ORNL-4037.txt

AK R}DGF NATfO‘NAL LABORATORY L

M!i i MHB i

3 445k 0548176 O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 
ORNT-40O3T

Contract No. W-Th0S-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMTANNUAI PROGRESS REPORT
For Period Ending August 31, 1966

R. B. Briggs, Program Director

JANUARY 1967

OAK RIDGE NATTIONAL LABORATORY
ODak Ridge, Tennessee
ocperated by
UNION CARBIDE CORPORATION
for the
U.8. ATOMIC ENERGY COMMISSION

K RIDGE NATIONAL LABORA

R

3 445L 054417k O

 

 

 

 

 

 

 

 

 

 

 

 
SUMMARY

Part 1. Molten~Salt Reactor Experiment

 

1. MSRE Operstions and Analysis

 

The reactor power lewvel wag increased in steps to the maximum at-
tainable value of 7.2 Mw. The power limitation wag imposed by the heat-
transfer capability of the air-cooled radiator, which was much lower
than the design value. The heat-transfer coefficients of the primary
fuel-to-coolant-salt heat exchanger were also substantially lower than
had been expected. Two periocds, each about two weeks long, of relatively
steady operation at the maximum power were achilewved.

Agide from the power limitation, the performance of the reactor
system was Tavorable. The inherent nuclear stability increased with
increaging power as had been predicted. The nuclear poisonlng vy 135%e
was only 0.3 to 0.4 of the expected value, apparently because of the
presence of circulating helium bubbles in the fuel salt which had not
been observed in earlier operations at similar conditions. Zero-power
reactivity balances showed a slowly increasing positive anomaly which
had reached a maximum of 0.3% 8k/k at shutdown. Radiation heating of
the primary-sgystem components was in the expected range, and radiation
shielding and containment were adequate.

A system ghutdown to remove irradiation specimens from the core,
which wag planned after an accumulated exposure of 10,000 Mwhr, was
advanced when a catastrophic fallure of one of the main radlator blowers
occurred at 7800 Mwhr. In addition to the removal and replacement of
the core irradiation specimens and the repair of the main blowers, a
number of other maintenance jJjobs were performed in the ensuing shutdown.
These included:

1. mechanical and electrical repalr of heat-induced damage to the
radiator enclosure,

2. replacement of the particle trap in the main reactor off-gas line,
which had developed intermittent high resistance to {low,

3. modification of the treated-cooling-water system to eliminate
radiolytic gas,

4. repair of water leaks in the reactor cell,
5

. modification of the component~cooling system to improve reliability,

N

general improvement of the in-plant electrical system.

The MSRE instrumentation and controls system continued to perform
well., There was the normally expected reduction in both malfunctions
and misoperation of instruments as instrument and operating personnel
gained experience and developed routines. While there were many design
changes, most of these were improvements and additiong to the gystem
rather than correctlive measures to the instruments and contrcls. A

iii
iv

disappointingly large number of faulty commercial relays and electronic
switches were disclosed. These faults were in the areas of both relay
deslgn and fabrication, and corrective steps have been taken.

2. Component Development

 

The operation of freeze valve FV-103 was improved by the deletion
of the hysteresis feature of one of the temperature-control modules,
which had caused thermal cycling of the wvalve before the ftemperature
reached equilibriunm.

The three conlrol rods have operated without difficulty. The
indiecated changes in rod length were random and were 0,05 in. for rod
1, 0.16 in. for rod 2, and 0.12 in. for rod 3.

Several small difficulties were encountered in the operation of
the control-rod drive units. In each case the difficulty was diagnosed
and adequate temporary changes were made to permit contiunued operation
of the reactor. The difficulties involved the failure of a synchro
transmitter and a reference potentiometer, which have been replaced
during the current shutdown. The excellent condition of the gears in
the drive units indicated that there was no high-temperature damage €O
the lubricant.

The radiator-door-operalting mechanism has performed satisfactorily
since the last modifications. Excessive alr leakage around the seals
ori the doors resulted from damsge sustained during the thermal cycling
during normal operation. Laboratory tests were conducted on several
arrangements of the metal seal surfaces, resulting in the choice of a
new hard-gceal scheme which was installed on the door. In addition,
alterations were made to the door structure to reduce bowing, and the
installation of a second soff,; resilient seal was proposed to back up
the existing seal.

The sampler-enricher has been used to isolate a total of 119 10-g
samples and 20 50-g samples and to make 87 enrichments to the fuel sys-
tem. The only major maintenance required has been replacement of the
manipulator boots, replacement of the drive-unit capsule latch, repair
of an open electrical circuit, and recovery of a capsule which had
fallen into the operational valve area. These repairs were performed
without the spread of airborne contamination or the exposure of the
personnel to significant radiation levels. An increase of the buffer
leakage in the operational valve was determined to be caused by particles
which fell onto the upper seal surface, and it was found that simple
cleaning of this surface was effectlive in reducing the leakage. Problems
resulting from an increase in the contamination level within the mech-
anism were solved by a partial cleanup of one aresa, the establishment of
a contamination control area at the sample withdrawal arca, and the
modification of the transport container to reduce contamination of the
upper part and to permit inexpensive disposal of the lower part.

Changes were made in the sampler~enricher control cireuit to
reduce the chance of rupturing the manipulator boots. In addition, a
fuse and voliage suppressor were installed to protect the electrical
leads {from excessive voltages and currents.

A total of 45 samples, including two 50-g samples, have been taken
using the coolant sampler. One electrical receptacle was replaced, and
the lezkage of the removal wvalve wag reduced by cleaning.

The design of the fuel processing sampler was essentially completed,
and installation i1s proceeding as craflt is avallable. The electrical
and instrument work is about 50% complebe, and the installation of
other equipment is over 95% complete.

The original off-gas Tilter in line 522 was replaced with one
which 1s designed to trap organic materials in addition to particulate
matter. Activated charcoal was chosen as the filtering medium, and
preliminary tests indicated that 1t had good efficiency for the removal
of Cg and heavier molecules. A prefilter was installed to remove the
radicactive particulate matter as well as the organic mists which might
exist. Since one of the purposes of the prefilter was to reduce the
heat load on the charcoal trap, heat dissipation was also a consideration
in the design. The entire filter—charcoal trap is cooled by immersion
in watey. Bince little was known of the character of the crganic ma-
terial at the time of the design of the filter system, one of the pur-
poses was to provide a method of diagnosing the problem. Therefore,
the particle trap will be removed and examined to gain information
needed in the design of a more permanent system.

Diffusion of activity into the fuel pump off-gas line resulted in
two short periocds of high activity at the off~-gas stack. A small char-
coal trap was installed to provide holdup times of approximately 2-1/2
days for krypton and 30 days for xenon.

Several remote maintenance jobs were performed during the periocd,
in which the accumulated reactor power increased fTrom 35 to 7822 Mwhr.
Several observations were made as a result of the work. Control of air-
borne contamination is not difficult, and the maintenance techniques and
systems prepared for the MSRE have worked well. The flexibility of the
maintenance approach was demonstrated by carrying oul unanticipated
tasks such as the installation of a thermocouple on a piece of pipe in
the reactor cell and the thawing out and clearing of a plug in one of
the service gas lines. The major tasks completed during the report in-
clude: (1) opening a section of the off-gas line, inspecting the inside,
and returning the line to operating condition; (2) removal and replace-
ment of both of the reactor-cell space coolers; (3) installation of =
new thermocouple on a horizontal seciion of the off-gas line; (4) repair
of the sampler-enricher electrical receptacle; (5) installation of tem-
porary piping in the main off-gas line to measure pressure distributions;
(6) removal of the graphite—flastelloy N surveillance samples; (7) re-
moval, repair, and replacement of two control-rod drive units; (&) re-
moval of a salt plug from & gas~pressure reference line by applying
pressure while heating the line.
Vi

3. Pump Development

The MSRE prototype fuel pump was operated for 2631 hr at 1200°F to
obtain data on the concentration of undissolved helium in the circulating
salt and on the hydrocarbon concentration in the pump-tank and catch-
basin purge gases. The spare rotary elements for the MSRE fuel and
coolant pump and the MK-2 Tuel pump were modified with a seal weld
between the bearing housing and shield plug to prevent oil from leaking
out of the leakage catch basin and down the outside of the shield plug
to the pump-~tank gas space. A lower shaft seal fallure was experienced
during preheat of the spare rotary element for the MSRE fuel pump in
preparation for hot shakedown. The spare rotary element for the MSRE
coolant pump was given cold and hot shakedown tests. The lubrication
pump endurance test was continued, and fabrication of the MK-2 fuel
pump tank was begun.

Operation of the PK-P molten-salt pump was halted by failure of
the drive motor. (In the summary section of a previous Progress Report,
the number of operating hours was reported in error as 22,622. Total
for four tests is 23,426 hr.) The gimbals support for the salt bearing
on the molten-salt bearing pump was modified, and a new bearing sleeve
and journal were fabricated.

4. Instrument Development

 

Performance of the temperature scanning system continues to be
satisfactory, although some problems were experienced with oscilloscopes
and mercury switches and some system instability was noted. Because
spare parts for the mercury switches used in the scanner can no longer
be obtained from the manufacturer, an effort is being made to find a
replacement for the mercury switch.

Further testing of the coolant-salt system flow transmitter which
failed in service at the MSRE confirmed that refilling the transmitter
with silicone oil had significantly reduced its temperature sensitivity.

The new NaK-filled differential-pressure transmitter ordered for
use as an MSRE spare was found to be excessively sensitive to pressure
and temperature variations during acceptance testing.

Performance of all molten-salt level detectors installed at the
MSRE, on the MSRP level Test Facility, and on the MSRE Prototype Pump
Test Loop continues to be satisfactory.

To correct excessive frequency drift present in the excitation
oscillator supplying the ultrasonic lewvel probe, a number of minor
changes in components and circuitry were made in electronic equipment
associated with the probe.

Modification and/or repalir of two defective helium control valves
was completed.

The feasgibility of using sliding disk valves for control of very
small dry-helium flows is being investigated.
5. Reactor Analysis

 

Rod drop experiments, performed during MSRE run No. 3, were analyzed
and compared with rod worths determined from other independent measure-
ments. Theoretical time~integrated flux trajectories following rod
scrams were calculated, based on negative reactivity insertions obtained
by integrating differential worth measurements. These trajectories were
found to compare closely with experimental reccrds of the accumulated
count following the scram. We have concluded that an approximate 5%
band of self-consistency can be assigned to the control rod reactivity
worths inferred Irom these two independent calibration technidues.

Part 2. Materials Studies

6. MSRP Materials

The grade CGB graphite and Hastelloy N specimens were removed from
the core of the MSRE after 7800 Mwhr of operation. Thelr macroscopic
appearances were essentially wunchanged by this exposure. Some of the
specimens were damaged physically as a result of differences in thermsl
expansion of parts of the assembly. A new core gpecimen array was as-
sembled with modifications to correct these difficulties.

A metallurgical investigation was conducted to determine the effect
of aluminum-zinc alloy contamination on the Hastelloy N tubing of the
MSRE salt-to-ailr radiator. Contamination occurred from a blower fallure
during which shrapnel was blown across the hot radiator tubes. ILabora-
tory tests showed that, in general, an aluminum oxide coating contained
the aluminum, even in the molten state, and interaction did not occur.
When the oxide skin wag broken from mechanical abrasion, shock, or other
reasons, wetting occurred. Moderate interaction to a depth of about
0.010 in. occurred in a wetted sample held at 1200°F for 5 hr. The tubes
in the radiator were inspected, and those which were contaminated were
carefully cleaned. As a result of the investigation and cleanup pro-
cedure, the radlator system was Judged to be satisfactory for further
operation. ;

Examinations of new grades of both anisotropic and isotropic graphite
indicate that these do not yet meet the reguirements of molten-salt
breeder reactors.

Results of a variety of graphite creep experiments performed over
a wide temperature range support a Cottrell model for irradiation creep.
Use of this model will permit easier exbrapolation of data. Implied in
this concept is the conclusion that as long as the stresgs acting on the
graphite does not exceed the fracture stress, the graphite will continue
to absorb the creep def'ormation without loss of mechanical integrity.

The experimental brazing alloy 60 Pd—35 Ni—5 Cr (wt %) was evaluated
for joining graphite to metals. Although it exhibits relatively poor
wettability on high-density graphite, 1its marginal behavior is enhanced
by preplacing it as foll in the joint. Graphite-to-molybdenum joints
brazed with this alloy preplaced in the joint were thermally cycled
between 200 and 700°C, and metallographic investigation showed that no
deterioration had occurred. Two graphite-to~-molybdenum~-to-Hasgtelloy-N
viii

transition joints were brazed using a tapered joint design. Visual
examination revealed no cracks, and evaluation i1s continulng.

A new brazing alloy, 35 Ni—60 Pd—5 Cr (wt %), for joining graphite
to molybdenum had less than 2 mils attack after exposure to LiF-~-Belp-
ZrF,~ThF,~-UF, at 1300°F for 5000 hr in a Hastelloy N container.

Since zirconium and titanium have been found to improve the resis-
tance of Hastelloy N to effects of neutron irradlation, the 1nfluence
of these elements on the weldability is being evaluated. Titanlium ap-
pears to have no deleterious effects. Zirconium in concentrations as
low as 0.06 wt % causes hot cracking. However, reasonably good welds
have been made by the use of filler wire that contains dissimilar metal.
The level of zirconium that can be tolerated has yet to be determined,

Our studies of the behavior of the Hastelloy N under neutron irra-
diation have been concerned with evaluating the heats of material used
in the MSRE and in evaluating several modified heats of Hastelloy N for
use in an advanced system. In-reactor stress-rupbure tests on several
heats of material suggest that there may be a stress below which es-
sentially no neutron damage occurs. Tests on specimens exposed to var-
ious ratios of fast flux to thermal flux indicate that the dsmage cor-
relates with the thermal flux. We believe that the damage 1s due to
helium produced hy the (n,&) reaction with 98, We have produced several
heats of material with very low boron, but have had to change from air
to vacuum melting practice. If the low-boron material is irradiated cold,
we find that the properties are superior to those of the higher boron
material; if irradiated hot, the properties are as bad or worse. Thus
the postirradiation properties are not uniquely dependent upon the boron
content, and other factors such as the distribution of boron and the
presence of other impurities must be very important. The addition of
titanium to Hastelloy N has been very effective in improving the prop-
erties.

Limited creep-rupture tests have been run on Hasteloy N to deter-
mine its suitability for use as a distillatlon vessel for molten salts
at 282°C. This work has resulted in a determination of the strength
propertice to times of 1000 hr. The formation of a second phase was
observed that may influence the ductility at lower temperatures. The
oxidation characteristics under cyclic temperatures remain to be deter-
mined.

Thermal convection loops containing Tused salt of MSRE composition
and fabricated from Hastelloy N and type 446 stainless steel have con-
tinued cperation for 4.5 and 3.2 years, respectively, with no sign of
difficulty. A slight decrease in cold-lcg temperature has been noted in
an Nb—1% Zr loop after 0.6 year. A Hastelloy N loop has circulated a
secondary coolant salt for 3C00 hr, whereas a Croloy 9M loop with the
same salt plugged in 1440 hr.

7 Chemistrz

The fuel and coolant salt have not changed perceptibly in composition
gince they were first circulated in the reactor some 16 months ago. The
ix

concentration of corrosion products has not increased appreciably. The
average oxide concentration in the fuel was 54 ppm, which is reassuringly
low.

The viscosgity and density of molten Bels were measured; the vis-
cosity was about 10% greater than previously reported, and the density
of the ligquld is not very different from that of the solid.

Vapor eguilibria that are involved in the reprocessing by distil-
lation have been measured. Decontamination factors of the order of 1000
Tor rare earths were evidenced.

Thermophysical properties have been estimated for the sodium potas-
aium fluoroborate mixture that dis a proposed coolant for the M3BR. The
vapor pressure, due to evolution of BF;, reaches 229 mm at the highest
operating temperature. Interim estimates for density, specific heat,
and viscosity of the proposed coolant were made available.

Possible reprocessing methods were studied 1ln greater detail. Fun-
damental studies related to the thermodynamics of the reduction of fission
product rare earths into a bismuth alloy were carried out. The exceed-~
ingly low activity coefficients of rare earths in the bismuth explained
the feasibility of the process. Further attention was pald to the re-
moval of rare earths by precipitation in a solid solution with UFs.

The removal of protactinium from blanket melts was studied in sev-
eral ways. These included an oxide precipitation with ZrO,, a pump loop
to transfer protactinium in a bismuth-thorium alloy, and attemptes at
electrolytic reduction from blanket melts. Moderate success was achleved
in these experiments, but more work is required to arrive at finished
and fully controlled procedures.

Analyses obtained from sampling assemblies that had been exposed
in the pump bowl of the MSRE showed thal noble-metal fission products
were being partially released to the gas space, presumably asg volatile
fluorides. At the same time, plating of noble metals from the liquid
was encountered. These puzzling phenomena were reflechbed in results on
surveillance specimeng of graphite and metal which were removed from the
MSRE. Some 10 to 20% of the yield of noble-metal fission products was
found to have entered the gas space in the graphite.

Analyses of xenon isotope ratios in concentrated samples of off-gas
from the MSRE showed that the burnup of 135%e wags about 8%; the remainder
escaped to the cover gas or decayed. This is in accord with the low
xenon polsoning indicated by reactivity behavior.

Preliminary estimates of xenon poisoning and cesium carbide for-
mation in the MSER indicate that cesium deposition will probably not be
a serious problem, but that stripping for iodine removal will probably
be required Lo keep poiscning within bounds. Oxide concentrations of
50 to 70 ppm were determined in fuel samples taken from the reactor
duriag operations at all power levels without apparent interference from
the activities of the samples. Techniques for the regeneration of elec-
trolytic moisture cells were developed to provide dependable replacementg
for the hot-cell oxide apparatus and components for future in-line ap-
plications.
Measurements directed toward the development of in-line speclro-
vhotometric methods disclosed additional wavelengths of potential analyt-
ical value in the ultraviolet absorption spectrum of U(III) and confirmed
the absence of interference from corrosion products. Investigation of
unusual valence states of rare-earth fission products indicates possible
interference from Sm(IT) but none from Eu(Il). An intease absorption
peak suitable fTor monitoring traces of uranium in coclant salt has been
found in the ultraviolet spectrum of U(IV). A modified optical system
has been ordered which will improve the spectrophotometiric measurcments
of molten Tluoride salts.

By voltammetric and chronopotentiometric measurements, the U(IV)
reduction wave was found to correspond to a one-electron reversible re-
action. Diffusion coeflicients and the activation energy were measurcd.
Repeated scans of this wave in gquiescent MSREE wmelts were reproducible to
about 2% over extended periods and better than 1% during short-term
measurements, A new voltammeter is being bulilt to improve the reproduc-
ibility and make possible measurement of flowing salt streams. Design
criteria are being considered for an in-line test facility for evaluating
three types ol continuous analytical methods.

Hydrocarbons were measured in helium from simulated pump leak ex-
periments, and an apparatus was developed for the continuous measurement
of hydrocarbouns in MSRE off~-gas.

Efforts were continued on the development and evaluation of equip-
ment and procedures for analyzing radioactive MSRE salt samples. The
remote apparatus for oxide determinations was installed in cell 3 of the
High~-Radiation-Tevel Analytical Laboratory (Building 2026).

In addition to the analyses performed on the salt samples, radio-
chemical leach solutions were prepared on silver and Hastelloy N wires
coiled onto the stalnless steel cable between the latceh and ladle.

The quality-control program was continued during the past period
to establish more realistic limits of error for the methods.

Part 3. Molten-Salt Breeder Reactor Studies

8. Molten-Zalt Breeder Reactor Design Studies

Further design changes were incorporated into the reference molten-
salt breeder reactor concept. The design of the primary heal exchangers
wag altered to eliminate the need for expansion bellows, Also, the flow
of fluid in the primary reactor circuits was reversed to lower the oper-
ating pressure in the reactor vescel.

The effect of lowering the feedwater temperature from 700 to 580°F
was evaluated. Tt was found that this change increased the plant thermal
efficiency from 44.9 to 45.4% and reduced plant construction costs by
$465,000 if there were no accompanying adverse effects. These savings
are canceled if the coolant used with the lower feedwater temperature
costs $2.4 million more than the coolant used with 700°F feedwater.

Molten-salt reactors appear well suited for modular-type plant con-
struction. Such construction causes no significant penalty to either
X1

the power-production cost or the nuclear performance, and it may permit
MSBER's to have very high plant-availability factors.

Use of direct-contact cooling of molten salts with lead significantly
improves the potential performance of molten-salt reactors and indicates
the versatility of molten salts as reactor fuelsg. However, in order to
attain the technology status required for such concepts, a development

program 1s necessary.

The molten~galt reactor concept that redquires the least amount of
development effort is the MSCR, but 1t is not 2 breeder system. The
equilibrium breeding ratio and the power-production cost of the MSCR
plant were estimated to be about 0.96 and 2.9 mills/kwhr (electrical),
regpectively, in an investor-owned plant with a load factor of 0.8.
Although this represents excellent performance as an advanced converter,
the development of MSBR(Pz) or MSBR plants appears preferable because of
the lower power-production costs and superior nuclear and fuel-conser-
vation characteristices associated with the breeder reactors.

9. Molten-8alt Reactor Processing Studies

The processing plant for an MSBR would use side streams withdrawn
from the fuel- and fertile-salt recirculating systems at rates that
vield a fuel-salt cycle time of approximately 40 days snd fertile-salt
cycle time of approximately 20 days. Among the significant steps in the
presently envisioned process are recovery of the uranium by continuous
fluorination and recovery of the carrier fuel salt by semicontlinuous
vacuum distillation. Alternative schemes are also belng conszidered.

Semicontinuous Distillation. Values of the relative volatilities
of NdF;, Lal;, and CeFs; in LiF are of the order of 0.0007. These are
from new measurements made using a recirculating equilibrium still.
Barlier measurements made by a cold-finger technique were zbout a factor
of 50 too high. The complexity of still design and operation is consid-
erably eased by these lower values. Retention of over 209 of the rare-~
earth neutron poisons in less than 0.5% of the processed salt can easily
be achieved.

 

Continuous Fluorination of a Molten Salt. The uranium in the fuel
stream of an MSER must be removed by continuous fluorination pricr to
the distillation step. The significant problems are corrosion of the
flucrinator and the possible loss of uranium. Shtudies are in progress
on continuous fluorinators constructed as towers with countercurrent
flow of fluorine to salt. Recoveries exceeding 99% have been consistently
attained with towers only 48 in. high. Higher recoveries with longer
towers are anticipated. Corrosion protection may be effected by the use
of a layer of frozen salt on the wall of the fluorinator. Feasibility of
this technique is based on successful experiments with batch systems and
simple heat transfer calculations. The heat generation oif the fuel salt
should be adequate to maintain an easily controlled layer of frozen salt
on the cooled metal wall.

 

Alternative Chemical Processing Metheds for an MEBR, Reductive co-~
precipitation and liquid-metal extraction are being studied as possible

 
x1i

methods for decontamination of MSBR carrier salt (LisBeF,) after uranium
removal by the fluoride volatilization process. Adequate removal of La
and Gd. is achieved by treatment with near-theoretical quantities of beryl-
lium metal to form beryllides of the type InBeq,. Either excess Be, up
to 243 times the theoretical amount, or a stronger reductant, Li, is nec-
essary Lo remove zirconium at trace level. The zirconium is removed as
free metal from 5 mole % ZrF, solutions in LipBeF,, but from dilute
solutions a beryllide, ZrBes, has also been identified.

Lithium~-bismuth liquid-metval extraction experiments were also con-
tinued. Significant removals were observed for La, Sm, G4, Sr, and Eu,
the latter principally by extraction into the metal phase, the others by
deposition as interface solids, as previously reported for other metal
extraction tests.
CONTENTS

Sm....l..ll..."..lll.li.!..tl....l‘.l" ------ > a &

IMRODUCTION.......Q......I.-.O...-’I.l.........'.-...-n'.fllifll'.. l

Part 1. MOLTEN-SALT REACTOR EXPERIMENT

1. MSRE OPERATTIONS AND ANATLY SIS.eseessessassscscsecsassasnsanancs '

1.1 Chronological Account of Operations and Maintenance..... 7
1.2 Reachivity BalanCe.eiessnesesnsssssrsssancassassnsscssnns 10
Reactivity Balances 8 POWer.ceeesseecacscerarssannons 11
Reactivity Balances at LOWw PoWwer..seeacscceccscnnense 12

1.3 Nenon POlcONingeesessocsssoccesnacossrrssssernsasnssnssns 13
Predicted Steady-State 12°Xe PoisOning.cevsseessssees 14
Analysis of Transient *2°Xe Poizoning..e.esseeeess--o 16

4 Cireulating Bubbles.sseveseseeerons ceeearrreaanaannn ceus 22
5 Salt TranspoTt.eeseesscecocrsaecane cessesccacscsarnuaeae 2%
Gradual Transfer Lo Overflow Tank....ceeeeesvescscess 24
OVEYTIi]l ) s ssoenasnssssasseansssnnarsassssossnsenasosssan 24

1.6 Power MeasurelenlS.ieeesenosassasssscaasasassonnn venesane 25
Heat BalolCO.seeesssscscassascssssossonsasssssssaansanas 25
Muclear INstrumentSseesessssecosssscssessossannesas . 26
Radiator ALlr FlOW.sesesseraanase frsesesntseasaessennes 26
1.7 Radiation Heabting..eeeeessceaacsssasserssossssssassssenns 27
Fuel-Pump Tank.eseeeseessae breees chessaesasssrssenb e 27
Reactor Vessel.eerneessrasssnaneas ceneaen eraaeas coees 28
Thermal Shield.ecseeacsssnssocsnssssvscscsaasvocassnnss 29
1.8 Reachor DynamicS.ceeessesesssscsssssssssrsssasenessassss 29
1.9 TEquipment Performance....ceeseasossssversscessonsocconss 35
Heat Trancleri.ieecscossasseans cenaa teeceesersenacnse 35
Main BlOWerSeesecescassorsocnsonsnas tessssasnesnerses 39
Radiator ENcloSUl€.casssevssesssasssasssssncssonssenssens 4.
Fuel Off-Gas Syslem..eescaeeassceorsesssseasnsoasnacs 43
Trested Cooling-Water System....eiieiecesercacsrsonns 47
Component~Cooling Systell.ceesssoscoassassnsascascsns . 48
Salt~Pump O1l SySTEmMS: ccveasessansasssesanccasssasnss 50
Electrical System...ceeecevoeas pacesrarsacarvasnaseans 51
Control Rods and Drives.se.sesesrasosnnsansss e aamesee 53
SaMP LB S e e aaessssssosssescasnassasssseassesssassersssss 53
Contalnment .. oeescessseressssassassasnsnnssssscnnnnas 54
Shielding and RadiafioNeesascesescsrssonsaas cascasaas 56
1.10 Instrumentation and ControlS..ssesessscscssssassesasonss 57
Operating Experience — Process and Nuclear
Instbruments.cesesenoscaaeasvas eessesarnsnoasrennnee 58
Daba SyShell.eeosenvssasnasesnscsssarsasoscecssssssnss 59
Control-System Designe cesosceassssasososssaasasassnas ol

2& COD’[PONEN‘I]:‘ DEVELOPIVEIq’mI‘l...l..fll..i.0‘t'fl.l..l...fl.l'i.flll'ln' 6-')

xiii
Xiv

2.1 Ireeze ValveS.ieaeeeieeersosssetseesstsscscnecssascnceacsscse 65
2.2 Control ROUSeecesesseasessssocsssscaasosacscsosassosossseasse 65
2e3 Control-Rod Drive UnitSeeeeesececsscsescssoasosssnossosesss 66
2od RAdiatior DOOT Sessesssssssrsssssssarssassasscssscvcssssaess o7
265 oD e -l GO s aseasseesasossessrsssssssnscsscsssnanssa 70
Replacement of Manipulator BOOLS.ssssssscsssesacsanna 70
Replacement of the Capsule LabCh eeeeseesesresccerons 71
Repair of an Open Electrical Circuit..sceececscsascas 7L
Recovery Of a CapslUlCasscesessossracsssscassecesansascea 72
Cperational Valve 1eakige.sececcscasscssasnosnaconcea 72
Contamination of Removal Valve Se8lSeiecescasscrssonsans 72
Miscellaneous ProblelS.ceecescessessserserssacrersannss 73
Changes to the Control Circuit..eceeeesecsacsoesasses 73
Coolant SampPleTesesessaessscsssesrssssssvsossscossrsonsssns 73
Fuel Processing System Sampler.sssecscscescacessnsansans 7
Off~Cas Tllter Ascemblyeseecesesoastassrsesorscaccssnessse "4
Filter MediUMeceesesssssasossrascassasseanacsnssessnaa T4
Heat Diceipalion,. caeessseescsssssaossososssntrsoncroases M
Particle TrDesscesessarsasscsaccasssssscoossesssscesess 75
Charc08l TraPeersscccsensssscecscscssesssanssessacssessss 76
2¢9 524 Charcoal BeQeeessssscsscsscassssssssnesnssnsssnsssossos 7
2.10 Remote MalntenanCE.ceeessesccsssessesessoaenssnasosansena 77

oMM
o -3 O

Pm@ .DEVEIJOP}JENT..."....'...‘..-.........-....I............l. 81

3.1 MORE PUlDSesessasesasoscsascssansscsssssosscesnastoscnsossanns g1
Molten~-Salt Pump Operation in the Prototype Pump
Test Facllifyeeesstssesaessassesscanssssnscacsncnss 81
Pump Rotary Element Modificabtions.esesseecesscceasoas 82
Iubrication SysteMececcesssssesssssessaseesssssnnesss 82
MK=~2 Fuel PUllDessessssvssscssasnsssconssensecscancecses 82
3.2 Other Molten-8alt PuMpPSeesacscececssescavoasososansssons 82
PK-P Fuel-Pump High~Temperabture Endurance Teot.eeeee. 82
Pump Containing a Molten-Salt-Lubricated Journal
Bear N escsaessessessssnessesnsassssncscsasnssssasans 82

INSTR[M]NTDEVELOPI%]NT...II.‘..C.....I..I..lDl...l.ll..l..'..l 84‘

4. 1 Ten‘.l)e 1‘.atux.e Sc al‘l-rler * o 5 o0 PSR PSR RS E T S e ST P e R e G S e e R 8ZE'
4e2  High-Temperature, NaK-Filled, Differential-Pressure
TransmittEr. 8 5 0 8 9 0 PR e SO TSrE S PO S RSP E SRS SN RS S e S 85

lee 3 Molten~3alt Tevel DeteCtOrSeeessosccssassessosscccssssssa 85
doet  Helium Control Valve Trim Replacellenl.eesesececasccosess 86

mACTORANA:LYSIS.-..'...‘......I....‘....‘....l.........'.-..l 88

5.1 Analysis of Rod Drop ExperimentS.cecececsrsenssoacsacsnss 88
Description of EXperimentS.eeeeeececscesscscoseasossas 88
Analysis ProCcelUreS.esessecssassarssssscsarscssassanas g9
RESULllSeesevseasessansssncssecsassocasacsasassarsnnsana 91
6. I&SI%P D&aterials.....“...I-I..‘D.ll

6.1
.2

o Oy
W

SOy Oy
-
~3 O\

6.8

XV

Part 2. MATERIALS STUDIRS

4 & 2 8 8 9% ¢ 90 & 00 S P DI R ST S A B

MERE Materials Surveillance Testing.sceesvessasssessacos
Evaluation of Possible MSRE Radiator Tubing

Contamination with Aluminum,..

2 9 o % 00

2 # & 2 2 @ % 3 2 2 & ¥y 00 > TR

Evaluation Df Graphitei..'I.I...C'.I.ll...l‘lllD‘.ll..l.
Internal Stress Problem in Graphite Moderator

Blocks...llI‘...ll..lll...'hfll
Brazing of Graphite.....cssevces

480 0

A8 ST S sERA BB F R RaE

Corrosion of Graphite~to-Metal Brared Jointo.i.eeeeeoveas
Welding Development of Hactelloy Nesecessseassascsaonasas
ffect of Irradistion on the Mechanical
Propertics of Hastelloy Neeseseseescnsscorscsocnsco
Characterivation of Hastelloy N for Service
At F82% Ciernnenonennnsarucnansasossnnsnsossnnassasss

Thermal Convectbion Loops...........,....................

r?. CI{EMISTI{I"....’..."...I....."....‘0............-l’i‘....’fi..‘

7el

7o

7.3

77- ZI’

765

Chemistry of the MORE. eeevecoacsassssssnsnsnconssnsnass
Behavior of Fuel and Coolant Salt.sssesesesvacsssnces

Physical Chemistry of Fluoride Mellt.eeeeeesersseancanes
Viscosity and Density of Molten Beryllium

F:Lu.oride-p..---o..--oaoounn

------

Transpiration Studies in Support of the Vacuum
DiStillationPI‘OCGSS.'..IO......B...III.I..'O..O..’
Estimated Thermophysical Properties of MSBR

Coolan-t Salt‘.......l...-‘.

& & 9P oS s S PR e T eSS

Separation in Molten FluorideS.sasseccsssessescsssssccee
Extraction of Rare Tarths {rom Molten Fluorides
1nto Molten MebtalSeeeesossesonssassssnasass-snnesnss
Removal of Rare Earths from Molten Fluorides by
Simultaneous Precipitation Wwith UFesesscsnsssssaas
Removal of Protactinium from Molten Fluorides by

Oxide Precipitationesscecsass

& ® ¢ g » o & 5 & 0" a0 a8 9 &

Extraction of Protactinium from Molten Fluoridesgs
into Molten Mebtialo.ieearsesesoessssnnoacsnmesssnacass
Protactinium Studies in the High-Alpha Molten-
Salt Laboratoryesescaseessssscrsacsorssssscasnesssoa
Radiation Chemistry.iesereaeesonsocsncssesasrsssassarsnans
Xenon Diffusion and Possible Formation of Cesium
Carbide in an MSBReesessesaoreesa
Fission Product Behavior in the MORE..eeesreesossovas
Development and Evaluation of Analytical Methods
Tor Molten-5alt ReaC b0 e eerassssocescsesssaracsnseres
Determination of Oxide in Radicsactive MSEE

* 0 4 8 & F 0 88 S A S e 0 ad

SampleSO.C"l....l..'.l’l."..ll...ll.l'...'.@fl!.-.

Spectrophotometric Studies of Molten-3alt Reactor

mels....ll.....,.l...l.-l.

......

® & * g % b &9 S5 @0 e s P

9'7
o7

103
108

110
112
115
117
117
125
130

134
134
134
139
139
140

143
142

142
145
147
148

156
158

158
165

191
191

193
740

xvi

Voltammetric and Chronopotentiometric Studies of

Uranium in Molten LiF-BeFy-ZrF,.
In-Line Test Faclilityeeeareoeecans
Analysis of Helium Blanket Gas....

2 5 & 8 % &8 P & s 2

Development and Evaluation of Equipment and Procedures
for Analyzing Radiocactive MSRE Salt Samples..ieeseasss

Sample AN8lySeS.isesssssossvnsansonss

Q,ualitwaO'fltl"Ol PI‘OgI‘Elm. « 8 0P 08 s e

Part 3. MOLTEN~SALT BREEDER BEACTOR STUDIES

MOLTEN-SALT BREEDER REACTOR DESIGN STUDIES..s.e..

8.1
8.2

8
8.

™~ W

Design Changes in MSBR Plant.....eeee

Modular-Type Plant..eeeseessessssonese

¢ 8 2 & ¢ o9 5 0 0 b e

oteam Cycle with Alternative Feedwater Temperalure......

Additional Design ConceplSeesscscesas

MOLTEN~-SATT REACTOR PROCESSING STUDIES.....

9.1
9.2
9.3

Semicontinuous Distillations.eeeeceess
Continuous Fluorination of a Molten Salt..

» = » 9 8

Alternative Chemical Processing Methods for an MSBR.....

Reduction Precipitation.c.ceeeeess
Li_Bj_ A.lloy EXJGl"aCtiOIl. & % 48848 s

195
196
19

199
200
200

207
207
212

217
223

227

228
232
233
235
237
INTRODUCTION

The Molten-Salt Reactor Program is concerned with research and de-
velopment for nuclear reactors that use moblle fuels, which are solu-
ticns of fissile and fertile materials in suiltable carrier salts. The
program is an outgrowth of the ANP efforts to make a molten-salt reactor
power plant for aircraft and is extending the technology originated there
to the development of reactors for producing low-cost power for civilian
uses.

The major goal of the program 1s to develop & thermal breeder re-
actor. Iuel for thils type of reactor would be 233UF4 or 235UF4 dissolved
in a salt of composition near 2LiF-BeF,. The blanket would be Thl, dis-
solved in a carrier of similar composition. The technology veing devel-
oped for the breeder is applicable to, and could be explolted sooner in,
advanced converter reactors or in burners of fissionable uranium and
plutonium that also use fluoride fuels. Solutions of uranium, plutonium,
and thorium salts in chloride and fluoride carrier salts offer attractive
possibilities for mobile fuels for intermediate and fast breeder reactors.
The fast reactors are of interest too, but are not a significant part of
the program.

Our major effort is being applied to the cperation and testing of a
Molten-Salt Reactor Experiment. The purpose of this experiment 13 1o
test the type of fuels and materials that would be used in the thermal
breeder and the converter reactors and to obtain several years of ex-
perience with the operation and maintenance of a small molten-~salt power
reactor. A successful experiment will demonstrate on a small scale the
attractive features and the technical feasibility of these gsystems for
large civilian power reactors. The MSRE operates at 1200°F and at
atmospheric pregsure and was intended to produce 10 Mw of heat. Initially,
the fuel contains 0.9 mole % of UF,, 5 mole % ZrF,, 29.1 mole % Bel'y, and
65 mole % LiF, and the uranium is about 30% “2°U. The melting point is
340°F. In later operation, we expect to use highly enriched uranium in
the lower concentration typical of the fuel for the core of a breeder.

In each case, the composition of the solvent can be adjusted to retain
about the same liquidug temperature.

The fuel circulates through a reactor vessel and an external pump
and heat-exchange system. All this equipment is constructed of Hastelloy
N,l a new nickel-molybdenum-chromium alloy with exceptional resistance

to corrosion by molten fluorides and with high strength at high tempera-
ture. The reactor core contains an assembly of graphite moderalor bars
that are in direct contact with the Tuel. The graphite 1s a new material?
of high density and small pore size. The fuel salt does not wet the
graphite and therefore should not enter the pores, even at pressures well
above the operating pressure,.

 

1a1s0 sold commercially as Inco No. 806.
2Grade CGB, produced by Carbon Products Division of Union Carbide
Corp.
Heat produced in the reactor is transferred to a coolant salt in
the heat exchanger, and the coolant salt is pumped through a radiator
to dissipate the heat to the atmosphere. A small facllity is installed
in the MSRE building for occasionally processing the fuel by treatment
with gaseous HF and Fs.

Design of the MSRE was begun early in the summer of 1960. Orders
for special materials were placed in the spring of 196]. Major modifi-
cations to Building 7503 at ORNL, in which the reactor is installed,
were started in the fall of 1261 and were completed by January 1963.

Fabrication of the reactor eguipment was begun early in 1962. Some
difficulties were experienced in obtaining materials and in making and
installing the equipment, but the essential installalions were completed
so that prenuclear testing could begin in August of 1964. The prenuclear
testing was completed with only minor difficulties in March of 1965.

Some modiflcations were made before beginning the critical experiments

in May, and the reactor was first critical on June 1, 1965. The zero-
power experiments were completed early in Jduly. Additional modifications,
maintenance, and sealing and testing of the containment were required be-
fore the reactor began to operate at appreciable power. This work was
completed in December.

Operation at a power of 1 Mw was begun in January 1966. At that
power level, trouble was experienced with plugging of small ports in
the control valves in the off-gas system by heavy liquid and varnish-
like organic materials. Those materials are belileved 1o be produced
from a very small amount of oil that leaks through a gasketed seal and
into the fuel salt in the pump tank of the Tuel circulating pump. The oil
vaporizes and accompanies the gasecus fission products and helium cover
gas purge into the off-gas system. There the intense beta radiation
from the krypton and xenon polymerizes some of the hydrocarbons, and the
products plug small openings. This difficulty was largely overcome by
installing an absolute filter in the off-gas line ahead of the control
valves.

The full power capability of the reactor — about 7.5 Mw under design
conditions — was reached in May. The power was limited by the perfor-
mance of the salt-to-alr radiator heat-dump system. The plant was oper-
ated until the middle of July to the equivalent of =zbout one month at
full power when troubles with the blowers in the heat-dump system re-
quired that the operation be Iinterrupted for maintenance.

In most respects the reactor has performed very well: +the fuel has
been completely stable; the fuel and coolant salts have not corroded the
Hastelloy N container material in several thousand hours at 1200°F; and
there has been no detectaple reaction between the fuel salt and the
grapnite in the core of the reactor. Mechanical difficulties with equip-
mentl have been largely confined to peripheral systems and auxiliaries.

Because the MSRE 1s of a new and advanced type, substantial research
and development effort is provided in support of the operation. Included
are engineering development and testing of reactor components and sys-
tems, metallurgical development of materials, and studies of the chemistry
of the salts and theilr compatibility with graphite and metals both in-
pile and out-of-pile. Work is also being done on methods for purifying
the fuel salts and in preparing purified mixtures for the reactor and
for the research and development studies. Some studles are being made
of the large power breeder reactors for which this technology is belng
developed.

This report is one of a series of pericdic reports in which we de-
scribe briefly the progress of the program. ORNL-3708 is an especially
useful report because it glves a thorough review of the design and con-
struction and supporting development work for the MGRE. It also de-
scribes much of the general technology for molten-salt reactor systems.
Other reports issued in this series are:

ORNIL~2474 Period Ending January 31, 1958
ORNI~-2626 Period Ending October 31, 1958
ORNI-2684 Period Ending January 31, 1929
ORNI-2723 Period Ending April 30, 1959
ORNI~2799 Teriod Ending July 31, 1959
ORNI~2820 Period Ending October 321, 1959
ORNI~2973 Periods Ending January 31 and April 3G, 1260
CRNL-3014 Period Ending July 31, 1960
ORNI~3122 Period Ending February 28, 1961
ORWNL~3215 Period Ending August 31, 1961
ORNI~3282 Period Ending February 28, 1962
ORNT~3369 Period Fnding August 31, 1962
ORNI~3419 Period Ending January 31, 1963
ORNI1-352% Period Ending July 31, 1963
ORNTL-3626 Period Ending January 31, 19064
ORNI~-3708 Period Ending July 31, 1904
ORNI~3812 Period Ending February 28, 1965
ORNI-3872 Period Ending August 31, 1265
ORNI-3936 Period Ending February 28, 1966
Part 1. MOLTEN-SALT REACTOR EXPERIMENT
1. MSRE OPFRATTIONS AND ANAIYSIS

P. ¥N. Haubenreich

1.1 Chronological Account of Operations and Maintenance

 

R. H. Guymon H. C. Roller

J. L. Crowley R. C. Steffy

T. L. Hudson V. D. Holt

P. H. Harley A. T. Krakoviak
H. R. Payne B. H. Webster

W. C. Ulrich C. K. McGlothlan

R. Blumberg

At the beginning of this report period the reactor was down to cope
w1th the problem of plugging in small passages in the fuel off-gas sys-
tem. Investigation showed that the material causing the trouble was a
mixture of hydrocarbons, probably pump lubricating oil that had been
affected by heat and radiastion in the off-gas line. With this informa-
tion, a device 0o clean up the gas stream was deslgned, bullt, and in-
stalled in the off~gas line upstream of the fuel pressure control valve.
It consisted of a trap for particles and mist, followed by a charcoal
ped for vapor adsorpticn. Also during thisg shutdown, the pressure con-
trol valve and the charcoal-bed inlet valves were replaced with valves
having larger trim. ‘

Power operation was resumed in mid-April, and the program of in-
vestigating the performance up to full power was completed by the end of
May. (This period of operation was designated run 6.) During this time,
observations were made as planned on radiation heating, heat transfer,
xenon poisoning, and reactor dynamics. With the exception of heat
transfer, the system behavior was within the expected limits, although
xenon poilsoning was somewhat less than predicted.

The power escalation waz interrupted twice, once when the anomalous
behavior of the control system led to a drain and again by an electrical
failure in the fuel sampler, whose repair required the fuel be dralned.
The approach to full power also disclosed problems in several areas:

1. After about six weeks of operation, pressure drop across the newly
nstalied trap in the off-gas line had bullt up to 10 psilg, and it
seemed 1t would have to be replaced. But during further operation
at high power, the pressure drop went down to less than 1 psig.
Temperatures indicated considerable retention of fission products in
the trap but no evidence of poisoning or deterioration.

o

After each step up in power the pressure drop across the main char-
coal bed inlets increased, hut each time this occurred, bvackblowing
with helium proved effective in restoring the pressure drop to an
acceptable level, ‘

7
3. Radicactive gas from the fuel drain tanks diffused back into the
helium supply lines outlside the shield, requiring & periodic purge
of these lines to keep down radiation in the North Eleciric Service
Area.

4. When the power reached 5 Mw, radiolytic gas began o accumulate in
the thermal shield, displacing part of the water. Efforts to elimi-
nate the gas by venting the shield slides and increasing the water
flow were unsuccessful.

5. The water circulated through the shield contained lithium nitrite as
corrosion inhibitor, and oxidation of the nitrite ion to nitrate was
also observed at the higher power.

6. A discrepancy arose belween the reactor power indicated by the heat
balances and by the neutron instruments (calibrated at low power ) .

1t was found later that the neutron instruments gave errconeously
high readings at high power because of an 1lncrease 1n the temperature
of water in the shaft around the instruments.

The climax of the power escalation was reached when the coolant
heat removal system (air-cooled radiator with adjustable doors, bypass
damper, and two blowers) was extended to its limit, and the capability
proved to be only 7.5 Mw. The coefficient for heat transfer in the pri-
mary heat exchanger was also below expectations, imposing a lower limit
of about 1210°F on the fuel outlet temperature at the maximum power.

In late May, power operation was halted when 1t became evident that
apparent inleakage of air into the reactor cell had begun to exceed the
specified limit. The cell was pressurized to 20 psig for leak hunting,
and no leak of any consequence was found. It was discovered that the
apparent cell leak was actually nitrogen leaking into the cell from
pressurized thermocouple penetrations. When this was taken into account,
the measured reactor cell leakage at 20 psig was acceptably low.

While the cell leak hunt was in progress, attempts were made to de-
termine the source of water which had appeared in the cell atmosphere
(1 to 2 gprd had begun to condense in the component-coolant pump domes,
the coolest spot exposed to the cell atmosphere). 'The source was nob
located because elevated temperatures in the cell prevented meaningful
leak rate measurements on separate portions of the water system.

Operation at full power was resumed on June 13 to Investigate the
chemistry and reactivity behavior of the reactor and to increase the ex-
posure of the core specimens before their removal, scheduled at 10,000
Mwhr. This was ruan 7.

Fuel-salt samples, taken at a rate of three per week, showed no
change in the main constituents of the salt, very low oxide (about 50
ppm), and no appreciable growth of corrosion-product chromium in the
salt. The sampler-enricher was also used to expose metal wires briefly
to the gas in the pump bowl, and thesge showed more fission products than
expected.
Reactivity transients following changes in power indicated that
the 139%e roisoning was only 0.4% 6k/k at full power. The presence of
circulating gas bubbles was suggested by the low xenon poilsoning, and
experiments on the effect of pressure on reactivity confirmed that there
was ag much as 1% by volume of bubbles in the core salt. The system re-
activity showed very 1little net change connected with power operation
other than that atiributed to xenon, even though the calculatsed samarium
poisoning increased to zbout 0.3% 8k/k.

Ancther anomalous behavior was the slow, continuous accumulation of
fuel salt in the overflow tank at pump-bowl levels below that at which
accumulation had heen observed before.

There was some encouragement from the treated-water system. Partial
deaeration of water going to the thermal~shield slides significantly re-
duced the holdup of radiclytic gas, leading to plans for more effective
deaeration. Decomposition of the nitrite corrosion inhibitor leveled off
at an acceptable level.

Power operation was interrupted for 14 hr after a shaft coupling on
one of the main blowers falled. Another interruption, this one for four
days, resulted when an electrical short in a component-cooling pump
caused the fuel system to drain. Power operation ended on July 17, when
the hub on one of the maln blowers broke up, reducing the air flow and
causing pieces of hub and blades to fly into the radiazator. The coolant
system was immediately drained and cocled down to inspect the radiator.
After some experiments at low power, the fuel was also drained to begin
the planned program of malntenance. Integrated power up to the shutdown
was 7823 Mwhr. Some of the operating data are summarized in Table 1.1.

The Tirst step was to flush the fuel system with Clush salt in
preparation for the specimen removal. During an experiment to verify the
fuel pump overflow point, flush salt got into scome of the gas lines on
the pump bowl and froze there, adding another job to the shutdown.

The agsembly of graphite and Hastelloy N specimens that had been in
the core since September 1965 was removed for examination. These showed
practically no corrosion or deterioration. The samples were sub/Jected
to & program of analysis and testing with emphasis on fission product
deposition.

Two control-rod drives were removed and faulty position indicators
repaired. Testing showed that the water leak in the reactor cell was
from one of the space coolers, and it was removed and repaired. Special
teols and heaters were devised, and the frozen salt was thawed from the
plugged gas lines on the pump. All the work in the reactor cell was done
through the maintenance shield because of radiation, which ranged from 1
to &0 r/hr over openings in the top shield. Inspection of the radiator
showed only inconseguential damage due to the blower failure, but con-
siderable repair of the enclosure and the door seals was necessary be-
cause of heat-induced deformation and fallures. Replacement rotary ele-
ments with stronger hubs and lighter blading were procured for the main
blowers. -
10

Table 1.1. bBumnary oi Some MSRE Operating Data

 

February 28, 1966  August 31, 1966

 

Time critical, hr 292 1775
Integrated power, Mwhr 35 7823
Fuel loop time ahove 900°F, hr
Circulating helium 1916 2780
Circulating salt 2836 4691
Coolant loop time above 900°F, hr
Circulating helium 1933 2020
Circulating salt 1809 5360
Heating cycles
Fuel system 5 5
Coolant system 5 5

i1l cycles

Fuel system 13 20
Coolant system 7

Power cycles

Fuel system 6 20
Coolant system 3 17

 

1.2 Reactivity Balance

 

J. R. Engel

The reactivity balance, as calculated for the MSRE, is a summation
of all terms which affect the system reactivity. Tt is routinely com-
puted every 5 min while the reactor is critical for the purpose of re-
vealing any anomalous effects that may be developing. The value of the
reactivity balance in revealing unexpected effects is strongly dependent
on the accurate calculation of all the known terms. Very early in the
power operation it became apparent that the 13°Xe poisoning was much
iess than expected. Accordingly, this term was eliminated from the re-
activity balance until the xenon effect could be evaluated and an accu-
rate representation programmed into the computer. (The considerable
effort directed toward evaluating the xenon effect is described in Sect.
1.3.) 1In spite of this omission, considerable useful information was
derived from the reactivity balances.
11

Reactivity Balances at Power

 

Figure 1.1 shows the history of the partial reactivity balance (no
135%e correction) during the approach to full power from 1 Mw and steady
operation at high power. The power history for this report period is
included for reference.

Xenon Polsoning. The most prominent feature of the reactivity-
balance behavior ig the transients that accompany significant changes in
power. The directions of the transients and their time dependence cor-
respond to the expected behavior of 13536 poisoning, but they are much
smaller in magnitude. We used these transients for a tentative evalu-
ation of the xenon polsoning effect.

 

Power Coefficient of Reactivity. A power coefficient of reactivity
is used Lo account for the fact that the average graphite temperature is
higher than that of the fuel at high power. Initially we calculated an
effective temperature difference of 44°F at 10 Mw. This leads to a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

""" - yrtTTTE LT i i
O|'* .‘,,,l:w I e, H?‘,fi% ........ Longdee oo ’ e _1 4_* ST b L. J ‘..‘
s b "»-A T""_‘L;" e, ,“ ’ le, | | f.v‘...“ Lo &L l 4 b ‘ F
St LT L T LR s e
0L Jr L NETREACTIVITY | e e || PR a#,,,w ,,,,, |
oal L 11 1] Lmll L mmlmLJ“ _____ tmk“ | |
72 N
- i
| i J
1 | D
9 i1 13 15 17
MAY, 1966
x 4* v el g T T b
S i [ e b w — ! Pl L ¥ o3
0P et N_ETREACTW'T%__I___ T ]
¥ _oal = 1 o { ,,,,, ‘[J AAAAA ‘ __J, ,,‘,’"%.af L 1 ,
: L ol * E L

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

19 21 23 25 27
JUNE, 1966

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

JULY, 1966

Fig. 1.1l. Medified Reactivity Balance During Power Operation.
12

predicted power coefficient of —0.006(% 8k/k)/Mw when the reactor outlet
temperature is held constant. Observations of control-rod posltions
immediately before and after large power changes indicated a smaller
power coefficient of reactivity. Careful measurements in times that were
short relative to the larger reactivity effects gave a value of —0.001 *
0.001(% 8%k/k)/Mw. This value corresponds to an effective temperature
difference between the graphite and Tuel of 30°F at 10 Mw. Although the
contribution of the power coefficient to the net reactivity is small, all
the reactivity balances were corrected to the observed value of this
parameter.

Reactivity Balances at low Power

 

A somewhat less-obvious feabture of the general reactivity behavior
was a gradual drift in the positive direction of balances taken at very
low power when xenon poisoning was negligible. By the time the reactor
was shut down in July, this drift had raised the apparent net reactivity
to +0.1% Sk/k. However, the reactivity balances do not contain a term
to correct for circulating bubbles, and there is some evidence (see
Sect. 1.4) that the bubble fraction increased from essentially zero to
about 1% during this period. Bince circulating voids have a negative
effect on the reactivity, proper compensation for the bubbles would re-
veal an even larger positive drift in the net reactivity. If the bubble
fraction did increase by l%, the net positive shift in reactivity during
the period in question was about 0.3% 8k/k.

Several of the major terms in the reactivity balance are accurately
known, as demonstrated by the good balances obtained in the initial periocd
of low-power operation.l These terms include the temperature coefficient
of reactivity, compensation for the operating 235y concentration, and
control-rod poisoning. The terms which were introduced by extended oper-
ation of the reactor at high power are polsoning by low-cross-section
fission products, 235y burnup, and samarium polisconing. (Poisoning by
T35%e has not yet been included because of the discrepancy between pre-
dicted and cbserved behavior.) One possible explanation for the upward
drift in reactivity is overcompensation for one or more of these negative
terms.

Low~-Cross-Section Fission Products. Since the yields and effective
cross sections for these Tission products are reasonably well known, this
poisoning term can be accurately calculated. Furthermore, at the burnup
achieved so far, this term accounts for only —0.004% 8k/k.

 

235y Burnup. The cross sections for 2350 were adeguately treated in
the MSRE calculations, as evidenced by the prediction of the initial
critical concentration and the concentration coefficlent of reactivity.
Therefore, compensation for the burnup of 235U, which amounted to —0.128%
8k/k when the reactor was last shut down, is probably not a major source
of error.

Samarium Poisoning. We estimate that the equilibrium samarium
poisoning in the MSRE will be about 1% &k/k. Both **%°gym and *5'Sm con-
tribute to this effect, and, since different time constaals are associ-~
ated with each isotope, the two components are calculated separately and

 
13

then combined to give the total reactivity effect. The caleulated sa-
marium poisoning when the fuel salt was drained on July 24 was —0.364%
5k/k. |

Causes of Anomaly. The apparent reactivity anomaly at low power
must result from an unknown positive reactivity effect, overcompensation
for negative effects, or a combination of the two. The magnitude of this
anomaly appears to be at least 0.1% 8k/k, but it may be as large as 0.3%
8k/k. The uncertainty is related to the uncertainty in the circulating
void fraction during operation.

 

To date there has been no evidence, either chemical or physical,
that the items which make up the negative reactivity terms are behaving
differently than was assumed in the calculations. However, additicnal
detalled analyses of fulure operations will e required to verify the
predictions of samarium poisoning and burnup. There has also been no
outside evidence of any unforeseen phenomenon that could have a positive
reactivity effect of the magnitude observed.

1.3 Xenon Poisoning

 

B. E. Prince R. J. Kedl J. R. Engel

Before the MSRE was operated at power, estimates were made of the
13%%e poisoning that would be encountered.” These estimates were based
on detalled models postulated for the xenon behavior in the salt, helium,
and graphite in the reactor. Values for important parameters were ob-
tained wherever possible from various development tests and, in the case
of the circulating bubble fraction, from experiments in the MSRE. (These
early tests indicated practically no bubbles.)? On this basis the |
steady~-state reactivity effect due to 135%e was estimated to be —1.08%
5k/k (poison fraction = 1.44%) at 7.5 Mw. The detailed equations which
describe xXenon behavior were also used to make predictions about the time
dependence of this phenomerlon.4

When the reactor was operated at power, the reactivity transients
that were attributed to 1355q were found to have time constants that were
close to the predlicted values, bul the magnitudes of the transients were
only about 0.2 to 0.4 of the predicted values. A possible explanation
for the low xenon polsoning appeared when there began to be substantial,
independent evidence that there were now circulating helium bubbles in
the Tuel loop (see Sect. 1.4). As a result of these observations, ad-
ditional studies were performed in an effort to obtain bether correlations
between our predictions and experience.

Two geparate studies were performed using the same basic set of
equaticons but with somewhat different objectives. The first of these
was an extensive parasmeter study of the steady-state 135%e polsoning.
The obJjectives were (1) to see if the MSRE experience could be explained
with reasonable values of the parameters and (2) to provide information
14

about the effects of various parameters that would be required for the
extension of this treatment to other molten-salt-reactor concepts. The
second study was aimed primarily at the time dependence of the Xenon
poisoning with fewer variations in the physical parameters. The objec-
tive here, in addition to providing a better insight into the behavior
of xenon in the MSRE, was the development of an adequate computational
model for the inclusion of xenon in the on-line calculation of the re-
activity balance.

Predicted Steady-State 13556 Poisoning

 

The steady-state 13%%e model described previously2 was modified to in-
clude the effects of bubbles of hellum circulating in the salt, and calcu-
lations were made. The results are shown in Figs. 1.2 and 1.3. Signifi-
cant parameters which were constant for t{hese plots are as follows:

Reactor power level = 7.5 Mw

Salt stripping efficiency = 12%

Mass transfer coefficient to bulk graphite = 0.060 ft/hr

Mass transfer coefficient to center-line graphite = 0.38 ft/hr

Diffusion coefficient of xenon in graphite = 1 X 104 ft2/hr

some of these parameters have been changed scomewhat from those reported
in the last semiannual report, but this is an updating and does not
change any of the conclusions. While the calculations included transfer
of '?°Xe from the salt to the bubbles, they omitted any transfer of
bubbles between the salt and the surface of the graphite. There are
arguments to support this omission, but the effect might later prove to
be significant.

Figure 1.2 shows the reactivity as a function of circulating void
fraction. The band shows the variation when the bubble diameter is fixed
at 0.010 in. and the mass transfer coefficlent to the bubble covers a
range of 1 to 4 ft/hr. It also shows the variatlon when the mass transfer
coefficient is fixed at 2 ft/hr and the bubble diameter covers a range of
0.005 to 0.020 in. For this figure the bubble stripping efficiene¥ is
fixed at 10%. The bubble stripping efficiency is the fraction of '?°Xe-
enriched bubbles that burst in going through the pump bowl and are re-
placed with pure helium bubbles. The actual value of the bubble stripping
efficiency is completely unknown, and 10% was picked only because this 1is
approximately the value derived from some development tests for Lhe ef-
ficlency with which dissolved gas would be stripped from the salt.

Flgure 1.2b shows the variation of reactivity with void fraction at
various values of bubble stripping efficiency. For this plot the bubble
diameter is fixed at 0.010 in. and the bubble mass transfer coefficient
is fixed at 2.0 ft/hr. The conclusion that can be drawn from these
figures is that circulating bubbles certainly can, and probably do, ac-
count Tor the discrepancy between measured and predicted values of 135%e
15

ORNL-DWG 66-14436

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-15 ]
FIXED PARAMETER: FIXED. PARAMETERS:
BUBBLE STRIPPING EFFICIENCY = 10% BUBBLE DIAME TER =0.010in.
l BUBBLE MASS TRANSFER COEFFICIENT
3 10 A =20 ft /hr o ]
- |
E ———SEE TEXT FOR DESCRIPTION
5 OF BANDWIDTH
g BUBBLE STRIPPING
x ~05 — e
N EFFICIENCY (%)
T
:22222222==u- —
) (@) | (&) _.28
¢ 0.5 1.0 15 0 05 1.0 15

CIRCULATING VOQID FRACTION (%)

Fig. 1.2. 13°¥e Poisoning as a Function of Cireulating Void Fraction
and Other Parameters As Indlcated

ORNL *PWG 66-41437

 

 

 

 

 

 

 

 

 

 

 

D) g
~otl—

;6 _

> BUBBLE DIAMETER = 0.010 in. ~ T

g BUBBLE MASS TRANSFER COEFFICIENT = 2.0 it /e

& BUBBLE STRIPPING EFFICIENCY = 0% —

&

Ll

& Q0

- 0.001

 

 

 

 

04
CIRCULATING VOID FRACTION {vol %)

Fig. 1.3. Contribution of Systems to Total 135%e Poisoning.

poisoning. To prove this conclusively, one would have to have accurate
knowledge of the circulating void fraction, bubble stripping efficiency,
and other factors which are unavailable at the present tinme.

Figure 1.3 shows the contribution of '3°Xe in each phase (salt,
graphite, and bubbles) to the tobal computed *??Xe reactivity effect.
16

The wvarious parameters are fixed as indicated. This plot shows how circu-
lating bubbles work to decrease the loss of reactivity to T35%e. As the
void fraction is increased, most of the dissolved xenon migrates to the
bubbles, dropping the dissclved xenon concentration greatly. The con-
centration potential necessary for 135%e to migrate to the graphite is
reduced accordingly. DNow, since 13%%e in the graphite was the largest
contribution to reactivity in the "no circulating bubble" case, the over-
all loss in reactivity drops also.

