NAT_MSREexperience.txt

EXPERIENCE WITH THE MOLTEN- SALT

REACTOR EXPERIMENT

PAUL N. HAUBENREICH and J. R. ENGEL

 

KEYWORDS: molten-salt re-
actors, operation, performance,

testing, MSRE

Oak Ridge National Labovatory, Oak Ridge, Tennessee 37830

Received August 4, 1969
Revised September 19, 1969

 

The MSRE is an 8-MW(th) reactor in which
molten fluoride salt at 1200°F circulates through a
cove of graphite bars. Its purpose was to demon-
Strate the practicality of the key features of
molten-salt power reactors.

Operation with “°U (33% envichment) in the fuel
salt began in June 1965, and by March 1968
nuclear operation amounted to 9000 equivalent
full-power hours. The goal of demonstrating re-
liability had been attained—over the last 15 months
of ¥°U operation the reactor had been critical 80%
of the time. At the end of a 6-month run which
climaxed this demonstration, the reactor was shut
down and the 0.9 mole% wuranium in the fuel was
stripped very efficiently in an on-site fluovination
facility. Uranium-233 was then added to the
carriev salt, making the MSRE the world’s first
reactor to be fueled with this fissile wmaterial.
Nuclear operation was rvesumed in October 1968,
and over 2500 equivalent full-power hours have
now been produced with 3U.

The MSRE has shown that salt handling in an
operating veactor is quite practical, the salt
chemistry is well behaved, there is practically no
corrosion, the nuclear charvacteristics are very
close to predictions, and the system is dynami-
cally stable. Containment of fission products has
been excellent and wmaintenance of vradioactive
components has been accomplished without un-
reasonable delay and with very little radiation
exposure.

The successful operation of the MSRE is an
achievement that should strengthen confidence in
the practicality of the molten-salt reactor concept.

 

NUCLEAR APPLICATIONS & TECHNOLOGY

INTRODUCTION

The paper by Rosenthal et al.! describes the
origin of the molten-salt reactor concept and how
the encouraging results of the aircraft reactor
program led to recognition of the potential of
molten-salt reactors for economical production of
electricity. The role of the Molten-Salt Reactor
Experiment (MSRE) was to demonstrate the prac-
ticality of this high-temperature fluid-fuel concept
which seemed so promising on the basis of
materials compatibility information and calculated
fuel cycle costs. When design of the MSRE was
initiated in 1960, therefore, a primary objective
was to make the reactor safe, reliable, and
maintainable. How well these efforts succeeded is
told in this paper.

DESCRIPTION OF THE MSRE

The MSRE was designed® to use essentially the
same materials as the proposed molten-salt
breeders but, for economy and simplicity, there
was no attempt to make the reactor actually
breed. The core is small (54 in. in diam X 64 in.
high) so that the neutron leakage is high, and there
is no blanket of fertile material. The fuel salt
contains no thorium because at the time the
reactor was being designed we were thinking in
terms of the two-fluid breeder and we made the
MSRE salt similar to the anticipated breeder fuel
salt. The power level was to be limited to 10
MW(th) or less, but we wanted to try fairly large
molten-salt pumps. As a result, the temperature
rise of the salt as it passes through the core is
<50°F. The average temperature of the fuel salt
was to be 1200°F in the range proposed for the
power breeders. Even at this temperature, the
vapor pressure of the salt is <0.1 mm Hg, so the
pressure of the gas blanket over the salt was set

VOL. 8 FEBRUARY 1970
at only 5 psig. The flowsheet of the MSRE (Fig. 1)
shows the normal operating conditions at 8 MW,
the maximum heat removal capability of the air-
cooled secondary heat exchanger. The special
materials used in this reactor system are listed
in Table I.

The physical arrangement of the salt systems
is shown in Fig. 2. The building housing the
reactor is the one in which the Aircraft Reactor
Experiment was operated in 1954. The cylindrical
reactor cell was added for the Aircraft Reactor
Test (which was never built) and was adapted for
MSRE use.

Details of the MSRE core and reactor vessel
are shown in Fig. 3. The 54-in.-diam core is
made up of graphite bars, 2 in. square and 64 in.
tall, exposed directly to fuel which flows in
passages machined into the faces of the bars. The
graphite was especially produced® to have low
permeability, and, since salt does not wet the
graphite, very high pressure would be required to
force any significant amount of salt into the
graphite. Some cracks developed in the manu-
facture of the graphite, but cracked bars were
accepted when tests showed effects attending heat-
ing and salt intrusion into cracks were incon-
sequential.

All metal components in contact with molten
salt are made of Hastelloy-N (formerly called
INOR-8). Metal corrosion by salt mixtures con-
sists of oxidation of metal constituents to their
fluoride salts, which do not form protective
films.* Attack is therefore limited only by the

Haubenreich and Engel = EXPERIENCE WITH MSRE
thermodynamic potential for the oxidation re-
action, and is selective, removing the least-noble
constituent, which in the case of Hastelloy-N is
chromium. However, the diffusion coefficient of
the chromium in the metal is such that there is
practically no chromium leaching at temperatures
below 1500°F. Impurities in the salt, such as
FeF,, react with Hastelloy-N, but this is a limited
effect which goes to completion soon after the
salts are loaded. The metallurgy and technology
of Hastelloy-N have been throughly developed® and
it has been approved for construction under ASME
Unfired Pressure Vessel and Nuclear Vessel
Codes. Hastelloy-N is stronger than austenitic
stainless steel and most nickel-base alloys but,
like these metals it is subject to deterioration of
high-temperature ductility and stress-rupture life
by neutron irradiation. (These effects are due to
accumulation in grain boundaries of helium pro-
duced by 7z, o reactions.) In the MSRE neutron
spectrum the fast neutron reactions with nickel
are insignificant compared to the slow neutron
reactions with impurity boron. Careful analysis
of stresses and neutron fluxes in the MSRE® led to
the conclusion that the service life of the reactor
vessel would extend at least 20 000 h beyond the
point at which the properties of the metal began
to be seriously affected by the neutron exposure.

The control rods are flexible, consisting of
hollow cylinders of Gd,Os;-Al;O3; ceramic, canned
in Inconel and threaded on a stainless-steel hose
which also serves as a cooling-air conduit. An
endless-chain mechanism, driven through a

TABLE I
MSRE Materials

 

Fuel Salt
Composition:

Properties at 1200°F (650°C)

Density

Specific heat
Thermal conductivity
Viscosity

Vapor pressure

Liquidus temperature

Coolant salt?
Moderator
Salt containers

Cover gas

 

 

"LiF-BeF;-ZrFs-UF4 (65.0-29.1-5.0-0.9 mole%

141 1o/t 2.3 %/cm?’
0.47 Btu/lb-°F 2.0 x 10° J/kg-°C
0.83 Btu/h-ft-°F 1.43 W/m-°C
19 1b/h-ft 29 kg/h-m
<0.1 mm Hg <1 x10™* bar
813°F 434°C

"LiF-BeF2 (66-34 mole%)
Grade CGB graphite
Hastelloy-N (68 Ni-17 Mo-7 Cr-5 Fe)

Helium

 

 

aA nother batch of salt of this composition is used to flush the fuel system before it is opened to minimize fission
product escape and again after it is resealed to pick up moisture that may have entered.

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

FEBRUARY 1970 119
EXPERIENCE WITH MSRE

Haubenreich and Engel

d38 3aiyon14
WNIAOoS

ANVL
39vYH01S
13n4

ANVL
NIvya
1NVI00D

 

 
 

 

 

S431714 ‘H

31N10S8Yy
o1

e

d01vIavy

m——-

SV9- 440 3AILOVOIAVY —————
SVO HIA0D WNNITIH +eeeeseenenns
1vS INVI009 ez
17VS 1304 eem—

aN3937

 

Y31dNVS

"HYSIN 9Y3 JO 309ys molJ udrsa(g

 

 

 

 

coo i o

 

 

 

o_m.a S

(dAl) 39NVId 373344

e se st st e ot e sttt taoueTy]
............

 

 

seseeseeeSHI M4 31NT0SAY OL

dANd
1NVT002

 

 

 

 

‘T "8ig
Z ON MNvL b 'ON MNVL
NIvyQa
13Nn4d
IVODHVHD
xXNnv
Cererenerenes @reccnnnneceniesleciiiiiniinnndonnn, ... @eeeennns
lllllllllllll —

(dAL) INIVA 323394~

ANVL MOT483A0

d3IHIIYNT
-431dNVS

N

 

 

   

 

 

 

 

aas
WOJHVHD
NIVN

 

 

 

 

[
_
|
|
|
soeeNJoo o ol
MIN 8 “ I
Y3MOd | 35537 | NOILY I LN3A
HOLOV3Y _ 9078
'Wd'9 002} _
4o O} "
_
_
1012} |
| dNAI0oH
D 5 | ||sv9-140
o b _
L—

bis

coe@ oo

Q *ercee

 

 

 

 

S

 

 

 

 

 

NV4  MOVLS

o
SY314

31N70S8v

 

 

 

 

 

FEBRUARY 1970

VOL. 8

NUCLEAR APPLICATIONS & TECHNOLOGY

120
  
 
  
   

REACTOR CONTROL
ROOM

REMOTE MAINTENANCE
CONTROL ROOM

Haubenreich and Engel = EXPERIENCE WITH MSRE

 

 

 

 

 

 

 

 

   

 

 

Fig. 2.

clutch, raises and lowers the rods at 0.5 in./sec.
When scrammed, the rods fall with an acceler-
ation of ~12 ft/sec?.

