ORNL-3708.txt

AR

Al ORNL-3708

UC-80 — Reactor Technology
TID-4500 (35th ed.)

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT

FOR PERIOD ENDING JULY 31, 1964

OAK RIDGE NATIONAL LABORATORY

operated by

UNION CARBIDE CORPORATION
for the
U.S. ATOMIC ENERGY COMMISSION

£ SEARCH LIBRAT
DOCUMEN OLLECTION
LIBRARY LOAN €O

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT

For Period Ending July 31, 1964

R. B. Briggs, Program Director

NOVEMBER 1964

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

ORNL-3708

il QNI

.

o
1ii

CONTENTS

INTRODUCTION.............‘....l...........O.....I.........‘.......

THE PLACE OF MOLTEN-SALT REACTORS IN THE AEC PROGRAM —

El E. Sinclair............'...l........................l....l..

- MOLTEN-SALT POWER REACTORS AND THE ROLE OF THE MSRE IN THEIR
DEVELOPM:NT—R. B. Briggs.............l.....................ll

MSRE DESIGN -AND CONSTRUCTION - W. B. MCDOnaldl LB BN BN BE BL BN BN BN BN BN BN BN B BE BN B B BN RN

NUCLEAR CHARACTERISTICS OF THE MSEE — R. Js Engelesecececececccens

INSTRUMENTATION AND CONTROIL OF THE MSRE - J. R. TallacksONeeseesoss.

PUMP DEVELOPMENT — A. G. Grindell and P.‘G. SMithesecseeeccsscscass

COMPONENT DEVELOPMENT IN SUPPORT OF THE MSRE —
D‘Llr]lap Scott, Jr..............................................'.

REMOTE MAINTENANCE OF THE MSRE — Robert BIUmberges.esessseesseces.
FUEL, PROCESSING FACILITY — R. B. LinGBUETeeeeeeeesereecnses ceeeees
PLANS FOR OPERATION OF THE MSRE — P. N. Haubenreicheesesseseesssss
CHEMICAL BASIS FOR MOLTEN-SALT REACTORS — W. R. GrimeSeeeececeeocss

EFFECTS OF RADIATION ON THE COMPATIBILITY OF MSRE MATERTALS -
F. F. Blankenship.............. ........ 6 8 & 5 8 59O OO OO0 OSSO S PS e S PSR

PREPARATTON OF MSRE FUEL, COOLANT, AND FLUSH SALTS —

J. H. Shaffer..................‘................................

FU.III.UE CH.EMICAL DEVELOPN[ENT_HQ F. MCDllffie......................
ANATYTICAL CHEMISTRY 'FQR THE MOLTEN-SALT REACTOR — J. C. White....
N[ETAILURGICAL DEVELOPD"]ENTS"‘A. Taboada;..........................

Msm GR-APHITE—W. H. Cook........................................

115

147

167
190
201
205
214

252

288
304
320
330

373
INTRODUCTION

This semiannual report is a collection of papers that were presented
at a general information meeting on the Molten-Salt Reactor Experiment at
the Oak Ridge National Laboratory, August 18 and 19, 1964, It describes
the design and construction of the Experiment, the related research and
development, and is intended to bring the reader up to date on the status
of the general technology of molten-salt thermal-breeder reactors.

Previous progress reports of the Molten-~Salt Reactor Program are
listed below:

bS]

ORNL~2474 Period Ending January 31, 1958
ORNL-2626 Period Ending October 31, 1958
ORNL-2684 Period Ending January 31, 1959
ORNL=~2723 Period Ending April 30, 1959
ORNL~2799 Period Ending July 31, 1959
ORNL-28°0 Period Ending October 31, 1959
ORNL-2973 Periods Ending January 31 and April 30, 1960
ORNL-3014 Period Ending July 31, 1960
ORNL-3122 Period Ending February 28, 1961
ORNL-3215 Period Ending August 31, 1961
ORNL-3282 Period Ending February 28, 1962
ORNL-3369 Period Ending August 31, 1962
ORNL-3419 Period Ending January 31, 1963
ORNL-3529 Period Ending July 31, 1963
ORNL-3626 Period Ending January 31, 1964

THE PLACE OF MOLTEN-SALT REACTORS IN THE AEC PROGRAM

E. E. Sinclair

This meeting was organized so that interested parties may learn about
the status of molten-salt reactor technology and inspect the Molten-Salt
Reactor Experiment (MSRE) before it goes into operation. As you know,
the MSRE is just about ready for prenuclear operation, and the timing for
this meeting was selected with that fact in mind.
As a representative of the Atomic Energy Commission, which is sup-
porting the development of molten-salt reactor technology, I have agreed
to describe "the place of molten-salt reactors in the AEC program.” As
a matter of fact, this is wvery easily done by reference to the 1962 AEC
Report to the President, which delineates the objectives of this Nation's
reactor development programs. In discussing the program for the long-
range future (i.e., the development of breeders which will fully utilize
nuclear resources), the Report states in part: "Although breeding in the
thorium-uranium 233 cycle can build upon experience gained with less ad-
vanced reactors ... vigorous and specific efforts will be required to at-
tain breeding on a significant scale. Both fuel and blanket systems must
be pushed. Attention should be directed at methods of continuous removal
of fission products, including the use of fluid fuels (such as fused ura-
nium salts) and blanket materials. Experimental reactors designed to
breed must be built and operated. Hopefully, within the next several
years, the program will achieve the stage where operating prototypes will
be appropriate.”

From this statement and the AEC support of the molten-salt program
here at ORNL, it is evident that the AEC looks to molten-salt reactor
technology for the near-term accomplishment of breeding on a significant
scale in the Th-?33y cycle. There is, of course, no obvious reason why
molten-salt technology could not be' developed for fast breeding in the
uranium-plutonium cycle, and perhaps this will someday be done. The salt
system that has been developed here, however, quite naturally lends itself
to thermal breeding in the Th-233y cycle; and it is the objective of this
program to develop and demonstrate a reactor system which will exploit
this capability.

Although the forthcoming operation of the MSRE will not achieve this
objective alone, it is an exceedingly important and necessary step toward
the goal. The MSEE, being a small single-region machine, is not designed
to breed, but it does tie all the pertinent technology developed thus far
together into a relatively simple operating system. Its operation is ex-
pected to supply the information and confidence needed to proceed with
the next and, perhaps final, step of the demonstration program: the de-
sign, construction, and operation of a two-region thermal breeder suffi-
ciently large to permit scale-up to commercial sizes.

Tt would be wishful thinking to suppose that there will be no "bugs"
in the MSRE as designed and built. The history of reactor development
suggests that there will be bugs or minor difficulties unforeseen in de-
sign or caused by errors in construction. These are unwanted but are more
or less expected., It is also possible that operation of the MSRE will
uncover some truly unexpected major problems. I personally do not expect
any, because the technology base that has been worked out for the MSRE -
is very broad and thorough. In my view the MSRE is much better off in
this respect than has been the case for some past reactor experiments.
3

Today and tomorrow, this technological base which has been developed
for the MSRE will be explained in detail. I hope you are as favorably
impressed as I am, by the broad coverage, yet thoroughness, of the devel-
opménts that have been conducted here at ORNL. I venture to predict that
there will be another information meeting here in the not too distant
future to introduce the operation of a molten-salt breeder prototype. I
hope to see you all here at that time.

MOLTEN-SALT POWER REACTORS AND THE ROLE OF THE MSRE
IN THEIR DEVELOPMENT

R. B« Briggs

Realization of a system that makes full use of the potential energy
in thorium to produce cheap electricity is the primary mission of reactor
development at the Oak Ridge National Laboratory. That system must be
an efficient breeder system. An advanced converter may be a worthwhile
step in the development, but an advanced converter does not reach the
goal., No matter how good the conversion ratio, if it is significantly
less than 1, the amount of uranium that must be mined to make up the def-
icit in fissionable material is greater than the amount of thorium that
must be mined to compensate for the thorium converted to 2337 and burned.
' For example, if the conversion ratio is 0.90, the 237U from 20 tons of
natural uranium will be burned with each ton of thorium consumed. Even
with a conversion ratio of 0.99, the 237U from 2 tons of uranium must be
supplied with each ton of thorium., '

One feature of the Th-?23U fuel system is that breeding should be

. possible with thermal and intermediate reactors as well as with fast re-
actors. Many technical and economic factors combine to favor the thermal
breeders; so we have chosen to emphasize them in our program. Our studies
lead us to believe that molten-salt reactors are the most promising of

the several possible thermal breeder reactors for achieving a satisfactory
breeding gain and producing cheap electricity.

Our estimatel of the breeding performance of a two-fluid, graphite-
moderated thermal breeder is shown in Table 1. The neutron yields and
losses are known with less accuracy than the numbers imply, but the table
was prepared to show some of the smaller losses and to balance.

We see that the uranium and thorium concentrations and the ratio of
uranium to carbon atoms can be adjusted to obtain a good balance between
the neutron yield and the parasitic absorptions in moderator and in car-
rier salts. This leaves a net of about 12 neutrons per 100 neutrons ab-
sorbed in fuel for production of excess ?32U, The losses to #23Pa can
Table 1.

Performance Estimate for

Molten-Salt Breeder Reactor

Fuelhsalt composition, mole %

~ Blanket salt composition, mole %
Moderator-fuel ratio, N(C)}/N(U)

Volume fraction of fuel salt in core

Vblume‘fraction of fertile selt in core

Thorium inventory, metric tons for
1000 Mw (electrical) :

Fertile stream cycle time, days
Fuel stresm cycle time, days
Total 223y, 2357, and 233Pa inventory,

kg
Net neutron yield, fié
Neutron losses:
233y absorptions
235y absorptions
2321 fission
233pa absorptions (x2)
236y absorptions
237Np absorptions
135%e absorptions
Sm absorptions
Other fission product absorptions
Corrosion product absorptions
Moderator absorptions
Fuel carrier absorptions
" Thorium carrier absorptions
Loss of delayed neutrons
Leakage
Chemical processing losses
Net breeding gain
Net fuel yield, % per full power year

0.071

35
11
880

2.2137

0.9172
0,0828
0.0019
0,0120
0.0106
0.0004
0.0050
0.0001
0.0218
0.0008
0.0291
0.0284
0.0177
0.0043
0.0016
0.0022
0.0778
8.6

0.070

35
23
860

2.2131

0.9150
0.0850
0.0019
0.0117
0.0115
0.0010
0.0050
0.0001
0.0267

0.0008.

0.0286
0,0302
0.0199
00,0043

0.0016

0.0022
0.0676
8.0

0. 4UF;~63LiF-36,6BeF,
15ThF,~67LiF-18BeF,
5100
0.16 _
0.069  0.068 0.068 0.066  0.065
270
35 50 100 . 200 200
3 55 55 72 84
860 900 1020 1220 1220
2.2128 2.2120 2,2115 2.2103 2.2097
0.9139 0.9111 0.9095 0.9053 0,9034
0.0861 0.0889 0.0905 0.0947 0,0966
0.0018 0.0018 0.0018 0.0018 0.0018
0.0116 0.0113 0.0112 0.0108 0,006
0.0119 0.0132 0.0140 0.0164 0.0176
0,0010 0,0008 0.0008 0.,0010 0.0011
0,0050 0.0050 0.0050 0,0050 0.0050
0,0001 0.0002 0.0003 0,0005 0.0005
0.0317 0.,0422 0,0442 0.0532 0.0631
0.0008 0.000¢ 0.0008 0,0008 0,0008
0.0286 0.0287 0.0287 0,0287 0.0287
0.0302 0.0302 0.0291 0.0301 0,0301
0.019¢ 0.0196 0.0195 0,0192 0,0190
0.0043 0.0043 0,0043 0,0043 0,0043
0.0016 0.0016 0.0016 0.0016 0.0016
0.0022 0.0022 0.0022 0.0022 0,0022
0.0622 0,0501 0.0480 0.0347 0.0232
7.3 5.6 5.0 2.8 1.9

0.066
140

50
50
880

2.2095

0,9021
0.0979
0.0018
0.0214
0.0165
0.0010
0.0050
0.0003
0.0478
0.0008
0.0287
0.0301
0.0192
0.0043
0.0016
0.0022
0.0288
3.4

be kept low by providing an ample inventory of thorium. Losses to fission
products must be kept low, and this can be done by processing the fissile
and fertile streams on short cycles — cycles that would be prohibitively
expensive if they invélved storage, shipment, dissolution, extraction,

and refabrication of solid fuel elements, but that should be acceptably
‘cheap if they involved only simple on-site physical and chemical separa-
tions from liquid fuels and blankets.

We can readily obtain a breeding gain and fuel yield of 5%, a spe-
cific power of 1 Mw (electrical) per kilogram of fissile material and
4 Mw (electrical) per metric ton of thorium. In the studies from which
the data in Table 1 were taken, we assumed that 233pa could not be sepa-
rated easily from the fertile stream; thus we were limited to rapid re-
moval of 233U by the fluoride volatility process. Our chemists have evi-
dence that 233Pa in very low concentration can be separated from the
fertile stream on solid adsorbents. With developments such as this and
use of the shorter fuel cycles, we should be able to push the breeding
gain to 8%, the yield to 10%, and the specific powers to about 2 Mw (elec-
trical) per kilogram of fissile material and 10 Mw (electrical) per metric
ton of thorium. Any type of breeder reactor with performance character-
istics equal to the more conservative ones listed above would be an ac-
complishment of major importance. One with the more advanced character-
istics would be an outstanding achievement.

Our belief that molten-salt reactors can produce cheap electricity
is founded, in part, on favorable estimates of costs for conceptual de-
signs — fuel costs?® of 0.5 to 1.5 mills/kwhr and capital costs?s % of $125
to $l45/kWhr. These estimates were made several years ago and were equal
to, or less than, those for competitive systems. We have observed no
later development, technical or economic, that should adversely affect -
the outlook for molten-salt reactors.

The costs for a large molten-salt reactor really cannot be estimated
accurately enough to be the primary basis for developing a new reactor
system. To develop a power reactor technology is a long and expensive
business. A new type of reactor merits consideration for such development
only if it has basic characteristics which give it considerable long-term
promise., Early large-scale versions must produce competitive power, but
there must be potential for bettering the competitive position through
improvements which lead to larger unit power output, higher thermal effi-
ciency, and better utilization of fuel. Molten-salt reactors have these
important basic characteristics.

Low capital costs are promoted by the:following characteristics.
Fuel and blanket salts have vapor pressures below 1 atm up to 2600°F, and
they have good heat-transfer and fluid-flow properties. Thus, molten-salt
reactors are low-pressure, high-temperature heat sources, and units can
be designed to supply steam to one or more of the largest modern turbine- -
generator sets. The salts do not undergo violent chemical reactions with
air or with a variety of coolants. Volatile fission products can be
purged continuously. These characteristics, when combined with the low
pressure, can be used to simplify containment.
6

Uranium, thorium, and plutonium fluorides are sufficiently soluble
in appropriate carrier salts for the fuel and blanket liquids to be true
solutions, The solubility increases with temperature, so the solutions
are thermally stable; they are also stable to radiation. There is no
need to resort to slurries or to provide equipment for recombining decom-
position products.

The fuel solutions generally have a high negative temperature co-
efficient of reactivity. ZXenon-135 can be removed continuously. Fuel
can be added while the reactor is at full power by means of relatively
simple equipment. Only a small amount of excess reactivity must be con-
trolled., These attributes should lead to simplified nuclear control and
safety systems. :

Several characteristics promote low fuel costs. The good radiation
stability permits long burnup where desirable. On the other hand, there
are no fuel elements to fabricate, the reactors can be designed for good
neutron economy, and the fluids can be handled conveniently; so we have
the opportunity to remove bred material and fission products on short
cycles and to obtain good breeding ratios at low cost. For example, the
fluoride volatility process can be used to remove 233y from the fluoride
salt of a thorium breeder blanket without otherwise changing the liquid.

Low operating costs are obtained by building large units that operate
continuously. Materials are available to contain the salts at high tem-
perature with little corrosion. Well-designed equipment, carefully fab-

- ricated of these materials, should require little maintenance,

-Some characteristics of the salt systems tend to make the cost high.
Since the salts melt at 850 to 950°F, we must make provision for preheat-
ing the reactor equipment to a high temperature before the salts are ad-
mitted and we must prevent them from freezing. We must provide an inter-
mediate heat-transfer system or a specially designed steam generator to
keep water from coming in contact with the salt and to isolate the reactor
from high pressure in the event of a leak. The chemical reaction on mix-
ing is not violent; but hydrogen fluoride is generated and it is corro-
sive., Also, zirconium and uranium oxides precipitate from the salts in
contact with moisture.