Analysis of Transient 13550 Poisoning

The equatlions given in ref. 4 were modified to include the possible
presence of minute helium gas bubbles circulating with the liguid salt.
The equations describing the transient concentrations, modified to in-
clude mass transfer of xenon to the gas bubbles, are given below. To
simplify the description, we will assume that the neutron flux is flat
throughout the core. The method described in ref. 4 for correcting for
the spatial dependence of the 135%e poisoning within the graphite-
moderated region is directly applicable to the modified equations given
below. As in ref. 4, we have used a one-dimensional model for the fuel
channels in order to simplify the calculations. The differential equa-
tions governing the concentrations of *2°Xe in the liquid salt, graphite,
and circulating gas volume are as follows:

dNi Vc 2
® T TN (1)

s v
Xe c £ c £
at 7XeP'V; } AINI - (:BXe * Rs " dXe® §7"> NXe

 

 

 

_ hvc <.N£ __RT 8 ] )
yOVL e HXE € Xe x=0

6
_figbww@wb), (2)

Xe er Xe

 

- - ¥ g
3t T e dx2 (%Xe + UXeQ) “xe 7 ()
The symbols used

4
.
,NI)Xe(t)
b oy
Nxe(t)

n%e(x, )

€,

71,%e

I,Xe
Xe

Xe

Jo

T

17

yi
g 1 g
NXe == 0( n e(x, t) dx . (5)

in these equations are defined as follows:

= average concentration of 35T and 1?%Xe in the tiquid

fuel salt, atoms per cublc centimeter of liquid;

average concentration of 13%%e in the clreculating gas

phase, atoms per cubic centimeter of gas;

local concentration of 135%e at position x within the
graphite stringer, measured from the graphite-salt
interface, atoms per cublc centimeter of graphite;

average concentration of 135%e over the graphite volume
assoclated with a single fuel channel, atoms per cubic
centimeter of graphite;

fission density in the core galt, fissions per second
per cublic centimeter of ligquid;

average thermal-neutron flux in reactor core, neutrons
cm™? sec“l; '

fission yields of 1351 ang lBsXe;

radicactive decay constants for *3°I and 135Xe, sec“l;

average absorptlion cross section of 135%e for thermal

neutrons in MSRE, cm?;

diffusivity of xXenon in graphite, square centimeters
of graphite per zecond;

ratio of volume of salt in core to effective mass
transfer surface area of graphite, cm;

ratio of volume of graphite to mass transfer surface
area of graphite, cm;

ratio of volume of s=alt Within reactor core to total
volume of circulating salt;

mass transfer coefficient for liguid film alt the
graphite-salt interface, cm/sec;

mass transfer coefficient for liquid £ilm at liquid-
gas bubble interface, cm/sec;

effective volume fraction of helium bubbles 1n circu--
lating fliuild;

bubble diameter, cm;

- universal gas constant, 82.07 em? atm mole ™ (°K)“l;

average temperature, °K;
18

HX = Henry's law coefficient, cm® atm mole™;
e

€ = graphite porosity, cubic centimeters of vold per cubic
centimeter of graphite;

kS = effective removal constant for stripping of 135%e from
liquid, sec™t;
Asb = effective removal constant for stiripping of 135%e from

gas bubbles, sec™t,

Equations (1) and (4) are identical in form with those given in vef. 4.
Fquation (2) has been modified by the addition of the last term on the
right-hand side of (2) to represent the net mass transfer of 135%s from
the liquid to the gas phase. Transfer of bubbles between the liquid and
the surface of the graphite was not included. Egquation (3) governs the
time dependence of the 135%e in this phase. The boundary conditions re-
gquired in the solution of Eq. (4) are:

3 g
n?(\,. — O , (6)
Ox
X=¥1
ONE
Xe _ £  RT g
~ % 5% ”h<NXe e fxe ) ' (7)
x=0 e x=0 ,

Apgroximate values of the basic parameters governing fthe mass trans-
fer of “?°Xe to the graphite were obtained from krypton injection ex-
periments.” No direct experimental data are available for the effective
mass transfer coefficients for the bubbles, Iy, or the effective bubble
sizes; however, representative values and probable ranges for these pa-
rameters could be estimated.® The dependence of the volume fraction of
circulating bubbles on the reactor operating conditions (fuel pump-bowl
level, temperature, pressure) is not yet well understood. In this study,
volume fractions between zerc and 1% were considered. Numerical values
for the mass transfer coefficients, geometric parameters, and graphite
characteristics used in most of the calculations are given in Table 1.2.
The graphite diffusivity and porosity are approximate values for the MSRE
graphite; however, it can be shown that the calculated transients are
relatively insensitive to these parameters.

The effective removal constants for external stripping, Ag and Ash s
were calculated in accordance with the formila

_ Qk
7\8”—71':’

where Q/VL is the ratio of the bypass flow rate through the xenocu-strip-
ping spray ring to the volume of the circulating fluid and E is the
19

Table 1.2. Numerical Values of Parameters Governing Mass
Transfer of T2°Xe in MSRE

 

Value Assumed in

 

Parameter This Study
Yo, cm 1.41
yi, cm 1.38
h, cm/sec 5.46 x 1074
Iy, , em/sec 0.017
d, cm 0.0254
Dye, cm?/sec 2.15 x 1074
Hye, €’ atm mole™ 0.33 x 10°
e, P 10

 

stripping efficiency. Based on the krypton injection experiments, the
value of E 1s expected to be about 107. For this study, this parameter
was varied in the range of 10 to 20%.

Relative values of the calculated time constants in Eg. (3) for mass
transfer, stripping, decay, and neutron absorption indicate that the 135%e
in the circulating gas phase should ve very nearly in equilibrium with the
135%e in the liquid, independently of changes in the reactor power level.
Hence, this approximation [setting the right-hand side of Eq. (3) equal to
zero for all times] was made to simplify the numerical calculations.

Some typical transient reactivity curves calculated with the pre-
ceding formulas are given in Figs. 1.4 through 1.6. Filgure 1.4 corre-
sponds to a step increase in power from zero to 7.2 Mw at time zero.

The separate curves indicate the effectiveness of increasing the volume
of circulating gas in reducing the net 135Xe‘poisoning. Fach of these
curves corresponds to a fixed stripping efficiency of 10% tor xenon both
in the liguid salt and the circulating bubbles. The transient at 7.2 Mw
wag chosen, since a sustained run was made at this power level during
run 7. In Fig. 1.4, we have alSo plotted the smoothed experimental data
obtained during this run. These data represent the magnitude of the
apparent 135%e poisoning, determined by subtracting all other known
power-dependent reactivity effects from the reactivity represented by
movenent of the control rod. We should emphasize that the experimental
evidence concerning the reproducibility of transients of this type is as
yvelt insufficient for conclusions to be drawn concerning the amount of
gas bubbles which may be in circulation. The data in Fig. 1.4 suggest
that from 0.5 to 1.0 vol % circulating gas can account qualitatively for
the obsgerved transient polsoning. However, the experimental curve seems
to exhibit a slower approach to equilibrium than is predicted by the
theoretical model.
ORNL-DOWG 66— 11428

 

|

14~ [ ! | | : : ] J —‘

 

| a,, ,VOLUME FRACTION

 

0B f e e — ‘ — —

0.6 | ] . = . . - S s

REACTIVITY MAGNITUDE (% Ai/k)

0al— . " 4o e e

   

0.2

 

 

 

O 4 8 12 16 20 24 28 32 36 40 44
TIME AFTER INCREASE IN POWER LEVEL (hr)

 

 

Fig. 1.4. Effect of Volume Fraction of Circulating Gas on Transient
135%e Reactivity. ©Step change in power from O to 7.2 Mw; stripping ef-
ficiency, 10%.

ORNL-DWG  6G-11438

 

 

 

 
 

 

  

 

 

 

 

 

 

 

040 e e e e s
< T | | ] 1 | | | ‘
st i :
< * EXPERIMENTAL DATA ,OBTAINED | ‘ é; (%o)
& 030F— DURING MSRE RUN NO.7. o —— e —— e - 0
! | ®

5 : . * * v 5
& 020 1+0— T T T e e g R T *—I— — 3 oot
g ‘ l o~ e 20
> | e L
T I0pb——m—— g B —_— _— —l— —_
= e
2 = - |
!
o O ® | i ]

O 4 8 12 16 20 24 28 32 36 40 44

TIME AFTER INCREASE IN POWER LEVEL ()

Fig. 1.5, Effect of Bubble Stripping Efficiency on Transient 335¥c
Reactivity. Step change in power from 0 to 7.2 Mw; volume fraction of
bubbles, 0.005.

The influence of the bubble stripping efficiency on the caleculated
transients is shown in Fig. 1.5. Here, the separate curves represent the
variation of this parameter within a range of 10 to 20% for a constant
bubtble volume fraction of 0.005. The experimental data of Fig. 1.4 are
also plotted in this figure.

kach of the transients shown in Figs. 1.4 and 1.5 may be separated
into components which correspond to the 135%e contained in the liquid,
the helium bubbles, and the graphite pores. The composition of a typical
transient, corresponding to a circulating gas volume Tfraction of 0.005
and a strilpping efficiency of 10%, is shown in Fig. 1.6. For the poison-
21

ORNL-DWG €6-11440

 

   
  

 

  

 

 

 

0.5 et e
T I I ]
,,,,,,,,,,,,, o TOTAL ¥9Ke POISONING |
W
| " [ GRAPHITE (FLAT FLUX)
02 o — e s ...,......_._......_....._........_..;;7;... bttt _ T P |
— T r
GRAPHITE (EFFECTIVE)
O [ flo e L
< A7
X .
<005
®
it
Q .....................
D
=
S 0.02
<{
=
2
S 0.0F o g
z ST SOOI SRS
2
<L
I"L" --
29
0.005 Ff
0.002 [ — f
0.00!

 

 

 

 

 

 

TIME AFTER INCREASE IN POWER LEVEL (hr)

Fig. 1.6. Components of Transient +37¥e Reactivity. Step change
in power from O to 7.2 Mw; volume fraction of bubbles, 0.005; stripping
efficiency, 10%.

ing in the graphite, calculations are given both for a flat-flux approxi-
mation {dashed curve) and the weighted poisoning for a flux distribution
approximating that in the actual core (s011d curve). Figure 1.6 illus-
trates that nearly all the 135%e is contained in the bubbles and the
graphite pores, the bubbles providing the additional "sink" which reduces
the effective mass transfer to the graphlte. It also illiustrates the
difference in the time constants for buildup of these components. The
approach to equilibrium of the 13%%e in the graphite is slower than that
in the liguid, owing to the effective time lag introduced by the mass
transfer between the liquid and graphite.

As more experimental evidence is accumulated concerning the tran-
sient behavior of the *3%Xe poisoning, attempts will be made to refine
the preceding model. This willl include attemplis to determine the de-
pendence of the volume of circulating gas on the reactor operating con-
ditions and the degree to which a fixed gas volume might be expected
during further operation of the reactor.
22

1.4 Circulating Bubbles

 

R. J. Kedl J. R. Ingel

Early in the development of the MSRE salt pumps, some evidence ap-
peared that the circulating fuel salt would contain a small but measurable
guantity of helium bubbles introduced by the xenon stripping device. This
was a significant finding because of the potential influence of these
bubbles on the reactor dynamics (through a pressure coefficieant of re-
activity) and on xenon poisoning. Theoretical analyses led to the con-
clusion that the dynamic effects would not be serious but that bubbles
would substantially reduce the xenon polsconing. Because of the im-
portance of the bubbles, experiments were planned to measure the circu-
lating void fraction.

The experiment designed to measure the void fraction consisted in
subjecting the Tuel system to a rapid pressure release. 1In these tests
the fuel system would first be pressurized to about 15 psig (normal over-
pressure is 5 psig) with helium and allowed %o establish a new steady
state. The excess overpressure would then be rapidly released by vent-
ing the gas to a previously vented, empty drala tank. Expansion of any
gas 1n the loop forces salt into the surge space in the pump bowl, where
it can be measured by the level instrument. In addition, the expulsion
of salt from the core reduces reactivity, which can be measured in the
critical reactor by observing control-rod motion. A third method of
evaluating the circulating voids, a salt densitometer, wss installed for
use in the initlial tests prior to operation of the reactor at significant
DOWer .

Three pressure-release tests were performed as part of the zero-power
experiments in July 1965.° Two of these tests, at normal system tempera-
ture and salt level in the pump bowl, indicated that there was essentlially
no gas in the core. However, the third test, which was conducted at an
gbnormally low pump-bowl level {obtained by lowering the operating teme
perature to 1050°F), showed a void fraction of 2 to 3%. It appeared from
this test that bubbles did nol begin to circulate in the loop until the
pump-vowl level was reduced to about 50% on the bubbler level elements.
Since only one such experiment was performed, the individual effects of
pump-bowl level and salt temperature were not separated.

The results of these zero-power tests led to the establishment of a
minimum pump-bowl operating level of 50% to prevent circulation of helium
bubbles. In additlion, the bubble term was removed from the eguations
used to predict xenon poisoning. When the reactor was operated at high
power, the observed xenon transients were much smaller than those pre-
dicted by the "no bubble" equations, implying the presence of bubbles.

In addition, a reactivity disturbance occurred on June 19, 1966, which
again emphasized the importance of bubbles. On that date (see Fig. 1.1)
the net reactivity began to decrease sharply, and it was noted that the
lower of two pump-bowl level indications had dropped below 50% because

of the continuous transfer of salt from the pump bowl to the overflow
tank. The reactivity recovered rapidly when the sall was returned to the

e
pump bowl from the overflow tank. Following this event, small increases
in reactivity were noted each time salt was returned from the overflow
tank even though the pump-bowl level was kept above 50%.

Several pressure-~release experiments were performed near the end of
run 7 (July 1966) to determine whether the circulating void fraction was
higher than that observed a year ago. Since, by this time, salt was con-
tinuously transferring to the overflow tank at the normal pump-bowl level,
experiments could be performed at different pump-bowl levels with the
same reactor outlet temperature. Experiments were performed at both high
and low power levels, and one was performed at an abnormally low temper-
ature.

Preliminary evaluations have been made of the vold fractions at each
condition, and the results are shown in Table 1.3. The results of two of
the earlier tests (performed July 2, 1965) are included for reference.
Although additional experiments are necesgsary Lo elucidate the situation,
some conclusions can be drawn. It appears that the circulating veld
fraction at the end of run 7 was substantially higher than during the
zero-power experiments. The results also show a substantial dependence

Table 1.3. Circulating Void TFraction Measurements in MSRE

 

 

 

Conditions Void Fraction (%)
Date
of Reactor Reactor Outlet Initial Pump- From From
Test Power Temperature Bowl lLevel Pump-Bowl Control-~Rod
(37 ) (°F) (%) Ievel Rise@ Motion
7/2/65 107 1197 59.6 0 0
7/2/65 1077 1050 50.0 2.69 2.05
7/8/66 7.3 1210 53.0 0.65 0.81
7/12/66 7.2 1210 61.2 0.24~0.54 0.34
7/21/66  0.01 1191 56.2 0.87-1.01 1.94
7/22/66  0.01 1190 51.3 0.50-0.72 1.24
7/23/66  0.01 1189 61.5 0.48-0.62 0.96
7/23/66  0.01 1124 52.8 2.68-2.91 2.39

 

aRange reflects current uncertainty in compensating for pressure
sensitivity of level elements.
24 :

on salt temperature, at least at pump-bowl levels above 50%. This de-
pendence may be related to the effect of salt viscosity on the bubble rise
velocity.

There are some inconsistencies in the measured void fractions. These
may be due in part to the uncertainties in compensating for side effects
during the pressure release, for example, pressure sensitivity of the
level-indicating devices and transient nuclear periods during control-rod
adjustment. There is also a basic difference between the vold fractions
measured by the two methods in Table 1.3. The pump~bowl level measure-
ments indicate the average void fraction in the entire primary loop, while
the control-rod measurements indicate only the average void fraction in
the core. Thus, 1f there were any tendency for volds to concentrate in a
particular location, either in or outside the core, differing results
could be expected. Furthermore, there may be some dependence on the im- -
mediate past history of the reactor. If, for example, the bubbles circu-
lating at low pressure are very stable and are inefficiently stripped, a
level change Jjust prior to a test could produce misleading results. De-
tailed evaluations of thege and other effects will be included in future
analyses. -

1.5 Salt Transoort

II. B. Piper

Gradual Transfer to Qverflow Tank

Farly operation of the reactor showed that by some unexplained
mechanism salt gradually accumulated in the fuel pump overflow tank even
when the salt level in the pump bowl was well below the overflow point. -
The transfer rate depended on salt level, and the transfer ceased when
the level was about 3 in. below the overflow point. This situation ex-
isted until about April 1966, when transfer began to be observed at lower
salt levels. The rate appeared to increase gradually as time went on
until it leveled off in June and July at 0.57 1b of salt per hour, in-
dependent of salt level as far down as 4.7 in. below the overflow point.
The change occurred at the time of the stepwise increase in power, but
no mechanism connecting the two has been identified. This transfer has
no 111l effect on operation of the reactor other than imposing the require-
ment thalt the overflow tank be emptied three times a week to keep the
levels in the desired operating range.

Overfill

At the conclusion of operation in July, the fuel loop was filled
wilth flush salt to rinse out residual pockets of fuel salt, thus re-
ducing the radiation levels for scheduled work in the reactor cell. We
decided to transfer flush salt into the overflow tank for two reasons:
25

to flush out the residual fuel and to check the indicated level at the
overflow point. Transfer began when the indicated level was 9.6 in.

(In January 1965 the indicated level at the overflow point had been 2.2
in.) Approximately 0.6 £t2 of salt had been transferred when suddenly
the fuel pump level rose sharply offscale. Rising level in the overfliow
tank triggered a drain, but not before salt had entered some of the lines
connected to the top of the pump bowl. What had happened was that the
salt level in the flush tank had been lowered too far, allowing the pres-
surizing gas to enter the fill line and pass up into the reactor vessel.
The gas expanded rapidly as it rose through the salt, causing salt to
flood the pump bowl. The pump-bowl bubbler reference line was plugged
with frozen salt, and enough salt was frozen in the sampler tube to
obstruct passage of the latch. A thermocouple indicated that some salt
also entered the off-gas line, but this line was not plugged. Salt also
froze in the annulus around the fuel-pump shaft, preventing its rotation.

This incident was caused by human error. The level to which the
overflow tank wag to be filled is within the range easily altainable with
fuel salt. But the volume of flush salt in the reactor is less, and, in
order to reach the specified level in the overflow tank, it was necessary
that the salt temperature be about 1200°F or above. This was overlooked
when the procedure was specified, and the loop was filled at 1140°F. This
temperature 1s within the range for a normal fill, and the szpeclal re-
quirement for this experiment was not recognized until affter the incldent
had occurred.

Immediately after this mishap the overflow tank was emptied and the
loop was refilled. The pump tank was heated to 1200°F and the shaft was
freed. Subsequently electric heaters were applied to the outside of the
lines to melt out the salt in the bubbler reference line and the sampler
tube. The short flexible portion of the off-gas line was replaced be-
cause of uncertainty over the possible effects of salt in the convolutlons.

1.6 Power Measurements

 

H. B. Piper C. H. Gabbard

Heat Ralance

The nuclear power produced In the MSRE is determined by an overall
system heat balance. During early operation at low power (below ~1.0 Mw)
uncertainties in measuring small temperature differences yielded a large
percentage error in the computed heat balance., However, as the power '
was ralsed to intermediate levels, the uncertainties in the heat balance
caleculation became less and good confidence was established in the heat
balance power. Since the heat balance is the primary power standard,
all nuclear power instruments were calibrated to agree with it. As the
power was raised, the calibration of the nuclear instruments changed, and
a disagreement in power indication existed between the nuclear instru-
ments and the heat balance. (This disagreement will be'discussed later
in this section.)
The heat~balance calculation in the computer remains unchanged, but
the metheod for obtaining a zero-power base line nas been modified. Ini=-
tially, a constant "heat loss" term along with zero-power hemperature
bilases were used to establish the heat balance zero power. It was noted,
however, that with the reactor producing zero power, the computed heat
balance might vary from run to run by as much as ~ 200 kw. We believe
this occurs because many thermocouples are sensitive to heater settings
and the heaters may not be set exactly the same for each run. This
difference represents an error of only +2.7% when we run at 7.5 Mw.
Nevertheless, at the beginning of each run, several heat balances are
taken, the value of the discrepancy is determined, and this value is
entered in the heat-balance calculation as a new value of the "heat loss"
term, thus yielding a corrected power Tor the remainder of each rua.

Nuclear Instruments

 

The agreement between the heat-balance power and the power indicated
by the nuclear instruments was better than *= ~10% from 2 to about & Mw.
(Heat balances have a scatter of * ~200 kw, and this is independent of
power; below 2 Mw the percentage scatter is large.) Above this level the
two power indications began to diverge, with the nuclear instruments ine
dicating 15 to 20% high at a heat-balance power of 7.5 Mw. After in-
vestigating, with negative results, several mechanisms which might explain
the disagreement, it was postulated that the effect might have been

caused by a rise in water temperature in the nuclear instrument pene-
tration (NIP) as the power was increased. This was found to be the case.
When the ratio of heat-balance power to nuclear-instrument power was
plotted vs NIP temperature, a good correlation with a negative slope re-
sulted. This plot showed that an increase of 10°F in the NIP water tem-
rerature produced an increase of 450 kw, or about 6%, in the power indi-
cation from the nuclear instruments at a level power of 7.5 Mw. This
anomaly could be caused by Temperature sensitivity of the nuclear instru-
ments, but we believe the most likely explanation of the temperature
sensitivity 1s the change in the attenuation characteristics of water with
temperature. On June 27, a heat exchanger system was placed in operation
to cool the water in the NIP. As the NIP water temperature decreased, the
ratioc between heat-balance power and nuclear-instrument power approached
unity. The heat exchanger used was immediately available, but it does

not have the capacity to hold the NIP water temperature constant at all
power levels. From zero to full power, there is now a temperature rise

of ~18°F as compared to ~72°F before the heat exchanger was installed.
There i1s still a shift of about 5% between the two power measurements

from zero to full power.

Radiator Air Flow

 

Because of the lack of agreement between the heat-balance power and
that indicated by the nuclear instruments, we attempted to verify the
power by determining the amount of heat removed from the radiator by the
air. The inherent uncertainty in determining the reactor power by a heat
2'7

balance around the alr side of the radiator is great because of diffi~
culties in measurement of both the air flow and the temperatures, These
difficulties arise from the fact that the coolant stack is not long enough
(L/D = 7) for symmetrical flow to be established. Even so, caldculations
vere made using the best data available for this evaluation. The heat
removed by the alr was determined for reactor power levels of 4, 5.8, and
7 Mw, as measured by the overall system heat balance, and was found to be
1% low, 16% high, and 15% high respectively. This is considered to be =a
reasonable confirmation of the system heat-balance method.

1.7 Radiation Heating

 

C. H. Gabbard H. B. Piper

The effects of radiation heating on some of the M3BE components were
evaluated both during the approach to full power and during sustained
operation at high power. Of particular interest in this area are the
fuel-pump tank, the reactor vessel, and the thermal shield. (The effects
of fission product radiation heating in the off-gas piping are discussed
in connection with other observations of that system in Sect. 1.9, sub-
section entitled "Fuel Off-Gas System').

Fuel-Pump Tank

The upper portion of the fuel-pump tank is subject to substantial
heating from fission products in the gas space above the salt. Since
the useful life of the pump tank is limited by thermal-stress considera-
tions at the junction of the volute support cylinder with the spherical
top of the tank, close control was maintalned over the temperstures in
this region. Design studies’ had indicated that the maximum 1ifetime
would result 1f the Junction temperature were kept about 100°F pelow the
temperature on the tank surface & in. out from the Junction. Component
cooling air is provided to maintain this temperature distributicn. A
secondary conslderation in controlling the temperatures was a desire to
keep as much of the pump tank as possible above the liquidus temperature
of the salt.

In operating the reactor, it would be ideal if a fixed flow rate cof
alr over the pump tank would provide a satisfactory temperature distribu-
tion for all conditions. Farly design calculations indicated that this
condition could be met with an air flow of 200 cfm. However, temperature
measurements on the pump~test loop and during the initial heatup of the
MERE indicated that only about 50 c¢fm would be required and that the air
would have to be turned off when the pump tank was empty.

To minimize the temperature effects when the cooling air is turned
o, alr flow during power operation should be the minimum that gives the
desired temperatures. It was found that an air flow of 30 cfm provides
a satisfactory temperature distribution at all power levels up to 7.5 Mw.
Figure 1.7 shows the temperatures in the two regions of interegt as a
function of reactor power level with this air flow. The wvariations in
28

ORNL-DWG 66-11441

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1200 ! """ T T T T T T [ """ B
| | | .
| i
MO0 P - ! e
L : |
& | A
= | A
5 1000 T ‘ — A
0 1 4
S
= | 8 :
00 g— A Fo T e
': i s
. -
8 OO l [ l e ] e e
0 { 2 3 4 5 6 7 8

POWER (Mw)

Fig. 1.7. Variation of Fuel-Pump Tenk Temperatures with Reactor
Power. Cooling-air flow, 30 cfm. -

the individual temperatures are caused by variations in pump~vowl level,
salt temperature, and air flow. Both the individual temperatures and

the temperature difference increase linearly with power, as expected.
Although the temperature difference would exceed 100°F at powers much
above 7.5 Mw, the reactor could be operated at 10 Mw with the 30-cfm air
flow without significantly reducing the life of the pump tank. However,
variation of the air filow could also be employed to obtain closer control
of the temperatures.

Reactor Vessel

Since radiation-produced heat in the reactor-vessel walls must be
transferred to the salt for removal, the outer surfaces of the vessel
are hotter than the adjacent salt by an amount that is proportional to -
the reactor power. Any deposition of solids on the inner surfaces would
reduce the heat transfer (and increase the heat production) and lead to
still higher temperatures at the outer surfaces. There are two locatlions
in the reactor vessel where golids would tend to accumulate 1f they were
present in the circulating salt. These are the lower head and the lugs,

Just above the inlet volute, which support the core matrix. Deviations
from the normal surface temperatures might also indicate changes in the
core flow pattern.

Although there has been no evidence, whatever, of solids in the
molten salts, the differences between the reactor-vessel temperatures
and the salt inlet temperaturec have been carefully monitored. Since
the salt inlet temperature can be measured only at the inlet line and
there 1s some fission heat generation in the salt as it flows to the
areas in question, the observed Al''s do not represent the actual tem-
perature drops across the walls. However, the proportionality to power
should exist, and there should be no change with time.
29

Throughout the power operation of the reactor, the temperature dif-
ference between the vessel wall at the core-support lugs and the salt
inlet has been 2.0 * 0.1°F/Mw, and that between the lower head and the
salt inlet has been 1.5 #* O.2°F/MW. There has been no indication of
deviation from linesrity with power or of changes with time in either
temperature difference.

Thermal Shield

The function of the thermal shield around the reactor vessel is to
reduce the radiation level and radiation heating in the rest of the re-
actor cell. As a result, a substantial amount of fast-neutron and gamma
energy 1s deposited in this shield. The cooling system for the thermal
shield was designed to remove up to 600 kw to allow for uncertainties in
the rate of heat generation at power. Measurements of the heat load on
the thermal shield gave a zero-power value of 40 kw due to heat losses
from the reactor furnace and an additional 17 kw per megawatt of reactor
power due to radiation heating.

1.8 Reactor Dynamics

 

T We Kerlin 5. J. Ball

The inherent stability of the reactor system was investigated by
frequency response tests at eight power levels from zero to full power.
The testing methods, analysis procedures, and results of tests at powers
to 1 Mw are described in detall in the last progress report.8 Subse-
quently, tests were conducted at 2.5, 5.0, 6.7, and 7.5 My using pseu-
dorandon binary reactivity insertions, pulse reactivity insertions, and
step reactivity insertions. At each power level, essentially equivalent
results were obtained from the different tests.

A readlily obtainable result is the natural period of oscilliation of
the reactor power following a disturbance in reactivity. This information
can be obtained by simply observing the flux transient if the response 1is
lightly damped or by determining the frequency of the peak 1n the ampli-
tude of the frequency response. The experimental results agree very well
with previous theoretical p‘redictions,9 a5 shown in Fig. 1.8.

The measured freguency ressonse results for power levels of 2.5 Mw
and higher are shown in Figs. 1.9 to 1.12 along with the theoretical
predictions. The theoretical phase predictions agree with the experi-
mental results within the experimental scatter. All the tests at a given
power level give magnitude ratios having the same shape but differing in
their absolute values. TFurthermore, the portions of the frequency re-
sponses above 0.3 radian/sec should he the same for all power levels,
gince feedback effects are small in this frequency range and the zero-
power frequency response should dominate. The experimental results for
30

various power levels show the same shape in this frequency region but
different absclute amplitudes. Both of these inconsistencies indicate a
bilas problem which is apparently due to eguipment limitations.*™® These
equipment difficulties are not surprising, since we wanted to be able to
position the control rod with an accuracy better than 0.050 in. in the
dynamics tests; this far exceeded design specifications.

We previously -t;hought8 that a simple adjustment of several paramaters
in the theoretical model could force agreement between theory and experi-
ment. An adjustment made o minimize the error at 1.0 Mw also produced
agreement at 0.075 Mw and 0.465 Mw. This now appears to have been for-
tultous, since the same parameter adjustments increase the discrepancies
at power levels greater than 1.0 Mw.

The net conclusions are that the dynamic characteristics of the sys-
tem are guite satisfactory and in reascnable agreement with predictions.
The system 1s stable at all operating power levels, and the stabllity in-
creases with power level. The theoretical model appears to be satis-
factory in spite of unfortunate equipment limitations which prevent de-
talled parameter fitting.

CRNL-DWG 66-102385

 

 

 

 

 

 

 

 

 

 

 

 

PERIOD OF OSCILLATION {min)

 

 

 

 

 

 

 

100

POWER LEVEL (Mw)

Fig, 1.8. Comparison of Predicted and Measured Natural Periods of
Oscillation.

¥
31

ORNL-DWS 86-10042

 

 

 

 

 

 

 

Bty
B4k,

 

 

‘ .
POWER LEVEL - 2.5 Mw
o STEP TEST
——+t——l..l ® 127 BIT PRB5 - DIRECT ANALYSIS ||
N 127 BIT PRBS - INDIRECT AMALYSIS
4 511 BIT PRBS - DIRECT ANALYSIS
T ¥ 51 BIT PRBS ~ INDIRECT ANALYSIS

 

et el

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10? -
0.001 0002 0005 00! 002 005 ot _ 05 10
FREQUENCY (radicns/sec)
o0
J— POWER LEVEL - 2.5 Mw
[
80 // .............. — ]
O STEP TEST
70 NG - o] ® 127 BIT PRBS-DIRECT ANALYSIS
0 A 127 BIT PRBS - INDIRECT ANALYSIS
o e . L
A 511 BIT PRBS -DIRECT ANALYSIS
50 V 541 BIT PRBS - INDIRECT ANALYSIS
40 e R
30 e - L
G 20t Je e N —
Q
T
w 10 - | 11
)
T |l
I e i _ . e _ R Y \
I o
40 --%/THEORETICAL 4]
oo b N L
\L
IO [ Q O
-40 ?fi&
A
_so -
_60 ........................... — L | 4
(5 |
-70
0.0014 0.002 0.005 0.014 0.02 0.05 C.4
FREQUENCY {radians/sec)
&N /Mg
Fig. 1.9. Frequency Response: Magnitude Ratio of m for Ny =

Mv (a) and Phase of

i R s T

&N/ Ng

K/ Ko

for Ng = 2.5 Ma (b),

2,5
 

32

ORNL-DWG 66-10041A

 

T

| s
Sk/, %,

 

POWER LEVEL - 50 Mw
o STEP TEST
e 127 BIT PRBS -DIRECT ANALYSIS
L A 127 BIT PRBS - INDIRECT ANALYSIS
& 511 B!IT PRBS - DIRECT ANALYSIS

“e WYL 7T v 511 BIT PRBS - INDIRECT ANALYSIS

 

 

 

 

’ . %" ~. 1
. %j ELg h&‘v\{;%;( jt/meomflm_
A

 

 

 

 

 

 

 

005
FREQUENCY (radions/s

 

ec)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Q0 1
L ] ' 1 POWER LEVEL-50 Mw
80 - % .
J \G O STEP TEST
). o _ |
70 /—40 © o U—e ® 127 BIT PRBS-DIRECT ANALYSIS |
© L Q A 427 BIT PRBS-INDIRECT ANALYSIS
60 — L . v
o . §[ A4 511 BIT PRBS-DIRECT ANALYSIS
50 15 - ’ $ ””” ¥ 541 BIT PRBS-INDIRECT ANALYSIS
40 _ )7__J 1 AN -
50 | » &t :
gl m ——_— L. 44 et sam e
g 2° #fi}Jfi y
wo 40— — . AN . W
2 ] Z\
r ot———t 1t V4+vq§ —_ -
an v —_— -
_10 . ,'_—. T_i#_r__.__ — %
ol brit— it
_30 - - L e e — 1 -]t
_40 . fl‘ 1 4‘_._4 _ .-
-50 ft———— T T 4t +—
__60 }'—— ----- -t r———#—a[— --------- —1 -
(5) ’ l
-70
0.001 0.002 0.005  0.04 0.02 0.05 0 0.2 0.5

FREQUENCY (radi

ans/sec)

oN/Ng

Fig. 1.10. Frequency Response:; Magnitude Ratio of -S-W for Ng = 5
0

BN/ Ng
Mw (a) and Phase of TR, for Ng = 5 Mw (Db).
0
 

2
Sk/ke

PHASE {deg)

 

33

ORML-DWG &6-10032A

POWER LEVEL ~ 6.7 Mw
o STEP TEST NO. 1
& STEP TEST NO.2
v 51t BIT PRBS - INDIRECT ANALYSIS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

107
0002 0005 001 002 005 ol 02 05 10
FREQUENCY (rudians/sec)
80
M
v
70 '::[lmg & ™
/ Ad * N 'i POWER LEVEL-6.7 Mw
60 o"! é ® STEP TEST NO. 4
. v 3
ol * e g\ A STEP TEST NO.2
v ¥ 541 BIT PRBS — INDIRECT ANALYSIS
v 7
40 —@ v
,\
30 3
20 ‘%&
w
10 a}i THEORETICA L
v\ |
0 3
v N
-{0 “'1 v / ‘
4 v
_30 ‘—
v ave 4 A v{
¥ A “’
-40 &e. @
¥ a A |A &
A -
~50 A
60
(6) L
-70
0.004 0002 0.005 0.04 0.02 0.05 0.1 0.2 0.5 1.0
FREQUENCY (rodians/sec)
SN/Nq

Fig, 1.11.

I}

Frequency Response:

Megnitude Ratio of o for Mg
S/No BK/Ko

6.7 Mw (a) and Phase of SRS Tor Wg = 6.7 Mv (b).
0
 

 

34

ORNL-DWG ¢6-10040A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2
! | POWER LEVEL - 7.5 Mw
1 o STEP TEST
e 127 BIT PRBS - DIRECT ANALYSIS
o3l oo e ] 4& 127 BIT PRBS - INDIRECT ANALYSIS
. o —t— | 1 & 511 BIT PRBS - DIRECT ANALYSIS
[ e LT e @_gég{#fi?fi[gg}mss~mDmECTANAusm
| - G ,i?f‘ Ay _C_g./\ &
b VVITaSS
| 5 e b e el T T8
""" - v
Ml o ATl
Ak |
/% J -
2 | ,’,g/ ................. _L — _
4?/ | |
v
v A :
ol BT |
102 / I 1 =
0002 0.005 0.0 002 005 01 02 05 10
FREQUENCY (radians/sec)
80 .
POWER LEVEL -7.5 Mw
70 a”/f’ !
pz///,/’ d STEP TEST
60 L ol 127 BIT PRBS-DIRECT ANALYSIS |||
o © 127 BIT PRBS-INDIRECT ANALYSIS
0 -
> o 514 BIT PRBS-DIRECT ANALYSIS
40 | o 514 BIT PRBS-INDIRECT ANALYSIS |-I-
30 L 4 ..
20 e e
v &
10—yt ] A\ - e .
o g _——THEORETICAL -
tal O ,,,,,,, e <
g &
T | A -
a 40 A fl'“\‘ -------------------
20 @%3
_30 u _ e
_40 —
-50
_60 ........................
( 6}
-70 !
0.001 0.002 0.005 0.01 0.02 0.05 0.1
FREQUENCY (radians /sec)
BN /N
Fig. 1.12., Freguency Response: Maenitude Ratio of for N
BN/ N,
7.5 Mw {a) and Phase of for Ny = 7.5 Mw (b) -
. O - -
5K/ K,
35

1.9 Equipment Performance

 

Heat Transfer — C. H. Cabbard, H. B. Piper, and R. J. Kedl

As the reactor power was raised, the heat-transfer capability of the
alr-cooled radiator was found to be less than expected and, in fact, to
1limit the attainable heat removal to about 7.5 Mw. The overall heat-
transfer coefficient of the primary heat exchanger was also below the
predicted value, resulting in somewhat larger fuel-coolant temperature
differences than had been planned. fter the filrst indications of low
hea®t transfer, we reexamined the reactor data to determine heat-transfer
coefficients as accurately as possible and to see if the coefficients
varied with power level or operating time. Meanwhile, we reviewed the
original design work to see if there were errors in the caleculational
method or physical properties that could account for the discrepancy be-
tween the predicted and observed performance.

Primary Heat Exchanger. Because of the elevated temperatures and
relatively small temperature differences in the primary heat exchanger,
g proper accounting for thermocouple errors is essential to an accurate
calculation of heat-transfer coellicient. The performance was evaluated
by two procedures, which differed mainly in the method of handling thermo-
couple biases.

 

The essentisl feature of the first procedure, used routinely at the
MSRE, is a statistical fit of a theoretical relation to a large nmumber of
temperature measurements at different power levels. The formulation 1s
such that biases in thermocouples, if constant, do not significantly af-
fect the outcome and need not be evaluated. The effect of random error
in thermocouple output is minimized by using many sets of data. The
relation used to evaluate the overall heat transfer coeflficient, U, is

 

 

where

T = measured salt temperatures,

=i
il

mass flow rates,

Cp heat capacity;
36

subscripts
= Iuel,
¢ = coolant salt,
i = heat e¢xchanger inlet,
c = neat exchanger outlet.
During operation, the terms in the derivative on the left are computed

and logged by the on-line computer. These values can be retrieved and a
slope determined from the plot of one against the other.

The other procedure, used as a check on the results of the first,
ig In some respects more straightforward. A set of temperatures is meas-
ured, and a value of U is calculated from the conventional heat-transfer
cqualbion

 

where
U = overall heat-transfer coefficient,

Q@ = heat transferred, computed from a coolant-salt heat
balance,

A = 279 Tt? of total tube surface area, including the return
hends, based on the tube 0D,

ATm = log mean temperature difference.

Temperatures used in the calculation of Al and @ are obtained by adding
a bias correction to each thermocouple indication. DBilases are determined
by logging a complete set of thermocouple readings when the reactor is
operating at a steady, very 1ow power, so0 that the salts should be iso-
thermal throughout the fuel and coolant loops. The bilas for each thermo-
couple is then taken to be the differeace between the average of all the
thermocouples and that particular thermocouple.

Heat~transfer coefficients calculated by the two methods are shown
in Fig. 1.13. The continuous curve for the "derivative" method was ob-
tained by Titting data points taken over two periods of time: three
weeks in April and May and seven weeks in May and June. The two sets of
data gave identical results. The "conventional" points were obtained on
May 26. A dependence of heat-transfer coefficients on power level, as
exhibited in the results of both methods, is to be expected. As the power
is raised with the core outlet temperature held constant, the average tem-
peratures of the salts in the heat exchanger decrease. According to con-
ventional formulas for heat-transfer coefficients, The changes in salt
properties (primarily specific heat and viscosity) could account for
practically all the observed variation.
L
~J

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T
&
©
T:: S | i OTNL—DWG |66—11442
2 METHOD OF ANALYSIS (SEE TEXT):
C 800 s e e CONVENTIONAL _...
2 n .
. Lt DERIVATIVE,
Fig. 1.13. Observed Overall o
- - - b ! e ___________

Heat Transfer Coefficients in @ 700 T N T T
< \\

MSRE Primary Heat Exchanger. iy T — b
& BOD o e e | e
= e e
& ™
«
|<—1 sopl-—{+—— L i L
W
T
—
|
& 400 el
b9 1 2 3 a 5 6 7 8
o

REACTOR POWER (Mw)

Design calculations had predicted an overall heat-transfer coeffi-
cient at 10 Mw of 1100 Btu hr~t £t~2 (°F)~*. (This was for the straight
portion of the tubes, in a cleen condition.)} This is about a factor of
2 above the coefflcients observed at temperatures approximating those
used in the design. In an effort to resolve the discrepancy, the design
calculations were carefully reviewed to see 1f the proper procedures were
used and to check for errors in the calculations. The results of this
design review indicated that the heat exchanger had been properly designed
using conventional procedures and that the design should have been con-
servative by about 20%. References 11 and 12 indicate that conventional
heat-transfer relations are valid for molten salts, and therefore the
conventional design procedures should be applicable for the MoRE neat ex-
changer. The most likely explanation for the discrepancy appears Lo be
erronecus valueg of salt conductivity in the deslign computations. Recent
measurements of 2 salt similar in composition to the MSRE fuel salt have
shown the thermal conductivity to be about one-third of the value that
was believed to be correct at the time of the heat exchanger design (see
Chap. 7). No recent data are yeb available on the conductivity of the
coolant salt; but 1f a similar discrepancy exists, this would essentially
account for the reduced performance of the heat exchanger.

Radiator. The design and instrumentatlion of the air radiator pre-
clude a calculation of the overall heat-transfer coefficient at all but
two power levels. At most reduced power conditions the radiator doors
are partially closed or the bypass damper is open; thus the effective
tube area, the air flow through the radiator, and the downstream air tem-
perature are unknown. Therefore the reduced performance was not obvious
until the reactor was raised to full power.

The overall coefficient for the radiator was evaluated with both
main blowers running, the doors full open, and the bypass damper fully
closed (full~power conditions). The air flow measured at these conditions
was equal to the design flow rate of 200,000 cfm. Another evaluation was
made with similar radiator conditions except that only one main blower
was ruflnin%. The heat-transfer coefficients were 38.5 and 28.5 Btu hr-t
=2 (°F)™ and the power levels were 7.4 and 5.6 Mw for these two con-
ditions. The observed heat-transfer coefficients varied with the 0.575
L
Q

power of the air flow rate, agreecing with a theoretical exponent of 0.6.

. e . _ o oo o s
But the predicted coefficient at full-power conditions was 58 Btu hr
172 (°F)"t, far above the observed valuc.

The radiator design was reviewed Lo determine the reason for the un-
expectedly low performance. The design procedures followed conventlional
practice except in two important points: the evaluation of air properties
and the allowance for error in the predictions. In the calculation of
the air-slde coefficient the paysical properties of air were evaluated at
the tube surface temperature instead of the prescribed "film" temperature,
defined ags the mean of the surface temperature and the bulk alr tempera-
ture. Had the "film" temperature been used, the overall coefficient would
have been lower by 14%. Use of an erroncous value for the conductivity of *
the salt had little effect on the overall coefficient, since the heat-
transfer resistance on the inside of the tubes is less than 5% of the
total in any case. Thus, even when the air-side calculation followed the
prescribed Tormula, the observed overall coefficient was still only 66%
of the predicted wvalue. This is greater than the usual error in heat-
transfer predictions, but 1is probably not unreasonable considering the
configuration of the radiator and the very large temperature difference .
between the bulk of the air and the surface of the tubes. The radiator
was designed when the nominal design power for the reactor was 5 Mw. The
components were designed for operation at 10 Mw to ensure sufficient
capacity; and when it was later decided to call the MSRE a 10-Mw reactor,
this left them with little or no allowance for uncertalnty. The actuzal
tube area is only 4% more than the minimum required for 10-Mw operation
according to the original calculations.

Effects of lLow Heat Transier on Reactor Operation. In the main heat
exchanger the reduced heat transfer causes a larvger temperature difference
between the fuel and ccoolant salts Tor any given power level. The origi-
nal design temperatures for 10-Mw operation were 1225 and 1175°F for the
fuel salt entering and leaving the heat exchanger and 1025 and 1100°F for
the coolant salt. Actually, operation at a maximum fuel temperature of -
1225°F and a minimum coolant temperature of 1025°F results in a heat-
transfer ratc of 7.5 Mw. IT would be possible to operate the heat
exchanger at higher power levels by increasing the heat exchanger fuel
inlet temperature and/or reducing the coolant inlet temperature. How-
ever, for long-term operation the heat exchanger fuel inlet temperature
is limited to a maximum of 1250°F by thermal stress and stress rupture
considerations, and the coolant inlet temperature is limited to a minimun
temperature of 1000°F by the possibility of freezing the radiator. Op-
eration at these temperature conditions would approach 10 Mw. The heat-
transfer rate could also be improved by lncreasing the fuel and coolant-
salt flow rates. The Tlow rates could be increased either by installing
larger~diameter impellers or by increasing the pump speeds by using a
higher~frequency power supply.

 

The reduced performance of the coolant radiator, however, imposes a
definite limit on the heat~removal rate from the coolant system, and there
is no convenient method of increasing the heat removal. The mean tem-
perature difference between the coolant salt and the bulk-air temperature
is 920°F, and the coolant-zalt temperature would have to be increased
39

significantly before a gain in reactor power could be realized. The fuel-
inlet temperature to the heat exchanger would then be pushed to an un-
acceptable level.

In summary, the maximum reactor power level is limited by the air-
side heat transfer from the coolant radiator. There is also a less severe
restriction at the main heat exchanger which could pe clrecumvented by op-
erating the reactor system at off-design tenmperatures. The radiator heat
tranzfer can be improved only by relatively extensive modifications.

There will be a small increase in maximum power level during the winter
months because of the lower amplent air temperatures.

Main Blowers — C. H. Gabbard

Aerodynamic Performance. The aecrodynamic performance of the main
blowers was satisfactory. The vane angle on both blowers had veen set
at 20°, as originally specified by the manufacturer for the design con-
ditions. However, this vane setting did not fully load the drive motors;
80 when the maximum reactor power was Tound to be below the expected ‘
value, the vane angle was increased to 22.5° to increase the air flow and
heat removal. As expected, this setting loaded the drive motors to their
capacity and increased the air flow sbout 10%. The effect was to raise
maximum reactor power by slightly less than 1/2 Mw.

 

Motors. The blowers are driven by 250-hp wound-rotor induction
motors that have four stages of external rotor resistance. Auvtomatic
stepping switches shunt out these resistances in a timed sequence during
the startup of the blower to limit the starting current of the motors.
During the early stages of power operation, difficulty was periodically
experienced in the start of main blower No. 1. The motor current during
the starts was erratic, and sometimes the current was high enough to trip
the circuilt breaker. One of the cast-iron resistance grids in the ex-
ternal rotor resistance on MB-1 was found broken. A broken grid was also
found in the MB-3 starting resistors, but this grid was in & less critical
location in the starting sequence and its effect had not been noticed.
Both the grids were weld-repaired, and no further difficulties of this
type have been noticed. ’

Coupling Failure. The blowers are comnected to their respective
drive motors through short floating shafts with disk-type flexible
couplings on each end. On June 14, while the reactor was operating at
full power, the couplings on MB-1 failed. The shaft coupling at the
motor end apparently. failed first, and the coupling on the blower end
was then torn apart by the resulting shaft whip. The shaft destroyed
the coupling gaard and damaged the sheet-metal nose that covered the
front hearing of the blower. Debris from the two couplings was scaltered
throughout the fan room, and scratches on the fan vlades indicated that
some of the material had gone through the blowers into the radiator duct.

 

The coupling on the motor end of the shaft showed evidence of O~
erating in a partially failed condition for some time. The nuts holding
the flexible disks were severely worn, indicating that the motor torque
had been transmitted from the motor flange directly to the shaft flange
40

by the bolts rather than through the flexible disks. The initial failure
of the disks was attributed to fatigue where some incorrect, flat washers
had caused high stresses.

The couplings on MB-1 were rebuilt, and new Tlexible shims were in-
stalled in the MB-3 couplings. Operation was resumed after the radiator
duct was cleaned and inspection of the hol radiator tubes showed nc dam-
age Lo the tubes.

Blade and Hub Failure. Power operation was brought to a premature
end on July 17 by the catastrophic failure of the rotor hub and the
blading of MB~1l. The outer periphery of the rotor hub disintegrated, and
all the blading was destroyed. Most of the fragments were contalined in
the blower housing, but numerous fragments of the cast aluminum-alloy hub
and blades entered the radiator duct and some actually passed through the
radiator. The reactor was taken to very low power, and the coolant was
drained to determine the cause of the failure, to inspect the other blower,
and to examine the radiator for possible damage.

 

Inspection of the broken pieces of MB-1 revealed numercus "old"
cracks in the blades and in the hub as evidenced by darkened or dirty
areas on the fractured surfaces. One blade in particular had falled along
a large "old" crack. The hub had contained short, 1-1/2 to 2 in., cir-
cumferential cracks at the base of & of the 16 blade sockets, and the
failure generally followed those cracks. Figure 1.14 is a piece of the
hub showing the darkened areas at the base of the blade sockets.

Since the failed blower had contained these old cracks, the top
casing was removed from MB-3 to permit a careful inspection of its hub
and blading. The Iront hub casting contained a large, continuous crack
wnich extended about 35% around the circumference. There were also several
of the short cracks similar to those that had been in the MB-1 hub. The
blades were dye-penetrant inspected and found to be satisfactory. A re-
placement blower that had not been in service also contalned some minor
surface cracks in the hub.

There is no general agreement on the cause of the failure. The
original soundness of the hub castings is in question because of the
old appearance of portions of the fracture surfaces and the presence of

AHO O 84352

 

Fig. 1.14. Fragment of Failed Impeller Hub from Main Blower No. 1.
41

cracks in the other blower hubs. Shock forces produced during the
coupling failure and vibration due to slight imbalances are suspected
as contributing factors.

The three blowers are being rebuilt in the manufacturer's plant.
The hubs are being reinforced to relieve the bending moment at the base
of the blade sockets, and the centrifugal blade loading is being reduced
by substituting magnesium alloy vlades which are 35% lighter. The three
complete rotor assemblies will be gilven a 30% overspeed test with dye-
penetrant inspections before and after the tests. In the installation,
close tolerances will be imposed on alignment and vibration, and instru-
mentation will be provided to monitor vibration during operation.

Radiator Enclosure — T. L. Hudson, C. H. Gabbard, and M. Richardson

 

Operation of the reactor at power provided the ultimate tegt of the
radiator enclosure, involving for the first time the operation of the
radiator at the maximum capability with the doors fully open and with
forced air circulation. Operation at power levels up to 1.0 Mw had
veen achieved during the last report periocd. This operation had indi-
cated thalt the radiator door seals had become less effective during op-
eration. The loss in door-seal effectiveness continued during this re-
port perlod. '

Power Level at Various Radiator Conditions. During the power es-
calatlion phase of operation, the radiator conditions were adjusted to
obtain preselected power levels. Therefore, the complete heat-removal
characteristics of the radiator are not known because the reactor has
operated at steady-state conditions at relatively few power levels. How-
ever, the reactor power levels for some of the key settings are listed
in Table 1.4. All intermediate power levels can be obtained by the
proper adjustment of tne doors or the bypass damper.

Table 1.,4. Radiator Conditicns and Reactor Power

 

Radiator Conditions

 

 

Inlet Oufilet Bypass Main Blowers Power (1)
Door Door Damper Running
Closed Closed Open None 0-0.05%
Cpen 15 in. .Open Open 1 2.5
Cpen Cpen Open 1 4ol
Open Open Closed 1 5.8
Open Open Closed 2 7.2 O.Zb

 

iDepends on heat leakage and heater settings.
Fxact value depends on ambient alr temperature.
42 -

Salt Frozen in Tubes. One of the main considerations during the
design of the radiator enclosure was to avold the freezing of salt in
the radiator tubes. The expansion of salt during thawing was helieved
capable of rupturing a tube if the thawing first occurred in the center
section of a tube, so that the molten salt was confined between two fro-
zen plugs. A "load scram,"” which dropped the radistor doors and stopped
the main blowers when the radiator salt outlet temperature dropped be-
low 900°F, was provided to prevent salt from being frozen in the radia-
tor. However, salt was frozen in the radiator tubes on two occasions
and was successfully melted out with no apparent damage to the radia-
tor.

 

In both cases, the freezing occurred as a result of a rod and load
scram combined with a stoppage of the coolant pump. The first freezeup
occurred when a defective relay caused a 'rod scram' Trom a 5.0-Mw power
level. The radiator load was scrammed manually at 1000°F because of the
replidly decreasing temperatures, but the coolant pump was stopped by a -
low level in the pump bowl which resulted from the reduced temperature.
With circulation stopped, heat losses froze some salt in 30 or more of
the 120 tubes. The second freezeup occurred as a result of an electrical
failure which caused a stoppage of the coolant pump and a scram of the
rods and load. In both cases the coolant system was drained immediately,
but the drain-tank weight Indicated that some salt had remained in the
coolant systen.

Recent data indicate that the volume change during the thawing of
coolant salt is relatively small. This low volume change, plus the
fact that the normal temperature distribution during heating of the ra-
diator would cause progressive thawing of the tubes from the top to the
bottom, gave confidence that the radiator could be thawed without dam-
age. In the first freezing incident, the radiator was reheated slowly
and all. the remaining salt was recovered in the drain tank. A pressure
test was then conducted on the entire coolant system to check for rup- -
tured tubes. There was no indication of leakage. After the second )
freezing incident, increased heat leakage from the radiator enclosure
prevented a portion of the radiator from reaching the melting point, -
and some salt remained frozen in the {ubes. However, the lowest tem-
peratures were sufficlently near the melting point that the coolant
system was Tilled and circulated. The radiator temperatures indicated
that four or five of the tubes were blocked at first, but after several
minutes of salt circulation all the tubes thawed and reached the tempera-
ture of the flowing salt.

After the first freezing incident, the control system was revised
so that a rod scram would alsc cause a load scram. The low-bemperature
set point for a load scram was increased from 900°F, which is only 60°F
above the liguidus temperature, to 990°F. Other revisions were made s0
that the coolant pump would not be turned off unless absolubely neceg~
sary. A scram test was conducted from full power, and these revisions
were adequate to prevent freezing of the radiator as long as the coolant
pumnp remained running.
43

Damage and Deterioration. The radiator was subjected to possible
damage on two occasions from mechanical failures of main blower No. 1.
Pleces of stainless steel shim stock were hlown into the radiator duct
when the shaft coupling failed. The radiator tubes were visually in-
spected from the inlet side while salt was circulated at operating tem-
perature, and no evidence of damage from the coupling failure was found.
A loose sheet-metal cover from an electrical cable tray was found against
the tubes and was removed. Reactor operation was resumed after the ra-
diator duct was cleaned to remove the debris from the coupling.

The coolant system was drained and the radiator was cooled and
thoroughly Inspected after the shutdown that followed the hub failure
of main blower No. 1. ©Several radiator tubes were dented, possibly by
aluninun fragments from the blower, and a few small pieces of aluminum
were stuck to the tubes. These pleces were easily removed, and the
tubes were cleaned to remove any adhering aluminum. Visual inspection
and a helium leak test at 20 psig indicated that none of the tubes had
been seriously damaged by the blower fragments, and tests were run that
indicated that the exposure to aluminum would not endanger the life of
the radiator if the tubes were cleaned.

In addition to the damage that had been caused by the blower failure,
the radiator enclosure was 1n need of other repair. Heating the empty
radiator to an acceptable temperature distrivution priocr to a fill had
become increasingly difficult, and on the last £ill some of the tubes
could not be heated above the melting point of salt until the radiator
was filled and circulation started. The radiator doors had warped a
little, and the seal strips had been severely distorted. The geal strips
had also been torn loose in some places by the operation of the doors.
Numerous sheet-metdl screws had broken or had worked loose, allowing
sheet-metal cable-tray covers and sheet metal on the inside surfaces of
the enclosure to come loose. There were also numerous broken electrical
insulators.

Although there was excessive heat leakage from the radiastor enclo-
sure, this wasg mainly around the doors, and the overheating of electrical
leads and thermocouple leads that had occurred previously did not recur.

The repair of the radiator enclosure is in progress. Deslgn of
the seal strip has been improved to reduce the thermal distortion and
the overall warpage of the doors. This seal strip is segmented and
free to move to allow for differential thermal expansion, and the T-
bar which holds these gegments has been slotted to relieve thermal
stresses which were causging the door to warp. The loose sheet metal
and the loose ceramic heater elements are being held in place with
welded c¢lipse rather than the sheet-metal screws.

Tuel Off-Gas System — A, N. Smith

 

The difficulties encountered with the fuel off-gas system during
the initial operation of the reactor at power, the findings and con-
clusions relative to the causes of thes difficulties, and the system
modifications proposed to prevent the recurrence of the difficulties
Al

were all described ia the last progress report. The modificaitions,
which consisted essentially in using valves with larger flow areas and
installing a particle trap and an organic-vapor trap upstream of the
pressure-control valve (PCV~522), were all completed before resuming
power operation of the reactor in April 1966. Observable pressures
and temperatures in the off-gas system were carefully monitored during
all subsequent operations, partly to evaluate the effectiveness of the
changes but primarily to identify and correct any undesirable conditions
before they became unmanageable. Bome difficulties were encountered
with buildup of pressure drop, but none were serious enough to force a
shutdown. In addition, there was no observable loss in efficiency of
the primary function of the off-gas system, the retention of gaseous
fission products.

Detailed analysis of the performance of some components, particularly
the particle trap and the organic-vapor trap, was hampered somewhal by the
scarcity of accessible pressure-meagsuring devices in the system. This was
aggravated by the fact that the pressure drop across the main charcoal
beds occasionally exceeded the 3-psi range of the installed (and inacces-
sible during operation) measuring instrument and also by the failure of
this instrument ten days before the shutdown on July 17.

Line 522 Holdup Volume. When plugging occurred at several points
in the off-gas system shortly after the power was first raised, it was
suggested that the dependence on power might be related to radiation
heating of the off-gas holdup volume in the reactor cell. Off-gas
samples taken while the reactor was shut down in March with the reactor
cell at different temperatures showed more hydrocarbons at nigher tem-
peratures, lending support to the hypothesis that there was a reservoir
of hydrocarbens in the holdup volume. t was nol practicable to clean
the 68-ft-long, 4-in.-diam pipe; but the off-gas line was disconnected
at the fuel pump and in the vent house, and large quantities of helium
were blown through the line in the forward and reverse directions at
velocities up to 20 times normal. Very little visible material was
collected on filters at the ends of the line, but there were fission
products, and the amount doubled when the cell was heated from 120 to
175°% . Visual cobservation showed that the head end of the holdup volume
was clean except for a harely perceptible dustlike film. A thermocouple
was attached to the holdup pipe near the hesad end for monitoring tem-
peratures during power operation. When the power was subsedquently
raised, the temperature rose from cell air btemperature (about 130°F)
at zero power to about 235°F at 7.5 Mw. The rise in temperature after
a setup from zero power occurred with a time constant of about 30 min,
not inconsistent with bulldup of gaseous fisgion products in the line.

 

Line 522 Filter Assembly. The filter assembly that was installed
upstream of the fuel pressure control valve (PCV-522) consists of two
separate units in series. First is a filter to remove partliculates and
mist; next, a small charcoal bed Lo remove organic vapors. (See Compo-
nent Development, Chap. 2, for a detailed description.) As a result of
the experience during this period of operation, the filter part of the
assembly will be replaced with a new one of the same design. The old
unit will be examined in a hot cell to determine, if possible, the cause
for the variaticons in pressure drop that were cbserved.

 
45

In the new condibion, the pressure drops across the particle trap
and charcoal bed were <0.05 and 0.7 psi respectively. With these in~
stalled, the total pressure drop in the fuel off-gas sgystem was about
2.3 psi at the normal gas flow rate of 4.2 std liters/gin and PCV-522
wide open. Thus, when the reactor-system overpressure was controlled
at 5 peig, a 2.7-psi pressure drop occurred at the throttling valve.
Because there were no measurements of pressures at intermediate points
between the pump bowl and the upstream ends of the main charcoal beds,
the pressure drop across the filter assembly was known only to be below
3.4 psi. This condition prevailed until May 9, during operation of the
reactor at powers up to 5 Mw. There was no indicatlon that the pressure
drop across the filter assenmbly reached the detectable limit of 3.4 psi
during this time.

During most of the operations after May 9, PCV~522 was kept wide
open, allowing the fuel overpressure to follow the total pressure drop
through the other parts of the off-gas line. This mode of operation
permitted the monitoring of the pressure drop across the filter assembly.
For the first ten days the overall trend in the pressure drop was upward,
with increases after the power was raised and decreases after it was low-
ered. On May 19, with the power at 5 Mw, the pressure drop was up to 8
pei. The power was shut down to redistribute the electrical load, but
the pressure drop continued on up, even though the gas fiow was reduced.
On May 20, shielding was removed, and a pressure gage was temporarily
attached to a tap between the particle trap and the charcoal filter to
determine which was responsible for the high flow resistance. The drop
measured across bthe trap was 9.9 psi, while the drop across the charcoal
was only 0.5 psi. Thus the increase in resistance was due entirely to
the particle trap. System pressure was reduced by venbting from the drain
tanks, and the power was ralsed to determine maximum power. With the re-
actor operating at 7.5 Mw the filter pressure drop decreased to about 3
psi. Then vhen the power was lowered Lo zero, the pressure drop came
down over a periocd of a day to less than 1 psi. The pressure drop re-
mained low until July 12, when it began to increase gradually. The in-
crease continued after the shutdown, and the unit will be replaced.

The particle trap was immersed in a tank of water for ccoling by
natural convecticn. Thermocouples on the outside of the trap responded
to changes in power and gas flow, but the maximum temperature rise was
only about 25°F.

It was expected that accumulation of organic material in the char-
coal filter would result in progressiwve poisoning along the length of
the trap. Such poisoning would shift the location of maximum fissicon
product absorption and produce a shift in the temperature profile of
the trap. Figure 1.15a shows two plots of the charcoal tempersture
profile, one on May 10, when about 1200 Mwhr had been accumulsted, and
one on July 7, when about 7000 Mwhr had been accumulated. Except for
the upward shift due to the increased power level, the basic shape of
the profile 1s the same for both periods, indicating that significant
poisoning had not occurred during this interval. Figure 1.15b shows
the effect of variations in pump-bowl pressure on the trap tempera-~
ture profile. As would be predicted, the trap temperatures, partic-
ularly near the inlet, vary inversely with system pressure, because
 

 

 

 

 

CRNL-DWG  85-11443
P

 

 

 

 

 

 

FP DATE POWER PRESSURE
DATE POWER PRESSURE  A6--24-56 7.2 Mw 30 psig
0 5-10-66 5 Mw 4.6 psig & 7-7-66 7.2 Mw 5.0 psig
® 7766 7.2 Mw 5.0 psig 15-14-66 75 Mw 7.0 psig
140 ey e i e
| - W | |
| |
L ] L
130—— | e L B
o o \ : A& ‘
o |
o 120 N\ S N <L :
o e | 2 \
2 N | } ! ~ ‘
% HoL— \i_‘» - _{_ — . A, } _
= I\ © N ‘ 4
fifiookzl\I\- s «;TL - T
T-sfi, .» a 4\ '
j .
! \ ;
{b)
O 20 40 a0 80 100 120 0O 20 40

CISTANCE FROM

Fig., 1.15.

 

The pressure drop across the ch
tendency to increase during power op
ization experiments established that
lets to the bheds, probably where the
packing of steel wool above the char

Temperature Profiles

 

TRAP INLET (in)

Along Lire 522 Charcoal Trap.

as the pressure increases, the residence time of the gas between the
pump bowl and the trap increases, pe

short-lived activity to decay before

Main Charcoal Beds. The fuel o
tions JA and 1B for the entire perio
ception of three days at the very en
used.

rmitting a greater fraction of the
reaching the trap.

fi-gas was routed througt hed sec-~
d of power operation with the ex-
d, when sections 24 and 2B were

arcoal beds shoved a persistent
eration. Pressurization and equal-
the restrictions were at fthe in-
1/4-in. gas line opens into a
coal. It was found that the pres-

sure drop could be reduced, usuvally to near the normal 1.0 psi, by blow-

ing helium backward througn the bhed.,
sure drop through the two beds in pa
1B plugged more often, but sometimes
tions. The plugging of the beds ccc
during tne approach to full power.

This was done whenever the pres-
rallel approached 3 psi. Section
restrictions buill up in voth sec-
urred each time the power was raised
The plugging hecame less frequent

later, but at the end of power operation it was still necessary to back-

blow the beds about cnce a week.