The bowl of the fuel pump is the surge space
for the circulating loop. Dry, deoxygenated he-
lium at 5 psig blankets the salt in the pump bowl.
About 50 of the 120 gal/min discharged by the
pump is sprayed into the gas space to provide
contact between salt and cover gas to allow '*Xe
to escape from the salt. (The solubility of xenon
and krypton in the salt is very low,) A flow of
4 liters/min STP of helium carries the xenon and
krypton out of the pump bowl, through a holdup
volume providing ~40-min delay, a filter station,
and a pressure-control valve to charcoal beds.
The charcoal beds consist of pipes filled with
charcoal, submerged in a water-filled pit at

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

 

 

 

g6 'z é"’g
s - -/ y
» | -7
!
/ |
|
i
¢ ,
Vot
8
.~ 1. REACTOR VESSEL 7. RADIATOR
2. HEAT EXCHANGER 8. COOLANT DRAIN TANK
3. FUEL PUMP 9. FANS
4. FREEZE FLANGE 10. FUEL DRAIN TANKS
5. THERMAL SHIELD 1. FLUSH TANK
6. COOLANT PUMP 12. CONTAINMENT VESSEL

13. FREEZE VALVE

Layout of the MSRE,

~90°F. The beds are operated on a continuous-
flow basis and delay xenon for ~90 days and
krypton for ~7 days. Thus, only stable or long-
lived gaseous nuclides are present in the helium
which is discharged through a stack after passing
through the beds.

All salt piping and vessels are electrically
heated to prepare for salt filling and to keep the
salt molten when there is no nuclear power.’ In
the reactor and drain-tank cells, where radiation
levels make remote maintenance necessary,
heater elements and reflective metal insulation
are combined in removable units. Thermocouples
under each heater monitor temperatures to avoid
overheating the empty pipe. The radiator is
equipped with doors that drop to block the air duct
and seal the radiator enclosure if the coolant salt

FEBRUARY 1970 121
Haubenreich and Engel @ EXPERIENCE WITH MSRE

   
   
   
   
  
 
 
  
 

GRAPHITE SAMPLE ACCESS POR

 

N

FUEL OUTLET

CONTROL ROD THIMBLES

 

CORE
CENTERING GRID

\ 't
\ g Il e
GRAPHITE-MODERATOR" £ o)
STRINGER _. _

FUEL INLET/

REACTOR CORE CAN—T

REACTOR VESSEL—]

 

ANTI-SWIRL VANES
VESSEL DRAIN LINE

Fig. 3.

circulation stops and there is danger of freezing
salt in the tubes.

Around the reactor-vessel furnace is a shield,
16 in. thick consisting of a tank of stainless steel
filled with steel balls and circulating water. The
shield absorbs most of the energy of neutrons and
gammas escaping the reactor vessel (20 kW/MW of
reactor power). It also cuts down on neutron
activation of components in the reactor cell,
facilitating maintenance. The cooling-water sup-
ply for the shield is deaerated to remove radio-
lytic gas.

Neutron chambers are located in tubes in a
36-in.-diam water-filled shaft that slopes down
through the reactor cell to the inner surface of the
thermal shield. Included are 3 uncompensated ion
chambers driving safety channels, 2 compensated
chambers, and 2 servo-driven fission chambers

122 NUCLEAR APPLICATIONS & TECHNOLOGY

  
   

FLEXIBLE CONDUIT TO
CONTROL ROD DRIVES

COOLING AIR LINES

ACCESS PORT COOLING JACKETS

REACTOR ACCESS PORT

OUTLET STRAINER

FLOW DISTRIBUTOR

   

 

MODERATOR
SUPPORT GRID

Details of the MSRE core and reactor vessel.

that provide a 10-decade power indication. The
compensated chambers are connected to multiple-
range ammeters and a flux- or power-servo
system. Any one of the three rods can be
connected as a regulating rod to the servo system.
The fuel salt, which is a mixture containing both
alpha emitters and beryllium, is itself a strong
neutron source, but there is also an Am-Cm-Be
start-up source in a thimble in the thermal shield.

There are no mechanical valves in the salt
piping. Instead, flow is blocked by plugs of salt
frozen in flattened sections of the lines. Temper-
atures in the ‘‘freeze valves’’ in the fuel and
coolant drain lines are controlled so they will
thaw in 10 to 15 min when a drain is requested. A
power failure of longer duration also results in a
drain because the cooling air required to keep the
valves frozen is interrupted.

VOL. 8 FEBRUARY 1970
The drain tanks are ~4 ft in diameter, but the
molten fuel is safely subcritical because it is
undermoderated away from the core graphite.
Water-cooled bayonet tubes extend down into
thimbles in the drain tanks to remove up to 100
kW of heat if necessary. Steam produced in the
tubes is condensed and returned by gravity to
provide reliability.

The reactor cell and drain-tank cell are con-
nected by a large duct so they form a single
containment vessel. The tops of the two cells
consist of two layers of concrete blocks, with a
weld-sealed stainless-steel sheet between the lay-
ers and the top layer fastened down to contain
internal pressure. The drain cell and the top of
the reactor cell were designed for 40 psig and
were tested at 48 psig. The sides and bottom of
the reactor cell, built to house the Aircraft
Reactor Test, were tested at 300 psig before the
top was modified for the MSRE. The cell atmo-
sphere is kept at 130°F by water-cooled, forced-
air space coolers. »

The cooling ‘‘air’’ for the freeze valves and the
fuel-pump bowl is actually reactor-cell atmo-
sphere, compressed and cooled. Some of the
blower output is discharged past a radiation
monitor and up the ventilation stack to keep the
reactor cell and drain-tank cell at -2 psig during
operation. A small bleed of nitrogen into the cell
keeps the oxygen content at 3% to preclude fire if
fuel-pump lubricating oil should spill on hot
surfaces. The cell inleakage is calculated from
purge flow-in, discharge rate, and differential
pressure between the cell and a temperature-
compensated, sealed volume in the cell.

All the components in the reactor and drain-
tank cells are designed and laid out so they can be
removed by the use of long-handled tools from
above. When maintenance is to be done, the fuel
is secured in a drain tank and salt plugs frozen in
the connecting lines. The upper layer of blocks is
removed, a hole is cut in the stainless membrane
and one or two lower blocks are removed over the
item to be worked on. A large duct from the
reactor cell to the upstream side of the ventilation
filters is opened to draw air down through the
shield openings. Tools, lights, and viewing de-
vices are inserted through fitted openings in a
steel work shield. Items removed from the cells
are bagged in plastic and removed for storage in
another cell in the reactor building, inspection, or
disposal.

In the same building, adjacent to the drain-tank
cell, there is a simple facility for processing the
fuel for flush salt.? The purpose of twofold: to
remove oxide contamination from the salt if this
should be necessary and to recover the uranium
from the salt at the end of the experiment. One

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

Haubenreich and Engel EXPERIENCE WITH MSRE
whole batch of salt (~75 ft®) is transferred into a
tank where it is sparged with gas—either hydrogen
fluoride to remove H,O and convert the metal
oxides to fluorides, or fluorine gas to convert UF,
to the volatile UFg which leaves the salt to be
trapped on a bed of sodium fluoride pellets.
Ancillary equipment includes gas-handling equip-
ment and a large filter to remove solids from the
salt after processing.’

CHRONOLOGY OF THE MSRE*

Design of the MSRE began in the summer of
1960 and 18 months later fabrication of the
Hastelloy-N primary-system components was
started. Production of the newly developed, low-
permeability graphite proved to be the critical
path but by the end of 1963 it was ready for
assembly in the reactor vessel. Installation of the
salt systems was completed in the summer of
1964. Activities during the start-up period that
followed are outlined in Fig. 4.

Prenuclear tests showed that all systems oper-
ated well and there were no unanticipated prob-
lems in handling the molten salt. The fuel carrier
salt that was circulated in the final prenuclear
test contained depleted uranium (150 kg). The
reactor was made critical by adding 69 kg of
highly enriched **U in the form of a UF,-LiF
eutectic salt containing 61% uranium by weight.
Most of the uranium was added to the salt in the
drain tanks, with the final amounts added, 85 g at a
time, through the sampler-enricher to the circu-
lating fuel salt. After the initial criticality on
June 1, 1965, more uranium was added while the
control rods were calibrated and the reactivity
coefficients that would be needed in later analysis
of the system were measured.

At the end of the zero-power nuclear experi-
ments, the reactor was shut down while final
preparations for power operation were made.
This consisted mainly of finishing construction;
the only repair or modification dictated by the
previous testing was replacement of the heat-
warped radiator doors with doors of an improved
design.

