, D.C. 20000

Research and Technical Support Division, ORO

M. Shaw, USAEC, Washington, D.C. 20000

W. L. Smalley, USAEC, Oak Ridge Operations

W. A. Spraker, Engineering Research Division, Scott Paper
Company, Philadelphia, Pa. 19100

H. A. Stahl, Cameron Division, Ingersoll-Rand Company,
Phillipsburg, N.J. 08865

‘B. Sternlicht, Mechanical Technology, Inc., Latham, N.Y. 12100

G. M. Wood, Pratt and Whitney Aircraft Corporation, East Hartford,
Conn. 06100

Manager, Technical Information Center, AEC

Technical Information Center (AEC)

Director of Division of Reactor Licensing, Washington, D.C.
20545

Director of Division of Reactor Standards, Washington, D.C.

20545