Lithium fluoride is a constituent of fuel and blanket salts, and the
separated isotope'7Li must be used to obtain low parasitic absorption of
neutrons., Stainless steel and Inconel will contain the salts satisfac-
torily at 1000 and possibly 1200°F, but more expensive materials must be
used at higher temperature. INOR-8 is preferred at 1300 to 1500°F. Re-
fractory metals — molybdenum and niobium — and graphite appear to be the
best container materials for higher temperatures. The graphite must be
coated or otherwise specially processed to have low permeability to salt
"and gaseous fission products.

Since the reactors involwve the handling of large quantities of mobile
fission products, some special precautions must be taken in the contain-
ment. Also, the equipment in the circulating systems becomes radioactive
and must be serviced by remote maintenance methods.
We believe that the advantages of 'molten-salt reactors can greatly
outweigh the disadvantages in well-designed and well-developed plants.
When demonstrated, they should assure the molten-salt thermal breeders a
prominent place in our nuclear power industry.

With the preceding as an introduction, I would like to discuss
briefly some general features of a molten-salt thermal breeder reactor.
Typical compositions of fuel and blanket salts for the breeder are shown
in Table 2. They consist of uranium and thorium fluorides dissolved in
a lithium fluoride—beryllium fluoride carrier salt., We use beryllium
fluoride to obtain mixtures with low liquidus temperatures and lithium
fluoride to obtain good fluid properties. Fluorine, beryllium, and 714
all have low neutron absorption cross sections, and the fluorides offer
advantages in processing the fuels.

Although these salts have moderating properties, additional moderator
is required to make a thermal reactor. Graphite is the preferred moder-
ator because it can be used in direct contact with the salts, thus elimi-
nating the need for a low-cross-section cladding material. The salts do
not wet graphite and will not enter the pores at pressures as high as
- 200 psi if the pore openings are less than about 0.4 p in diameter. An
experimental grade of graphite that satisfies this requirement was ob-
tained for the MSRE; additional development and improvement of that ma-
terial should produce one that would meet all the requirements of a large
power reactor. ‘ o

The costs of the salts should be compared with those of other reactor
materials. Typical coolants costs are a few cents per cubic foot for Hy0;
$1.20/ft3 for helium at 500 psi; $12/ft> for sodium; $92/ft> for lead;
$110/ft3 for a heat-transfer salt composed of lithium, sodium, and potas-
sium fluorides; and $1400/ft? for Dy0. Fuels cost about $3500/ft> for
natural uranium oxide, $45,000/ft3 for a 2.2% enriched uranium oxide for
a boiling or pressurized-water reactor, and $32O,OOO/:E"t3 for a UO,-Puls
mixture in a large fast breeder. The costs for fuels do not include fab-
rication. Thus we see that the fuel salts are relatively cheap fuel mix-
tures, but they are expensive heat-transfer fluids and must be used effi-
ciently.

A concept of a two-fluid breeder reactor’ is shown in Fig. 1. We
generally favor this type, in which the fuel is pumped through the reactor
core and then through external heat exchangers. The core can be made com-
pact without the necessity for providing large amounts of heat-transfer
surface in a small volume, and it can be made almost entirely of high-
purity graphite. The heat exchanger can be made compact and efficient
and of the most corrosion-resistant materials without regard for their
neutron cross sections. Gaseous fission products can be stripped from
the circulating fuel rapidly and continuously to hold the loss of neutrons
to 13%%e to a very low level. The cost and the high radiation level of
the circulating fuel dictate that the components be closely coupled.

Figure 1 represents a reactor that would produce 1200 Mw of heat for
generating 500 Mw of electricity. The reactor has a core about 8 ft in
diameter by 8 ft high, completely surrounded by a 3-ft-thick blanket of
Table 2. Compositions, Properties, and Costs of Ty?ical Fuel and Blanket Salts

Core of Two-Region

Blanket of Two-Region

Cost Breeder Breeder
($/1b)
Mole % Wt % $/ft> Mole % Wt % $/ft3
salt constituent
UF,2 (74% 233U and 23°U) 4200 0.3 2.7 13,700
ThF,, ' 5.50 15 64 770
LiFP 16 63 48.0 920 67 24, 840
BeF,, 7 36.7 49.3 410 18 12 180
15,000 1790
Density, 1b/ft> 120 220
Liquidus temperature, °F 850 930
Heat capacity, Btu ft=3 (°F)~1 65 65
Thermal conductivity, Btu ft™! hr~! (°F)~1 3 3
Viscosity at 1200°F, 1b ft~! hr~l 20 15

®Fissile material values at $5450/1b.
PLithium-7 (99.996%) values at $55/1b as LiOH.
fertile salt, A l-ft-thick graphite reflector is installed between the
blanket and the reactor wvessel.

The core is an assembly of graphite prisms 7—1/2 in., square and about
8 ft long. The corners of adjacent prisms are machined to form vertical
passages of circular cross section about 5 in., in diameter. The fuel salt
flows in bayonet tubes of impermeable graphite which are inserted into
these vertical passages. The outer tubes are 3.75 in. ID with walls 0.5
in. thick and they are brazed to a metal plenum. The inner tubes are 2.4
in., ID with walls 0,25 in, thick. They are joined to the inner plenum
by mechanical joints since some leakage of fuel can be tolerated at this
point,.

The graphite-moderator assembly is supported by graphite structural
members in the blanket. The annulus between the fuel tubes and the mod-
erator and the regions between the core and the reactor vessel are filled
. with blanket fluid, which is maintained at a slightly higher pressure than
that of the fuel stream. If a leak develops, fertile material will enter
the fuel stream, and the reactivity will decrease.

Fuel salt enters the core from the inlet plenum at 1125°F, passes
down through the annulus in the bayonet tubes, rises through the inner

UNCLASSIFIED
ORNL-LR-DWG 46040RA

SECONDARY
COOLANT

b 1
a *_J%F
¢

T
4

_é = ]E//\:/—_ PUMP

(

- . T |l — HEAT
EXCHANGER

{

z
|

}

L]
\— ]

B

FUEL SALT“:::::;‘lPHHIHH [T
(PRIMARY COOLANT)

GRAPHITE-METAL
JOINT

Za——REFLECTOR

7
5; fj———GRAPHlTE

P—
MODERATOR

GRAPHITE TUBES
VN

T 7
Ya——BLANKET

Fig. 1. Molten=-Salt Breeder Reactor.

h VESSEL

AVAY.

)
10

tubes, and is discharged at an average temperature of 1300°F. It is col-
lected in the outlet plenum and passes up through a duct to the impeller
chamber of the pump, then through the heat exchanger, and back to the re-
actor inlet. Changes in volume of the fuel are accommodated in the space
in the pump tank above the impeller. The flow rate of the fuel is 40,000
gpm., This is one-third to one-half the flow rate specified for water-
cooled and sodium-cooled reactors with the same electrical output. The
maximum velocity in the graphite tubes is 25 fps, and the maximum temper-
ature is 1400°F.

The blanket salt is pumped. through an external heat exchanger at a
rate of 5000 gpm. The temperatures are about the same as those for the
fuel stream.

The coolant salt (a mixture of lithium, potassium, and sodium fluo-
rides) in an intermediate circuit is pumped through the fuel and blanket
heat exchangers at a rate of 50,000 gpm (entering at 950°F and leaving
at 1100°F) and then through steam generators, superheaters, and reheaters.
The steam cycle can be varied over a considerable range to suit local con-
ditions and the preference of the user. In our studies, we have held the
steam temperature to 1050°F, and the pressure has been varied between 1800
and 3500 psi. With one stage of reheat and several stages of feedwater
heating, the net thermal efficiency is in the range of 42 to 45%.

Reactors of this general configuration can be conceived with multiple
pumps and heat exchangers and heat outputs much greater than 1200 Mw,
Our engineers estimate that as much as 5 Mw of heat (equivalent to 2 Mw
of electricity) can be extracted from the reactor for each cubic foot of
fuel salt in the reactor core and heat-removal systems.6 Plants with high
performance like this are the goal of the thermal-breeder development and
can be built when we learn that the graphite structures and the heat ex-
changers will operate for many years without maintenance.

Farly wversions of the breeder may have to sacrifice some in perform-
ance in order to make use of lower flow velocities and more conventional
equipment and to provide better access for maintenance. One such concept
of a 2400-Mw reactor is shown in Figs. 2 and 3. The graphite region is
about 12 ft square by 15 ft high. It consists of graphite bars that are
7—3/4 in. square over a 1l0-ft length in the center and are reduced to 5
in. in diameter over the end sections. The reduced sections are in the
end blankets. Groups of nine bars are assembled into 2-ft-square modules.
Fuel salt flows through 3-1/2-in,-diam holes in the centers of the bars,
up through one bar and down through an adjacent bar in the same module.
Pipes which are brazed to the bars are connected to concentric inlet and
outlet headers at the bottom of each module. The inlet headers are welded
or brazed into a plenum in the bottom of the reactor vessel, and the out-
let header is connected by a mechanical Jjoint. The modules are fixed and
supported by the piping at the bottom and are guided by a grid at the top.
We show access to the bottom of the reactor from a subpile room for making
the connections. However, the center bar of each module has no fuel chan-
nel and can be withdrawn separately from above. Our experience with re-
mote and semiremote maintenance gives us assurance that tools can be de-
vised to make and break the Jjoints from either end.
UNCLASSIFIED
ORNL-DWG 64-6959R

11

BLANKET

am\\\\&\? it

e

BT {f Aot
fesns

T e
MIRSERIES

I . . . .

Two-Region Molten-Salt Breeder.

Fige 2
12

The graphite structure is contained in a reactor vessel that is about
20 ft in diameter and 25 ft high. A graphite reflector is installed next
to the vessel wall. The interstices of the core and the reflector and
the space between the core and the reflector are filled with blanket salt.
The atmosphere over the blanket is helium or argon,

UNCLASSIFIED
ORNL-DWG 64-7465

g

18 ft

10ft

o

2 ft

OUTLET \

FUEL INLET

Fig. 3« Core Module for Molten-Salt Breeder Reactor.
»

8.

13

The reactor vessel is centered in a 36-ft-diam containment wvessel.
Heat exchangers, pumps for fuel and blanket salts, and heaters are in-
stalled in the annulus, with access from above for maintenance. The con-
taimment vessel is insulated and surrounded by concrete shielding.

In an arrangement such as this we can attain a performance of 2.5 Mw
of heat (1 Mw of electricity) or better per cubic foot of fuel salt.

The reactors become high-performance breeders when they are directly
connected to processing plants that can remove low concentrations of bred
materials from the blanket salt and fission products from the fuel salt
at moderate cost. In fact, our early enthusiasm for the molten-salt re-
actor as a breeder was based largely on the development of the fluoride
volatility process — a process that makes it economically feasible to
hold the inventory of 233U to a very low level in a molten-salt blanket.

A flow diagram7 of a processing scheme for a molten-salt breeder re-
actor is shown in Fig. 4. Krypton and xenon are removed in the reactor
by sparging the fuel salt with helium in the pump bowl. The off-gas is
passed through charcoal beds, where the fission products are absorbed and
the helium is recycled to the reactor.

UNCLASSIFIED
ORNL-LR-DWG 54677A

He SPARGE OFF -GAS (Xe,Kr,He}

SPENT FUEL

HF

ThF; MAKEUP !
UFg +Fp EXCHANGER REDUCTION
FERTILE
L
. STREAM >
H "I FLUORINATOR FLUORINATOR
EXCHANGER 90 % HF
REDUCTION g . HF DISSOLUTION
Hpd £la PROCESS
W [V
@ Lo Fission PRODUCTS
PRECIPITATE
FERTILE SALT SALT TO WASTE
DISCARD | MAKE-UP {MAINLY RARE EARTHS}
(20-yr CYCLE} - SALT DISCARD TO
UF, FUEL REMOVE SOLUBLE
MAKE-UP UFy FISSION PRODUCTS

AND RESERVE

— PRODUCT

Fig. 4. Fuel Processing Cycles for Molten-Salt Breeder Reactor.
14

- 8ide streams of fuel and blanket salt are removed from the reactor
for processing in a facility at the reactor site. The complete fuel and
blanket charges, respectively, are processed in 10 to 60 days (3 to 15%
burnup of fissile material) and 35 to 200 days (equivalent to 300 to 1700
Mwd per metric ton of exposure) to obtain the corresponding breeding gain
shown in Table 1. The fuel salt is first treated with fluorine. The UF,
is converted to volatile UFg and swept out of the fluorinator. The UFg
is then reduced back to UF, with hydrogen in a flame reducer and returned
to the reactor.

The uranium-free salt, containing fission products, is dissolved in
90% hydrofluoric acid. Rare earths and zirconium salts are insoluble and
are separated by filtration. The solution containing lithium and beryl-
lium fluorides and some soluble fission products (principally Cs, Rb, Sr,
Ba, Te, Se, Nb, Cd, Ag, and Te) is evaporated to dryness. Part of the
salt is discarded to rid the system of the soluble fission products, and
the rest is melted, combined with the recovered uranium, and returned to
the reactor.

The blanket salt is fluorinated to remove the bred uranium as the
hexafluoride, sparged with helium to remove traces of fluorine, and re-
cycled directly to the blanket. Since such a small fraction of fissions
occurs in the blanket, the fission-product-removal step is unnecessary.
The fission product poison level is controlled by replacing the blanket
salt on a 20-year cycle. Protactinium is not removed from the fertile
stream in this process; its concentration rises until the decay rate
equals its production rate, Loss of neutrons to protactinium is con-
trolled by adjusting the volume of the fertile salt.

The UFg from the fertile stream fluorinator is reduced to UF, with
hydrogen and blended with the fuel stream for recycle to the reactor.
Excess production is sold.

Because the methods outlined above involve only minor chemical ma-
nipulations to the bulk of the salts, we believe that they will permit
us to achieve high conversion ratios and low fuel costs in plants of 1C00-
Mw (electrical) capacity and possibly in smaller plants. However, these
methods may well be superseded by even better ones, We mentioned earlier
that the 2°3Pa precursor of 233U can be adsorbed from the blanket salt
and the effect of this on the reactor performance, Within the past month,
M. J. Kelly, one of our chemists, has shown that the carrier salts can
be separated from rare-earth fission products by vacuum distillation.,
This promises to be a substantial improvement over the HF dissolution
method and has other important implications regarding methods of operating
and controlling the reactors.

Except for the fluoride volatility process, we have been discussing
reactor and processing concepts. Concepts have real value when enough
basic technology is developed for us to convert them into authentic engi-
neering designs.,
15

Molten=-salt reactors are not entirely new., Much of their technology.
was developed here at ORNL in 1951 to 1958 in the Aircraft Nuclear Pro-
pulsion Program, A proof-of-principle experiment with molten-salt fuel —
the Aircraft Reactor Experiment — was built and, in 1954, operated suc-
cessfully for several hundred hours at 1200 to 1500°F, with periods of
nuclear operation at power levels as high as 2.5 Mw. But a large power
breeder reactor differs in many ways from an aircraft power plant. The
fuels and some materials must be different to obtain good breeding ratios.
The power levels must be considerably higher; methods must be available
for processing fuel and blanket salts efficiently; and costs are important.
On the other hand, the weight and size are not restricted. A reactor for
producing power for a modern turbogenerator can operate at considerably
lower temperature than a power plant for a supersonic aircraft. Extreme
compactness is a virtue but not a necessity.

Research and development for civilian power reactors was undertaken
in 1957, By 1960 enough favorable experimental results were obtained to
indicate that a molten-salt thermal breeder should be feasible, but that
experience with an operating reactor was essential to prove the feasi-
bility. At about this same time the importance of breeding to the long-
range utilization of nuclear fuels, and particularly to the utilization
of thorium, was gaining increased attention. Interest in the breeding
potential of the molten-salt reactors, their capacity to use the heat
efficiently to produce electricity, and in the characteristics that should
permit this to be achieved at low cost led the Atomic Energy Commission
to authorize design and construction of the Molten-Salt Reactor Experiment
(MSRE) in the summer of 1960, -

The MSRE is needed to investigate some essentials of chemistry, ma-
terials, engineering, and operation of the molten-salt reactor concepts.
~The major physics questions are related to uncertainties in the breeding
ratios. They can finally be resolved only by a material balance over a
large plant.

The central problem of the thermal breeder reactor is to prove the
suitability of graphite for the core structure. The graphite must have
low permeability to salt and fission products; it must have long life
under reactor conditions; and it must be obtainable in large size. In
1960 this kind of graphite was a laboratory curiosity. Procurement of
material for the MSRE was a step in development of graphite for large re-
actors. Operation of the MSRE will provide us with essential data about
the long-term compatibility of graphite with salts and structural material
in the reactor enviromment.

We know a great deal about the chemistry of molten salts in the ab-
sence of radiation and have some knowledge of radiation effects. But the
complex environment of a power reactor cannot be duplicated in a simple
capsule or in-pile loop. Operation of fuel and blanket salts in a reactor
is essential to the investigation and demonstration of their compatibility
with graphite and structural materials under radiation. The predicted
behavior of fission products must be confirmed, and their distribution
in the reactor systems must be established.
16

INOR-8 (Hastelloy N or Inco 806), a new nickel alloy containing 17%
molybdenum, 7% chromium, and 4% iron, was developed specifically to hawve
good strength and to contain reactor fuel and blanket salts with little
corrosion at high temperature. This is a preferred material for molten-
salt power reactors. Experience is needed with the construction of re-
actor components and piping and with the behavior of commercial materials
during long exposure in a reactor plant.