After tests in the kilowatt range showed that
the dynamics of the system were as expected, the
approach to the full power was started in January
1966. Trouble was encountered immediately—
within a few hours, plugs developed at several
points in the fuel off-gas system and the reactor

 

2 A ctivities in the Molten-Salt Reactor Program, includ-
ing the MSRE, are described in a series of semiannual
progress reports issued by Oak Ridge National Lab-
oratory.

FEBRUARY 1970 123
Haubenreich and Engel =~ EXPERIENCE WITH MSRE

INSPECTION AND
PRELIMINARY TESTING

PRENUCLEAR TESTS
OF COMPLETE SYSTEM

FINAL PREPARATIONS
FOR POWER OPERATION

 

N

FINISH LEAKTEST,
INSTALLATION  PURGE & HEAT
OF SALT SYSTEMS SALT SYSTEMS
- T8 ORI

 

 

TEST AUX. SYSTEMS
77772

 

 

INSTALL
CONTROL RODS
SAMPLER-ENRICHER
22227777773

 

 

N f )

INSTALL CORE SAMPLES
INSPECT FUEL PUMP

HEAT-TREAT CORE VESSEL
TEST SECONDARY

CONTAINMENT
FINISH VAPOR-COND. SYSTEM )

TEST TRANSFER, OPERATOR MODIFY CELL PENETRATIONS  ADJUST & MODIFY
OPERATOR TRAINING FILL & DRAIN OPS. TRAINING REPLACE RADIATOR DOORS  RADIATOR ENCLOSURE
——— — 222 B 77277777777 22272723 2772772777773
INTO DRAIN TANKS LOAD 8  IN ZERO-POWER LOW- POWER
COOLANT FLUSH  CIRCULATE CIRCULATE NUCLEAR (0-50 kw)
SALT SALT  C & FL SALTS CARRIER EXPERIMENTS EXPTS.
2222727727777 T

Fig. 4.

was shut down. Investigation showed that some-
thing like a varnish mist had plugged the small
passages and a small filter in the off-gas system.
There had evidently been some 0il in the off-gas
holdup pipe in the reactor cell and this had been
vaporized and polymerized by the heat and radia-
tion from the fission products in the fuel off-gas.
Only a few grams of material had caused the
trouble and the installation of a more efficient,
larger filter (‘‘particle trap’’) downstream of the
hold pipe brought the problem under control.
Almost three months were spent in investi-
gating and remedying the off-gas problem before
operation was resumed. As shown in Fig. 5, in
the approach to full power, the reactor was
operated at several intermediate power levels to
observe dynamics, xenon behavior, and fuel chem-
istry. The only significant delay in this period
was to repair an electrical short in the fuel
sampler-enricher drive. In the final stages of the
power escalation it was discovered that the heat
removal capability of the air-cooled secondary
heat exchanger was substantially less than ex-
pected. As a result the maximum steady-state
power of the MSRE was restricted to 8 MW. (At
the time, heat balances indicated the maximum
power was ~7.2 MW. Later the coolant salt
specific heat was measured and found to be 11%
higher than the original value, which had been
obtained by extrapolation of data on other salts.)
Shortly after full power was reached, the
indicated air inleakage into the reactor cell rose
and, as shown in Fig. 5, the reactor was shut down
to investigate the difficulty. No excessive leakage
path to the atmosphere was found; the inleakage
was from pressurized thermocouple headers and

124

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

1965

MSRE activities, July 1964~-December 1965,

was corrected. Full-power operation was then
resumed. In the next five weeks the only inter-
ruption of more than a few hours was a four-day
delay when an electrical short in the power lead to
a component cooling blower caused a salt drain.,

High-power operation was abruptly halted in
July when the hub and blades in one of the main
blowers in the heat removal system broke up.
The cast aluminum hubs in the other blower and in
the spare unit were also found to have some
cracks, so new units had to be procured. Through
the vigorous efforts of the manufacturer, the hub
and blade castings were redesigned and a new unit
was built, tested, and installed at the MSRE within
11 weeks.

While the reactor was down, the array of
graphite and metal specimens was removed from
the core and a new array was installed. During
the flushing operations before the specimen re-
moval, the fuel-pump bowl was accidentally over-
filled and some flush salt froze in the attached gas
lines. Temporary heaters were installed remote-
ly to clear the lines.

Operation with one blower resumed in October,
but it was soon found that the fuel off- gas line had
not been completely cleared. A restriction near
the pump bowl became worse and the reactor was
shut down. When temporary heaters were put on
the off-gas line and pressure was applied, the plug
partially blew out, and the pressure drop returned
to near normal. While the reactor was down, the
second new main blower was installed and when
the reactor was started up again it was to go to
full power. Again it was found that the off-gas
line was not clear. Once more the reactor was
shut down to take positive steps to clean out the

FEBRUARY 1970
SALT IN
FUEL LOOP POWER

 

 

 

 

0O 2 4 6 8 10
POWER (MW)

FUEL XXX
FLUSH | |

NUCLEAR APPLICATIONS & TECHNOLOGY

 

DYNAMICS TESTS

INVESTIGATE
OFF=GAS PLUGGING

REPLACE VALVES
AND FILTERS

RAISE POWER
REPAIR SAMPLER

ATTAIN FULL POWER
CHECK CONTAINMENT

FULL- POWER RUN

MAIN BLOWER FAILURE

REPLACE MAIN BLOWER
MELT SALT FROM GAS LINES
REPLACE CORE SAMPLES
TEST CONTAINMENT

v

 

RUN WITH ONE BLOWER
) INSTALL SECOND BLOWER

ROD. OUT OFF-GAS LINE
CHECK CONTAINMENT

30-day RUN
AT FULL POWER

} REPLACE AIR LINE
DISCONNECTS

SUSTAINED OPERATION
AT HIGH POWER

REPLACE CORE SAMPLES
TEST CONTAINMENT

} REPAIR SAMPLER

Fig. 5.

VOL. 8

FEBRUARY 1970

Haubenreich and Engel EXPERIENCE WITH MSRE

SALT IN

FUEL LOOP POWER

XENON STRIPPING
EXPERIMENTS

 

MAINTENANCE

W } INSPECTION AND
REPLACE CORE SAMPLES

TEST AND MODIFY
FLUORINE DISPOSAL
SYSTEM

o

1968 —

| PROCESS FLUSH SALT

 

{
B }» PROCESS FUEL SALT
4} LOAD URANIUM-233
T REMOVE LOADING DEVICE
233y ZERO - POWER
PHYSICS EXPERIMENTS

INVESTIGATE FUEL
SALT BEHAVIOR

CLEAR OFF=~GAS LINES

 

. A
Sy

REPAIR SAMPLER AND
CONTROL ROD DRIVE

233\, DYNAMICS TESTS

INVESTIGATE GAS
IN FUEL LOOP

 

g gt

HIGH-POWER OPERATION
TO MEASURE 233U o /oq

 

 

 

 

REPLACE CORE SAMPLES
REPAIR ROD DRIVES
CLEAR OFF-GAS LINES
J
A
S
0
N
D
L L L1
O 2 4 6 8 1
FUEL RO POWER (MW)
FLUsH [

MSRE activities, 1966-1969.

125
Haubenreich and Engel = EXPERIENCE WITH MSRE
off-gas line and also to investigate an apparent
high inleakage of air.

When the off-gas line was opened, crusts of
frozen salt were found in the flanges where the
heaters had not been effective. These deposits
were rodded out. The cell inleakage proved to be
from leaking service-air-line disconnects. Since
these did not represent a leakage path to the
outside, provisions were simply made to measure
their input so the cell leak rate could be calculated.

The next run began in mid-December and
continued for 30 days at full power. A shutdown,
which had been scheduled for the 1000-h inspec-
tion of the new blower, was extended to permit
replacement of all the air-line disconnects in the
reactor cell and installation of an improved par-
ticle trap in the fuel off-gas line.

It turned out that the bothersome problems that
had interfered with sustained operation during the
autumn had been finally overcome. Nuclear oper-
ation resumed on January 28 and continued for 102
consecutive days until a shutdown that was sched-
uled to permit removal of specimens from the
core. As shown in Fig. 5, during this period the
reactor was at full power nearly the full time (93%
of the time from February 1 through May 8). The
longest interruption was four days at low power
while a zero-xenon reactivity measurement was
being obtained and a rough bearing on a main
blower was replaced. During this run a total of
761 g of **U was added through the sampler-
enricher, marking the first time fuel had been
added with the reactor at full power.

The reactor was down for 39 days while the
core specimens were replaced, annual tests of
containment and instrumentation were conducted,
and miscellaneous maintenance jobs were done.
Again full-power operation was resumed. More
“®U was added during the run (18 capsules con-
taining 1527 g of **U in seven days) in preparation
for six months of power operation with no further
uranium additions to obtain a measure of the
capture-to-fission ratio of *°U. However, after
42 days of full-power operation, the run was
cut short by a malfunction in the fuel sampler-
enricher.