Stack monitors indicated no bre
ten-year %°Kr at any time. That thi
unbalanced flow through the parallel

akthrough of activity other than
5 was so, despite the sometilmes
beds and the extra volumes of gas

introduced by backblowing, is an indication of a capacity that exceeds

expectations.

Line 524 Charcoal Bed.

 

In the fuel pump, part of the gas admitted

to the shaft annulus (about 100 cc/min) flows up along the shaft to pre-

vent oil fumes from diffusing down into the pump bowl,

This gas then
47

flows oub through the cateh basin and line 524. Originally the line-
524 flow joined the main off-gas stream just downstream of PCV~522,
whose pressure drop supplied the driving force for the gas flow through
524, Line 524 was rerouted to come in downstream of the main charcoal
bed for two reasons: o get more pressure drop when PCV-522 was open
and to eliminate a possible way that hydrocarbons could enter the char-
coal beds. Some manipulations of the system pressure caused actlvity
to get into line 524, resulting in closure of the radiation-block valves
at the end of the off-gas line. Therefore a small charcoal bed was added
in line 524 to hold up fission gases and prevent stack releases by this
route (see "Component Development,' Chap. 2).

Auxiliary Charcoal Bed. Prior to the startup on April 11, 4if-
ficulty in venting through the auxiliary charcoal bed was Tound to be
due to a check-valve poppet which had beccme lodged in the bed ianlet
line. The poppet had apparently vibrated loose from a check valve at
one of the drain-tank outlet lines and had been carried downstream dur-
ing subsequent venting operations. After removal of the poppet, venting
operations through the auxiliary bed were uneventful through most of
the power operations. However, an intermittent restriction was noted
early in July, and, after the shutdown on July 17, the restriction be-
came continuous and more severe. Furthermore, the situation was not
relieved by reversing the gas flow (backblowing). Pressure and Tlow
tests indicated that the plug was in the same area ag in the main char-
coal beds, namely, at the place where the gas header connects to the
charcoal bed.

 

An attempt will be made to remove the restriction by local external
heating. At the same time, we are designing and bullding a replacement
bed, protected by an inlet filter, that can be installed and connected
with a minimm of reactor shutdown time.

Treated Cooling-Water System — R. B. Lindauver

 

Chemical Treatment. During power operation of the reactor, radio-
lytic decomposition of treated cooling water in the thermal shield pro-
duced hydrogen peroxide in a concentration of 300 to 500 ppm. This
caused oxidation of the lithium nitrite corrosion inhibitor to 1lithium
nitrate., Additional nitrite was added pericdically until equilibrium
was reached st approximately equal concentrations (~700 ppm) of nitrite
and nitrate. This occurred at about 3000 Mvhr, and no additional ni-
trite addition was required after that time. The presence of the ni-
trate ion has no effect on the corrosion inhibition.

 

Radiolytic Gas Formation. Because of the inhibiting effect of the
corrosion inhibitor on the recombination of radiolytic gases, approxi-
nately 2 ft3/hr of hydrogen was formed in the thermal shield during op-
eration at full power. At steady state, about & 13 of radiolytic gas
accumulated in the thermal shield and thermal-shield slides. About one-
fourth of this accumulated gas was removed by partially deaerating, in
a small vented tank, the cooling waber supplied to the slides. A larger
degassing tank is being installed to deaerate the entire thermal shield
flow., This is expected to keep the radiolytic~gas concentrations below

 
48

the limits of solubility and thus prevent gas pockets from forming.

The gas that is stripped from the water will be diluted with nitrogen

to below the explosive limit and vented to the area stack. A continuous
hydrogen analyzer will be installed in the off-gas stream at the tank to
ensure adequate dilution.

In-Cell Leaks., Before full-power operaticn, while the reactor cell
was open, reactor cell cooler No. 2 was cobserved to have a small water
leak. The cooler was removed, repalred, and replaced. Some time after
the cell was resealed, a steady accumulation in the cell of 1~l/2 gpd
vas observed. This water was condenged from the circulating air stream
in the cool suction line of the component-cooling pumps. Afber reactor
shutdown in dJuly, the various water-contalining components in the cell
were lsolated and leak tested. ALl components were leak-Light except
reactor cell cooler No. 1. It was removed, decontaminated, and repaired.
As in RCC-2 earlier, the leak was 1n a brazed joint between a tube and
a header.

Treated-Water Cooler. About one week before the reactor was shut
down, the total leakage from the treated-water system suddenly increased
from ~4 gpd to 3 o 5 gph. Since there was no visible leak or increased
accurmlation in the cells, it was suspected that the water was leaking
torough the treated-waler cooler to the cooling-~tower-water systen.
After the treated-water system was shut down, the cooler was opened for
inspection and leak checking. A large amount of gray solids was found
in the shell (tower water) side arcund the tubes and in Llow-velocity
areas. All tube-to-tube sheet Joints were leak tested with alr and were
leak free, but a bydraulic test of the tubes indicated that 17 of the
360 tubes had leaks at the inside surface of the tube sheet. After these
tubes were plugged and the gasket on the floating head was replaced, the
heat exchanger was leak-{ight.

 

Component-Cooling System ~ P. H. Harley

The two component~cooling pumps (CCP), cne of which is a standby
unit, supply cell gas Go cool in-cell equipment, circulate gas past &
radiation monitor, and discharge gas to keep the cell at a negative
pressure. CCP-1 operated for 1584 hr and CCP-2 operated 1759 hr during
this report period.

During the previous report period the multiple matched belts on
each CCP were replaced by a single poly-V-belt. Blower operation, al-
though improved, was not completely satlisfactory because the output
was low at best, there were times when the operating pump failed com-
pletely to meet the demand, and the standby unit was sometimes slow in
building up pressure. Drive belt slippage caused the loss of a blower
on three occasions. On CCP-1 the belt had to be retightened after 450
hr of operabion and replaced after 870 hr because of damage due to slip-
ping.

The poor performance caused some inconvenience in reactor operation,
and the second loss of CCP-L contributed to a reactor drain. AL the
time of this fallure, the containment enclosure of CUP-2 was isclated
49

and undergoing a leak test, An attempt was made to get CCP-2 back in

operation, but the system began to drain before the blower could be
started.

At the shutdown in July the belt on CCP-1 had operated for 1084 hr
and the one on CCP-2 had operated for 2013 hr. Both belts showed some
wear and minor cracking, probably caused by overheating. Both had re-
laxed, so the tension was significantly less than the original adjust-
ment.

Even when the belts were not slipping, the output of the blowers
was low. The oubput at the speed at which the blowers were operating
was supposed to be 5920 scfm, bubt conservative flow calculations indi-
cated only 400 scfm. Various leaks might account for the difference.
A rupture disk is being installed at the discharge of the pressure-
relief valves, which are known to leak, The check valves will be in-
spected and repaired if necessary. (One of the check valves failed
completely in 1965.) Leakage through the valve which vents excess
gas to the reactor cell to control blower discharge pressure might
also account for part of the losses.

In August new, larger sheaves were installed on the drive motors
t0 raise the nominal oubput of each blower to 740 sefm. Part of the
increase in output will be discharged into the drain-tank cell to pro-
vide better mixing of air between that cell and the reactor cell.

The modifications should result in extended belt life. The larger
drive sheaves will lower the stress on the belts, and stopping the leak-
age through the pressure relief valves should lower the ambient temper-
ature in the enclosures. Another change being made specifically to
lower belt temperatures 1s the addition of a simple deflector to direct
cool incoming gas over the belts.

After the space cooler began to leak water into the reactor cell,
condensate accumulated in the 10-in, suction line to the CCP-1 dome at
a rate of 1 to 2 gpd. (The water was vaporizing in the reactor cell
and condensing in the suction line, the coolest surface exposed to the
cell atmosphere.) Simple drains were installed, but these could be used
only when the reactor was subcritical so that radiation in the coolant
drain cell permitted entry. Handling of the drained water was compli-
cated by the tritium (up to 915 pe/ml) produced from the %14 in the
treated-water corrosion inhibitor. During the shutdown in August, pip-
ing was installed to permit draining the domes during power operation
should leakage in the cell require it. Water condensed in the suction
lines now will drain to a tank in the sump room, from whence it can be
transferred to the liquld-waste tank.

Moisbure in the component-cooling domes may have contributed to
an electrical failure in wiring to the CCP-1 motor which happened on
June 27. A short in a seal on the end of a copper-sheathed magnesia-
insulated cable destroyed the seal and caused a large breaker to open,
and the fuel drained before services could be restored. Breakdown of

he epoxy potting in the seal is another possible explanation for the
failure. The CCP~1 wiring was replaced at that time, and during the
August shutdown CCP-2 was rewired to eliminate that epoxy-Tfilled seal.
50

oalt-Pump O0il Systems ~ J. L. Crowley and H, B. Piper

 

The performance of the salt-pump oil systems during the period was
acceptable although not trouble free. The heat transfer in the coolers
deteriorated because of cooling-water scale, which had to be removed,
and the tendency for oil to collect in the salt-pump motor housings
under certain conditions contlnued to be a nulsance. ILeakage by the
seals in the salt pumps continued to be very low, although there was
some increase in leakage by the lower shaft seal in the fuel pump.

Coclers. The oll is cooled by cooling-tower water passing through
coils in the two oil reservoirs. During March the temperature of the
coolant-pump oil supply gradually increascd from 1ts normal 134°F to
140°F ., After inspection showed there was scale in the coil on the
coolant-pump oil cooler, it was flushed with a 15 wt % solution of
acetic acid. Considerable material was removed, and the oll tempera-
ture was reduced 7°F. The fuel-pump 0ll cooler was given the same
treatment, and the oil temperature was reduced 3°F. During subsequent
operation the temperature of the fuel-pump oll gradually rose, and in
August the acetic acid treatment was again given both coolers.

Holdup in Motor Cavity. Holdup of oil in the salt-pump motor hous-
ings continued to occur under certain conditions of flow and pressure.
In July, during a pressure test of the coclant system, the pressure on
the coolant-pump oll tank was increased from 7 to 20 psig. This caused
58 liters of oil to accumulate in the wotor cavity, and it took five
days for all of it to draln back to the tank.

 

Shaft Seal Leakage., 011 leaking past the lower shaft seal in a salt
punp drains through a catch basin in the pump to an external catch tank
provided with a level indicator. Changes in the coolant-pump oil catch
tank indicated a steady accumulation of less than 2 cm3/day'throughout
the period. Until April the accumulation from the fuel pump was also
less than 2 em?/day. It increased then to about 6 cm?/day and again
in mid-June to about 20 em®/day. This indicates some deterioration of
the punp shaft seal, but the rate is still far below the 1000 cmg/day
considered tolerable,

 

Leakage past Static Seals. 0il can leak past a static seal around
the shield plug and into the salt in the pump bowl. This leakage cannot
be measured directly but, in principle, is detectable by the decrease
in oil ldnventory. 'This technique is limited by the accuracy of the
tank level measurements and variations in the amount of oil held up in
the salt-pump motor housings, which cannot be measured. The latter ef-
fect 1s canceled when data are averaged over a long period of time.

The expected accuracy of the level measurenents is equivalent to 1.2
liters. Data for the four-month pericd from April through July are
shown in Table 1.5. {(Inventories for the separate syshems were not
obtalned in March pecause of transfers while the ©0il coolers were being
flushed.) The apparent losses are apout one-fourth the probable error
in the level measurements.

 

Cil Replacement. No deteriocration of the oil was observed during
operation, but after the reactor was shut down in July, both systems vere
drained and refilled with fresh oil.

 
Table 1.5. 01l Systems Inventory Changes, April-July, 1966

 

Change (cm3)

 

 

Ltem Fuel Coolant

System System
Samples removed 8834 8834
Increase in cateh tank (shaft seal) 1005 181
Total accounted for 2839 9015
Decrease in reservoir 9536 9450
Apparent loss —303 335

 

Electrical System — T, L. Hudson and T, F., Mullinix

The MSEE plant includes an elaborate electrical subsystem to pro-
vide ac and dc¢ power to the various components. Failures within or
associated with the electrical system accounted for six unscheduled
interruptions in the power operation of the reactor during this report
period. These lnterruptions varied in length from a few minutes to
several days. In all cases the necessary repalrs or modifications
were made, and no damage to reactor eguipment was incurred.

AC Power Supply. Five of the six electrical failures that caused
reactor power interruptions were associated with the ac supply system:
three were caused by electrical storms, one by a wiring failure at a
component-cooling pump, and one by a simple overload of the main process-
power breaker,

 

On two occasions while the reactor was operating at power, momentary
electrical outages during storms caused control-rod scyrams by range-switch-
ing the nuclear power safely channels because of dips in the fuel-pump-
motor current. In both these cases the nuclear power was quickly restored
to the value that existed just prior to the outage. OSince the low range
of the safety channels is used only while filling the reactor (when the
fuel pump 1s off), raplid response of the range switching is not required.
Therefore, time-delay relays were incorporated in these circuits Lo pre-
vent their activation on momentary power dips.

The third storm~induced failure occurred when lightning struck a
power substation and parted one wire of the main MSRE feeder. In this
case the electrical lcad was automatically transferred to an alternate
feeder, and all essential equipment was restarted in time to prevent
draining either the fuel or coolant system. However, operation at high
nuclear power could not be resumed until service was restored on the
main feeder. The entire supply system has since been reviewed and im-
proved to make it less susceptible to damage by storms.

The short in the component-cooling pump cable seal is described in
Sect. 1.9, subsection entitled "Component Cooling System." When this
52 ;

snore occurred, there was a massive flow of current and the breaker
supplying the entire bus tripped hefore the individual breaker for the
motor.

During the initial full-power operation of the reactor the main
process-power hreaker (breaker R) was loaded to near its seb~point
value, When the load on the coolant-radiator main-blower mohors was
increased by increasing the piteh on the blower fan blades (see Sect.
1.9, subsection entitled "Main Blowers"), the increased load on breaker
R cauged 1t to trip on overcurrent. This condition was relieved by
transferring about 200 kw of auxiliary load from the main process trans-
fTormer to an existing auxiliary power transformer. The load transfer
decreased the current through breaker R from about 1700 amp/@hase to
about 1550.

After the shutdown in July all switchgear breakers were removed,
tested, and calibrated.

DC-AC Inverter. To improve the reliability and increase the ca-
pacity of the reliable ac power supply, motor generator 4, a 25-kva
single~phase rotary inverter, was replaced with a new 62-kva three-
phage static inverter. The total load on the reliable power supply,
including the on-line computer, which has some three-phase load, is
40 kva. The static inverter offers a number of advantages over the
old rotary inverter, including higher efficiency (increased 35%),
smaller size, no moving parts, quiet operation, and excellent voltage
and frequency regulation. Prior o the installation of this eguipment,
both the capacity and the voltage regulation of the reliable power supply
were inadegquate for the operation of the on-line compuber.

During the checkout and testing of the static inverter in April,
after the installation had been completed, trouble occasionally devel-
oped that blew the load fuses. Two timesg the manufacturer's field en-
gineer made exhaustive tests and could not find any trouble, but all -
sympboms indicated that the trouble was in the low-voltage loglc power
supply, which supplies 24-~v dc control power. On April 22 the inverter
failed again while the reactor was operating at low power, causing a
control-rod scram. The inverter load was automatically transferred to
the normal ac power supply. After this failure a new power-supply mod-
ule was installed, and no further trouble has oceurred.

The inverter output voltage regulation has been 208 =* 1/2 v, and
the output freguency better than 60 cps + 0.01%. On one occagion the
inverter was operated 2~l/4 hr from the 250-v battery system withoub
the dc gernerator operating. Although the input to the inverter varied
20 v, the output varied only 0.5 v.

Diegsel Gererators. Three diesel-power generators, each with a ca-
pacity of 300 kw, supply emergency ac power to mobors and heaters in the
MSRE. During this report period there were taree occasions on which
this emergency power was necded. The diesel generators were started and
operated without difficulty except in June, when the component-coolant-
pump breaker was closed after the fault in the junction box. This caused
an overload on DG-3, and it was manually shut down.

 
aa e 8 B e S e

53

In February a crack was found in the block of DG-3 at one of the
bolt holes. All attempts at repairs were less than completely success-
ful, but the unit was operable. Diesel generator 3 is being replaced
with a surplus diesel generator from ancother installation. The replace-
ment unit is similar to the original except that it is started by com-
pressed air instead of by a storage battery.

Control Rods and Drives — M, Richardson and R. H. Guymon

 

The control rods continued to operate reliably although there were
fallures in position instrumentation.

During this report period there were 16 rod scrams while the reactor
was critical. Bix were caused by power failures, six resulted from in-
strument malfunctions, two resulted from operator mistakes, and two were
deliberate experiments. In additicn, the rods were scrammed 18 times
in circuit tests and 42 times to measure drop times. In no case did
any rod ever fail Lo scram, nor was any drop time in excess of the
gspecified maximum.

Drop times for rods 1 and 2 were consistently below 800 msec, The
drop time on rod 3 increased from 900 to 960 mgec, still well below the
1.3-sec limit set by safety considerations.

Measurement of the fiducial zeros showed no appreciable change in
rod length nor any shift in position indication. The coarse position
synchro on rod drive 3 failed on July 2 because of an internal short,
and it wag necessary therealter to count turns of the fine synchro in
positioning this shim rod. The potentiometer on rod drive 1, supplying
a position signal Tto a fill-permit interlock, became inoperative because
of worn windings.

The defective potentiometer and synchro were replaced in August.

bamplers — R. B. Gallaher and R. H. Guymon

The sampler-enricher was used to obtain 35 fuel salt samples, 10 of
which were 50-g samples for oxide analysis. Most of these were taken
without incident. A brief account of difficulties follows; detalls are
covered ia Chap. 2.

On April 29 a latch with reduced diameter was installed to eliminate
binding which was encountered at the lower bend in the sample line. While
it was beilng tested, wires to the drive motor shorted out inside the san-
pler. The fuel was drained from the reactor, the sampler was removed
and repaired, and operation wag resumed in ten days.

AfGer the latch replacement the operation of the sampler went with-
out a hitch except for one time (July 6), when the capsule stuck for
some unexplained reason and no sample was obtained.,

Once it was necessary to retrieve an empby capsule which had been
dropped accidentally to the operational valve. A magnet on a cable was
used to retrieve the magnetic latch key with capsule attached.,
24 -

The interior of the transfer box larea 3A) gradually became contami—
ted, and on two occasions minor contamination of The outside resulte
as the tLransport container was belng removed througn the top. More strin-
P
« . | . i : ! . . o
ent procedures, including the establishment of a 'contamination" zone
£ P 2 =
at the sampler, were adopted to prevent resurrence.
2

B

a
B

On July 24 accidental overfilling of the pume with flush salt forced
salt into the sampler tube. The tube was Tound Lo be obstruetad aboutb
2 ft abvove the pump bowl, preventing sampling. IExternal heaters applied

remotely were used to melt out the salt.

The coolant sampler was used wibhout difficuliy to obtain 18 coclant-

w

salt samples. -

Containment - H. B. Piper -
Several aspects of contalnment are of interest in the overall op-

eration of the MSRE, In addition to the primary consideration, that

of maintaining a low-leakage envelope to prevent the release of exces-

give activity in the unlikely event of a reactor accident, 1t is also .

necessary to prevent activity releases and conbtamination during main-

tenance periods, when the normal secondary containment and even the

fuel loop itself may be open. The latter consideratlon is particu-

larly important with a Tluld-fuel reactor, where fission product ac-

tivity is distributed throughout the fuel loop. The high chemical

toxicity of MSRE salts (due to their high berylliwas and fluoride con-

tent) also requires careful measures to prevent the release of large

quantities of even nonradiocactive salis.

During reactor operation the secondary contaimment, or primary
accident contaimment, system consists of the reactor and drain-tank
cells, which are sealed and kept at ~2 psig. The reactor bullding
serves as a third barrier, since it 18 kept at slightly subatmospheric
pressure and the exhaust alr goes througn roughing and absolute £il- -
ters before belng released from the conbainment stack. When the main
containment barrier and the reachbor must be opened for in-cell main- -
tenance, primary activity containment is provided by diverting the
building ventilation air into the cell opening and out to the stack by
way of the absolute filters. The reactor bullding then becomes the
second contalnuent barrier.

Secondary Contaimment System. The initial testing of the secondary
contaimment system has been covered in detail.*® The leakage rate from
the secondary contalinment has been monitored throughoul the reporting
period. The leakage rate of the containment is determined by measuring
either the change in inventory (cell atmosPQerfi) or the change in pres-
sure as a function of time and then correclting this value for the known
and measured flows both lato and cut of the systewn.

 

A higher than acceptable leakage rate was measurced when tue con-
talnment system was reburned to sexrvice in March after a period of in-
cell maintenance. A check valve in the alr-steam line leading to the

reactor-cell sump Jeb was found Lo be leakling. There is a similar check
55

valve in the drain-tank-cell sump Jjet line, and it was suspected of leak-
ing also, so both lines were parted and capped to assure leak-tightness.
After this the contaimment was found to be acceptable, with a measured
in-leakage rate of 19 ftB/day‘at —2 psig. The highest acceptable leak-
age rate at this pressure would be ~7/5 ftB/day. Starting in late April
and continuing through May, it appeared that a large leak had developed
in the containment envelope, and a great deal of effort was made to find
and stop this leak. It was finally necessary, however, to pressurige

the cell to 20 psig and do extensive leak hunting.

At -2 psig it looked as if in-leakage had increased by about 100
ft3/day, and after pressurizing to 20 psig the data still indicated a
leak into the cell. The leak was found to be in one or more of the
eight pressurized thermoccouple headers. HRach header is constructed as
a box, with cne wall of the box Tacing the contaimment cells and the
other facing atmosphere. Thermocouple leads coming from the contain-
ment penetrate first the inner wall and then the outer wall of the box,
with the space between belng pressurized with nitrogen o a pressure
greater than that of the contaimment. Three of these headers were
found tc be definitely leaking. Assurance was cobtalned that the headers
were nobt leaking to atmosphere by soap-bubble checking all penetrations
in all headers while they were pressurized. No leaks were found. A
calibrated rotameter was placed in the nitrogen line which is used to
pregsurize the headers so that the leakage Trom the headers to the con-
tainment can be continuously monitored and corrected for in the leakage-
rate calculation. After this work was completed, the leakage into the
containment wag again measured at -2 psig and found to be ~25 42 /aay,
an acceptable value,

The leakage rate has been monitored daily since that time, and,
until the reactor containment was opened for maintenance, the contain-
ment remained acceptable. Although some leaks did occur during the re-
port period, none was large, and all were found and remedied.

Conbtainment for In-Cell Maintenance. Under normal conditions,
whether the reactor is operabting or not, the reactor building (high bay)
is kept at subatmospheric pressure, ~—0.2 in. H»20, Air from the reactor
building is discharged from the 100-£t containment stack after particu-
lates have been removed by an "sbsolute' filter. When in-cell main-
tenance is carried on (the reactor cell is open), a large cell exhaust
line to the filters is opened o assure that alr flows down into the
contaimment cell with a veloeity at least 100 fps through the openings,
thus preventing any uncontrolled release of airborne activity,

 

During the report period no detectable particulate activity was re-~
leased. There was a total of 97.3 mc of gaseous activity (iodine) re-
leased, with 74 me being released during the two weeks that the graphite
samples and flexibile Jjumper in the off-~gas line were being removed. The
total amount released in six months is 0,016 of the total permissible re-
lease from the five stacks in the ORNL area.

Filters and Stack Fans. In October 1965 the filter pit was over-
hauled, and new roughing and absclute filters were installed. ELfi-
ciencies measured by the standard ORNL dioctyl phthalate (DOP) test

 
56 N

were 99.994 to 99.999¢% for the three banks (99.95% is the minimum ac-
ceptable). The filters were again DOP-tested in March 1966, and mea-
sured efficiencies were 99.995 to 99.999%. GHvidently neither the ab-
solukbe filters nor the new silicone caulking had deteriorated. From
the time of installation through August, the pressure drop across the
absolute fillters remained unchanged, while drop across tne roughing
filters gradually increased to 5.5 in. H30, causing a decrease 1in stack
flow of 15%. The flow remained above the value assuwed in safety anal-
yses for stack dispersion, but tne roughing Tilters will be replaced
gfter in-cell mainbtenance 1s completed,

The hearings were replaced in the west stack Tan in March. Ilodi-
cations were that insufficient lubrication had contributed to the failure,
so the lubrication system was changed from grease to light oll. The same
lubrication revision is planned for the east fan. The fans are driven -
through V-belts by an external motor, which, as originally installed,
had adjustable sheaves. These sheaves apparently cause greater than -
normal belt wear, and even though no belt failure has occurred, rc-
placement with a properly sized solid sheave seemed advisable. This
was done on the west sback fan in July and is also planned for the -
east unit.

Beryllium. Air is continuously sampled at 15 locations through-
out the MSRE area for beryllium contamination. This is done by draw-
ing alr through paper filters upon which the beryllium would be collected
and then determining the amount of contaminant that is deposited. These
samples are collected and analyzed every working day. The lower limit
of detection is 0.05 pg per sample, and each daily (24 hr) sample rep-
resents the amount of beryllium in 14 m? of air. This represcnts a
lower limit of detectable concentrabion of 0.004 ug/m®. The maximum
permissible concentration of beryllium for continuous occupaacy of an
area for an 8~hr work day is 2 ug/m3. There has heen no detectable re-
lease of beryllium during this report period.

Alr drawn from the coolant-radiator stack is monitored continuously
vwhile the reactor is in operation., A beryllium detector which samples
and analyzes con line is used fTor this purpose, The limits of detect~
ability are the same as those descrived above. Again, there has been
no detectable release during this report period.

Snielding and Radiaticn — H., B. Piper

 

Complete radiation surveys were made of the reactor area as the
power was raised from 1 Mw to full power, and with the exception of
the areas discussed in the following paragraphs, The shielding was
found to be adequate.

When the reactor power was first raised o 1 Mw in April, the ra-
distion level in the North Electric Service Area (NESA) was found to
be high: 20 mr/hr on the baleony and 8000 mr/hr at the west wall. In-
vestigation showed that there was radioactive gas in the lines through
which gas is added to the drain tanks. Two check valves in each line
57

prevented the gas from getting beyond the secondary contalnment enclo-
sure, but the enclosure, of l/Q—in. steel, provided 1little gamma shield-
ing. The pressure in the fuel system at that time was controlled by

the newly installed pressure control valve with rather coarse trim, and
the pressure fluctuated around the control point (normally-5 peig) by
+~2% . These pressure fluctuations caused fission product gases to diffuse
mere rapidly into the drain tanks and back through the 1/4-in. lines
through the shield into the NESA., The radiation level was lessened by
installing a temporary means of supplying an intermittent purge to the
gas~addition lines to sweep the fission product gases back into the

drain tanks. At ~7.5 Mv on May 24, the radiation levels were 5 mr/hr

on the balcony and 2000 mr/hr at the west wall. During the June shutdown,
o permanent purge system was Ilnstalled to supply a continuous helium purge
of 70 cc/min to each of the three gas~-addition lines. This was proved suc~
cessful by subsequent full-power operation in which the general background
in the NEGSA was <1 mr/hr.

There are now three areas in which the dose rates at full power
are too high for continuous cccupancy. Limiting accesgs to these areas
ig not considered a hindrance to the orderly operation of the reactor
and so no further remedial action is planned.

Regetor Cell Top. Two very narrow beams, presumably coming from
cracks between shield blocks, reading 10 mr/hr garma, and 60 millirems/
hr fast neubtrons, were found on top of the reactor cell blocks; these
areas were properly marked.

Coolant Drain Tank Cell Access Ramp. FEven though shielding was
added inside the coclant drain cell door, the dose rales oubside the
door (700 mr/hr gamma, 125 millirems/hr fast neutrons, >75 millivems/
hr thermal neutrons) and halfway up the access ramp (35 mr /hr RIS, ,
2 millirems/hr fast neutrons, 25 millirems/hr thermal neutrons) are
high during power operation. This area 1s clearly marked with radia-
tion zone slgns at the entrance and halfway down the ramp.

Vent House. ©OStacked concrete vlock shielding has been added peri-
odically to keep dose rates low in this area. Even so, the background
radiation level is ~7.0 mr/hr at full power, and the area is & condi-
tional-access radiation zone.

1,10 Instrumentation and Controls
J. R, Tallackson R. L. Mocre

The MERE instrumentation and controls system continued to perform
well. There was the normally expected reduction in both malfunctions
and misoperation of instruments as instrument and operating persconnel
gained experience and developed routines. While there were many design
changes, most of these were improvements and additions to the system
rather than corrective measures Lo the instruments and controls. A dis-
appointingly large number of fauwlty commerclal relays and electrconic
switches were disclosed. These faults were in the areag of both relay
design and fabrication, and corrective steps have been taken.
58

Operating Experience — Process and Nuclear Instruments — C. E. Mathews,
. N. Fray, R. W. Tucker, and G. H. Burger

Control-System Relays. All 115 of the 48-v de-operated relays in
both the safety- and control-grade circuits will be replaced because
of heat damage to their Bakelite frames. This problem was discussed
with the manufacturer, who stated that overheating is a common problem
with all relays of this particular model if they are continuocusly ener-
gized for long periods (MERE relays have operated for two years). It
is a borderline conditiorn, which he says has been corrected by changing
the design to reduce by 20% the total power dissipated in the operating
coil and the serieg~connected dropping resistor. An order was placed
for 139 relays of the latest design.

 

Four new spare solenold coll assemblies were purchased for the spe-
cial weld-sealed electric solenoid valves., Approximately 40 valves have
been in service on the fuel- and coclant-salt circulating pumnp level
measuring systems and the fuel sampler-enricher for two years. The first
and only coil failure ocecurred recently.

Pressure Transducers. A differential pressure cell used Lo obtain
pressure drop in the heliwm flow through the charcoal bed shifted its
range selblting. The cell has been removed, but the cause of this range
shift has not yet been determined.

 

Thermocouples. Thermocouple performance has continued to be excel-
lent. Only one thermocouple failure occurred during this period. This
brings the total number of failures since the start of MSRE operation to
5 out of over 1000 couples in use.

Flectronic Switches. The Electra Systems switches®? used for alarm
and control of temperatures in the freeze-valve system performed without
malfunction during this period. This improvement in performance is at-~
tributed to modifications reported previously;l5 vo the esgtablisnment
and enforcement of more rigorous test and periodic checking procedures;
to the stabilization, by aging, of critical resistors in the switch mod-
ules; and to a better understanding, by operating personnel, of their
use in the system. A check showed that out of 109 switch set points,
83% had shifted less than 20°F over the six-month period. No swiltches
with double set points were found, and it is likely that this malfunction
will not reappear.

 

Sporadic malfunctions in control loops containing a particular model
current-actuated commercial electronic switch have been a source of an-
noyance, This faulty behavior was, apparently, associated most frequently
with ambient temperature changes obrought on by alir-conditioning failures
and with excesgsive vibration. Thorough inspection, made possible by a
system shutdown, revealed that out of 61 switches in service, 38 had one
or more faulty internal connections. Tnese faulty connectlions, origi-
nated during manufacture, were Imperfectly soldered joints or joints
which had never been scldered.

Nuclear Instrumentation. Water leskage via the cable into the
counter-preanplifier assembly caused several failures in the wide-range
59

counting chamnnels. The cause has been diagnosed as excessgive strain and
flexing of the cable, and a redesigned cable assembly will be installed.

Personnel Monitoring System. The reactor building radiation and
contamination warning system was again revised to correct some defi-
ciencies and improve its effectiveness. Four additional beacon lights
were installed. These were located in the coolant cell, blower house,
diesel house, and motor-generator room. The power circuits for the
lights were revised to improve reliability. The test procedures for the
system were revised to include monthly tests that actuate the entire
system, including the air horns, and to actuate and test the systen
trouble alarm features.

 

Data System — G. H. Burger and C. D. Martin

Several new analog input signals were added to the data system,
bringing the total close to the maximum capacity of 350. These were
added to obtain more information about the operaticn of the off-gas
system, to measure the reactor cell leak rate, and to monitor the
bearing temperatures of the main blowers and the water temperature in
the nuclear instrument penetration.

New programs were added to supply more reactor operating informa-
tion and to retrieve operating information previously stored on the
magnetic tapes. The two operating information programs calculate an
average of the fuel- and coolant-salt outlet temperatures and the re-
actor cell leak rate. The average outlet temperature caleulations
(OAFOT, OACOT) are used extensively by the operators as a guide for
the operation and control of the reactor. The data-retrieval prograns
were written to retrieve and process the stored information on line or
at the ORNL computer center. Both programs were used many times to
list and plot old data and were very useful in helping to determine
the time, cause, and effect of several reactor shuldowns. These data
also provided information which was used to redesign the off-gas sys-
tem and resulted in changing some of the reactor cperating procedures.
The retrieval program for the compuber center was written so that all
stored information including calculation results could be processed
and listed or plotted. The use of this program ig becoming a routine
operation, and data-retrieval requests are handled by a Jjob order card.
The on-line retrieval programs are more specialized and handle only
certain inputs or calculation results. These programs have to be pub
into the gystem as they are needed and reguire some operator time when
they are executed. The analysis group is now being trained to handle
rebrieval with these programs. A new general-purpose on-line program
similar to the compubter cenber program is planned.

Revision of operating programs ccentinued as new requirements were
determined during reactor operation. The heat-balance program wasg re-
vigsed and is now used to determine the true level of reactor power.
The reactivity-balance program was revised but is still not entirely
correct due to uncertainties in some of the parameters used in the
calculations,
60

In addition to the routine and periodic collection of operating

ta, the data system was used to instrument and control further re-
actor dynamics bests similar to those last reported,t® Tt was also
vsed Lo contyrol the temperature of a waterial surveillance test stand
by a program which simulated three three-mode analog controllers. The
control program compuses the surveillance specimen temperature control
set point from the reactor power, fuel outlet temperature, and an off-
set temperature To match the temperature profile of the material speci-
mens in the reactor core. An error signal is generated by using this
set polnt and the temperature of the test-stand specimens. An output
signal is then generated by the computer to control the test-stand
speclmen temperatures by changing the voltage to the heaters by means
of compressed-air-actuated autotransformers.

During the period covered by this report, the data system has be-
come a virtually indispensable tool for the operators and analysts.
The acceptance and confldence in this equipment which is now shared by
the operators and analysts are the result of the system's ability to
supply reliable and accurate information Lo assist the operators in

CRNL-DWG 66-9740

[[T"oN"TIME EB SCHEDULED SHUTDOWNS SCHEDULEGC SHUTDOWNS
FOR MAINTENANCE AND FOR MAINTENANCE OF

MCDIFICATIONS PERIPHERAL. EQUIPMENT

MSRE COMPUTER TIME AVAILABLE

e —m— p TTTTTTR TR L
R FF -a-‘_,,,[,:-* - T ‘k‘"“"’_h_‘fi } - 7&_LA »

 

 

100

 

N-57.31 %
INTEGRATED
»——’_ "ON" TiIME

 

95 i \,,‘E
80
85

7
-
=1

sof-
75
70l
ssr
80
55

N

 

501

 

TIME (%)

45
40F
35
301
25}~
201
151

 

 

10

il _J

 

 

 

 

 

 

 

 

2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7

 

 

 

7 e
o .. R ki D b - A
1 B 1522292} 5 12 19 26| 3 10 17 24 31 < 118 25|11 B 52229 6 32027

 

7 14 21284 11 1825 310147 24’ 1 8152229 5 121 26

 

 

 

 

 

MAR APRIL MAY JUNE JuLy AUG

 

 

oo : CA8Bh e e 1966 — —— - —— — —,-rfi%

 

WEEK BEGINNING

Fig. 1.16. MSRE Data System Service Record from Date of Acceptance,
October 1, 1965, to September 1, 19606.
61

controlling the reactor and to assist the analysts in evaluating the
operation. During this period the system continued to show a steady
improvement in overall reliability. Figure 1.16 shows the availability
of the system on a weekly basis. OSome of the downtimes in March and
April were caused by ac power difficulties as a result of installing
and getting the static inverter power supply operating. ©Since April,
after the inverter shakedown was completed, the system has had only
one unscheduled downtime. For the 11 months the system has been in
operation, the availability is in excess of 97¢, and for the past 6
months is about 98%. The system downtimes to data total 19 for 220
hr 40 min.

The data system has met all the objectives for which it was orig-
inally intended, and the operation is approaching routine status. Most
of the programming is complete with the exception of data-retrieval pro-
grams and the reactivity-balance program. It is expected that Tubture
programning and other system changes will be minimum and will only re-
sult from requirements generated by continued operation of the reactor.

Control~System Design — A, H. Anderson, D. G. Davis, and P. G, Herndon

 

Control Instrumentation Additions and Modifications. Further addi-
tions to and modifications of the instrumentation and controls systems
were made o provide additional protection, improve performance, or pro-
vide more information for the operators. One hundred twenty-five re-
quests for changes in the iInstrumentation and controls system (or in
systems affecting instrumentation and controls) were received and re-
viewed during the past report pericd. Of these, 52 requests resulted
in changes in instrumentation and/or controls, 19 were canceled, 8 did
not require changes in instrumentation or controls, and 25 are active
requests for which design revisions are either in progress or pending.
Prior to the initiation of design changes, the requests were reviewed
by persons responsible for operating the reactor and for the original
design., Changes in the reactor system were not made until the neces-~
sary approvals had been obtained, Some examples of these changes fol-
low.,

 

fuel and Coolant Pumps. The six relays that monitor the fuel-pump
motor current were replaced to prevent unnecessary shutdowns. Relay
chatter was causing spuriocus operations of the reactor flux scram set-
point circuits. Modifications to the existing relays and thelir current
settings did not correct this condition; therefore, new relays were in-
stalled. Some spurious operations occurred during thunderstorms. It
ig believed that lightning-induced transients on the power distribution
system caused the very fast-acting current relays to drop oub and scram
the reactor. To prevent this, the existing relays in the reactor flux
geram set-point circuite were replaced with time-delay relays.

 

After the faillure of a lube oil Tlow switch interlock had shut down
the coolant-salt circulating pump during a load scram, it was proposed
that all protection interlocks be removed from both the coolant- and
fuel-salt circulating pump control circuits. This would prevent un-
necessary punp stoppages and provide additional protection against freez-~
ing of salt in the radiator. After careful congideration it was decided
62

that these interlocks should remsin in the circults, because the protec-
tion they provide for the pumps oubwelghs their disadvantages. However,
new operating criteria stipulate that prevention of possible freezing
of salt in the radialtor is more important than protecting the punp from
possible damage resulting from loss of lube ©il or cooling-water flow.
To satisfy these nevw requirements, we presently plan to install a manual
switerd which (after proper administrative approval) may bne operated Lo
override the pump protective interlocks. Since pump shutdown may resuli
from other causes, such as actlion of overcurrent Lrip in the circuit
breakers, loss of TVA power, etc., additional changes in the power dis-
tribution and switechgear systems will be reguired 1f maximum obtainable
puping reliability is required. DManual switeh ecircuits that will over-
ride all protective interlocks in this purp control circult are now belag
esigned.

A weld-sealed electric pressure transmitter was installed on the
fuel-pump helium supply line 516 at a point hydraulically near the fuel
pump. This measurement will bte compared to the punp-~-bowl cover-gas
pressure to determine the pressure drop across the lube-0il static seal
between the Tuel-pump shield and the impeller shaft housing. This in-
formaticon should help o determine if there are periods of conditions
that should favor leakage of an abnormal amount of oil into the pump
bowl,

Master Control Circuits. A new jumper and asscciated circultry
were installed to provide a bypass around the freeze valve 111 Trozen
pernissive contact in the drain-tank helium supply valve countrol cir-
cuit. The Jjumper will make i, more convenient to operate through the
salt-transfer frecze valves. 1t was previously necessary to freeze
and thaw freeze valve 111 three times when blowing out the Lransfer
lines after a Tuvel-salt transfer.

 

To prevent radioactive gas backup into the drain-tank helium supply
lines, a continuous hellum purging system, compatible with safety re-
quirements, was designed and installed. Fach supply line is purged .
through a capillary flcow restrictor which is sized to limit the purge
flow rate to 0,07 liter/min. The capillaries are supplied from line -
519 at a peint downstream of the contalnment block valves.

It was proposed that spurious operations of interlocks in the RUN
mode, OPERATE mode, main blower No. 1, and main blower No. 3 control
circuits that were causing unnecessary shutdowns be prevented by re-
placing the existing relays in these circuits with time-delay relays.
This proposal was canceled after investigations indicated that the real
cause of the trouble was erratic orveration of the new 60-kva system
power supply. The performance of the power supply has been improved
considerably, and these circults are now operating normally.

Load Control.. To satisfy established operating criteria, addi-
tional control-grade circuits were designed to provide avtomatic load
setback action when the reactor is in the manual load control mode.
The dnstallation of these circuits is beling delayed pending a complele
review of the automatic load control system.
63

Additional safety-grade circuits were installed to provide auto-
matic load scram whenever the reactor control rods are scrammed. The
purpose of this revision is to help prevent salt from freezing in the
radiator.

The design of a system to measure the vibration of the radistor
cooling-air blowers and motors is now in progress. Surplus vibration
instruments suitable for this application are on hand.

Thermocouples. The thermocouple system did not change appreciably,
although a few thermocouples were added. Twelve with in-cell type re-
mote disconnects were installed on the fuel system off-gas filter and
particle trap. These are read on a multipoint recorder located in the
vent house. Thermocouples were also added to the new filters in off-
gas line 524, to the gas holdup volume tank in line 522, and to the
bearings on the main blowers.

Weigh System. The problem of leaking pneumatic selector valves in
the drain-tank welgh system readout was studied. Manometer readout is
accomplished by selecting parbicular weigh cell channels with the se-
lector valves. The valveg are composed of a stacked array of individual
valves operated by cams on the operating handle shaft. ILeaks from these
switches cause a slight error in the manometer reading. Efforts to stop
the leaks permanently have not been successful. Leak tests on a quick
digsconnect device indicate that it would be a satisfactory replacement
for each valve in the switch assembly. The valves were not replaced
because the problem is not severe enough at this time to Jjustify the
expense,

Auxiliary Systems. A pressure-reducing valve, a flowmeber, and a
containment block valve were installed in the reactor cell therxmocouple
nitrogen-pressurizing supply header. The normal operating pressure was
reduced from 50 to 5 pslg because of the excessive leak rate into the
reactor cell,

 

The component-coolant-pump control circuits were cross~interlocked
to prevent both pumps from being energized at the same time. This was
accomplished previously with contacts mounted directly on the circuit
breaker starter, but this arrangement would not allow one pump to op-
erate normally when the breaker for the other pump is racked out for
maintenance.,

A new design of instruments and controls is under way for a 350-
gal-capacity degassing and surge tank which was added to the treated-
water system to remove gases generated in the reactor thermal shield.,
A tank level measurement and possibly some flow control are required.,
An air purge system, which includes a pressure regulator, flow indi-
cator, and two solenoid-operated containment block valves, is alsc re-
quired.

Other minor revisions and additions were as follows:

1. The beryllium monitor was relocated from the vént house to the high-
bay area.,

2. The range of helium flow measuring loop FE-524-B was increased.
3. The range of the differential pressure transmitter wag increased from
O3 psig %o C~10 psig. Also, the indicator in this neasuring loop
was replaced with a strip-chart recorder.

4. A flowmeter was installed in treated-water line 877,

5. A new instrument power distribution panel was installed to provide
additional circuits required by other modifications and for future
expansion.

0. Manual switches were added in the main radiator blower damper con~
trol circuit so that the dampers can be closed when the blowers are
not rumning.

7. Radiator duclt blower air flow switches were added to the annunciator
circuits.

8. A time delay was provided ian the high-~bay area containment pressure
annunciator.

References

1. MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, pp. 10-12.

2. Ipbid., pp. 69-71.

3. MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, pp. 22-23.

%o MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, ppr. 87-92.

5, MSR Program Semiann. Progr. Rept. Ang. 31, 1965, ORNL-3872, pp. 55-56.
. R. J. Kedl, personal communicabtion, June 1966.

o
7. C. H. Gabbard, Thermal-Stiress and Strain-Fatigue Analyscs of the
MSRE Fuel and Coclant Pump Tanks, ORNL-TM-78 (October 1962).

8. MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, pp. 12-23. i
1.3_—'2-3 . 2

9. S. J. Ball and 1. W, Kerlin, Stability Analysis of the Molten-Saltl
Reactor Experiment, ORNI-TM-1070 (December 1965).

 

 

 

 

 

10. T. W. Kerlin and 5. J. Ball, Experimental Dynamic Analysis of the
Molten-Salt Reactor BExperiment, ORNL~TM-1647 {to be published,.

11. H. W. Hoffmen and S. E. Cohen, Fused Salt Heat Transfer - Part 111;
Forced Convection Heat Transfer in Circular Tubes Containing the
Salt Mixture NallO,~-NaNO3-KNO3, ORNL-2433 (March 1960).

 

 

 

 

 

12. R. E. MacPherson and M. M. Yarogh, Development Testing Performance
Evaluation of Liquid Metal and Molten Salt Heat Exchangers, ORNL-
CF~60-3-164 (Mar. 17, 1960); also ANS Meeting in 1959 (Washington,
DIC.)I

13. MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3930, pp.29-34.

 

 

 

14‘. Ibid-o 3 pp - '?8—'79.
15. MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, p. 72.
16. MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, p. 49.

 

 
2. COMPONENT DEVELOPMENT

Dunlap Scott

The development group continued to assist in the operation and test-
ing of the reactor. Much of their effort consisted of helping in diag-
nosing problems and in devising and installing equipment used to solve
she problems. The operational performance of the equipment is covered in
Chap. 1, but the description of problems and their solutions is given
below.

2.1 TFreeze Valves

M. Richardson

Operation of the freeze valve since the addition of the modulating
air controllers has been without incident. Operation of FV-103 was im-
proved by deleting the hysteresis feature of the FV-103-1A module, which
caused thermal cycling of the valve before the valve temperature reached
equilibrium. The module FV-103-1A now operates as an off-on switch which
alarms at 1000°F.

2.2 Control Rods

M. Richardson

The three control rods have operated without difficulty. Fiducial
zero positions are listed below. Changes are caused by changes in rod
length.

 

 

 

 

Rod 1 Rod 2 Rod 3 Date

1.74 1.55 1.40 2/12/66 (startup)
1.73 1.43 1.49 L2466

1.78 1.57 1.52 7/14./66

Rod~drop times for 51-in. fall for rods 1 and 2 remain at < 0.8 sec.
Drop time for replacement rod 3 remains between 0.9 and 0.95 sec since
installation.

65
2.3 Control-Rod Drive Units

 

M. Richardson J. BR. Tallackson

The control-rod drive units were operated from installation in
February until shutdown in August with the following difficulties.

1. The coarse-position synchro torqgue transmitter of the No. 3
drive Tailed. The transmitter was deenergized, and the drive unit con-
tinued operation using the fine-position synchro transmitier and the po-
tentiometer to indicate position.

2. The "Fill Permit" position potentiometer of the No. 1 drive
began to transmit erratic readings. 'The potentiometer was Jjumpered out
of the circuit, and the drive continued to operate satlsfactorily.

3. The high-temperature (200°F) switches which are mounted 6 in.
from the bottom of the drive-unit housings went into alarm on No. 1 drive
and remained in alarm under normal operating conditions. The switches
cleared when the reactor was drained and the cell temperature was lowered.

The potentiometer and synchro were replaced during the current sihut-
down. FExaminatlon showed that the potentiometer had developed a region
of cpen wiper contact. The synchro, mildly radiocactive, has not yet been
disassembled to determine the exact cause of Tailure. Resistance measure-
ments indicate a coll-to-coil short circuit in the stator windings, and
the partially melted plastic end cap is positive evidence of excessive
temperature.

The thermostatic temperature switches® in the lower end of the drive
housing are intended to indicate the approach of high-~temperature con-
ditions which would damage the motor and gear box assenmbly at the uppe
end of the housing. The switches are inaccessible, and it was not possible
to determine absolutely if the high-temperature indication from the No. 1
drive was correct or if one or both of the switches had misoperated. Since
neither of the other drive units showed a similar indication, there was
room for doubt. Measurements were taken of the winding resistances of the
drive motor, the fan motor, and the tachometer generator and compared to
those of similar units at room temperature and in the other two rod drive
units. These indicated that temperatures of the sensitive parts of the
gquestionable unit were pormal. Operation was continued with the switches
bypassed but with periodic checks of these winding resistances.

Inspection of the No. 1 and No. 3 rod drive units in August revealed
that the switches on the No. 1 unit were operating correctly, aand they
showed evidence (discoloration) that they had been exposed to high tem-
perature. Motors, gear boxes, and other parts of the drive units showed
no evidence of overheating or mechanical difficulty. The gear cases were
opened and were in excellent condition. The APL grease was soft and
adhering to the worm and worit gear. The lubricant had darkened slightly,
but there was no evidence of tarring or lumping of the grecase. The wire
insuvlation was in good condition.
67

It is reasonable to conclude that the air stream introduced at the
upper end of the No. 1 rod drive housing, while sufficient to keep the
gear-box components coocled, was not large enough to prevent the develop-
ment of the nigh temperature at the lower end of the housing caused by a
part of the heated gas stream rising from the thimble. The temperature
switches on rod drives 2 and 3 went into alarm during periods when the
reactor cell was operated above 150°F. Thie indicates that the air flow
to all three rod drives is marginal. The good condition of the gears,
motors, and lubricant makes immediate action unnecessary, but a study
will be made of methods of increasing the alr flow to the drive housings.

2.4 Radiator Doors

M. Richardson

Modifications® to the radiator-door-operating mechanism permitted
satisfactory operational performance of the doors through the last run.
However, excessive leakage of air into the radiator enclosure due to
poor seals became evident after the doors had been thermally cycled
several times during operation.

Examination of the doors after shutdown revealed that the metal seal
surfaces had been severely damaged. The metal seals, in turn, damaged
the mating soft seal mounted in the face of the radiator. The metal
strips had buckled between weld points, broken at some points, and been
torn away completely at the top of the outlet door. Some of this damage
is shown in Figs. 2.1 and 2.2.

The door structure and insulation boxes were in good condition ex-
cept for bowing of the 4~in. H-beam structural members. The maxinum bow,
which was at the top of the outlet door, was 5/16 in.

Laboratory tests were conducted on several arrangements of the metal
gsurfaces. A method of thermal cycling the test sections was devised
which would duplicate the distortion found in the seal strips on the doors
after the previous operation. Figure 2.3 shows three of the strip ar-
rangements that were tested; the strip support member (or T-bar) was the
same in each case. In the case of sections 1 and 2 of Fig. 2.3, these
strips were welded to the T-bar at intervals of 2 to 4 in. Because of
the heat loss through the door, these strips were operated about 500°F
above the temperature of the support bar, resulting in severe distortion
of strips between the welds. In some places the welds had broken, per-
mitting the seal strips to protrude away from the plane of the seal sur-
face. A similar test was run on a third arrangement of this seal which
consisted of 2-in. segments of the seal held in place at one end by a
plug weld and at the other end by an overlapping tab that interlocked
with the adjacent segment. This arrangement is shown as section 3 in
Fig. 2.3. It was found that thermal cycling did not affect the alignment
of these segments. The segments were spaced 0.031 in. along the T-bar.
68

PHOTO B5416

j ¢———SEAL STRIP SUPPORT MEMBER,
SEAL STRIP TORN AWAY

INSULATION BOXES

BUCKLED METAL SEAL STRIP

 

   
   

g )
SEAL STRIP —0 <«——SEAL STRIP SUPPORT (T-BAR)

Fig. 2.1. Radiator Outlet Door Showing Buckled, Broken, and Torn
Seal Strip (a) and Top Seal, North Side i

PHOTO 854147

BROKEN GASKET
RETAINER
RADIATOR FACE

SOFT GASKET

Fig. 2.2. BSoft Seal and Retainer Outlet Side of Radiator, Showing
Buckled and Broken Soft Seal Retaining Strip.
69

PHOTO 73612

.l ®

   

 

SECTION 1

   

BROKEN PLUG
WELD

 
  
  

SECTION

 

Fig. 2.3. BSeal Strip Tests — Radiator Door.
70

It was evident throughout all the test program that the seal-strip
support member bowed about 3/16 in. along the 32-in. length of test
section. It seemed reasonable to assume that bowing of this member would
also occur along the solid 10-ft 4-1/2-in. length of the T-bar in the
radiator door and contribute to the stressing of the door structure. The
door repair work included cutting the T-bar into segments; when this was
done, the bowing of the door structure was reduced from 5/16 in. to 1/8 in.

Corrective measures to the seals and T-bar are presently in progress.
These include:

1. Modification of the T-bar to minimize bowing by installation of ex-
pansion joints in the bar.

2. 1Installation of a new hard seal which is composed of 3-1/4-in.-long
overlapping links. The short links will be plug welded to the T-bar
by a single weld per link with a l/32—in. expansion Jjoint between
overlapping segments.

3. The existing soft-seal retainer, which is mounted on the face of the
radiator, 1s being modified to permit thermal expansion of the metal
retainer.

4. It is not anticipated that bowing of the T-bar will be completely
eliminated. Thermal expansion of the radiator face may also con-
tribute to the air leakage. An additional seal in the form of a
resilient, soft seal is being proposed as backup to the existing
seals. This seal would form a closure at those points along the
hard seal which are not in contact with the existing soft seal.

2.5 Sampler-Enricher

 

R. B. Gallaher

During runs 6 and 7, 35 samples were isolated with the sampler-en-
richer, 10 of which were 50-g oxide samples. To date 119 10-g and 20
50-g samples have been taken, and 87 enrichments made to the fuel system.

In the past six months, four major maintenance Jjobs were done on the
equipment. They were (1) replacement of the manipulator boots, (2) re-
placement of the drive unit latch, (3) repair of open electrical circuits,
and (4) recovery of a capsule which had fallen onto the gate of the op-
erational valve. A brief discussion of each Job follows.

Replacement of Manipulator Boots

 

A 12-psi pressure difference which was accidentally placed across
the manipulator boots caused a small puncture in the inner boot. To re-
place the boots the manipulator assembly was removed from the sampler.
The radiation level 3 in. from the hand was 10 r/hr as removed, but this
71

was reduced to 1 r/hr by scrubbing with soap and water. About 2 hr was
required for the Job of replacing both boots. The boots had been used
during the 48 sampling cycles since the operation at power was started.

Replacement of the Capsule latch

 

Just after the boots were replaced, the drive-unit motor stalled as
the latch which holds the sample capsule to the cable was being retrieved
through the lower bend in the transfer tube. After several tries the
letch was completely withdrawn. Testing showed the latch was Jamming in
the transfer tube near the top of the lower bend. The design was changed
to provide additional 1/8 in. clearance between the tube wall and the
lateh. The old latech and part of the cable were pulled through the access
port and the removal valve with the reactor operating at less than 100 kw.
The old latch was pulled into the sample transport cask to reduce the
radiation level in the work area. It was disconnected from the drive unit
cable and the new latch was installed.

Repailr of an Cpen Electrical Circuit

 

While operating the drive unit to be certain that the new latch would
pass around the lower bend without Jamming, the cable position indicator
stopped at 15 £t 2 in. from full withdrawn position, and the upper limit
switeh activated. FElectrical continuity checks showed open circuits to
both insert and withdraw windings of the drive-unit motor and to the upper
limit switch. All three of these l€ads penetrated the containment wall
through a common 8-pin receptacle.

When the drive nmotor stopped, the cable extended almost to the pump
bowl, preventing closing of the operational and maintenance valves.
Therefore, the reactor had to be drained before maintenance work could be
started. After the reactor was drained, the 15 ft 2 in. of cable was
pulled out of the transfer tube without exposing the fuel system to air by
using the one-hand manipulator to pull and the transport container and
access port operators to hold the cable. Then the operational and main-
tenance valves were closed to provide the containment barrier to allow
the motor assembly to be removed for repalr.

The assembly, which contains the drive-unit motor, was removed from
the sampler-enricher and placed in the equipment storage cell for decon~
tamination and repair. Shadow shielding and partial decontamination were
used to reduce the radiation level enough to permit direct maintenance.
Partial disassembly revealed that three connector pins had burned off one
8-pin receptacle. The damaged receptacle was removed and a new one was
welded in its place. All six receptacles in this location were filled
with epoxy resin Lo provide additional electrical insulation and mechanical
strength to the connector pins.

As the drive-~unit cable was being pulled from the transfer tube it
was bent in several places and needed to be straightened. The cable was
=

serubbed with soap and water until the radiation level was less than 5
r/hr at 3 in. Then mogt of the kinks were straightened. The remaining
bends in the cable did not appear to adversely affect the operations of
the drive unit when it was operated prior to reinstallation.

The parts were reassembled. All electric circults were checked for
continuity and for grounds, and all gas lines were Jeak checked. The
assembly was then reinstalled in the sampler-enricher and the normal
sampling was resumed.

Recovery of a Capsule

 

As a capsule key was being inserted into the latch, the manipulator
slipped and the capsule was Jerked through the access port and fell onto
the gate of the operational valve in the transfer line. A magnet was
lowered into the transfer line; the mild steel key3 on the capsule was
picked up by the magnet, and the capsule assembly was removed as the
magnet was withdrawn. The radiation level of the magnet after this op-
eration was 10 r/hr at 2 ft.

These four maintenance Jobs involved handling eguipment that had
been in contact with fuel salt. All work was done in contamination con-
trol zones. Contamination was found outside these zones only twice, and
both times it was from contaminated shoes. Apparently this type con-
tamination is not readily airborne. Personnel exposures did not exceed
a dose of 40 millirads/day for any one individual.

Operational Valve leakage

 

While the assembly was removed for replacement of the electrical
receptacle, the upper face of the operational valve gate was cleaned as
muich as possible with the valve closed. Several small, dark particles
were observed on the gate prior to cleaning. After cleaning, the leak
rate of helium through the top seal decreased for a while. After the
capsule had peen dropped onto the gate, the leak rate suddeniy increased
again, indicating that another particle had lodged in the sealing area.

Contamination of the Removal Valve Seals

 

The surface of the transport container used during the withdrawing
of the cable with the manipulator became very contaminated from contact
with the cable. When the transport container was remcved, part of this
contamination was transferred to the removal area seal. During subsequent
sampling procedures particulate contamination was spread to the top of
the sampler-enricher. Since the removal aresa was cleaned using damp rags
and the top of the sampler-enricher was made a contamination control zone
during sampling operations, there have been no further problems with con-
tamination.
73

Miscellaneous Problems

 

During one sampling sequence as the top of the transport container
was being lowered over the bottom part, which contalned a 50-g salt
sample, the steel wire connecting the capsule to the key caught on the
edge of the transport container top and was pulled down into the threads.
When the two pieces were threaded together, the wire caused the threads
to gall before the pleces sealed. The sample and the transport container
were ruined. ©Since then a long plastic sieeve has been placed in the
bottom of the transport container to hold the wire out of the way. The
sleeve also reduces the amount of contamination transferred from the
sample to the inside of the top piece of the transport container.

About 4 hr is presently required to decontaminate a transport con-
tainer sufficiently to be returned for use. Mild steel bottom pleces
have been fabricated which will be used one time and thrown away without
decontaminating.

Changes to the Control Circuit

 

Three changes were made to the control circuit.

1. The annunciator, which alarmed when capsule area pressure was
greater than manipulator area pressure, was removed. A permissive light
was installed to indicate when capsule area pressure wags equal to or less
than manipulator area pressure. This condition is necessary before the
access port can be opened but is not necessary at other times.

2. A Tuse was added to the drive-unit motor circuit to protect the
motor and the electric receptacles from excessive currents. The fuse is
located on one of the panelboards.

3. Voltage suppressors were placed across the two motor windings to
limit any high voltage peaks during starting and stopping of the motor.

2.6 Coolant Sampler

 

R. B. Gallaher

During runs 6 and 7, ten 10-g salt samples were isolated from the
coolant pump bowl. A total of 45 samples including two 50-g samples have
been taken using the coolant sampler.

One pin on an electrical receptacle shorted during a sampling cycle.
This pin was in the circuit to the indicator light which showed when the
capsule had been withdrawn 18 in. from the pump bowl. The receptacle was
removed and a new one welded in place.

A leak rate from the lower seal of the removal vaive increased. The
valve was disassembled, cleaned, and reassembled. The valve then sealed
properly.
2.7 Fuel Processing System Sampler

 

R. B. Gallsher

Design of the fuel processing system sampler has been completed ex-
cept for the shielding. The sampler-enricher mockup egquipment was modi~
fied to conform with the design. All parts were received, and the re-
visions tc the mockup panelboards were completed. The equipment is being
ingtalled at Building 7503 whenever craft are avajlable. The tubing in-
stallation is 95% complete, and the electrical and instrument work is
about 50% complete. The final assembly and leak checking remain to be
done.

2.8 0Off-Gas Filter Assembly

 

A. N. Smith

The off-gas filter in line 522 was originally provided to protect
the very fine trim of the reactor pressure control valve, PCV 522, from
becoming fouled by particulate matter. Following the analysis of plugging
difficulties in February and March 1966, efforts were directed toward the
design of a filter which also would trap organic materials. Primary con-
siderations were (1) choice of filter medium and (2) heat dissipation.

Filter Medium

The organic material was presumed to exist in a range of forms ex-
tending from low-molecular-welght vapors to suspensions of droplets and
polymerized solids. Charcoal was selected as the best available medium
for the removal of the heavier organics. Preliminary tests rveported in
Chap. 7 confirmed that the charcoal would have good efficiency for removal
of Cg and heavier molecules. Tt was assumed that lighter molecules would
prass through the main charcoal beds without adverse effects.

Heat Dissipation

 

The hezat load at the Tfilter comes primarily from beta decay of
krypton, xenon, and their daughter products. The contribution due to
krypton and xenon alone is estimated at 0.1 kw/liter. The contribution
due to the daughter products would depend on the assumed physical model.
For purposes of the filter design, the most pessimistic viewpolnt was
adopted, namely, that all daughters formed between the pump bowl and the
filter remain in suspension and are transported to the filter. Since the
efficlency of the charcoal trap would vary inversely with temperature,

a prefilter or particle trap was added for removal of solid daughters and
the entire assembly was water cooled.
75

Particle Trap

The design of the particlie trap is illustrated in Fig. 2.4. Gas
from the reactor pump bowl enters at the bottom of the unit through a
central pipe, reverses direction, and passes in succession through two
concentric cylinders of porous metal (felt metal), the first somewhat
coarser than the second, and a bed of inorganic (Fiverfrax) fibers. A
layer of stainless steel wool is inserted at the bottom of the unit to
serve as an impingement surface for material which might be thrown out
of the stream by centrifugal action. A bellows section is provided in
the felt-metal zone for thermal expansion. The Fiberfrax (a poor con-
ductor) is compartmented between perforated metal plates to provide for
better transmission of heat to the walls of the filter housing. Three
thermocouples are attached to the outside of the filter housing - one
near the top, one adjacent to the lower end of the Fiberfrax section,
and one adJjacent to the middle of the felt-metal section.

Specific design data are as follows:

Felt-Metal Section

 

 

Coarse Fine
Material Stainless steel Stainless steel
Manufacturer®s type FM 225 FM 204
(Fuyck Metals Co.,
Milford, Conn.)
Element size 2-5/8 in. diam X  3-5/8 in. diam X
9 in. long 9 in. Jlong
Filter area 0.5 £t? 0.7 ft2
Removal rating 98% > 1.4 u 98% > 0.1 p
Clean pressure drop 1-1/4 in. Hs0
at 4.2 liters/min
for coarse and fine
in series
Fiberf{rax Section
Material Carborundum Co. Fiberfrax, long staple
Tivers, 5 p mean dlameter
Packing density 8-1/2 to 9 1b/ft’
Total weight 189 g
Geometry Nine annular compartments, 4.05 in.