The practicality of molten-salt reactors depends on our ability to
maintain the radiocactive equipment by remote or semiremote methods., We
have had considerable experience with maintenance of radioactive equipment
in aqueous homogeneous reactors, but the problems encountered depend very
much on the types of fuels and materials used in the reactors and on their
interactions. We need some experience with radioactive molten-salt sys-
_tems to establish criteria for the design of large reactors.

The salts in the reactor equipment and piping must be kept molten
at all times. This should be easy in all but the steam generators after
some fission products have accumulated in the salts. However, heaters
will be needed for preheating the systems and for maintaining the high
temperatures in the absence of fission product heat. We require experi-
ence with the design, operation, and maintenance of heating equipment for
radioactive systems.

Finally, we need general operational experience with a molten-salt
reactor in order to establish the criteria for the design of a practical
power plant. We cannot predict the detailed behavior of the plant and
equipment or the day-to-day problems that can arise with its operation
and with the containment of radiocactivity or the numerous small improve-
ments that, in total, can greatly simplify the operation and increase the
confidence in a new device,

We have designed, developed, and are now preparing to operate the
MSRE to test much of the technology of the thermal breeder reactor. A
flow diagram for the reactor systems is shown in Fig. 5. The reactor
produces 10 Mw of heat while operating at 1200°F. The base pressure in
the system is 5 psig at a free surface in the pump bowl. Fuel salt is
pumped at a rate of 1250 gpm through the shell of the heat exchanger and
through the reactor vessel, The average increase in temperature as the
fuel passes through the core is 50°F at 10 Mw.

A coolant salt in an intermediate circuit circulates through the
tubes of the heat exchanger to remove heat from the fuel, The heat is
dissipated to the atmosphere in an air-cooled radiator. Power recovery
is not an objective of this experiment, so we have included no power-gen-
eration equipment.

Drain tanks are provided for storing fuel and coolant salts when the
reactor is not operating. Drainage of the fuel to the storage tank is
the primary shutdown mechanism.
17

FUEL
PUMP

UNCLASSIFIED
ORNL—-LR-DWG 591584

COOLANT
PUMP

HEAT , 552°C 52 liter /sec

!
| |
| 663°C | 1025°F 830 gpm
| \pager EXCHANGER Ii o
i 625°C ' H - ~\
[ . 1250 | $110C°F
| 1#75°F } gpm | 593°C
| 80 liters/sec | i
| | COOLANT SALT
| REACTOR | LiF-66%
| VESSEL | BeF; 34%
: |
| | 300°F
| REACTOR | 150°C
i CELL |
| | . -
b — T ————————— S T
r
I I 167,000 cfm
I I 100°F
L_'—_‘_"_I 79m3/sec
DRAIN TANK , 389C
l CELL |
| FREEZE |
| VALVE I
I [
I I
| ‘ ’ |
@;;;;:;:: qjizj;:o ' FLUSH TANK COOLANT
AND DRAIN DRAIN TANK (68 ft3) DRAIN TANK
TANK (68 ft7) (68 it3) 1.9 m? (401t3)
1.9 m? 1.9 m3 {1 m?
Fig. 5. MSRE Flow Diagram.

A drawing of the reactor vessel
about 5 ft in diameter and 8 ft high,

The core is an assembly of graphite bars 2 in. square by 5 ft long.

is shown in Fig. 6. The wvessel is
including the nozzle on the top.
Fuel

salt circulates in direct contact with the graphite through channels ma-

chined in the sides of the bars.
installed through the access nozzle.

radiating graphite and metal specimens in salt in the core.

length core bars can be withdrawn by

A1l equipment that contains salt is made of INOR-8.

Three control rods operate in thimbles

A channel is also provided for ir-.
Five full-
removing the thimble assembly.

The graphite

is a new material that satisfies many of the requirements for graphite

for the breeder.

A1l wvessels and pipes that contain salts are heated.

Considerable design and development effort was spent in making essential

equipment replaceable.

An elevation of part of the MSRE building is shown in Fig. 7.
reactor vessel, pump, and heat exchanger are installed in one cell,

drain tanks are -in an adjoining cell,

ondary containment and will be sealed when fuel is

coolant system is in a shielded area.

are controlled, but no other secondary containment

The
The -

constitute the sec-

in the reactor. The f

and access to the. area

1s necessary.

These cells

Ventilation
18 -

. The MSRE is primarily an experiment in reactor technology. We re-
alize the importance of the fuel and blanket processing to achievement . .
of the final goal but have chosen to prove the feasibility of the reactor
systems before proceeding with development of some of the processing con-
cepts. However, it proved to be advisable to install equipment at the
reactor site for fluorinating and hydrofluorinating the fuel. This equip-
ment is being installed in a cell adjacent to the drain tank. We believe

 UNCLASSIFIED
ORNL-LR-DWG 61097RIA

FLEXIBLE CONDUIT TO

GRAPHITE SAMPLE ACCESS PORT .,Fjlja y CONTROL ROD DRIVES

COOLING AIR LINES

ACCESS PORT COOLING JACKETS

FUEL OUTLET REACTOR ACCESS PORT

SMALL GRAPHITE SAMPLES
HOLD-DOWN ROD
OUTLET STRAINER

CORE ROD THIMBLES
LARGE GRAPHITE SAMPLES

CORE CENTERING GRID

FLOW DISTRIBUTOR
VOLUTE

GRAPHITE —~MODERATOR § %
STRINGER '

FUEL INLET -/

REACTOR CORE CAN —

~~— CORE WALL CCOLING ANNULUS

REACTOR VESSEL —

>
e MODERATOR
SUPPORT GRID

ANTI-SWIRL VANES
VESSEL DRAIN LINE

Fig, 6. MSRE Reactor Vessel.
19

UNCLASSIFIED
ORNL-DWG 63-1209R

REMOTE MAINTENANCE
CONTROL ROOM

. REACTOR VESSEL 7. RADIATOR

. HEAT EXCHANGER 8. COOLANT DRAIN TANK

. FUEL PUMP 2. FANS

. FREEZE FLANGE 10, FUEL DRAIN TANKS

. THERMAL SHIELD | 11, FLUSH TANK

. COOLANT PUMP 12. CONTAINMENT VESSEL
13. FREEZE VALVE

Fig. 7. Elevation of Part of MSRE Building.

that its use will help considerably to demonstrate the advantages of on-
site processing for liquid fuels.

The MSRE was fitted into an existing building; the reactor was in-
stalled in an existing containment vessel that was modified to provide a
flat top. The drain tank cell and some other areas are new, but most of
" the facility was modified only slightly for the MSRE. This resulted in
some compromises in the design and an installation so compact as to give
some the impression of complexity. The power level was held to 10 Mw as
a sensible maximum for an experiment; but criticality considerations dic-
tated the size of the reactor vessel, and the size of the reactor vessel
dictated the flow rate of the fuel. Except for the amount of heat-trans-
fer surface in the heat exchanger and radiator, this small ingtallation
is more nearly representative of a 40~ to 60-Mw reactor.
20

Suppose that good operation of the MSRE and detailed engineering
studies of large plants giwve convincing evidence of the promise of molten-
salt reactors for breeding and producing low-cost power. Where do we go
from the MSRE? Three areas need special attention in progressing to large
power breeder plants: (1) the graphite core structure needs additional
work; (2) the fuel and blanket processing must be brought to a satisfac-
tory stage of development; and (3) large-scale equipment must be developed
and operated., Pilot-plant demonstration of elements of the core and of
the fuel and blanket processes should be done in a modification of the
MSEE or in a thermal breeder test facility. But, primarily, we want
molten-salt reactors to be useful, and they will be useful when operated
by a utility to produce power.

We believe that the MSRE can be followed by a power reactor proto-
type. Much of the large equipment can be modeled after equipment being
developed for sodium reactors., The graphite development and the process-
ing development do not hawve to be complete, A reactor similar in design
to the concept shown in Fig. 2, but with graphite throughout the blanket,
can be fueled with a mixture of the fuel and blanket salts, shown in Table
2, to make a one-fluid advanced converter reactor. This reactor would
require very little in the way of processing to have very low fuel costs.
It could be used to test graphite elements as well as other major equip-
ment for the breeder. It could eventually be changed into a breeder by
installing proved graphite assemblies in the core and additional facili-
ties for processing fuel and blanket streams. Such a prototype is the
keystone of any plan for making an early profit on the effort invested
in the development of molten-salt reactors.

Up to this point we have mentioned only the thermal breeder and a
possible precursor (the advanced converter) and the relationship of the
MSRE to them, However, there are other interesting and important possi-
bilities for molten-salt reactors. dJ. A. Lane has suggested that‘235U
and plutonium may become very cheap in the future and that a molten-salt
reactor with its high thermal efficiency and simple fuel would be ideal
for burning this material in small reactors to generate 50 to 250 Mw
(electrical). P. R, Kasten has proposed the development of an interme-
diate breeder reactor with the molten-salt fuel cooled internally by the
molten-salt blanket.

For many years fast reactors with molten-salt fuels have been judged
to be impractical. The inventory in externally cooled versions was always
‘prohibitive., Internally cooled designs had too low a specific power, or
the salt temperature was too high, or too much structural material was
required., E. S. Bettis has suggested molten lead as a direct-contact
coolant for molten-salt reactors. He has proposed some designs to achieve
very high specific power in large fast reactors. Preliminary experiments
have shown molten lead to be immiscible and unreactive with fluoride and
chloride salts.,

None of the reactors that have received serious consideration in the
past have made use of the full high-temperature potential of the molten-
salt fuels, There is no need to go to temperatures above 1400°F in the
21

reactors to provide the highest temperature steam that can be used to ad-
vantage in today's steam cycles. However, A, Fraas® suggested that the
potassium vapor cycle being dewveloped for space applications should one
day be a good topping cycle for the conventional steam cycle to raise the
thermal efficiency to about 55%. He proposed a molten-salt reactor op-
erating at about 1900°F as a good heat source for the combined cycle.

To propose that the MSRE will solve all the problems of all these
reactors would be to exaggerate. But most of the experience with the re-
actor and with the handling of the fuels will apply to all. We have en-
countered problems in building the reactor and will encounter more in
starting and operating it. However, our experience with the test loops
and construction gives us full confidence that we have a good reactor.

We expect the MSRE to provide an impressive beginning for the development
of a variety of useful power reactors.

References

l. L. G. Alexander et al., Thorium Breeder Reactor Evaluation. Part I.
Fuel Yield and Fuel Cycle Costs in Five Thermal Breeders, ORNL-CF-
61-3-9, p. 150 (March 1961).

2 L. Go Alexander et al., Thorium Breeder Reactor Evaluation., Part I.
Fuel Yield and Fuel Cycle Costs in Five Thermal Breeders, ORNL-CF-
61-3-9, p. 147 (Maréh 1961).

3. Oak Ridge National Laboratory, Capital Investment for 1000 MWe Molten
Salt Converter Reference Design Power Reactor, Sargent & Lundy — 1994
(Dec. 27, 1962).

4. L. G, Alexander et al., Cost of Power from a 1000-MWe Molten-Salt Con-
verter Reactor, ORNL report (in preparation).

5. Le Go Alexander et al., Thorium Breeder Reactor Evaluation, Part I.
Fuel Yield and Fuel Cycle Costs in Five Thermal Breeders, ORNL-CF-
61-3-9, p. 69 (March 1961).

6. L. G. Alexander et al., Thorium Breeder Reactor Evaluation. Part I.
Fuel Yield and Fuel Cycle Costs in Five Thermal Breeders (appendices),
ORNL-CF-61-3-9, p. 10.

7. L. G. Alexander et al., Thorium Breeder Reactor Evaluation. Part I.
Fuel Yield and Fuel Cycle Costs in Five Thermal Breeders (appendices),
ORNL-CF-61-3-9, p. 87. ‘

8. A. P. Fraas et al., A Potassium-Steam Binary Vapor Cycle for Nuclear
Power Plant, “ORNL-3584 (May 1964).

22
MSRE DESIGN AND CONSTRUCTION

W. B. McDonald

Introduction

The Molten-Salt Reactor Experiment (MSRE) is a circulating fuel,
graphite-moderated, single-region reactor designed for a heat generation
rate of 10 Mw, The fuel employs a molten mixture of lithium, beryllium,
and zirconium fluoride salts as a solvent for uranium or thorium and ura-
nium fluorides. Heat generation occurs as the fuel flows through machined
passages in the graphite core, and heat is transferred from the fuel salt
to a similar coolant salt in a shell-tube heat exchanger. Finally, the
heat is dissipated to air in an air blast radiator. ©Since power recovery
1s not an objective of this experiment, no electric power generation
equipment is utilized.

It is intended in this section to give a general description of the
design of the MSRE systems, pointing out the most salient features and
presenting data of general interest, and to describe the major features
of the reactor and facility construction. The reader having interest in
a more detailed exposition of the MSRE design is referred to ORNL-TM-728,
Molten-5alt Reactor Experiment, Part I, Reactor Design.

Description of MSRE Equipment and Process

The simplified drawing in Fig. 1 shows the general arrangement of
major equipment items. The primary fuel system consists of the reactor
vessel in which heat is generated, a heat exchanger for transfer of heat
from fuel salt to coolant salt, a sump-type centrifugal pump for circu-
lating the fuel salt, and connecting piping.

The secondary coolant system consists of a coolant-salt pump, a ra-
diator in which heat is transferred from coolant salt to air, and piping
which connects the pump, heat exchanger, and radiator.

Both the primary and secondary systems are connected to drain tank
systems for storage of the fuel, coolant, and flushing salts.

Reactor Vessel and Graphite Core

The reactor vessel is a 5-ft-diam, 8-ft-high tank containing a 55-
in.-diam by 64-in.-high graphite core structure. A cutaway drawing of
the reactor is shown in Fig. 6, p. 18. Design data for the reactor vessel
and the graphite core are presented in Table 1.
23

At the design point of 10 Mw (thermal) the fuel salt enters the flow
distributor near the top of the wvessel at 1175°F and 20 psig. The fuel
then spirals downward in turbulent flow through a l-in. annulus between
the vessel wall and the core can, cooling the vessel wall to within 5°F
of the bulk temperature of the entering fuel salt. Straightening vanes
welded to the bottom head of the reactor vessel remove the rotational
component and direct the fuel upward in laminar flow through the 0.4- by
l.2-in. channels machined in the 67-in.-long graphite bars of the core
matrix. There are about 1140 of these flow passages in which heat is
generated by the fissioning of the 37U constituent of the fuel salt.

When—fidleds—the core, having a nominal volume of 90 ft3, contains
\20 ft2 of fuel salt/ and 70 ft2 of graphite. At 10 Mw, with no fuel ab-
sOrbed by the graphite, 1.4 Mw of heat is generated in the fuel outside

UNCLASSIFIED
ORNL-DWG 63-1208R

REMOTE MAINTENANCE
CONTROL ROOM

REACTOR CONTROL
ROOM

. REAGTOR VESSEL 7. RADIATOR

!
T 2. HEAT EXCHANGER 8. COOLANT DRAIN TANK
’ 3. FUEL PUMP 9. FANS
4 FREEZE FLANGE 10. FUEL DRAIN TANKS
5. THERMAL SHIELD 11. FLUSH TANK
6. COOLANT PUMP 12. CONTAINMENT VESSEL

13. FREEZE VALVE

Fig. 1. MSRE Flow Diagram,
24

Table 1. Reactor Vessel and Core Design Data and Dimensions

Construction material ' INOR-8
Inlet nozzle, sched 40, in., iron pipe 5

size
Outlet nozzle, sched 40, in., iron pipe 5

size

Core vessel

0D, in. 59-1/8 (60 in. max)

ID, in. 58

Wall thickness, in. ~ 9/16

Overall height, in. (to center line 100-3/4

of 5-in. nozzle)

Head thickness, in. 1

Design pressure, psi 50

Design temperature, °F ‘ 1300

Fuel inlet temperature, °F 1175

Fuel outlet temperature, °F 1225

Inlet Constant area
distributor

Cooling annulus ID, in, 56

Cooling annmulus OD, in. 58

Graphite core

Diameter, in. 55-1/4

Number of fuel channels (equivalent) 1140

Fuel channel size, in. 1.2 X 0.4 (rounded
corners )

Core container

ID, in. | 55-1/2
0D, in. ' 56
Wall thickness, in. 1/4

Height, in. 68

25

the nominal core volume, 0.6 Mw 1s generated in the graphite, and 8.0 Mw

is generated in the fuel within the core, giving an average power density
of 14 kw/liter in the nominal core. The maximum power density is calcu-

lated to be 31 kw/liter.