The drive cable in the sampler had become
tangled and, in attempts to untangle it, the cable
was severed, dropping a sample capsule and the
cable latch into the pump bowl. In the next five
weeks the latch was retrieved and the drive unit
was replaced. The capsule was left in the pump
bowl.

Full-power operation was barely resumed when
one of the component cooling pumps went down
with a lubricating-oil leak. Rather than start what
we hoped to be a 6-month run without a standby,
component cooling pump, we shut down and fixed

126 NUCLEAR APPLICATIONS & TECHNOLOGY

it. A new start was made on September 20. This
time everything went well and it was 188 days
before the fuel was again drained.

The long run was devoted primarily to study of
fission-product behavior. The reactor was oper-
ated for about two of the six months at a reduced
power, 5.6 MW, to permit the fuel temperature to
be varied in a study of the effect of various
operating conditions on °°Xe stripping. The only
interruption of any significance was a week in
November when the power was lowered (sub-
critical for two days) while an electrical short in
the fuel sampler-enricher was being repaired.

The shutdown on March 26, 1968 was the end of
nuclear operation with **°U, For several months,
preparations had been underway to strip the
uranium from the fuel salt and replace it with 233U,
The uranium was to be removed on-site by the
fluoride volatility process. For the ?**U addition,
equipment had been set up in the nearby Thorium -
Uranium Recycle Facility (TURF) at ORNL*° to
produce half a cubic foot of UF,-LiF eutectic
containing 35 kg of ***U. (This uranium contained
~220 ppm ***U, which made it so radioactive as to
practically prohibit any other use.)

In the first month of the shutdown the core
specimens were replaced, and the fuel piping and
vessels were surveyed using remote gamma-ray
spectroscopy to determine the distribution of
fission products. At the same time necessary
maintenance was completed. This included repair
of two heaters from the primary heat exchanger,
rodding out the fuel off-gas line at the pump bowl
and fishing (unsuccessfully) for another sample
capsule that had been accidentally dropped into the
pump bowl. Attention then focused on the salt
processing system where the system for disposal
of excess fluorine was as yet untried.

In the preparations for uranium recovery,
difficulties were encountered in obtaining the
required efficiency in disposal of fluorine by
reaction with SO: gas, and eventually the process
was changed to reaction in a caustic solution.
Processing of the flush-salt charge started on
August 1; eleven days later the 6 kg of uranium in
the salt had been fluorinated out as UF¢ and
collected on sodium fluoride pellets, corrosion-
product fluorides had been reduced to filterable
metals, and the salt had been filtered on its return
to the reactor. Fluorination of the fuel salt to
recover the 218 kg of uranium present took only
47 h of fluorine sparging over a 6-day period.
Corrosion products were reduced and filtered in
another 10 days. Two cubic feet of the stripped
carrier salt, still loaded with fission products,
was left in the processing vessel for a future test
of a vacuum distillation process. The remainder
was returned to the reactor.

VOL. 8 FEBRUARY 1970
Ioading of ***U began immediately. The bulk of
the uranium required for criticality was loaded
into the carrier salt through equipment attached
temporarily to one of the drain tanks. Cans of
frozen eutectic, containing up to 7 kg ***U each,
were lowered into the hot tank where the salt
melted and poured into the carrier salt. Neutron
multiplication measurements were made with the
salt in the core after the addition of 21, 28, and
33 kg U. After the last of these, the addition
equipment was removed and the cell was closed
and leak-tested before the addition of the re-
maining 400 g required for criticality. This
amount was added in capsules through the sam-
pler-enricher and on October 2 the MSRE was
first critical with 23U fuel. Capsule additions
were continued and on October 8, U.S. Atomic
Energy Commission Chairman Seaborg took the
reactor power to 100 kW for the first time
after ceremonies marking the world’s first
power operation with **U fuel. Over the next
month another 1.7-kg U was added while the
control rods were calibrated and the temperature
and concentration coefficients of reactivity were
measured.

Early in the zero-power physics experiments
the amount of gas entrained in the circulating fuel
was observed to increase from <0.1 to ~0.5 vol%.
The run was therefore extended two weeks for
experiments aimed at determining the cause of the
increase. The initial increase was coincident with
the addition of beryllium reductant to the fuel salt
and further additions confirmed that beryllium had
an effect on bubble behavior. Magnets dipped into
the salt showed that small amounts of finely
divided, reduced iron were floating on the surface
in the pump bowl. This evidence was later fitted
with other observations to conclude that minor
changes in surface tension occurred when beryl-
lium was exposed in the salt containing some
unreduced corrosion products. These minor
changes affected the behavior of the bubbles
churned into the fuel in the pump tank by the xenon
stripper sprays just enough to affect the entrain-
ment of gas into the circulating stream.

In preparation for precision assays of the
uranium to measure the ***U capture-to-fission
ratio during subsequent power operation, the re-
actor was shut down to mix the uranium in the
loop and in the drain tanks. While it was down the
fuel off-gas line, which had become restricted near
the pump bowl, was rodded out. The reactor was
started up, but before power operation was begun,
it was shut down again to repair a gear in the
sampler drive mechanism. This work extended
into January 1969.

As shown in Fig. 5, the reactor power was
stepped up to 1, 5, and 8 MW over a two-week

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

FEBRUARY 1970

Haubenreich and Engel EXPERIENCE WITH MSRE
period in January, with dynamics tests and obser-
vations of reactivity and radiation heating at each
level. As expected, the system was quite stable
over the entire range of power, with dynamic
response characteristics very close to
predictions. .

During the approach to full power, there were
observed for the first time sporadic small in-
creases (~5 to 10%) in nuclear power for a few
seconds, occurring with a varying frequency
somewhere around 10/h. The perturbations in-
volved brief reactivity increases of 0.01 to 0.02%
5k/k, and temperature and pressure changes too
small to be of consequence. The characteristics
of the transients pointed to changes in gas volume
in the fuel loop and it appeared that they were
most likely caused by occasional release of some
gas that collected in the core. This hypothesis
was supported when, late in February, a variable-
frequency generator was used to operate the fuel
pump at reduced speed. The gas entrainment in
the circulating fuel decreased sharply (from 0.7 to
<0.1 vol %) and the perturbations ceased entirely.
Upon resumption of full-speed circulation the
perturbations appeared again but gradually be-
came smaller and less frequent, until by the end
of March they were indistinguishable from the
continuous noise in the neutron level (£ 2% of
power). The disappearance was tentatively attri-
buted to gradual, slight changes in salt surface
tension and circulating gas fraction.

In May, restrictions began to show up again in
the fuel off-gas system. Partial restrictions ap-
peared at several points including two places in
the main fuel off-gas line where they had occurred
previously as well as two other locations that had
not plugged before. Although the restrictions
were somewhat more severe than in the past, they
did not prevent operation of the reactor at high
power until the scheduled end of the run on June 1.
The power operation in this run burned up enough
2337 to provide a good measure of the ratio of
neutron captures to fissions in this fissile ma-
terial. (Samples taken for this purpose were
prepared for high-precision mass-spectrographic
analyses at the time of this writing.)

The primary purpose of the shutdown was
to permit removal and replacement of surveil-
lance specimens in the core and to investi-
gate the distribution of fission products in the
primary loop by gamma scanning. Preparations
were also made for more extensive gamma scan-
ning during and after the next period of power
operation. In addition, maintenance operations
were performed on the reactor control rods
and drives and to relieve the off-gas restrictions.
The return to power operation was scheduled for
August.

127
Haubenreich and Engel = EXPERIENCE WITH MSRE

SUMMARY OF EXPERIENCE

Operation of the MSRE has served to demon-
strate and emphasize the basic soundness of the
molten-salt reactor concept. It has also provided
some valuable new information. The following is
a discussion of the experience in each of several
important areas.

Fuel Chemistry

Fuel chemistry is an area of vital importance
to a fluid-fuel reactor comparable to fuel struc-
ture, cladding integrity, and coolant stability in a
solid-fuel reactor.

In assessing the safety of fluid-fuel reactors
perhaps the first question that comes to mind is
the possibility of uranium separating from the
fluid. This has occurred in aqueous homogeneous
reactors, where several mechanisms exist for
such separation and operating conditions may be
close to a region of chemical instability. Such is
not the case in molten-salt reactors, however.
The MSRE fuel is subject to only two mechanisms
for uranium separation: the contamination of the
salt with moisture or oxygen to the point that
uranium oxide precipitates, and the reduction of
s0 much UF, to UF; that the UF; begins to
disproportionate into UF, and insoluble uranium.,
Neither condition is approached during operation.*

Oxide formation in the MSRE salt is controlled
by providing a blanket gas (helium) that has had
oxygen and moisture reduced to <10 ppm by
passage through a 1200°F titanium sponge. When
the core access is opened for specimen removal,
moist air intrusion is minimized by first filling
the system with denser argon and working through
a standpipe filled with dry nitrogen. Then before
the fuel salt is put back in the core, the flush salt
is first circulated to react with any moisture that
may have entered. Evidence of the effectiveness
of these measures is the extremely low oxide
level in the fuel after four years in the reactor:
analyses consistently show only ~60 ppm oxide.
This small amount is in solution: the chemistry
of the fuel salt is such that the first oxide to form

is ZrO; and its solubility is ~700 ppm. The ZrF,
was put in the fuel as a cushion; without it, the
UO, that would have been formed would still have
been far below its solubility limit of 1000 ppm.
The success of the MSRE of keeping oxide
contamination down argues that ZrF, need not be
included in the fuel of future molten-salt reactors.