OD x 0.84 in. ID, with thickness

of two each at 1/4 in., two each

at 1/2 in., four each at 1 in., and
one at 1-1/4 in.
76

ORNL-DWG 86-11444

FINE METALLIC FILTER — NICKEL SPACER (TYP).,

N\ FIBERFRAX
‘ (LONG FIBER)—, N y~ THERMOCOUPLE (TYP)

COARSE METALLIC FILTER—,
N

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

" STAINLESS STEEL MESH / NICKEL BAFFLE (8} — A

s

STAINLESS STEEL BAFFLE STAINLESS STEEL BAFFLE >

Pig. 2.4. Particle Trap of Line 522 Filter Assembly.

Diactyl phthalate (DOP) efficiency tests® were run on the felt metal
(coarse and fine in series) and on the Fiberfrax with the following re-
sults:

 

 

 

Aerosol
s . i . Efficiency
Material Geometry Pype Av Particle (%)
P Size !

Felt metal Coarse and fine  Polydisperse 0.8 u 96.7
in series, 4 .
in. dianm .

Fiberfrax 4 in. diam X Monodisperse J.3 pn 99.4 -
'7"]_/4 in.

long, bhed
density 9-1/2
b/ft>

The efficiency of the assembled particle trap using the polydisperse
acrosol was determined to be 99.9%.

Charcoal Trap

The charcocal trap consists of 13 14 of l-in. IPS stalinlesgss steel
pipe, arranged in three hairpin sections of approximately equal length.
The trap was loaded with 1092 g of Pittsburgh PCB charcoal to give an
average bed density of 0.5 g/em?. Six thermocouples were installed at
5, 12, 51, 59, 105, and 113 in., respectively, from the bed inlet. The
77

entire charcoal trap 1s enclosed in a sealed tank through which water is
circulated at a rate of 10 gym and an. inlet temperature of 65 to 70°F.
The diameter of the trap piping was selected on the basis of theoretical
temperature profile caleculations.? One-inch IPS was selected as the size
which would permit the lowest {rap operatlng temperature consistent with
an acceptable pressure drop.

Flow pressure~drop tests gave the following results at the normal
reactor off-gas flow of 4.2 liters/min:

 

psi
Felt material 0.04
Fiberfrax 0.004
Total particle trap 0.06
Charcoal trap 0.85
Total filter assembly Q.95

The performance of the off-gas system after the installation of this
filter is described in Chap. 1.

2.9 524 Charcosl Bed

 

A. N. Smith

Diffusion of activity into the fuel pump upper off-gas line (line
524 ) resulted in high stack activity on two occasions during the report
period and dictated the need for a small charcoal bed in line 524 (see
Chap. 1).

A small upflow of gas (about 100 cmB/min) is maintained at the fuel
pump shaft ©o inhibit back-diffusion of oil vapors into the pump bowl.
Iine 524 serves to transport this flow to the off-gas system.

The bed consists of 9 £t of 3-in. sched 10 pipe arranged in hairpin
shape and loaded with 15.8 1b of Pittsburgh PCB charcoal. Bulk density
of the bed is 0.47 g/cmB. Holdup time at 100 cm3/min helium flow is
estimated as 2—1/2 days for krypton or 30 days for xenon. The bed was
placed in service on June 9, 1966.

2.10 Remote Maintenance

 

R. Blumberg

At the start of this reporting period, radiation levels encountered
by maintenance crews were mild due to the small accumulated power-hours
produced. This low radliation level permitted the use of temporary, and
78

at times casual, shielding, and in-cell manipulations were thus unhampered
by the restricticns of the shielding, so that the time required for com-
pletion of these early Jjobs was shortened. However, by the middle of
July, when the reactor was shut down for maintenance after having produced
some 7800 Mwhr, the radiation level was an important factor, adding to the
difficulty, time, and expense of the work. As a consequence the portable
maintenance shield had to be set up for each job, and extensive health
physics precautions, especilally in the area of contamination control
(housekeeping), had to be employed. As of this writing, the remote main-
tenance work of the shutdown is about 60% complete.

Based on the experience gained to date one may make the following
observations. (1) Contamination control is not a major difficulty. We
have handled several pieces of eguipment that had much wipable and trans-
Terable contamination withoul spreading contamination throughout the
working area. The contamination doces not become alrborne readlly. Never-
theless, the practice has been to use conservative handling measures;
these involve the use of plastic bag coverings, blotter paper, gas masks,
ete. (2) An encouraging sign is the ease of executing some of the routine
maintenance jobs. There have been many instances where removal and re-
placement of in-cell components have gone very smoothly and with minimuam
supervision. In general, these are items which have standard electrical
and piping disconnects and supports, such as the space cocler, and wnich
represent a considerable percentage of the equipment that must be main-
tained remotely. (3) Thus far, almost all of the maintenance technigues
or systems have been tried and have worked well. These are the portable
maintenance shield, graphite sample system, vent house shielding, the
remote maintenance control room, and repair techniques in the decontami-
nation cell. The remote maintenance tasks where we do not yet have ex-
rerience are in general associated with the handlling of the large compo-
nents. The shielding provided for maintenance has been guite adequate.

The following 1s a summary of the tasks performed during the Jlast
period. The accumulated hours of power operation, together with an in-
dication of the radiation level in the immediate area Just below the work
shield, are given for each task. 1In general, the work shield was ef-
fective in reducing the background radiation level above 1t to less than
5 mr/hr except while a tool penetration was partially open, and then the
radiation level at the hands would be near 200 mr/hr. The maxizum ac-
cumulated deosage Lo any one worker was less than 1/4 of' the maximum per-
migsible level over the periocd.

March 1 - 35 Mwhr

 

Removed the flexible Jumper which connects the off-gas Jine
of the pump bowl to the permanent in-cell part of line 522, in-
spected the Tlange faces and inside of the piping with a bore-
scope and binoculars, obtalined specimen of a residue from the
piping for chemical analysis, and restored tihe system. This
task was part of the effort to diagnose the restriction problem
in the off-gas system. The radiation level was 200 mr/hr at
floor level.
79

March 10 — 35 Mwhr

 

Removed and replaced the reactor cell east space cooler.
This operation was accomplished mostly by the craft forces.
Readings were 500 mr/hr at floor level.

March 12 - 35 Mwhr

 

Installed a thermocouple on a horizontal section of off-
gags line 522. A C-clamp was modified to provide the contacting
force to hold the couple onto the 4-in. pipe. The thermocouple
leads were connected to a spare thermocouple lead-out bhox.
Radiation was 500 mr/hr at floor level.

March 31 — 35 Mwhr

 

The revised PCV 522 valve and {ilter unit descrived ahove
were installed in the vent house area. Changes in the shielding
and containment arrangement were done to make possible the re-
mote replacement of any one of three component parts, that is,
the valve, the particle trap, or the carbon bed, rather than
having to remove the entire assenbly. The new installation re-
quired larger maintenance shielding and stronger structural
support for the shielding.

April 30-May 6 — 820 Mwhr

 

Repaired the sampler-enricher. A large subassembly of the
equipment was removed to the decontamination cell, where it was
cleaned and repaired using local shielding. This was the first
real use of the decontamination cell, which had heen prepared
for this type of work, and it was found to work quite well.
There were contact readings of above 100 r/hr, but the radiation
was soft enough that a little shielding reduced the field to ac-
ceptable levels.

May 20 — 1932 Mwhr

 

Installed temporary instrumentation piping at the PCV 522
areca for particle trap, carbon bed, and valve pressure drop
neasurements. Tubing connections were made up remotely; this
was a new maintenance procedure. Readings were 100 r/hr at
floor level and 10 r/hr through an open tool penetration.

July 27 — 7822 Mwhr

 

This was the start of the shutdown which has extended to
the end of this report period. The reactor was shut down on
July 17, and the graphite—~INOR-& surveillance samples were re-
moved from the reactor core on July 28, 11 days later. The
work required three days. The estimated radiation level while
the sample was being transported was 1500 r/hr at 1 ft. The
remote control room for the crane was used for the first time
and performed satisfactorily. The tools used for the operation
were contaminated to a level greater than 100 r/hr.
3.

‘4’0

80

August 510 — 7822 Mwhr

 

Removed, repaired, and replaced the Nos. 1 and 3 control
rod drives. The detalils of the repair work are described in
bect. 2.3. The remote handling operation of the removal and
replacement went smoothly. Radiation was 500 mr/hr at floor
level.

August 15 — 7822 Mwhr

 

Removed and replaced the reactor cell west space cooler.
The Jjob went routinely and was accomplished mostly by the
craft forces. Radlation was 1 to 6 r/hr at floor level.

August 17 — 7822 Mwhr

 

A gas pressure reference line which became filled with
a salt plug was cleared by applying pressure while heating -
the line. This task was unanticipated and required some de-
sign, Tabrication, and testing prior to installing a heater
on a long-handled tool and safely operating it remotely.
Readings were 60 r/hr at floor level and 10 r/hr through a
tool penetration.

References

MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNI~3936, p. 54.

 

Thid., pp. 54—56.
Tohid., p. 58.

E. C. Parrish and R. W. Schneider, Tests of High Efficiency Filtlers
and Filter Installation at ORNL, ORNIL-3442 (May 17, 1963).

 

 

Personal communication from 5. J. Ball and J. R. Engel.
3. PUMP DEVELOPMENT

- P. G. Smith A. G. Grindell

3.1 MORE Pumps

Molten-2alt Pump Operation in the Prototype Pump Test Facility

The prototype pump was operated for 2631 hr, circulating the salt
LiF-Bel'p~ZrF 4~ThE 4~UF (68,424 .6~5.0~1.1~0.9 nole %) at 1200°F., The
fuel-salt pump impeller of ll~l/2 in. OD (size installed in the MSRE
fuel pump) wags used. Meagurements were made of the concentration of
helium bubbles in the cilrculating salt, and various tests were con-
ducted on the pump-tank and the catch-basin purge gas lines,t

The radiation densitometer* was used to determine the concentration
of helium bubbles circulating in 1200°F salt during operation at normal
pump-tank liquid level and with a fuel-pump impeller of 11-1/2 in. di-
ameter running at 1170 rpm. The helium concentration in the salt was
0.1% by volume., This value contrasts with 4.6% by volume measured pre-
viously?® in the prototype facility using a 13-in.-diam fuel impeller,
and compares favorably to a barely detectable concentration in the MaRL
fuel circuit.

Tegts were performed which included analyses to debermine the hy-
drocarbon content of the pump-tank and catch-~basin purge gas streams
and. the composition of hydrocarbons. They were done as part of the
effort to establish whether the fuel pump was the likely source of hy-
drocarbons in the off-gas system of the MSRE. During these tests the
Tlow of gas down the shaft annulus at constant supply pressure decreased,
and this was taken ag an indication of a partial plug in the ammulus.
The parbial plug increased the pressure difference acrosg the gasketed
joint between the shield plug and the bearing housing from O to 5 psi
at the design purge flow of 4 liters/min. Under these circumstances
the off-gas from the pump tank contalined several hundred parts per
millicn of hydrocarbons. This indicated that the pressure difference
was forcing oil from the catch basin through the gasketed seal, down
the outside of the shield plug, and into the pump tank. The partial
plug could be removed temporarily by stopping the pump for a short
time or by ralsing the temperature of the circulating salt from 1200
to 1250°F. When the plug was not present, the concentration of hydro-
carbons in the off-gas stream from the pump tank was very low. BSamples
of the plug material were obtained when the pump was dismantled. Ex-
amination showed them to be salt of the composition that was being cir-
culated by the pump. The mechanism by which the salt reached the shaft
apnulus is being investigated.

Other tests were run in which oil was injected into tThe pump bank
through the sampler-enricher nozzle. Results of the analyses apnd other
information about both types of tests are reported in more detail in
dect. 7.5, subsection entitled "Analysis of Hellum Blanket Gas.'

 

*Densitometer was provided and operated by V. A. McKay of the In-
strumentation and Controls Division.

81
g2

Pump Rotary Element Modification

The modification® that replaces Lhe gasketed seal between the
shield plug and the bearing housing with a seal weld, which should
eliminate the oil leakage into the pump bowl, was completed on the
spare rotary elements for the MSRE fuel and coolant pumps and on the
rotary element for the MK-2 fuel pump. The spare element, for the fuel
pump was tested in the cold shakedown stand and installed in the pro-
tobtype hot test stand for shakedown. During preheat of the system and
pump, approximately 1 qt of oil passed through the lower rotary seal
into the catch basin. The cause of the poor performance of the seal
is being investigated, and the rotary element is being cleaned and re-
assembled,

Cold and hot shakedown tests of the spare rotary element for the
MSRE coolant pump were completed. The hydrocarbon content in the pump
tank off-gas was not more than 15 ppm,

Tubrication System

The lubrication pump endurance test® was continued, and the pump -
has now run for 27,000 hr, circulating oil at 160°F and 70 gpm.

MK-2 Fuel Pump
The pump tank? is being fabricated, and, when completed, it will
be installed in the prototype pump test facility. Then the MK-2 fuel

punp will be tested at operating conditions gsimulating those required
by the MGRE.

3.2 Cther Molten-Salt Pumps

 

PK-P Fuel-Pump High-Temperature Endurance Test .

 

Endurance operation3'was halted as a result of a failure of the

drive motor. The pump had operated continuously for 7830 hr, circu-
lating the salt TiF-BeF,-ThF 4-UF, (65-30-4-1 mole %) at 1200°F, 800
gpm, and 1650 rpm., The rotor windings of the wound rotor mobor had
seized against the stator. The pump has operated for a total of 23,426
hr in four tests.

Pump Containing a Molten-Salt-Lubricated Journal Bearing

 

The gimbals support for the salt bearing4 was modified, and a new
bearing and journal were fabricated. Performance of the salt bearing
Wwill be investigated with an oll lubricant prior to molten-salt lubri-
cation.
83

References

MSR Program Semiann. Progr. Rept. Feb. 28,
75.

MSR Program Semiann. Progr. Rept. Aug. 31,
65fi

MSR Program Semiann. Progr. Rept. Feb. 28,
Tpid., p. 76.

1966, ORNL-3936, pp. 74—

1966, ORNL-3936, p. 75.
4. INSTRUMENT DEVELOPMENT
R. L. Moore G. . Burger J. W. Krewson

4.1 Temperature Scanner

 

Performance of the tcmperature scanning systeml conttinued Lo be

generally satisfactory, although some problems were experienced with
the oscilloscopes and mercury switches and some system instability was
noted.

The problems with the oscilloscopes were only continuations of
those previocusly experienced due to the age and design of the scopes.
The scopes are about 12 to 15 years old and have been a continuing
source of trouble. Manufacture of the scopes was disconbinued, so .
1o spare parts were available. Two new solid-state-circuit scopes
were ordered and installed. The new scopes have apparently elimi- -
nated that problem, although more time is required to evaluate thelir
rerformance.

Although the mercury switches have contlnued to give much better
service than expected, some problems due to normal wear developed.
Upcn ordering replacement parts for the switches, it was discovered
that the switches were no longer manufactured and no spare parts
existed. Some spare parts and switches were obhtained from the ORNL
Reactor Divisicon, but it was apparent that a replacement switch was
needed. A possible replacement was found at Union Carbide Corpora-
tion, Olefins Division, in South Charleston, West Virginia. The Olefins
Divigion has developed a solid-state multiplexer as a direct replacement
for the mercury switch and has agreed 4o sell one to ORNL for test and
evaluation. An order has bheen placed for one unit, with delivery expected
about January 1967. 1In the meantime the Reactor Division will coatinue
to supply mercury switch spare parts as long as they are available and -
can be released for our use. -

The scanner instability problems mostly resulied from operating
the system outside its design range. <The system was adjusted to ob-
taln more display resclution and to provide easier readout for the
operators. In order to do this the alarm discriminator was required
to operate outside its design range, causing alarm set point insta-
bility. To eliminate this problem while continuing this mode of op~-
eration would require redeglign of the discriminator, so operation of
the system was changed back to comply with the original design. A
calibration and test unit is being designed and will be installed
shortly to provide the operators with an easy method of calibrating
the system and of checking its stability. This test unit will pro-
vide & variable, accurately calibrated signal to one point on any of
the selected scanners and will allow the operators to check the scan-
ner calibration and alarm set points during routine cperation of the
scamnner system.

84
85

4.2 High-Temperature NaK-Filled Differential-Pressure Transmitter

Testing of the coolant-salt system flow transmitter which failed
in service at the MSRE and which was subsequently refilled with sili-
cone 0il? was continued. Further testing at temperatures over the range
from room temperature to 1200°F confirmed that the temperature sensi-
tivity had been significantly reduced by the refilling operation but
was still excessive. Before refilling, the zero shifts cbserved were
of the order of 15 to 20 in. (water column) per degree Fahrenheit change
in body temperature. After refilling the shifts were of the order of
1 in. HyO per degree Fahrenheit. Zero shifts of the order of 0.1 in.
HoO per degree Fahrenheit or less are required to obtain the accuracy
desired in measurement of MSRE coolant-salt flow. In all cases the
temperature-induced shifts observed were zero shifts. No shifts of
span calibration were observed. Also, the instrument did not exhibit
a sensitivity to statiec pressure changes either before or after refill-
ing.

The remaining shifis are presently believed to be due Lo the con-
tinued presence of some gas vold in the silicone-oil-filled cavities.
Attempts to eliminate these voids will be made by repeating the £ill-
ing operations. To assure that all remaining gas has been removed, at-
tempts will be made to obtain a "harder" vacuwn than the 28 in. Hg
vacuum previously obtained. After refilling, the temperature sensi-
tivity tests will be repeated.

The new NaK-filled differential-pressure transmitter ordered for
use as an MSRE spare was received; however, acceptance tests showed
that the instrument was extremely sensitive to temperature and static
pressure variations and would, therefore, nol meel specifications.
Negotiations with the manufacturer for repalr or replacement of the
transmitter are in progress.

 

4.3 Molten-Salt Level Detectors

Performance of all molten-salt level detectors installed at the
M3RE, on the MSRP Level Test Pacility, and on the MSRE Prototype Punp
Test Loop continues to be satisfactory.2’3 No failures have occurred
in any of the level devices during the past report period, and, except
for the changes made in electronic equipment associated with the ultra-~
sonic level probes reported below, no modifications have been required.
The corrections made in the calibration of the vall-float-type trans-
mitter installed on the MSRE cocolant-salt pump4 were apparently ade-
quate, and no further adjustments have been necessary.

To correct the excessive frequency drift present in the excitation
oscillator supplying the ulirasonlie level prsbe,5’6 a number of minor
changes were made in the components and clireultry of the oscillator and
detector amplifier circults. In general, these changes involved the re-
placement of a numwber of critical components with high-grade components
having very small temperature coefficients, increasing the size of
86

coupling capacitors so that they would have smaller reactance at the
25-kilohertz operating frequencies, and making several minor circuit
revisions to increase the stability of the B+ supply. These changes re-
sulted in a considerable improvement in the frequency stability. Before
the changes were made the oscillator drifted randomly, with a maximum
deviation from the center or resonant frequency of 300 hertzes. This
deviation was sufficient to cause the instrument to become inoperative.
When the changes were made, the maximum drift observed after warmup was
5 hertzes. These drifts were observed in an air-conditioned laboratory
and are expected to be greater in the field; however, it is expected
that the increased stavility obtained will be adequate. A complete
check on the operability of the ultrasonic level probe requires an
actual variation of the level of molten salt in the fuel storage tank

of the MSRE. No salt has been transferred to or from this btank since
the probe circulbry was modified, and field checks on the effectiveness
of the modifications have not been made at this time. These checks will
be made when salt transfers can be made without interference with MSRE
operations.

4.4  Helium Control Valve Trim Replacement

Modifications and/or repalr of two weld-sealed hellum control valves
which failed in MSRE service during the previous report period7 have been
completed. The spline-type trim previously used in the main helium supply
contbrol valve, walch had failed twice, was replaced with a taper trim.

The use of taper trim was permissible and desirable because of the rela-
tively high flow rates involved.

There have been no further fallures of helium control valves during
the past six months.

To determine the feasibility of controlling very low flows of dry
heliun with sliding disk valves, a mockup of such a valve was constructed
and tested. The active member of the gliding disk valve is a flat disk
attached to the valve stem and held against a seating surface by spring
and pressure action. The surface between the disk and seat is lapped
to light band flatness to assure tight shutoff. Throttling control of
helium flow is accomplished by means of a tapered V-notch in the disk,
which connects a port in the disk to a port in the valve seat. Rota-
tion of the disk causes the effective cross-sectional area of the pas-
sage between the ports to vary in proportlon to the degree of rotation.
This type of construction offers promise in the control of dry-helium
flow because of an inherent self~-wiping action and because of the pos-
sibility that the valve might be operated without the use of lubricants.

Preliminary results of bench tests were inconelusive because of
excessive leak rates between the lapped surfaces of the disk and seat.
The tests will be repeated after additional lapping, cleaning, and re-
assembly of the valve are completed.
MSR Program Semiann.
Tbid., pp. 7778,

MBR Program Semiann.

87

References

Progr. Rept. Feb. 28, 1966,

Progr. Rept. Feb. 28, 1965,

 

MSR Program Semilann.

Progr. Rept., Feb. 28, 1966,

 

Ibid., p. 77.
MSR Program Semiann.

MSR Program Semiann.

Progr. Reptb. Aug. 31, 1965,
Progr. Rept. Feb, 28, 1966,

 

ORNL-3936,

ORNL-3812,

- ORNL~-3936,

ORNL-3872,
ORNIL-3936,

p. 80.

Pp. 4243,
p. 78.

pp . 66—'70 »
p. 79.
5. REACTOR ANALYSIS

B. E. Prince

5.1 Analysis of Rod Drop Experiments

 

Desceription of Experiments

 

The series of zero-power nuclear experiments made during MSRE run
No. 3 included several rod drop experiments as part of The control rod
calibration program. The purpose of these experiments was to provide
an alternative and independent measurement of the control rod reactivity
worth, which could be compared with worths determined by other methods -
(rod-bump period measurements and equivalent €357 addition). Three sets
of rod drop experiments were performed, after additions of 30, 65, and )
87 capsules of excess 23%U had been made. Each set of experiments in-
cluded rod drops both with the fuel circulating and with circulation
stopped. We have analyzed the experiments made with the Tuel pump
stopped, and the results are summarized below.

The rod drop experiment consists of the intentlonal scram of a rod,
or rod group, from an initially critical configuration, and the recording
of the decay of the neutron flux as a function of time following the
scram. One then determines the amount of negative reactivity required
to produce this fTlux decay trajectory. The trajectory is characterized
by a sharp drop in the flux immediately feollowing the scram, which corre-
sponds closely with the actual fall of the rod. The curve rapldly and
continuously evolves into one with a much slower rate of flux decrease,
governed by the decay of the initial distribution of delayed-neutron
precursors.

Due to the reguirement imposed by tnls experiment of accurately
recording the flux while it is rapidly falling by about two decades, the
graphical record obtained from the trace of a log n recorder is of limited
usefulness. The combined requirements of fast response, good counting
statistics, and reproducibility can be served, however, by recording the
integrated count as a function of time following the drop.

In order that the integrated flux-time curve could be recorded with-
out requiring scalars with special accurately timed cycles for count ac-
cumulation and recording, three scalars, together with a photographic
technique of rapid recording, were employed in the following way. One
of the scalars was driven by one of the fission-chamber channels. The
other two were operated as 60-cps timing devices. One of these Limers
was synchronized to start with the scram of the rod, while the other
timer and the count-accumulating scalar were synchronized and started a
few seconds before the scram was initiated. The three scalars were
stacked together in a vertical array, and a rapld-action camera was used
to simultanecusly photograph the records on the three scalars approxi-
mately once every second, starting a few seconds before the scram and
ending about 30 sec after the scram. From these photographs, the count

g8
89

rate at the initial critical condition, the time of the scram signal,

and the integral of the flux, or count rate, as a function of time fol-
lowing the scram could be accurately determined. In these experiments,
the position of the fission chambers was adjusted to obtain an initial
count rate at criticality of approximately 30,000 counts/sec, which was
low enough to result in quite small counting rate losses due to dead-time
corrections. Also, the remote position of the MSRE instrument shaft,
relative to the reactor core, would be expected to minimize errors due to
changes in the spatial flux shape during the rod drop experiment.

Analysis Procedures

 

The method used to analyze the experiments performed with the fuel
pump stopped was that of comparison with theoretical flux decay curves.
These latter curves corresponded to a negative reactivity lnsertion with
magnitude determined from the integral worth vs position curves already
obtained from period measurements and critical position comparison tech-
niques,l and with rate of insertion corresponding to the fall of the rod
Trom its initial critical position. The theoretical time~integrated Tlux
decay curves were calculated using the MATEXP program2 to integrate the
standard space-independent reactor kinetics eqguations. Since this program
is designed to integrate a general system of first-order differential
eguations, the theoretical time-integrated flux following the scram could
be obtained by solving the system:

 

 

Wit
leét) = n(t) , (1)
am(e) () =B (), p oA o (5) (2)
dt A . i i ?
i=1
dCi(t) Bin(t)
-T:-—?\ici(t)-k-—j\—‘_w’ i=131,2, . .., 06, (3)
where

¥(t) = time-integrated count rate following the scram,
n(t) = detector count rate following scram,
Ci(t) = delayed-neutron precursor concentrations, normalized

to detector count rate,

reactivity addition vs time following scram,

i

25 (t)

Bi = production fraction for ith group of delayed neutrons,
A; = radiocactive decay constant for ith group precursors,
A = promplt neutron generation time,

6
B: Bi
i=1
90

The initial conditions for performing the integrations of Egs. (1), (2),
and (3) are

n{0) = initial count rate at critical condition , (5)
B, n(0)
c;(0) = ”"Ki;"' . (6)

The time variation of the reactivity is governed implicitly by the
equations "

so(t) = plyo) = oly(t)] (7)
y(t) = vo , 0Os+tst,, -
1 -
= yo — 5 at? b, ST St 4t

wWwhere

v = rod position, inches withdrawn (0 = y £ 51 in.); initial
position, yg; scram position, Yo

o{y) = magnitude of reactivily worth corresponding to rod position
vy, normalized to zero reactivity at fully withdrawn -
position; , -

t = effective lag time between scram signal and start of rod
drop (~20 msec);

a = acceleration during fall (~l5 ft/sec2);

t = time to fall to scram position.

As discussed previously, the value of p{y) used in this analysis was
determined by integration of period differential-worth measurements. We
have also used an algebraic representation of these experimental curves,
obtained by a least-squares curve-fitting procedure.?

In applying the above analysis to the experiments in which rod groups
were dropped, the magnitude of the total negative reactivity insertion
was determined by combining the calibration curve for the single rod with
the results of rod-shadowing experiments (comparisons of critical positions
of single rod and rod groups).l Typical rod-group drop experiments con-
sisted of the simultaneocus scram of the regulating rod (rod No. 1) from
its initial eritical position and one or both of the shim rods from the
91

fully withdrawn position. The normalized shape of the reactivity in-
gertion curve for the rod group was approximated by that corresponding
to a single rod, falling from the fully withdrawn position.

Results

The results of the rod drop experiments made after additions of 30,
65, and 87 capsules of enriching salt are shown in Figs. 5.1, 5.2, and
5.3 respectively. In all three sets of experiments, the analysis of the
single rod drops shows very good consistency with the rod worth cali~
bration obtained by integration of the differential worth measurements.
A similar consistency was obtalned from the experiments involving the
scram of all three rods. In the case of the two-rod experiments, results
obtained after 65 and 87 capsules appear to be slightly anomalous with
respect to the experiment with the same rod group after 30 capsules.
Multiplication of the magnitude of the negative reactivity insertion by
a factor of agbout 1.05 brings the caleculated and experimental curves for
these experiments into better agreement. However, another source of this
discrepancy could be in the approximation of the shape of reactivity in-
sertion vs time for the tandem fall of the rods by that corresponding to
a single rod. This source would be expected to have its greatest ine
fluence on the two-rod experiments.

The sensitivity of these experiments to variations in the magnitude
of the reactivity insertion is illustrated in Fig. 5.4. Here the analysis
of the experiment where rod No. 1 was scrammed after addition of 30 cap-
sules 1g reproduced from Fig. 5.1; the theoretical curves are added which
correspond to an increase and a decrease of 5% of the total magnitude of
negative reactivity insertion. ¥For this particular experiment, this
corresponds to an increment of 0.007% Bk/k in the magnitude of reactivity.
With the exception of the apparent ancmaly in the experiments involving
the scram of two rods, the results of all these experiments were well
within the 5% band of self-consistency with the rod calibration results
obteined from the differential worth measurements.t

Work will continue on the analysis of these experiments and similar
cnes performed with the fuel eirculating, in an effort to assess the Tull
potential of the rod drop experiment as a routine method for rapid periodic
redetermination of the shutdown worth of the MSRE control rods.
92

ORN{-OCWG 66-11445

 

 

  
   

 

(x10%) 7 —— ey . ,
| ; |
' o ea EXPERIMENTAL POINTS
CALCULATED BY INTEGRATION OF REACTOR
10 KINETICS FQUATION; REACTIVITY DETERMINED -
FROM INDEPENDENT CALIBRATION MEASUREMENTS.
! |
/
8 [ ST
— f |
; \
D 1
O
a ‘ 1
§ 6l | ROD NO.f SCRAMMED o]
i
i3 ! RODS NO.1 AND 2 SCRAMMED..—*
!/
4 e : C e e aa
/'T /A/
,/
‘TODS 1,2,8ND 3 SCRAVIVED

 

 

20
TIME AFTER SCRAM (Sec)

Fig. 5.2. Results of Rod Drop
Experiments After 65 Capsule Addi-
ticns.

Fig. 5.1. Results of Rod Drop
Experiments After 20 Capsule Addi -
tions.

 

ORNL-DWG €6 -11446

   
   
   
  
  
 

 

 

  
  

 

 

 

 

(x *04) ..............................
o,8,8 EXPERIMENTAL POINTS
12 | — CALCULATED BY INTEGRATION (F —
REACTOR KINETICS EQUATIONS:
REACTIVITY DETERMINED
FROM INDEPENDENT CAL -
IBRATION MEASUREMENTS,
i ;
10 / -
. ROD NC.1 SCRAMMED
: i
i
|
Y
= _
L8 ; ]
> |
S ;
O |
4 !
a |
3 .
o | ~
o8 RODS NO 1 AND 2 SCRANM Y
=z
4 - - B
i '__',,,.—-»--"'"‘“
#M/M,ffib,fi,,
RODS NO. 1.2, AND 3
SCRAMMED
2 b f % - -]
'» - | ; |
i | |
i ! |
; | ‘
: | T
o L :
0 5 © 15 20
TIME AFTER SCRAM (sec)
(x 1073

INTEGRAL COUNT

 

93

ORNL DN\J B6E—11447

 

   

    

RODS 1 AND 2 SCRAMMED

TRODS ¢,2, AND 3
QCF?,’-\MMED

 
 
 
  
   
 

o,s, s EXPERIMENTAL POINTS

CALCULATED BY INTEGRATION OF — —
REACTOR KINETICS EQUATIONS;
REACTIVITY DETERMINED FROM
INDEPENDENT CALIBRATION

MEASUREMEN'L

 

 

Fig. 5.3. Results of Rod Drop
Experiments After 27 Capsule Addi-
tions.

 

 

 
  
 

 

 

 

 

5 {0 5 20
TIME AFTER SCRAM (sec)
ORNL-DWG 66-11448
umfl( T ‘
‘ # EXPERIMENTAL POINTS g
=== CALCULLATED BY INTEGRATION OF B
REACTOR KINETICS EQUATIONS,
8 [ e e |
’_
% 6 l T - o
S L7
~d 72 "
<1 Vv
& S
e i
; 7
- ROD NOA SCRAMMED AFTER 30 CAPSULE
ADDITIONS.
: NECATIVE REACTIVITY INSERTION,AP, DETER-
MINED FROM INTEGRATION OF DIFFERENTIAL ...
WORTH MEASUREMENTS.
ECATIVE REACTIVITY INSERTICN,1.05 Hp.
C: N:CATIVt REACTIMITY INSERTION, 095Ap
0
TIME AFT!—,H SCRAM isec)
Fig. 5.4. BSensitivity of Rod Drop Experiment to Changes in Magnitude

of Reactivity Insertion.
ez

References

P. N. Haubenreich et al., MSRE Zerc Power Physics Experiments, ORNL
report in preparation.

S. J. Ball and R. K. Adams, MATEXP — A General Purpose Digital
Computer Program for 3olving Nonlinear Ordinary DifTerential
Equations by the Matrix Exponential Method, ORNI~IM report in

 

preparation.

MBR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNI-3936, pp. S2-87.

 
Part 2. MATERTATS STUDIES
6. MSRP MATERTALS
G. M, Adamson, Jr.
6.1 MSRE Materials Surveillance Testing
W. H, Cook

Specimens of grade CGB graphite and Hastelloy N (INOR-8) were re-
moved from the core of the MSRE after 7800 Mwhr of operation as part of
the materials survelllance programwl The visual appearance of the metal
and graphite wag good. The external surfaces were virtually free of salt.
The metal was dull gray, and the graphite appeared unchenged by the expo-
sure, The Reactor Chemlstry Division was supplied with graphite speci-
mens for fission product analyses from the top, middle, and bottom of
thie asgembly. The results of thelr analyses are reported in Chap. 7.
Other exsminations scheduled for the graphlite and metal are in progress.

T™e objective at this stage of sampling was to remove one stringer
(one-third) of the assembly and return the others, plus a replacement,
to the reactor. However, approximately one-seventh of the assenbly iua
a wone just above the midplane of the reactor was severely damaged.
Graphite specimens were buckled and broken, and the tensile rods were
beut. Apparently, the assembly had been locked together by frozen salt
during the reactor cooldown, and the large differences in the coeffi-
cients of thermal expansion of the graphite and metal created high
stresses which led to the mechanical damage. The damsge in this zone
occurred in all three Hastelloy N and graphite stringers, requiring that
all three be replaced. The graphite and Hastelloy N in the other por-
tions of the array were suitable for the intended property evaluation
tests.

A slightly modified replacement for the damaged reactor core gpeci-
mens has been installed in the MSRE. The modifications were made to re-
duce mechanical stresses in the assembly by reactor heating and cooling
cyeles, The Hastelloy N tensile specimen rods in the new set are made
from reactor vessel wall and head materials plus two modified alloys of
Hastelloy N, one containing 0.52 wt % Ti and the other 0.43 wt % Zr.
These additions were made for the purpose of reducing the effects of
radiation on the alloy.

In general, the molten sgalts drained well from the graphite and the
mebal specimens, as shown in Figs. 6.1 and 6.2. The near absence of salt
ajded the disassenbly.

The overall mechanical damage 1n the reactor core specimens is shown
in Fig. 6.2 and in more detail in Fig. 6.3, A region of minor damsge
near the bottom of the assembly is shown in Fig, 6.3a, and the region
of maximum damage is shown in Fig. 6.3b. Approximately 25% of the
graphite specimens were broken in each of the three stringers. The
general appearance of the graphite was the same as it was when it was
machined. Note the reflection of the cross pin in the surface of the
graphite in Fig. 6.3a.

27
28

R-30975

200000
000000

 

Fig. 6.1. Basket of Reactor Core Specimens Removed from the MSRE
After 7800 Mwhr of Operation.
   
  

99

R-31011

 

Fig. 6.2. Overall View of the Reactor Core Specimens After Removal

from Basket.

SLEEVE SHOULDER

 

FLUX MONITORS SHEATH

 

Fig. 6.3. Details of the Mechanical Damage to the Graphite and
Hastelloy N Specimens in Zones of (a) Moderate and (b) Severe Damage.

The graphite specimens had been assembled so that they formed tongue-
and-groove joints that were bound together by Hastelloy N straps, as shown
in Figs. 6.2 and 6.3. Slip-fit sleeves had been spot-welded to the metal
straps that hold the rods of tensile specimens in position. The graphite
was held within the cage of tensile specimen rods by cross pinning through
the shoulders of the tensile specimens (Fig. 6.3a).

The total assembled length of the graphite stringers at room
temperature was 62 in. At the reactor operating temperature, 1210°F,
the graphite and Hastelloy N had expanded approximately 0.06 and
100

0.55 in. respectively. At the freezing temperature (850°F) of the

flush salt, the Hastelloy N was approximately 0.32 in. longer than

the graphite. During any heating or cooling of the assembly, the greater
expansion of the Hastelloy N tensile specimen rods requires that their
shoulders slide within the sleeves, which were spot-welded to the straps
that bind the graphite specimens.

On disassembly, it was found that most of the graphite tongue-and-
groove joints had separated during the heatup and allowed salt to become
entrapped in the joints. As the reactor cooled, the salt had frozen in
small rectangular platelets, as shown in Fig. 6.4. These ranged from
0.022 to 0,065 in. in thickness. Their typical thickness was 0.03 in.,
which means that the room-temperature length of each stringer of graphite
and salt was approximately 62.21 rather than the normal 62.00 in.

The third item found was that some of the sleeves were locked to the
shoulders of tensile specimens by frozen salt. The most strongly bonded
sleeves were at the ends of the severest damage,

PLATELET OF FROZEN SALT &

 

Fig. 6.4. Typical Platelets of Frozen Salt Trapped in the Tongue-
and-Groove Graphite Joints.
101

We conclude that the Hastelloy N tensile specimen rods sought to
return to their original room-temperature length during the reactor cool-
down, but they were obstructed by the sleeves that were attached to the
rods with frozen salt and by the increase in room-temperature length of
the graphite columns created by the frozen salt trapped in the joints.

The graphite specimens were end loaded, and they buckled and broke in

the manner of slender colums. Calculations based on 45,000 psi, the
lowest value, for the yield strength of Hastelloy N, showed that the

end loading would be sufficient to buckle such slender colummns of graphite,
assuming that the ends were fixed and the lcads were axial. As one might
expect, the thinnest specimens were broken more frequently than the thicker
ones.

We believe that the damage was unavoidably magnified during the re-
moval of the assembly from the perforated basket. We assume that a broken
piece of graphite became tilted in the region that showed the severest
damage, and this tilted piece of graphite acted as a yoke to bind the
assembly in the basket when approximately one-third of the assembly had
been withdrawn from the basket. The additional damage was probably created
by this yoke when the assembly was forcibly extracted the rest of the
way out of the basket.

Because many finished specimens were avalilable for use in a replace-
ment assembly and time was short, we decided not to make major revisions
in the design of the assembly. We did, however, make three minor modi-
fications to the basic design to alleviate these problems.

1. To avoid the entrapment of salt within the tongue-and-groove joints,
0,040-in.-diam pins of Hastelloy N were used to pin the joints so
that the graphite would not separate during the heatup.

2. To minimize the distances that the shoulders of the tensile rods must
glide within their sleeves, the middles of the graphite stringers were
pinned tc the middles of the tensile specimen rods. Since expansion
is now forced to work both ways from the pinned middles, the differ-
ential movements are halved. However, the maximum movement, for the
sleeves and shoulders farthest from the middle, is still as much as
0.2 in.

3. To minimize friction, misalignment, and salt entrapment, the sleeves
were shortened and split longitudinally. Compare those of Figs. 6.3a
and 6.5,

The weakest point in the modifications is that salt may freeze in
some of the annular spaces between sleeves and shoulders and that the
pins in the tongue-and-groove joints may shear.

The graphite specimens in the modified assembly were made from grade
CGB graphite that is inferior to the graphite in the first assembly in
that it has more cracks. The Hastelloy N tensile specimen rods of this
second assembly were made from the reactor vessel wall and head materials,
heat Nos. 5085 and 5065, respectively, plus two Hastelloy N alloys modi-
fied with 0.52 wt % Ti and 0.43 wt % Zr, heat Nos. 21545 and 21554 re-
spectively. The modified alloys are attempts to reduce the damaging
102

Y-75011

 

Fig. 6.5. ©Shortened and S1it Sleeves to Provide Better Salt Drainage
and to Alleviate Binding.

effects of irradiation. The specimens are included in the series in
order to study the effects of the alloying additions on damage by irra-
diation and corrosion by molten salt.

Flux monitors of 0.020-in.-diam wires of type 302 stainless steel
and pure iron and nickel had been sealed under vacuum in a protective
sheath of Hastelloy N as shown in Figs. 6.6 and 6.3a. In the postirra-
diation examination, the stainless steel and iron wires were readily
separated from the sheath, although the stainless steel showed some signs
of bonding to it. The nickel wire had solid-phase-bonded throughout its
length to the sheath.

The new set of flux monitors installed in the MSRE with the new re-
actor core specimens are of the same materials. However, €0 guard against
a repetition of the solid-phase bonding, the flux monitor wires and the
inside of the sheath were heat treated in air to produce thin oxide coat-
ings on them. To minimize vaporization of oxide, the sheath was sealed
with 1 atm of air at rocom temperature.
103

ORNL-DWG 66-11449

o

 
 
 
  
  

.100-in- M x i _ 1, -
0.100-in-DIAM x “g-in. LONG PLUG, INOR-8 g in. 1, in] I ‘
4 3, in.
8
TIG WELD 0.125-in-0Dx 0.020-in.-WALL INOR-8 &0 ‘l/ in i
i
R el 2T 777
AN —
ST SISO SIS I I IO, IRON NICKEL TYPE 302

 

STAINLESS STEEL

DETAIL OF END WELDS
0.020-in-DIAM FLUX MONITCR WIRES

0o Y% NOTE: WIRES ARE FOLDED BACK AT ONE END

SCALE' ! IINCH AS ILLUSTRATED ABOVE TO FACILITATE
THEIR IDENTIFICATION. THE BENT ENDS

ARE AT THE TOP OF THE SHEATH.

PINCH CLAMP MARK (PINCH CLAMP HELD VACUUM WITHIN
THE SHEATH DURING THE SEALING OF THE FINAL END WELD.)

 

 

 

 

 

 

 

 

/SHEATH
&= _'I:O
ToP
O [T = 3O
‘1/ ]«—13/ —J<— 60", * o 1y e
—T 8 8 a
et 62 —

 

 

¥ THE FLUX MONITORING WIRES EXTEND ALONG THIS SPACE OR DIMENSION.

Fig. 6.6. MSRE Flux Monitoring Wires and Protective Sheath.

6.2 Evaluation of Possible MSRE Radiator Tubing
Contamination with Aluminum

D. A, Canonico D, M. Haseltine

On Sunday, July 17, a cast Al-5 wt % Zn alloy blower failed at the
MSRE site, throwing small shrapnel across the Hastelloy N tubing of the
salt-to-air radiator. The failure occurred at about 10:00, while the
radiator was operating at 1070°F. The protective doors were closed while
the salt was still circulating; the temperature then rose to approxi-
mately 1200°F. This temperature was maintained until 14:30, at which
104

time the salt was drained and the heat was turned off. At 19:00, the
temperature had decreased to 250°F. The doors were opened and the tem-
perature dropped further to 150°F at 20:00.

A metallurgical investigation was conducted to determine the effect
of the aluminum-zinc alloy on the radiator tubing. Since the actual
tubing could not be removed and sectioned metallographically, only spec-
imens simulating the exposure could be prepared and evaluated. Small
segments of the failed blower were tested in compatibility experiments
with representative samples of the tubing.

Laboratory experiments were conducted to determine the influence of
temperature and time upon the Hastelloy-N—aluminum-alloy interaction. Tem-
peratures of 1150, 1175, 1190, 1200, 1225, and 1250°F were investigated,

Y-74029

1250°F

 

INCH

15 MINUTES AT TEMPERATURE
Fig., 6.7. Hastelloy N—Aluminum Alloy Compatibility Test Specimens.,
105

and times at temperature ranged up to 5 hr. Figure 6.7 shows the samples
after being held at temperature for 15 min. Figure 6.8 shows the samples
with the aluminum removed from the tubing. The refractory aluminum oxide
coating had contained the aluminum, even in the molten state, and inter-
getion did net cecur. The rounded ¢orners ipdicgte Lhat a lLiguld phase
was present at 1175°F and above; metallographic examination verified this.

When the oxide skin was broken from mechanical abrasion, shock, or
other reasons, wetting occurred. Figures 6.9a and 6.9 show the appear-
ance of a specimen where an angular drop of aluminum alloy had broken
through its skin and become sattached to the Hastelloy N tube. The spec-
imen was held for 5 hr at 1200°F. A metallographic section through the
cone and tube revealed the moderate interaction shown in Fig. 6.9c. At-
tack to a depth of approximately 0.010 in. had occurred.

Since these experiments indicated some possibility of penetration
of aluminum into the MSRE radiator tubes, they were inspected carefully
for both dents and attached shrapnel, The difficulty of this task can
be shown by Fig. 6.10, a photograph of the face of the radiator. The
close-packed spacing of the tubes made the use of mirrors and elaborate
lighting necessary. It is estimated that 80% of the surface area of
the tubing was examined.

Y-74028

   

1200°F

   

o2 0°F

    

  

g TR Bk

IOk 0 1 12B0°F

INCH
15 MINUTES AT TEMPERATURE

Fig. 6.8. Hastelloy N—Aluminum Alloy Compabibility Test Specimens.
106

Y-74022

 

(e}

Fig. 6.9. (a) Specimen Where an Angular Drop of Aluminum Had Broken
Through Its Skin and Become Attached to a Hastelloy-N Tube. (b) Closeup

of drop (10X); (c) metallographic section through cone and tube (50X).
As polished.

Eight smears of aluminum were found on the air-inlet side of the
radiator, none on the outlet side. No cone-shaped deposits of aluminum
were discovered, although one small angular piece was found wedged be-
tween a lower tube and a heater. Three dents were noticed in the outer
10%

HOTO 84553

 

Fig. 6.10. TFace of MSRE Radiator, Showing Close Packing of Tubes.

tubing on the air-inlet side, but no aluminum was detected 1in these
areas. It is thought that the dents resulted from small rocks, etc.,
which the blowers had drawn in at an earlier date.

The tubes on which aluminum was found were marked and subseguently
cleaned. Stainless steel wool was used to remove most of the aluminum
from a deposit area; then emery paper was employed. The radiator was
then brushed thoroughly to dislodge any pileces of aluminum that might
have been missed during inspection.

As a result of this investigation and cleanup procedure, it appears
that the radiator system is satisfactory for further operation. Actual
adherence of the aluminum to the tubing was observed in only a few cases.
Inspection and cleaning of accessible Tubes assured that aluminum was re-
moved from those most likely to be contaminated. If intimate contact
occurred in scattered areas, interaction dces not appear to be rapid.
Longer time effects are being studied by visual and metallographic
analysis of specimens from tests of duration up to 1000 hr.
108

6.3 Evaluation of Graphite

 

W. He Cook

Work has continued on the evaluwation of anisotroplc and isotropic
grades of graphite as potentilial materials for molten-salt breeder re-
actors.? The MSRE graphite properties are being used as a basis for
comparison. The needle-coke, anisotropic graphite has the more desir-
able properties for excluding moiten salts and gases but is less stable
under irradiation. Some isotropic materials show promise, but many of
the isotropic grades of graphite tend to be too amorphous for nuclear
use. Isotropic grades of graphite are new, and past development has
not been directed toward the production of material having the low gas -
permeabilities and small pore entrance diameters required of graphite
for molten-salt breeder reactors. :

Table 6.1 is a comparison of the physical properties of the various
grades of graphite. The specific resistances of these indicate that grades -
CGB, CGB-LB, 1425-64-1, and H-315 are fairly well graphitized; the rest
are too amorphous. For isotropic graphite, we desire specific resis-
tance less than 900 microhme em? cm™+. The gas permeabilities are all
very high. We are seeking permeabilities approaching 1077 cma/sec.

The pore entrance diameters are important in that (1) they must be
small enough to prevent penetration by the nonwetiting fluoride salts
and (2) they must be grouped to minimize the number of impregnations
necessary to fabricate the base stock into a low-permeability graphite.
For the latter, it has been reported that pore entrance diameters should
be less than 1 u and be concentrated in a narrow range.3’4 The grades
of graphite shown in Fig. 6.11 are finished grades rather than base
stocks, but only grades CGB, CGB-LB, and EP1924-1 appear amenable to
improvement by impregnation treatments. The third grade had been al-
ready ruled out because of its amorphous nature.

The standard salt screening test was made in which 0.500-in.-diam -
by 1.500-in.-long specimens were exposed for 100 hr to molten salt at
1300°F under a 150-psig pressure. The results are shown in Table 6.1.
The MSRE grades CGB and CGB-LB are included for reference purposes. Of
the new grades, only grade 1425-64-1 showed a pore structure approaching
the quality of the MSRE graphite.

Work is in progress to obtain additional anisotropic and isotropic
grades of graphite with properties superior to those of the MSRE graphite
and approaching those required for the molten-salt breeder reactors.
Table 6,1. Ccmparison of Varicus Physical Properties of Anisotropic Needle-Coke and Isotropic Grades of Craphite

 

Specific Resistance Bulk Volunme

 

Cross~-Section Buik Helium Accesgible Surface . 2 - Permeability of Graphite
; . . \ . microhms cm® cxm A -
Grade Shape Dimensions Type Density  Density Poros;ty Area to Helium Pilled

(in.) (g/cn?) (gfem®) (ch )2 (2 /g )8 l!b e (cre? /fsec) Wi%h)galt

- %
X 104

CGB Bar 2 X2 Needle-coke 1.86 2.03 12.7 0.462 610 1200 3 0.2

CGR-LB Bar 1X 1.6 Needle-coke  1.86 2.12 12.4 595 1100 0.4 (c)wéc:"/—
0.8
1425-64-1 Pipe 3.6 0D X 2.5 ID Neecdle-coke 1.82 1.91 9.2 0.171 690 1215 4500 0.5
H-315A Pipe 4.7 00D X 3.5 1@ Isobropic 1.83 2.04 11.7 0,165 820 980 89 7.2
EP-1924-1 Cylinder 4.1 diam Isotropic 1.80 2.14 5.8 0.332 1385 2600 12.5
2020 Block 4.0 X 4.1 Isotropic 1.71 2.02 17.2 0.316 1760 1660 240G 14.6
HCTE-Y12  Cylinder 7.3 diam Isotropic 1.88 1.95 loste 0,046 13C0 1810 9460 4.0

 

She measurements were made on 0.250-in,-diam X 1.C00-in.-long specimens.

Measured in the direction of the ap axes for needle-coke, anisctropic grapnite; no specific directions indicated for isotropic graphite ex-
cept that measurements were mutually perpendicular.

Measured in the direction perpendicular to the ag axes for needle-coke, anisovropic grapbite; no specific direction indicated for isotropic
graphate except that two measurcments were made that were mutuaily perpendicular,

Evacuated specimens, 0.500 in, diam X 1.500 in. long, exposed for 100 ar to molten salt at 1300°F and a pressure of 150 psig; a standsrd
geresning Legt,

Tmoregnation was not uniform; the valunes in parentheses indicate the range observed.

60T
110

ORNL-DWG 66~ H450

5o

e L el e L]
Z ”CGB—LB | ) _:"_ 4T‘* I l:] NJ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

28 -
LT
= (CGB ™ i
d ——— .- - .
Qo 10[_; -1 T ][ : - i’:,:
s ! |
2 e B
g 2020 i |
v o b =T - L
5} —_— _.-.1””7 cmiaaae
20— — — — — —
- .
sl
= EPI924-1 |
E_:J 0 el
E {eh """" r .
‘ |
| |
6 — — — — 44#4‘7 —
1425-64-19 ] ‘
o T T ] . ,
209 8.0 1.0 030 003 coz ot

PORE ENTRANCE DIAMETER ()

Fig,., 6.11. Comparison of the Distributions cof the Pore Entrance
Diameters Tor Various Grades of Graphite.

6.4 The Internal Stress Problem in Graphite Moderator Blocks

 

C. R. Kennedy

One of the major conceras in the use of graphite-moderated reactors
is the stress generation caused by differential growth. The stress gen-
eration is moderated considerably and generally maintained at safe levels
by the ability of the graphite to ereep under drradiation. The problem
resolves itself {0 a determination of the balance hetween the differential
growth rate that produces stress and the creep that reduces stress. Al-
though the creep-~rate coefficient may be such that the gtress level does
not exceed the fracture strength, a second concern is the ability of the
graphite toc absorb the creep strain indefinltely without failure. Cer-
tainly, a fallure criterion based upon the fracture stress of the mate-~
rial is accurate., Therefore, it is of great importance Lo know both the
restrained growbh rate and the creep coefficient for the material under
the operabing conditions. Our purpose has been {o determine the general
creep behavior of graphite under irradiation.

Creep experlments were performed at 700 and 1000°C for comparison
with the previous data’ obtained at lower temperatures. The high-tem-
perature experiments weve again similar to the previous experiments in
111

ORNL-DWG 66~11451

 

2 ‘ :
" SINGLE SPECIMEN }
/

-4

(vt )

 

4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 r— —
Fig. 6.12. Effect of Temper- 2 s ]
ature on the Creep Coefficient of ¢ f
Graphite Under Irradiation. b T
Ll
o 2 1
[
L .
i .
o i ‘
o 1027 G
a v Sl v AGOT -
w -m%_------k—f—# — © EGCR-TYPE AGOT—
S iifl‘“ VALUES | 7771 a coB o
5 ,,,,,,,,,,,,,,,,, d s JA
0 200 400 600 80C 1000 1200 1400

TEMPERATURE (°C})

that cantilevered parabolic-beam specimens were used. The main differ-
ence was in the uge of four-zone furnaces to obtaln the desired temper-
atures. The number of specimens was reduced from nine to six because
of the space requirements of the furnsces.

Results from all creep experiments performed from 150 to 1000°C do
demonstrate a generalized creep behavior, as shown in Fig. 6.12. This
type of behavior strongly supports a Cottrell model for irradiation
creep, which allows extrapolation of these data to most reactor grades
of graphite. The Cottrell model for irradiation creep, as given by
Anderson and Bishop,6 is

K = Ajfo_ (1)
y
where
K = creep coefficient,
A = accommodation factor,
¥ = ghear rate due to anisotropic growth (GC *'Ga)’
0&_: yvield strength of the crystallites,

GC and Ga are the growth rates in the ¢ and a directions.

The shear rate ¥ of graphite can be derived from measurements made
on pyrolytic graphites through estimates of polycrystalline graphite
growbh rates. The accommodation factor A primarily reflects the degree
of accommodation by microecracks and, in general, the vold volume in the
graphite. It is not necessarily constant and is expected to vary with
temperature and neutron exposure. The variation of A with temperature
should not be very large; however, A should exhibit a rather significant
increase as microcrack closure occurs. The closure of microcracks re-
gquires a dose of approximately 1022 neutrons/émz; thus the value of A
will be essentially constant for at least half this dose. The value of
112

the yield strength or flow stress, 0, of the crystallites will undoubt-
edly vary under Iirradiation and with irradiation temperature. The value
of 0y, like that of A, will not vary with temperature, however, as sig-
nificantly as ¥, the shear rate.

The creep-rate coefficient should therefore strongly reflect the
variation of ¥ with temperature. This is demonstrated in Fig. 6.12,
where the creep~rate coefficient exhibits a minimum around 350°C. The
creep-rate coefficient K is not exaclly proportional to the shear rate
¥ because of the variations of A and Oy. The value of A would be es-
sentially independent of the graphite grade: thus the effect of Gy on
the creep-rate coefficient can be demonstrated by a comparison of var-
ious grades at the same temperature, This was demonstrated previouslys
by comparing the modulus of elasticity of six grades of graphite to their
creep-rate coefficients at 400°C. Although this correlation strongly
supports £q. (l), or the Cottrell model, for the crecp of graphite, one
glaring discrepancy in the correlation is the test result at 1000°C for
the CGB grade of graphite. It should be noted that this particular ma-
terial demonstrated a one-~third decrease in its modulus of elasticity,
which is also unlike the behavior of previously tested grades. This
particular test result requires confirming data before specific conclu-
sions can be reached.

One should recognize that the Cottrell model for creep does not sug-
gest an actual mechanism by which the graphite deforms plastically. It
does describe the creep as a process of continuous yilelding, which is
not thermally activated as the term creep generally implies. Also, the
stress-induced jmbalance of the internal straining occurring in all poly-
crystalline graphites due to anisotropic growih. This implies that the
limit of creep deformation can be as large as the internal straing that
must occur in polycrystalline graphites irradiated without stress. Poly-
crystalline graphite can accommodate 16% shear strain and polycrystalline
carbons 160% shear strain’ without a loss of mechanical integrity. The
obvious conclusion is that as long as the stress acting on the graphite
does not exceed the fracture stress, the graphite will continue to ab-
sorb the creep deformation without loss of mechanical integrity.

6.5 Brazing of Graphite

 

J. M. Jones We J. Werner

The jolning of graphite to structural metals such as Hastelloy N
is of prime interest in advanced molten-salt reactor concepts. Studies
are under way Lo develop methods for joining large graphite pipes to
Hastelloy N tube sheets. OSuch joints will be needed for test assemblies
and for the reactor core.

The current studies have two primary objectives: (1) to develop a
corrosicn-resistant brazing alloy for graphite which does not suffer from
113

the transmutation problem associated with the gold-contalning alloys and
\2) to develop means for making transition joints with one or more ma-
terials having expansion coefficients intermediate between those of the
graphite and the Hastelloy N.

Work was focused on the evaluation of an experimental precious-metal-
base alloy having the composition 60 Pd—35 Ni~5 Cr (wt %). This glloy
appears to have some application for Jjolning graphite to the refractory-
metal portion of a transition piece. Unfortunately, it exhibits rela-
tively poor flowability on high-density graphite (Fig. 6.13a); however,
its marginal behavior 1s enhanced by preplacing it as foil in the Jjoint
(Fig. 6.13b). Photomicrographs of such a joint at low and at high mag-
nification are shown in Figs. 6.14a and 6.14b. The needle-like carbides
of Fig. 6.1l4a are clearly evident in Fig. 6.14b. The extensive diffu-
sion zone along the molybdenum—razing-alloy interface in Fig. 6.l4a
suvggests that considerable molybdenum was taken into solution in the
brazing alloy.

A series of graphite-to-molybdenum Jjoints brazed with this palladium-
nickel-chromium alloy preplaced in the joint were thermally cycled ten
times between 200 and 700°C. Metallographic investigation indicated that
no deterioration had oceurred.

omall arc-melted buttons of other alloys in this ternary system,
but with higher chromium contents, are being prepared in an effort to
improve wetting of graphite without reducing the overall joint proper-
ties. Alloys containing other corrosion-resistant carbide formers, that
is, nicbium and molybdenum, are also being prepared.

ITwo graphite-to-molybdenum-to-Hastelloy-N transition joints were
successfully brazed using the tapered joint design reported previously.8
The 1-1/4-in.-0D by 3/1l6-in.-wall graphite tubing was joined to the mo-
lybdenum with the 60 PA—35 Ni—5 Cr (wt %) alloy preplaced in the Jjoint;
copper was used to braze the molybdenum to the Hastelloy N. Visual ex-
amination revealed no cracks, and metallographic examination will be used
to further evaluste the Jolits.

Y-75419

 

denum Jointe Brazed with 60 Pd—35 0 1
Ni—-5 Cr (wt %) at 1250°C in Vacuum. Lok o

Y-75420

   
1l

 

 

. 0385 INCHES
ra \OO)(

T
|

 

 

 

 

 

 

Fig. 6.14. (a) Graphite-to-Molybdenum Joint Made with 60 Pd—35 Ni—5
Cr Alloy Preplaced in Joint. Good bonding is evident. Etch: 10% oxalic
acids (b) View of area enclosed in (a). The large carbide needles are
evident. Etch: 10% oxalic acid.
115

6.6 Corrosion of Graphite-to-Metal Brazed Joints

W. H, Cock

As was discussed in the previous section, some success has been
obtained in joining small pieces of graphite to metal by brazing the
graphite to molybdenum and, in turn, brazing the molybdenum to Has-
telloy N.%'L0 TIn addition to making a good joint, the braze must be
resistant to corrosion by molten fluoride salts., Corrcsion of the
brazing alloy is being investigated by exposing Jjoints of grade CGB
graphite bragzed toc molybdenum to static LiF-BelF,-Z2r¥,-ThF,-UF, salts
for 100, 1000, 5000, 10,000, and 20,000 hr at 1300°F in Hastelloy N
capsules. The brazing alloy, 35 Ni—60 Pd—5 Cr (wt %), was not at-
tacked by the salt during exposures for as long as 5000 hr. However,
layers of metallic-like crystals appeared preferentially on the braz-
ing alloy in the 1000- and 5000-hr tests. The 10,000- and 20,000-hr
exposures are still in progress.

The method of fabrication of test specimens is shown in Fig. 6.15.
The lateral surfaces of the machined pieces were polished prior to cut-
ting the pieces into 0.,62-in.-long specimens. The microstructures of
the left and right edges of the sides (the sides are perpendicular to
the plane shown) of an untested braze in Fig. 6.16a show that this pro-
duces straight, smooth sides suitable for references for determining the

ORNL-DWG 66-11452

CGB GRAPHITE

 

ALL DIMENSIONS ARE IN INCHES.

Fig. 6.15. Grade CGB Graphite Brazed to Molybdenum with 35 Ni—60
Pd=5 Cr (wt %). (a) As brazed; (b) as machined into three test specimens.
The lateral surfaces are polished.
116

GRAPHITE

s <

 
 
 

0.030 INCH

BRAZING ALLOY

 

(a) O hr (5) 100 hr (c) 1000 hr (d) 5000 hr

Fig. 6.16. Microstructures of the 35 Ni—65 Pd—5 Cr (wt %) Brazing
Alloy Used to Join Grade CGB Graphite to Molybdenum. (a) As brazed and
(b, ¢, d) After Exposures to LiF-BeF,-ZrF,-ThF,-UF,; (70-23.6-5-1-0.4
mole %) at 1300°F. The graphite was tilted during the brazing which pro-
duced the different thicknesses of braze in the sets of photomicrographs
that show both edges of a specimen. Etch: 10% oxalic acid. 100x.

Y-75395 7]

]|

 

HON| LCOC

53

 

 

 

Fig. 6.17. Microstructure of a Metallic-Like Deposit on the Graphite
of the Graphite-Molybdenum Joint Exposed for 5000 hr to LiF-BeF,-ZrF, -
Th.FA"UFé_ at lBOOoF .
117

extent of attack. Tor the metallographic examination, this specimen
and all others were cut into halves perpendicular to their 0.62-in.
dimension, and these cut surfaces were polished and photographed. This
technigque was used to compare the control specimen with those tested
for 100, 1000, and 5000 hr (Fig. 6.16). There is no microscopic evi-
dence of attack on the braze. Deposits approximately 0.5 and 1 mil
thick are evident on the sides of the brazes exposed for 10C0 and 5000
hr respectively. These layers are crystalline and metallic 1in appear-
ance. They have not been identified. The chemical analyses of the
salt from the 1000-hr test did not detect any corrosion products; the
analyses for the 5000-hr test are in progress.

Interpretation of the 5000-hr test is further complicated by a thin
metallic-like deposit on the graphite. It is unlike the layers on the
braze alloy in that it does not exhibit crystalline faces. The deposit
was heaviest on the slde of the graphite parallel with the brazed Jjoint.
Its microstructure is shown in Fig. 6.17. The identification of this
layer 1s being sought. It is suspected that this may be CriCy. In
previous tests that involved only salt, Hastelloy N, and graphite, there
were no deposits on the graphite. The longer-term tests appear necessary
to explain the cause of these deposits.

6.7 Welding Development of Hastelloy N

H. E. McCoy D. A, Canonico

Titanium and zirconium additions to Hastelloy N appear to improve
the resistance of the alloy to high-temperature embrittlement in an ir-
radiation field.