The thermal conductivities of both the fuel salt and the graphite
are such that, even with laminar flow, the maximum graphite temperature
is only about 60°F above the temperature of the adjacent fuel, The nu-
clear averagé and maximum temperatures of the graphite are estimated to
be about 1255 and 1300°F respectively. The fuel exit temperature in the
hottest channel is about 1260°F. The mixed fuel leaves the plenum at the
top of the reactor at 1225°F and 7 psig.

A 10-in.-diam access port on top of the reactor vessel is designed
to permit the insertion of graphite and metal samples and three 2-in,-diam
thimbles for control rods into the core at the region of highest neutron
flux., The control rod poison elements are composed of short l-in.-diam
cylinders of Inconel-clad gadolinium oxide, assembled on a flexible In-
conel hose to permit passage around two bends that form an offset in each
thimble, The control rods have an individual worth of 2.8% Ak/k and a
combined worth of 6.7% Ak/k. At the maximum withdrawal rate of 0.5 in./
sec, normally limited to one rod at a time, the change in reactivity is
0.04% (Ak/k)/sec. Simultaneous insertion of the three rods at a rate of
0.5 in./sec' causes a change in reactivity of 0.09% (Ak/k)/sec., The con-
trol rods and rod driwves are cooled by circulation of ambient cell atmos-
phere through the flexible hoses and the thimbles,

Fuel Pump

A sump-type centrifugal pump, shown in Fig. 2, having a vertical
shaft with an overhung impeller, takes its suction directly from the re-
actor. The impeller rotates at 1160 rpm to discharge 1200 gpm at a head
of 49 £t of fluid. The 75-hp drive motor is directly coupled to the
shaft, and the complete assembly is about & ft high., The pump and motor
are hermetically sealed to prevent the escape of volatile fission products
to the reactor cell.

The 36-in.-diam pump bowl serves both as the pump sump and as the
normal expansion volume for the fuel salt. Gas-bubbler-type level indi-
cators are used to accurately determine the position of the free fluid
surface in the pump bowl. An overflow tank having a volume of 5.5 ft3
is located beneath the pump and is connected to the gas space above the
free surface of the fuel in the pump bowl. Although this tank is not
normally used, its volume is sufficient to provide for the free expansion
of the fuel salt under all emergency conditions,

A sampling and enriching system, located in a hot cell in the high-
bay area above the reactor, is connected directly to the pump bowl and
provides means for extracting a 10-g sample of salt for analysis or adding
90 g of fuel to alter the fuel concentration or to compensate for fuel
burnup.
26

About 65 gpm of the pump output is recirculated in the pump bowl to
facilitate the release of entrained or dissolved gases from the salt,
These gases, containing highly radioactive xenon and krypton, are carried
" to the off-gas disposal system by about 200 ft?/day (STP) of helium flow-

ing through the gas space of the pump bowl.

This helium also serves as

a cover gas to exclude alr and water vapor from the salt.

A sealed external pumping system supplies oil to lubricate the ball

bearings and rotary seals on the pump shaft and to remove heat,

generated

UNCLASSIFIED
ORNL- LR- DWG-56043-AR

[
o I )
1 —
SHAFT q% WATER-
COUPLING : 7 PR, COOLED
. & v e A ; MOTOR
SHAFT SEAL __':25
I D oo —— == T — &

LEAK DETECTOR-

LUBE OIL INLET

SHAFT SEAL

LUBE :
OIL OUTLET

LEAK DETECTOR

BUBBLER TYPE
LEVEL INDICATOR

= P ———

BUOYANCY-
TYPE =
LEVEL

INDICATOR—

LUBE OIL BREATHER

SWEEP GAS INLET

__—GAS VENT
! XENON STRIPPER
4

_=_—_rl B ]
el FUEL QUTLET

N

o -

FUEL INLET

- MSRE FUEL PUMP

Fig. 2.

Cross Section of MSRE Fuel Pump.
27

by friction or nuclear radiation, from metal parts not cooled by the salt.
The helium purge enters the pump below the lower shaft seal. A small part
of it flows upward to sweep out oil wvapor leaking through the seal, and
the remainder flows downward along the shaft to the pump bowl, preventing
.the radiocactive gases from reaching the oil. The latter flow eventually
enters the off-gas system.

The drive motor and the pump rotary assembly are separately remov-

able, This facilitates remote removal and replacement of the rotary as-
sembly, if required.

Primary Heat Exchanger

The horizontal shell-and-tube heat exchanger shown in Fig. 3 is about
16 in. in diameter and more than 8 ft long. It contains 159 1/2-in.-OD
by 0.,040-in,-wall U-tubes approximately 15 ft long, with an effective heat
transfer surface of ~254 ft*. At the 10-Mw performance level the salt
discharged by the fuel pump flows through the shell side, where it is
cooled from 1225 to 1175°F. The coolant salt circulates through the tubes

UNCLASSIFIED
ORNL-LR-DWG 52036R2

FUEL INLET

THERMAL-BARRIER PLATE
TUBE SHEET
COOLANT INLET

16.4-in. OD x O0.2-in. WALL x 8-ft LONG

COOLANT-STREAM

A
COOLANT OUTLET
SEPARATING BAFFLE fl

FUEL OQUTLET

Fig., 3. Primary Heat Exchanger for MSEE,
28

at 850 gpm, entering at 1025°F and leaving at 1100°F. In the event of a
heat exchanger tube failure, it is desired that leakage be from coolant
to fuel. So the shell side of the primary heat exchanger is at lower
pressure than the tube side.

The shell side, contains 25%-cut cross-baffles on 12-in. spacing to
achieve cross flow through the shell. INOR-8 spacing bars are tightly
laced in two directions through spaces between the triangular-pitched
tubes to eliminate tube flutter and vibration. A flow-impingement baffle
is placed on the downstream side of the shell entrance nozzle to prevent
direct impingement of the high-velocity salt stream upon the tubes.

. The most significant design data for the primary heat exchanger are
presented in Table 2.

Coolant Pump

The coolant salt is circulated by a sump-type centrifugal pump iden-
tical in most respects to the fuel pump. Major differences are the elim-
ination of provisions for removal of radioactive gases and the overflow
tank. The pump has both float-type and gas-bubbler-type level indicators
in the pump bowl. The pump is driven by a 75-hp, 1750-rpm motor and de-
livers 850 gpm at a head of 78 ft of fluid.

Radiator

Heat transported by the coolant salt is discharged to air in the ra-
diator shown in Fig. 4. Seven hundred square feet of cooling surface is
provided by 120 tubes, 0.75 in. in diameter by 30 ft long, arrayed in 10
banks, each containing 12 tubes, The tubes in each bank are butt-welded
at both ends to common headers which are in turn connected to larger flow
distribution and collection headers at the enfrance and exit ends.

Cooling air is supplied to the radiator by two 250-hp axial blowers,
having a combined capacity of 200,000 cfm. At a power level of 10 Mw,
the salt enters the radiator at 1100°F and leaves at 1025°F, resulting
in a cooling-air temperature rise of 200°F. To guard against freezing
of the salt in the radiator tubes, quick-closing doors, automatically ac-
tuated upon sudden reduction of reactor power, are provided to shut off
the air flow, and the radiator is heated by banks of electric heaters in-
side the enclosure. The doors can be adjusted at any position between
fully open and fully closed, and remotely actuated dampers can be adjusted
to bypass air around the radiator to regulate the heat removal rate.

Pertinent design data for the MSRE radiator are given in Table 3.

Drain Tank Systems

Four tanks, fiith assoclated piping, heating and cooling equipment,
and valving, are provided for safe storage of the salt mixtures when the
fuel- and coolant-salt circulating systems are not in operation. A fill-
29

Table 2. Design Data for Primary Heat Exchanger

Construction material
Heat load, Mw
Shell-side fluid
Tube-side fluid
Layout

Barfle pitch, in.
Tube pitch, in.
Active shell length, ft
Overall shell length, ft
Shell diameter, in.
Shell thickness, in.
Average tube length, Tt
Number of U-tubes
Tube size, in.
Effective heat transfer surface, ft?
Tube-sheet thickness, in.
Fuel salt holdup, ft5 ‘
Design temperature, °F
Shell side
Tube side
Design pressure, psig
Shell side
Tube side 5
Allowable working pressure, psig
Shell side
Tube side
Hydrostatic test pressure, psig
Shell side
‘Tube side
Terminal temperature, °F
Fuel salt
Coolant
Effective log mean temperature
difference, °F
Pressure drop, psi

Shell side
Tube side
Nozzles

Shell, in. (sched 40)

Tube, in. (sched 40)
Fuel-salt flow rate, gpm
Coolant-salt flow rate, gpm

INOR-8

10

Fuel salt

Coolant salt

25%-cut cross-baffled
shell with U-tubes

12

0.775 triangular

~6

~8

16

1/2

14

159

1/2 0D, 0.042 wall

~254

1-1/2

6.1

1225 inlet; 1175 outlet
1025 inlet; 1100 outlet

133

24
29

5
5

1200 (2.67 cfs)
850 (1.85 cfs)

®Based on actual thicknesses of materials and stresses allowed by

ASME Code.
UNCLASSIFIED
ORNL-LR-DWG 55841R2
PENTHOUSE
s i
DOOR DRIVE MOTOR AND : .
GEAR REDUCER —__;
4y
WIRE ROPE SHEAVE
ok
W
i .
|
SUPPORTING STEEL -1,
“WIRE ROPES -
Sl
DOOR CAM GUIDE —=;
BLDG. 7503, FIRST FLOOR .
(ELEV. 852 ft-Oin.) S
e -'|I:
: . |
' N |
|
%,
o) l &
AIR BAFFLES ———fiT—=
8 / | M= B ;
I : e
AIR INLET DUCT —_ | : ~~AIR OUTLET DUCT : -
§ 7N gl - LR
/ AIRDUCT FLANGE — 2§ ' .
I // A ; . _fl;':h . -
-~ ih & N > ) ;
MAIN AIR BYPASS DUCT ] ’ N - RADIATOR ENCLOSURE
kit
RADIATOR TUBES
AIR FLOW P DAMPER
i
/
/
/
/
Fig. 4. MSRE Radiator Coil and Enclosure.
31

and-drain line connects two fuel-salt drain tanks and one flush-salt tank
to the reactor vessel. One drain tank is provided for the coolant salt.

During operation, salt is held in the fuel and coolant systems by
means of freeze valves. In simplest form, a freeze wvalve consists of a
flattened section of pipe, associated heating and cooling devices, and
instrumentation which permit freezing and melting a salt plug as required
for unmodulated flow.

Table 3, Radiator Design Data

. Construction material - INOR-8
Duty, Mw ' 10
Temperature differentials, °F
Salt Inlet 1100; outlet 1025
Air | Inlet 100; outlet 300
Air flow, cfm at 9.9 in. Hy0 ‘ 200,000
Salt flow (at average temperature), gpm 830
rEfféctive mean AT, °F Ww§62fm
Overall coefficient of heat transfer, 58.57
Btu ft™* nr~1 (°p)"1 e
Heat transfer surface area, ft° : 706
Design temperature, °F 1250 .
Maximum allowable internal pressure _ 350
at 1250°F, psi
Operating pressure at design point, psi 75
o Tfibe,diameter, in. 0,750
_wai‘l thickness, in. | 0.072
Tube length, ft 30
‘Tube matrix | 12 tubes per row, 10
rows deep
- Tube spacing, in. 1-1/2
Row spacing, in. . 1-1/2
Subheaders, in., iron pipe size, sched 40 2-1/2
Main headers, in., ID (1/2 in. wall) 8
Air side AP, in. of H30 2.9

Salt side Ap, psi 19.8

32

Provisions have been made for remotely cutting the drain tank piping
at preselected points for the removal of components, and for making brazed
Jjoints at those points.

A fuel-salt drain tank is shown in Fig. 5. The tank is 50 in, in
diameter and 86 in. high. Its 80-ft? volume is sufficient to hold in a
noncritical geometry all the salt that can be contained by the fuel-salt
circulating system., The tank is provided with a low-pressure, gravity-
circulated cooling system, capable of removing 100 kw of fission product
decay heat by boiling water in 32 bayonet tubes that are inserted in thim-
bles in the tank.

The flush-salt tank is similar to the fuel-salt tank, except that
it has no thimbles or. cooling system, New flush salt is similar in com-
position to fuel salt, but without fissile or fertile materials. It is
used. to wash the fuel-salt circulating system before fuel is added and
after fuel is drained, and the only decay heat is from the small quantity
of fission products that it removes from the equipment.

The coolant-salt tank resembles the flush-salt tank, but it is 40 in.
in diameter by 78 in. high and has a volume of 50 ft°.

The tanks are provided with devices to indicate high and low levels
and with weigh cells to indicate the weight of the tanks and their con-
tents with sufficient accuracy to determine the quantity of salt in the
cilrculating system at any time.

Piping and Flanges

The major components of the fuel-salt and coolant-salt circulating
systems are interconnected by 5-in. sched 40 piping. The fill-and-drain
lines are 1-1/2-in. sched 40 pipe in which the freeze valves are located.

Since equipment and techniques have not been developed for remote
welding of piping, special flanges are used in the fuel-salt circulation
system to permit removal and replacement of radioactive components. These
flanges, designated freeze flanges, utilize a ring of frozen fuel salt
between the flange faces to prevent liquid leakage and a conventional
O-ring-type, helium-buffered, leak-monitored joint to prevent the escape
of fission-product-contaminated gases.

Heating Equipment

All the components and connecting piping in the salt circulating
systems are heated electrically to maintain all portions of the system
above the salt liquidus temperature of 840 to 850°F, The equipment is
preheated before salt is added, and the heaters are continuously ener-
gized during reactor operation  to ensure that there is no uncontrolled
freezing of the salt in the piping and that the salt can be drained when
necessary. The total capacity of the heaters is about 1930 kw; however,
the actual power load is less than half of this. Normally, the power is
UNCLASSIFIED
ORNL-LR-DWG 61719

INSPECTION, SAMPLER, AND
LEVEL PROBE AGGESS

STEAM QUTLET
STEAM DOME

CONDENSATE RETURN

WATER DOWNCOMER INLETS

CORRUGATED FLEXIBLE HOSE

BAYONET SUPPORT PLATE
. STEAM RISER

T
O

:

STRIP WOUND FLEXIBLE
HOSE WATER DOWNCOMER

BAYONET SUPPORT PLATE
HANGER CABLE

B

< (@

' GAS PRESSURIZATION

AND VENT LINES INSTRUMENT THIMBLE

FUEL SALT SYSTEM
FILL AND DRAIN LINE

SUPPORT RING

S T NN
BN NN N

FUEL SALT DRAIN TANK
BAYONET HEAT EXCHANGER
. THIMBLES (32) TANK FILL-LINE
- 32 3
- 2|
- 16\
8
0.
0
INCHEBS L

THIMBLE POSITIONING RINGS

FUEL SALT SYSTEM
FILL AND DRAIN LINE . TANK FILL LINE

Ffig. 5. Fuel-Salt Drain Tank.
34

supplied by TVA lines; however, about 300 kw can be provided by the diesel
electric emergency power supply. ,

The heaters used are generally of three types: (1) molded ceramic
units with embedded Nichrome elements, (2) Inconel-sheathed tubular
heaters with Nichrome elements and magnesium oxide electrical insulation,
and (3) resistance heating of the pipe wall by connecting the pipe across
the secondary taps of low-voltage, high-current transformers. Most heat-
ing units in the reactor and drain tank cells can be replaced, if re-
quired, by remote manipulation. Heating units in areas having negligible
radioactivity during shutdown are replaceable by direct contact mainte-
nance only. '

To minimize dusting, multiple-layer reflective-type thermal insula-
tion is generally used with remotely replaceable heating units. The per-
manent heaters are insulated with low-thermal-conductivity ceramic fiber
or expanded silica materials. '

At least one thermocouple is provided on the piping for each heater

unit. Anticipated 'cold spots" in the piping have additional thermo-
couples, -

Materials of Construction

The salt-containing piping and components are fabricated from INOR-8
— a special nickel-molybdenum alloy of the Hastelloy family, having good
resistance to corrosion attack by the fuel and coolant salts at tempera-
tures at least as high as 1500°F. The mechanical properties are superior
to those of the austenitic stainless steels, and the alloy is weldable
by established procedures. Most of the INOR-8 equipment was designed for
1300°F and 50 psig service, assuming an allowable stress of 2750 psi.

Stainless steel piping and valves are used in the helium supply and

off-gas systems. INOR-8, Inconel, and Monel are used in the fuel reproc-
essing system,

Major Auxiliary Systems

In addition to the fuel-salt and coolant-salt circulating systems,
three major auxiliary systems are required for operating the MSRE, These
are the cover- and off-gas systems, instrumentation and control systems,
and the fuel reprocessing system. These systems are discussed in detail
elsewhere and are only briefly described here.