The second possibility, precipitation of re-
duced uranium, was the reason for specifying that
the uranium to be used in the original power
operation by only ~33% **U instead of highly
enriched. At that time we did not know precisely

128

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

what the average total valence of the products of
one fission would be in the reactor environment,
We expected it to be close to four, but if it had
been more it would, in effect, tie up more than the
four fluorine atoms made available by the de-
struction of the uranium atom. The ingrowth of
fission products would then gradually reduce some
of the UF, to UF,;. With more moles of UF, in the
fuel charge, more fissions could be sustained
before the UF,/UF, ratio reached the point of
disproportionation., Thus the composition of the
fuel was specified to contain 0.9% total uranium
even though <0.3 mole% of highly enriched ura-
nium would have been sufficient to make the MSRE
critical. Operation of the MSRE has shown that
in fact the average total valence of the products of
a fission is slightly less than four. This means
that the tendency is actually away from reduction
of UF4 to UFj;, thus eliminating the possibility of
uranium precipitation.

Although the chemical effect of the fission
process in the MSRE is in the direction of excess
fluorine, as long as there is any UF; present, the
result is simply the gradual conversion of UF, to
UF4. Therefore, a small fraction of the uranium
is maintained as UF; to guarantee a reducing,
noncorrosive environment. This is accomplished
simply by occasionally exposing a rod of Be metal
to the salt in the fuel-pump bowl. In a 10-h
exposure, ~10 g of Be forms BeF,, reducing two
moles of UF, to UF; and counteracting the effect
of the fissions over more than 10 000 MWh of
operation. After the fuel processing in September
1968, some unreduced corrosion-product fluorides
from the fluorination process were apparently left
in the salt, and the first three additions of
beryllium were consumed in reducing these fluo-
rides to the metals. Further additions then pro-
duced the desired UF,.

Corrosion products in the salt in the reactor
have never amounted to more than ~200 ppm and,
with the exception of the possible interference
with the UF,-UF; reduction just mentioned, they
have had no significant chemical effects.

Fission product behavior in the molten-salt
reactor environment was a question on which the
MSRE operation has shed new light.* The noble
gases, xenon and Krypton, are stripped into the
off-gas even more efficiently than had been ex-
pected. Only small fractions of their radioactive
isotopes decay in the salt or in the pores of the
graphite. As expected, most of the other fission
product species remain entirely within the fuel
salt. The only exception is the ‘‘noble-metal’’
group—Mo, Ru, Te; and Nb, which do not form
stable fluorides in the normal MSRE environment,
Analyses of salt samples typically show only one
percent or less of the noble metals circulating

FEBRUARY 1970
with the salt. (An exception was during the **°U
start-up when niobium showed up in the salt until
beryllium additions increased the reducing power
of the salt.) Specimens from the core array
showed that around half of the noble-metal fission
products were plated out on graphite and metal
surfaces exposed to the salt. The remaining
substantial fraction of the noble metals goes into
the cover gas above the fuel salt as a smoke or
aerosol of submicron particles. The deposition
and ‘‘smoking’’ of the noble metals create no
problems in the MSRE, but forewarn of afterheat
problems to be dealt with in the breeder design.

Also of interest for future reactors is the
behavior of plutonium in molten salt. The stable
form is PuFs, which is sufficiently soluble in the
salt to be of interest as a potential fuel for
molten-salt reactors. The MSRE operation with
partially enriched uranium produced ~600 g of
plutonium and analyses of samples show that all
has stayed with the salt.

Materials Compatibility

Analyses of several hundred samples of fuel
salt taken over a period of three years and
examination of specimens exposed in the core for
many thousands of hours have demonstrated the
compatibility of the salt, the moderator, and the
container material. |

Chromium in the fuel salt is the best indicator
of corrosion of the Hastelloy-N, since corrosion
selectively attacks the chromium in the alloy and
the product, CrF,, is quite soluble in the salt.
The MSRE fuel is sampled and analyzed for
chromium at least once a week. Results showed
an increase in the three years between May 1965
and March 1968 from 38 to 85 ppm. This corre-
sponds to the amount of chromium in a 0.2-mil
layer of Hastelloy-N over the entire metal surface
in the circulating system. However, the data
suggest that much of the chromium appeared in
in the salt while it was in the drain tank between
runs so that the chromium was leached from even
<0.2 mil over the circulating loop surfaces in
three years. This extremely low rate in the loop
is less than was predicted from thermodynamic
data and diffusion in the metal. One hypothesis is
that the corrosion is so low because the surfaces
are protected by the film of deposited nooble-metal
fission products, which averaged ~10 A in thick-
ness by March 1968.

The chromium indication of very low corrosion
in the fuel loop during the *®U operation was
substantiated by the condition of Hastelloy-N sur-
veillance specimens exposed in the core. There
was no sign of localized attack, and metallo-
graphic examination showed no appreciable corro-

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

Haubenreich and Engel EXPERIENCE WITH MSRE
sion. Carburization of metal specimens in contact
with graphite extended to a depth of only 1 mil.

During the first 3 months of fuel circulation in
the 2%*U start-up, the chromium concentration
came up from 35 to 70 ppm and then virtually
leveled out. This behavior was attributed to the
presence in the salt of some FekF, from the
fluorination process that reacted with the chro-
mium in the walls until it was reduced by addi-
tions of beryllium metal.

Graphite specimens exposed in the fuel salt for
as long as 22 500 h (2.5 years) showed no attack
by the salt. There was no change in the surface
finish and no cracks other than those present
before exposure. Only extremely small quantities
of salt were found to have penetrated the graphite
either through pores or cracks. (Assuming the
specimens are typical, there is <2 g of uranium in
the 3700 kg of graphite in the entire core.)

The physical properties of the Hastelloy-N
were affected by irradiation in the MSRE as
expected on the basis of prior irradiation experi-
ments.!! There was little change in ultimate
strength, yield strength, and secondary creep rate.
Rupture ductility and creep rupture life in speci-
mens of the heats of Hastelloy-N used in the
fabrication of the MSRE were reduced as had been
anticipated in the design. Specimens of improved
Hastelloy-N (containing small amounts of tita-
nium) irradiated in the MSRE showed very good
ductility and life at 1200°F, even surpassing
unirradiated standard Hastelloy-N.

Chemical Processing

Fluorination to recover the original charge of
uranium was accomplished in the simple equip-
ment depicted in Fig. 6. Mixtures of fluorine and
helium were bubbled at rates up to 33 std liters/
min of fluorine through a 1-in. dip tube immersed
64 in. in the tank of salt. At the beginning of the
processing there was a high utilization of fluorine
as the UF, was rasied to UFs. Then volatile UFs
began to be formed and the fluorine utilization
decreased to ~30%. The effluent gas stream
passed through a bed of sodium fluoride pellets at
750°F to remove volatile impurities and then into
a series of cannisters filled with NaF pellets at
200 to 300°F where the UFg was absorbed.

The uranium concentration in the flush salt was
reduced to 7 ppm in 7 h of fluorination; in the fuel
salt processing, 47 h of fluorination removed 218
kg of uranium from the 4730 kg of salt, leaving a
concentration of 26 ppm U. The UFg has not yet
been desorbed from the NaF for an accurate
material balance, but the weight gain of the
absorbers checks the expected UFeg recovery
within the accuracy of the measurement.

FEBRUARY 1970 129
Haubenreich and Engel = EXPERIENCE WITH MSRE

GAS ADDITION

STATION SAMPLER

  
   
 

=W

 

 

FUEL STORAGE TANK

FLUORINE REACTS WITH
DISSOLVED UF4 TO FORM
VOLATILE UFg

Fig. 6.

Decontamination of uranium from the fission
products was very effective; the processing tank
read over 2000 R/h, while cannisters containing
12 kg U read <0.002 R/h, The measured decon-
tamination factors for gross beta and gross
gamma activities were 1.2 x 10° and 8.6 x 10%,
respectively. This permitted direct handling of
the absorbed cannisters containing the recovered
uranium,

Corrosion of the tank during fluorination was
less than had been observed in fluoride volatility
processing in other equipment. This was probably
because the processing at the MSRE was done at a
lower temperature (850°F, just above the salt
liquidus) and also because the MSRE tank is
relatively large so that fluorine concentrations at
the wall tended to be lower. The depth of corro-
sion for the whole campaign, averaged over the
surface exposed to the salt, was 11 mil. At this
rate the tank would last for processing many
batches, although the dip tube would likely have to
be replaced after each few batches.