We have made welds in several experimental heats to evaluate the
influence of these alloy additions on the weldability. Welds have been
made in 1/2-in. plates of Ni—12 Mo—7 Cr-0.05 ¢ (wt %) with titanium
contents of 0,15, 0.27, 0.33, 0.45, and 0.55 wt %, These welds appear
to he sound, and test specimens have been made for evaluation.

Alloys containing Ni-12 Mo—7 Cr—0.05 C (wt %) and either 0.06 or
0.43 wt % zirconium have also been studied. The alloy containing 0.06
wt % zirconium exhibited extensive weld cracking. It was impossible to
get a complete pass in the alloy containing 0.43% zirconium without a
crack propagating the entire length of the pass. Some attempts have
been made to weld these heats with dissimilar weld metal.

Effect of Irradiation on the Mechanical Properties of Hastelloy N —
H. E. McCoy

The foremost objective of this effort has been to determine the in-
fluence of neutron irradiation on the mechanical properties of the Has-
telloy N used in constructing the MSRE. We have utilized postirradiation
tensile tests, in-reactor creep-rupture tests, and postirradiation creep-
rupture tests to evaluate the performance of several representative heats
of material used in the MSRE. Portions of this study have been reported

: 11-14 . . o .
previously and have been used in setting a minimum safe operating
ORNL-DWG 66-5779R

I NH I 11T

 

70

 

60

 

 

 

 

~ {
UNIRRADlAT ED
lrl

o
o
/

 

 

 

 

 

 

 

 

 

 

 

 

STRESS (1000 psi)
W S
Qo O
! !
( . I

b ""_
o
4
-

/¢
(%]

o | +

: —

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1

20 L <1> 6X1O n \H-: | \\1: i

| i 1-51\3‘—

! H 2.4%%' ;
10 ; | | (\“

|
0 L
10° 2 5 0 2 5 102 2 5  10° 2 5  40°

RUPTURE LIFE (hr)

Fig. 6.18. Creep-Rupture Properties of Hastelloy N at 650°C, Heat
5065. Numbers indicate fracture elongations.

life for the MSRE. One of the most interesting findings in this study

is illustrated by the data in Fig. 6.18. Although the rupture life and
ductility are both reduced by irradiation, the data indicate that there
may be a stress below which essentislly no irradiation damage occurs.

One of the current theories of high-temperature irradiation damage
predicts this type of behavior, since some minimum stress is requlred
to allow the bubbles of transmuted helium to grow without limit. 15 The
existence of a stress below which one could design without concern for
gross reductions in ductility due to neutron damage is of extreme im-
portance and needs to be studied further.

The surveillance specimens were removed from the core of the MSRE
after 7800 Mwhr with an effective integrated dose of about 7 X 10%°,
So far as neutron dose is concerned, these specimens should be repre-
sentative of the pressure vessel after the reactor has operated about
150,000 Mwhr (eguivalent to 15,000 hr at a power level of 10 Mw). The
specimens have been disassembled, but testing will not begin until after
examination with an optical comparator to separate the bent cnes. Limited
creep and tensile tests will be run to determine whether the properties
fall in line with those determined previously for material irradiasted to
the same dose.

The second objective of this program has been to understand the
mechanism of the damage in Hastelloy N and to determine a means of
making a basic Improvement in its resistance to irradiation damage.

We have found that the damage is dependent on the thermal-neutron flux.
This is illustrated by the data in Table 6.2. Specimens were exposed
to neutron enviromments of various energles, and it was found that

the damage was fairly independent of fast dose and depended primarily on
the thermal dose. It is believed that the specific thermal-neutron re-
action is that of the transmutation of °B to helium.
119

Table 6.2. Tensile Ductility of Hastelloy N (Heat 5065) at 760°C
& = 0,002 min™*t

 

 

o, =1 X 108 nvt, 0., =1 x10%% ave, o =5 X 10 nvt,
éf = 5 % 10 nvt ¢f =1 X 10*8 nvt ¢f =1 X 10 nvt
7.1 6.2 12.0

 

Hence, the first approach Lo the problem has been that of reducing
the boron level. Figure 6.19 shows the postirradiation creep-rupture
properties of several air-melted heats of Hastelloy N. 'The materials
contain between 20 and 40 ppm boron. The rupture ductilities are given
in parentheses. There seems to be little effect of irradiation temper-
alure on the rupbure life or ductility. Figure ©.20 shows the proper-
ties of two vacuum-melted heats containing 7 ppm (2477) and 9 ppi horon.
(65-552). When the materials are irradiasted at a low temperature, theilr
properties excel those of the alr-melted heats. When these same heats
are lrradiasted hot, they exhibil properties comparabvle with those of the
alr-melted heats.

ORNL-DWG 86-11453

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

70 - T T T T T —
- L | ] N 1T Tfl
80— \\%Sjnidmr(bu)“ T N
50_ 1 1 1 -1
‘ - TYPICAL
C ELTED
— 40 — Fwmwos - o e N
2 [ ‘ e
U’) v
W . !
E i ‘ (2.0
= i
» 30 |— uig | -
ALLCYING
HEAT NO. ELEMENT
o 21541 W
® 24548 - ANNEALED 100 hr
20 1421542 W AND Nb | AT 874°C PRIOR TO [ = S e e e
A 21545 Ti IRRADIATION
021543 Nb
®65-552 — WORKED AT 871°C
o || IRRADIATED AT €50°C B o
<1>m=3.5x1020 vt
OL _____ | m I _,L\ i m \J \ L
o 1 10 100 10,000

 

RUPTURE LIFE (hr)

Fig. 6.19. Postirradiation Creep of Several Alloys at 650°C. Num-
bers indicate fracture elongations.
120

ORNL-DWG 66-10882

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  
   

 

 

 

    
  

 

 

 

 

 

 

  

 

 

 

 

o T T T T 1T
-]CLOSED POINTS: COLD (RRADIATION
OPEN POINTS: HOT IRRADIATION
| 1] 1 L
60 lm 2 —— - i
! |
] //,UNRRADMTED PROPERTIES
i |
T < J\”
_ 40_#‘ | W N J‘
; | : 2.2 |31 _L
Q ! | . o | ¢ s ™
T 30 - e o 34 3.2
v f { !
il |
20 —— DOSE IRRADIATION e !
HEAT NO.  ®,,=  TEMPERATURE (°C) J . ’-
o 2477 5x 1020 43}‘ | ‘
10l ¢ 65-552  5x1020 150 i | ]
A 2477 251020 650 | |
O 65-552 2.6x10°° 650 | | ANNEALED 1hr AT 4477 °C PRIOR TO | |
¢ 2477 5 %1020 150 | | IRRADIATION |‘ ! | !
0 e vl | LU
1 10 100 1000 10,000
RUPTURE LIFE (hr)
Fig. 6.20. Postirradiation Creep-Rupture Properties of Several

Vacuum-Melted Heats of Hastelloy N at 650°C.
elongations.

Numbers indicate fracture

Several of the alloys were irradiated Lo various doses, and the
helium content was calculated from the dose and the original boron con-
tent of the alloy. Figure 6.2l shows the rupture ductility in a tensile
test at 650°C and 0.002 in./min as a function of helium content. There
is considerable deviation about the line drawn, with some of it probably
being significant. The two vacuwm-melted heats (65-552 and 2477) again
show lower ductilities when irradiated hot. The two Llow points for heat
4065 indicate a possible influence of high irradiation temperature on
this heat. However, the most important observation is that, for a rea-
sonsble *°B content and a reasonable lifetime, the helium content will
be between 1072 and 107% atom fraction. The variation in ductility over
this range of helium contents 1is only aboub 25%, agsuming that the line
drawn 1s reasonably correct.

A similar plot is shown in Fig. 6.22 which relates the helium pro-
duced in the alloy to the fracture dvctility. The line drawn seems 4o
be representative of the higher-boron air-melted heats (5065, ete.).
The vacwm-melted low-boron heats (65-552 and 2477) exhibit better duc-
tility if they are irradiated cold but actually have lower ductility
when irradiated hot. Hence, the role of boron in these alloys is not
at all clear., We have not prepared a serles of alloys where everything
elge has been held constant except the boron content. 1L seems that
the variations 1in composition and melting practice must influence the
ORNL-DWG 66-11454

 

 

 

        
  
 

 

 

 

LT

ETR 41-30
BSR
ETR 41-3f
ORR -149
ORR -155
ORR-153
ANNEALED 1 hr AT 1177 °C
PRIOR TO IRRADIATION ’
TESTED AT 650 °C, 0.002 min~" : |
~ GLOSED POINTS: COLD IRRADIATION “ \ 1 i
| \ | i

 

 

 

 

 

TOTAL ELONGATION (%)

&

 

 

 

 

 

 

 

 

 

OPEN POINTS: HOT IRRADIATICN

el

1078

 

 

 

 

 

 

 

 

 

 

ol
107 ?

 

1077
He CONTENT (otom fraction)}

Fig. 6.21. Variation of the Postirradiation Tensile Properties of
Jeveral Heabs of Hastelloy N with Helium Content.

ORNL-DWG &6-10880

 

 

 

 

 

 

 

 

 

 

Fig

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

14
I \ T T T
{ CLOSED POINTS: COLD IRRADIATION | A ORR-153
| OPEN POINTS: HOT IRRADIATION 0 BSR
[
ey L B
: i 2477 ORR-141
UNIRRADIATED v2d .
: ORR-148
DUCTILITY
N M | ANNEALED 1 hr AT
10 poormrmeto R 4 1H 4177 °C PRIOR TO-
l } IRRADIATION
_ \ ~\ L || TESTED AT 650 °C
Hi
> 8 \ kwl WI” .- L 1‘ ‘ ‘_&*
S
5 T
2 2477 Ny 5065
C ol AEPY ‘ 324 | L
) | |
5065 .
alo— L. i 400 \hl\
\
‘ ’ ‘ 65-552 \lg
| 3247
it LL : ||
2 | ‘ | | 324
\ :
i ‘ ‘: { 4q -.4 \
o - l
1078 1" -
He CONTENT ({atom fraction)

O.22.

Variation of Postirradiation Creep Ductility of Hastelloy

N with Helium Content. Numbers indicate stress in ksi.
122

distribution of boron or the effect that the transmuted helium has on
the properties so greatly that the role of boron per se is masked. The
great effect of irradiation temperature on the postirradiation proper-
ties of the vacuum-melted alloys supports this supposition.

Although we have known for some time that the ductility of an ir-
radiated test specimen varies greatly with temperature, we have made
extensive use of postirradiation tensile tests for screening purposes
even when we were interested in long-term applications. Figure 6.23
showeg how misleading this approach can be.

The rupbture ductility in postirradiation tensile and creep tegtse
is plotted to illustrate the influence of zirconium content on the
properties. The two sets of points at the left were obtained by tensile
tests at two different strain rates (0.05 and 0,002 in./min). The trend
is evident that the ductility improves with increasing zirconium content.
However, the duetility in the creep tests (represented.by*the polints at
the right) shows no systematic dependence on the zirconium content. This
material contained extremely high boron (approximately 200 ppm), and we
believe that the data do not warrant the conclusion that zirconium has
no beneficial influence.

The important point is that tensile tests may not be adequate even
for screening purposes in nickel-base alloys. The more expensive creep
test may be necessary.

ORNL—DWG &65--10885

 

 

 
  
 
  
   
 
 
 
  
 

 

 

 

 

 

 

 

 

T I | T TETTT [
| E? Zriwt %)
® — ° 0.05 T
. 013
{ [l:2
18 - Lgd | 1| & 0.43 | REMELTS OF HEAT .|
g o 0.52 NO. 5065 ’
* " 08 ‘
1 v 44
12 | — . Ty 1.2
@, =2x1020 A/
. IRRAD TEMPERATURE = 650°C
1 L T

 

 

P TEST TEMF’ERATUREZGSO"C - .

 

 

 

 

 

      

ELONGATION (%)
@

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RUPTURE LIFE (hr)

Fig. 6.23. Influence of Zirconium Content on the Postirradiation
Ductility of Hastelloy N.
123

Several small commercial melts have been procured and evaluated.
The basic alloy is Ni—12 Mo~7 Cr—0.05 C, the reduction in molybdenum
content being made to suppress second-phase formation. The alloy addi-
tions investigated include nicbium, tungsten, and titanium. To date,
these alloys have been investigated in only one metallurgical state, 40%
cold working followed by 100 hr at 871°C. This treatment produces a very
fine grain size and probably does not result in the optimum properties.
Figure 6.24 shows the postirradiation creep properties of several of
these alloys at 650°C. The very encouraging alloy is No. 21545, which
contains 0.5 wt % titanium. The rupture life is better than the other
alloys, but the rupture ductility is far superior.

Figure 6.25 shows a compilation of all the in-reactor creep tests
run on various heats. All the data are contained in a scatter band
bounded by heats 5065 and 5085. With few exceptions, all the heats ex-
hibit fracture ductilities of 1 to 3%, with no systematic heat-to-heat
variation evident. Three of the heats lend support to the idea that
there may be some critical stress below which irradiation damage be-
comes minimal. One must face the question of why heat 21545 locks ex-
cepbionally good in postirradiation creep tests (Fig. 6.24) and not in
in-reactor creep bests (Fig. 6.25). With a closer examination, one sees
that the test at 21,000 psi (Fig. 6.25) did not fail during the experi-
ment. This point indicates that the threshold stress for damage 1n this
alloy may be higher than that for the other alloys (approximately 15,000
psi)., Thus the in-reactor creep tests do suggest that the titanium-
bearing alloy may be superior.

ORNL-DWG 66-10879

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 ‘ ‘
° N | T T T T T T 1T
e ‘ ! ALLOYING
AN HEAT NO. ELEMENT
14 |- : o 21541 w 4
N a 21546 — .
o 21542 W AND Nb g;l;{g%g%éogom Al
\ ¢ 21545 Ti {RRADIATION
12 T N T A 21543 Nb
\ \\‘ \, a 65-552 - WORKED AT 871°C
N N
10 7\ N .. IRRADIATED AT 650°C i
E ;§\ \ N Gy = 3.5 X AT vt
z \ % ™ l
S 4 N N
= NG T —t i
< N A
2 N ™ fi\\ 21545
O i N | ‘
d 6 ___B.u. \ \3‘ “ 1. \"""-- ]
\\ N \4\\~ N ‘
N \ TH\\ . |
4 - \\ \J\ N e, \‘ ______
\fi \ \\\\\ /
~ % e
™ | \ o /
65-552 ST X
! 1 \ A | - Nl
LT g
o AL o LI
oo 04 4 10 100

RUPTURE TIME thr}

Fig. 6.24. Postirradiation Ductility of Several Alloys at ©50°C.
124

ORNL-DWG 66-11473

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

  

 

 

 

 

70 . - -
TNea T T | N
| \\ ' '
60 1 | E S - e
TYPICAL EX-PILE P RT
\\\\ /////,, ‘ 1 }ROPER IES
\.<"“
50 bomrm e L . 1]
\\\
&
Qo !
o 40 -~ -
o
e
= !
@ \\ta.zl:: \
30 (13)+,N- ---------- ‘
U'_;. (10 ‘ \
‘ | -
ALLOYING | H Hotedsin .9 \\ (125
201 HEAT NO. ELEMENT | 1 e {22 gzfinm—mqm ot
o 2065 _— fhr AT 1177 °C LR N
® 5085 e ‘ (1‘0‘)‘ (2.8)N(4.6) e \\
o 7304 e fhr AT 4177 °C+8hr AT 871 °C o s A
s 51245 VAP VL roonat 871 oc | o
|

 

10 |- & 21545 ! |
‘ l ‘ IRRADIATED AT 850 °C r ‘

 

 

 

Kfim= 6x10"3 v |
RUPTURE DUCTILITY IN ( )
L J L I |

 

 

 

 

 

 

 

 

 

 

 

 

. |

1 10 100 1000 10000
RUPTURE LIFE (hr)

 

Fig. 6.25. Comparison of In- and Ex-Pile Creep Rupture Properties of
Hastelloy N at 650°C.

ORNL-DWG 66-10878

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1800 (- *T—— ] | '
e [ e i i
1600 | I ‘ s 53
14900 1 — Ll L #100 l
58 ’ ’
1200 |— | |
£ , ‘
W 1000 L | ANNEALED 1 hr AT 4{77°C
L PRIOR TO IRRADIATION
Lyt
% 800 [#12.5 —— —— IRRADIATION TEMPERATURE=650°C-
£ UNIRRADIATED, Dy = 2.5 X 1070 s
5 STD HASTELLOY N
™ 600 - o572 | TEST TEMPERATURE=650°C T
STRESS=32,350 psi
400 { | J
200 | g o —— P
o4 ‘ ’
O ‘ | ,,J ........... L ,,,v....L,,,,,,,..._L..._______...‘.,,i

O ©Ot 02 03 04 05 06 07 08 09 10
Ti CONTENT (wi %)

Fige. ©.26. Influence of Titanium on the Postirradiation Creep Prop-
erties of Ni—12 Mo—7 Cr-0.05 C. Numbers indicate fracture elongations.
125

Figure 6.26 shows the results of postirradiation creep tests on sev-
eral laboratory heats containing various amounts of titanium., All the
alloys are superior to irradiated standard Hastelloy N with respect to
both rupture life and ductility. Several of the alloys exhibit rupture
lives in excess of that of uwnirradiated Hastelloy N and ductilities as
high as 10%.

Characherization of Hastelloy N for Service at 982°C - H. E. McCoy

 

Since Hastelloy N is being consldered as the structural material
for a molten-salt distillation vessel, the Mechanical Properties Group
wag asked to determine the creep properties of this alloy at 982°C.
The results of this brief study are presented, and several potential
problems associated with the use of Hastelloy N ab such an elevated
temperature are polnted oub. Although the vessel presently being built
will be used only with nonactive salts, it is anticipated that a similar
distillation apparatus will be an integral part of a Molten-Salt Breeder
Reactor. For this reason the questions concerning the use of Hastelloy N
for this application should ve resolved.,

The creep~rupbure properties of Hastelloy N are well documented® 7
over the temperature range of 600 to 800°C where it is commonly used.
However, data at 982°C were nonexistent, since this is above the normal
service temperature for this alloy. The results of a brief creep-rupture
program undertaken to supply data at 982°C are shown in Table 6.3, Plots
of these game data are shown in Figs. 6.27 to 6.29. These data were ob-
tained on test specimens of a typical heat of air-melted Hastelloy N (heat
5065). The test specimens were small rods having a gage section 1 in.
long X 0.125 in. in diameter. All tests were carried out in dead-weight
creep machines in air. The creep properties shown in Fig. 6.27 indicate
that the stresses to produce rupture and 5% strain in 1000 hr are rea-
sonably well defined. However, the stresses for 1 and 2% strain in 1000
hr are not defined sufficiently by experimental data. The minimum creep
rate data in Fig. 6.28 correlate very well, but these numbers are not
very useful for design purposes, since the minimum creep rate only applies
during a small fraction of the life at a gilven stress. The plot of the
Aductility shown in Fig. €.29 exhibits a ductility minimum for test spec-
imens having rupture lives over the approximate range of 50 to 200 hr.
However, the minimum ductility observed still is slightly in excess of
20%. The reduction in area decreases rapldly with increasing rupture
life to a value of about 15%. The increase in the elongation for long
rupture lives is thought to be associated with the very extensive in-
tergranular cracking that cccurs.

The specimen from test 5768 was examined metallographically. The
extent of intergranular cracking is illustrated by Fig. 6.30. It was
also noted that the quantity of precipitate present after testing was
much greater than that present before testing. Filgure 6.31 shows a
typical area of this heat of Hastelloy N prior to testing. Tigure 6.32
shows the tested specimen. The precipitate marked "A" is believed to
be that present in the starting material. The larger precipitate, marked
"B, " is thought to be induced by the thermal and mechanical history.
Table 6.3, Creep Properties of Hastelloy N at agz°c®

 

Time to Indicated Strain (hr)

 

 

o G Che ) Teme
: 1% 2% 5% 20% Rupture
5768 2,000 35 73 185 495 649 0.0276  41.50 16.22
5765° 3,000 1 7.6 33 117 123.9 0.123 22.36 13.4
5762° 4,000 0.7 2.0 11 47 51.5 0.282 21.88 15.47
5761 6,000 0.75 1.3 3.2 10.5 16.15 1.65 48 . 4ds 36.57
5763 10,000  <0.1 0.2 0.5 1.9 2.95 9.0 50.0 37.86
5867 1,880 35 72 165 407 520.8 0.026 52.0 15.6
5868 3,000 12 25 60 154, 157.9 0.081 22.22 15.6

91

 

“Heat 5065 tested in the as-received condition (1/2 hr at 1176°C mill annesl).
Temperature control not adeguate at start of test.
127

ORNL-DWG 66- 11455
20,00G |-y - ]H\

    

HASTELLOY N
HEAT 5063

 
    

10,000

5000 Fig. 6.27. Creep-Rupture Prop-

------ T erties of Hastelloy N at 982°C.

STRESS ( psi )

   

2000 ;
o RUPTURE

  

1000

 

 

 

 

 

 

O.1 i 10 100 1000
TIME {hr)

ORNL-DWG 6611458

 

10,000
" HASTELLOY N
HEAT 5065
- 982°¢C
5000 |- L L.
e . o . a
Fig. 6.28. Minimum Creep Rate =
w
vs Stress for Hastelloy N at 982°C., W
>
2000
1000
0.0 0.1 { 10

MIN CREEP RATE (% /hr)

ORNL DWG 686-11457

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6O [ ; : 60
| | |
| 3
50 - : : / 50
240 |— b ‘ ‘L e /o a0
— T =
2 | 2
an j | 3O< Fig. 6.29, Ductility of
+ T EE - et =
. | "\ = Hastelloy N at 282°C.
m .
> HEAT 5065 \ ‘ e
a O ELONGATION bl L[| 2o 2
Z 20 - ] 20 2
A REDUCTION IN AREA 2
DUCTILITY OF HASTELLQOY ~ Al
N AT 982°C a
10 F—1 g | 10
1 100 1000

 

RUPTURE TIME (hr)
128

 

.U INCHES
M 10CX

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 6.30. Photomicrograph of Hastelloy N Tested at 982°C and
2000 psi.

 

0.035 INCHES
10CX

M

s

 

 

 

Fig. 6.31. Photomicrograph of Hastelloy N, Heat 5065, in the As-
Received Condition.
129

- Y-74240

 

0.035 INCHES
M 1oCxX

 

 

[_..

 

iCn

 

£ 500X

 

0.007 INCHES

 

 

Fig. 6.32. Photomicrographs of Hastelloy N Tested at 982°C and
2000 psi.
130

Staining with a KMnO4-NaOH solution indicated that the precipitates were

of two different compositions. Although this precipitation dces not ap-
pear to impair the ductility at 982°C, it remains to be established whether
the low-temperature ductility is affected. BSince the vesgsel will be ex-
posed to numerous thermal cycles, the high- and low-temperature ductilities
are both of interest.

The results of thisg study are hardly adequate for the design of a
distillation system that is to be an integral part of a reactor system.
This study has shown the need for (1) strength data extending to longer
times, (2) tests to evaluate the influence of the precipitation on the
ductility, and (3) oxidation data under conditions of constant and cyclic
temperatures.

6.8 Thermal Convection Loops

 

A, P. Litman G. M, Tolson

We are continuing to study the compatibility of structural materials
with fuels and coolants of interest to the Molten-Salt Reactor FProgram,
Natural-circulation loops described previouslyl8'19 are used as the standard
test 1n these studies.

Currently, four loops are in operation. Three thermal-convection cir-
cults containing simulated MSRE fuel salt and fabricated from either Has-
telloy N, type 304 stainless steel, or Wb—1% Zr alloy with type 446 stain-
less steel external cladding have operated for approximately 4.5, 3.2,
and 0.6 years respectively. Operating conditions for the loops are de-
tailed in Table 6.4. The Hastelloy N and type 304 stainless steel loops
have continued to circulate salt without incident. However, the refrac-
tory-alloy circuit has shown some degradation of late. This was evi-
denced by the cold leg gradually losing temperature despite the addition
of insulation. It is suspected that the smaller internal diameter of this
loop (0.3 in.) has contributed to the operational problems.

The fourth operating loop (loop 10) is fabricated from lastelloy N
and contains a proposed secondary coolant (NaF-KF-BF3, 48-3-40 mole %)
for the reference-design MSBR. This circuit has operated for over 3000
hr with the hot leg at 1125°F and a temperature differential of 265°F.
During this reporting period, a Croloy M loop (loop 12) circulated the
proposed secondary MSBR coolant for 1440 hr, after which time it was
shut down due to plugging. X rays taken of the loop disclosed spotty
high-density regions in the cold leg. A few suspicious areas were also
seen in the hot leg. The loop piping and heat-transfer fluid will be
sub jected to complete chemical and metallurgical analysis.

A Croloy 9M (loop 8) natural-circulation loop containing lead with
230 ppm magnesium as an inhibitor plugged after operation for 2950 hr.
A section through the cold leg in the plugged region is shown in Fig.
6.33. Examination of the loop components is proceeding.
Table 6.4. Thermal Convection Loop Operation Through September 30, 1966

 

 

Hot-leg Maxdimum AT Time
Loop Material 5 . Heat Transfer Medium Temp. o Operated
pecimens o % (°F) -
(°F (tr)
Hastelloy N Hastelloy N + LiF-BeF 5 ~Zrf ,-UF,-ThF, 1.300 160 39,400
2% Mo (70-23-5-1-1 mole %)
Type 304 stainless steel None LiF-BeF »-ZrF ,-UF 4~ThF ., 1250 180 28,125
(70-23-5-1-1 mole %)
Type 446 stainless-steel- None LiF-Belp-ZrF 4~UF, 1400 300 5,235
clad No=1% Zr (65-29.1-5-0.9 mole %)
Hastelloy N None NaF -KI'-EF 5 1125 265 3,110
(48-3-42 mole %)
Croloy 9M Croloy OM Nal -KF-BF 5 1125 260 1,440
(48-3-49 mole %)
Croloy 9M Croloy OM Po + 230 ppm Mg 1100 200 2,950°

 

%Ioop plugged on 9/26/66.
Loop plugged on 6/9/66.

TeT
132

PHOTO 73481

 

Fig. 6.33. Matching Halves of Portion of Cold ILeg from a Croloy
OM Loop Which Circulated Lead with 230 ppm Magnesium as Inhibitor for
2950 nr at 1100°F,

To date, all uninhibited lead systems constructed of carbon, low-
alloy, and stainless steels have tended to plug due to the formation of
dendritic crystals of iron and chromium in the cold regions of the loops.
The hot-leg attack has consisted of uniform surface removal with isclated
pits extending to a greater depth. While Nb—i% Zr alloy has exhibited
no measurable hot-leg corrosion during test, niobium crystals have been
found in the cold leg of a loop which operated for 5000 hr at 1400°F.

All these findings are discussed in detail in a report sumarizing re-
sults to date on the circulating-lead loop program.ao

References

1. MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, pp.
87—92.

2. MSR Program Semiann. Progr. Rept. Feb., 28, 1966, ORNL-3936, pp.
107-8.

3. W. P. Etherly et al., Proc., U.N. Intern. Conf. Peaceful Uses At. En-
ergy, 2nd, Geneva, 1958, 7, 38901,

4. W. Watt, R. L. Bickerman, and L. W. Graham, Engineering 189, 110-11
(Janvary 1960).

5. C. R. Kennedy, Metals and Ceramics Div, Ann. Progr. Rept. June 30,
1965, ORNL-3870, pp. 194—97.
10.

11.

12.

13.

14.

15,
16.

17.

18.

19.

20,

133

R. G. Anderson and J. ¥. W, Bishop, "The Effect of Neutron Irradia-
tion and Thermal Cycling on Permanent Deformations in Uranium Under
Load,"” pp. 17-23 in Uranium and Graphite, Monograph 27, The Insti-
tute of Metals, London, 1962.

J. C. Bokros and R. J. Price, Radiation-Induced Dimensicnal Changes
in Pyrolytic Carbons Deposited in a Fluidized Bed, GA-6736 (November
1965). To be published in the Journal of Applied Physics.

MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, p. 104.
R. G. Donnelly and G. M. Slaughter, Welding J. fi;fl5), 461-69 (1962).

MSR Program Semlann. Progr. Rept, Feb. 28, 1966, ORNL-3936, pp.
l O]...-‘.Zl' .
W, R. Martin and J. R. Welr, Effect of Elevated Temperature Irradia-

tion on the Strength and Ductility of the Nickel-Base Alloy, Hastel-
loy I, ORWL-TM-1005 (February 1965].

 

 

 

 

 

 

W. R. Martin and J. R. Weir, Postirradiatlon Creep and Stress Rupture
Properties of Hastelloy N, ORNL-TM-1515 (June 1966).

MSR Program Semiann. Progr. Rept. Aug, 31, 1965, ORNL-3872, pp. 94—
1.05 *

MSR Preogram Semlann. Progr. Rept. Feb. 28, 1966, ORNL-3936, pp. 111~
21.

D. R. Harries, J. Brit. Nucl. Energy Soc. 5(1), 74 (January 1966).

 

 

 

Jo. Te Venard, Tensile and Creep Properties of INOR-8 for the Molten-
Salt Reactor Experiment, ORNL-TM~1017 (February 1965).

R. W. Swindeman, The Mechanical Properties of INOR-8, ORNL-2780
(Jan. 10, 1961).

G. M. Adamson, Jr., et al., Interim Report on Corrosion by Zirconium
Base Fluorides, ORNL-2338 (Jam. 3, 1961).

MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, pp. 81—
g7.

G. M, Tolson and A. Taboada, A Study of the Lead and Lead-Salt Cor-
rosion in Thermal-Convection Loops, ORNL-TM-1437 (April 1966).

 

 

 

 

 

 
7. CHEMISTIRY

7.1  Chemistry of {he MSRE

 

Behavior of Fuel and Coolant Salt — R. E. Thoma

 

Several refinements of analytical methods were made during the low-
power operating period of the MSRE which led us to believe that the
composition and purity of the reactor salts can be ascertalned accurately
and economically on a routine basis.’ Within the last few months we have -
attempted to confirm this belief by demonstrating that samples of the
reactor salts could be obtained and analyzed on a routine basis with the
reactor operating at full power. In addition we have sought to establish
from the results a practical optimal frequency of sampling. Comparisons
of the current data with previous results indicate that the fuel and
coclant salts have not undergone perceptible composition change since
they were first circulated in the reactor some 16 months ago. The con-
centration of corrosion products has not increased appreciably in these
salts within the report period.

fuel Salt Composition and Purity. Twenty-~five samples of fuel salt
were removed from the MSRE during full-power runs FP-5, -6, and -7 for
compcosition anaLysis.z The results of thnese analyses, listed in Tables
7.1 and 7.2, show clearly the chemical stability of the fuel solution;
it has maintained constant composition since it was constituted in the
reactor in 1965.

 

Structural-Metal Impurities. The concentrations of structural-metal
corrosion products found in the MSRE fuel salt in the period January to
August 1966 are compared with previous assays in Table 7.3. Since chro-
mium is the most chemically active constituent of Hastelloy, all cor-
rosion reactions should, on proceeding to equilibrium, result in the .
production of CrFo. The extent of corrosion in the MSRE is currently
monitored by analyzing for the chromium content of the salts. In its
full-power operation during the period PFebruary to July 1966, the MSRE
has behaved well with respect to corrosion chemistry; no apparent cor-
rosion has occurred within the fuel or coolant systems. We have con-
cluded in the past that corrosion might be influenced by the presence
of a small amount (perhaps approximately 1%) of UF; formed in the IilF-
UF, enriching salt during preparation. When present, UF; is readily
oxidized to UF, in reaction with CrFj:

 

CrlF, + UF3 = cr® + Uk, .

The presence of MoF, or Mol's in the fuel salt, as inferred from the re-
sults of recent examinations of surveillance specimens removed from the
reactor core,3 is puzzling in connection with the likelihood that UFj
sti1ll remains in the fuel salt. The notable freedom from corrosion
during this period therefore is reassuring.
135

Table 7.1, Results of MSRE Fuel 8alt Analyses,
Runs FP-5, -6, and -7

 

Concentration (wt %)

 

 

u;gple Mirhr -

. Li Be Zr U* F 5
FP5-1 26.5 10.65 6.53 11.45 4625 68,18 101..44
FP6-2 11L.43 D.22 10.80 4.622 68.97 102,04
TP6=5 10,45 6.37 11,12 4,605 69,85 102.40
FPe-6 10.45 646 11.32 4625 68.56 101.52
Pe-7 180 10.32 6 41 11.42 4o 047 69 .67 102.47
FPo-8 10,43 649 11.2%9 4655 68,20 101.77
FP6-9 10,48 6.066 11.52 4 684 68.92 102.26
FP6-10 10.30 6. 70 11.54 4,595 685,81 102,00
FPe-11 10.30 6442 11.28 4,012 70,02 102.70
Fre-13 1000 10.50 6.41 11,06 4.628 66.79 99,38
Pe~14 10.50 645 11.33 4,617 68,18 101.08
TPG6-15 10.55 6.86 11.35 4 601 68.1.8 101.54
P6-16 10.55 6.58 11.31 %« 629 67.88 100.90
FPe-17 11 .44 O .64 11.05 4.652 68.26 102.04
FP6-19 10.40 6.88 11.72 4667 69.1.0 102.77
FP7-1 2920 10.60 6.78 11.16 e g GA4TT 69.26 102.45
FP7-3 10.55 6.56 11 .44 4 .656 68 .24 101 .45
FP7 -4 10.50 6.40 11.61 4 640 67.32 100,97
FP7-6 4626 10.60 6.63 11.35 4614, 69.22 102.41
Fp7-7 10.55 6.65 11.13 4 o 641 69.60 102,57
FP7-8 10.60 .59 11.67 4,663 67.87 101.39
FPr7-10 10.63 6.91 11.21 4 . 609 68.20 101..56
FP7-11 6900 10.45 7.00 11.22 4 .630 69.48 102.78
FP/7-12 1.0.50 6.65 11.04 4 o 640 68 .44 101.27
FP7~14 10.55 6.50 11.26 4 .660 68.79 101.76
¥FP7-1.6 7823 10.55 6.71 11.60 4,638 67.59 101.08

 

The average concentrations of the structural-metal impurities which
have been found in the MSRE fuel sall since the beginning of reactor op-
erations are given in Table 7.3.

An interesting trend is evident in the above results. As impurities,
iron and nickel are presumed toc exist as colloidally dispersed metallic
particles. The concentration of iron has decreased slowly bubt steadily
during reactor operation, while nickel analyses remained essentially
constant.

Oxide Analyses. FEight samples of MSRE fuel were analyzed for oxide
content during the recent full-power operations. etails of the proce-
dures employed are given elsewhere.® The average value of oxide concen-~
tration was found to be 54 ppm, significantly lower than that found in
136

Table 7.2. Summary of MSRE Fuel Composition Analyses

 

Component. Nominal Boock FP4 P5-7 FP4~7

 

Mole Percent

 

 

 

 

LiF 65,00 64 . 88 63.36 £ 0,567 62.20 £ 0,764 ©3,29 * 0,721
BelF o 29.17 29.26 30.65 + 0.583 30.76 = 0.746 30.70 + 0,695
Zriy 5.00 5 .0¢ 5.15 = 0,116 5.22 £ 0.129 5.19 £ 0,126
UF 4 0.83 0.82% 0.825 + 0,011 0.822 + 0.011° 0.824 + 0.0LL
Weight Percent
Li 10.95 10.93 10.51 + 0.137 10.54 * 0,017 10.52 +* 0,178
Be 6.32 ©.34% 6.55 + 0,161 6.60 * 0.183 6.57 * 0,179
Zr 10.97 11.06 11.14 £ 0,295 11.33 % 0.220 11.24 £ 0,271
U 4,73 L6462 4642 t 0.028  4.635 t 0.022°  4.638  0.025
9237.003 = 33.241 wt % 235U,
©237.009 = 33.06 wt % 235U.
Table 7.3. Concentration of Cr, Fe, and Ni in MSRE Fuel Salt
Run Nugber of Concentration (ppm)
No. Samples O Fe Ni.
¥p-3 51 37 £ 8 154 + 55 48 * 19
Fr-4 22 48 + 7 131 + 65 20 £ 20
rp-5,6 14 50 £ 7 108 * 44 54 * 25
rp-7 11 48 + 6 79 + 38 48 + 23

 

the zerc-power experiments (95 ppm). Since fluorine can be evolved from
frozen irradiated fuel, it was necessary to consider the possibility that
oxlde is released by the reaction

7S + 0,7 = =08 + 2F .

o

Sample specimens were analyzed in one instance aflter minimum delay {ap-
proximately a 7-hr interval between sample removal to HF purge) and in
another instance after a 48-hr delay. Results of the two analyses were
coincident within experimental error. Although the minimum delay period
is probably not sufficilently long to allow any of the oxide to he con-
verted to oxygen, plans are under way to test the infrinsic capability
of the method by standard additions of oxides to future samples.4
137

Table 7.4. Isotopic Composition of Uranium in MSRE
Fuel Salt Specimens

 

Isotopic Composition of U (wt %)

 

 

Sample Mirhr

No. 2347 235y 236(; 238y
#P2-10-13 0 0.356 33.716 0.145 65,783
Fre-14 1000 0.352 33.534 0.144 65.970
FP6-19 2900 0.345 33 Ak 0.151 66.060
FP7-8 5000 0.352 33.400 0.163 66.085
FP7-12 7000 0,349 33.329 0.174 66.148
FP7-15 7600 0,347 33,161 0,177 66,315

 

Isotopic Analysis of Uranium. Varistions in the relative concen-
trations of uranium isotopes in the MSRE fuel are determined on a regular
basis by mass spectrometric analysis.5 A summary of the isotopic analyses
of the MSRE fuel is given in Table 7.4, which shows qualitative evidence
of 23y burnup and ingrowth of 238, The limit of experimental accuracy
(#0.1% of individual values) and the low burnup fraction (approximately
1%) render the current values of little use for accurate caleculation of
M5RE power levels at present. 1Tt is anticipated that as a greater
fraction of *3%U is consumed, such determinations will assume greater
significance in corroborating burnup computations which are based on
rerformance data.

 

Coclant Salt Composition. Coolant salt was circulated in the MSRE
for some 1200 hr during the prenuclear test period. Since flush and
coolant salts were supplied from the same reservolr, the coolant salt
was analyzed during this period only for impurities. At the end of pre~
nuclear testing, concentrations of the structural-metal impurities, Cr,
Fe, and Ni, were established to be approximately 30, 90, and 7 ppm,
respectively, which are remarkably low considering that the circuit had
not been Tlushed previously with molten salt. The results of spectro-
chemical analyses made at that time showed that no significant amounts
of additicnal impurities were introduced during this period.6 As the
reactor was brought to [ull power early in 1966, coolant salt was again
circulated, sampled regularly, and subjected to compositional and im-
purity analysis. The results obtained from these analyses are given in
Table 7.5.

 

In addition to the components, ceolant salt was analyzed for, and
found free of, zirconium by wet chemical and spectrochemical methods.
The intention was to confirm that no fuel salt had been transferred to
the coolant circuit.

Analyses of coolant salts are conducted in the General Analysis
laboratory of the ORNL Analytical Chemistry Division, while those for
the fuel are obtained from the General Hot Analyses laboratory of that
138

 

 

 

 

Table 7.5. Results of MSRE Coolant Salt Avalyses, Runs CP-4 to -7
Concenbration
1 Total Time
S;mp © in Pump Bowl Wt % ppm
e (hr)

Li Be F 5 Cr Fe i o
CP4-1 36 65 <5
CP&4-2 11 26 83 <5 130
CP4~3 59 13.78 .91 76,70 99.39 41 41 37 185
CPL—d 67 53 46 24
CP4-5 180 14.20 g8.87 '76.80 99.87 50 50 20
CP4~6 229 38
CP5-1 13.86 8.85 76,7 29.41 35 5 <10 150
CP5-2 13.82 8.55 76.6 98.97 35 <2 <10 110
CP5-3 12.76  10.51 76.04 99,31 46 57 64 65
CP5-4 13.00 9.17 76.8 98.97 35 119 14 <20
CP5~5 12.60 9.15 76.7 98.45 42 87 28 1.20
CP5-6 13,01 10.06 77.2 100,27 20 100 8 260
CP5~7 855 13.17 9.49 76.4 99.06 24 112 <2 171
CP5~8 12.70 10,22 772 100,12 29 36 <z 71
CP5-9 12.77 9.43 76.7 98.90 32 44 <2 35
CP5-10 12.97 2.59 76,9 99,46 33 31 <2 <20
CP5-11 12.92 9.6l  76.4 98.93 36 49 53 <20
CP5-12 12.75 9.74  76.4 98.89 44 57 8 <20
CP5-13 1358 13.10 9.98 76.2 99,38 32 73 10 162
CPo-1 13,09 9.73 76.1 298.92 30 43 4] <20
CP6~2 12.76 9.56 76.4 °08.64 38 28 9 <20
CP6-3 12.82 9.13 76.8 98.75 45 76 47 210
CP6-4 12.76 9.63 76.0 98.39 57 68 15 210
CP6-5 12.70 9.79 74.8 97.29 20 57 o 135
CP6-6 3135 9.65 74,1 28 37 6 <40
CP7~-1 13.64 8§.72  '74.65 97.01. 32 60 <2 <20
Cp7-2 12.16 9.88  74.5 9.54 31 5L <2 300
CP7-3 12.39  10.24  74.8 97.43 51 80 <2 <145
CP7-4 3462
Average 13.03 Q.40 76.16 28.69 36 58 16 107
Standard deviation 0.46 0,49 0.83 10 27 17 79
Nominal 13.94 9.24 76,82  100.00

 

3KBrF4 method.

division.

are employed.

It is apparent from the contrast of material balances shown
in Tables 7.1 and 7.5 that minor differences in analytical procedures
These differences were appraised with respect to their
possible influence on compositional analysis and were found to he in-

significant, since compositional analysis is computed from cation assays
without use of the fluoride values.
139

On the basis of the analytical data listed in Table 7.5, the present
composition of the coolant salt is T1iF-BeF, (63.93-36.07 mole %),
practically identical with values obtained in earlier analyses6 of the
flush salt (63.80-36.20 mole %). The unresclved disparity between the
design composition (66.0 * 0.25-34.0 = 0.25 mole %) and that found by
chemical analysis continued to be a puzzling matter. The excellent re-
producibility of chemical analyses throughout the operational period of
the MSRE is, however, of significance in affording the opportunity for
chemical trends to be observed.

7.2  Physical Chemistry of Fluoride Melts

Viscosity and Density of Molten Beryllium Fluoride — S. Cantor and
C. T. Moynihan¥*

The viscosity of molten Belp was measured over the temperature range
573 to 985°C using Brookfield LVT and IBF5X viscometers. In this tem-
perature range, the viscosity, n, varied from 10 to 10% poises and was
about lO% greater than previously reported.v

The plot of log n vs l/T(°K) showed only slight curvature. A least-
squares it of the data to an equation quadratic in l/T yielded the
following:

1.1494 x 10% , 6:39 x 107
T(°K) T2 ?

 

log n (poises) = —8.119 +

standard error in log n = 0.013. This corresponds to an apparent in-
crease in the activation energy for viscous flow from 57.3 kcal/mole at
985°C to 59.6 keal/mole at 575°C. These are relatively small activation
energy changes, and thus the temperature dependence of the viscosity of
Befs is basically Arrhenius over the range (10 to 10° poises) covered in
this investigation.

Density measurements were attempted via the Archimedean method, by
determining the appareant loss of welght of a platinum sinker when im-
mersed in molten BeF,;. In such viscous melts, the balance is extremely
slow in coming to equilibrium; hence, the method® of extrapolating to
zero velocity the plots of rate of ascent or descent of the sinker under
various balance loads (measured by timing the travel of the balance
pointer) was used to determine the equilibrium weight of the sinker in
the melt. In neone of the Tour attempts to measure the Bel, densily was
it possible to eliminate the few small bubbles that adhered to the
sinker. 'The buoyant effect of the bubbles leads to low values of the
apparent weight of the sinker and, hence, to high values of the density.

 

*Chemistry Department, California State College al ILos Angeles; summer
participant, 1966.
140

If a surface tension of 200 dynes/cm is assumed for BeF,, the best
of our deunsity results yielded a value of 1.96 * 0.01 g/cm3 for BeFs at
850°C. This result must be considered only as an upper limit to the real
Bel's density, but it may be compared to the value of 1.95 * 0.01 g/cm3 at
800°C reported by MacKenzie® and to a value of 1.968 g/em3 measured for
the Bel's glass at room temperature. These results suggest that the thermal
expansion coefficient of liquid BeF, is quite small.

Transpiration Studies in Support of the Vacuum Distillation Process —
5. Cantor

 

To determine the equilibrium vapor separation of rare-earth fission
products from M3SR carrier salts, a series of vapor pressures have bheen
measured by the transpiration (i.e., gas-entrainment) method. The melts
were composed of 87.5-11.9-0.6 wole % LiF-BeF,-1aF3. These concentrations
of IiP and BeF, are approximately those expected in the still pot of the
vacuum distillation process, but the lanthanum concentration is many times
greater than would ve permitted as total rare-earth concentration in the
still. This high concentration of lanthanum in the melt was required in
order to give lanthanum councentrations in the vapor that were high enough
to analyze.

 

Measurements were carried out in the temperature interval 1000 to
1062°C; dry ar§on, the entraining gas, flowed over each melt at the rate
of sbout 30 em”/min. Salt vapor, condensing in a nickel tube, was ana-
lyzed by spectrochemical and neutron activation methods. The latter
method gave higher, more consistent, and probably more reliable lanthanum
analyses. To date, the more consistent results have been obtained at
only two temperatures:

 

Temperature Decontamination Factor®
(°c) for Ianthanum
1000 210
1028 1150

%Defined as (mole fraction of lanthanum
in liquid)/(mole fraction of lanthanum in con-
densed vapor).

At six other temperatures, transpiration pressure measurements yielded
much higher (up to 7300) decontamination factors; however, these deter-
minations either were hased on insufficient sample for proper analysis
or else duplicate analyses gave widely different results. It did appear
that the higher the temperature, the higher the decontamination factors.

Although much more study is required before the vacuum distillation
process can be shown to be practical, it seems that decontamination fac-
tors close to 1000 can probably be demonstrated.
141

Estimated Thermophysical Properties of MSBR Coolant salt — 5. Cantor

 

Preliminary phase equilibrium studies have indicated that a mixture
containing 47.5 mole % NaF, 48 mole % BFi3, and 4.5 mole % KF (melting
point, 365°C) is sufficiently low melting for use as the secondary coolant
of the MSBR. To aid investigations (such as heat-transfer studies) that
will likely be carried out in the near future, estimates are given below
for some thermophysical properties of this mixture. It is anticipated
that certain thermophysical properties will soon be measured. The esti-
mates given hereln are meant to serve until such measurements can be
made .

Measurements of dissociation equilibria have shown that KBF410 is

much more stable than NaBF4.ll The composition given in the first para-
graph can be recast as a mixture of [luorcoborates whose composition is
(in mole %): NaBF,, 83.65; KBF,, 8.65; and NaF, 7.7. This recast form
is useful because it permits estimates of properties to be made in terms
of components which are all molten (or supercooled liquids) in the tem-
perature range of interest (BF; is a gas in this tenperature range ).

Vapor Pressures. These were estimated by assuming (1) that the
fluoroborate mixbure behaves in accordance with Raoult's law and (2) that
the only vapor species is BF3. Since the dissociation pressure of NaBF4ll
is many times greater than that of KBF4,10 then, given the two assump-
tions, the total vapor pressure P may be expressed as

 

” _ < 0
P(mm) = 0.8305PN8,BF/+ s (1)
where FO is the dissoeclation pressure (in millimeters) of pure NaBr.

NaBF,
From the log P vs l/T equation for pure N&BF4,11 the temperature depen-
dence of the proposed MSBR coolant is easily derived and ig given by
log P{mm) = 6.51 — 3650/T(°K) . (2)

By appropriate substitution in Eq. (2}, Table 7.6 was generated.

Table 7.6. Estimated Vapor Pressure of Proposed
MSBR Coolant Salt

 

 

Temperature Pressure
(°c) (s )
370 (lowest possible temperature in 6.8

coolant circuit)

454 (mean temperature of coolant 27
going to heat exchanger)

607 (highest normal operating 229
temperature )

 
142

Density. The liquid densities of the fluoroborates have not as yet

been measured. For an estimate of the density at 454°C, the following
assunptions were made:

1. The fluoroborate mixture is additive with respect t¢o liquid molar
volumes of the components.

2. The molar volumes of liguid N&BF411 and KBF412 are 1.2 times the
X-ray molar volumes of the respective solids.

3. The molar volume of supercooled liquid NaF at 454°C can be ob-
tained by extrapolation of the stable ligquid values.®3

From these three assumptions a value of 2.1 g/cm3 is calculated as
the density of the coolant at 454°C. This density is probably uncertain
by 10%. Because expansion coefficients of fluorcoborates are unkrown,
estimations of the temperature coefficlent of density would be premature
at this time.

Speclfic Heat. By using the rulel® +that each gram-atom of a flu-
oride mixture contributes about 8 cal/°C to the molar heat capacity, the
molar heat capacity of the coolant eguals approximately 48 cal/°C. But
since the coolant contains the complex ion [BF4]1”, some of the oscil-
lational degrees of freedom of the individual atoms are lost; my guess
is that this loss amouats to about 7 cal/°C. Thus, the estimated specific
heat is 41 cal/°C divided by the molar weight of the coolant, 106 g, or
ebout 0.4 cal g™t (°C)™t. This estimate is provably good to +25%.

Viscosity. The viscosity of NaBF, was measured™’ somewhat crudely
at two temperatures. The results were 7 * 2 centipoises at 466°C and
14 + 3 centipoises at 436°C. Since the coolant is predominantly NaBF.,
these rough values can serve as a gulde Lo viscosity until further ex-
perimental values become available.

7.3. Separation in Molten Fluorides

 

Extraction of Rare Earths f{rom Molten Fluorides into Molten Metals —
J. H. Shaffer, F. F., Blankenship, and W. R. Crimes

 

This experimental program envisions a liquid-liquid extraction
process for removing rare-earth fission products from the fuel solvent
of the reference-design MSBR and a similar back-extractlion process for
concentrating them in a second salt mixture for disposal or further
utilization. This process is an alternative to distillation aund like
the distillation process would follow removal of the uranium by the
fluoride volatility process. The results of previous experiments have
shown that a mixture of bismuth containing 0.02 mole fraction of lithium
metal sufficed for removing essentially all cerium, lanthanum, and neo-
dymium and substantial gquantities of samarium and eurvopium from a mixture
of IiF-BeF, (66-34 mole %).1® The second phase of the reprocessing
143

method would then involve the back extraction of rare earths from the
molten metal mixture into another salt mixture. Hydrofluorination of
such a two-liquid-phase mixture should be very effective for this step,
but the overall process would also result in the removal of some lithium
from the fuel. Since isotopically pure “Ii is used in the fuel, the con-
servation of this material is of economic importance. According to tne
experimental data, the removal of 1 g of neodymium would result in the
loss of about 2.2 g of lithium. However, material balances on experi-
ments in which lithium was used as the reducing sgent showed that only

25 to 50% of the lithium added to the extrasction system was found as a
dissolved component of the metal phase. The gquantity of lithium required
for the stoichiometric reduction of rare-earth fluorides was negligible
by comparison. Therefore, additional studies of the behavior of lithium
metal in the extraction system have been conducted.

Duplicate experiments dealt with the material balance of lithium in
bismuth in the absence of a salt phase. As 1n previous experiments,
about 2.35 kg of bismuth was contained in low-carbon steel at 600°C.
Filtered samples of the molten metal were taken after each incremental
addition of 1lithium metal to the systems and during prolonged eguilibration
periods (100 hr) during which one mixture was maintained under static
helium and the second was sparged with dry helium. The results of spec-
trographic analyses of these samples showed that, as lithium was added,
its concentration in each vessel increased in amounts which corresponded
to about 75 to 80% of the amount added to each system. However, upon
standing, the concentration of Iithium in each vessel became constant at
values which corresponded, very nearly, to the amount of lithium added.
These experiments illustrated the stapility of lithium~bismuth mixtures
contained 1n low-carbon steel equipment and indicated that the rate of
dissolution of lithiuvm Into bismuth in the extraction systems was some-
what slower than anticipated.

The foregoing results showed that the unexplained 1ithium losses
were not attributable to the analytical procedure nor to the behavior of
lithium in the molten metal phase of the extraction systems. Accordingly,

the next phase of this experimental series was an examination of the ef-
fect of the salt pnase on the concentration of lithium in the metal phase.

A previously purificd mixture of ILiF-BeFf, (66-34 mole %) was now
added to each of the above mixtures of lithium snd bismuth in three sue-
cessive tared increments of about 1 kg. IFiltered samples of the metal
phases were taken during 24-hr egquilibrium periods after each salt ad-
dition. The results of spectrographic analyses of these samples showed
losses of lithium from the metal phase after each salt addition. As
shown by Fig. 7.1, this loss corresponded to about 0.11 mole of lithium
per kilogram of the salt mixture and was independent of the concentration
of 1lithium in the metal phase. Although this loss may be attributed to
a number of possible clrcumstances, it should not materially affect the
econoimics of the rare-earth extraction process.

In earlier experiments, tne extracticn of rare earths from a salt
phase into molten bismuth was achieved by the addition of beryllium metal
to the system. This reduction process also resulted in an increases of the
lithium concentration of the molten metal phase. Accordingly, further
¢
i

‘mole fraction)

LiITHUM FOUND IN 8:SMUTH

study of the reaction

2IiF + Be©

in the two~phase extraction system has

144

peen initiated.

T QLL'LD + Bel's

The first experi-

ment of this series was conducted by adding clean beryllium metal turnings

to a two-liquid-phase system of 3.17 kg of LiF-BeF. (66-
2.30 kg of bismuth contained in low-carbon steel at 600°C.

34 mole %) and
Filtered

samplies of the metal phase were withdrawn at 4- and 24-hr intervals after

each incremental addition of about 1 g of beryllium metal.

These samples

wvere analyzed speotrogriphlfially for their lithium and beryllium content

by groups after 5 g of Be”
Fig. 7.2,

had been added to the systen.
the lithium concentration jin bismuth inecreased after each ad~

As shown by

dition of beryllium until a value of abvout 0.16 mole fraction lithium was

attained.

Beryllium concentrations in bismuth remained below the 1imit
of detection (less than 0.0001 wt %) of the analytical method.

Thus, the

analytical results for lithium in bismuth shonld reflect dissolved lithium

metal concentraticons rather than the dissolution or suspension of
olnce the distribution of rare
two liquid phsses of the extraction system was found dependent upon

in the molten metal phase.
the

the lithium concentration in bismuth, €

limits for estimating extraction parameters.
fraction of lithium in bismuth 1s assumed to be in equilibrium

0.16 mole

salt
earths between

esults provide extrapolation
If this limiting value of

with beryllium metal at unit activity, the activity coefficients y of

Fig. 7.1.

Rate of Lithium Loss

Trom Molten Bismuth into LiF-Ber,

(66~34 Mole %) at 600°C.

LITHUM «;OSS ENTO SALT PHASE

ORNL-DWG 65—11458

 

 

 

 

 

ORNL-DWG 66-11459

 

 

 

 

 

 

 

 

 

 

 

0.20 e : |
ZIGHT OF SALT PHJIQSEZ 347 kg [
WEIGHT OF METALL PHASE 2.30 kg ®
016 ; | ; | "?L__g...—-.'..
| , ‘/
o ——— -w~-} } ;7# Fw \ -
: 1 J | }
|
; | »/?‘ o
0C8 |- _— * L
i ( Lo® 3 | |
| 8 o {
ooa | L
] |
44 | | |
._,_,.3/3 ! !
o =¥ L i 4
0 Q.2 0.4 C.6 0.8 1.0 1.2 14 16

BERYLLIUM ADDED TC SYSTEM (moles)

L 03— e -
5 | |
5 Lfi | E |
€ o EXPT Li—6 i
g02 T e ExPT Lim? ——1— T
2
- » | {
: .
O 0o =1 —i»f-'” ————
5 e |
v \
2 ,
Q
£ 1 ;
oL L | -
0 0.02 0.04 0.06 0.08 0.0
LITHIUM CONCENTRATION IN BISMUTH (mole fraction)
Fig. 7.2. Reduction by Beryllium

Metal of LiF-BeF, (66-34 Mole %) in
Contact with Molten Bismuth at 600°C.
145

lithium and some rare earths in molten bismuth can be estimated. On the
basis of this lithium concentration and thermodynamic data, Hill
found* 7' 18 the following activity coefficients of metals dissolved in
molten bismuth at 600°C:

 

Metal ¥
Ti g x 1072
La 1.2 x 107+
Ce 2.9 x 10™t#
Nd g8 x 10713

Comparable results at 500°C which differ by no more than a factor of 10
have been reported.19’2o Additional experiments are in progress to better
define the equilibrium reactiocn.

Removal of Rare Barths from Molten Fluorides by Simultaneous Precipitation
with Uf3 — J. H. Shaffer and H. ¥. McDuffie

 

The relatively low solubility of UF3 in fluoride mixtures of interest
to the MSR Program,zl together with the known similarities of the crystal
structure of rare-earth trifluorides with UF3,22 provides a baslis for
studies of the precipitation of solid solutions of these compounds Trom
fluoride melts. Fissicon product rare earths represent a major portion
of the poison fraction in the fuel of a molten-~salt nuclear reactor; thils
study has been oriented toward the development of suitable reprocessing
methods for rare-earth removal. Initial experiments conducted in this
program considered the reduction of UK, contained in the reactor fuel
mixture to UFs and the simultaneous precipitation of rare-earth tri-
fluorides with UF; as the temperature of the fuel mixture was reduced.

A second series of experiments 1s in progress to examine the precipitation
of rare earths from a simulated fuel solvent upon addition of solid UFj.

If all the UF, contained in the current MSRE fuel mixture, LiF-Bel's-
7rF,-UF, (65.0-29.1-5.0-0.9 mole % respectively), were reduced to UFj,
the solution would be saturated with UF; at approximately 725°C. By
lowering the melt temperature to 550°C, approximately 83.5% of the uranium
would be precipitated from solution. Results of preliminary experiments
designed to investigate this reprocessing method demonstrated that ILalk's,
CelFs, and NdF; could be precipitated with UF3. Europlium and samarium
were probably reduced to thelr divalent states by the in situ reduction
of uranium with added zirconium metal and showed 1ittle or no loss from
solution during the precipitation of UFj5. OSubsequent experiments with
excess reducing agent showed that cerium removal could be related to the
U2t concentration in solution by the equation

In Npp = k In Wys+ + const (3)

wnere NRE and NU3+ are respective mole fractions of rare earth and tri-
valent uranium. As illustrated in Fig. 7.3, a value of about 0.55 was
146

obtained for k in Eg. (3) for the simultaneous precipitabtlion of CelFj.
Further investigation would be needed to verify this experimental rela-
tionship for other rare earths of interest to the program.

tracer.

A more recent experimental program has been concerned with the re-
tention of rare-earth trifluorides on a bed of solid UF3 as an alternative
reprocessing tecnnique.
ments to approximately 2.2 kg of IiF-BeF, (66-34 mole %) that initially
contained 107% mole fraction of CeFs with about 1 mc of l440e as a radio-

In this experiment, UF; was added in 30-g incre-

Filtered samples of the salt mixture were taken approximately

48 hr after each addition of UF3 and analyzed radiochemically for cerium.
As shown by Fig. 7.4, these results illustrate a somewhat linear decrease
in cerium concentration as UF; was added which corresponds to a solid

phase which contained about 1 mole % CeFj3.

N @
O O

a
O

4.0

1o
O

N
o

CERIUM FOUND IN SOLUTION {mole fraction x105)
&

50 -

 

 
    

%Lff‘wrmfififlimwjm

EXPT Ce-2 __ |

 

 

 

 

 

 

TEMPERATURE RANGE: 850-550 °C
INITIAL U CONCENTRATION
o Hwt%

) 5wt %

i

 

CRNL-DWG 66-11460

 

 

   

?

 

 

 

 

2 3 4 5

6 8

15 20

URARIUM FOUND IN SOLUTION (mo'e fraction x103)

Fig., 7.4.

with Solid UF; at 550°C.

Removal of CeF,; from
LiF-BeF, (66-34 Mole %) by Exchange

25

CERIUM FOUND IN SOLUTION {mole fraction X1O4)

lated MSRE HFuel Mixture.

1.0%--

o8

06

04

02

 

Fig. 7.3.

Simultaneous Precip-
itation of CeFg and UF; from Simu-

 

 

 

 

ORNL ~DWG €6-11461

WEIGHT OF SALT
SOLVENT: 246 kg _ |

 

 

 

 

 

 

 

 

 

 

 

04

Uf-”3 ADDED (moles)

0.6
147

Removal of Protactinium from Molten Fluorides by Oxide Precipitation —
J. H. Shaffer, ¥. F. Blankenship, and W. R. Grimes

 

In a previous experiment, protactinium was removed from solution in
a solvent mixture of LiF-BeF, (66-34 mole %), containing ZrF, (0.5 mole
per kg of salt), by the addition of Zr0O, at 600°C.%2 An interpretation
of the experimental data according to the equatlon

1. D
e = ] 4 e ¢ W (4)
n 1 7 J
Fpg Vaalt 2r0z

where D = (Pa)oxide/(Pa)salt, Fo, = fraction of Pa in the salt, and W =

welght of the designated phase, showed that the distribution of pro-
tactinium between the two phases remained constant over the protactinium
concentration range of the experiment. These results could be inter-
preted as the formation of labile oxide solid solutions or as surface
absorption of “?2Pa on the solid ZrO,. Further studies of this oxide
precipitation behavior were conducted in the same fluoride solvent with
Zr0, powders having varied surface areas.

Zirconium dioxide used in the original experiment was purchased
commercially and had a surface arca of about 19.6 m?/g. Material having
higher surface areas was prepared from Zr(0H), by dehydration.24 sSuf-
ficient ZrO, for this experimental series was fired at 600, 700, and
1000°C in separate batches that yielded average surface areas of 80, 20,
and 1.32 mz/g respectively. About 3.55 kg of a salt mixture having a
nominal composition of LiF-BeF,-ZrF, (64.8-33.6-1.6 mole %) with about
1 me of 233Pa {as irradiated Thoz) was prepared in a nickel vessel by
our usual HF-H, treatment at 600°C, followed by H, sparging at 700°C for
further purification and dissclution of protactinium as its fluoride
salt. Selected ZrO, was added to the salt mixture in 10-g increuments;
the mixture was then sparged with helium at a rate of about 1 liter/min
during 24-hr equilibration periods. Filtered samples of the salt mixture
were taken after each equilibration period and analyzed radiochemically
for %23pa by counting its 310-kv photopeak on a single-channel gamma
spectrometer. At the conclusion of the experiment, the mixture was
hydrorfluorinated to convert added Zr0, to itls fluoride salt and to restore

233 pg, activity in the molten-salt phase. This experimental procedure
was repeated with the same salt mixture for all three lots of Zro0,.