Cover- and Off-Gas Systems

A helium cover-gas system protects the oxygen-sensitive fuel from
contact with air or moisture. Commercial helium is supplied from a tank
trailer, located outside the building, and is passed through a purifica-
tion system to reduce its oxygen and water content below .1l ppm before it
35

is admitted to the reactor systems. A flow of 200 ft?/day (STP) is passed
continuously through the fuel pump bowl to transport the fission product
gases to activated charcoal adsorber beds. The radioactive xenon is re-
tained on the charcoal for a minimum of 90 days, and the krypton is re-
tained for 7-1/2 days, which is sufficient for all but the 85Kr to decay
to insignificant levels. The 85Ky is maintained well within tolerance,
the effluent gas being diluted with 21,000 cfm of air, filtered, moni-
tored, and dispersed from a 3-ft-diam, 100-ft-high steel containment ven-
tilation stack.

The cover-gas system is also used to pressurize the drain tanks to
move molten salts into the fuel and coolant circulating systems. The gas
effluent from these operations is passed through charcoal beds and fil-
ters before it is discharged through the off-gas stack.

Instrumentation and Control Systems

Nuclear and process control are important to the operation of the
MSRE. The reactor has a negative temperature coefficient of 6.4 to 9.9 X
1075 (Ak/k)/°F, depending on the type of fuel that is being used. The
excess reactivity requirements are not expected to exceed 2 X 107° Ak/k
at the normal operating temperature. The three control rods have a com-
bined worth of 5.6 to 7.6 X 107% Ak/k, depending upon the fuel .composi-
tion. Their major functiong are to eliminate the wide temperature varia-
tions that would otherwise accompany changes in power and xenon poison
level and to make it possible to hold the reactor subcritical at a temper-
ature 200 to 300°F below the normal operating temperature. They have some
safety functions, most of which are concerned with the startup of the re-
actor. Rapid action is not required of the control rods; however, a mag-
netic clutch is provided in the drive train to permit the rods to drop
into the thimbles with an acceleration of 0.5 g as a convenient way of
providing insertion rates that are more rapid than the removal rates.
Burnup and growth of long-lived fission product poisons are compensated
by adding fuel through the sampler-enricher. Complete shutdown of the
reactor is accomplished by draining the fuel salt into the storage tanks
in the drain tank cell.

When the reactor is operated at power levels above a few hundred
kilowatts, the power is controlled by regulating the air flow, and thereby
the rate of heat removal, at the radiator. The power level is determined
by measuring the flow rate and temperature difference in the coolant-salt
system. The control rods operate to hold the fuel outlet temperature
from the reactor constant, and the inlet temperature is permitted to vary
with power level. At low power the control rods operate to hold the neu-
tron flux constant, and the heat withdrawal at the radiator or the input
to the heaters on piping and equipment is adjusted to keep the temperature
within a specified range.

Preventing the salts from freezing, except at freeze flanges and
valves, and protecting the equipment from overheating are among the most
important control functions. Temperatures are monitored in about 400
36

places in fuel and coolant systems, and the heating and cooling equipment
is controlled to maintain temperatures within specified ranges throughout
the systems.

Digital computer and data handling equipment are included in the in-
strumentation to provide rapid compilation and analysis of the process
data. This equipment has no control function, but gives current infor-
mation about all important wvariables and warns of abnormal conditions.

Fuel Processing System

Batches of fuel salfi or flush salt which have been removed from the
reactor circulating system can be processed in separate equipment to per-
mit their reuse or to recover the uranium.

Salts that have been contaminated with oxygen to the saturation point
(about 80 ppm of 0p), and thus tend to precipitate the fuel constituents
as oxides, can be treated with a hydrogen—hydrogen fluoride gas mixture
to remove the oxygen as water vapor. These salts can then be reused.

A salt batch unacceptably contaminated with fission products, or one
in which it is desirable to drastically change the uranium content, can
be treated with fluorine gas to separate the uranium from the carrier
salt by volatilization of UFg. In some instances the carrier salt will
be discarded; in others uranium of a different enrichment, thorium, or
other constituents will be added to give the desired composition.

The processing system consists of a salt storage and processing tank,
supply tanks for the Hs, HF, and F, treating gases, a high-temperature
(750°F) sodium fluoride adsorber for decontaminating the UFg, several
low-temperature portable adsorbers for UFg, a caustic scrubber, and as-
sociated piping and instrumentation. All except the UFg adsorbers are
located in the fuel processing cell below the operating floor.

After the uranium has been transferred to the UFg adsorbers, they

are transported to the ORNL Volatility Pilot Plant at X-10, where the UFg
is transferred to product cylinders for return to the diffusion plants.

General Plant Arrangement

The general arrangement of Building 7503 is shown in Fig. 6. The
main entrance is at the north end. Reactor equipment and major auxiliary
facilities occupy the west half of the building in the high-bay area.

The east half of the building contains the control room, offices, change
rooms, instrument and general maintenance shops, and storage areas. Ad-
ditional offices are provided in a separate building to the east of the
main building. ~
UNCLASSIFIED
CRNL-DWG 64-872EA

T

B S
T =
, l H BUILDING 7509
(OFFICES}
1 2 3 4 5 6 7 8 9
— INSTRUMENT
OFFICE OFFICE OFFICE OFFICE — STORE
— ROOM INSTRUMENT
— SHOP
=l il cHane :
‘J‘ ROOME OFFICE
© ‘ = NUCLEAR
| | INSTRUMENT
e N SHAFTS /<
MAIN
AUXILIARY
DATA ROOM O TR CONTROL ROOM
| EEEHOT e
HSAMPLER—
, CHANGE ROOM
I A A I”ENRICHER -
EQUIPMENT o AT~ r’;/’ / H
MOTE MAINT, - g I N I N
PRACTICE | CELL STORAGE CELL | SPARE CELL A LT SO 0 SPECIAL
% N —
) 0 —=='| EQUIPMENT
DRAIN TANK /1 \H ROOM
LL:
\ l i [b|=|=|=|=|=|==|:
/ P
- \\ < \ . Ay
' \ > \—\tv N 7‘}#
\ | R pd RADIATOR
/ \ Y L STACK
lINZ
1 / = = REACTOR L.
LIQUID WASTE CELL v
] REMOTE MAINT. CELL
| ] | CONTROL ROCM
DECONTAMINATION CELL/ k —‘ H \COOLANT CELL

FUEL PROCESSING CELL

Fig. 6. MSRE Plant Layout, Plan (Above 852-ft Level).

LE
38

Equipment for wventilation of the operating and experimental areas
is located south of the main building. A small cooling tower and small
buildings to house stores and the diesel-electric emergency power equip-
ment are located west of the main building.

The reactor primary system and the drain tank system are installed
in shielded, pressure-tight reactor and drain tank cells, which occupy
most of the south half of the high-bay area. These cells are connected
by an open 3-ft-diam duct and are thus both constructed to withstand the
same design pressure of 40 psig, with a leakage rate of less than 1 vol %
per day. A vapor-condensing system, buried in the ground south of the
building, is provided to keep the pressure below 40 psi during the maximum
credible accident by condensing the steam in vapors that are discharged
from the reactor cell. When the reactor is operating, the reactor and
drain tank cells are sealed, purged with nitrogen to obtain an atmosphere
that is less than 5% oxygen, and maintained at about 2 psi below atmos-
pheric pressure. :

The reactor cell is a carbon steel containment vessel 24 ft in di-
ameter and 33 ft in overall height. The top is flat and consists of two
layers of removable concrete plugs and beams for a total thickness of
7 ft. A thin stainless steel membrane is installed between the two layers
of plugs and welded to the wall of the steel vessel to provide a tight
seal during operation.

The reactor cell vessel is located within a 30-ft-diam steel tank.
The annular space is filled with a magnetite sand and water mixture, and
there is a minimum of 2 ft of concrete shielding around the outer tank.
In addition to this shielding, the reactor vessel is surrounded by a l4-
in.-thick steel and water thermal shield.

The drain tank cell adjoins the reactor cell on the north. It is a
17-1/2-ft by 21-1/2-ft by 28-ft-high rectangular tank, made of reinforced
concrete and lined with stainless steel. The roof structure, including
the membrane, is similar to that of the reactor cell.

The coolant cell abuts the reactor cell on the south. It is a
shielded area with controlled ventilation but is not sealed.

The blowers that supply cooling air to the radiator are installed
in an existing blower house along the west wall of the coolant cell.

Rooms containing auxiliary and service equipment, instrument trans-
mitters, and electrical equipment are located along the east wall of the
reactor, drain tank, and coolant cells, Ventilation of these rooms is
controlled, and some are provided with shielding.

The north half of the building contains several small shielded cells
in which the ventilation is controlled, but which are not gastight. These
cells are used for storing and processing the fuel, handling and storing
liquid wastes, and storing and decontaminating reactor equipment.
39

The high-bay area of the building over the cells mentioned above is
lined with metal, has all but the smaller openings sealed, and is provided
with air locks. Ventilation is controlled, and the area is normally op-
erated at slightly below atmospheric pressure., The effluent air from this
area and from all other controlled-ventilation areas is filtered and mon-
itored before it is discharged to the atmosphere. The containment wven-
tilation equipment consists of a filter pit, two fans, and a 100-ft-high
stack., They are located south of the main building and are connected to
it by a ventilation duct to the bottom of the reactor cell and another
along the east side of the high bay.

The vent house and charcoal beds for handling the gaseous fission
products from the reactor systems are near the southwest corner of the
main building. The carbon beds are installed in an existing pit that is
filled with water and covered with concrete slabs. The wvent house and
pit are also controlled-ventilation areas. Gases from the carbon beds
are discharged into the ventilation system upstream from the filter.

Maintenance of equipment in the fuel circulating and drain tank sys-
tems will be by removal of one or more of the concrete roof plugs and use
of remote handling and viewing equipment. A heavily shielded maintenance
control room with viewing windows is located above the operating floor
for operation of the cranes and other remotely controlled equipment. This
room will be used, primarily, when a large number of the roof plugs are
removed and a piece of highly radioactive equipment is to be transferred
to a storage cell,

Equipment in the coolant cell cannot be approached when the reactor
is operating, but since the induced activity in the coolant salts is
short-lived, the coolant cell can be entered for direct maintenance
shortly after reactor shutdown.

Description of Major Areas; Status of Construction
and MSRE Installation

General

A general view of Building 7503 is shown in Fig. 7. This area was
originally constructed for the Aircraft Reactor Experiment and later mod-
ified for the Aircraft Reactor Test, which was canceled in 1957. Some
accommodations in the MSRE design were necessary to fit the experiment
into the existing structures, and extensive modifications were required
in the sections of the building that house the reactor and fuel-salt cir-
culation system, the coolant-salt circulation system, and the fuel-salt
drain tanks. A shielded remote-maintenance control room was added. The
existing reactor containment wessel height was increased by about 8-1/2
ft, and the shielding walls, roof plugs, cell wall penetrations, supports,
and other structural features were extensively modified. Considerable
excavation was needed within the high bay to make room for the drain tank
cell., The penthouse for the coolant system and the radiator duct were
extensively modified.,
40

Fig. 7. DNorth View of Building 7503.

In modifying the existing buildings for the MSRE, the areas were
divided into five classifications:

Class I. These areas have high radiation levels at all times once
the reactor has operated at power and highly radioactive fuel or wastes
have been handled in the equipment. They include the reactor cell, drain
tank cell, fuel-processing cell, liquid waste cell, charcoal bed pit,
etc. The equipment in these areas must withstand relatively high radia-
tion levels and, in most cases, must be maintained by remote-maintenance
methods. Direct maintenance will be possible in the fuel-processing and
liquid-waste cells, but the equipment must first be decontaminated.

Class II. Areas in this classification are not accessible when fuel
salt is in the primary circulating system but can be entered within a

41

short time after the salt has been drained. The coolant-salt area, which
includes the radiator, coolant pump, and coolant-salt drain tank, is in
this category. The west tunnel and the unshielded areas of the blower
house are other examples. Iquipment in these areas can be repaired by
direct approach.

Class III. These are areas that are accessible during periods of
low-power operation of the reactor, such as the special equipment room
and. south electric service area, but cannot be entered if the power is
above 1 Mw. The equipment in these areas can be inspected and repaired
without draining the reactor.

Class IV. These are areas that are accessible or habitable at all
times, except under the conditions described below in Class V. These
areas include office spaces, control rooms, etc.

Class V., The maintenance control room will be the only habitable
area during maintenance operations when large, radioactive components are
being removed from the reactor cell. The rest of the MSRE site must be’
evacuated. This shielded room contains remote control units for the
cranes and TV cameras. |

Building

Above grade, Building 7503 is constructed of steel framing and as-
bestos cement corrugated siding with a sheet metal interior finish., Re-
inforced concrete is used in almost all cases below the 850-ft elevation.

Floor plans at the 852- and 840-ft levels are shown in Figs. 6 and
8. The general location of equipment is also shown in Fig. 6, and an
elevation view is shown in Fig. 9. |

The west half of the building above the 852-ft level is about 42 ft
wide, 157 ft long, and 33 ft high. This high, or crane, bay area houses
the reactor cell, drain tank cell, coolant-salt penthouse, and most of
“the auxiliary cells,

The eastern half of the building above the 852-ft elevation is 38
ft wide, 157 ft long, and about 12 ft high. Offices for operational per-
sonnel are located along the east wall of the north end. The main con-
trol room (Fig. 10), auxiliary control room, and a room used for the data
logger, computer, and by the shift supervisor on duty are located across
the hall on the western side. The large hall provides ample space for
an observation gallery. Windows behind the control panels enable the op-
erating personnel to view the top of the reactor cell and other operating
areas o0f the high bay. The change rooms are located near the center of
the building, The southeastern corner is used for an instrument shop,
instrument stores, and officés for instrument department supervisory per-
sonnel.,

Most of the western half of the building at the 840-ft level is oc-
cupied by the reactor cell, drain tank cell, and auxiliary cells. The
FUEL PUMP
LUBE OIL SYSTEM/

UNCLASSIFIED
ORNL-DOWG 63~ 4347A

BATTERY

ROOM MAINTENANCE

SHOP

I

QFFICE

CHEMICAL [
LABORATORY [}

INDUCTION

EQ
MAINTENANCE :
PRACTICE STCOERLALGE
CELL

REGULATOR

CELL

LIQUID WASTE
CELL

INSTRUMENT
TUBE

ROOM

SPECIAL /
EQUIP 4

COOLANT PUMP
LUBE OIL SYSTEM

TO FILTERS
AND STACK

CELL VENTILATION
DUCT AND BLOCK VALVE

TO VAPOR

o

COND. SYSTEM
COOLANT PUMP

SALT DRAIN
TANK

DECONTAMINATION / A N (
CELL FUEL SALT TANK NO.2. THEAMAL
FUEL SALT/  TANK FUEL |
o FLUSH TANK
STAT!ON FUEL
PROCESSING FUEL DRAIN
CELL TANK NO. 1
REACTOR
“ELEC. SERVICE AREA BELOW CELL ANNULUS
BLOWER ~
HOUSE

RADIATOR

DUCT BLOWERS

[~~RADIATOR
BLOWERS

Fig. 8. MSRE Plant Layout at 840-ft Level.

cy
UNGLASSIFIED
ORNL-DWG 64-5974

f\@
ik 2 3 4 5 6 7 8 9
| T
:T$E;?;§%:;r~3o—mn CRANE ;:iIIIIZIIES”H3 AND 10-ton CRANES
i STACK
@ COOLANT SALT
=N PUMP
|
| | MAINTENANCE ‘: | SHIELD
| CONTROL ' BLOCKS
DECONTAMINATION REACTOR |, ROOM ] _
CELL e e :
CELL FUEL SALT NS | 7 |
SHIELD BLOCKS ADDITION SHIELD - FUEL PUMP =
STATION BLOCKS,

| 1 I Il ]
| i i i 1L

KDL AR K AP
KDL

——RADIATOR

5
] LiouID WASTE
: CELL

T

DRAIN TANK CELL |

RADIATOR
BYPASS DUCT

FUEL
STORAGE TANK

COOLANT SALT
DRAIN TANK

Jei
FUEL CRALI V

FUEL DRAIN REACTOR
TANK NO.1 FLUSH LINE VESSEL
TANK ' .
FUEL DRAIN THERMAL
TANK NO. 2 SHIELD

Fig. 9. MSRE Plant Layout, Elevation.

134
ncLASSIFIED
PIGTO 66394

Fig. 10. MSRE Main Control Room.

emergency nitrogen cylinder station is located against the west wall in
the northwest corner of this level. Switch boxes used in the heater cir-
cuits are located across the aisle between columns A and C (see Fig. 6).
Behind these switch boxes is an 8- by 40-1/2-ft pit with a floor elevation
of 831-1/2 ft. This pit contains heater circuit induction regulators with
some heater transformers mounted above. It is accessible from the 840-ft
level by stairs located at the east end. Additional induction regulators
and rheostats are located at column line C between columns 2 and 3. The
heater control panels and thermocouple scanner panels are installed along
column line C between columns 3 and 4.