At the end of fluorination the fuel salt contained
850 ppm Ni, 410 ppm Fe, and 435 ppm Cr as
fluorides. The fluorides were reduced by treat-
ment with hydrogen and finely divided zirconium,

130 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

ALT IS SAMPLED
O MEASURE UFy4

CONTAMINANT
FLUORIDES
ABSORB ON

NaF PELLETS
AT 750°F

UFg ABSORBERS

UFg IS ABSORBED ON
NaF PELLETS AT 250°F

   

 
 

FILTER '

 
   
 

 

 

    
 

TRAPS

TRACES OF

FLUORINE

AND 10DINE
ARE REMOVED

CAUSTIC SCRUBBER

FLUORINE REACTS WITH
KOH. IODINE IS ABSORBED

MSRE salt fluorination process.

The agglomerated reduced metals were then re-
moved by a 9-ft® fibrous-metal filter as the salt
was returned to the reactor. Samples from the
reactor showed only 60 ppm Ni, 110 ppm Fe, and
35 ppm Cr, indicating the effectiveness of this
part of the processing. As discussed before,
however, it appears that at least some of the iron
that got through was in the form of the fluoride.

Reactivity

The MSRE has gone through two series of
nuclear start-up experiments: once'® with 2®U
and again with ***U. In each case the critical
concentration of fissile material was carefully
determined and the control rod calibrations and
reactivity coefficients necessary for analysis of
the nuclear operation were measured. Table II
summarizes the most important results. Com-
parison of predicted and observed values shows
that the available data and procedures were quite
adequate for calculating the nuclear character-
istics of the MSRE at start-up.

During routine reactor operation, the system
reactivity is monitored by a calculation executed
every five minutes by an on-line digital computer

FEBRUARY 1970
Haubenreich and Engel = EXPERIENCE WITH MSRE

 

 

 

TABLE II
Summary of MSRE Nuclear Parameters with 235 and ?**U Fuels
%0 Fuel 283 Fuel
Parameter Units Calculated Measured Calculated | Measured
Initial critical concentration in salt g U/liter 33.062 32.85 +0.25% 15.30b 15.15 + 0.1P |
Reactivity loss due to circulation
of delayed-neutron precursors % ok/k 0.222 0.212 + 0.004 0.093 c
Control-rod worth at initial critical loadingd % 6k/k
1 Rod 2.11 2.26 2.75 2.58
3 Rods, banked 5.46 5.59 7.01 6.9
Temperature coefficient of reactivity ok/k 5
. . (x107)
at operating loading °F
Total -8.1 -7.3 £0.2 -8.8 -8.5
Fuel -4.1 -4.9 £+ 2.3 -5.7 e
Concentration coefficient of reactivity Zogi/lg 0.234 0.223 0.389 0.369
0

 

 

 

 

 

 

 

a235U only

bUranium of the isotopic composition of the material added during the critical experiment (91% 233U).

“Measurement obscured by effect of circulating voids.
dNormal full travel of rod(s).
€Not separately evaluated.

at the reactor site.'* The computer is program-
med to take signals of control rod position,
neutron flux, and fuel temperature and compute a
reactivity balance based on present conditions and
the power history. The basic equation solved by
the computer is

_KRES = KTEMP + KPOW + KXE + KSAM
+ KFP + KU235 + KROD,

where

KRES = the deviation of the observed re-
activity from the expected behavior

KTEMP = the reactivity variation associated
with variations of the system tem-
perature from a reference value
(1200 °F)

KPOW = the reactivity difference caused by
the calculated temperature distri-

bution within the core at power rel-
ative to the isothermal core

KXE = the reactivity effect of **Xe

KSAM = the reactivity effect of ***Sm and**'Sm

KFP = the reactivity effect due to buildup of
nonsaturating fission products and
other long-term isotopic changes
(e.g., Pu buildup, U isotopic changes,

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

FEBRUARY 1970

i burnout in fuel, '°B burnout in
graphite)

- KU235 = the reactivity effect of changes in the
concentration of 2*U and ?**U

KROD = the reactivity effect of the control
rods relative to a baseline condition.

If the deviation, KRES, exceeds preset limits
(usually + 0.25% 6%/k)the computer automatically
annunciates the fact. In any event the complete
balance is printed out hourly for the guidance of
the reactor operators. The results are used to
watch for unexpected short-term changes in re-
activity and also to study the long-term behavior
of the reactor. The precision of the results has
been excellent, with the scatter being generally
<0.02% 6k/k. No significant deviations from the
expected short-term behavior have been observed.

A major term in the on-line reactivity balance
at power is the description of '*Xe poisoning.
Xenon and other noble gases are only slightly
soluble in the molten salt and therefore tend to
transfer into any gas phase in contact with the
fuel. Contact with the helium cover gas is pro-
vided by the 50-gal/min spray in the fuel-pump
bowl, which not only produces salt spray in the
gas space but also carries bubbles of helium into
the salt, some of which enter the circulating loop.
Although most of the '*Xe is swept out into the
off-gas system, some diffuses into the pores of the

131
Haubenreich and Engel = EXPERIENCE WITH MSRE
core graphite and some remains in the salt until it
decays or captures a neutron. The various com-
peting mechanisms operate to give a Xenon poi-
soning in the MSRE around 0.3% 6%/k, only + to
s of the value that would exist if all the xenon
remained in the fuel salt. The reactivity measure
of xenon poisoning has been confirmed by deter-
minations of the '**Xe/'**Xe ratio in off-gas sam-
?Slses which show the effect of neutron captures in

Xe.

The steady-state xenon poisoning in the MSRE
varies somewhat with system temperature and
pressure and, to some extent, with the level of
salt in the fuel-pump tank. These variations are
accompanied by small changes in the volume
fraction of circulating helium bubbles in the fuel
salt, and it appears that they are caused by
changes in gas-stripping efficiency in the pump
tank., The correlation between stripping efficiency
and the various operating parameters has not been
fully described. Instead, the on-line calculation of
Xenon poisoning uses average values for stripping
efficiencies and transfer coefficients obtained em-
pirically from analysis of many xenon poisoning
transients.

Reactivity balance calculations are done only
when the reactor is critical with the fuel circu-
lating, so the reactivity effects of circulation are
part of the baseline condition. In the operation
with ®*U, circulation reduced reactivity by
0.21% 6k/k due to the loss of delayed neutrons and
only 0.02% 6k%/k or less due to the effect of gas
bubbles entrained in the fuel. Since the salt
processing, however, gas bubbles are carried
deeper into the salt in the pump bowl by the xenon
stripper jets and more gas gets into the loop. As
a result, the void effect attending circulation in
the recent operation has amounted to around 0.25%
6k/k. An exception has been during operation at
reduced pump speeds when the circulating bubbles
are eliminated by a 15% reduction in speed. The
reactivity decrease due to loss of delayed neu-
tron amounts to ~0.09% 6 2/k with the ?3%U fuel.

The long-term reactivity behavior is most
accurately defined by measurements made in the
absence of '*Xe poisoning. During the *U op-
eration, such measurements showed the reactivity
decreasing very slightly more than predicted.
(The deviation was -0.05% 6%/ at the end of 9006
equivalent full-power hours.) After the end of the
*®U operation, a careful reexamination of the
calculation revealed small errors in some of the
long-term effects. At about the same time a more
reliable value for the heat capacity of the coolant
salt was obtained which shifted the heat-balance
power of the reactor from 7.2 to 8 MW, thus
increasing the calculated burnup and fission-
product inventories. When these errors were

132 NUCLEAR APPLICATIONS & TECHNOLOGY

corrected the observed-calculated reactivity de-
viation showed a gradual increase to + 0,2% 6k/k
in 9006 equivalent full-power hours of operation
over a period of more than 2 years. Although it
had been anticipated that distortion of the core
graphite by neutron irradiation might have a
positive reactivity effect, no attempt had been
made to predict the effect quantitatively because
of the limited amount of data originally available
on MSRE graphite. After the **U shutdown, with
the benefit of more data on graphite distortion, we
evaulated this effect and found that at most it
could account for only a small part of the calcu-
lated reactivity drift. The drift must, therefore,
be assigned to inaccuracies in calculating other
long-term effects on reactivity.

In more than 2000 equivalent full-power hours
of operation with ?**U, there has been no distin-
guishable long-term trend in the deviation between
observed and predicted reactivity.

Reactor Dynamics

The MSRE is stable and self-regulating with
regard to changes in heat load, with a response
that becomes quicker and more strongly damped
as the power level is increased. Responsible in
large part for this behavior are the strong nega-
tive temperature coefficients of reactivity associ-
ated with both the fuel salt and the graphite
moderator. When the reactor is operated at low
power, because of the very large temperature
difference between the salt and air, only a very
small area of the secondary heat exchanger need
be exposed. This introduces a long time constant
in the system thermal equations and produces a
rather sluggish system at low power.

Loss of delayed neutrons by precursor decay
outside the core significantly reduces the effective
delayed-neutron fraction in the MSRE. In fact,
with the salt circulating, the effective fractions
are 0.0045 and 0.0017 for the **U and 2%°U fuels.
Despite these low fractions, the system response
to perturbations is quite acceptable with either
fuel.