In each experiment the addition of ZrOp; to the fluoride mixture re-
sunlted in the loss of protactinium from solution. However, as shown by
Fig. 7.5, a plot of the reciprocal fraction of 22°Pa in solution vs ZrQo
added, according to Eq. (4), yielded distribution coefficients for <°3Pa
between the two phases which varied continuously as the precipitation
reaction approached completion. These experimental results are contrary
to those obtained previously with respect to the constancy of the )
distribution ccefficients; they indicate that 233pa removal from the salt
mixtuire is not proportional to the surface area of the added oxide parti~
cles and also does not conform to an eguilibrium solid solutlon. Experi-
meants in progress will attempt to further elucidate the characteristics
of the oxide precipitation process.
ORNL-DWG 66-114562
25 | " I . T ]
® SURFACE AREA 7r0, =80 m?/qg 1 1
O SURFACE AREA ZrO, =50 m?/g
A SURFACE AREA Zr0,=132 m?/g

 

 
 
 
   
   
 
 
 
    

20 |--
WEIGHT OF SALT MIXTURE: 3.55 kq

W.
ZrQ
'I!:— = +D(W r,..?,..,)
Pa SALT
~ WHERE FPu ~— FRACTION OF Pa IN LIQUID
W= WEIGHT OF DESIGNATED PHASE
/- CONC OF Pg IN SOLID PHASE

~ CONCOF Pa IN LIQUID PHASE
L

 

233

 

RECIPROCAL FRACTION OF 23%pg IN SOLUTION

 

 

 

 

Zr0, ADDED (g)

Fig. 7.5. Effect of Surface Area of Zr0O, on the Removal of 223Pa
from LiF-BeF,~ZrF, (64.8-33.6-1.6 Mole %) at 600°C.

Extraction of Protactinium from Molten Fluorides into Molten Metals -
J. H. Shaffer, F. F. Blankenship, and W. R. Grimes

 

The removal of protactinium from solution in IiF-BeF,-ThF, (73-2-25
mole %) by adding thorium metal directly in the salt mixture or by con-
tacting the salt with molten lead or bismuth in which thorium had been
dissolved has been demonstrated in several experiments.25 More recent
experiments have examined methods which could be used for recovering
protactinium from the fertile blanket of the reference-design MSBR. The
results of several batch-type laboratory experiments led to the design
and operation of a small pump loop which has demonstrated, in principle,
the removal of protactinium from molten fluorides by a liquid-liquid ex-
traction technigue.

In early experiments a molten fluoride mixture containing dissolved
233ps gnd molten bismuth containing dissclved thorium metal were left in
contact in low-carbon-steel containers under static conditions. Subse-
quent examinations of the container indicsted that most of the prot-
actinium had deposited on the vessel walls that were in contact with the
salt phase. The disappointing fallure of the protactinium to cross the
boundary of the two liquid phases may have resulted from nonwetting be-
havior. The effects of such behavior could be greatly reduced by the im-
proved contacting oobtainable in a pump-loop system.

In other experiments in which no salt phase was used, solutions of
233pg in molten lead and bismuth were obtained by the addition of irradi-
ated thorium metal to the 1liguid metals in steel containers. Radio-
chemical analyses of filtered samples taken from these vessels showed
149

that 23°Pa did not form stable solutions in either liquid metal. About
40% of the °33Pa activity was retained in bismuth, and less than 5% was
retained in lead during 48-hr contact periods. Much of the 233Pa was on
the walls of the vessels, with the distribution shown in Fig. 7.6. The
larger amount near the bottom of each vessel lmplies loss of 233pg via
sedimentary deposits of insoluble materials rather than by surface ad-
sorption. It is of interest that the concentrations of thorium in olsmuth
were found, spectrographically, to be less than the 1000 ppm added and
well below reported solubility values of 3000 to 4000 ppm.=27

The fractions of 222Pa and thorium found in solution iu vismuth were
roughly equal. The concentrations of both increased for about 5 hr after
the irradiated thorium was added to the liguilid bismuth and then decreased
with time. The behavior in lead appeared to be considerabvly different.
Although as much as 80% of the thorium dissolved in the lead, the fraction
of protactinium in solution remalined very small.

We concluded from these data that bismuth is the preferable extractant
and that formation of slightly soluble compounds can complicate the ex-
traction of protactinium from fluoride salts by causing the effective solu-
bility of protactinium in liquid bismuth to be low. Although a high solu-
bility seems desirable, the liguid-metal pnase in the process acts only
as a carrier to transfer the protactinium from one salt to another. High
solubility of protactinium in the liquid metal i1s not an essentlal condi-
tion for attaining adequate transfer rates. Thus, a pump-loocp experiment
was tried in an effort to achieve transport of protactinium from a blanket
salt while maintaining a low concentration in the liquid bismuth.

An experimental loop, shown schematically in Fi%. 7.7, was designed
to contact a fluoride mixture containing dissolved 233Pa with =a recircu-

lating stream of molten bismuth contalning thoriuwm. The thorium was intro-
duced by passing the bismuth over thorium chips Just before its entry into

ORNi.~DWG 6611463

100 W T —‘

80l e ]

 

 

o BiSMUTH
o LEAD

 

 

. Fig. 7.6. Distribution of 223pa
| in Low-Carbon Steel Used to Contain
) Bismuth and Lead at 600°C.

 

 

INTERFACE BETWEEN LIQUID -
METAL AND COVER GAS

 

 

 

23350 ACTIVITY FOUND ON CONTAINER WALL (%)

 

 

 

 

 

 

!
N— ,_,u]l_ .................. ]
\\L ‘
0 4' ‘ \“'6"— v o | o o |
O 2 4q 6 8 10

AVERAGE HEIGHT ABOVE BOTTOM OF CONTAINER {in.)
150

ORNL-DWG 8611464

HELIUM
SUFPLY

EXHAUST
J{V-‘i -

      
    

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

}_.
o
s . ) VARIABLE ; O
Fig. 7.7. Diagram of 233pa SPEED : .
1 TO _» 7 9
. MOTOR ] z
Extraction Pump Loop. 7 V-3 EXHAUST Sl e
Ry "TTTtrTeemstToTTTTTT ’ § Z
g s -
"‘ 1 Ty 5
THORIUM 8 g 12
ACDITION-] & o
o =
= He a 5‘ -
- i 2e B
N | = =
E- B
] v v-7 Yoz ! .
~ S0l gl saLT @
T >0 4 =
az .
] Lo 5 .
1 PuvP oF o
] X0, Q
<< I Bi nqz
l__
)
=0 S

 

 

 

 

 

 

the extraction vessel. Thus, at low flow rates thorium would be introduced
at a rate regulated by its solubility in bismuth. The alloy was then
sprayed into the salt mixture so the protactinium would be extracted at

the surfaces of free-falling droplets of bismuth. A pseudo-first-order
rate of extraction was eXpected.

Although the blanket reprocessing method involves stripping the prot- )
actiniun from molten bismuth into a second salt mixture by hydrofluori- )
nation, this feature was not included in the initial pump-loop design.

Instead, the recirculating molten-metal stream was pumped through a bed -
of steel wool to provide for the collection of protactinium by absorption

on the iron surfaces or by filtration of suspended particles. This ex-

pedient was based on results of the experiments described above and on

another in which thorium metal was added to a blanket-salt mixture that

contained ???Pa. The protactinium was found uniformly distributed on

steel wool that had been immersed in the salt.

Accordingly, the steel wool column was designed to provide a largs
surface area of 1lron relative to iron surfaces exposed 1o the liguid
phaseg elsewhere in the system without unduly restricting the flow of
bismuth. The extraction vessel was also fitted with an open cylinder of
niocbium for primary coatainment of the salt mixture. The centrifugal
pump was the same as that designed and operated by F. S. Bettis, Reactor
Division, 1n similar molten-salt—molten-lead systems.

Pump-Loop Operation. Several related experiments were carried out
in the pump loop. The loop was first charged with 13.5 kg of bpismuth that
151

had been previously treated with hydrogen at 600°C for oxide removal. This
material was circulated through the system to ascertain operational pro-
cedures and pump performance characteristics. The steel wool colunr was
then prepared on its fixture, inserted into the system, and fired with
hydrogen at 700°C in situ while bismuth was static, and the balance of
the system was protected from oxide contamination by flowing helium.

The column contained 12.5 g of grade 1 steel wool having a total surface
area of 0.49 m? or about ten times that of the geometric surface of iron
that was elsewhere exposed to bismuth in the circulating system. The
salt mixture IiF-BeF,-ThF, (73-2-25 mole %) was spiked with about 1 me of
233pa from irradiated ThO, by our usual HF-Hp treatment in nickel at
600°C and with Hs; alone at 700°C. Approximately 6.7 kg of this material
was transferred into the niobium-lined extraction vessel.

Bismuth was at first circulated without a reducing agent, to estab-
1ish the stability of 233Ps in the salt mixture. Then thorium metal was
introduced in small amounts by submerging a basket of thorium chips in
the bismuth stream at the outlet of the column. Five successive additions
of thorium were made in this manner during this phase of the experiment.
The radiochemical results obtained from filtered samples of the two liquid
phases during some 30 hr of pump operation are shown in Fig. 7.8. It is
interesting to note that the 233pg, activity in the sall phase was exX-
tremely stable during the first 17 hr of loop operation and the first
three additions of thorium. After the fourth addition of thorium, the
233py activity in the salt phase decreased at a measurable rate until
apparent exhaustion of the thorium metal and continued after the fifth
thorium addition until 57% of the *°2Pa was removed. As shown by Fig.

ORNL-DWG 656-11465

 

 

 

 

 

 

23300 ACTIVITY FOUND (%)

 

 

 

 

 

 

 

120 e
£
-
o
100 ¢ - 5 e e o
»
2
© SALT PHASE Z
80 % o METAL PHASE e ) i 233 .
5 Fig. 7.8. Behavior of Pa in
~| Pump Loop During First Extraction Ex-
w - .
6o | I B . 5l periment (Th® Added to Bismuth As
! =z
? ? \ 0| Noted).
. = = z | .
5 g8 o 3 z £
o < <1 < . a _ 2
o M O_c < o.c HDI: g‘; < 8
Ew Fe FZrx o8
. - o £ =0 o
b & b = g
20 e ' — s F]
i ) |
' ¢ # + - T L
® : =
os ; J ./ Ok L. »
o L e—fpe-s-a— ¢ —o @5 * O oo
Q 5 10 15 20 25 30

LOOP OPERATING TiME (hr}
CRNL-DWG 661466

 

Wi |

 

 

Fig. 7.9. Loss of 22%Pa from
LiF-BeF,-ThF, (73-2-25 Mole %) by
Reduction with Thorium in Bismuth
as a Pseudo-first-Order Reaction.

 

 

 

 

mex—— THY ADDED AT £:=0

 

 

 

233py REMAINING IN SALT PHASE (%)

 

 

 

ol
| o0 apep AT r=%fl-fin1 |
| | { -
10 —— b p— N
0 2 4 6 8 10

EXTRACTION TIME (hr)

7.9, this removal rate recasonably agreed with a first-order rate relation
postulated for the reduction of protactinium by bpismuth droplets of con-
stant thorium concentration while falling through the salt phase. The
run was interrupted at this point to correct a malfunction in the pump.
The apparent collection of solids about the pump rotor resulted in a
seizure which could not be overcome by the low-torgue l/3-hp motor.

A compilation of material balance data on thorium added to the sys-
tem is shown in Table 7.7. Since lhe thorium baskets were coated heavily
with bismuath, the actual quantity of thorium introduced ianto the system
was estimated by volume loss rather than by weight balance. Spectro-
graphic analyses of samples taken prior to each thorium addition showed
that very little thorium or reduced lithium was retained as a soluble
component of the metal phase. Chemical analyses of salt samples from -
corresponding periods showed that the chromium concentration of the salt
rhase was reduced from 164 to less than 2 ppm after the first thorium »
addition and remained at that level for the duration of the experiment.
Concentrations of iron and nickel were virtually unchanged at about 125
and 30 ppm respectively.

During the interruption, the column of steel wool was removed from
the system for examination. Although the column had gained considerable
weight, only the head-end sections (the bottom of the column) contained
appreciable gquantities of solids. A gquantitative analysis for 233Pa in
the column was made by E. I. Wyatt, Analytical Chemistry Division, by
dissolving the entire assembly 1into an aqueous solution. Radiochemical
analyses of aliquots of this solution were related to the total volume
cf solution and the quantity of 233pg, present originally in the salt sys-
tem. On this basis, the column accounted for abhout 14% of the 237pa.
Since 43% of the 222Pa activity remained in the salt and 3% remained in
the metal, about 39% of the 233pg was apparently deposited elsewhere in
the system.
Table 7.7. Material Balance of Thorium Metal Added to #22Pa Extraction Loop

Cumuletive Equivalence

 

 

Additi Thorium Thorium horium Th Found Li Found Praction of Th
’ o0 ~otacted  Withdrawn — Added  4As Th® As Th As 10 in Metal in Metal  Accounted For
No. () () (g)  in Metal  in Salt  dn Metal (pom) (ppm) in Metal

(ppm) {ppm) (ppm)
1 8.37 2.0 6.3 468 936 57 60 18 0.45
2 8.39 2 6.4 94,3 1836 114 30 13 0.13
3 5.9 ly oty 1.5 1055 2109 127 30 20 0.17
4 5.9 1.0 4.9 1419 2836 171 30 17 0.11
5 5.0 5.0 1790 3579 216 <30 17 0.0%

 

£ot
154

The second steel wool column was assembled with coarser grades of
steel wool in the inlet section in an attempt to retain more of the
solids. In the previous column the filterable solids were stopped by the
first three of eight sections of steel wool. It seems possible that
solids held by inertia on the head end of the column settled in the pump
bowl when the pump was stopped. The steel wool on the new column had a
net weight of 18.2 g and a surface area of 0.87 m?.

The results of radiochemical analyses of filtered samples from the
two liquid phases taken during this second extraction experiment are
summarized in Fig. 7.10. Two additions of thorium during 15 hr of pumping
made no change in the 233pn concentration in the salt rhase. Because the
pump was operating poorly and the flow rate of bismuth was irregular,
subsequent thorium additions were made through the sampling port directly
to the salt phase. Three l-g additions of thorium reduced the ?23Pa con-
tent of the salt phase to about 27% of its original value before the run
was Interrupted to change out the steel wool column. During this op-
erational interval, 16% of the 22?Pa was removed from the salt phase, 4%
was in the metal phase, and about 7% was found on the column.

The steel wool column for the final extraction experiment was pat-
terned arter the preceding one. A total of 28.7 g of steel wool from
grades 3 through O provided a surface of 1.4 m?. Thorium metal was added
in three 2-g increments during 12.5 hr of pump operation. Radiochemical
analyses of filtered samples of the liquid phases are shown in Fig. 7.11.
Protactinium was removed from the salt phase until about 4% of the 233pg
remained at apparent equilibrium. During this period the pump was op-
erated at its maximum allowable speed in an attempt to wash protactinium
from the walls of the extraction vessel and to better suspend any prot-
actinium-bearing solids in the molten bismuth. Material-balance calcu-
lations show that 23% of the original quantity of 223Pa was removed from

ORNL-DWG 56-11467
MOSTLY STATIC OPERATION WHILE PUMP
// WAS SEIZED PRIOR TO 7=5

 

 

 

 

 

 

 

 

 

 

 

100 o I
-t o @ ‘ |
L |
|3 = Q. g
o - 1
80 |- - .J_ ] ;i{ . ..I__ ; S L; (2 &_
z 5 2y =i Fig. 7.10. Behavior of ?33pa j
2 L 5z e 1g. /.10, ehavior of Pa in
—_ I < > - . -y -
2 o 2 %5; | g;; g Pump T,oop During Second Extraction
0 - o o A o 7 EI o . i 0 .
2 0 |- e E “fl*sfi | e gl Experiment (Th” Added to Bi Except
o O
2 L S , i,e‘ ! Where Noted),
> o i
= A £ Y
> ’ T l 3
i 0 o 2
<<'[) aon — o . L# o = L ac
Q _J
0 SALT PHASE S
0 ® METAL PHASE | éa
l s
| -
Ll
i L
o —es”
r— - @
o L ‘ . . . . . ® N
2

0 5 1O 15 20
LOOP QPERATING TIVE (hr)

o
155

ORNL—DW3 66--11468

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O SALT PHASE
4o—§——~ ® METAL PHASE -— e e
<5
g e
Fig. 7.11. Behavior of ??°Pa o |3 i |
, . . 5 30 e S I B —
in Pump Loop During Third Extrac. @
tion Experiment (Th® Added to £
Salt Phase). 5 g
4 —-Q T —_
o oL
rlgn. o
~ T
~
T
0 ( . [
0 25 50 75 0.0 12.5

LOOP OPERATING TIME (hr)

the salt phase. About 19% of the 23°Pa was on the column, giving an
82.6% recovery for this third experiment.

Chemical analyses were obtalned on salt samples taken during the
second and third extraction experiments. Concentrations of iron ranged
randomly from 100 to 170 ppm, while chromium and nickel were virtually
absent at reported values of less than 10 ppm. The nicbium content of
the salt phase was consistently below the detectable level of 4 ppm.
Values for the concentration of bismuth in the salt mixture ranged from
67 to 264 ppm in a random manner; the arithmetic average concentration
from all samples was 119 ppm. However, it was not possible, under the
experimental conditions, to ascertain whether these values represented
dissolved bismuth or droplets suspended in the salt mixture. The values
did not, however, reflect the pumping speed of the bismuth or a dependence
on time.

Discussion and Conclusions. The protactinium bhalance for the com-
plete experiment shows that 26% of the 233pg, originally in the salt phase
was removed by thorium metal additions. At least 43% was pumped as 8 so-
Jution or suspension in molten bismuth and deposited on rather small
volumes of steel wool. Since an additional 4% of the 232pa remained in
each of the two ligquid phases, we can account for about 51% of the prot-
actinium.

 

Although large amounts of thorium were added to the system, the con-
centration in bismuth remained very low. Considerable thorium had to be
added in the first experiment bvefore any protactinium was extracted from
thie salt. Additional thorium was consumed without extraction of prot-
actinium at the beginning of the second experiment, which followed the
changeout of the steel wool column. Reduction of some metal Tlucride
impurities from the salt was evident, but the amount was not sufficient
156

to account for all the thorium losses. BHome high-melting metallic plugs
were taken from the system, and the spectrograpnic analyses showed them
to contain high concentrations of thorium asscocliated with iron and
chromium.

The collection of protactinium on the steel wool columns appeared
to be primarily a filtration process, although some surface absorption
also was apparent. We expected some mass transfer of iron in the poly-
thermal loop system, but it appears that the transfer was aggravated by
the presence of thorium. We believe that iron, chromium, and thorium,
accompanied by some of the protactinium, formed compounds of low solu-
bility in bismuth at the operating temperature. Precipitates formed,
some of them collected on the filter, but the remainder collected in
other parts of the system to account for the thorium and protactinium
losses. This suggests that a more resistant material, such as niobium,
would be desirable for use in the extraction equipment.

The pump-locp experiments wlll be continued. The next one 1s planned
to demonstrate recovery of protactinlium from the liquid-metal stream by
contacting the bismuth with a molten salt that is saturated with a mixture
of HF and Hp.

Protactinium Studies in the High-Alpha Molten-3Salt Iaboratory — C. J.
Barton

 

An experiment on the removal of protactinium from a breeder-blanket
mixture LiF-Th¥, (73-27 mole %) having an initial concentration of 25
ppm of 231pg was described in the previous report.27 Reduction of prot-
actinium was eflected by metallic thorium in the presence of lead at
about 625°C. The fact that only a small fraction of the reduced prot-
actinium was found in the liquid lead encouraged study of other reduction
techniques.

Reduction with 3o0lid Tnorium. Several experiments were performed to
study the reduction of protactinium in IiF-ThF, (73-27 mole %) by solid
thorium in the form of a rod or turnings. Three different container ma-
terials were used: nickel, copper, and graphite. The first experiment
wags conducted in a nickel container with a 3/8-in.-0D thorium rod. The
231ps content of the salt mixture, based on analysis of filtered samples,
dropped from 11.1 to 0.09 mg during the initial 65~min exposure of the
rod to the molten fluoride mixture at 625°C. A further 5-hr exposure at
the same temperature produced an apparent increase in 2°*Pa content to
0.54 mg. A large fraction {(approximately 70%, or 28 g) of the part of
the rod that was immersed in the molten-salt mixture was lost during the
experiment. We believe that the thorium rod was in contact with the
bottom of the nickel pot during this experiment, causing a current flow
that corroded the rod electrolytically. The salt mixture removed from
the nickel pot contained a large amount of black material, part of which
was magnetic. The analysis of the magnetic part showed 45% Ni, 30% Th,
and 0.015% ??1Pa. The black, nonmagnetic material contained 22% Ni,

50% Th, and 0.027% 23 pa.

 
157

A second experiment, performed under conditions described above ex-
ceplt that care was exerted to avolid contact of the thorium rod with the
nickel pot, gave more readily understandable results. The 2?1Pa concen-
tration of a filtered sample of the salt dropped to 30% of the initial
concentration (32 ppm) after a 1-hr exposure and to 19% after the second
hour of exposure. A ground sample of the uvnfiltered salt removed from
tne pot at the conclusion of the experiment had a higher concentration of
231ps than the initial filtered sample. This may have been due, in part,
to the fact that the precipitated protactiniwum was redissolved by HF-H;
treatment in 2 smaller volume of fused salt than was present at the be-
ginning of the experiment because of the removal of a significant fraction
(about 2.5%) of the salt in each filtered sample.

The next experiment was similar to the previous one except that
thorjum turnings supported in a nickel-plated copper screen were exposed
to the melt for 65 min and the 2?1Pa concentration of filtered salt de-
creased to 9% of the initial value. The screen came loose from its sup-
porting rod when we attempted to remove it from the pot, and it remained
in the molten mixture. Conseguently, 1t was necessary to dissolve the
thorium metal by prolonged treatment with an HP-H, mixture in order to
redissclve the precipitated protactinium. The 231ps concentration of a
filtered sample was 92% of the initial concentration after a 170-min
HE-H, treatment, and, after an additional 43-min treatment, an unfiltered
semple showed a content equal to 96% of the initial value.

In the fourth thorium reduction experiment, we agaln exposed the
melt containing 12 ppm of 231pg o thorium turnings held in a nickel-
plated copper screen and found that a 120-min exposure removed 97.5% of
the protactinium from solution, as determined in a filtered sample. This
time, however, we removed the basket and turnings from the melt, cooled
to room temperature with a helium atmosphere in the pot, and disassembled
the apparatus to determine the distribution of reduced protactinium. The
recovered salt contained 51% of the amount of 231py present at the be-
ginning of the experiment, the basket and contents had 20%, the wall had
7%, while the dip leg and magnetic material removed from the salt (plus
particles produced in sawing through the nickel pot) each contained avout
1%. About 16% of the protactinlum was unaccounted for.

Fxperiments very much like that deseribed in the previous paragraph
were conducted in copper and graphlite containers to study the effect of
container material on distribution of reduced protactinium. In both
cases, approximately 60% of the recovered protactinium was found in the
ground, unfiltered salt (and untreated with HF and H,), although only 5%
of the initial protactinium concentration was present in the final i1-
tered sample of reduced salt in the graphite container experiment and
29% in the copper container. It appears, therefore, that a large fraction
of reduced protactinium remains suspended in the molten LiF-Thi', mixture
regardless of the container material.

Electrolytic Reduction of Protactinium in IiF-ThF, (73-27 Mole %).
A series of exploratory experiments on the electrolytic reduction of
protactinium in molten IiF-ThF, (73-27 mole %) has been conducted with a
variety of electrode arrangements. None of these arrangements has been
explored in detail, and the preliminary results obtained in some cases

 
158

ars more confusing than enlightening. Consequently, these experiments
will be sumrarized very briefly.

Graphite anode — nickel wvessel cathode — 3.0 v and 0.5 amp: The
protactinium content decreased during the first hour of electrolysis and
then increased during the second and third hours for reasons that are not
all clear.

Silver ancde — grapnite liner or nickel dip leg as cathode -~ 3.0 v
and about 1 amp: About 20% reduction in 231ps concentration after elec-
trolyzing for 2 hr.

Thorium rod anode — thorium rod cathode — graphite liner — 3.0 v
and about 1.6 amp: Removed 95% of 231pg during 70-min electrolysis
(filtered sample), but 95% of the initial ??'Pa concentration was found
in the unfiltered salt.

Nickel rod immersed in bismuth connected to negative side of 6.0-v
battery — nickel rod immersed In the molten fluoride connected to positive
pole of battery — 5.0 v and about 2.7 amp: No reduction in protactinium
content of salt, and only a trace amount was found in the bismuth.

Another construction of the previous experiment with bismuth cathode
(i.e., nickel dip leg contacting both salt and bismuth) — graphite anode
contacting fluoride mixture only — graphite liner ~ 5.0 v and 0.5 amp:
Again, there was no significant reduction in the protactinium content of
the salt. About 0.02% of the protactinium was found in the last filtered
sample of bismuth as compared to 0.4% in the unfiltered bismuth. The
latter also contained 6.3% nickel.

Conclusions. Protactinium dissclved in molten IiF-Th¥, can be re-
duced to a form that does not pass through a sintered copper filter, but
a large part of the precipitated protactinium remains suspended in the
molten mixture.

Exploratory experiments on electrolytic reduction of protactinium
have not produced encouraging results to date, but we are continuing to
pursue this approach to the protactinium removal problem because of its
potential simplicity.

7.4 Radiation Chemistry

 

Xenon Diffusion and Possible Formation of Cesium Carbide in an MSBR -
C. F. Baes, Jr., and R. B. Evans IITI

 

Previously, several investigators have considered the neutron polson-
ing effect caused by diffusion of 135%e into the graphite moderator of a
molten-galt reactor.29732 It is the present purpose to consider as well
the effects of the cesium which is born within the graphite by decay of
the various fission product xenon nuclides which have diffused there. In
159

particular, it is of interest to estimate whether or not sufficient con-
centrations of cesium might occur to form (lemellar) cesium carbides, as
by

Cso(g) + nC(s) = Can(s) 5

and whether or not a sufficient amount of CsC, could be formed to damage
the graphite in a full~scale MSBER.

In an attempt to answer these questions, the partial pressure of
cesium within the graphite void spaces was calculated using the following
model and assumptions:

1. Diffusion of gaseous xenon into the graphite void spaces and dif-
fusion of gaseous cesium out of the graphite were approximated as
one dimensional; that is, the moderator was represented as a slab
of graphite infinite in two dimensions, with a specified thickness
(21), immersed in the fuel salt.

2. All cesium born in the graphite was assumed to be in the gaseous
elemental form. Cesium born in the fuel salt or reaching the fuel
salt by diffusion from the graphite was assumed to be oxidized to
CsT and to remain in the salt.

3. Steady-state conditions were assumed.

The resulting expressions and the parameters employed (which corre-~
spond approximately to the present MSBR reference design??) are summarized
in Table 7.8. The steady-state partial pressure of each cesium nuclide
(PCS) was a function of the depth inlo the graphite (x), the porosity of
the graphite (e), the partial pressure of the parent xenon at the salt-
graphite interface (P ), the diffusion cocefficients of cesium and xenon
(D, assumed to be the Seme for both), and the appropriate decay constants
(2) and neutron capture cross sections (o). The xenon partial pressure
at the salt-graphite interface (P%P) was, in turn, a function of several
terms: Y corresponds to the xenofiJproduction rate; S reflects the xenon
loss from the fuel salt by decay, stripping, and burnup; G reflects dif-
fusion of xenon into the graphite; and, finally, ' is a factor reflecting
the effect of the film coefficient H at the salt-graphite interface.

(This film coefficient is defined by the relationship for the flux JXP,

_ A0
JXe - H(CXe CXe) ?

wherein CXe is the concentration of the nueclide in the bulk of salt and

C§e is the concentration in eguilibriwan with P§e
terms 5, F, and G also appear in the 135%e poison Tactor, which 1s the
ratio of the 13°Xe poison fraction under the conditions specified to the
maximum possible poison fraction, approximately 0.005.

at the interface.) The

The rather cumbersome expresslions in Table 7.8 were evaluated with
a computer. The effects of variations in the gas stripping rate (AfiT),
O
160

Teble 7.8, Caleuwlation of Cs® Partial Pressure and of 1°3°Xe Poisoning

 

 

Parametersa Assigned MSRE
Values Values

¢ Average thermal-neutron flux, 7 % 10%% 1.5 x 1013
cm < sec”

Cas 235U concentration in fuel, 2 X 1074 1.8 X 1074
moles/cm3

T Temperature of core, °K 873 873

Qi Henry's law distribution ccefficient 6000 6000
for xenon

ViV, Ratio of total fuel volume to fuel 4,37 4.0
volume in flux

A/Vc Ratioc of graphite area to fuel volume 2.0 25
in flux, em™t

L Half thilckness of graphite slab, 0.5 2
cm

€ Poreosity of graphite 0.05 0.09

Ag Fraction of fuel stripped per 0.001~0.1  0.0004~0,002

' second, sac”*

D Effective diffusion coefficient for 1078074 1.3 X 1074
both Xe and Cs, cmz/sec

H Film coefficient at salt-graphite 0.002-0.02 0.00045

interface, cm/sec

 

Partial Pressure of Cs Nuclide in Craphite

0 ey\)(e

PCs - PXe —M_*““~—_~"~E*3 (ECS _-EXe) [ Six fii (ZIFX)} ( ZBiL

2
DCS (BXe 505 Ei =1 e + 2

Partial Pressure of Xe Parent at 5alt-Graphite Surface

 

 

 

 

 

PO = F"S’%“@ ) Y = ¥y U25C250RT
135y%e Poison Factor a = (VT/VC) (AXe * KST) T Oy ?
T e (1350 = Mss) Gi 2By L
135 e oD, B (e *° 1)
P, = —e—— g | o> e xe
Jr3s¢ + 7\135 H F =1 + 7 SET
5 + F‘g (e Xe 4 l)
2P, L
AD. B (e e _ 1)
G = Xe Xe
< 7 25 T
© (¢ ¢ 1+ 1)

 

a, . .
¥, 0, and A denote, as usual, the yield, neutron capture cross section, and
decay constant of a glven nuclide.
1el

Fig. 7.12. Calculated Steady-
State Pressure of Cesium at Center
of Graphite Slab as a Function of
the Gas Stripping Rate (NST)’ the
Diffusion Coefficient (D) and the
Film Coefficient (H). Other condi-
tions are specified in Table 5.8.
The upper curves show the corre-
sponding variation in the 1°5Xe
roison factor.

35%e POISON FACTOR

4
}
:

CESIUM PRESSURE {atm

ORNL — DWG 66-11468

 

 

 

H
{cm /sec) /

.02

 

 

 

ESTIMATED PRESSU

 

   

o0

)‘ST (sec~1)

 
  

 

o

" fem?/sec)

106

 

RE AT WHICH CsCy IS FORMED

the diffusion ccefficients (D and the film coefficient (H) were deter-
2

mined.

Cesium Carbide Formation. The maximum total pressure of cesium (at
the center of the graphite slab) was found to increase ama D was decreased
(Fig. 7.12). This is a varadoxical result since it is usually thought
desirable that the graphite possess the lowest possible D values. In the
range of D values tested here, however, the rate~controlling step in the
diffusion of parent xenon into the graphite appeared Lo he at the salt-
graphite Iinterface (see below) and did not depend significantly upon D.

 
162

As a result, lowering D for xenon and cesium appreciably decreased only
the diffusion of cesium out of the graphite, causing the steady-state
accumulation of cesium_(PbS) to be higher.

Included for comparison in ¥ig. 7.12 is the partial pressure of
cesium at which carbide formation might be expected at 600°C, based on
an estimate by Manowitz.?% Tt appears that in all cases tested the cesium
partial pressure in the graphite was high enough to produce carpide for-
mation. As a consequence, it is likely that the actual quantities of
cesium which could accumulate within the graphite will be higher than
estimated in the present calculations. While no detailed calculation of
this effect was attempted, considerable comfort may be Tound in the Tol-
lowing observations.

1. By the present calculations the amount of cesiwn accumalation in the
graphite is very small; even with a xenon partial pressure as high
as 0.01 atm, for example, there would be only 1 atom of cesium present
for every 25,000,000 atoms of carbon. At acceptably low *7°Xe poison-
ing levels, the calculated maximum cesium concentration would be ap-
proximately 100-fold (10~% atm) lower than this. Thus, although the
present estimates of cesium accumulation would probably be increased
if carbide formation occurs, an enormous increase (e.g., 10°-fold)
in the accumulation could occur before it would become a matter of
concern.

2. Bven if 1t were assumed that all the cesium born in the graphite were
to remain there, the rate of accumulation would be low. Thus, under
the most unfavorable conditions tested (H = 0.02 cm/sec, A, = 0.001

ST
Sec"l, and D = 1076 cmz/sec, which gave a 135Xe poison factor of 0.9),

the total of all xenon nuclides entered the graphite at a rate

v
= JXe = E%K 5 GP%e = 1.29 x 107t mole cm™? sec™* 3

so low that even if it all were to decay to cesium, 150 days would
be required to produce a cesium-to-carbon ratio of 1:1000. With

more realistic calculations and a more realistic calculation of the
geccumulation rate — in particular, a calculation which includes the
diffusion of Can as well as gaseous cesium out of the slabh — much

lower accumulation rates undoubtedly would result.
12°¥e Poisoning. For most of the values of D and H chosen here,

the rate-determining step was the transfer of xenon across the film at
the salt-graphite interface (i.e., F >> 1 and G/F — AH/V&QH). Hence

 

the poison fraction was not very dependent upon D, but rather was deter-
mined almost entirely by the magnitude of H and K%T (Figs. 7.13 and 7.14).

This result, it should be noted, 1s compatible with the assumption of one=-
dimensional diffusion in the graphite; that is, 1if the rate step instead
had been in the graphite, such a crude model would have been more question-
able.
163

ORNL-DWG 86-11470

 

H {cm /sec)
! o 0.02
0.0+
0.005
0.002
05
0.2
r
o
©
a
£
O
?
S 0
Q)
>
©w
"
0.05
0.02
0.04
1078 07 108 1073 10?
0 (cm®/sec)

Fig. 7.13. Dependence of the Xenon Poison Factor on the Diffusion
Coefficient.

The 35%e polson factor, as well as the cesium partial pressures,
could of course be reduced either by increasing KOT or decreasing H (Fig.
v

7.14); however, it seems unlikely that in practice the stripping rate
could greatly exceed 0.01 (corresponding to a removal half-time of ap-
proximately 1 min) or that H would normally be kept below 0.01 cm/sec.
Approximately these conditions evidently must be met 1f the poison factor
is to be held below 0.1 (poison fraction approximately 0.005).
ORNL-DWG 66-11471

 

     
   

 

 

 

 

 

 

OO‘F """" "“T """" CTTT T
0005 | i j{
—. 0002 |- ‘{
O
& MAXIMUM 'O xe
£ POISON FACTOR =
~ 0.001 L e
00005 | — ] ’
0.0002 |— T —_t
L1
0.0001 i e L I
0001 0002 0005 0.01 0.02 0.05 04

Agr (sec™!)

Fig., 7.14., Approximabte Boundary Values of The Stripping Rate and
the Film Coefliclent Calculated for Various Xenon Poisoning Limits in
the Absence of Todine Stripping.

Todine Removal. Removal of iodine (as HI) by HF sparging of the
salts would be an effective way of reducing the 135%e poison fraction,
provided the overall removal rate is large compared with the decay rate
of *3°T (2.89 x 107% sec™, t1;p = 6.7 hr) — and it seems likely that this
could be done.?? The amount of the reduction in the case of <220 fission-
ing could be as much as & factor of 5, limited by the 18 * 3% of the “*°Xe
being produced directly Trom fission.

This treatment is relatively ineffective in reducing the cesium
partial pressure, however, because the principal contributors to it are
mass numbers 137 and 138, which have short-lived iodine precursors (A =
0.03 and 0.12 sec™ respectively).

Salt Impregnation. Tapregnation of the graphite to a limited depth
with the fuel salt would be an effective means to reduce the rate of
xerion diffusion into the graphite by virtue of the fact that both the
xenon diffusion coefficient and the xenon concentration would be roughly
four corders of magnitude lower in an interstitial salt phase than in an
interstitial gas phase.

 

 

This situation can be repreéesented within the framework cof the present
calculations by the simple procedure of considering the interpenetration-
salt—graphite barrier as the film to which H refers. Comparison of the
equation which defines H in a normal film,

J =1 (-~-c?),
165

with the equation which would define the diffusion coefficient in the salt-
graphite barrier,

Y-

shows that 1t is reasonable to approximate H for such a barrier by
H~ D/t ,

where © is the depth of salt penetration. The effective D for this salt-
graphite barrier should be in the range 1077 to 108 cmz/sec for MSRE-
type graphite (i.e., D for xenon in the molten salt times the porosity/
tortuosity factor). Assuming a penetration depth of 0.1 cm, we can then
place H in the range 107° to 1077, which is at least three orders of
magnitude below the minimum film factor estimated for an MSBR.

The consequences of such a small film coefficient are that diffusion
of xenon into the graphite is no longer significant in determining the
135%e poison factor (i.e., the terms containing F in the expression for
the poison factor become negligible). The 133ye poisoning is then deter-
mined only by %ST:

0135®
"o (VT/VC) (M35 + ?\ST) + 01350

 

P.F

Summary. The present calculations indicate that cesium carbide Tor-
mation can be expected to occur in an MSBR, but in such small amounts as
to be of little concern. In addition, these calculations indicate that
in the absence of iodine removal, xenon poisoning in a full-scale MSBR
will he controlled primarily by a film coefficient H and will be reduced
most effectively either by iodine removal or by some method which in ef-
fect reduces this film coefficient.

Fission Product Behavior in the MSRE — S. S. Kirslis

 

The behavior of fission products in the M3SRE is being studied to
derive information bearing on the practical concerns of corrosion and
neutron poisoning. Significant clues to the basic nature of Hastelloy N
corrosion in a fissioning molten-salt environment are provided by obser-
vation of the volatilization and plating behavior of fission products
whose oxidation states are relatively easily changed. The chemical fate
of a number of fission product poisons 1s important in determining their
contributions to the overall polsoning of a reactor. Thus, Tor example,
it is of interest to determine what fraction of the T3%Xe produced by
fission in the melt 1s released to the cover gas before 1t becomes 136%e
by neutron capture. Iikewise, it is 1mportant to determine whether
noble-metal fissicon products which are moderate neutron poisons (Mo, Ru,
Te, Nb, Pd, and Rh) remain in the circulating fuel, volatilize, or deposit
166

on graphite or Hastelloy N. The rare-carth fission product poisons, be-
cause of the chemical stability of their nonvolatile fluorides, are ex-
pected to remalin in the circulating lfuel.

There arsc unique advantages in carrying out fission product studies
in an operating reactor rather than in small-scale laboratory or in-pile
tests. Iirst, in small-scale tests it 1is difficult to mock up realisti-
cally the geometry, the fuel circulation conditions, and other possibly
significant factors of the reactor environment. Second, although the MSRE
was not primarily designed for the chemist's convenience, several features
make it rather well adapted to chemical studies. Most importantly, a
facility exists for taking sizable samples of fuel salt from the pump bowl
during reactor operation. In the same facility 1t is possible to expose
selected metal sampies to both the cover gas and the fuel melt in the
punp cowl. Provisions exist for taking samples of the reactor cover gas
and are being improved to permit continuous monitoring and the taking of
concentrated samples. Finally, provisions werec made for Inserting re-
movable long-term surveillance specimens of Hastelloy N and graphite in
the reactor core of the MSRE. These facllities and advantages are avail-
able only with great difficulty in small-scale in-pile tests. The prin-
cipal lacks from the chemist's viewpoint are facilities for direct sam-
pling of the pump bowl cover gas and for exposing metal and graphite sam-~
ples to -the pump bowl atmosphere and fuel phase for longer periods of
time.

This report on fission product behavior will cover the initial re-
sults from fuel salt sampling, from the exposure of metal samples to the
pump bowl gas and liquid phases, from the first examination of the long-
term survelllance specimens of graphite and Hastelloy N, and from the
analysis of the first sample of reactor cover gas taken during reactor
operation.

Ffuel Salt bamples. Using the pump bowl sampling facility, a large
number of 10- and 50-g fuel salt samples were taken during reactor op-
eration to monitor changes in the concentrations of oxide, bulk fuel
components, and corrosion products. Metal samples were attached to the
10~g sampling assembly for observation of the volatilization and plating
behavior of Tission products during five of the salt samplings. These
five salt samples were delivered to the analytical hot cells within a
few hours of sampling and were rapidly powdered, weighed, dissolved, and
analyzed radiochemically for the 15 isotopes listed in Table 7.9. The
strontium and cerium isotopes were delermined as fission monitors, since
these elements have convenient half-lives and stable fluorides which
ghould remain in the melt. The ruthenium and molybdenum isotopes were
representative of noble metals. The tellurium and icdine isotopes were
of interest for their volatilization and plating properties as precursors
of 13%Xe. The 239Np and *29Pu isotopes were measured as indicators of
the epithermal fiux in the MSRE.

 

The data listed in Table 7.9 show that the strontium and cerium
isotopes behaved in a regular and expected manner. The °lor and *43ce
isotopes appear to be the most reliable fission monitors for interrupted
power operat.ion. The average of the four values for °lsr at the nominal
 

 

 

Table 7.,9. TFission Products in MSRE Salt Samples
Sarple TPo=-17 P6-19 FP7-7 FP7-10 FP7-12
Date and sampling time 5-23, 040 5-26, 0400 6=27, 0243 7-6, 0208 7-13, 0336
Operating time,a days 2.0 2.5 13.3 el 1.9
Nominal power, Mw 5.6 7.3 ey 7.2 7.2
Fission .
Yield Disintegrations per Minute per Gram of Sal1t°
(%)
9.67-ar “iSe 5.81 1.08 x 10+ 1,20 x 0% 1.16 x 10t 1,31 x 10t 1.32 x 10t
2.6-hr °%gr 5.3 9.8 x 10 1,19 x 0% 9,70 x 1019  1.51 x 10*? 1.49 X 10+t
51-day 2%r 4,79 1.80 X 10%0 2,23 x 10*° 2.96 x 0%  2.93 x 100 3.96 x 10%9
33-day *4iCe 6.0 2,65 % 1040 6.10 % 10*°  6.69 x 10*° 6.88 % 10%°
33-hr *43Ce 5.7 1.45 x 0% 1.5 x 10 1,43 x 10t 1.32 x 10%*
66-nr ?Mo 6.C6 .68 X 10%9 3,51 x 10 2 9,51 x 10*9 1,10 x 10* 3.15 x 10%9
39,7-day *%%Ru 3.0 1.81 x 0% 7,05 x 10° 2.42 X 107 6.02 x 107 7.14 X 107
lidB-hr *O9Ry 0.9 4.9 X 10%% 2,47 x 10** 2 3.50 x 100 9.67 x 10%° 3.76 % 1049
1.0L-year 19%Ry 0.38 0 2.13 x 10"
77-nr 1327 AR 3,12 ¥ 10F0 4,21 % 10%° 5.15 x 16*% 3,64 x 10%° 3,81 X 10%°
8.05-day 13T 3.1 2.45 X 100 4,20 x 100 5,0 x 10Y0 4,50 x 1010 5.36 x 1049
20,8-hr 1331 6.9 1.08 x 10% 1,31 x 10+t 1.35 X 10**  1.40 x 10** 1.45 x 10
6.7-nr *3°1 5.1 9.5 x 10*% 1,50 x 0%t 1.17 X 10%* 1.11 x 10tt
2,33-day 231 5,35 X 10% 1.0 x 102 1.08 x 10%2
2.44 X 10%-year Pu 7,00 X 107 ¢ 8,77 x 10° ¢ .20 x 10% ¢

 

a . . X . . - . e s .
Continucus cperating time since previcus shutdown of more than 12 hr duratlon or change in power

level.
b

Calculated at sampling time.

C .
Alpha counts per minute per gram.,

LOT
168

Table 7.10. Bummary cof Pump Bowl Test Results

Averages of five runs

 

 

99MO lSZTe lO'SRU. 1033ua lOéRu 135I lBBI 1311
Salt, % of theoretical 60 30 >1.00 30 15 90 1.00 98
Lateh (Ni) 8 14X 10X 10X — 0.5X  6X 02X 1.5%
Silver 2X 6X 3% 3% — 1X 2% 01X 0.9x%
Hastelloy N 1% 7 3% 25X — 0.3% 11X 0 2x 0.5%
Liguid phase (ss) 4K X 5% 40X — 1X 1X 0 2X 0.8X

 

Spercent 2Ry deposited on metal samples uniformly decreased from first to fifth

run.
bX = disintegrations per minute of given isctope per gram of salt.

7.2-Mw power level is 1.25 x 101 dis min™?t g™*. Using the formula

Mw = [(1,25 x 10 dis min"l ¢71) % (4.68 x 10° g)

L

x (200 Mev/fission) x (1.6 x 10719 . M )J 3+ [(fractional
ev/sec

fission yield of ?lgr) x (60 sec/min)] ,

the calculated fission power density is 5.4 Mw, or about 75% of the

nominal total power density. Using the average of the four 14306 values

at the nominal 7.2-Mw power, the calculated power is 6.3 Mw, or 87% of

the nominal value. -

For molybdenum and ruthenium isotopes, the concentrations found in
salt samples showed much more scatter, which may well be real. The first
row of data in Table 7.10 shows the averages of the amounts of these
isotopes Tound compared to the amounts calculated to be formed using
91gy (arbitrarily) as the [ission monitor in each run. The reason for
the impossibly hign values for 20°%Ru is being sought. The 1327¢ values
varied little among the 7.2-Mw runs but represented low fractions of the
total amount formed (Table 7.10). The iodine isotopes showed relatively
little scatter, and the amounts found corresponded well with the amounts
calculated from the 2'Sr values.

The epithermal flux in the MSRE has not yet been calculated from
the neptunium and plutonium values in Table 7.9.

Pump Bowl Volstilization and Plating Tests. Advantage was taken of
the experimental possibilities of the pump bowl salt sampling facility
to carry out some gualitative tests to detect the presence of chemically
reactive fission product species in the pump bowl cover gas and to deter-
mine which fission products would plate on clean metal surfaces from the
fuel salt phase.

 
169

The copper fuel salt sampling ladle 1s attached by a double stain-
less steel cable to a nickel-plated irvon latch which fits into a mecha-
nism for lowering and raising the assembly in a pipe leading from the
pump bowl to the sampling cubicle above the reactor. To detecli the
presence of chemically reactive fisgion products in the gas phase, colls
of 0.015-in.-diam silver and Hastelloy N wire were wound on the small
sbainless steel cables fTor a distance of 2 1n. below Lhe bottom of the
latech. The latch and the wire colls were later leached and analyzed for
gagsaous Tisslon products. The lower 2~in. lengths of the stainless steel
cables Just above the copper ladle were leached and analyzed similarly
to determine which fissgion products plated from the fuel melt. Figure
7.15 shows the sampling assembly with wire coils attached. The gampling
device was submerged in the pump bowl for times verying from 1 min to
10 min. It was thsn-ralo d 2 ft and allowed to cool for 10 min and then
ralised to the sampling cublcle. The metal and salt samples were delivered
in & carrier to the hot analytlcal laboratory, usually within 3 hr of the
sampling time. The salt sample was prepared for analysis by the procedure
summarized above. The stainless steel cables were clipped to provide
separate samples of the lateh, the silver coll, the Hasztelloy N coil, and
the stainless steel cable exposed to the fuel melt. The metal samples
were leached Tilrst with an alkaline mixture of Versene, boric acid, and
citric acid to remove iodine without volatilization. Then they were
leached with a warm mixture of HNO3 and HC1 until the radiocactivity was
less than 1% of the original reading {(ususlly greater than 500 r/hr at
contact). Dilutions of the original leach solutions were sent to the
radiochemical separations group for overall gamma scans and estimations
of’ several lfldLVldUdl Tzotopes.

The overall gamma scans showed that the principal sctivities in the
metal samples exposed to the pump bowl gas phass were 1227 and ]321
with smaller amounts of 21T and ??Mo. The *4PRa-*40ra and 2%7r- 95Nb
peaks, which were prominent in the gamma, spectra of fuel zalt samples,
were much lower in the metal samples. This indicates that the observed
activities were not due to condensation of fusl salt mist on the metal
specimens. Similarly, the gamma spectra of the submerged stainless steel
samples were quite different from those of Tuel salt. Correspondingly,
no adhering particles of fuel salt were visible on any of the mebal speci-
mens. It was later found {Table 7.10) that the amounts of tellurium,
rutheninm, and molybdenum deposited on the metal samples corresponded to
the amounts in srfgrdW grams of fuel salt.

The gamma scans indicated that quantitative radiochemical estimations
for *2%0e, T and Y%Mo would ve useful. In addition, estimations were
made of 1331 1357, leRu 105Ru) and °%Ru. The results of these analy-
BC8 are oummdrized in Table 7.10. Although the exposure times for the
five runs were 1, 2, 5, 10, and 10 min, there was no corresponding vari-
ation of amounts of the several isotopes deposited from either the gas or
the Jiguld phases. The scatlter of res 1t3 for a given isotope between
runs was large, often a factor of 2 and sometimes a factor of 4 or more.
For these reasons, the averaged results given in Table 7.10 are more

casily interpreted than =z tabulation of the individual results. As In
the case of some of Lhe salt analyses, the oObserved scatter is much
greater than the normal error of radicchemical analyses and undoubtedly
170

PHOTO 85018

 

 

0 1 .. "

Fig. 7.15. Pump Bowl Deposition Testing Assembly.

represents real large variations in the amounts deposited. The causes
for these variaticns are as yet unknown.

The results in Table 7.10 reveal a striking degree of volatilization
and plating of molybdenum, ruthenium, and tellurium on clean metal sur-
faces. Iarge amounts of égMo were found on the metal samples in both
the gas phase and fuel phase of the pump bowl. Correspondingly, only
about 60% of the theoretical yield of ?°Mo was found in the fuel salt
171

samples. The results for ruthenium were similarly spectacular. The

fractions of the three ruthenium isotopes remaining in the salt decreased
with increasing half-1life, suggesting that slow reactions are removing
ruthenium from the melt. In successive runs, the amount of 41-day 1°3Ru
that deposited on metals steadily decreased from the first run to the last
(although the exposure time of the samples increased from 1 to 10 min), as
if the amount in solution were decreasing with time. This behavior is not
explained.

For 122Te the amount remaining in the melt is also low (30%), and
the corresponding gas- and liquid-phase depositions are high. For 1311,
1331, and 13511 most of the element remains in the melt, although high
activities of 1321 were observed on the gas-phase specimens. The short-
lived 221 is, however, a daughter of the relatively long-lived l32Te,
which was shown to volatilize readily. Rather less 1337 and 1317 was
found on the metal specimens than of molybdenum, ruthenium, and tellurium.
However, no 1351 at all was detected on the metal specimens. Corre-
spondingly, the tellurium precursor of 1357 has a half-1life of only 0.4
min, while 133Te and *?!Te have half-lives of 63 and 24 min. Apparently
very little tellurium can volatilize or plate in 0.4 min. It thus appears
that the volatilization and plating characteristics of iodine are deter-
mined by the behavior of the precursor tellurium.

Tellurium and iodine had been detected in the gas lines of previous
in-pile tests, but the volatilization of molybdenum and ruthenium was
surprising. The only volatile compounds of molybdenum and ruthenium are
those of valence 4 or greater. The plating behavior of tellurium, mo-
lybdenum, and ruthenium also indicates that these elements are present
with valences greater than zero. However, all positively charged ions
of these elements should be reduced to the metallic state by the metallic
chromium in Hastelloy N. These considerations suggest that the Hastelloy
N container vessel and piping of the MSRE may be protected by a layer of
noble-metal fission products, permitting the existence of higher oxida-
tion states in the melt. If it is assumed that 30% of the noble metals
produced by fission in the MSRE operating at 7.2 Mw are deposited on an
estimated metal surface area of 10° cm?, it may be calculated that the
metal film would grow at the rate of about 0.3 A/hr. It is thus not un-
likely that all the metal surfaces exposed to fuel in the MSRE are
presently coated with several hundred angstroms of noble metals.

It is not a simple matter, however, to produce adherent pinhole-free
electroplates, so that a plate hundreds of atoms thick might not protect
against the reducing action of the base metal. Furthermore, 1t has been
calculated on a very reasonable basis that approximately 1% of the uranium
in the fuel in the MSRE should exist in the trivalent condition. Noble-
metal ion concentrations should be infinitesimal in the presence of this
concentration of the strongly reducing U3+, These very basic puzzles in
elucidating noble-metal fission product behavicor In the MSRE have not yet
been solved.

Future pump bowl experiments will involve looking at the behavior of
several more elements (noble metals, other cations with volatile fluorides,
neutron-activated Hastelloy N corrosion products, etc.) and attempts to
172

take small samples of the pump bowl cover gas. The determination of the
reducing power of the fuel salt would also be helpful in explaining fis-
sion nioduct behavior.

Examination of the Graphite Survelllance Specimens. A package of
MSRE graphite and Hastelloy N surveillance specimens has been exposed to
a fissioning molten-salt environment along the central axis of the MSRE
for 7800 Mwhr of power operation. After the reactor shutdown of July 17,
1966, the package was removed from the reactor and disassembled in a hot
cell. Rectangular bars of graphite, 5 to 9 in. long, 0.66 in. wide, and
0.47 in. thick, from the bottom (inlet), middle, and top (outlet) of the
reactor core were made available for detailed examination. The surveil-
lance specimens were contained in a perforated cylindrical tube of
Hastelloy N, 5-1/2 ft long, 2 in. in diameter, and 0.030 in. thick. Rings
of this tube approximately 11/16 in. in height and 10 g in weight were .
sawed out of the bottom, middle, and top regions of the tube to provide
Hastelloy N specimens for fission product deposition studies. -

The graphite bars were first sectioned transversely with a thin
Carborundum saw to provide specimens for photographic, metallographic,
autoradiographic, x-radiographic, and surface x-ray examination. Speci-
mens were saved for other possible tests (spark spectroscopy or electron
probe). The remainders of the bars, 7 in. long for the middle specimen,
2-5/8 in. long for the bottom sample, and 2-7/8 in. long for the top sam-
ple, were used for milling off successive surface layers for fission
product deposition studies.

(a) Visual, Autoradiographic, and X-Radiographic Examinations. The
graphite specimens showed no visible signs of corrosion or chemical
change. No metallic or salt films were discernible under low-power mag-
nification, which revealed the original machining grooves on the graphilte.
The middle sample (Y-2) was known to have a crack near one end before in-
sertion into the reactor. The crack was still visible after removal but
had not propagated further. Some of the graphite specimens in one region
of the package had cracked due to mechanical stresses caused by the un-
even thermal expansion and contraction of the tightly packed Hastelloy N
and graphite specimens. A small piece of the bottom sample (VA-1) had
cracked off during the disassembly of the package. These cracks should
not reflect on the overall integrity of the irradiated graphite speci-
mens. Prints of an autoradiograph and an x radiograph of each of the
three graphite specimens are shown in Figs. 7.16 to 7.18. The x radio-
graphs were taken on thin (0.020- to 0.060-in.) transverse slices of the
graphite bars. The autoradiographs were taken on adjacent transverse
slices after mounting in epoxy resin and polishing for metallographic
examination.

The x radiographs show no general deposition of fuel salt nor of
heavy elements on the surface of the graphite. A hint of penetration or
deposition is visible near one corner each of the bottom and top graphite
samples. The crack in the middle sample is clearly penetrated by fuel
salt. The edge of the cracked bottom sample shows no foreign material,
indicating that the crack occurred after removal from the reactor.
 

(a)

ig. 7.16., Autoradiograph (a) and X Radiograph (b) of Middle CGraphite
(v-2).

The autoradiographs indicate a thin film of highly radicactive ma-
terial on the exposed surfaces of the graphite and confirm the presence
of salt in the crack of the middle specimen. These pictures also show
that the penetration of the radlioactive material into the interior of the
graphite is by no means unliform. The nonuniform nature of the porosity
of graphite has been demonstrated in gas permeability tests.

The observations described here are generally in accord with those
from previous in-plle tests. [The graphite is not visibly affected, and
there are no signs of chemical attack, film formation, or salt genetration.

(b) Metallographic Examination. Metallographs of transverse sections
of the three graphite bars are shown in Figs. 7.19 to 7.21. The stiructure
of the graphite in all cases appeared normal and undamaged under both
bright-~-field and polarized-light illumination. No metallic, carbide, or
sall films were visible on the surfaces of the specimens. No sign of salt
penetration was observed except in the case of the cracked middle speci-
men. Here voids were oObserved which probably had been filled with salt
before polishing with a water suspension of the polishing compound. The
other graphite specimens were polished under CCl, to avoid dissolution of
fuel salt.

 
 

7.18.

 

s
S
0

.&‘.&%"‘
R
&‘s*::'%

i
"t

O
o

e : ,_,&,:?:,—j_“‘g. o
.

: S e 2 :
fi%fi%‘#@ : i R

: oI

-

e
e
.

!

S '-‘.;.%

 

Autoradiograph (a) and X Radiograph (b) of Botbtom Graphite
175

O

™|

 

X008 o«

SAHINI LCOC

 

 

sl X00§ =
SR OO0

 

)

L]

:,

/M

—~

@

o

§

b

~

@

2

o

g

£

o

&

O

w

—

T

ke

i

0

0

w0

=

a2y

3 .

& £

O

el

— —l

@

LT

M J
o
~

+ @

oy~

—~ O
- By

o~
P

D

o

By en
T
l
T

T
AR
R

o

Tha

e

 

e

 

 

 

 

X¢

y

05 o
LO0C

R

G

e

e,

 

 

 

 

 

t field;

) Brigh

a

(

20. Metallographs of Top Graphite {(VH
zed light

7

Fig.
(b) polar

i

i
177

 

 

D.007 INCHES
EQOX

=

 

 

 

 

CHzS
ox

J

0.007 1k

n

 

 

 

Fig. 7.21. Metallograpns of Bottom Graphi .
. 4 . phite (VA-1). o
field; (b) polarized light. ( ) (a) Bright
178

A fuel-salt-exposed surface of the middle graphite bar was mounted
for hot-cell x-ray examination of the surface for metallic films or other
contamination. Only graphite lines were observed. Because of the nega-
tive result of this test, the other graphite specimens were nol examined
by the x-ray technique.

(¢) Milling of Surface layers of Graphite. The valusble contribu-
tions of J. G. Morgan, M. F. Osborne, and H. E. Robertson in planning,
developing hot-cell methods for, and starting the work on the sampling of
the graphite specimens are gratefully acknowledged. A very ingenious
"planer" was designed and built by the Hot Cells Operation Group for mill-
ing thin layers from the four long surfaces of each of the graphite bars.
The cutter and the collection system were so designed that a large fraction
of the graphite dust removed was collected. By comparing the collected
weights of the graphite samples with the weight loss calculated from the
initial and final dimensions of the graphite bars and their known den-
sities, the average sampling losses were 4.5% for the middle bar, 18.9%
for the top bar, and 9.1% for the bottom bar.

 

The patterns of sampling of the graphite surface layers are shown in
Fig. 7.22. An identifying groove was cut along the length of the middle
of the graphite surface which was pressed against another grapnite sur-
face in the original surveillance package. The other ftThree surfaces were
exposed to circulating Tuel salt. The numbers of the layers in Fig. 7.22
indicate the order in which the layers were cut from the graphite bars.
After each layer was cut from a bar, the milling apparatus and the bar
were vacuumed to avoid cross contamination of samples. Rach powdered
graphite sample was placed in a small capped plastic bottle and weighed.
The middle bar was measured with a micrometer to determline the depth of
each cut. This time-consuming operation was omitted for the other two
bars, since it became evident that the depth of cut could be more
accurately calculated from the weight of the sample removed. The average
depth of cut was 00,0075 in. for the middle bar and 0.011 in. for the
other two bars. A clean unirradiated MSRE graphite specimen was sampled
with the milling device in the standard manner after the ninth cut and
after the last cut on the middle graphite bar to provide comparison sam-
ples to indicate the level of cross contamination in the hot-cell milling
operation. As seen in Fig. 7.22, the salt-exposed surfaces of the middle
bar were sampled to a depth of six or seven layers, or about 0.050 in.
The surfaces in contact with graphite were more thoroughly sampled in the
other bars.

(d) Fission Product Analyses in Graphite Samples. The powdered
graphite samples were delivered to the Radiochemical Separations Group,
and weighed portions were dissolved in a hot mixture of HNO; and ,50,.
The gases evolved were passed through a condenser, a charcoal trap, and
an alkaline solution to recover any volatilized iodine or ruthenium.
Aliquots of these solutions were analyzed radiochemically Tor the isotopes
listed in Table 7.11. The first milled sample of each graphite surface
was also analyzed for uraniuwa by the fluorometric method. No uranium was
detected with a sensitivity limit of about 30 ug of uranium per square
centimeter of graphite surface. This finding will be confirmed by more
sensitive neutron actlvation methods.

 
179

ORNL-DWG 66-11378

TOP GRAPHITE

 

e IDENTIFYING GROOVE

 

 

 

O O Mepleo]es

MIDDLE. GRAPHITE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- 0470 in,
|
|

 

 

 

.| BOTTOM GRAPHITE
A

 

 

 

 

Fig. 7.22, Scheme for Milling Graphite Samples.

The radiochemical data so far reported are given in Tables 7.11 to
7.13 for the middle, top, and bottom graphite bars respectively. The
data are generally internally consistent for a given bar, and the agree-
ment between bars is good. Very few individual values are out of line,
which is to the credit of those involved in the sampling and in the anal-
ysis of the samples.

From the neutron poison staandpoint, the results of most interest are
those for Mo, *°2Ru, and ?°Nb. It is seen ia Tables 7.12 to 7.13 that
the concentrations of these isotopes are considerable in the first layer
and that the activities fall off by a factor of about 100 in the seccnd
layer. Tellurium-132, which is also noble in the sense of possessing
fluorides of only moderate stability, behaves rather similarly, but its
concentration drops off less rapidly with penetration distancs. DBeyond
a penetraticn depth of about four cuts, the activities of molybdenum,
180

ruthenium, and nicbium approach the contamination background (samples 10
and 24 in Table 7.11), while that of 1327 is distinctly higher. It is
pregsently thought that the deposition of molybdenwn, ruthenium, and proba-
bly niobium in the graphite may be due to the reaction of volatile hexa-
fluorides or pentafluorides with graphlite, depositing lower fluorides or
carbides. There is no chemical reason 1o expect the lower fluorides of
molybdenum or ruthenium to plate on graphite. On the other hand, the
pump bowl experiments definitely established the volatility of molybdenum
and ruthenium, probably in the forms of MoFg and Rufs.

The behavior of l4oBa, SQSr, lélCe, 14408, and 13705, all of which
have xenon or krypton precursors, is distinctly different. The gradient
of activity with penetration distance is much less, and it appears to bear
a relationship to the half-1ife of the rare-gas precursor involved. Thus
the gradient for 14084 with a 16-sec 140%e precursor is much steeper than
that for 89Sr, which has a 3.2-min %Kr precursor. Apparently, the
longer-halif-1life rare gases can achieve a Tlatter gradient by diffusion
before they decay to the observed isotope, which is assumed to remain
where deposited. The penetration data for 1%YBa and 8%Sr fitted well a
simple diffusion model which led To Xenon and krypton diffusion coeffi-
cients of about 1 x 107 ftz/hr in MSRE graphite. To explain why 13708,
which has a 3.8-min '?7Xe precursor, has a much flatter profile than 8%sr
in Table 7.11, it may further be postulated that ?7Cs itself diffuses,
rather than remaining where it was born. The internal *?7Cs concentration
of 2 x 107 dis min™* g%, or about 1 atom of *27Cs per 108 atoms of graph-
ite, should have negligible chemical effect on the integrity of the graph-
ite structure. Only a few values have yet been obtained for %7r. ‘This
element is expected to remain in the fuel salt, with little tendency to
volatilize or plate. Thus the amount found in the graphite should repre-
sent only injection by fission recoil. Correspondingly, the amount in
the graphite is low compared to the noble metals.