The batteries for the 48- and 250-v dc emergency power supply are in
an 18- by 18-ft battery room in the northeast corner. The motor gener-
ators and control panels for the 48-v system are in an area west of the
battery room. The main valves and controls for the fire protection sprin-
kler system are installed along the north wall.

A maintenance shop area is provided between columns 2 and 4. The
process-water backflow preventer is installed on the east wall between
columns 2 and 3.

45

The area between column lines 4 and 6 houses the main lighting break-
ers, switch boxes and transformers, the intercom control panel, water
heater, and air conditioners for the main control room and transmitter
room, as well as a lunchroom and meeting room for maintenance personnel.
The transmitter room is located between column lines 5 and 6.

The service room, 16 by 27 ft, located at the northeast corner of
the 840-ft level, serves as a small chemistry laboratory and as an access
to the service tunnel. The instrument panels for the fuel and coolant
lube o0il systems are located in this area.

Reactor Cell. The reactor cell, shown in Fig. 9, is a cylindrical
carbon steel vessel 24 £t in diameter and 33 ft in overall height (ex-
tending from the 819- to 852-ft elevation), with a hemispherical bottom
and a flat top. The lower 24-1/2 ft (819- to 843-1/2-ft elevation) was
built for the ART in 1956, It was designed for 195 psig at 565°F and
was tested hydrostatically at 300 psig. The hemispherical bottom is 1
to 1-1/4 in. thick. The cylindrical portion is 2 in. thick except for
the section that contains the large penetrations, where it 1s 4 in. thick.

This vessel was modified for the MSRE in 1962 by lengthening the .
cylindrical section 8-1/2 ft (843-1/2- to 852-ft elevation). Several new
penetrations were installed, and a 12-in. section of 8-in., sched &0 pipe,
closed by a pipe cap, was welded into the bottom of the vessel to form a
sump. The extension to the vessel was 2 in. thick, except for the top
section, which was made as a 7-1/4- by 14-in. flange for bolting the top
shield beams in place. The flange and top shield structure were designed
for 40 psig and the completed vessel was tested hydrostatically to 48
psig, measured at the top of the cell., Both the original vessel and the
extension were made of ASTM A0l grade B fire-box-quality steel. Material
for the extension was purchased to specification ASTM A300 to obtain steel
with good impact properties at low temperature. Steel for the original
vessel was purchased to specification ASTM A201, :

All the welds on the reactor cell vessel were inspected by magnetic
particle methods if they were in carbon steel, or by liquid penetrant
methods if they were in stainless steel. All butt welds and penetration
welds were radiographed. After all the welding was completed, the vessel
was stress relieved by heating to 1150 to 1200°F for 7-1/2 hr,

Calculated stresses in the vessel were well below those permitted
by ASME Boiler and Pressure Vessel Code Case 1272N-3. Allowable stresses
for AST™M A201 grade B steel were taken as 16,500 psi for general membrane
stresses, 24,750 for general membrane plus general bending plus local mem-
brane stresses, and 45,000 psi for combined primary and secondary stresses.,

Two layers of 3-1/2-ft-thick shield blocks for the reactor cell are
supported by a channel formed into a ring and welded to the inside of the
cell wall. The cavity in the channel is filled with steel shot to provide
shielding for the cracks between edges of the blocks and the cell wall.

- One-half-inch steel plate stiffeners are installed at 9° intervals. The
top of the channel is at the 848-1/2-ft elevation.
46

Two beams provide the rest of the support for the bottom layer of
blocks. These beams were built of 36-in. WF 150-1b I-beams with angle
iron and steel plate stiffeners. The cavities were filled with concrete
for shielding. The beams rest on a built-up support plate assembly, which
is welded to the side of the cell at the 847-ft 7-in. elevation.

Offsets, 6-1/4 by 26 in., are provided in the ends of the bottom
blocks to fit over the support beams. Guides formed by angle iron assure
proper alignment. Several of the bottom blocks have stepped plugs for
access to selected parts of the cell for remote maintenance. '

The sides of the blocks are recessed 1/2 in., for 14 in. down from
the top. With blocks set side by side and with a l/2—in. gap between, a
1—1/2-in. slot is formed at the top. One-inch-thick steel plate 12 in.
high is placed in the slots for shielding. The 1l-gage ASTM-A240 type
304 stainless steel membrane is placed on top of the bottom layer of
blocks and is seal-welded to the sides of the cell., Cover plates are
provided over each access plug., These are bolted to the membrane and are
sealed by neoprene O-rings. A l/8-in. layer of Masonite is placed on top
of the membrane to protect it from damage by the top layer of blocks.

The top blocks are beams that reach from one side of the cell to the
other., The ends of these blocks are bolted to the top ring of the cell
with fifty 2-1/2-in. No. 4 NC-2 studs, 57-1/4 in. long, made from ASTM-
A320 grade L7 bolting steel. These studs pass through holes that were
formed by casting 3-in. sched 40 pipes in the ends of the blocks. Cold-
rolled steel washers, 9 in. long by 9 in. wide by 1 in. thick, and stand-
ard 2-1/2-in. No. 4 NC-2 nuts are used on each stud.

A removable structural steel platform (elevation 823-1/2 ft at top)
forms a floor in the reactor cell vessel and a base for supporting major
pieces of equipment.

The reactor cell wvessel is installed in another cylindrical steel
tank that is referred to here as the shield tank. This tank is 30 ft in
diameter and 35-1/2 ft high (elevation 816-1/2 to 852 ft). The flat
bottom is 3/4 in. thick, and the cylindrical section is 3/8 in. thick.
The shield tank sits on a reinforced concrete foundation that is 34-1/2
ft in diameter by 2-1/2 ft thick. .The reactor wvessel cell is centered
in the shield tank and supported by a 15-ft-diam, 5-ft-high cylindrical
skirt made of* 1-in,-thick steel plate, reinforced by appropriate rings
and stiffeners. The skirt is joined to the hemispherical bottom of the
reactor cell in a manner that provides for some flexibility and differen-
tial expansion and is anchored to the concrete foundation with eighteen
2-in,-diam bolts,

From elevation 816-1/2 to 846 ft, the annulus between the shield
tank. and the reactor cell vessel and skirt is filled with magnetite sand
and water for shielding. The water contains about 200 ppm of a chromate
rust inhibitor, Nalco-360, A 4-in,-diam overflow line to the coolant cell
controls the water level in the annulus.
47

The region beneath the reactor cell vessel inside the skirt contains
only water, and steam will be produced there if a large quantity of salt
is spilled into the bottom of the reactor cell. An 8-in.-diam vent pipe
is provided to permit the steam to escape at low pressure. This pipe con-
nects into the skirt at the junction with the reactor cell and extends
to elevation 846 ft, where it passes through the wall of the shield tank
and terminates as an open pipe in the coolant cell.

From elevation 846 to 852 ft, the annulus between the reactor cell
and. shield tank is filled with a ring of magnetite concrete. The concrete
ring improves the shielding at the operating floor level and provides some
stiffening for the top of the reactor cell wvessel, The concrete ring is
supported from the wall of the shield tank, and the reactor cell wall is
free to move through the ring and to expand and contract relative to the
shield tank.

Numerous penetrations were installed through the walls of the reactor
cell and shield tank to provide for process and service piping, for elec-
trical and instrument leads, and for other accesses. The penetrations
are 4- to 36-in.-diam pipe sleeves welded into the walls of the reactor
cell and the shield tank., Since the reactor cell will be near 150°F when
the reactor is operating and the temperature of the shield tank may at
times be as low as 60°F, bellows were incorporated in most of the sleeves
to permit radial and axial movement of one tank relative to the other
without producing excessive stress. The bellows are covered with partial
sleeves to prevent the sand from packing tightly around them,

Several other lines are installed directly in the penetrations with
welded seals at one or both ends, or they are grouped in plugs which are
filled with concrete and inserted in the penetrations. The major openings
are the 36-in.~-diam neutron instrument tube and drain tank interconnection
and the 30-in.-diam duct for ventilating the cell when maintenance is in
progress. 'The original tank contained several other 8- and 24-in.-diam
penetrations, and they were either removed or closed and filled with
shielding.

A1l the components and connecting piping, which were first fitted
together on a precision assembly jig outside the cell, have been installed
in the cell, as shown in Fig. 11. All thermocouple and heater leads have
been installed and connected to the thermocouple and heater remote dis-
connects. All leak detector tubing, helium, air, and water piping, to-
gether with associated valving, hawve been installed. No additional in-
stallation work is required for prenuclear testing with flush salt. Before
the addition of fuel salt and meking the MSRE critical, the sampling and
enriching system .and the control rods and rod drives must be installed.

Drain Tank Cell., The drain tank cell shown in Fig. 6 is 17 £t 7 in,
by 21 £t 2-1/2 in., with the corners beveled at 45° angles for 2-1/2 ft,
The flat floor is at the 8l4-ft elevation, and the stainless steel mem-
brane between the two layers of top blocks is at the 838-ft 6-in, eleva-
tion, The open pit extends to the 852-ft elevation.

48

The cell was designed for 40 psig, and when completed in 1962 it

was hydrostatically tested at 48 psig (measured at the elevation of the
membrane, 838-1/2 ft).

The bottom and sides have a 3/16-in.-thick stainless steel liner,
backed up by heavily reinforced concrete, magnetite concrete being used
where required for biological shielding. The liner is welded to an angle-
iron grid work at approximately 8-in. spacing, with 1/2-in. plug welds.
The angle irons are welded to reinforcement rods embedded in the concrete.

Vertical columns in the north and south walls are welded to hori-
zontal beams embedded in the concrete of the cell floor. The tops of the
columns are welded to horizontal 36-in. WF 160 I-beams running east and
west. These have stiffeners welded to them on approximately 6-in. centers.
A slot is provided at the top of these I-beams at elevation 842 ft 1 in.
to 842 £t 4-3/8 in. by welding a 1-1/4-in. plate to the web of the beam.
BEighty-two 3-1/4- by 4-1/2- by 10-in. steel keys are wedged into this
slot to hold down the top blocks.

The top of the cell is composed of two layers of reinforced concrete
blocks with an 11-gage (A-204 type 304 stainless steel) membrane between

NCLAsSIFIED

Fig. 11. Fuel-Salt Circulation System in Reactor Cell.

49

them. Both layers of blocks are ordinary concrete (density 150 1b/ft?).
The bottom layer is 4 ft thick and the top layer is 3-1/2 ft thick.

The blocks are arranged to facilitate remote maintenance. One side
of the lower blocks, which run east and west, is supported by a ledge at
an elevation of 834- 1/2 ft. The other side of these and the ends of the
north-south blocks are supported by beams that extend from the east to
the west side of the cell, These beams are built up of 24-in. 104.9-1b
I-beams with angle-iron and steel-plate stiffeners. The cavities are
poured full of concrete for shielding. These rest on a bullt-up support
plate assembly, which is welded to the side of the stainless steel liner
and is anchored into the concrete walls., Offsets, 4- 1/4 by 25- 1/2 in.,
are provided in the ends of the bottom blocks to fit over the support
beams. V grooves formed by angle irons assure proper alignment. The
sides of the bottom blocks are recessed 1/2 in. for 14 in. down from the
top. With two blocks side by side and with a l/2-in. gap between the
sides at the bottom, a 1-1/2-in. slot is formed at the top. One-inch
steel plates, 12 in. high, are put into the slots to provide biological
shielding. :

The 11-gage (ASTM A-240 type 304 stainless steel) sealing membrane
is placed on top of the bottom blocks (elevation 838 £t 6 in.) and is
seal-welded to the sides of the cell. A 1/8-in. layer of Masonite is
placed on top of the seal pan to protect it from damage during installa-
tion of the top blocks. The top blocks are beams that reach from the
north side to the south side of the cell. They are held down by the
eighty-two 3-1/4- by 4-1/2- by 10-in. steel keys discussed previously.
These are inserted into slots in the top beams of the north and south
sides of the cell and are driven in approximately 4 in., leaving approxi-
mately 5 in. bearing on the ends of the blocks. There are approximately
four keys for each end of each block. A 1- 1/2 -in. 6 NC square nut is
welded to the top of each key to aid in removing and handling the Keys.

The floor elevation is 814 ft at the highest point along the west
wall and slopes 1/8 in. /ft to a trench along the east wall., The trench
slopes 1/8 1n./ft toward the south and terminates in a sump in the south-
east corner. The sump consists of a 10-in.-long section of 4-1n. sched
40 pipe and a 4-in. butt-welded cep (347 stainless steel).

Numerous lines are installed through the walls of the drain tank
cell to provide for process and service piping, electrical and instrument
leads, and for other accesses. These enter through 3/4- to 6-in. pipe
or pipe sleeve penetrations that are welded to the stainless steel liner
and cast into the concrete walls. ITines are installed in these individ-
ually or grouped in plugs that are filled with concrete and inserted.

Figure 12 shows the salt storage vessels and associated equipment,
installed in the drain tank cell, All tanks, tank furnaces, steam cooling
equipment, and associated piping have been installed in the drain tank
cell. Ieak testing and other required check-out tests have been conducted
on installed equipment. Except for the installation and check-out of
some remotely removable pipe heater Insulation units, this cell is ready
to accept the flush salt. )
50

UNCLASSIFIED
PHOTO 65477

Fig. 12. Drain Tank Cell Equipment Installed.

n

51

Coolant Cell and Coolant Drain Tank Cell. Except for two lines in
the reactor cell which connect the coolant-salt circulating system to the
heat exchanger, all coolant-salt piping is located in the coolant and
coolant drain tank cell, shown in Fig. 6. The top section of the coolant
cell is a concrete enclosure (penthouse) which rises from the southwest
corner of the high bay to an elevation of 863 ft 3 in. The walls are
ordinary reinforced concrete 24 in. thick except for the south corner,
which is 15-1/2 in. thick. Additional shielding for this corner is pro-
vided by the 2-ft-thick base for the stack, which goes to the 859-ft ele-
vation. A 4-in. ledge at elevation 861 ft 9 in. on the southeast and
northeast sides of the penthouse supports the bottom layer of the top
shield blocks. These are constructed of 12-in.-thick reinforced ordinary
concrete. The 12-in.-thick top shielding blocks rest on the bottom blocks.
The l/2-in. gaps between plugs in each layer are staggered for better
shielding. Iedges at the 862-ft 9-in. elevation on the northeast side
and at the 86l-ft 9 in. elevation on the other sides are prov1ded for
shielding. -

The bottom of the coolant radiator is at elevation 836 ft 1-3/4 in.
The floor of the coolant stack enclosure, located south of the radiator,
is at elevation 835 ft 5 in., and the 2-ft-thick reinforced regular con-
crete walls extend to the 859-ft elevation. : .

Al- l/4—1n. steel grating and/or 11-gage sheet metal pan at elevation
834 ft 11 in. separates the coolant cell from the coolant drain tank cell.
Thus, the coolant drain tank cell is below ground level on,all sides. The
elevation of the concrete floor is 820 ft. It is bounded by the reactor
cell annulus on the.northeast and by the special equlpment room on the ‘
east, .

Two penetrations from the reactor cell terminate in the coolant cell
or coolant drain tank cell. The wall between the coolant and coolant
drain tank cells and the special equipment room is 16 in. thick to the
835-ft 5-in. elevation and 12 in. thick above this. Existing openings
between the cells are closed by stacked magnetite concrete blocks to en-
able proper ventilation of both areas and to provide adequate shielding.
A 6-ft-wide concrete ramp with 2-ft stairs in the center (slope 7 in 12)
extends from the northwest corner of the coolant drain tank cell to the
blower house at the 840-ft elevation. This is closed off by stacked mag-
netite blocks. ' : e i ‘

Shielding from the coolant pump and from piping is provided by 8 ft
of barytes blocks stacked above the 835-ft elevation between the radiator
housing and reactor shield.

A1]. equipment for the coolant-salt system has been installed. Figure
13 shows the coolant pump and the radiator door-lift mechanism. The in-
stalled radiator is shown in Fig. 14, and the main blowers are shown in
Fig. 15. The installed coolant drain tank equipment, located at the level
below the radiator, is shown in Fig., 16. This system has been checked
out and is ready to receive the coolant salt.
UNCLASSIFIED
PHOTO 66397

Fig. 13. Coolant Pump and Radiator Door-Lift Mechanism.

UNCLASSIFIED
PHGTO 66404

Fig. l4. MSRE Radiator, as Installed.

uncLAssiFiED
PUGTO 6690

Fig. 15. Main Blowers for MSRE Coolant System. |

Fig. 16.

54

Coolant Drain Tank.