The dynamic behavior of the MSRE was ex-
tensively examined, by theoretical techniques®®
before the reactor was operated and by experi-
ments'® during the operation. Calculations had
indicated that the reactor would be inherently
stable at all power levels and that the degree of
stability would increase with increasing power,
and experimental measurements of the reactor
dynamic response agreed very closely with the
predictions. In addition, measurements made
throughout the operation with ?*U fuel showed that
there was no change in dynamic behavior with
time.

VOL. 8 FEBRUARY 1970
Similar theoretical and experimental evalu-
ations were made of the dynamic behavior with
2331y fuel. The calculations indicated that, despite
the lower delayed-neutron fraction, the reactor
stability would be greater with ***U—-due primarily
to the larger negative temperature coefficient of
reactivity of the fuel salt. Experimental mea-
surements of system transfer functions and tran-

sient response are in good agreement with

predictions.

Because of the good self-regulating character-
istics of the MSRE, the system is quite simple to
control. In more than 15 000 h of critical op-
eration, not once have the nuclear power, period,
or fuel temperature gone out of limits so as to
cause a control-rod scram.

Equipment Performance

Some of the MSRE equipment items have un-
usual features but none differ radically from
components tested and proved before the MSRE.
In addition to conservative design, great care was
exercised in quality assurance for every part of
the primary systems. These factors account in
large measure for the high degree of reliability
attained by the MSRE.

Salt Pumps. The fuel and coolant salts are circu-
lated by centrifugal pumps of conventional hydrau-
lic design. The vertical shafts have oil-lubricated
bearings above the salt surface. Oil leakage past
the lower seal is drained out of the pump housing
and a split purge of helium into the shaft annulus
prevents oil vapors from going down into the pump
bowl or gaseous fission products from coming up
the shaft. The two original pumps are still per-
forming satisfactorily. The fuel pump has circu-
lated salt for more than 19 400 h; the coolant
pump, more than 23 500 h.

The only problem with the salt pumps has been
with oil leakage into the pump bowl. There is
evidence that in the fuel pump a small amount of
oil (on the order of 1 cm?®/day) leaks from the seal
drainage passage, past a gasketed joint,and into
the pump bowl, The oil does not react with the
salt or affect it in any way, but the oil is ther-
mally decomposed and its products are drawn into
the off-gas line where they have caused problems
as discussed below. Spare pump rotary elements
have a seal weld that prevents oil leakage, but the
off- gas problems have been adequately handled
otherwise, obviating the need for pump replace-
ment. -

Off-Gas System. The primary function of the fuel
off-gas system is to discharge cover gas from
which radioactive materials have been filtered or

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

Haubenreich and Engel = EXPERIENCE WITH MSRE
allowed to decay. Tests have shown that the times
required for krypton and xenon to work their way
through the 240-ft-long charcoal beds are at least
7 and 90 days, respectively. As a result, there
has been no significant activity in the effluent
other than 10-year 3°Kr.

The problem with the oil decomposition prod-
ucts first appeared when the power was raised to
1 MW. A light fraction that had condensed in the
holdup pipe was vaporized by the heat from the
fresh fission-product gases and the intense beta
radiation transformed it into a cross-linked poly-
mer that plugged a small sintered-metal filter and
a throttling valve and partially plugged the steel
wool in the entrances of the charcoal beds. A
large filter, consisting of coarse stainless-steel
mesh, sintered fibrous-metal filter media, and
Fiberfrax, was designed and proved quite effec-
tive. The first unit developed excessive pressure
drop when the trapped fission products raised
internal parts of the filter above 1200°F, causing
the inlet pipe to expand against the filter medium.
No more troubles of this sort were experienced
after a revised unit was installed in January 1967,

Salt mist is produced in the pump bowl by the
xenon stripper spray and some drifts into the
off-gas line where it freezes into tiny beads (1to
5u diam). The rate is very low (probably a few
grams per month) but the frozen mist does

gradually build up in the off-gas line at the first

cool point just outside the heated enclosure around
the pump bowl. Accumulations there were rodded
out in 1966, in April 1968, and in December 1968.
Intermittent heating at the point of accumulation
proved effective in a similar situation in a salt-
pump test loop and a small heater unit that can be
installed remotely in the MSRE has been de-
signed and tested.

Heat Exchangers. MSRE operation has shown that
conventional design calculations adequately predict

heat transfer in molten-salt heat exchangers and

there is no change in heat transfer over an ex-
tended period of operation. In the primary heat
exchanger, which is of conventional shell-and-tube
design, the predicted overall heat transfer coeffi-
cient is 600 Btu/h-ft?-°F and that observed is
656 + 15 Btu/h-ft>-°F. Values measured over
more than 15 000 h of salt circulation since the
beginning of power operation show no detectable
change in the coefficient. In the air-cooled secon-
dary heat exchanger the overall coefficient, which
is very close to the air-side coefficient, proved to
be lower than the design value by ~27%. Part of
the reason for the disparity was traced toan error
in the choice of air properties used in the design
computation. Even after correction of this error,
the design coefficient of 51.5 Btu/h-ft>-°F is still

FEBRUARY 1970 133
Haubenreich and Engel = EXPERIENCE WITH MSRE

higher than the observed value of 42.8 Btu/h-ft2-
°F. This discrepancy is most likely attributable
to inapplicability of the basic heat transfer corre-
lation used in the design to the specific geometry
of the air flow through the MSRE heat exchanger.

The tubes in the primary heat exchanger are
welded and back-brazed into the tube sheet. In
the secondary heat exchanger, the tubes are
welded to stubs forged in the headers. There has
never been any leakage in either exchanger.

The doors on the secondary-heat-exchanger
enclosure warped during prenuclear testing at
1200°F so that they could not be opened. Revised
doors with the structural members protected from
high temperatures did not warp. The gasketed
closure was not perfect at first and air inleakage
caused salt to freeze in several of the 120 tubes
on two occasions when the coolant-salt circulation
stopped and the doors were shut. The tubes were
thawed without damage, using the heaters built
into the enclosure (the coolant-salt density
changes very little on freezing and thawing). A
segmented metal strip pressing against an asbes-
tos-filled braided Inconel gasket gave a tight door
seal and no more freezing has occurred.

Salt Samplers. The samplers on the fuel pump,
the coolant pump, and the processing tank are all
similar. A motor-driven windlass in a shielded
containment enclosure is used to lower and re-
trieve capsules latched to the end of a drive cable.
Isolation valves close off the sampler tube before
a sample is withdrawn from the containment en-
closure. The samplers have operated very well
except for two occasions when a capsule being
lowered in the fuel sampler hung up, causing the
drive cable to tangle. Both times the capsule was
accidentally dropped down into a cage in the pump
bowl. Efforts at retrieval were unsuccessful. Two
other brief delays in operation were occasioned by
electrical shorts in the wiring penetrating the
containment enclosure to the drive motor. On
another occasion, loose setscrews in a drive gear
rendered the sampler inoperative until repairs
were made. Altogether the three salt samplers
have been used for over 580 sampling operations
and the addition of 141 uranium enriching cap-
sules. Never has there been any significant
release of activity into the operating area.

Other Equipment. The unique flexible control rods
and drives developed for the MSRE proved satis-
factory. In four years after the beginning of nu-
clear operation, no control rod ever failed to
scram when requested. One rod hung one time on
June 1, 1969 and the trouble was traced to galling
on an air tube which slides inside the hollow rod.

134

NUCLEAR APPLICATIONS & TECHNOLOGY

The ‘“freeze valves,’’ which take the place of me-
chanical valves in the salt piping, operated as
intended after the cooling-air controls were prop-
erly adjusted during the prenuclear start-up of the
reactor. The valves in the transfer lines are slow
to use, however, because a gas-tight seal requires
molten salt on the high-pressure side of the
frozen plug and sometimes this takes several
hours to set up. The fuel and coolant drain valves
operate reliably, with thaw times between 10 and
15 min. Removable heaters on the salt piping and
vessels in the reactor cell have had a low
incidence of failure: only 4 units of ~92 have ever
been out of service. Although not a part of the
primary systems, the positive-displacement blow-
ers used to circulate cell atmosphere for cooling
the control rods and freeze valves are essential to
reactor operation. Failure due to belt slippage,
lubricating oil leakage, and a short in the power
wiring have shut the reactor down at various
times. These problems have been corrected and
the blowers are now considered to be quite
reliable.

Containment

The MSRE design aimed at zero leakage from
the system of piping and vessels that is the
primary containment for the fission products. In
addition, a secondary containment system was
provided to limit the release of fission products to
the environs in the event of a failure in the
primary containment. Stringent leakage criteria
were set for the secondary containment by the
assumption that the fuel salt system and the
water-cooled thermal shield could rupture simul-
taneously, mixing all the salt with the amount of
water that would give the maximum steam pres-
sure in the cell. This was calculated to be 39 psig
and the reactor cell and drain-tank cell are
required to leak at a rate <1% per day at this
pressure.