The concentration of T2'I in the graphite is alsc low at the surface,
and its §radient is similar to that of *??Te. It is possible that the
24-min 21Te diffused into the graphite as TeFg or TeF, and then decayed
to AL, However, it is difficult to see why the iodine did not diffuse
vack out of the graphite, unless 1t formed a nonvolatile compound with
other fission preoduct atoms in the graphite. While it is not possible at
this time to account satisfactorily for the Tact that the ilodine concen-
trations in graphite are low, the implication of the low 1317 concentra-
tions is that the 12°I concentrations will be similarly low. Thus little
13%%e will be born in graphite due to the previous immigration of 1357,

Since the depths of cut of the many graphite samples were not the
sane, especially for different bars, it is not casy to compare the amounts
of the various isotopes in the first layer from the data in Tables 7.11
to 7.13. BSince, also, the main practical interest is in the noble-metal
fission products, the bulk of which were deposited in the first layer,
Table 7.14 was prepared, giving the amount of each isotope in the first
layer in disintegrations per minute per square centimeter of graphite
surface. The values in Table 7.14 were obtained by multiplying the values
in Tavles 7.11 to 7.13 (disintegrations per minute per gram) by the weight
of sample analyzed (corrected for weight loss in sampling) and dividing
by the measured area of the surface from which the sample was cut. The
181

Table 7.11. Radiochemical Analyses of Middle Graphite Bar (Y-2)

 

 

 

 

Weight Depth of Disintegrations per Minute per Gram of Graphite
Sample Cut ,
(e mits) 990 132.0, 1035, 95\ 1314 95 144, 89g, 140, 1410, 137
VWide Face Exposed to Circulating Fuel
1 0.8463 6.02 1.88 x 1012 1.09 x10*2  2.57 x 10! 6.93 x10!' 128 x10°°  1.12 x10'? 5,08 x10° 128 x10'! 144 x10't 3.7 x10'® 8.39 x107
4 1.2737 9.27 1.28 x10*°%  5.19 % 10'?  2.40 x10° 9.16 x 10° 9.45 x 108 8.43 x 107  1.10 x10'!  6.57 x10'?  6.53 x10° 2,02 X 107
7 0.9814 7.50 2.35 x 10° 2.33 x 1019 4.24 x 108 7.76 X 10° 4.17 x10° 2.80 X107  7.65 x10'%  2.75 x10*%  1.43 x10° 1.98 % 107
11 0.9145 6.94 5.53 x 108 1.07 x10'°  1.52 x 10% 2.94 x 108 1,77 x 108 2.22 x 107 7.08 x10'%  1.65 x10'°  7.33 x 108 2.17 x 107
14 0.8962 6.98 1.20 x 109 8.60 x 107 3.17 x 108 4.28 % 10° 3.18 x 108 4.07 X107 4.55 x10'% 9,71 x10° 5.55 % 108 1.86 x 107
17 1.0372 8.25 1.13 x 10° 8.27 x 10° 2.18 x 10° 3.16 x 107 1.79 x 108 2.46 X 10° 430 x10'%  7.60 x10° 3.90 x 108 1.96 x 107
23 0.8176 6.64 1.12 x 10° 8.76 x 10° 1.73 x 10% 2.43 x 10% 7.74 % 10°
Side Face Exposed to Circulating Fuel
2 0.7583 7.68  1.40 x10'?  1.04 x10'?  2.14 x10'! 833 x10'' 1,20 x10'° 2.48 x 107 1.41 x10**  3.38 x10'°
6 0.9720  10.10  2.70x10'° 462 x10'?  s5.92 x10® 152 x10'%  8.98 x10° 7.68 x 10'°
8 0.5139 5.43  3.73x10° 2,41 x10'%  6.33 x 108 2.79 x 10° 3.00 x 107 3.86 x 10 2,50 x10°
12 0.3395 3.65  2.44x10'° 379 x10'%  4.55 x10° 1.36 x 109 2.99 x 1010
15 0.6976 7.61  466x10” 111 x10'%  7.38 x10° 2.75 x 10% 2.54 x 107 1.52 x10'%  8.39 x 10°
18 0.3737 4.15 2,41 x10'% 305 x10'%  4.76 x10° 6.62 % 10° 2.29 x 10"
Other Side Face Exposed to Circulating Fue!
3 0.6135 6.21 1.73 x 1012 1.09 x10'?  2.60 x10''  6.36 x10'"  1.61 x10'7 2.45 x 10° 1.75 x 10" 3.35 x10'°
5 2.7543 7.84  5.94x10'°  7.01 x10'®  8.55x10°  2.80 x10'%  1.21 x10? 7.18 x 1010
9 0.6198 6.55  853x10° 196 x10'®  5.52 x10° 1.79 x 10% 2.18 x 10’ 4.28 x10*°  3.01 x10°
13 1.1652 12.53  6.26 x10° 870 x10%  1.15 x10° 2.05 x 10° 1.00 x 10%7
16 0.8469 9.24 561 x10°  8.02x10°  7.25%10° 1.41 x 108 2.04 x 107 4.98 x10°  3.77 x10°
19 0.9406  10.45  5.08 x10°  5.65x10°  6.89 x 10° 6.99 x 107 4.23 x10°
Face in Contact with Graphite
20 1.1381 9.25  3.22x10'!  s5.82x10'"  7.88x10'% 104 x10''  1.36 x10'?  6.68 x10°  1.65 x10° 8.96 x10*°  2.15 x10'°
Unexposed Graphite Blanks
10 4.63 % 10° 2.62 x 10% 7.18 x 107 2,46 % 10° 3.26 % 107 4.48 % 10° 1.07 x10°  2.03 x10° Low 1.41 x 10°
24 6.47 x10°  3.22x10® 811 x107 328 x10° 126 x10°  1.31 x10°  8.06 x10° 3.21 x107  6.38 x10°
NOTES: 1. The samples are arranged in order of successive cuts on each face (see Fig, 7.22).
2. 'The sample weights given here have been corrected for the average 4.5% loss during milling,
3. The depths of cut were calculated from the sample weights, areas, and the known graphite density.
4. Additional analytical results are forthcoming on selected samples for 9SNb, 89Sr, QQTC, 952r, 147Nc§, 136Cs, 137Cs, 91Y, °3Ni, and ?Fe.
5. The activities tabulated are corrected to the time of shutdown, 11:00 AM, Jfuly 17, 1965,
182

Table 7,12, Radiochemical Analyses of Top Graphite Bar (VH.5)

Disintegrations per Minute per Gram of Graphite

 

 

 

Sample  Weight  Depth of
{g) Cut {mils) %90 132m 1055, 95N 1314 957, 1440, 89g, 1405, t4la, 1374
Wide Foce Exposed to Circulating Fuei
25 0.3602 6.23 1.54 x 1012 @89 x10°!  g.02x10'!  is6x 10! 4.87x10°  1.28x10%  440x10®  1.19x 10!t 1.04x10''  860x10°  2.24 x 107
29 0.4355 7.93 5.45x 107  9.24 x 10° 6.30 x 10° 1.06 x 10° 4.92 x 107 5.74x 101%  1.60 x 161° 1.22 x 107
58 0.5260 9.94 5.95 x 10° .39 x 10° 7.76 x 108 9.54 x 103 6.54 x 107 6.19x 10'?  1.98x10'®  1.87 x10%
60 0.2916 5.51 1.73 x 10° 3.95 x 10° 1.64 x 108 3.58 x 107 3.95x 101" 7.19 x 10°
62 1.0783 20.38 6.22 x 10° 1.63 x 107 7.96 x 167 1.34 x 107 1.5: x 1089  1.67 x 10° 3.54 x 107
Side Foce Exposed to Circulating Fue!
26 0.4615 11.41 4.94 x 1017 438 x10'!  1.37x10%! 2.37% 10% 2.59 x 10% 4.96 x 10'® 4,94 x 10°
31 0.4564 11.88 2.24 x 10° 9.06 x 10° 2.82 x 108 7.26 x 107 6.11 x 108 1.45x 10" 2,13 x 108
Other Side Face Exposed to Circulating Fuel
28 0.6703 17.11 6.10 x 10'1 5.7t x10%'  7.81 x 10'° 5.44 x 107 2.69 % 108 6.38 x 10'% 5,38 x 10°
33 0.5404 14.43 2.20 x 107 5.39 x 10° 2.32 x 108 3.01 x 107 8.27 x 10°
Wide Face in Contact with Graphite

27 0.6422 11.32 3.86 x 10*Y 2,80 x 10'Y 514 x10'%  ss50x10't 3.07x107  1.24x10°  2.74 x 10® 3.29 x 1010 4,03 x 10°
32 0.5375 9.95 1.956 x 107 1.03 x 10'°  1.91 x 10 2.28 x 108 9.26 X 107 8.51 x 10°
59 0.3154 5.96 1.81 x 107 1.24 x 10*® 2,01 x 10° 6.25 x 108 7.31 x 107 8.75 x 10° 2.56 x 10°
61 0.5835 11.03 9.17 x 108 3.95 x 107 1.17 x 10° 1.86 x 107 2.76 x 107
63 0.7310 13.82 6,08 x 10% 2.41 x 10° 8.16 x 107 1.1t x 107 1.58 x 10° 2.17 x 107

NOTES: The samples are arranged in order of successive cuts on each face (see Fig. 7.22).

Additional analytical results are forthcoming on selected samples for QSNb, 898r,

The sample weights given have been corrected for the average 18.9% weight loss during milling.

The depths of cuf were calculated from the sample weights, areas, and the known graphite density.

99TC' 957, 1475a, 138cs 137cs 91y S3Ni and S9Fe.
The activities tabulated are corrected to the time of shutdown, 11:00 AM, July 17, 1966,
183

Table 7.13. Radiochemical Anciyses of Bottom Graphite Bar {VA-1)

 

 

 

 

Weight Depth of Disintegrations per Minute per Gram of Graphite
Sample - Cut ;
(g) (mns) QQMO 132Te 103Ru QSNb ].EH.1 QSZr 144Ce SQSr 140Ba I41Ce 137CS
Wide Face Exposed to Circulating Fuel
34 0.8032 15.04 5.30 x 102! 4.39 x10'1 8.30 x10'%  3.34 x10'Y 4.25 x10° 2.40 % 10° 6.39 x10%  4.17 x10'?  3.34 x10'%  8.11 x107 2.80 X 107
38 0.5979 11.64 5.67 x 10° 3.26 x10'?  8.22 x10® 2.49 x 10° 4.44 x 108 3.86 x10'?  1.65 x 1019 1.09 x 107
64 0.2323 4.68 9.5 x10° 1.91 x10'%  9.76 x 107 3.12 x io® 2.63 x 108 3.52 x10'%  1.05 x10'°  6.92 x10°
56 0.3120 6.28 8.65 x 10°% 1.91 x10*%  1.20 x 10% 1.40 x 10% 3.62 x 1019 1.13 x 100
69 0.7183 14.49 3.39 x 108 1.06 x10*?  5.14 x 107 8.01 x 107 3.08 x10!%  5.87 x10° 1.84 x 10°%
Side Face Exposed to Circulating Fuel
35 0.3904 10.70 6.62 x10'*  5.31 x 10! 9.44 x10'° 7.09 x 10° 8.32 x 10° 4.18 x10'%  1.04 x 1019
39 0.4480 12.98 7.56 x 10° 4.12 x101%  s5.86 x 10° 5.44 x 108 2.21 x 1029
Other Side Face Exposed to Circulating Fuel
37 0.5480 15.39 4.3% x 10'Y  3.64 x 10!  5.11 %3010 4.58 x 10° 6.19 x 10° 6.12 x101%  9.19 x10°
41 0.3520 9.62 3.95 x 107 1.81 x10'% 4,95 x 108 1.43 x 108 1.38 x 109 7.25 x 10%
Wide Face in Contact with Graphite
36 0.4810 9,12 0.31 x 10! 5.97 x10''  1.00 x10''  4.20 x10'!  6.01 x10° 3.47 % 10° 9.49 x 108 4.48 x10'%  1.07 x10'°
40 0.5936 11.77 8.23 X 10° 3.70 x10'%  1.06 x 10° 2,50 % 10° 5,01 x 108 2.04 x 1019
65 0.4756 9.58 1.57 x 10° 2.44 x16°  1.74 x 108 3.07 x 10% 4.31 x 108 1.48 x10*? 1,10 x 108
67 0.4025 8.10 1.83 x 10° 2.15 x10'%  1.48 x 108 1.31 x 108 1.24 x 101¢
68 0.6260 12.61 2.66 x 108 6.90 x 10 3.83 x 107 4.96 x 107 4,50 x 10° 1.03 x 108
NOTES: 1., The samples are arranged in order of successive cuts on each face (see Fig. 7.22).
2. The sample weights given have been corrected for the average 9.1% weight loss during milling,
3. The depths of cut were calculated from the sample weights, areas, and the known graphite density.
4, Additional analytical resulits are forthcoming on selected samples for QSNb, 898:‘, 99Tc, 95Zr, 147Nd, 136Cs, 13'7Cs, le, 63Ni, and SgFe.

5. The activities tabulated are corrected to the time of shutdown, 11:00 AM, July 17, 1966.
184

Table 7.14. Fission Product Deposition on Surface® of MSRE Graphite

 

Top Graphite

Middle Graphite

 

 

Disintegrations per

Disintegrations per

Bottom Graphite

 

Disintegrations per

 

fsotope Minute per Square Percent of Total® Minute per Square Percent of Total? Minute per Square Percent of Total®
Centimeter Centimeter Centimeter
9910 3.97 x 1010 13.36 5.14 x 1017 17.24 3.42 x 1010 11.5
1327 3.22 x 101° 13.84 3.26 x 10*° 13.60 2.78 x 10'° 12.0
193gu 8.34 x 107 11.40 7.53 x 10° 10.32 4.75 % 10° 6.30
Nb 4,62 x 109 12.00 2.28% 101? 59.2 2.40 x 10°° 62.4
131 2.00 x 108 0.162 4.22 x 108 0.328 3.27 x 108 0.252
Szr 3.78 x 107 0.326 3.14 x 108 0.270 1.72 x 108 0.148
e 1.58 x 107 0.0516 8.26 x 107 0.268 4.36 x 107 0.142
89gr 3.52 x 107 3.24 3.58 x 10° 3.30 2.99 x 10° 2.74
140p, 3.55 x 10° 1.38 4.76 x 10° 1.85 2.93 % 167 1.14
4ice 3.16 x 108 0.194 1.03 x 10° 0,632 5.83 x 10° 0.356
137¢s 6.62 x 10° 0.070 2.35 x 108 0.248 2.01 x 10° 0.212
60(."10 in
adjacent 4.37 x 10° 1.57 x 109 6.69 x 10°

Hastelloy N,

a
i
i

dis min~ " g

-1

 

a u " s
Average on the three salt-exposed surfaces for first milied cut.

Ppercent of total amount of isctope in reactor system which is deposited in the 2 x 106 cm? of graphite surface that is in the reactor.
185

deta were condensed by averaging the usually well-agreeing amounts Found
on the three salt-exposed surfaces. The amounts found on the other
graphite-contacting surfaces were, on the average, 38% lower. The latter
figure indicates rather free circulation of salt between the contacting
graphite surfaces. Oalt may be expected to circulate egually well in the
interstices between the graphite stringers of the MSRE.

It is seen from Table 7.14 that the amounts of noble metals in the
top, middle, and bottom samples of the graphite agree rather well, with
pernaps usually somewhat larger values for the middle graphite. For the
other isotopes, also, the middle graphite usually holds the highest sur-
face concentrations, though seldom in the ratio of the prevailing neutron
fluxes. The last row in Table 7.14 gives the °°Co activities in the
Hastelloy N samples adjacent to the three graphite samples. These numbers
are proportional to the thermal-neutron fluxes. The latter have not yet
been calculated since the chemical cobalt analyses have not yet been re-
ceilved.

Also listed in Table 7.14 are the percentages of the total isotope
produced by Tission found on the graphite surface, on the assumption that
511 2 x 10° cm? of graphite held the surface concentrations listed to
their left in the table. The total amcunt of each isotoove was calculatad
for a fission rate based on the °'gr activity (1.32 X 101+ dis min~t =)
in the last salt sample taken on July 13, 1966. The percentages in Table
7.14 would be sbout 20% lower if the nominal power bistory of the reactor
were used to calculate the total amount of each isotope produced by fis-
sion. It should be recognized that the percentages glven actually repre-
sent the behavior of each izotope only during a few half-lives of that
isotope. Thus the figures for 66-hr “?Mo charscterize only a period of a
week or so before the July 17, 1966, shutdown. It is quilte conceivable
that the deposition behavior of the various isotopes is changing with
time. Several sets of survelllance specimens must be examined before [irm
conclusions may be drawn.

Sizable percentages of the total r”9I\/Io, 132Te, lOBRu, and 2°Nb were
found on the graphite surfaces. Many of the values were close to 13%, but
two of the ?7Nb values were near 60%. Much lower percentages of the other
isotopes deposited on the graphite, the highest being 3.2% for 893y and

1.8% tor T49pa,.

An attempt vwag made to znalyze two of the surflace graphite samples
for all fission product isotopes of molyovdenum by mass spectrometbry, after
spiking the samples with °2M0 or 28Mo. The results indicabed somewhat
more molybdenum deposition (20 and 32% of the total) than was measured
radiochemically with “?Mo. However, confidence in thesze resulbs was
lessened by the fact that distinetly different combinations of fission
product molybdenum, natural molybdenum, and natural zirconium (used as a
carrier) had to be used to fit the mass analyses of the two samples.

(e) Fission Product Deposition on Hastelloy N. It is interesting
to compare the deposition of fission products on graphite with that on
the adjacent Hastelloy N. The samples of the perforated Hastelloy N ob-
tained as described above were weighed and dlssolved in a warm mixture of
HNG3z and HC1L. From the known dimensions and density of Hastelloy N, it
was calculated that the surface-to-welght ratio of the perforated metal

 
Table 7.15. Deposition of Fission Produets on Hastelloy N

 

Top

Middle

 

Disintegrations

per Minute

Percent of

Hastelloy N

 

per Square Total® Graphite

Centimeter
o 2.12 % 10%! 42,8 5.3
1326 5.08 x 10°} 131 15.8
19304 3,55 % 10 © 29.3 4.3
13y 8.24 » 107 3.82 39,4
957, 1.85 % 10° 0.96 48.9
lag 5.22 % 107 0.019 .14
144, 1.07 x 107 0.020 0.68

60Co in
adjacent 9
4,37 < 10

Hastelloy N,
:: o—1 =1
dis min g

Disintegrations
per Minute
per Square

Centimeter

2.76 % 10" °
3.41 x 1011
2.55 x 190
2.7 x 107
1.84 x 10°
2.24 x 10°

8.95 % 107

Percent of

Tota

55.6

88

1.84
0.95
0.07

0. 17

1.57 x 10

Bastelloy N

1 Graphite

5.3

i.4

3.4

9.4

5.9
i 0.22
5 1.1
10

 

per Minute
per Square
Centimeter
2.04 % 101

4.27 % 10"}

2.32% 100
5.24 % 10°
2.5% % 107
@

1.50 % 107

3.51 % 107

Disiniegrations

Bottom

Percent of Hasteiioy N

 

 

Total Craphize
41.2 5.9
110 15.3
19 4.9
2.44 5.0
1.32 15.0
0.055 0.25
0.068 9,81
6.69 x 10

.. , . . . : o . . . 2 . : : . ;
APercent of total isotope produced by fission in the reactor which was deposited on 1.2 % 107 cm” of Hastelloy N surface in the reactor, assuming
deposition on all surfaces is the same as on Hastelioy N in the core.

hRatio of the dis min_l

em™?% of each isctope on Hastelloy N to that in the first milled cut of the adjacent graphite.

981
187

was 4.30 em®/g. Using this figure the anslytical results (in dis min~?*
o~ %) could be converted to dis min™t em™%. Table 7.15 gives the radio-
chemical results obtained to date from the solutions of the Hastelloy N
specimens. It is seen that large percentages of the total 2“Mo, 132Te,
and 103Ry deposited on the Hastelloy N. Also given in Table 7.15 are

the ratios of the surface concentrations of each isotope on the Hastelloy
N and on the graphite. Between 3 and 14 tTimes as nuch molyodenum,
tellurium, end ruthenium deposited on metal as on graphite. It Is in-
teresting that a falr material balance is obtained for 29Mo by adding the
vercentages of the total produced found in the salt (approximately 60%,
Table 7.10), in the graphite (approximately 14%, Table 7.14), and on the
Hastelloy N (approximately 45%, Table 7.15). The agreement for the other
isotopes is not so good.

Miuch lower percentages of the total 1311 deposit on the Hastelloy N,
but the amount compared to that on gravhite i1s high. The data suggest
that the precursor tellurium deposits on the metal and that not all of
the 1317 is able to leave when the tellurium decays. The amount of 237
on the metal was low, butbt higher than the surface concentration on graph-
ite. Very little 14iCe and f440@ were found on the metal, in fact, less
than on graphite. The fioding of more cevium isotopez in graphite may be
due to their short-lived xenon precursors.

General Discussion of the Deposition Results. The principal prsctical
interest 1n the fission product deposition results lies in thelr implica-
tions regarding neutron poiscning by noble metals in the graphite cores
of molten-salt reactors. In most studiez of the physics of reactors such
as the M3BR, it has been assumed that the ncble-metal fission products
would either plate out instantanecusly on metal surfaces Or e removed
periodically from the fluoride fuel by the Tluoride volatility orocessing.
Under these conditions, the poisoning effect of noble metals would be of
minor practical concern. However, it has been calculated®® that if all
of the elements 3e, Br, No, Mo, Te, Te, and I remained in the core of the
- MSBR, the neutron poiscon fraction (capture by these elements per absorption
in fissile material) would be 0.0969 in two years, 0.1652 in five years,
0.2104 in ten years, and 0.2585 at equilibrium. IT 10% of these materials
remainad in the core, the respective poison fractions would be 0.0106,
0.0172, 0.0226, and 0.0275. If this group of slements remained in the
fuel and was periodically removed by fiuoride volatlility processing, the
average poison fraction would remain constant at 0.0015. The isotopes of
mo lybdenum and 9%Te gecount for about 90% of the polsoning effect of this
group of elements. The nuclear breeding ratio for the MEER is expected
to be 1.05 to 1.07 in the absence of deposits of noble metals in the core.
The numbers zmbove indicate that the actual breeding ratio or the time
that the graphite can be left in the core of a breeder can be considerably
influenced by the deposition of this group of elements.

 

1t was also reported above that five to six times more 29Mo was de-
posited on Hastelloy N surfaces than on adjacent grapnite surfaces. In
the MSRE there is approximately 2 X 1 % em?® of graphite surface (about
half in the fuel channels and hall in the flats pressing against adjacent
flats) and 1.2 x 10°% cm? of Hastelloy N surface exposed to the circulating
fuel. In the current MSBR design the ratio of metal surface to graphite
188

surface exposed to circulating fuel is about 2. It seems likely that the
fraction of molybdenurm depcsited on the gravhite in the MSBR core would
be correspondingly lower than in the MSERE. The ratio of metal to graphitce
surface arca coula be further increased, but at the expense of fuel in-
ventory, by circulating the fuel through a chamber packed with finely

divided metal.

Anotner approach to the problem of molybdenum deposition in graphite
is suggested by considerations of the probable chemistry involved. The
volatility of 290 demonstrated in the pump bowl tests suggests the
presence of Morg Or MoF; in the circulating Tuel. The gaseous form of
molybdenum would explain 1ts observed penctration to appreciable deptiis
into the graphite surfacoe. On the basis of available thermcchemical in-
formation, it is not clear what chemical reacticn occurs to fix the
molybdenum to the graphite. For example, the free energies of the reac-
ticns of Molbg with graphite to form CI, and molybdenum or Mol, are posi-
tlve by a foew kilocalories. The same is true for Rul's and NbFs. Only in
the case of tellurium is there a definltely negative free-energy change
for the reaction of the [luoride with graphlte. Possible chemical ex-
planations of the observed depositions are the formation of stable mixed
alloys of the noble-metal fission produclts or the reactions of the [luo-
rides with reducing impuritics present in the grapnite. In any case,
these reactions take place inside the graphite pores near the surface.
The penetration of volatile molybdenum could be decreased by making the
graphite surface less permeable. 1t is hoped that grapghite speclmens
whose surfaces have been made less permeable by impregnation treatments
can be included in futurc surveillance specimen packages to test whether
molybdenum deposlltion in graphite can te decreascd in this way.

It the deposition of noble-metal Tission proaucts in grapnite does
indeed reguire thelr prior conversion to high-valent [luorides, a third
method of deposition control suggests itself: +the fuel melt may be made
more reducing by increasing the U3+/U4+ ratio. Fxisting thermochemical
data indicate that a sizabkle percentage of the tetravalent uranium could
be converted to the UPF state without causing uranium carbide formation.
A small percentage of the urarium could be made trivalent by adding a few
percent of Hz to the helium cover gas of a reactor. Alternatively, when
fresh fuel is added to compeansate for burnup, the added uranium could be
partly trivalent. One or both of these procedures may be Teasible in the
MSRE. The effects on volatilization and plating behavior could be easily
observed by pump bowl tests with metal and graphite specimens.

The second major area in wnich illumination is desired from fission
product behavior is Hastelloy N corrosion. As indicated above, 1t 1is
difficult to reconcile all the observations with a single consistent
chemical pvicture. The principal difficulty is that the high-valent forms
of molybdenum, ruthenium, and tellurium indicated by thelr volatility and
plating behavior may not exist in equilibrium with the appreciable con-
centrations of U’ estimated to be present in the MERE fuel salt. The
oxidizing effect of burning up about 400 g of ?3°U to date i1s equivalent
to only one-tenth of the UeT originally present in the Iuel.

If it is assumed that the U’V concentration was actually mucen lower
than estimated, the presence of high-valent noble metals in the fuel im-
plies that the Hastelloy N surfaces have been protectively plated with
189

noble metals. Experimental observations of such platings were reported
apove. This protection would explain the observed low corrosion rate of
Hastelloy N as indicated by chromium analyses of the fuel. However, the
absence of iodine velatilization is then difficult to explain. In the
presence of MoFg, RuFs, and Telg, it is expected that icdides would be
oxidized to Ipz. It is not expected that the solubility of I, in the fuel
is high encugh to prevent its velatilization.

These problems of consistency make it clear that the nature of
Hastelloy N corrosion reactions in an operating reactor is not under-
stood as well as would be desirable. Of great help in solving some of
these problems would be the development of a method to measure the Ut
concentration in radloactive fuel salt samples.

MSRE Cover-Gas Analyses. Samples of MSRE cover gas were Ilsolated
on June 23, 1966, during steady 7.2-Mw operation in the three shielded
gas-sampling bombs provided for this purpcse. The three samples (500,
500, and 1500 em’ ) represented pump bowl cover gas which had passed
through the first holdup volume (68 ft of 4-in. stainless steel pipe),
the particle trap, and the charcoal filter. The latter should have re~
moved all heavy hydrocarbons and other impurities more easily adsorbed
than xenon.

 

The gas analysis of prime interest was that for the l36Xe/134Xe
ratio, from which the fractional burnup of 135%e in the MSRE could be
determined. To obtain high-sensitivity mass spectrometric analyses for
136ye and 134Xe, it was necessary to concentrate the sampled gas by a
factor of at least 50. The concentration was accomplished by adsorbing
the impurities in a helium cover-gas sample on a small volume of molecular
sieve sorbent at liguid-nitrogen temperature, then warming the sorbent €O
500°F, and flushing the liberated impurities with helium into & small (20-
cem®) sample bottle. In this way, one 500-cm? sample was concentrated by
a factor of 25 and the 1500-cm® sample by a factor of 75. At the same
time small unconcentrated samples of the cover gas were taken for gamma
spectrometry (1 em?) and for mass spectrometry (2 cmB) to detect impuri-
ties such as H;0 which are not readily liberated from the warmed molecular
gsieve sorbent. The latter sample also provided more reliable analyses for
very low-boiling impurities like Hp, which would not be completely sorbed
by the molecular sleve.

After cooling for more than two months, the radicactivity of the gas
samples was low, and no activity problems were encountered in the direct
gampling and concentration procedures. The activity in the gamma spectro-
metric sample, taken in a thin-walled l-cm? glass bulb, was preponderantly
S.27-day 1334, The observed counting rate indicated a concentration of
about 7 ppm of 133%e in the cover gas at the time of sampling. Minute
tracez were also detected of a2 0.16-Mev activity which might be 12-day
13Myve and of a 0.5-Mev activity which might be 10.3-year 85Kr, 40-Aday
103ry, or 1.0-year 106Ry.

The results of the mass analyses of the unconcentrated sample, the
concentrated 500-cm’ ssmple, and the concentrated 1500-cm> sample are
shown in Table 7.16. From the ratio of the analyses for 136xe ang **%Xe
for the concentrated 300-cm? sample, it was calculated that 7.7% of the
120

+35%e produced in the fuel melt captured neubrons to become *30Xe. The
corresponding more accurate result from the 1500-cm® sample was 7.9% T35%e
burnup, with a standard deviation of about 0.5% caleculated from the esti-
mated analytical accuracy. It should be emphasized that this method of
determining 1355 oburnup 1s not affected by sample contamination or by

the exact value of the concentration factor, provided the concentrated

Table 7.16. Mass Spectrometric Analyses of MSRE Cover Gas

 

Sample No. OG-8 0G-5 oG-7
Concentration factor 1 25 75

Constituent, P

Hy C.022 0.81 0.49
He 99.71 93.65 S5.27
CH, < 0.005 0.44 0.23
H>0 < 0.005 0.058 0.021
Hydrocarbons 0.005 0.010 0.004
N, + CO 0.1 3.87 2,70
Oz 0.066 0.49 0.26
Ar 0.003 0.043 0.028
CO» 0.006 0.53 Q.54
Kr < 0.005 0.018 0.083
Xe < Q.01 0.078 0.38

Isotopic Analyses
Sample No. Theor. 0G-5 OG-6

Constituent, %

83yr 14.1 14.0 14.05
84y 25.93 26.8 26.52
85gy 7.60 7.6 7.4
8oy 52.37 51.6 51.96
131ye 13.42 8.8 9.08
132ve 20.06 14,7 14.72
134%e 36.92 41.1 40.90

136ye 29 .59 35,4 35.30

 
-

191

seample contains sufficient xenon for a good measurement of isotople ratios.
The low +2°Xe burnup values are consistent with indications from MSRE re-
activity measurements and are much lower than values predicted theoreti-
cally for the case of no helium bubbles circulating with the fuel salt
through the MSRE core. Recent measurements indicated an appreciable
bubble volume fracticon in the circulating fuel.

From Table 7.16 it is seen that the observed fractions of 1?'Xe and
132%e were much lower than the fractions calculated from the flssion
yields. In the case of 131Xe, this is well acccunted for by the fact that
the precursor &.05-day 1317 had not yet reached its equilibrium concentra-
tion when the sample was taken. For 132Xe, the observed value was some-
what lower than is indicated by a similar explanation involving the 77~hr
132me precursor. The observed krypton isotopic percentages matched closely
those calculated from fission yields (Table 7.16). The observed ratios of
total xenon to total krypton were lower than the theoretical 5.7, proba-
bly because of the xenon precursor effects mentioned above. It may be
noted that the total xenon and total krypton concentrations were a factor
of nearly 5 higher for the 1500-cm® sample than for the 500-cm? sample,
whereas a factor of 3 was expected.

The concentrations of all impurities in the three analyzed samples
should have been in the ratiocs of their respective concentration factors.
This was generally not true. Except for total xenon and total krypton,
the impurity concentration in the 500-cm” sample was considerably more
than the expected one-third of the 1500-cm® sample concentration. Most
analyses for impurities are therefore uncertain by a factor of 2 or more.

The analyses for air (Np, 0,, and Ar) were higher than expected for
the thoroughly leak-checked sampling and concentration system. A pure
helium sample is being concentrated in the system to provide a quantitative
plank for the alr values.

The analyses indicate more than 60 ppm of Hp and COp and more than 30
pp of CH, in the MSRE off-gas. Presumably, these gases were generated
from the thermal and radiolytic decomposition of the organic materlals
which caused plugging in the off-gas system. The low concentrations of
nigher hydrocarbons indlcate that the particle trap and charcosl filter
were still effective on June 23, 19606.

7.5 Development and Evalvation of Analytical Methods
for Molten-Salt Reactors

 

 

Determination of Oxide in Radiozctive MSRE Samples — R. F. Apple, J. M.
Dale, and A. &. Meyer

 

The equipment for the determination of oxide?” in highly radiocactive
MSRE salts has been transferred to the High-level Alpha Radiation Labo-
ratory and installed therein. Since installation, one coolant salt and
192 -

Table 7.17. Oxide Concentrations of Coolant
and PFuel Salt from the MSRE

 

1 oncentration
Sample Code Oxide Conce o1

 

(ppm)
Coolant salt CP-4-4 25
Tuel salt I'p-6-1 £9
FP~6-4 53
FP-6~12 50 -
FP-6-18 457
FP-7-5 66
FP-7-9 59 .
rp-7-13 66 _
FP-7~16 56

 

eight fuel samples have been analyzed for oxide. The 50-g fuel salt sam-
ples were taken at levels from zero- to full-power reactor operations.
With an in-cell radiation monitor, the initial sample read 30 r at 1 ft.
This activity increased to 1000 r at 1 £t at the full-power level. Re-
sults of the oxide analyses are given in Table 7.17.

The oxide 1n the coolant sglt sample, 25 ppm, is comparable to a
value of 38 ppm obtained for a coolant salt sample taken on January 25,
1966, and analyzed in the laboratory after three weeks' storage. The fuel
analyses are 1n reasonable agreement with the samples analyzed on the
bench top before the reactor was operated at power. The oxide in these
nonradioactive samples gradually decreased from 106 to 65 ppm. Between
the FP-6 and FP-7 series the sampler-enricher station was opened for
maintenance, and the apparent increase in oxide concentration (ca. 15 ppm)
may represent contamination of the samples by residual moisture in the
sampling system; however, the number of determinations has not been suf- -
ficient to establish the precision of the method at these low concentra-
tion levels.

In an attempt to determine whether radioclytic fluorine is removing
oxide from the fuel samplies, sample FP-7-9 was removed from the transport
container and stored in a desicecator for 24 hr prior to analysis. Since
the oxide content, 59 ppm, is comparable to that of the remaining samples
for which analyses were started 6 to 10 hr after sampling, no significant
loss of oxygen 1s indicated. A more direct method of establishing the
validity of these results by measuring the recovery of a standard addition
of oxide will be attempted when reactor operations are resumed.

A method for verifying the performance of the capillary gas stream
splitter and the water electrolysis cell in the remocte oxide apparatus
was developed. A tin capsule containing a known amount of 5n0O, 15 heated
to 550°C in the hydrofluorinator as hydrogen is passed through the system.
e

193

The SnO, is reduced to the metal, and the water formed passes on to the
electrolysis cell. Two standard samples of S5n0,; were analyzed with a
four-month interim, and oxide recoveries of 96.1% and 95.6% were obtained.

It is probable that this slight negative bias is due to momentary
interruptions in the flow of the hydrofluorinator effluent gas through
the water electrolysis cell. Difficulty with cell plugging was en-
countered throughout the period of development of the oxide method. As
an attempt to eliminate the negative bias and also to provide a replace-
ment cell for the remote oxide apparatus, it was deemed necessary to find
& method of regenerating the electrolysis cell which would permit a
steady gas flow at relatively low flow rates.

The water electrolysis cell contains partially hydrated P05 in the
form of & thin viscous film in contact with two spirally wound 5-mil
rhodium electrode wires. The wires are retained on the inside of an
inert plastic tube forming a 20-mil capillary through which the sample
passes. The 2-ft-long tubing element is colled in a helix inside of a
5/8—in.-diam pipe and potted in plastic for permanence.

During the course of the investigation of the cell, 1t was found
that a wet gas stream in iteelf did not cause the electrolysis cell to
plug. It was also necessary for current to be flowing through the cell
for flow interruptions to cccur. This indicated that the hydrogen and
oxygen evolving from the electrodes create bubbles In the partially
hydrated P05 film, which then grow in size sufficient to bridge the
capillary and form an obstructing film.

After many unsuccessful approaches, an acceptable answer to the
problem was obtained by means of a special regenerating technique using
dilute acetone solutions of H3PO, as the regenerating solutions. This
provides a desiccant coating sufficient to absorv the water in the gas
stream and gives a minimal amount of flow interruptions during electrol-
ysis. The cells which have been successiully regenerated in this manner
have yielded oxide recoveries of 99.6 * 1.3% from standard Sn0, samples.

Spectrophotometric Studies of Molten-5alt Reactor fuels — J. P. Young

 

Studies pertaining to a continuous spectrophotometric determination
of U(IIT) in circulating MSBR fuels have continued. A general description
of the optical design of a facility for performing this determination has
been discussed.’® The facility will be used in conjunction with a com-
mercial instrument, a Cary model 14H recording spectrophotometer. During
this period further and more detalled discussions of the optical problems
involved have been carried out with the designer of this spectrophotometer
and have culminated in a purchase order issued to them for the development
of suitable sample-space optics. The apparatus which will result from
this requisition is to be delivered in six months and will provide optimum
light gathering power and optical design for use with a double convex
lens~-shaped drop of liguid. The apparatus will be used with existing in-
strumentation and will be compatible with the optical path extension which
willl be required in the final proposed reactor installation.
194

More detailed studies of the spectra of U(III) in molten 2IiF-Bef,
have provided better resolution of the various absorptions in the complex
ultraviolet absorption peaks of U(III). As reported before, the maximum
abvsorption is 360 nm, but shoulders are observed at approximately 310,
445, and 508 nm. These shoulders might be of analytieal value if as yet
unknown interferences vprevent the use of the absorption at 360 nm.

On the basis of caleculations made to estimate the probable valence
state cf Tission products, it is expected that dissolved fission products
will be in one of their more normal valence states and therefore will
cause no interference at the concentrations expected. Concerning unusual
oxldation states of rare-ecarth fission products, cursory spectral studies
have been carried out with Sm{IIl} in molten 2LiF*Be¥,. This lower valence
state of samariwa exhibits a strong, broad absorption peak with a maximum
absorption at 325 nm with a shoulder at approximately 470 na. The molar
absorptivity is not yet known, butl it is believed to be greater than 200.
TIf conditions are such that Sm(II) is rresent, possible interference with
a determination of U(IIT) at 360 nm may be encountered, depending on con-
centrations present. Interference of lower-valent rare earths will not
be a general problem, but only a problem with specific ions; for example,
Fu(Tl) exhibits nc absorbance at wavelengths above 300 nm.

Although it has been assumed from other specltral studies; mainly in
the solvent ILiF-NaF-KI', that corrosion product ions would not interfere
with the proposed determination, experimental verification has not been
available until this period. 'The molar absorptivity of Fe(Il), Cr(II),
and Ni(II) at their wavelengths of maximum absorbance 1s 5 at 1020 nm,

& at 760 nm, and 10 at 432 nm respectively. None of these dissolved
species will interfere at concentration levels of 10 to 100 times that
expected in the fuel salt. In general, the spectra of these 3d jons can
be interpreted as arising from essentially octahedral coordination in

the case of Ni(II) and Cr(II) and distorted octahedral symmetry in the
case of Fe(IT). A cursory spectral study of Cr(III) was made. The molar
absorptivity et 1ts wavelength of maximum absorbance, 706 nm, is approxi-
mately 7; Cr(III) is not expected to be present in the fuel salt.

Tn a study of the ahsorption spectrun of Fe(II) in several different
IiF-Bel's soluticns, an extranecus peak was noted at 432 nm. The position
of the peak suggested an Ni(IT) jmpurlty in the melt. Based on a subse-
quent spectral study of Ni(II) spectra in these melts, a molten~-salt
spectrophotometric analysis of Ni{II) concentration was possible. The
comparison of this determination to wet chemiceal analysis of the same
samples is given below.

Ni(II) (% w/w)

Molten-Salt Spectra Wet Chemical

P il

 

Sample

 

Fe-IBog 0.21 0.19
FE"IBBB 0.18 0-22
195

These melts were made in graphite contalners but were stirred with
a nickel stirrer. It would seem that this high nickel contamination,
found originally by the mollen-salt spectra, may have arisen from the
stirrer.

An extremely sensitive absorption peak attributed to U{IV) which
causes total absorption of most ultraviolet light has been found at 235
nm. The molar absorptivity of this peak is approximately 1500. The peak
has been observed in aqueous abscorption spectra st 207 nm,39 but has not
been reported previcusly in any molten-salt solvent. A possible analyt-
ical application of this absorption would be a continuocus spectrophoto-
metrlic monitoring of coclant salt for leakage of uranium-bearing fuel
salt.

Voltammetric and Chronopotentiometric Studies of Uranium in Molten ILil-

 

BeF,-ZrF, — D. L. Manning and Gleb Mamantov*

Electrochemical reduction and oxidation of U(IV) in IiF-BeF,-ZrF,
(64-34-1.8 and 65.6-29.4-5.0 mole %) was investigated by rapid-scan
voltammetry and chronopotentiometry. Well-defined and reproducible
current-voltage curveg and potential-time curves were cobtained at concen-
trations of uranium as high as 0.8 mole % (MSRE fuel). From the data
obtained so far, the results are very encouraging from the standpoint of
utilizing the voltammetric approach as a means for in-line monitoring of
uranium in molten Tluoride systems of interest to the molten-salt reactor
program. Utilizing a platinum indicator electrode, the relative standard
deviation of peak current Tor 41 runs over a period of 36 days was 2.0%;
considerably better precision (standard deviation 0.76% for 8 runs) was
ocbtained over a period of 2 hr. Other indicator electrodes tested in-
cluded pyrolytic graphite, molyodenum, tungsten, and tantalum. Of the
electrodes tested, better reproducibility was obtained at platinum or
platinum-10% rhodium.

At 500°C the reduction of U(LV) at platinum is a reversible one-
electron process, as determined from Nernstian log plots and the diag-
nostic criteria of linear sweep voltammetry. Alsc from chronopoten-
tiometry, a plot of the current-density—transition-time product (igT) vs
(transition time)l/z vielded a straight line, which is in agreement with
theory for a reversible electrode process. TFrom the slope of the line,
the diffusion coefficient for uranium was calculated to be about 1.5 X
1078 em?/sec, in good agreement with the value obtained by voltammetry
(1.5 to 2.0 x 10™° cm?/sec). The effect of temperature on the diffusion
coefficient was determined over the range 480 to 600°C; and from a plot
of log D vs 1/T the activation energy which corresponds to the reduction
of U(IV) to U(III) was found to be approximately 11 kcal/mole.

Additional data are being collected to evaluate further the precision
and reproducibility of the voltammetric measurements at different levels

 

*Consultant, Department of Chemistry, University of Tennessee,
Knoxville.
196

of uranium concentrations, from which it is hoped tec obtain a better
assessment of this approach as an analytical method for in-line deter-
mination of U(IV) in molten fluorides.

A new voltammeter is being bullt that will measure a 20-fold higher
current {100 ma) than existing equipment, so that electrodes with more
reproducible area can be used. With present equipment the electrode is
limited to a 20-gage platinum wire inserted only 5 mm into the fuel, so
that slight changes in melt level Introduce a significant change 1n re-
duction current. The new instrument will alsc provide a more rapld
voltage scan, up to 500 v/sec, to minimize flow effects in an in-line
cell.

In-Tine Test facility — R. F. Apple, J. M. Dale, J. P. Young, and A. 5.
Meyer

 

In view of the value and potential success of methods for the con-
tinuous analysis of circulating salt streams, a facility to provide salt
streams of known composition is being considered to provide a test of
equipnent under similated in~line measurements. The most practical
facility commensurate with the available space 1n a California hood con-
sists of a 20-kg salt reservoir fitted with a stirrer, ports for sampling
and for the addition of solid and molten constituents, purge streams and
electrodes for purification, and a helium gas 1ift to continucusly transfer
a stream of salt to an elevated constant-level vessel. From the constant-
level vessel, salt streams can be gravity-Ted to apparatuses for testing
electrometric and spectrophotometric techniques of analysis and for deter-
mining parameters for the continuous determination of oxide by counter-
current eguilibration of a salt stream with anhydrous hydrogen fluoride.
The analyzed salt stireams will then be returned to the reservoir. This
facility will also be used to test capillary and orifice techniques Tor
the control of low salt flows and to design small freeze valves. Tests
of various gas 1ift designs are being carried out with simulated molten
fuel (aquecus zinc chloride solutions) to determine whether a 6- to §-ft
1ift is practicsal.

Analysis of Helium Blanket Gas — C. M. Boyd, C. A. Horton, A. D. Horton,
and A. S. Meyer

 

The Analytical Chemistry Division, together with members of the Re-
actor and Reactor Chemistry Divisions, participated in various experi-
ments to determine possible sources of the organic depesits which have
caused pluggling of valves and filters in the MSRE off-gas system. The
analytical support included:

1. +the installation of a continuous hydrocarbon analyzer to monlitor gas
streams from Reactor Chemistry experiments4o to simulate oil leaks
into the MSRE pump and to evaluate trapping systems for reducing
hydrocarbon concentration, and to monitor hydrocarbons in the off-
gas from the Y-12 pump test 1oop;41
197

2. the determination by gasg chromatographic analysis and by selective
chemical reactions of individual hydrocarbons in "grab" samples
from the above experiments and in a trapped sample from a nonradio-
active MSRE purge line;42

3. the development of a technique to measure the concentration of hydro-
carbong collected on an experimental charcoal trap by pyrolysis of a
sample of the charcoal followed by gas chromatographic analysis of
the pyrolyzate. By sampling the charcoal bed at various depths, it
is possible to determine the distribution of individual hydrocarbons
as a function of trap length.

The continuous hydrocarbon monitor proved to be the most ussful of
these technlgques, particularly in measurements performed at the Y-12 test
loop with P. G. Smith, of the Reactor Division, and R. G. Ross, of the
Reactor Chemistry Division. Figure 7.23 is a flow schematic of the in-
Jection experiment. Although the experiment was designed to measure the
effects of deliberate injection of oil into the pump tank, the initial
results revealed an actual oll leak in the pump. The experiment there-
fore afforded an opportunity to study the oil leak and correlate the
hydrocarbon level in the pump tank off-gas with an operating variable
(Ap across the shaft annulus) and thus distinguish between possible leak
locations. A typlcal plot of the hydrocarbon concentration in the pump
tank off-gas is shown in Fig. 7.24. When the pump rotation was stopped,
the shaift annulus Ap decreased from 3 to 0.5 psi, and the hydrocarbon
level in the off-gas dropped immediately to less than 10 ppm methane
egquivalent. After about 25 hr the Ap began to recover, and the hydro-
carbon level started to undergo & series of rapid excursions that are
suggestive of discrete drops of oil entering the hot regions of the pump
tank. The average hydrocarbon concentration gradually returned to a level

ORNL--DWG, 66-6827A

SAMPLE
POINT 1

  
   
 

 

He - LN
41L

  

 

2 3
SAMPLE POINTS

PUMP TANK PURGE

 

 

 

 

A e

 

 
ORNL-DWG &5-10029A

 

 

 

 

 

1200 — —$}: I e T T T T : e ,_,
~ : :
o
= 1000 +~ — — - - -
o
'_
g |
S 2 800 — iy
= © .
oz :
S 3 i
2 Lo i
R 2 600 ‘ i
xr o
< £
O @
2 € ao0
o €
- o
T & ;
| I o
;I’ 200 p— — f+— o =
]_
O !
= ; P
0 — —— Do T o o ]
210 220 230 240 260 270 280

 

TEST TIME (hr!

Tig., 7.24.  Hydrocarbong in Y-12 Pump Tank Off-Gas.

of about, 300 ppm, in parallel with inecreasing Ap. Similar fluctuations

in hydrocarbon concentrations were cbserved during three additional
changes in pump operation to reduce the Ap, which was ultimately shown to
be caused by a salt plug at the bottom of the shaft annulus. Because this
differential pressure is exerted across the seals between the shield plug
and the catch basin, the position of the leak was defined as through
these seals rather than down the shaft annulus. This was confirmed, on
disassembly, by an oil film and pyrolysis stain on the outside of the
shield plug.

Injection of oil, Gulf Spin 35, into the pump tank showed that es-
sentially all the oil entering the tank appeared as hydrocarvons 1n the
off-gas. Table 7.18 shows gas chromatographic analyses of the off-gas
together with those from helium effluents from Reactor Chemistry experi-
ments in which the oil was injected into a helium stream entering an
empty nickel pot at 600°C. These results revealed that, at least in the
absence of radioactivity, oil entering the pump tank is predominantly
cracked to light hydrocarbons, methane, ethane, and unsaturates lighter
than Cs5, which are trapped ineffectively on charcoal traps at 100°C.
Conversely, in the Reactor Chemlstry experiments the cracking was incom-
plete to yield a substantial fraction of > Cg hydrocarbons, which are
trapped with high efficiency.

A thermal conductivity method to measure the total hydrocarbon con-
centration in the radiocactive off-gas of the MSRE continuously has been
developed with the Reactor Chemistry Division. In this method the off-
gas sample is passed over copper oxide at 700°C to convert hydrocarbong
to COs and H,O. This oxidized stream is passed through one side of a
thermal conductivity detector and thence to a trap containing Ascarite
and Mg(C10z), to yield a stream of inert gases which is directed through
the reference side of the thermal conductivity cell. In bench-top tests
the response of this apparatus was linear to total hydrocarbon concentra-
tion of 1000 ppm with a limit of detection below 10 ppm. A similar sys-
tem with the trap and copper oxide furnace designed for one year of reac-
tor operation will be installed in the gas-sampling station of the MSRE.
199

Table 7.18. Hydrocarbons Produced by 0il Injection

Injection rate, 16 cm®/day

 

Hydrocarbon Concentration {(ppm methane equivalents)

 

Y-12 Test Loop

 

Hy drocarbon Reactor Chemlstry

 

 

 

 

Simulated Ieak 2 hr After 20 hr After
Tests Start of Injection Start of Injection
Before After Before After Before After
Trap Trap Trap Trap Trap Trap
CHy, 25 30 210 210 320 250
CoHg &4 12 tidy 42 46 40
CoH, 140 140 600 590 700 690
C3Hg 100 120 170 158 230 210
Unsatd. Cu 56 94 98 3 130 160
Unsatd. Cs 78 4 10 7 24, A
and Cg
Aromatics 158 206
> Cg 300
Total 703 400 1290 1010 1666 1354

 

7.6  Development and Fvaluation of Equipment and Procedures for
Analyzing Radioactive MSRE Salt Samples

 

 

F. ¥X. Heacker C. . Tamb L. 7. Corbin

The remote apparatus for determining the oxide content of MSRE salt
samples was installed In cell 2 of the High-Radiation-ILevel Analytical
Iaboratory (ARLIAL) (Building 2026). Fluoride salt samples and tin oxide
standards were analyzed to check the apparatus and Tamiliarize the HRIAL
persomnel with the method. The results were satisfactory, and the phys-
ical manipulations required were performed adequately. DBight 50-g fuel-
salt samples have been analyzed remotely.

The time required to decontaminate the transport containsrs was
greatly reduced py using disposable mild steel plugs for each sample
submitted.
200

Sample Analyses

 

From January 1, 196G, through June 30, 1966, 43 MSRE fuel-salt sam-
ples were submitted for analysis. The samples were analyzed as shown
below.

 

Analysis Number of Determinations
Uranium 35
Zirconium 35
Chromium 35
Beryllium 35
Fluorine 35
Iron 35
Nickel 35
Molybdenum 7
Lithium 35
RCA Prep. 22
MSA Prep. 5
Oxide &
Carbon 3

Of the 43 samples submitted, two 50-g oxide samples were not analyzed due
to contaminated ladles. Tlhree samples were analyzed for carbon and found
to contain <50 ppm.

Several of the samples were received with a silver and a Hastelloy
N wire colled onte the stainless steel cable between the latch and ladle.
The latch, wires, and cable were separated and prepared for radiochemical
analyses.

Quality-Control Program

 

The quality-control program was continued during the first and second
guarters of 1966. A composite of the values obtained by four different
groups of shift personnel is shown in Tables 7.19 and 7.20. Molybdenum
values are not shown since it was not added to the synthetic solutions.

It was evident from the second-guarter control data that a positive
hbias of approximately 3% existed in the amperometric zirconium methaod.
The bias was attributed to poelymerization of the zirconium in the standard
solutions used to standardize the cupferron titrant. Although the hias
appears to have been eliminated by the preparation of new zirconium
standards, more data are needed o confirm it.

The positive bilas existing in the spectrophotometric nickel method
has been rediced to approximately 8% by cnanging the color filter in the
filter pnotometer and constructing a new calibration curve. Further ef-
forts are being made to e¢liminate the bias completely.
201

Table 7.19. BSummary of Control Results,
January Through February 1966

 

Limit of Error

(%)

Number of

 

 

 

Determinations

Determination Method Determinations
Fixed Found
Beryllium Fhotoneutron 27 5.0 2.43
Chromium Amperometric 25 15.0 7.95
Iron o-Phenanthroline 34 15.0 7.93
Nickel Dimethylglyoxime 30 15.0 12.74
Uranium Coulometric g 1.0 1.16
(high sen.)
Zirconium Amperometriec 30 5.0 5.74
Table 7.20. Summary of Control Results,
April Through June 1966
Limit of Error
a
Determination Method Number of (%)

Fixed Found

 

Beryllium Photoneutron 26 5.0 2 .60

Chromium Amperometric 36 15.0 12.46

Iron o-Fhenanthroline 41 15.0 7.5%

Nickel Dimethylglyoxime 48 15.0 7.67

Uranium Coulometric 127 1.0 1.04
(high sen.)

Zirconium Amperometric 57 5.0 5.48

 
n

10.

11.
12.

14

15.

17.
18.
19.
20.
21.
22 .
23.
24
25.

MSR Program Semiann. Progr. Repb. Feb. 28, 1966, ORNL-3936, p. 122.

 

Tone operational history of these runs is described in sect. 1.1 of
this report.

5. 5. Kirslis, this report, sect. 7.4, subsection entitled "Fission
Product Behavior in the MSRE."

R. 7. Apple, J. M. Dale, and A. S. Meyer, this report, sect. 7.5,
subsection entitled "Determination of Oxide in Radiocactive MSRE
Samples."

R. E. Eby, ORNL Isotopes Division.

R. E. Thoma to P. N. Haubenreich, "Chemical Analysis of MSRE Flush
and Ccolant Salts in Prenuclear Test Period," MSR-65-19 (March 19,
1965) (internal correspondence).

5. Cantor and W. T. Ward, Reactor Chem. Div. Ann. Progr. Rept. Dec.
31, 1965, ORNL-3913, pp. 27-28.

L. Shartsis and 5. Spinner, J. Res. Natl. Bur. Std. 46, 176 (1951).
J. D. MacKenszie, J. Chem. Phys. 32, 1150 (1960).

J. H. DeBoer and J. A. M. Van lLiempt, Rec. Trav. Chim. 46, 124
(1927).

L. J. Kinkenberg, Rec. Trav. Chim. 56, 36 (1937).

 

Structure Reports, vol. XITI, p. 342.
J. D. Edwards et al., J. Electrochem. Soc. 100, 508 (1953).

S. Centor, Reactor Chem. Div. Ann. Progr. Repb. Dec. 31, 1965,
ORNL-3913, pp. 29-32.

S. Cantor and W. T. Ward, ipid., p. 27.

MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNI~3936, pp.
141—45.,

 

 

D. G. Hill to W. R. Grimes, private communication, June 29, 1966.
D. G. Hill tc W. R. Grimes, private communication, July 1, 1866.
W. 5. Ginell, Nucl. Tech. 51, 185 (1959).

J. J. Egan and R. N. Wiswall, Nucleonics 15, 104 (1957).

Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591, p. 50.
Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1965, ORNL-3789, p. 16.
MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, p. 145,

 

 

 

Zr0p was prepared by H. H. Stone, Reactor Chemistry Division.

MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-39356, p. 147.

 
26 .

27,
28.

29.

30.

31.

32.

37.
38.
39.
40,

41,

42

203

R. P. Elliott, Constitution of Binary Alloys, First Supplement,
p. 203, MeGraw-Hill, New York, 1965.

MSR Program Semiamn. Progr. Rept. Feb. 28, 1966, ORNI~3936, p. 148.

W. D. Burch, G. M. Watson, and H. 0. Weeren, Xenon Control in Fluid
Fueled Resctors, ORNL-CF-60-2-2 (July 6, 1960).

I. Spiewak, "Xenon Transport in MSRE Graphite," M3R-60-28 (Nov. 2,
1960) (internal correspondence).

G. M. Watson and R. B. Evans III, Xenon Diffusion in Graphite; Ef-
fects of Xenon Absorption in Molten Salt Reactors Containing Graph-
ite, ORNL-CF-61-2-59 (Feb. 15, 1961).

J. W. Miller, Xenon Poiscning in Molten 5alt Reactors, ORNI~CF-61-
5-62 (May 3, 1961).

R. B. Evans III, "Xenon Poisoning in Molten Salt Reactors Contain-
ing Graphite," Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1961,
ORNI~3127, pp. 15-16.

P. R. Xasten, E. 3. Bettis, and R. C. Robertson, Design Studies of
1000-Mw(e) Molten-8alt Breeder Reactors, ORNI-3926 (August 1966).

 

 

 

 

 

 

 

 

 

 

B. Manowitz, quoted by P. G. Salgado in Nuclear Reactor Magneto-
Hydrodynamics Power Ceneration, LA-3368, p. 23.

MSR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNI-3872, p. 127.

 

 

 

A. M. Perry, internal memorandum, July 30, 1966.
MSR_ Program Semiann. Progr. Rept. Fep. 28, 1966, ORNI~3936, p. 154.

 

MSR Program Semiann. Progr. Repb. Aug. 31, 1965, ORNL-3872, p. 145.
Donald Cohen and W. T. Carnall, J. Phys. Chem. &4, 1933 (1960).
B. F. Hitch, R. G. Ross, and H. F. MeDuffie, Tests of Various Parti-

 

 

cle Filters for Removal of 0il Mists and Hydrocarbon Vapors, ORNI~
T™-1623 (to be issued). '

 

A. 5. Meyer, Investigation of Hydrocarbons in the 0ff-Gas from the
Y-12 Test Loop, ORNI~CF-66-8-30 (September 1966).

A. 5. Meyer, "Hydrocarbons ian Seal Purge of MSRE Pumps," MSR-66-10
(May 1966) (interunal correspondence).

 
Part 3. MOLTEN-SALT BREEDER REACTOR STUDIES
8. MOLTEN-SALT BREEDER REACTOR DESIGN STUDIES

Paul R. Kasten E. 8. Bettis Roy C. Rcbertson
J. H. Westsik H. F. Bauman H. T. Kerr

The MSBR reference design concept was presented previously,l and
is a two-region, two-fluld system, with fuel salt separated from the
blanket salt by graphite tubes. On-site fuel recycle is employed, using
fluoride-volatility and vacuum-distillation processing., In additiocn,
results were given for the case of direct removal of protactinium from
the blanket region; the associated concept was termed MSBR(Pa). Since
these studies, additional design conditions were investigated and
evaluated,? and these are discussed below.

8.1 Design Changes in MSBR Plant

 

In reactor design studies it often occurs that certain features of
the detaliled design undergo changes as more understanding is obtained of
the overall problems and as new ways are discovered to solve a given de-
sign problem, §Such changes have tsken place during the M3BR design
studies; the most important are those associated with the primary heat
exchanger designs and the pressures that exist in the wvarious circulating-
salt systems.

An objectionable feature of the MSBR heat exchanger design considered
previously was the use of expansion bellcows at the bottom of the exchanger.
These bellows permit tubes In the central portion of the exchanger to
change in length relative to those in the annular region due to thermal
conditions. Since such bellows may be impractical to use under reactor
operating conditions, a new design was developed that eliminated them.

Figure 8.1 shows the revised heat exchanger design. The expansion
bellows were eliminated, and changes in the tube lengths due to thermal
conditions are accommodated by the use of sine-wave type of construction,
which permits each tube to adjust to thermal changes., In addition, the
coolant salt now enters the heat exchanger through an annular volute ab
the top and passes downward through a baffled outer annular region. The
coolant salt then passes upward through a baffled inner amnular region
and exits through a central pipe.

In Mig. 8.1 the flow of fuel =slt through the pump is reversed from
that given previcusly® in order to reduce the pressure in the graphite
Tuel tubes., TFuel salt enters the heat exchanger in the inner annular
region, passes downward through the tubes, and then flows upward through
the tubes in the outer annular region before entering the reactor.

The blanket-salt heal exchanger was also revised to give 2 design
similar to that of Fig. 8.1. The general features of these exchangers
and their placement in the reactor cell are shown in Fig. 8.2. The

207
208

ORNT, DWG. 66-713%

—FUEL LEVEL (DUNP)

FUEL SALT DUMP TANK

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I
o —— — —— FUEL SALT PUMP
FUEL SALT 1O k b T I ‘
& FROM REACTOR |+ | i
N
N COCLANT SALT FROM
! ,,;-7“-' {  STEAM HEAT EXCHANGERS
rd / ;
’j:u-:.-xfl
| e
\\ W i ‘I —COOLANT PASS
| ] 1l | ALY SEPARATING BAFFLE
| ; ‘ it
A 1f | | ok ‘
[4-4 i ;-cl u_f b Wn T e lflnjt i
B L
i N 1
I g AL FLOW ARRANGEMENT
I il FUEL SALT-IN TUBE SIDE
fo DU L ‘ L ‘L ML COOLANT SALT-IN SHELL SIDE
] 1 T b |
‘ T i \ |
i 4% : \ ‘ “ e —';'ll’lol.—' ; i
| il | ’ |
L)y aH L8 0" DD il i R
TRl
| ! B \ |
| | | Il 1t
=i = \‘ { 7.):#?: : 't“
i ; D
! ‘ | : | ‘ ' : | \
. I L0
! VH’U’“” 1 Jl ‘{H il
] f\ Rlefeld 1 f\ii;j
\\_ - /f | ”\\.t___; _,7/.
10 FUEL———— i
DRAIN TANKS T \ ~COOLANT SALT TO

BLANKET HEAT EXCHANGER

Fig, 8.1. Revised Fuel Heal Exchanger for MSBER.
    
    

     
 
 
 
 

    

   

209
ORNL DWG. 66-7109
FUEL PUMP BLANKET PUMP
MOTOR CONTROL ROD MOTOR
DRIVE
CONSTANT -
SUPPORT
HANGERS

  
 
 
 
 
  
  
  
  
   
  
   
 
  
   
   

FUEL DUMP
TANK WITH

 

COOLING et
= COILS FOR _.'/.) o
AFTER HEZAT S
REMOVAL s o ABLANKET

72 S THEAT
e JEXCH.

= YL Smee———m——— - s el e .
[ ' 4
C o > & -5
s, -
2

> 4
______ a1

s 3
¥
a a

o .

2 ~

AR
< e — e

« |CORE

.;;u REACTOR
- VESSEL

 

i - , nc i
e 2 i

L. FUEL SALT
: IDIST.

: &
oo+ PLENUMS
e € &
L -
- . .
S -~
. e
e
. ‘,,‘ . ‘

PRIMARY =
HEAT -
EXCH. e

 

  

 

    

 
  
 

       

g :1.‘
1 ‘o
.S
]
R
. w
Ny TR AT B B g
T Y L LREACTOR CELL HEATERS
h, . R R AR L et
‘ <N 2k N : y

 

-

g. 8.2. MSBR Cell Elevation Showing Primary Heat Exchangers and
lacement,

-
210

blanket-salt pump was also altered so that blanket salt leaving the re-
actor now enters the suction side of tne pump.