UNCLASSIFIED
PHOTO 66401

55

Special Equipment Room. The major items of equipment located in the
special equipment room are the component coolant blowers, the containment
boxes for the fuel pump cover gas and bubbler lines, the 30-in.-diam cell
ventilation duct, and the rupture disks and lines to the vapor-condensing
system. This cell is located southeast of the reactor cell and, as dis-
cussed previously, has a common wall with the coolant cell and coolant
drain tank cell. The other walls are 12-in.-thick reinforced concrete
below ground level. The cell is 15 ft 11 in, by 17 ft. The floor eleva-
tion is 828 ft 7 in., except for a 5-ft by 15-ft 1l-in. by 3-ft-deep pit
which runs east and west, 2 ft 3 in. from the south wall.

The top consists of two rows of removable concrete shield plugs with
staggered joints. Each row of blocks is 1 ft thick. The bottom row is
supported by a 4-in. ledge at the 850-ft elevation. Containment is pro-
vided by ventilation from the stack fans.

Figure 17 shows the equipment in the special equipment room as in-
stalled, checked out, and ready for operation. Only the lines to the
vapor-condensing system are yet to be installed.

Pump Room. The pit pump and sump pumps are located in the pump room.
This room is under the north end of the special equipment room. It is
’7-1/2 ft wide by 15 ft 3 in. long by 6 ft 7 in. high with a floor eleva-
tion of 820 ft. A 3- by 3-ft sump, near center of the north wall, extends
to the 811-ft elevation. Access to the pump room is through a 3- by 4-ft
hatch near column C-8. This equipment is installed.

Service Tunnel. The fuel and coolant lube-oil systems shown in Fig.
18 and the reactor cell ventilation block valves are installed in the

UNELASSIFIED
PHGT0 435

Fig. 17. Equipment Installed in Special Equipment Room.

56 o

service tunnel. This tunnel is 7 ft wide by 11 ft high by approximately

67 ft long. It passes under the southeast corner of the 7503 building -
and extends south and west outside the building. The floor elevation is

833 ft 3 in. Normal access is from the service room on the 840-ft level;

however, a 3- by 3-ft hatch near the west end provides access from outside

the building. This hatch is normally closed by a sheet-metal-covered

wooden lid. Containment is controlled by ventilation.

UNCLASSIFIED .
PHOTO 66400

Fig. 18, Typical Lube-Oil System for Fuel and Coolant Pumps. .

57

Transmitter Room and Electric Service Areas. The area north and west
of the reactor cell annulus at elevation 831 ft (west tunnel) connects
by a narrow passage to a similar area north and east of the reactor cell
annulus (east tunnel). These extend eastward under the 840-ft floor to
a 24-in,~-thick concrete wall 12 ft north of column line 6. This entire
area is called the south electric service area. Six penetrations from
the reactor cell and eighteen from the drain tank cell terminate in this
area, The walls, except for those joining the reactor or drain tank
cells, are located below ground level. The ceiling, which connects with
the transmitter room, is 24-in.-thick reinforced concrete with a 3-ft
7-1/2-in. by 6-ft 7-1/2-in. by 24-in.-thick concrete plug for access to |
the area. This plug has a 4- by 1l2-in. offset on all sides for shielding.

The north electric service area is located north of the 24-in.-thick
concrete wall mentioned above. The floor of this room is at the 824-ft
elevation. Many of the drain tank cell penetrations terminate in this
area as well as some from the spare cell. The ceiling of this area and
the floor of the transmitter room above is 2- to 4-in.-thick reinforced
concrete poured on "Q" decking No. 3-16. Containment is by controlled
ventilation., A 2- by 3-ft manhole with a hinged steel door prov1des ac-
cess from the transmitter room.

The floor of the transmitter room is at the 840-ft elevation. The
5-in. floor of the high bay (elevation, 852 ft) is the ceiling of the
‘transmitter room. A 5-ft by 9-ft 3-1/2-in. plug can be removed from the
high-bay floor so that the high-bay crane can be used to remove the shield-
ing plug on the 840-ft level between the transmitter room and the south
electric service area., This room is approximately 16 by 20 ft and con-
taing the leak detector system and instrumentation for the drain tank
weigh cells, pump level and speed, sump level, component cooling air, etc.

Auxiliary Cells. Six smaller cells are located between columns 2-5
and A-C. Removable concrete blocks provide access from the high-bay area
at the 852-ft elevation. Containment of each is by control of ventilation
and discharge of the air through the filters to the stack. Descriptions
of their projected uses are given below. |

Fuel Processing Cell — The fuel storage tank and other in-cell equip-
ment for hydrofluorinating or fluorinating the fuel or flush salt are lo-
cated in this cell. The absorber cubicle is located east of this cell
at the 852-ft level.

Decontamination Cell — This cell contains a decontamination tank.
Both tank and cell are used for underwater maintenance and storage of con-
taminated equipment,

Liquid Waste Cell — The liquid waste tank, sand fllter, and cell sump
jet block Valves are located in this cell.,

" Remote Maintenance Cell — The llquld waste pnmp and waste tank ex-
haust blower are located in this cell. It will also be used asgs a remote
maintenance practice cell.
58

Hot Storage Cell — This cell will be used to store contaminated
equipment removed from the reactor or drain tank cell.

Spare Cell — The absolute filter in the chemical processing cell air
exhaust line and the flame arrester in the chemical processing off-gas
line are located in this cell.

High-Bay Containment Enclosure. The high-bay containment enclosure
between columns 2-8 and A-C is 42 by 136 ft with a ceiling elevation of
885 ft 4 in. The containment extends east near column 8 to include the
neutron instrument tube and to provide room for the fuel-salt sampler.

A false ceiling and contaimment seal is installed 15 ft above the
floor between columns 8-9 and B-C. The containment walls extend from
this elevation to 885 ft 4 in. along column line C from columns 8 to 9,
west along column line 9 to column A, then north along column A to column
8. Thus the coolant cell penthouse is included in the high-bay area and
the high-bay crane can be used for handling equipment and shielding blocks
in this cell., A 10- by 12-ft loading hatch and Bilco door in the false
ceiling of the area between columns 8-9 and B-C is used for moving small
equipment into or out of the containment enclosure and for removal of
fuel- and coolant-salt samples. A 12~ by 12-ft door located at column
line 2 is used for removal of larger equipment. The framework of the high
bay is Stran Steel construction with a 1l6-gage iron sheet metal skin skip-
welded to it. The cracks were sealed with fiberglas tape and Carboline
paint.

Emergency exits are provided at the southeast corner and near the
center of the east wall. Since this area is considered as a contaminated
zone, normal entrance and exit is through the hot change house, located
east of the high bay, and through the regular change room,

Maintenance Control Room. A 19-ft 8-in. by l4-ft 9-in., maintenance
control room is located along the west side of the high-bay containment
between the reactor and drain tank cells. The floor elevation is 862 ft
and the ceiling 870-1/2 ft. The west side is corrugated asbestos to match
the building. The roof is 12-in,-thick concrete, the south side 2-ft-
thick concrete, and the north and south sides 3- to 3-1/2-ft-thick con-
crete for shielding. A zinc bromide—filled viewing window is located on
the west side near the south end and another on the northwest corner of
the room for use during remote maintenance. Access to the maintenance
control room is via stairs which terminate at ground level west of the
building.

A 6-ft by 19-ft 8-in. electric equipment room located below the main-
tenance room has a floor elevation of 852 ft. This is accessible through
a hatch in the southeast corner of the maintenance control room, :

Off-Gas Area

The off-gas area, consisting of a vent house and absorber pit, is
shown in Fig. 8.
59

The vent house is 12 by 15 £t with an operating floor level of 848
ft. The sloping roof attaches to the south end of the 7503 building at
the 857-1/2-ft elevation.

A 5- by 9~ by 3-ft-deep contaimment enclosure is located along the
east side of the wvent house. This contains the reactor and drain tank
off-gas lines, instruments, valves, etc. The bottom of this enclosure
is at the 839-1/2—ft elevation. The east and west sides are reinforced
concrete; the north and south ends and top at the 842-1/2-ft elevation
are constructed of mild steel angle irons and 1/8-in. plate (ASTM A-7).
The Jjoints are caulked to prevent leakage. Hand valves have shielded
extension handles which permit them to be operated from the vent-house
floor. Pipe caps are installed for containment when the valves are not
being operated. Five feet of barytes blocks are stacked on top of the
containment enclosure. A 17-in,-diam removable plug is installed above
the control wvalves and check valves to facilitate remote maintenance.

The west side of the vent house has a floor elevation of 844-1/2 ft
and is used for off-gas lines not requiring shielding and for stacked
block shielding for the reactor and drain tank off-gas lines.

A 5~ by 15-ft valve pit with the bottom at an elevation of 841 ft
is located south of the wvent house. The off-gas lines pass through this -
pit before entering the absorber beds, shown after fabrication in Fig.
19. The outlet lines are also located in this pit. The inlet valves are
contained in a 2-1/2- by 4- by 2-ft-high containment box which has a top
elevation of 845-3/4 ft. Seventeen inches of steel plate is used above
this box for shielding. Shielded extension handles permit operation of
- the valves from the operating floor at an elevation of 848 ft. Pipe caps
are installed for containment when the valves are not being operated. The
liners between the charcoal beds and the pit go through a 9- by 18-in.
penetration, where they are grouted in place using Embeco expansion grout.

The charcoal bed pit has a 10-ft inside diameter with l4-in.-thick
concrete walls. The bottom of the pit is at an elevation of 823 ft 9 in.
The pit ie filled with water to an elevation of 846 ft for shielding.

Two 10-1/2-ft-diam by 18-in.-thick barytes concrete blocks are supported
by a ledge at the 846-1/2-ft elevation. The top block is caulked to ob-
tain an airtight seal. Additional barytes concrete blocks are stacked

on top of these to give a minimum of 5-1/2 ft of shielding. A 4-ft an-
nulus around the outside of the charcoal bed pit is filled with stabilized
aggregate from the 848-ft to the 849-ft 5-in. elevation.

‘ The off-gas from the charcoal bed goes through the valve pit men-
tioned previously, where unshielded extension handles permit operation
from the 848-ft elevation,

Installation of the charcoal édsorbers and assoclated valving and
piping is nearing completion,
60

Fig. 19. Charcoal Adsorber Beds for MSRE Off-Gas System.

Stack Area

The stack area is located south of Building 7503. All off-gas and
containment ventilation air passes through roughing and absolute filters
before being discharged up the stack.

These filters are located in a pit 60 ft south of Building 7503.
The pit is 26-1/2 by 18—3/4 ft with a sloping floor. The floor elevation
at the north end is 850 ft and the south end is 848 ft. The walls are
1-ft-thick concrete. The roof plugs are 18-in.-thick concrete with a
3-in. offset at the center. This offset rests on a ledge 9 in. down from
the top of the pit. The elevation of the top of the pit is 857 ft. The
roof blocks are caulked to prevent leakage.

A 3-ft 3-in. by 5-ft 6-in. valve pit is attached to the filter pit
on the southeast corner. This houses the filter pit drain valves and
water level indicators. The walls and removable roof plug are 8-in,-thick
concrete, The floor and roof are at 845- and 857-ft elevations.

The two 21,000-cfm stack fans and the associated 3-ft-diam by 100-
ft-high steel stack are located south of the filter pit.

A11 construction and equipment installations associated with the
stack, filter house, and ventilation ducts are completed.

61

Vapor-Condensing Tanks

The vapor-condensing system water tank, expansion tank, and associ-
ated piping are located west of the stack area. As mentioned previously,
the rupture disks and connections to the reactor cell exhaust duct are
located in the special equipment room. The tanks and piping for this
system are nearing completion, but are yet to be installed.

Blower House

The blower house is 36 by 43 ft with floor and ceiling elevations

- of 838 and 856 ft respectively. The north, south, and west walls are
louvered to provide air inlet to the two coolant blowers and two annulus
blowers located in the south half of the room. The air from the blowers
passes through the radiator area of the coolant cell and out the 10-ft-
diam by 70-ft-high steel stack located just north of the vent house. Top
elevation of this stack is 930 ft. .

Gas coolant pump No. 3, which supplies air to freeze valves in the
coolant drain tank cell and fuel processing cell, is located between
column lines 7 and 8.

A ramp leading to the coolant drain tank cell begins at the north-
west corner of the blower house.

The cooling water equipment room, housing the cooling-water control
panel, storage tanks, flowmeters, and treated-water circulation pumps,
is located in the southwest corner of the blower house.

The stairs leading to the maintenance control room are located just
west of the blower house. The area under the stairs is used for the flu-
orine gas trailer, hydrogen and hydrogen fluoride gas cylinders, etc.,
which are used in processing the fuel salt.

All equipment in this area has hbeen installed.

Diesel House

The 30- by 72-ft diesel house is located 36 ft west of Building 7503
between column lines 3 and 5 and adjoins the switch house on the east.
It has corrugated asbestos siding and a pitched roof with an elevation
of 858 ft at the center and 852 £t at the sides.

The three diesel generator units, Fig. 20, and their auxiliaries are
located at the east end of the north half of the building. . The 5000-gal
diesel fuel storage tank is located approximately 100 ft north of the
diesel house.

The helium cover-gas treatment equipment, Fig. 21, and emergency
helium supply cylinders are located just west of the diesels. The helium
supply trailer is located outside the west end of the building.
Fig. 20, Diesel Generator Installation.

Fig. 21. Helium Cover-Gas Treatment Equipment and Instrument Air
Compressor.

63

The tower-water-to-treated-water heat exchanger is located along the
west end inside the building.

The instrument air compressors and auxiliary air compressor are along
the south side (see Fig. 21).

Installation of equipment in this area is completed.

Switch House and Motor Generator House

The switch house, which is 30 by 36 ft, adjoins the diesel house on
the west and Building 7503 on the east. It has corrugated asbestos siding
with a flat roof at the 852-ft elevation. The floor elevation is 840 ft.

The main switchgear for the 480-v feeder lines and the diesel gen-
erator controls are located in thig area.

A 15- by 16-ft motor generator room, located north of the east end
of the switch house and accessible from the switch house, contains two
motor generators and their control panels. The process power substation
is located west of the motor generator room, -

Electrical installation in this area is completed.

Inlet Air Filter House

The high-bay inlet air filter and steam heating units are located
in the inlet air filter house. This 12- by 20-ft concrete block building
is located Jjust west of the 7503 building. The floor elevation is 840 ft
6 in., and the flat roof is at an elevation of 850 ft 11 in.

Procurement and Fabrication of Major Components

When project status was attained for the MSRE, it was thought de-
sirable to have the major components for the salt circulating systems
made by industrial fabricators who were experienced in proddcing reactor
components such as nuclear code vessels, heat exchangers, reactor con-
tainment vessels, etc. Engineering drawings and specifications, welding
qualification procedures, and inspection procedures were prepared which
were consistent with this approach. '

Requests for bids on the reactor wvessel, radiator, heat exchanger,
and the several salt storage tanks were sent out to several companies
considered to have the greatest amount of experience in nuclear component
fabrication. After these companies had reviewed the drawings and spec-
ifications, a prebid conference was held at which representatives of these
companies were briefed on the properties and fabricability of INOR-8 and
interpretations of the specifications were made, if unclear to any pro-
spective bidder.
64 -

Since INOR-8 was a new alloy, and since considerable precision ma-
chining and flaw-free welding were required, all companies invited de- .
clined to submit lump-sum bids for producing the components. Further
negotiations disclosed that most felt too many uncertainties existed to
justify lump-sum bids, although most had considerable experience with
other alloys of the Hastelloy family. Some of the companies expressed
an interest in producing the components under a cost-plus-fixed-fee con-
tract. It was decided, however, to fabricate the components in Union
Carbide Corporation shops, since craftsmen in these shops had several
vears' experience in INOR-8 fabrication and the shops were well equipped
for fabrication of units of this size. The major portion of the INOR-8
component fabrication was done in the Y-12 shops.

Special Materials Procurement

A1l contracts for special materials were fixed price and ordered on : .
the basis of competitive bids.

INOR-8., Most of the INOR-8 plate, bar stock, and tube rounds were .
produced. by the Stellite Division of Union Carbide Corporation. Some of
the ingots were produced by the International Nickel Company and Alvac
Metals, The extrusion of seamless pipe and tubing was done by Michigan
Seamless Tube Company, Wall Tube Company, and the International Nickel
Company. Forged pipe fittings were produced by Taylor Forge. Approxi-
mately two years were required to complete all INOR-8 procurement.

Typical INOR-8 prices were?

Rolled plate $3.00 per pound
Forged bar 4.50 per pound
Tubing (1/2-in. diam) | 6.00 per foot )
Pipe (5-in.-diam sched 40) 225,00 per foot
Graphite. The 6000 1b of special graphite bars for the MSRE core .

were extruded, processed, and machined by the Carbon Products Division
of Union Carbide Corporation for a fixed price of about $194,000,

This graphite, which has low salt absorption and low permeability
to gases, has been produced only in the laboratory and in small sizes
prior to this order. Approximately 2-3/4 years were required to produce
the final machined bars to the quality and tolerances required for the
MSRE core assembly.