The MSRE has met the criterion for primary
containment. By the use of welded construction
with a minimum of gasketed joints, and those
pressure-buffered, zero leakage has been attained
during all periods of operation. Annual tests and
routine measurements of air leakage into the
reactor cell during normal operation at -2 psig
indicate that the reactor has always operated
within a satisfactory secondary containment. On
two occasions in 1966, the reactor was shut down
when a high air leakage rate into the reactor cell
was measured. The sources proved to be leaks in
pressurized thermocouple penetrations and valve
operator air line disconnects. In neither case did
the leaks constitute an open path for outleakage
had the cell been pressurized.

VOL. 8 FEBRUARY 1970
Maintenance of Radioactive Systems

For the molten-salt reactor to be accepted as a
practical possibility for commercial power, it
must be maintainable at a reasonable cost. De-
velopment and demonstration of safe and efficient
maintenance is therefore a prime objective of the
MSRE. The reactor equipment was designed and
laid out with an eye to maintenance, and philos-
ophy, methods, and tools were developed in ad-
vance for all conceivable jobs of radioactive
maintenance. Many of the most important jobs
(fuel-pump rotary element removal, for example)
were practiced before the system became highly
radioactive.

Although there has been little trouble with the
major components of the reactor, enough jobs
have involved working in the reactor cell to
thoroughly test the maintenance scheme. Four
times the specimen array in the core has been
replaced. The fuel off-gas line has been rodded out
and a removable section replaced. Off-gas system
valves and filters have been replaced. Eighteen
disconnects in service air lines in the reactor cell
were replaced because of air leakage. Two large
heater units surrounding the primary heat ex-
changer were removed, repaired, and replaced.
All of these jobs were done with long-handled
tools working through penetrations in the mainte-
nance shield. Such work generally takes several
times as long as similar tasks done directly, but
with experience and with the help of detailed
procedures, it has proved possible to estimate
time requirements for remote maintenance as
for conventional jobs.

Through careful planning and use of temporary
containment enclosures, radioactive contamination
has been successfully controlled and cleanup dur-
ing and after radioactive jobs has required only
vacuuming and mopping or wiping. Exposure of
personnel to radiation has been held well below
permissible limits: the maximum exposure of any
individual in any quarter has been <0.5 rem.

CONCLUSIONS

The MSRE operated long enough with **U to be
a good test of the practicality of molten-salt
reactors. Some of the highlight dates and statis-
tics from this period are listed in Table III. In
retrospect, the 30-day run that started in De-
cember 1966 was the real beginning of the sus-
tained power operation phase of the test program.
Up to that time sundry problems related to start-up
had prevented any long run, After that point, over
the next 15 months until the shutdown, the reactor
was critical 80% of the time. Counting four weeks

NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8

Haubenreich and Engel = EXPERIENCE WITH MSRE
that were spent on core specimen replacement,
the reactor was available for planned eXperi-
mental work for 87% of that period. This is
obviously a good record, but when measured
against the yardstick of other reactors in a
comparable stage of development, it is seen to be
indeed remarkable.

From the months of operation and experiments,
a very favorable picture emerged. In properly
designed equipment, handling the high-melting salt
proved to be easy. Maintenance of the radioactive
systems was not easy, but there were no unfore-
seen difficulties, and control of contamination
was, if anything, less difficult than expected. Fuel
chemistry and materials compatibility lived up to
expectations, showing no adverse effects due to
the reactor environment. The noble gases, Xenon
and krypton, were stripped efficiently. It was
found that the noble-metal fission products, whose
behavior had hitherto been uncertain, partially
plated out and partially came off as a smoke, thus
providing important information for the design of
future reactors. Although the operation of the
MSRE showed that the design of some equipment
and systems could be improved, key components
performed well.

The on-site removal of the original uranium
from the fuel and the loading of the **°U into the
stripped carrier salt extended the usefulness of
the MSRE and the simplicity and efficiency of
these steps illustrated one of the virtues of the
fluid-fuel, molten-salt system.

TABLE III
Important Dates and Statistics in Operation of the MSRE

 

Dates

 

Salt first loaded into tanks
Salt first circulated through core

October 24, 1964
January 12, 1965

First criticality June 1, 1965
First operation in megawatt range January 24, 1966
Reach full power May 23, 1966

Complete 30-day run
Complete 3-month run
Complete 6-month run
End nuclear operation operation with **U
Strip U from fuel carrier salt
First critical with **°U
First operation at significant power
with U
Reach full power with **U

January 14, 1967
April 28, 1967
March 20, 1968
March 26, 1968
August 23-29, 1968
October 2, 1968

October 8, 1968
January 28, 1969

 

Statistics

 

 

 

 

 

 

 

235U a 233U b Totalb
Critical hours 11 515 3 910 15 424
Integrated power, MW(th) h T2 441 20 363 92 805
Equivalent full-power hours 9 006 2 549 11 555
Fuel pump circulating salt, h 15 042 4 363 19 405
Coolant pump circulating salt, h | 16 906 6 660 23 566

 

aGalt circulation times include prenuclear testing.
bThrough June 1, 1969.

FEBRUARY 1970 135
Haubenreich and Engel = EXPERIENCE WITH MSRE

The nuclear start-up experiments and operation
at power confirmed the adequacy of the data and
procedures used to predict the nuclear behavior.
The system was quite stable and easy to control
even during ***U operation with a delayed neutron
fraction lower than in any other reactor. Finally ,
burnup over an extended period at high power
should yield very accurate information on 23*U
cross-section ratios in a neutron energy spectrum
typical of molten-salt reactors.

A net result of the MSRE operation is enhanced
confidence in the practicality and performance of
future molten-salt reactors.

ACKNOWLEDGMENTS

This research was sponsored by the U.S. Atomic
Energy Agency Commission under contract with the
Union Carbide Corporation,

REFERENCES

1. M. W. ROSENTHAL, P. R. KASTEN, and R. B.
BRIGGS, ‘‘Molten-Salt Reactors—History, Status, and
Potential,”” Nucl. Appl. Tech., 8, 107 (1970).

2. R. C. ROBERTSON, ‘“MSRE Design and Operations
Report, Part I—Description of Reactor Design,”’ ORNL-
TM-278, Oak Ridge National Laboratory (January 1965),

3. W. H. COOK, ‘‘“MSRE Program Semiannual Pro-
gress Report, July 31, 1964, ORNL-3708, pp. 373-389,
Oak Ridge National Laboratory,

4. W. R. GRIMES, ‘“Molten-Salt Reactor Chemistry,”
Nucl. Appl. Tech., 8, 137 (1970),

9. A. TABOADA, ‘““MSRE Program Semiannual Pro-
gress Report, July 31, 1964, ORNL-3708, pp. 330-372,
Oak Ridge National Laboratory.

136 NUCLEAR APPLICATIONS & TECHNOLOGY

6. R. B. BRIGGS, ‘‘Effects of Irradiation on the Ser-
vice Life of the MSRE,” Trans. Am. Nucl. Soc., 10, 166
(1965).

7. T. L. HUDSON, ‘“‘Design and Operation of the 1200°F
Heating System for the MSRE,”’ Trans. Am. Nucl. Soc.,
8, 147 (1965).

8. R. B. LINDAUER, ‘““MSRE Design and Operations
Report, Part VII—Fuel Handling and Processing Plant,’’
ORNL-TM-907 Rev., Oak Ridge National Laboratory
(December 1967),

9. R. B. LINDAUER and C. K. McGLOTHLAN, ‘“De-
sign, Construction, Development, and Testing of a
Larger Molten Salt Filter,”” ORNL-TM-2478, Oak Ridge
National Laboratory (February 1969).

10. J. M. CHANDLER and S. E. BOLT, ‘““Preparation
of Enriching Salt "LiF-?3® UF4 for Refueling the Molten
Salt Reactor,”” ORNL-4371, Oak Ridge National Labora-
tory (March 1969).

11. H. E. McCOY, “MSRE Program Semiannual Pro-
gress Report, August 31, 1968,° ORNL-4344, pp. 216-
223, Oak Ridge National Laboratory.

12. R. B. LINDAUER, ‘‘Processing of the MSRE Flush
and Fuel Salts,”” ORNL-TM-2578, Oak Ridge National
Laboratory (July 1969).

13. B. E. PRINCE, S. J. BALL, J. R. ENGEL, P. N,
HAUBENREICH, and T. W. KERLIN, °‘‘Zero-Power
Physics Experiments on the Molten-Salt Reactor Ex-
periment,”” ORNL-4233, Oak Ridge National Laboratory
(February 1968),

14. J. R. ENGEL and B. E. PRINCE, ‘“The Reactivity
Balance in the MSRE,” ORNL-TM-1796, QOak Ridge
National Laboratory (March 10, 1967).

15. S. J. BALL and T. W, KERLIN, ‘‘Stability Analysis
on the Molten-Salt Reactor Experiment,”” ORNL-TM-
1070, Oak Ridge National Laboratory (December 1965).

16. T. W. KERLIN and S. J. BALL, ‘‘Experimental
Dynamic Analysis of the Molten-Salt Reactor Experi-
ment,’”” ORNL-TM-1647, Oak Ridge National Laboratory
(October 1966),

VOL. 8 FEBRUARY 1970