From the viewpoint cof reactor safely, it is important that the
blanket salt be at a higher pressure than the fuel salt.? Under such
circumstances, rupture of a fuel tube would result in leakage of Ffertile
salt into the fuel and a reduction in reactivity. In order to achieve
this condition with a minimum operating pressure in the reactor vessel,
the fluid flow was reversed from thet in the initial M3BR design as
stated above. The resulting flow diagram is shown in Fig. 8.3,

In addition, it 1s desirable that any leakage between the reactor
Tluid and coolant-galt systems be from the coolant system into the fuel
or blanket system. In order to achieve these conditions, the MSBR
operating pressures were revised to those shown in Table &,1.

Table 8.1. Pressures in Various Parts
of Revised MSBR Salt Circuits

Flow diagram given in Fig. 8.3

 

 

Nominal.
Location Pressure
(psig)
Fuel-salt system
Core entrance 50
Core exit 25
Pump suction 10
Pump ocutlet 150
Heat exchanger outlet 60
Blanket-salt system
Blanket entrance 66
Blankelt exit 65
Pump suction 64
Pump outlet 155
Heat exchanger outlet &
Coolant-zalt system
Pump suction before boiler-superheaters 130
Pump outlet before boiler-superheaters 280
Tnlet to fuel heat exchangers 220
Qutlet from fuel heat exchangers 160
Outlet-inlet to blanket heat exchangers 142
Pump suction before reheaters 130
Pump outlet before reheaters 240

Reheater cutlet 220

 
 
 
  
  

   

 

 

       
  
 

 

 

 

 

 

 

 

 

 

 

 

    
 

  
     

 

 

 

 

ORNL DWG. 66-7022
REACTOR VESSEL 5134 #
T T T T T e ™
;550 000es 10.067 # 7.152 4 /1518.5h-540p-1000°
;020p- oo T 1
15|3.5hj ; : 5142491-3515;: 1000° |
L f | !
S N i |
N | 3600p- 1000°F :!424h ! | J/’H’P//L GEN.
20 1t Yeoe | ' TuRINE [ !P TURBINE = 5272 Mwe
BOILER : | T L G
. ? 600 p- i !
= SUPERHEATERS 'S . izm"h; B L
i '190°F 000°F  qriendT }*’850 ******* S }r
3 T | :
o R EXCH AND PUMPS l ‘ i
‘ T PUMPS ] i - 5077 Mwe
iR FUEL SALT HEAT jik dhii — ERR TURB!NE TURB'NE Gross
==§mh’],;25 f EXCH. AND PUMPSL?!’ fi ?’30“%“ B : | i L,/\[
! 4 | Hes*F § i i |
| , ‘ U | Remear sTeaw L |
Wl e T | \T/ p ‘ i __
F j;'/ | | | REHEATERS (CONDENSER & FEEDWATER |
I 7 l | | ] 3800p-ThG°F i (307 84 L SYSTEMS ) -
| | L | [ ; | 7692h 3500 p-866°F 35000~ 5509 =
1 ! 1 | - 9
i ” e T 1 __l":l—f:j___’§§§§_*!"__' et 4 T T
, : 4 4 570p-650°F [ e
Pog £ ] P e 3475p-695°F ,
b R L Te64n MI;]T*_’T‘!;—‘
{ BOOSTER
PUMPS
/BLANKET SALT ) /" FUEL SALT /"COOLANT SALT PERFORMANCE
\_DRAIN TANKS \_DRAIN_TANKS _DRAIN TANKS NET OUTRUT 1000 Mwe
LEGEND ' GROSS GENERATION 034.9 Mwe
- BF BOOSTER PUMPS 9.2 Mwe
FUE%K_ — STATION AUXILIARIES 25.7 Mwe
?éoffi;f - REACTOR HEAT INPUT 2,225 Mwt
éTEAM i T NET HEAT RATE 7,601 Btu/kwh
o NET EFFICIENCY 44.9 %
o 16 b /hr
P psia
B Biu./th.
O °F
. Fresze Valve

Fig. 8.3.

Revised MSER Flow Diagram.
g
212

As given in Table 8.1, the minimum pressure difference between the
core and blanket regions is about 15 psi plus the static head differential,
or a minimum toltal difference of about 30 psi. If it is desirable to in-
crease this pressure differential, the blanket salt pump could be changed
so that it discharges into the reactor blanket region, giving a minimum
differential pressure between the core and blanket fluids of about 120
psi. Whether this change 1s necessary or whether it would increase the
reactor vessel design pressure is dependent upon the safety criteria
that need to bhe satisfied. A design pressure of 150 psia was used in
determining the thickness of the M3BR reactor vessel,

8.2 Modular~-Type Plant

 

Ann important factor in attaining low power costs 1s tThe abllity To
maintain a high plant-availability factor; design features tThat improve
this factor are desirable if these features do not themselves introduce
compensating disadvantages.,

In the MSBR plant, use is made of four heat exchanger circults in
conjunction with one reactor vessel in such a manner that if one punmp in
the fuel circuilt stops, the reactor is effectively shut dowa. I, on
the other hand, it were practicable to have four separate reactor cir-
cults, with each connected to one of the four heat exchanger circuitls,
stoppage of a fuel pump would shut down only one-quarter of the station
capacity, leaving 75% available for power production. In order to de-
termine the practicality of using a number of reactors in a single 1000-
Mw (electrical) station, the design features of a modular-type MSBR
plant, termed MMSBR, were invesgtlgated.

The MMSBR design concept conslders four separate and identical re-
actors, along with their separate salt circuits. The only connections
of the four reactors are through the fuel-recycle plant. The designs
of the heat exchangers, the coolant-salt circulits, and the steam-power
cycle remain esgentially as for the MSBER. Each reactor module gen-
erates the thermal power reguired for producing 250 Mw (electrical)
net.

The flow diagram given previously for the MSBR (Fig. 8.3) also is
essentially valid for the MMSBR. Salt flow rates and capacities of the
various compcnents remain as in the MSER design.

Figures 8.4 and 8.5 give plan and elevation views of the four dis-
tinet reactor cells, along with their adjacent steam-generating cells.
Any reactor module can be shubt down and serviced while the other three
remain operating.

The reactor core consists of 210 graphite fuel cells operating in
parallel within the reactor tank. The design of the graphite tubes
separating the fuel and blanket salts is similar to that used in the
MSBR. The reactor core region is cylindrical with a dlameter of about
6.2 ft and a height of aboult 7.9 ft. The reactor vessel 1s approximately
06-7114

ORNL DWG.

 

213

BLANKET
EXCHANGER
AND PUMP
REACTOR

~~REHEATERS
HEAT

 

 

L e

 

 

 

 

 

 

 

 

 

220"~~~

 

 

 

 

 

 

 

 

‘ZOF

 

 

 

 

 

 

 

 

 

 

 

 

 

 

130" ———

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ORS iy

 

 

  

b= — . 7
MAH = _D /
oy o T» Z
T.E Q. ey
02 5= e L
@ Ca oS =25 5
WMP pild
qr = w
=00 )
[Fap- ] W)
ool Z

A

Modular Plant Reactor Cell — Plan AA,

Mg, 8.4,
ORNL DWG. 66-7115

CONTROL ROD DRIVE -~
- ; —FUEL AND BLANKET

PUMP DRIVE MOTORS

COOLANT SALT PUMPS

     
 
  
  
   

 

/f_“STEAM GENERATORS

\ REACTOR— i
5 fi - REHEATERS

 

 

 

 

 

\ . Y
. BLANKET HEATZ
‘ £ XCHANGER .
Lo T3
60-0
=3 STEAM
PIPING
| =3
PRIMARY HEAT
EXCHANGER
.
i
| ,
P ! |

fig. 8.5. Modular Plant Reactor Cell — Elevation EB.

71
215

12 £t in diameter and sbout 14 Tt high. Except for the use of four re-
actor vessels instead of one, all design features of the MMSBR are similar
to those of the MSBR, The design conditions for one reactor module are
summarized in Table 8.2,

Table 8.2. MMSBR Design Conditions for One Module

 

Power generation, Mw

Thermal 556
Electrical 250
Thermal efficiency, % 45
- Plant factor 0. 80
Dimensions, It
Core
. Height .87
Diameter 6.3
Blanket thickness 2
Radial 2
Axial 0.5

Reflector thickness
Reactor volumes, ft°

Core 245
Blanket 1000
Salt volumes, ft3
Fuel
Core 4.5
Blanket 7
Plena 22
Piping Z
Heat exchanger and pump g2
Processing 75
Total 185
Fertile
Core 12
Blanket 1000
Heat exchanger and piping 25
Processging 24
Total 1061
Salt compositions, mole %
Fuel
"LiF 63.6
Bng 36,2
UF, (fissile) 0.22
Tertile
LiF 71
B@Fz 2
Thi, 27

Average power density in core fuel salt, kw/liter 4773

 
216

The nuclear and fuel-cycle performance of a Tour-module plant gen-
erating 1000 Mw (electrical) was studied botn for protactinium removal
from the blanket stream and for the case of no direct protactinium re-
moval, The same methods and bases as those for the M3BR studies were
employed. Analogous to previous terminology, these cases are termed
MMSBR(Pa) and MMSBR., The results obtained are summarized in Table 8, 3.
Comparison with the results obbained for the MSBR(Pa) and the MSBR indi-
cates ‘that the nuclear and Tuel-cycle performance of a modular-type
plant cowmpares favorably with that of a single-reactor~-type plant; the
modular plant tends to have slightly higher breeding ratlo, fissile in-
ventory, and fuel-cycle cost.

Table 8.3, Nominal Nuclear and Fuel-Cycle Perforuance
of 1000-Mw (electrical) Modular Plants

Investor-owned plant: 0.8 load factor

 

MMSBR(Pa) MMSBR MSBR(Pa)

 

Fuel yield, %/year 7.3 5.0 7.95
Breeding ratio 1.073 1.053 1.071
Specific fissile inventory, kg/MW (electrical) 0.76 0,80 0.68
Specific fertile inventory, kg/Mw (electrical) 125 310 1.05
Fuel-cycle cost, mil%s/kwhr (electrical) 0.38 0.48 0.35
Doubling time, years 13.7 20 12.6
Power doubling time, years 9.5 13.9 8.7

 

*Tnverse of fractional Tuel yield per year.
Pased on continucus investment of bred fuel in reactor pover plants.

Capiltal cost estimates were alsc made Tor the modular plant. The
primary difference between the MMSBR- and MSBR-Type plants is the use of
four reactor vessels and cells in the modular plant rather than the one
in the MSBR. However, the reactor vessels in the modular plant are
smaller, so that their combined installed cost 1s only about $1 million
more than that of the single large vessel., At Tthe same time, the modular
plant permits betler placement of cells and a reduction in building
volume. The resultant capital cost estimate for the modular plant was
about $114/kw (electrical) for a privately owned plant, which is about
the same as thal estimated for the single~reactor plant. Using this
cost estimate, along with the MSBR estimate for operation and maintenance
costs, and the fuel-cycle costs from Table €.3, glves the power-genera-
tlon costs summarized in Table 8.4. These costs are nearly the same as
those for the MSBR-type plants and thus Indicate the desirability of a
modular-type plant if the plant availability facltor is improved by its
use,
217

Table €.4. Power-Production Costs for Modular-Type
Molten-Salt Breeder Reactors

Investor-owner plant: 0,8 plant factor

 

Cost [mills/kwhr
(electrical)]

 

 

MMSBR(Pa.) MMSBR
Fixed charges 1.95 1.95
Operation and maintenance costs 0.34 0.34
Tuel-cycle costs®™ 0.38 0.48
Total power-production costs 2.7 2.8

 

aCapital charges of processing plant are included in fuel-cycle
COoses,.

8.3 Steam Cycle with Alternative Feedwabter Temperature

 

In the above molten-salt breeder reactor concephts, the leedwater
temperature entering the once-through supercritical boilers was 700°F,
and the temperature of the "cold" steam to the reheater was 650°F.

These temperatures were specified in order to avoid any freezing of the
intermediate-coolant saltl’? and involved diversion of prime steam. It
would be a significant adventage if it were not necessary to divert
almost 3% of the throttle steam for-heating of the feedwater and rcheat
steam, since this diversion leads to a loss of available energy. An
even more significant saving could be achieved if the 9.2 Mw (electrical)
of power redquired to drive the feedwater pressure-booster pumps could

be eliminated; alsoc, removal of the reheat-steam preheaters and the
booster pumps would reduce caplital investment requirements. Thus, sav-
ings can be achieved by lowering the temperature of the steam-cycle

fluld entering the boilers and reheaters. To determine the incentive for
developing a coolant salt having a low liquidus temperature, the MSBR
steam-power cycle was studied with conditions of 580°F feedwater tem-
perature and 550°F reheat steam. In order to differentiate and compare
cases, use of 700°F feedwater and 650°F recheat steam is designated case
A, while case B represents the alternative conditions.

The cycle arrangement for the case B conditions is shown in Fig. &8.6.
In this cycle the 552°F steam from the high-pressure turbine exhaust is
introduced 1nto the reheaters without preheating. fThe feedwater is
heated from 550 to 580°F by the addition of one more stage of feedwater
neating; steam extracted from the high-pressure turbine is used., The
condensate from this new heater 1s cascaded back through the feedwater
heaters to the deaseralting heater in the usual manner. The heat balances
and the analysis of the steam cycle with case B conditions were performed
in the same manner as for case A conditions. Table 8.5 compares the de-
sign data for the two cases.
 

BOILER -
SUPZRHEATER
1837.0 Mwt

SooraaT saT |

 

850

 

 

7. 460,460 #

 

3,730,230+
~3700 P-580.0° 5836 h

LEGEND
WATER

Frm———————

 

 

 

 

_,}-T—“‘“l

Co

! *
560:9°%- 5622 h

o

7,460,460 #
I 38007 — 000" 19240 h I
i ]
11,744,000 #
- s o e = s
vigs”
REHEATER
3360 Mwi
r— S
929.‘1&1’-5-.“-— — _J
g5%9°
]
—_— e _360489%
[~ 1396 P-(333.0 &

I
.k

T
w

|
L

>

-57i0 h

e
020 # [ g 750.5224

STEAM

COOLANT SALT
Lbs/ hr
Los/in®  aos.

Btu/ib
°F

NOTE

CONDITIONS FOR "8"

T e I TN S IS SEDED S SE— — DTSR

i

433~ 4512n

EQUIPMENT

(EXCEPT TUR-GEN) SAME AS FOR ‘A",

Fig. 8.0.

 

5055,950#
A L]

. -~ —
540P-1000 —-i518 5h

3515 F' IOOO

|
!
!
|
|
}
|
|
|
|

1314.0 h

1,579,000 #

|46P - 6937

4

 

I4240h

 

|
|

1P TURBINE

GENERATOR "A"

 

GROSS

 

 

OUTPUT

 

 

r__

    
    

 

530.4 Mwe

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T

 

   
   
 
 
 
  

 

ORNL, DWG. 66-7134

 

CONDENSER A" |

1.5 " Hg abs

32,660 #
103.0%-710h
HOTWELL
PUMPS

)IA‘I\

       
   
 

| CONDEMSATE 4
! BOOSTER |
| PLMPS

 

- —_——— e | —— e 2O
!_ T - ; 1000 3h
i i l |
T:;—Mfl GENERATOR "8"
i GROSS
[ OUTPUT
f 505.C Mwe
!
'/r—’/r\l\h*] |
| | Pyt | ;
' p | '*i
Py e
Lt Lt ..E_'.LJI — L%
— :J__'"i e
_d T snzos

 

 

 

 

 

 

 

405,330 #

[362P-aga e 1242 7h

 

1,560,480 #

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

220, 460%

 

 

 

 

 

 

 

321,8504

 

SUMMARY OF PERFORMANCE AT

NET

146.6

!
g i
i L 32B.5-12040h T T :
- T T — |
1 I 4-3 _i I 6-8 !
1 | | q8898004 | [ 1
| 2 ~—
f 77 l I
I : DEAERATOR | | | i
b ! g
! | : i ;
l n ;;J‘:_ _‘fi STORAGE | L_i . L_.____._‘
953,190+ ' ’ |
| 3761-349.3 1 = | | @;’i‘l";s.
| (357553295 2341° M 1734143 h
| | 202.4n
|
] |—  7es0s504 |
| 52202304 | 141,900%
| 183.4°- (5.3 h

QUTPUT

-1H85h

 

3BOOP -366.1°- 343.3h

Alternate Steam System Flowsheet for 580°F Feedwater,

ELEC. POWER FOR AUX

BF PRESSURE -BOOQOSTER PUMPS
GROSE GEN. QUTPUT
8F PUMPS

REACTOR HEAT INPUT
GROSS E£FF OF CYCLE
NET EFF

OF CYCLE

NET HEAT RAYTE

1436 -i11.5 h

RATED LOAD
10097 iAwe
25.7 Mwe
NONE
1035. 4 Mwe
30.6 Mw
2225 Nwt
47.9 %
454 %

7,518 Btu/Kwh
219

Table 8,5. MSBR Steam System Design and Performance Data for Case A
and Case B Conditions

 

Cagse A -

MSBR Steam Cycle

Case B —

MSBR Alternative

 

Total steam capacity, lb/hr

Temperature of inlet feed-
water, °F

Enthalpy of inlet feedwater,

tu/1b

Pressure of inlet feedwater,
peia

Temperature of exit gteam, °F

Pressure of exit steam, psis

Enthalpy of exit steam,

Btu/1b

Temperature of inlet coolant
salt, °F

Temperature of exit coclant
salt, °F

Average specific heat of
coolant salt, Btu 1b™1 °p™1
Total coolant-salt flow
1b/hr
cfs
gpm

10.068 % 10°
700

~3800

1003

~3600
1424.0

1125

850

0.41

58,468 x 10°

129.93
58,316

with 700°F Steam Cycle with
Feedwater 580°F Feedwater
General performance

Reactor heat input, Mw 2225 2225

Net electrical output, Mw 1000 1.009.7

Gross electrical generation, 1034.9 1035.4
Mw

Station auxiliary load, Mw 25.7 25.7
(electrical)

Boiler~feedwater pressure- 9.2 None
booster pump load, Mw
(electrical)

Boiler~feedwater pumnp steam- 29,3 30.6
turbine power output, Mw

Flow to turbine throttle, 7.152 x 10° 7.460 % 10°
1b/hr

Flow from superheater, 1b/hr 10.068 x 106 7.460 x 108

Gross efficiency, % 47.83 47,91

Cross heat rabe, Btu/kwhr 7136 7124

Net efficiency, % 4,9 45,4

Net heat rate, Btu/kwhr 7601 7518

Beiler-superheaters
Number of units 16 16
Total duty, Mw (thermal) 1931.5 1837.0

7,460 % 10°
580

583.6

~3800

1003

~3600
1424, 0

1125

850

Q.41

55.608 x 10°

123,57
55,463
Table &8.5.

(continued)

 

Case A —

MSBR Steam Cycle

with 700°F
Feedwabter

Cagse B ~

MSRR Alternative
Steam Cycle with
580°F TFeedwater

 

Steam reheaters

Number of units

Total duby, Mw (thermal)

Total stesm capacity, 1b/hr

Temperature of inlet steam,
°F

Pressure of inlet steam, psi

Fnthalpy of inlet steam,
Btu/1b

Temperature of exit steam,
°F

Pressure of exit steam,
psia

Enthalpy of exit steamn,

Btu/1b

Temperature of inlet coolant
salt, °F

Temperature of exit coolant
salt, °F

Average specific heat of
coolant salt, Btu 1b™1 °F™
Total coolant-salt low
1b/hr
cf's
o
Coolant-calt pressure drop,
inlet to ocutlet, psi

Reheat-steam prenheater

Nuntber of units

Total duty, Mvw (thermal)

Total heated steam capacity,
1b/hr

Tnlet temperature of heated
steam, °F

Exit temperature of heated
steam, °F

Tnlet pressure of heated
stean, psia

Exit pressure of heated
steam, psia

Inlet enthalpy of heated
steam, Btu/lb

Exit enthalpy of heated
steam, Btu/1b

Total heating steam, 1b/hr

Tnlet temperature of heating
steam, °F

8

293.5
5,134 x 10°
650

~570
1323.5

1000

~54.0
1518.5
1125

850

0.4

8.884 x 10°
19,742

8861
~50

8

100.45
5.134 x 10°
551.7

650

~580

~570
1256.7
1323.5

2.915 x 10°
1000

g

388.0

5.056 x 10°
551.,7

~500
1256.7

1000

~54.0

1518.5

1125

850

0.41

11,744 % 108
26.098

1714
~60

None
221

Table 8.5. (continued)

 

 

Case A - Case B —
MSEBR Steam Cycle MSBR Alternative
with 700°F Steam Cycle with
Feedwater 580°F Feedwater
Exit temperature of heating 866
steam, °F
Inlet pressure of heating 3515

steam, psia
Exit pressure of heating
steam, psia

Boller-feedwater pumps
Nurmber of units 2 2
Centrifugal pumps

Number of stages

Centrifugal pump

Feedwater flow rate,
1b/hr total

6 6

10,067 x 108

Feedwater flow rate, 7152 x 10° 7460 % 10°
1b/ar total

Required capacibty, gpm 8060 8408

Head, f% ~9380 ~2380

Speed, rpm 5000 5000

Water inlet temperature, °F 357.5 357.5

Water inlet enthalpy, Btu/1b 329.5 329.5

Water inlet specific volume, ~Q, 01.808 ~0,01808
££2 /1o

Steam~-turbine drive

Power required at rated 14,66 15.30
flow, Mw (each)

Power, nominal hp (each) 20,000 20,000

Throttle steam conditions, 1070/700 1070/700
psiaf°F

Throttle flow, 1b/hr (each) 413,610 431,400

Exhaust pressure, psia ~{7 ~ 77

Number of stages 8 &

Number of extraction points 3 3

Boiler-Teedwater pregsure-
booster pump
Number of units 2 None

 

Required capacity, gom {(each) 2500

Head, ft ~1413

Water inlet temperature, °F 695

Water inlet pressure, psia ~3500

Water inlet specific volume, ~(. 03020
£t2/1b

Vater outlet temperature, °F ~700

Electric-motor drive

Pover required at rate flow, 4. 587
Mw (each)

Power, nominal np (each) 6150
222

The elimination of the feedwater pressure-booster pumps required in
case A saves 2bout 9.2 Mw (electrical) of auxiliary power, which, to-
gether with the improvement in the cycle thermal efficiency due to the
additional stage of feedwater regeneration, makes about 9.7 Mw (electrical)
additional power available from the case B cycle., The overall net thermal
efficiency is thus improved frowm the 44.9% obtained from case A to 45.4%
in case B,

To complete the discussion of case A vs case B conditions, the cost
estimates for the affected items of equipment were compared; the results
are summarized in Table 8.6. As shown, the case B arrangement reguires
about $465,000 less capital expenditure, primarily due to removal of the
pressure-booster pumps. [In this cost study 1t was assumed that the
580°F liquidus-temperabure coolant salt has the same cost (about $1.00/1b)
as the MSBR coolant salt.] The lower construction cost reduces power costs
by about 0.008 mill/kvwhr (electrical), while the increased efficiency
lowers power cost by about 0.026 mill/kwhr (electrical) (private [inanc-
ing), to give a total saving of about 0.034 mill/kwhr (electrical) [0,021

Table 8.6. Cost Comparison of 700°F and 580°F Feedwater Cycles for MSBR

 

 

Number
of Case A — 700°F Cagse B — 580°F
Units Feedwater feedwater
Feedwater pressure-booster 2 $ 400, 000 None
pumps
Reheat-steam preheaters S) 180,000 Neone
Special mixing tee B 5,000 None
Feedwater heater No, O None $ 150,000
Charge for extra extraction None 45,000
nozzle on turbine for
heater No. 0O
Boiler-superheaters 16 6,000, 000° 5,900,000
Reheaters g 2,720,000" 2,880,000
39,305,000 £8,975,000"
Cost differential
Direct construction cost $320, 000
Total construction cost $465,000

 

a, , - . ;
Table shows only those costs different in the two cycle arrangements
and is not a complete listing of the turbine plant costs.

-

0 . . o

The high-pressure feedwater heater added in case B was designated
" "n = . 0 , . .
No. 0" in order not to disturb the heater numbers used in case A.

CEstimated on basis of $130/ft2.
dEstimated on basis of $140/ft2.
“Gstimated on basis of $125/7t2,

T .
Tndireect costs were assumed to be 41% of The direcet costs.
223

mill/kwhr (electrical) for public financing]. This saving in a 1.000-
Mw (electrical) plant (0.8 load factor) corresponds to aboub $238,000
per year. The present worth (6% discount factor) of this saving over

a 25-year period is about $1.5 million, For several MSBR power plants,
the saving would be proportionally greater. Thus, there is an economic
incentive Tor developing a coolant salt with a low liguidus temperature,
zo long as its inventory cost does not outweigh the potential saving.

Tf the inventory cost of the coclant salt for case B were about $2.4
million more than +that for case A, the potentisl saving would be canceled
by the increased coolant-salt inventory cost (for a privately owned
plant).

8.4 Additional Design Concepts

 

Other molten-salt reactor designs were studied briefly.? In general
the technology required for these alternative designs is relatively un-
developed, although there are experimental data that support the feasibility
of each concept. An exception is the molten-salt converter reactor (des-
ignated MSCR), whose application essentially requires only scaleup of MSRE
and associated Tuel-processing technology. However, the MSCR iz not a
breeder, although it approaches break-even breeder operation. The ad-
ditional concepts are termed MSBR({Pa-Pb), SSCB(Pa), MOSEL{Ps~Fb), and MSCR.
The MSBR(Pa-Pb) designation refers to the MSBR{Pa) modified by use of
direct~contact cooling of the molten-salt fuel with molten lead. Lead
is dmmiscible with molbten salt and can be used as a heat exchange medium
within the reactor vessel to significantly lower the Tisslle inventory
external to the reactor. The lead also serves as a heat transport medium
between the reactor and the steam generators.

The SSCB(Pa) deszignation refers to a Single-Stream-Core Breeder with
direct protactinium removal from the fuel stream.” This is essentially a
single-region reactor having fissile and fertile material in the fuel
stream, with protactinium removal Trom this stream; in addition, the core
reglon is encloged within a thin metal membrane and is surrounded by a
blanket of thorium-containing salt. Nearly all the breeding takes place
in the large core, and the blanket "catches" only the relatively small
fraction of neutrons that "leak" from the core (this concept iz also
referred to as the one-and-one-half region reactor).

The MOSEL(Pa~Pb) designation refers to a MOlten-Salt Epithermal,
breeder having an intermediate-to-fast energy spectrum, with direct
protactinium removal from the fuel stream and direct-contact cooling of
the fuel region by molten lead. No graphite is present in the core of
this reactor.

The M3CR refers to a Molten-Salt Converter Reactor that has the
Tfertile and fissile material in a zingle stream. No blanket region is

o

employed, although a graphite reflector surrounds the large core.

The fuel-cyele performance characteristics for these reactors are
summarized in Table 8.73; in all cases the methods, analysis procedures,
Table 8.7.

Summary of Design Conditions and Fuel-Cycie Performance for Reactor Designs Studied

 

Design Concitions

. . . &
Reactor Designation

 

 

MSBR(Pa) MSBR MMSEBR(Pa.) MSBR(Pa-Pb) SSCB{Pa) MOSEL(Pa-Fb) MSCR
Dimensions, ft
Core h C
Height 12.5 12.5 7.9b 12.5 16,0 3.OC 20.8
Diameter 10,0 10.0 6,3 10.0 9.8 6.5 15.5
Bianke®t thickness
Racdial 1.5 1.5 2.0 i.5 1.2 3.0
Axial 2.0 2.0 2.0 2.0 0.0
Volumne fractions, core
Fuel J.169 0,169 0.17 0.16%9 0.185 0.5 0.105
Fertile 0.073 0.074 0.05 0.076 0.0 0.0 0.0 Rfi
Moderator 0.758 0.757 0,78 0.755 0.807 0.0 0.895 A
Salt volumes, ft°
Fuel
Corse 166 166 166 166 230 63.5 4’76
External 551 547 574 120 60C 0.7 654
Total 717 713 740 276 830 E4.2 113C
Fertile, total 1217 3383 1570 1324 G835 758 0,0
Fuel-salt composition, mole %
LiF 63.6 63.6 63.6 63.6 71.0 7L.0 70.0
Beiy 36.2 36.2 36.2 36.2 23,1 0.0 13.0
Th¥, 0.0 0.0 0.0 J.C 8.68 24.0 26.55
UF,, {fissile) 0.22 0.23 ¢.21 0,23 0.23 5.0 0.45
Core atom ratios
Th/U 41,7 39.7 28.4 41,5 37.7 4,76 36.7
C/U 5800 5440 5980 5520 6280 0.0 6525
Teble 8.7. (continued)

 

. . . a
Reactor Tesignaticn

 

 

MSTR(Pa) MSBR MMSER(Pa} MSBR{Pa-Ph) SSCR(Pa) MOSEL(Pa-Pb) MECR
Power density, core averags,
ww/liter
Gross 80 80 80 80 66 618 17
Tn fuel s&lt Ak 475 473 A2 341 1236 165
Heytron flux, cor
newtrons o2 s
Thermal 7.2 x 104 6.7 X 10%% 7.3 x 104 6.8 x 1024 6.1 X 10%% 0.0 » 10%4 1.9 x 1034
Pasth 12,1 X 10%% 12,1 x 104 11,7 % 10%% 12,1 x 104 10,0 x 1014 72.2 % 1014 2.7 X 104
Fast, over 100 kev 3.1 x 1034 3.1 x 104 3.0 x 1034 3.1 % 104 2.6 X 1014 23.3 x 1014 0.7 % 1014
Neubron preduction per 2.227 2,221 2.229 24220 2.226 2.260 2,201
ahsoroticn {ne)
Tuciesr and fuel-cycle perfcrmance
Fuel yield, %/year 7.95 4,86 7.21 17.2 6,63 16.3
Breeding ratio 1.07 10.5 1.07 1.08 1.06, 114 0.9 N
Fuel-cycle cost, mills/kvhr 0.35 0.46 0.38 0.25 0,37 0.13 0.57 W
Specific fiseile inventory, 0.68 Q.77 G.76 0. 34 0.68 0.99 1.63

xg/Ma (electrical)

=

b

“See text for explanation of resctor designations.

The core dlmensionsg for thnis case refer to one module of & four-module station.

C . . . s < . o
For this case, the core had sanular geometry; the fuel ammulus inside diameter was 3 Tt, and the outside diameter was 6.5 ft.

Cn

Use of

of di
inventory to abouw

irect-contact lead cooling would lower the iuel-cyele cost to zbout 0.32 mill/kwhr (electrical) and the speeific fissgile
0.41 kg/Mw {electrical).
220

and economic conditions employed were analogous to those used in obtain-
ing the reference MSBR design data. In genecral, fuel recycling was based
on fluoride-volatility and vacuum-distillation processing; direct prot-
actinium removal from the reactor system was also considered in specified
cases.

The results indicate the potential performance of fluoride-salt
systems utilizing a direct-contact coolant such as molten lead and the
versatility of molten salts as reactor fuels. They also illustrate that
single-region reactors based on MSRE technology have good performance
charscteristics. Since the capital, operating, and maintenance costs of
the M3CR should be comparable with those of the MSBR, the power-produc-
tion cost of an investor-owned MSCR plant should be about 2.9 mills/
kwhr (electrical), based on a load factor of 0.8. However, the lower
power costs of the MSBR(Pz) and MSBR plants and their superior nuclear
and fuel-conservalbion characteristics make development of tThe breeder
reactors preferable.

References

1., MSR Program Semiann. Progr. Rept. Feb. 28, 1966, ORNL-3936, p. 172.

 

P. R. Kasten, E. 3. Bettis, and R. C. Robertson, Desgign Studies of
1.000-Mw(e) Molten-Salt Breeder Reactors, ORNL-3996 (August 1966).

 

 

3. P. R. Kasten, Safety Program for Molten-Salt Breeder Reactors, un-
published internal report (July 29, 1966).

 
9, MOLTEN-SALT REACTOR PROCESSING STUDIES
M. B. Whatley

A close-coupled facility for processing the fuel and fertile streams
of a molten-salt breeder reactor (MSBR) will ve an integral part of the
reactor system. BStudies are in progress for obtaining data relevant to
the engineering design of such a processing facility. The processing
plant will operate on a side stream withdrawn from the fuel stream, which
circulates through the reactor core and the primary heat exchanger. For
a 1000-Mr (electbrical) MSBR approximately 14 £t of salt will be proc-
essed per day, which will result in a fuel-salt cycle time of approxi-
mately 40 days.l

The probable method for fuel-stream and fertile-stream processing
is shown in Fig. 92.1. The salt will Tirst be contacted with ¥y for re-
moval of U as volatlile UFg. Purified UFg will be obtained from the filu-
orinator off-gas (consisting of UFg, excess Fp, and volatile fission
product fluorides) by use of NaF sorption. It may be necessary to dis-
card as much as 5% of the salt leaving the fluorinator for removal of
fission products such as Zr, Rb, and Cs. A semicontinuous vacuum dis-
tillation will then be carried out on the remaining salt for the removal

CRNL-DWG 85-1BOIRZA

Fo RECYCLE mmrmsrmnnnnscmeececceg - p—) e —- L RECYCLE

  
  
   

 
   

     
  

£
e T

MgF,  COLD
SORBER  TRAP

     
    
 

 

FiiTis 3
MaFs 81 ft Yday

colD
TRAP

     
  

 
   
 
 

BLANKET SALY

 
 

SPENT

 

 
 

 

 

NGF + MgF ] NaF + Magi,
UFg Fs Fa
LiF - BeF,~ UF,
r 15 $t%/day
WASTE .
FERTILE SALT I.IF - BeF,—UF, LiF MAKEUP UFg
015 f1¥doy 1 =

LiF+Bef,

 

 
 

PRODUC T
= 221 Vg/duy

 

?fl_@‘fifi&’!’ RECOMBINER

     

 

 

WASTE = 0,059 #¥/day
TLi=078 kg day
FP's

 

 

 

-

 

Fig. 2.1. MSBER Fuel and Fertile Stream Processing.

227
228

of the rare eartns, Ba, Sr, and Y. These fission products will he re-
moved from the still ia a salt volume equivalent to 0.5% of the strean.
The barren salt, the purified UFg, and the makeup salt will then be re-
combined. This step involves reduction of UFg to UFs, mixing of these
streams, and sparging the resultant material with an Hp-HF stiream, Fi-
nally, the salt mixture may be filtered before return to the reactor.

9.1 Semicontinuous Distillation

 

J. R. Hightower L. BE. MclNeese

New measurements of the relative volatilities of NdF; and LakFj in
LiF have been made using a recirculating equilibrium still. These values
are lower than earlier data by a factor of about 50 and are in a range
(around 0.0007) where the proposed distillation step in the MSBR proc-
esging plant should work very well.

The present concept of the distillation step in the MSBRE processing
plant uses a continuocus feed stream and vapor removal for separating
rarve-earth Tission products (FP“S) trom the fuel salt; the less volatile
rare-~cartn FP's will accumulate in the still pot and will be discharged
periodically.’ A measure of the decontamination achieved in this step
is the relative volatility of the less volatile FP's compared to the
carrier salt. The relative volatility of component A compared to com-
ponent B is defined as

Y.p = %’;’% ’ )
B
where
thB = relative volatility of A compared to B,
Y = vapor-phase mole fraction,
X =z liquid~-phase mole fraction.
For systems in which the concentration of compouent A is swall and X
is nearly unity, the relative volatility can be approximated by
0 o &Y, /X, (2)

To achieve good decontamination from tne less volatile fission products
their relative volatilitles must be small.

oince the major constituent of the still pot will be LiF, experi-
mental measurements have been made with mixtures of rare~earth fluorides
in LiF. A cold~finger technigue gave approximate values for the relative
volatilities of six rare-earth fluorides with respect to LiF (975 to
1075°C) ranging from 0.0l to 0.05.2 These were high enough to sericusly
1limit the effectiveness of the proposed simple distillation scheme. More
accurate meagurements of the relative volatilities were called for.
229

ORNL-DWG 66-3933

 

 

   
   
  
   
   
 
  

 

/‘f =
(
( NICKEL } TO VACUUM
) PIPE PUMP
«
D
/
q 5 3/8—1'n.
2-1n. (””’ NICKEL
NICKEL PIPE | ) TUBING
«
«

L

 

 

 

BOILING
LIQUID —. _ Hf//fC%PDENiFD
T~ P AMPLE
\\
THERMOWELL~_ DA

AIR-WATER
MIXTURE

 

NICKEL
TUBING

Fig., 2.2. Diagram of Reciraulating Equilibrium S5till.

A disgram of the recirculating equilibrium still used in recent
work is shown in Fig. 9.2. The boiling section is a 12~in. length of
2~in.~diam nickel pipe. The condensing sectlon is made from l-in. nickel
pipe wrapped with cooling coils of l/4-in. nickel tubing. In the bottom
of the condenser is a condensatbte trap where liguid collects and overflows
a welr to return to the still pot. The vacuum pump is connected near the
bobttom of the condenser section.

After charging salt of known composition, the still is welded shut
and purged with argon. The desired pressure 1ls set, and the still is
heated to the desired temperature while cocoling the condenser. After
operating the stiil for a period of time at the selected conditions, the
still is pressurized with argon, cooled to room temperature, and cub
apart for examination and sampling. The concentrations of the rare-earth
fluorides in the condensate trap and in the still pot are used to calcu-
late relative volatilities according to Eg. (2).
230

In these experiments the determination of absolute values for rela-
tive volatilities has been hampered because the vapor samples have rare-
earth concentrations below the apalytical limit of detection. However,
upper and lower limits of the relative volatilities for 0.01 to 0.02 mole
fraction CeF3, NdF3, and LaFj in LiF al 1000°C and 0.5 mm Hg were deter-
mined and are given below:

<
0.0013 < 0y ;o < 0.0029

0.00055 <:OfidF3~LiF < 0.00089

= 0.00069,

These values are substanhially lower than those previously reported. The
samples have been submitted for analysis by a more sensitive analytical
method (neutron activation). Higher concentrations of rare-earth fluo-
ride will be used in the sbtill pot in future work in ovrder that rare-
earth fluoride concentrations in the wvapor phase will be higher than the
1imit of detection.

Thne importance of relative volatility in determining the operating
characteristics of the distillation system is shown by the following cal-
culation. Consilder the reboiler of a single-stage distillation system
which contains V moles of LiF at any time and a gquantity of RBeFs such
that vapor in equililibrium with the liquid has the composition of MSBR
fuel salt. Assume that MSBR fuel salt containing Xg moles of rare~earth
fluorides (REF) per mole of LiF is fed to the still pot at a rate of F
moles of LiF per unit tilme where it mixes with the liquid in the system,
Let the initial REF concentration in the liguid be Xg moles of REF per
mole of LiF, and let the concentration at any time © be X moles of REF
per mole of LiI'. From a material bhalance on RER,

& (VX) = FXq ~ FO , (1)

where

V = still liquid holdup, moles of LiF,
X = BEF concentration in still liquid, moles of REF per mole of
Liy,
I' = LiF feed rate to still, moles per unit time,
Xo = REF concentration in feed, moles of REF per mole of LiF,
X = relative volatility of REF referred to LiF.

This equation nas the solution

x =20 L - (1-a) A (2)

The total quantity of REF fed to the system at time t is (Fb + V)X, and
231

the gquantity of REF remaining in the ligulid at that time is VX. Thus
the fraction of the REF not vaporized at time t is

e o vx___ {1/e) [L - o - /Y] ()
REF = (Ft + V)Xo — 1+ Ft/V) :

Values for the fraction of REF retained in the still as a Tunction
of dimensionless throughput (Ft/V) are given in Fig. 9.3 for various
values of O, Approximately 91% of the REF will be retained in the still
when 99.5% of the LiF has been recovered 1f the relative volatility of
the REF is 0.00Ll; a retention of greater than 95% can be obtained for
the same LiF recovery if & has a value of 0.0005.

ORNL-DWG 66-11472

   
  

 

]
2:0.0005

0001

 

 

FRACTION OF RARE-EARTH FLUORIDE RETAINED IN STILL

 

 

 

 

 

 

 

 

0 50 100 150 200 250
Ft/V, OIMENSIONLESS STiLL THROUGHPUT

Pig. 9.3. Fraction of Rare-Earth Fluoride Retained in otil1.
232

9.2 Continuous Fluorination of a Molten Salt
L. E. McNeege

Uranium present in the fuel stream of an MSER must be removed prior
to the distillation step since UF, present in the still would not ve
completely volatilized and would in part be discharged to waste when the
atill contents are dumped pericdically. Eduipment 1s being developed
for the continuous removal of Ury from the fuel stream of an MSBR by
contacting the salt with Fp in a salt-phase-continuocus system. This
egquipment will be protected from corrosion by freezing a layer of salt
on the vessel wall; the heat necessary for maintaining molten salt ad-
Jacent to frozen salt will be provided by the decay of fission products
in the fuel stream. Present development work consisgts of two parts:

(1) studies in a continuous fluorinator not protected by a frozen wall,
and (2) study of a frozen-wall system suitable for continuous fluori-
nation but with which an inert gas is used. Experimental work on the
nonprotected system is well under way; the protected system is being
designed.

The nonprotected system consists of a l-in.-diam nickel fluorinator
72 in. long and auxiliary eguipment (Fig. 9.4) which allows the counter-
current contact of a molten salt with ¥Fy. Experiments can be carried
out with molten-salt flow rates of 3 to 50 em /min with fluorinator salt
depths of 12 to 54 in. The system is constructed of nickel with the ex-
ception of molten-salt transfer lines, which are Hastelloy N. The flu-
orinator off-gas passes through a 400°C NaF bed for removal of chromium

ORMNL-DWG 65-93544
CHROMATOGRAPH
FOR UF;, F, AND N, ANALYSES

1

METERED F, ~ |

 

 

S0DA LIME TRAP

 

7 kg
METERED N, FOR
SALT DISPLACEMENT SALT SAMPLING ]
A VESSEL
gE - 1.5-in-DIAM OFF - GAS

  
 

  
  

SALT RECEIVER

SALT FEED TANK | - 14 liters

14 liters

 

 

 

 

 

 

NICKEL FLUORINATOR
{in. DIAM
T2in. LONG

Fig. 9.4, Equipment for Removal of Uranium from Molten Salt by
Continuous Fluorination.
233

fluorides, a 100°C NaF trap for removal of UFg, and a soda~lime bed for
Fo disposal. Analysis for Fp, Ufg, and Ny 1s made prior to the 100°C
Nal' bed with a gas chromatograph.

Iixperiments have been carried out at 600 to 650°C using an Nal'-LiF-
7Zr¥, mixture containing 0.2 to 0.5 wt % UF; and having a melting point
of ~550°C. Salt feed rates of 5 to 21 cm?/min and Fu rates of 75 to
250 em®/min (STP) have been used with molten-salt depths of 40 to 52
in. Uranium removal during one pass through the fluorinator has varied
from 95% to 99.6% as determined from salt samples. FEguipment operation
has been smooth although several plugs have developed in salt transfer
lines and in the fluorinator off-gas,.

During the best run to date (CF—lO), a molten-salt feed rate of
20.7 cmfi/min and an ¥, feed rate of 250 em?® /min {STP) were maintained
for a 2.5~hr period. The concentration of U in the feed salt was 0,32
wt %, and the fluorinator was operated at 650°C with a salt depth of
48 in. The U concentration in the salt discharged from the fluorinator
during the last hour of operation was 0.0020 wt % zs determined from
four salt samples taken at 15-min intervals. Based on inlet and exit
U concentrations in the salt, 99,4% of the U was removed by the flu-
orinator. The equipment operated smoothly during the run, salt and gas
feed rates were constant, and the system appeared to have reached sieady
state during the last hour of operation.

Continuous fluorination of the fuel stream of an MSBR with equipment
of the type studied is considered feasible. Study of continuous f{lucri-
nation will be continued, and the use of a frozen wall for corrosion pro-
tection will be demonstrated.

9.3 Alternative Chemical Procesesing
Methods for an MSBR

 

 

. I. Cathers Ce E. Bchilling

Liguid-metal extraction was studied in a previous search for a salti
reprocessing method to replace vacuum ddstillation. These tests with
LisBel'y salt showed that the lanthanides were removed in most cases by a
reductive coprecipitation with Be to yield refractory insoluble beryl-
lides deposited at the salt-metal interface. The present problem was to
develop a reductive precipitation using ninimum quantities of L1 or Be
as reductant and involving simple physical separation of the precipitated
metals by filtration or settling. Deconbtamination factors were determined
wsing initial concentrations of 30 to 1000 ppm of Zr, Nd, Ia, Sm, Fu, Gd,
and Sr either singly or in various combinations.

Further studies on the liguid-metal extraction method were made using
Li-Bi alloys to test removal of the same elements with the exceoption of
Zr and N4,
Table 2.1. Reductlve Precipitation of Neutron Poisons from LipBeF,

Initial concentrations: Zr = 70 pom; Nd = 400 ppm; others £00-900 ppm

 

_ ppm spike in original salt

 

Factor of r

 

 

T : - T s . ; o 2 - _T_ _ef. . ; =
Ixperiment Reductant Fxcess Tempgrata*e ppm spike in treated salt Speg1§§ denti 1e9 in Met l.
Nunber Reductant (°c) Precipitates by X-Ray Analysisg
Zr La Nd Gd
1 Be 1 550-60C 2o 2.3 Be, 0-Zr (trzcel}, and TaBeys (trace)
2 Be 4.2 (NA)  550-60 55 2.0 Not examined
243 (Zr)
3 Be oty 550-95 2.0 8.0 TaBe1s sud Be (trace)
4 Be 103 550-80C 52 &-Zr, ZrBes, Be, and NdBejs N
5 Be 12 55060 395 Be >
6 Be 104 800 36 Cu (from filter) + unidentified
Lines
7 Be 162 1000 1% Yot examined
g Li 1.28 55075 234° 1.1 1.3 o-Zr; LiF + LiyBeF, in salt
9 Li 105 550—80 51 Cu (from filter)
10 Li 4 550 i3 6 Be, LaBe1s, FeBes (from crucible)”
rocess required I 's 1.43 1.7 20 1.05

 

“Tnitial concentration
Initial concentration
separation by settliing.

400 ppm Zr,
11.7% Zr.
Reduction Frecipitation

The reduction-coprecipitation studies were made on 15-g (7.5-cc)
samples of LipBeF, spiked with cold neutron poiscns as the fluorides;
the samples were contained in mild steel under a protective abmosphere
of argon., A typical test consisted of: preparation and sampling of the
original salt while molten; reduction with between 1 and 243 times the
theoretical amount of Li or Be ab temperatures in the range 500 to 1000°C
for about 90 min with stirring by an argon sparge; filtration through a
sintered Cu filter stick; recovery of frozen samples of treated salt from
the filter and, when possible, precipitated metals from the bottom of the
reactor. oalt samples were analyzed for lanthanides and zirconium by neu-
tron actlvation analysis, spark source mass spectrometry, or emission spec-
trometry. The metal precipitates were examined by x-~ray diffraction.

The decontamination factors DF (La) and DF (Gd) shown in Table 9.1,
when compared with the indicated process requirement at the bottom of
each column, show that adequate removals could be achieved with both Id
and Be using excess reductant in the range of 1 to 4 times the theoretical
reguirement for LnBeis deposition. The Llow values obtained in experiment
8 reflect the incomplete removal of Zr, present in this one experiment
as a major component (11‘7%). In this case the final concentration of
Zr {500 ppm) was comparable to the initial concentrations of Is and Gd,
and therefore sufficient to inhibit their reduction. The species iden-
tified in the metal precipitates recovered {see Table 9.1, experiments
1, 3, and 10) indicate that reduction of lanthanides in LisBel, with
elther L1 or Be leads exclusively to berylilides of the type ImBeis. The
x-ray identifications listed in Table 9.) represent only the most likely
specific beryllide since all the InBejs compounds of the lanthanide series
are isostructural, with very nearly identical lattice parameters, and
therefore are indistinguishable from each other when codeposited from =
mixture.

 

Fxamination of zirconium DF's shown in Table 2.1 show it to be re-
moved easlly from solutions in LizBelF, over a broad range of concentra-
tion by treatment with elther Ii or Be. It ig deposited elther as free
metal or from very dilute solutions (70 ppm) as ZrBe, (Table 9.1, experi-
ment 4). The value of DF (Zr) = 11 obtained at 1000°C in experiment 7
is of special interest since it suggests a solution to the high volatility
of Zr¥F,, a major problem encountered at this temperature in the vacuum-
distillation salt recovery method. Lithium or beryllium could be used
to retain Zr in the still pot as nonvolatile free metal. In experiment
4, due to the low zirconium concentration (70 ppm), ZrBe, was ldentified
along with G-Zr.,

In addition to the elements listed in Table 9.1, Sm, Bu, and Sr
were studied as parts of the mixtures used in experiments 1, 3, and 8§
plus other tests not listed. Aside from a IF (Sm) = 2 obzerved in ex-
periment 3 and a DF (Sr) = 1.14 in an unlisted test uging Idi at 550°,
no encouraging resulbs were obtained by this method for these elements.
Tabie 9.2. Li-Bi Alloy Extraction of Lanthanides from Molten LipBelly

Tritial concentration: lsnthanides = 800-L000 ppm each; Sr ~200 ppm

 

Contacting Conditions

 

C . -
_ original sall

mole fraction in metal

 

 

 

Ailoy Composition = Ctre"ted et KD T mole fraction in salb
(at. %) Temperature  Tlme Volume Ratio = mEs
°n ) e ) - 3
(“c. (win)  (alloy/sale) .= g A Ge St e Sm s ad Sr
5,0 Li—94.1 Bi 55060 160 0.71 3.0 3.5 1.0 2.0 1.5 0.85 2.2 0.5 0.32 0.18
(C.. . =2.56 at. % Li)
final
30 Li-=70 Bi 55060 120 .l 6.0 2.0 8.0 5.0 2 1.1 3.1 17.1 2.00 Q.97
(C.. = 10.05 at. % Li)
final
DE reguired 1.7 5.0 1.43  1.05 1.04

 

9¢c
237

Li-Bl Alloy Extraciion

 

Liguid-metal extraction tests with sclutions of ILi in Bi were made
in the same equipment but without filtration of either phase. Samples
were recovered after settling, quenching to room temperature, and sac-
rificing the crucible. Both salt and metal slugs were cleaned of sur-
face deposits before sampling for analysis.

The decontamination factors (DF's) and distribution coefficients
(KD) obtained in liquid-metal extractions of spiked carrier salt using
Ii-Bi alloys are summarized in Table 2.2. TFor the spectrum of spikes
present (La, Sm, Bu, Gd, and Sr) adequate decontamination was achieved
for all but Sm. The generally high DF's cannot be explained in most
cases by a true extraction mechanism, a fact indicated by the low dis-
tribution coefficlents shown in Table 9.2. Also, poor material balances
were obtained through use of analytical results from samples of the salt
and metal phases. The explanation for removal of the elements in question
ig precipitation as interfacial solids which were later identified by
x-ray analysis to be beryllides of the InBeps type. The high Kj observed
for Fu indicates that this element was removed almost exclusively by ex-
traction when using 30 at. $ Li-Bi alloy.

References

1. C. D. Scott and W, L. Carter, Preliminary Design Study of a Continuocus
Mluerination ~ Vacuum-Distillation System for Regenerating Fuel and
Fertile Streams in a Molten Salt Breeder Reactor, ORNL-3791 (January
1966).

2. MSR Program Semiann., Progr. Rept. Feb. 28, 1966, ORNL~3936, p. 199.

 
239

OAK RIDGE NATIONAL LABORATORY
MOLTEN-SALT REAGCTOR PROGRAM

JUNE 15, 1965

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MRC

MEC
MEC
M&C
M&C
MBC
M3.C
Ma.C
M&C
M&C
M&C
M&C
M&C
M&C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R. 8. BRIGGS, DIRECTOR D
P. R. KASTEN, DEPUTY DIRECTOR R
: METALLURGY
MSBR DESIGN STUDIES MSRE OPERATIONS COMPONENT AND SYSTEMS DEVELOPMENT JAC DESIGN AND DEVELOPMENT REACTOR CHEMISTRY
E. S BETTIS R W. R. GRIMES* RC G. M. ADAMSON*
o ‘ = P. N. HAUSENREICA R DUNLAP 5COTT R R. L. MCGRE 1&C F. F. BLANKENSHIP RC W. H, COOK**
W. L. CARTER* cT £ G. B0HLMANN* RC B A CANONICO®
A. G. GRINDELL* R H. F. McDUFFIE* RC o
-G - E. A. FRANCO-FERREIRA
H. T. KERR R :
2 C ROBERTSON R MSRE ON-SITE CHEMISTRY C. R. KENNEDY
L [ = H. E. McCOY*
W. TERRY S. 5 XIRSLIS RC 1. L. GRIFFiTH*
). K. JONES R . P. D. NEUMANN® RC V. G. LANE"
COORDINATION NUCL EAR AND MECHANICAL ANALYSIS ENGINEERING ANALY SIS A.H. ANDERSON I&C E J LAWRENCE®
D. G. DAVIS 1&c IRRADIATION PROGRAM R G SHOOSTER®
C.H. GABBARD"* R J. R. ENGEL R 2 1 KEDL a P. G, HERNDON 18C E M. THOMAS®
CoH. GABBARD** R - j J. W. XKREWSON* 18C J. M. BAKER RC . b WADDELL*
3 PIPER R T. M. CATE 1&C E. L. COMPERE RC B & WILLIANS
1. A WATTS R B. J. JONES i&C H. C. SAVAGE RC e
J. E. SAYOLAINEN
J. ROHART RC
1. W. MYERS RC
R. M. WALLER RC
. REMOTE MAINTENANCE
OPERATION PHYSICAL AND iNORGANIC CHEMISTRY
R. H. GUYMON R INSTRUMENTATION AND CONTROLS R. BLUMBERG R C. F. BAES, JR RC
J.H. CROWLEY R - IR, SHUGAR™ R C. J. BARTON* RC
P. H. HARLEY R J R TALLACKSON I8C G. D. BRUNTON® RC
V. D, HOLT R R K. ADAMS" 15C S. CANTOR* RC
T. L. HUDSON R G H. BURGER a0 J. H. SHAFFER* RC
A. 1. KRAKOVIAK R e o MARTIN' 140 R E. THOMA™* RC
H. R. PAYNE R C E MATHEWS 12C C. £. L. BAMBERGER* RC
H.C. ROLLER R R. W. TUCKER 1aC KUCLEAR ANALYSIS F. A DOSS RC
R. STEFFY R ¢ £ ELLIS 1AC H. E. FRIEDMAN" RC
W. C. LLRICH R ¢ K KIRKWOOD* 12C MSRE COMPONENTS 8. E. PRINCE R L. 0. GILPATRICK* RC
T. ARNWINE R S e LeTELLIER* 1aC AN ST Q T. W. KERLIN® R J. D. REDMAN* RC
W. H. DUCKWORTH R T. SAMUELSEN 13,0 R' B' GALLAHER R K. A. ROMBERGER* RC
J. D. EMCH R V. R. SiLER" 1&C “' R'ICHARDSO'\i R D. R. SEARS™ RC
E. H. SUINN R R s = H.H. STONE" RC
T.G. HILL R T E AR 2 W. K. R. FINNELL RC
C.C. HURTT R C. A GIFFORD 8. F. HiTCH RC
1. C. JORDAN R C. E. ROBERTS RC
H. J. MILLER R . P. TEICHERT RC
L. E. PENTON R
W. E. RAMSEY R
R. SMITH, 2R. R CHEMISTRY
J. L. STEPP R p— FUEL PROCESSING DEVELOPMENT
J. L. UNDERWOOD R R. E. THOMA** RC
S, R, WEST g M. E. WHATLEY* cT
Wi - 5
J. E.WOLFE W. L. CARTER*" €T
J. R. HIGHTOWER cT
AC ANALYTICAL CHEMISTRY DIVISION R. B. LINDAUER** cT
€T CHEMICAL TECHNOLOGY DIVISION PUMP DESIGN AND DEVELOPMERT L. E. McREESE T
_ . . C. E. SCHILLING cT
D DIRECTORS DIVISION CHEMICAL PROCESSING A . GRINDELL" R J. BEAMS cr ANALYTICAL CHEMISTRY
s S - G ORI : ). SHIPMAN cT
14C :NSTRUMENTATION AND CONTROLS DIVISION DESIGN LIAISON L P G, SMITH R cJ LT CORBIN® AC
M&C METALS AND CERAMICS DIVISION — R. B. LINDAUER"* cT L. V. WILSON ™ R J. C. wHITE® AC
R REACTOR DIVISION €. K. McGLOTHLAN R A S MEYER* AC
. R. F. APPLE AC
£ EMl DIVISION
RC REACTOR CHEMISTRY DIVISIO e e
UT UNIVERSITY OF TENNESSEE J. M DALE AC
. - F. K. HEACXER AC
PART TIME ON MSRF AR A
DUAL CAPACITY MATERIALS W. L. MADDOX* AC
MAINTENANCE G. MAMANTOV uT
A A W. H, COOK** M&.C D. L. MANNING* AC
8. H. WEBSTER — J. P. YOUNG® AC
A. F. GILLEN R
L. ®. PUGH R

 

 

 

 
 

   
W30y H~WwWo R
>

i
O

11.
12.
13.
14.
15.
15.
17-26.
27
28.
29.
30.
31.
32
33.
34
35.
36.
37.
38.
39.
40
41.
4o
43

Lly

45,
40 .
47
48 .
49.
50,
51.
52.
53.
5
55.

2
20

Ga
L.
C.
J.
e
E.
De
.
R.
H.
A.
E.

G
M.

7
-

R.
.
Co
S
D.

E.
J
W.
L.
G
Ja
F.
dJde.
D.
W
Je
R.
D.
N.
J
E.
W.
De
A.
Je
C.

'W’

R.
R.
w.

241

ORNTL-4037

UC-80 — Reactor Technology

INTERNAL DISTRIBUTION

M. Adamson
7. Alexander
F. Baes

M. Baker

. Beall

S5. Bettis

S. Billington
F. Blankenship
Blumberg

F. Bauman

L.. Boch

G. Bohlmann
J. Borkowski
E. Boyd

A, Bredig

J. Breeding
B. Brigge

R. Bruce

H. Burger
Cantor

W. Cardwell
L. Carter

L. Compere
A. Conlin

H. Cook

T. Corbin
A. Cristy

L. Crowley
L. Culler

M. Dale

re Davis

W. Davis

He DeVan

G. Donnelly
A, Douglas
. Dunwoody
R. Engel

P. Epler

K. Brgen

E. Ferguson
P. Fraas

H TFrye, Jr.
H. Gabbard
R. Gall

B. Gallaher
G. Gilliland
R. Grimes

57
58.
59.
60.
61,
62.
63.
64 o
65.
66 .
&7.
68,
69.
70.
71,
72
73.
"4,
75-80.
81l.
g2.
83.
84 .
85.
86 .
87 .
88.
89.
90.
91.
92,
93.
D
95.
9.
V7
28.
99.
100.
101.
102.
103.
104
105.
106.
107.
108.

A.
R.
R.
P.
Ca
P.
P.
R.
M.

-

H.
Ve
P.
A.
A
T
H.
W.
P.
R.
M.
M.
C.
T
o.
A
.

hl
®

Je
C.
E.
T.
R.
A
R.
M.
R.
Ho
R.
N
We
C.
H.
H.
C.
L.

G.
H.
P.
H.
Se
N.
G.
.
R.
Ce
W.
De.
P.

Grindell
Guymon
Hammond
Harley
Harrill
Haubenreich
Herndon
Hibbs (Y-12)
Hill

Hige
Hoffman
Holt

Holz

Hollaender

S e

.T_ln

Householder
Hudson

Inowye

H.
R.
J.
T.
J.
R.
W
T.
T.
W.
e
A.
E.
Je
A,
B.
P
5
I.
N.
Ge
B.
Ca
R.
E.
B.
F.
K.
E.

Jordan
Kasten
Kedl
Kelley
Kelly
Kennedy
Kerlin
Kerr
Krakoviak
Krewgon
Lamb

Lane
Larson
Lawrence
Iincoln
Lindauer
i tman
Livingston
Taundin
-Ly—On
MacPherson
MacPherson
Maienschein
Martin
Mathews
McCoy
MeDuffie
McGlothlan
McNeese

F. C. Miller
109.
110.
111.
112.
113.
114,
115-116.
117.
118.
119,
120.
121.
122.
123.
124.
125,
126,
127.
128.
129.
130.
131.
132.
133.
134,
135.
136.
137.
138.

209.
210.
211.
212,

213.
214 .
215.
216.
217.
218.
219.
220,
221-492.

C.
We
R.
K.
J e
W
R.

P.
H.
We
H.
B.
Je.
M.

a.
M.
.
A.
D.
H.
J .
E.
M.
G.
A.
P.
A.

D.
A
W.
W.

T.
M.
E.
W.
Je
R.
M.

A. Miils
R. Mixon
I,.. Moore
Zi. Morgan
C. Moyers
R. Osborn
B. Parker
F. Parsly
Patriarca
R. Payne
B. Pike

B. Piper
E. Prince
L« Redford
Richardson

. C. Robertson

C. Roller

W. Rosenthal
C. Davage

W. Bavolainen
Scott

B. beagren
He Bhaffer
D. Shipley
J. Skinner
M. Slaughter
N. Smith

G. Smith

H. Snell

242

13%.
140.
141
142,
143
1.
145.
146.
147
148.
149,
150.
151.
152
153.
154 .
155.
156.
157,
158.
159.
160.
161.
162-163.
164-168.

169-171.
172-207.
208.

W
I.
R.
C.
C.
Je
E.
R.
Ge
D.
R.
We
D.
G.
B.
A,
J.
M.
Ge
J.
L.
G

F-

Spencer

oplewak
Steffy

E.
D.
R.
H.
E.
M.
B.
We
C.
C.
M.
H.
M.
R.
E.
C.
C.
V.
Je

Stevenson
susano
Tallackson
Taylor
Thoma
Tolson
Trauger
Tucker
Ulrich
Watkin
Watson
Webster
Weinberg
Weir
Whatley
Williams
White
Wilson
Young

Biology Library

Reactor Division Library

ORNI, — Y-12 Technical Library
Document Reference Section
Central Research Library
Laboratory Records Department
Laboratory Records, ORNL R.C.

EXTERNAL DISTRIBUTION

¥. Cope, Atomic Energy Commission, RDT Site Office (ORNL)
Gilambussc, Atomic Energy Commission, Washington

J. Larkin, Atomic Energy Commission, CRO

B. McDonald, Battelle-Pacific Nortawest Laboratory, Hanford
Washington
W. McIntosh, Atomic Energy Commission, Washington

Shaw, Atomic Energy Commission, Washington

E. Binclair, Atomic Energy Commission, Washington

L. Bmalley, Atomic Energy Commission, ORO

A. Swartout, 270 Park Avenue, New York 17, New York

. Sweek, Atomic Energy Commission, Washington

J. Wnitman, Atomic Energy Commission, Washington

Regearch and Development Division, AEC, ORO

Given distribution as shown in TID-4500 under Reactor Technology
category (25 copies — CFSTI)