Fabrication. Although some development of fabrication methods and
procedures was required, in general it was found that INOR-8 presented
no more problems than Inconel or various stainless steels if procedures
were closely followed.

65

The Y-12 shops, during peak periods, used ten welders qualified in
INOR-8 welding procedures. These welders made 1300 x-ray-quality welds
requiring only 55 repalrs. Out of 3300 welds which required liquid pen-
etrant inspection of each pass, only 1015 minor repairs to removed dye
indications were required.

To meet the peak demands, during the 2-1/2-year fabrication period,
approximately 40 craftsmen were utilized at Y-12 and 10 craftsmen at the
X-10 and Paducah shops.

Representative costs for making reactor-grade welds to MSRE specifi-
cations aret (1) precision fitting and welding of top head to reactor
“vessel (60 in, diam by 1 in. thick), about $2900, or $187 per foot of
weld; and (2) fitting to reasonable tolerances and welding of bottom head
to fuel storage tank (50 in. diam by 1/2 in. thick), about $620, or $47
per foot of weld.

Data concerning cost of major components are shown in Table 4.
Fabrication Process. A photographic presentation of the comple%ed

major components is made in the following pages, together with pictures
of the more significant stages of fabrication. ‘

Drain Tanks. ZFigure 22 shows the rolled shell for a typical drain
tank. These shells were rolled, machined, and welded to a tolerance of
approximately 1/16 in., in order to give good weld fit-up with the beads.
Figure 23 shows the bayonet tubes assembled to the top head of a fuel
drain tank, and ¥Fig. 24 shows the completed tank. The steam dome cooling
assembly is shown.in Fig. 25. '

Table 4. Cost of Major MSRE Components

gt o Corty TRO oSt Cont

(dollars)

Reactor 9,500 200,000 50,000 200,000
Heat exchanger ‘ 2,100 80,000 28,000
Radiator (less enclosure) 3,500 62,000 61,000
Fuel flush tank 3,200 22,000 20,000
Fuel drain tanks and steam 11,600 108,000 90,000

dome s2

Coolant drain tank 1,800 19,000 9,000

& wo duplicate units.
Fig. 22. Typical Rolled Shell for Fuel Drain Tank.

Fig. 23.

Fuel Drain Tank Bayonet Tubes Assembled to Top Head.

Fig. 24.

67

UNCLASSIFIED
PHOTO 39934

Completed Fuel Drain Tank.

68

PHOTO 39404

a
-]
2
g
z
g

Completed Steam Dome Cooler for MSRE Drain Tanks.

Fig. 25.

69

Heat Exchanger. The shell, entrance-exit plenum, and tube bundle
are shown in Fig. 26, and Fig. 27 shows the completed unit. The semi-
automatic tube-to-tube-sheet welding process is shown in Fig. 28.

=

Fig. 26. Components for MSRE Primary Heat Exchanger.

Fig. 27. Completed Heat Exchanger.

70

UNCLASSIFIED
PHOTO 39316

Fig. 28, Welding Heat Exchanger Tubes to Tube Sheet.

71

Pumps. The MSRE fuel and coolant circulating pumps are very similar,
Figure 29 shows the fuel pump bowl and volute in process. The rotary
element assembly is shown in Fig. 30. The completed pump and drive motor
assembly is shown during hot performance tests in Fig. 31.

Radiator. The radiator with its furnace enclosure, shown in Fig.
32, is one of the more complex MSRE components from the standpoint of
fabrication. The main header in its early fabrication stage is shown in
Fig. 33. A closeup of the tube-header assembly in process, with sheathed
thermocouples attached, is shown in Fig. 34. The completed coil assembly,
prior to its placement in the enclosure, is shown in Fig. 35.

Reactor and Core. As with all other MSRE components, extreme care
and critical inspection went into the fabrication of the reactor vessel
and core assembly. Figure 36 shows the machining of the bottom head for
the vessel, and Fig. 37 shows the attachment of vanes to remove the swirl
from the fuel and to direct its flow through the graphite core. Figure
38 shows the flow distribution and Fig. 39 the vessel ready for instal-
lation of the graphite core. The core assembly at the halfway point is
shown in Fig. 40 and the completed assembly in its can in Fig. 41. The
reactor access nozzle and the control rod thimbles are shown during fab-
rication in Fig. 42. The completed control rod thimble assembly is shown
in Fig. 43, prior to final attachment to the completed vessel, which is
shown in Fig. 44 ready for shipment to the reactor installation site.

UNCLASSIFIED
PHOTO 385969

Fig. 29. Fabrication of Fuel Pump Bowl and Volute.

72

Fig. 30. Fuel Pump Rotary Assembly.

Fig. 31. Completed Fuel Pump on Hot Test Stand.

UNCLASSIFIED
PHOTO 70358

Fig. 32. MSRE Radiator and Furnace Enclosure.
74

UNCLASSIFIED
PHOTO 58612

AMOLVHOBY WNOLLYN

Main Header for the MSRE Radiator In Process.

Fig. 33.
75

UNCLASSIFIED
PHOTO 39465

Fig. 34. Radiator Tube-to-Header Assembly with Thermocouples At-
tached to Each Tube.

Radiator Coil
Fig. 36. Machining Bottom Head for Reactor Vessel.

Fig. 37. Attachment of Flow-Straightening Vanes to Reactor Vessel.

Fig. 38. Reactor Vessel Inlet Nozzle and Flow Distribution.

Fig. 39.

Reactor Vessel Ready to Accept Graphite Core Assembly.
79

UNCLASSIFIED
PHOTO 70627

Fig. 40. Assembly of Graphite Bars into MSRE Core.

80

Fig. 41. Completed Graphite Core Assembly.

Fig. 42.

Reactor Access Nozzle and Control Rod Thimbles In Process.

81

[UNCLASSIFIED!
PHOTO 71111

Fig. 43. Completed Control Rod Thimble Assembly.

UNCLASSIFIED
PHOTO 71114

Fig. 44. Reactor Completed and Ready for Shipment to MSRE Site.
83

NUCLEAR CHARACTERISTICS OF THE MSREL

R. J. Engel

Basis of Core Design

The basic concept of the MSRE is a cylindrical wessel contalning a
graphite matrix for neutron moderation through which a molten-salt fuel
is circulated. To provide a logical basis for the core design, several
preliminary studies were made to establish the dependence of a limited
number of nuclear properties on important core parameters. Three combi-
nations of the components of the fuel mixture were considered in these
calculations; the nominal compositions and densities are listed in
Table 1.

Effect of Core Size

" The effect of core size on the critical concentration and mass of
uranium was examined for a series of cores containing 8 vol % of type I
fuel (Table 1). Calculations were made for two core heights (5.5 and
10 ft) and four diameters (3.5, 4.0, 4.5, and 5.0 ft). The resultant
critical concentrations (in mole % total uranium) and critical masses
(in kilograms of 22°U) are shown in Fig. 1. No allowance was made for
uranium inventory in the circulating loop outside the core.

Effect of Fuel Volume Fraction

Two preliminary studies of the effect of fuel volume fraction were
made in the range from 8 to 16 vol % using fuel mixtures I and II. These
were superseded by a more detailed study using fuel IIT after mechanical
design and chemical studies had established firmer values for the core
size and fuel composition. The core used had a 27.7-in. radius and was
63 in. high. Uniform fuel volume fractions from 0.08 to 0.28 were con-
sidered, and system inventories were based on a fuel holdup of 38.4 ft°
outside the core. In addition to the calculations of critical mass and
inventory, preliminary calculations were made of the fuel temperature
coefficient of reactivity and the reactivity effect associated with
permeation of 7% of the graphite volume by fuel salt.® The results of
this study are summarized in Table 2 for the fuel volume fraction range
from 12 to 28%. . |

Miscellaneous Studies

A number of calculations were made to evaluate at least the critical
concentrations and masses for several other core configurations. In some
cases neutron flux distributions were also obtained. The configurations
considered included:

*This fraction was at that time the estimated fraction of the graphite
volume accessible to kerosene.
84

Table 1. Nominal Fuel Compositions and Densities Used in
MSRE Survey Calculations

Fuel type I o II III

Composition, mole %

LiF® 64 64 70

BeF, 31 31 23

ThF, A 0 1

ZrF, 0

UF,P 1 1 1
Density, g/cm? 2.2 2.2 2 .47

80.003% °Li, 99.997% "Li.
g3, 59 235y, .59 238y,

UNCLASSIFIED
ORNL-LR—DWG 50592A

50
)
AN
= \\\\\\\ 10 ft H
= 30 —— 3
5 C
&) —g— o
= 20 ' 2 &
p— [«4]
x ) 5
(&) ® E
=
-
\ : 55ftH >
10 — ) 3
— 3
—b
0 0
35 4.0 45 50

CORE DIAMETER ({ft)

Fig. l. Critical Concentration and Mass as a Function of Core Size,
85

Table 2. Effect of Fuel Volume Fraction on Nuclear
Characteristics of MSRE

Core dimensions: radius, 27.7 in.; height, 63 in.

Nominal composition of fuel:; LiF-BeF,-ZrF,-~Tht,-UF,
(70-23-5-1-1 mole %)

Temperature: 1200°F
Fuel density: 2.47 g/cm’
Graphite density: 1.90 g/cm?

Fuel fraction, vol % 12 14 16 20 24 28
Critical fuel concen- 0.296 0.273 0.257 0.238 0.233 0.236
tration, mole % U
Cr%%%cal mass, kg of . 11.0 11.8 12.7 14.8 17.4 20.5
U

System 23°U inventory,a 51.0 . 48.6 47 .4 47.1 48.7 52.4
kg

Fuel temperature coeff., -3.93 -3.83 -3.70 -=3.44 =3.16 -—2.86
(8k/x)/°F (x 10%)

Permeation effect,P 7 11.4 9.7 g.3 6.1 4.6 3.5
% 5k/k

%Core plus 38.4 £t3 of fuel.
P permeation by fuel salt of 7% of graphite volume.

1. Cores containing two concentric regions with different fuel volume
fractions. Three combinations were evaluated.

2. Cores containing three concentric regions with different fuel volume
fractions. Thirteen combinations were evaluated.

3. Cores in which the solid moderator was all in a reflector or partly
in a reflector and partly in an island at the center of the core.
Graphite, beryllium, and beryllium oxide were evaluated as solid
moderators for both configurations.

4. Cores in which the fuel was confined within INOR-8 tubes. Three fuel
volume fractions were evaluated for each of three tube thicknesses.

None of these configurations showed any significant advantage over the
uniform, one-region core.
86

Choice of Final Confiliguration

At an early stage of the design it was decided that the core would
be approximately 4.5 ft in diameter and 5.5 ft high, after calculations
showed that the critical mass was relatively insensitive to core size in
this region (Fig. 1). A fuel volume fraction of 0.225 was chosen because
it was near the minimum critical concentration of uranium and had a
favorably low reactivity effect due to fuel permeation of the graphite.
(The nominal fuel fraction was 0.24, but this was reduced to 0.225 be-
cause of mechanical considerations regarding the graphite moderator bars.)

Basic Nuclear Properties

Description of Core

The final design of the core and reactor vessel is shown in Fig. 6
in the preceding paper. Fuel salt enters through a flow distributor near
the top of the vessel and flows downward through an annular region out-
side the graphite to the lower head. Antiswirl vanes in the lower head
direct the flow radially inward before it turns up to enter the fuel
channels. After passing through the moderated portion of the core, the
fuel leaves the reactor vessel from the upper head. The graphite moder-
ator is supported by an INOR-8 grid which is suspended from the core can.
The channeled region of the core consists of 2-in.-square, vertical
graphite stringers, with half-channels machined into each face to provide
fuel. passages. The regular pattern is broken near the axis of the core,
where three control rod thimbles and a sample assembly are located.
Figure 2 shows a typical fuel channel and the arrangement around the core
axis.

UNCLASSIFIED

REACTOR § ORNL-LR-DWG 6084541

% \\\\\\%7//,\\\\\\% S

TYPICAL
FUEL CHANNEL

R=REMOVABLE

4 GRAPHITE IRRADIATION /\\\\f/// % W STRINGER

SAMPLES ( ¥/g-in.DiA)

2in. TYPICAL

Fig. 2. MSRE Control Rod Arrangement and Typical Fuel Channel.
87

Methods of Calculation

Most of the nuclear characteristics of the core were calculated for
a somewhat simplified representation of the actual design which included
as much detail as was practical. The actual configuration was approxi-
mated by a two-dimensional model in R-Z geometry (cylinder with angular
symmetry) containing 20 regions with different compositions based on the
fractions of salt, graphite, and structural material in each. Limitation
of the model to two dimensions resulted in a large savings 1n computing
time, and the representation was considered adequate for most purposes.
The principal shortcoming of the model was the representation of the three
control rod thimbles. These were approximated by a single, cylindrical
shell of metal around a cylinder of low-density fuel to account for the
extra fuel around the thimbles and the air inside them. Special mathe-
matical models were used where accurate representation of the region
around the control rods was required.

The nuclear characteristics of the reactor were calculated for three
different fuel-salt mixtures with nominal compositions as shown in Table
3. In initial operation of the reactor a mixture very similar to fuel C
will be used.

Table 3. MSRE Fuel Salts for Which Detailed Nuclear
Calculations Were Made

Fuel type A B C

Salt composition, mole %

LiF® 70 66.8 65
BeF> 23.7 29 29.2
ZrF, 5 4 5
ThF,, 1 0 0
UF, {approx) 0.3 0.2 0.8

Uranium composition, at. %

2347 1 1 0.3
233y 93 93 35
236y 1 1 0.3
238y 5 5 YA
Density at 1200°F, 1b/ft? 144.5  134.5  142.7

#99.9926% 7Li, 0.0074% ©Li.
88

Multigroup Calculations. The critical uranium concentration for
each fuel was calculated by the use of MODRIC, a one-dimensional, multi-
group neutron diffusion code. This program was used with 32 fast-neutron
energy groups and a thermal energy cutoff at 0.4 ev. Fast group cross
sections for most elements were generated by the use of GAM-I. However,
the cross-section library that was available for this program did not
include data for ®Li, 7Li, and *°F. The slowing-down effects of these
isotopes were simulated with an equivalent amount of oxygen, and absorp-
tion cross sections were compiled separately from basic data. Average
cross sections for the thermal energy group were computed by the use of
another program, THERMOS.

In addition to the critical concentrations, the MODRIC calculations
produced sets of two-group constants for each region of the 20-region
model. Three one-dimensional calculations were required to provide the
necessary constants for all the regions with each fuel mixture. The
MODRIC calculations also produced a neutron energy spectrum for each
region (normalized to 1 neutron produced in the reactor).

Two-Group Calculations. Absolute spatial flux distributions at 10
Mw were calculated for the complete 20-region model using the two-
dimensional, two-group neutron diffusion code EQUIPOISE-3. This program
also served as a check on the critical concentrations produced by the
one-dimensional calculations. |

Other Calculations. The reactivity changes associated with changes
in fuel and graphite temperature, fuel and graphite density, uranium
concentration, and 135%e concentration and poison fraction were calcu-
lated for a one-region, two-group model of the core. This model was also
used to investigate the effect of the thermal-neutron cutoff energy on
the temperature coefficients. The latter study revealed that, for this
system, the temperature coefficients of reactivity would be underestimated
if the thermal energy cutoff were less than 1 ev.

Summary of Results

The basic nuclear characteristics of the MSRE with each of the three
fuel mixtures are summarized in Table 4. The critical concentrations
shown are the results of the MODRIC calculations. These concentrations
gave values for ke of 0.993, 0.997, and 0.993 for fuels A, B, and C,
respectively, when ised in the EQUIPOISE program. The thermal-neutron
fluxes are direct results of the EQUIPOISE calculations, as are the
prompt-neutron lifetimes. The reactivity coefficients.were obtained
from the single-region calculations. The temperature coefficients in
this table reflect the l-ev thermal energy cutoff. Fuel and graphite
temperature coefficients of reactivity were also computed for fuel C using
the EQUIPOISE results. The values for fuel and graphite were, respectively,
3 and 15% lower than those obtained with the single-region model. How-
ever, the multiregion results are not necessarily more accurate because of
assumptions that were made regarding the spatial dependence of the thermal-
neutron spectrum.
g9

Table 4. Nuclear Characteristics of MSRE with Various Fuels
Fuel A Fuel B Fuel C
Uranium concentration, mole %
Clean, noncirculating
235y 0.291 0.176 0.291
Total U 0.313 0.189 0.831
Operatinga
235y 0.337 0.199 0.346
Total U 0.362 0.214 0.890
Uranium inventory,b kg
Initial criticality
2327 79 48 77
Total U 85 52 218
Operating®
235y 91 55 92
Total U 28 59 233
Thermal-neutron flux,® neutrons
cm < sec”
Maxcimum 3.31 x 10Y%  5.56 x 1012 3.29 x 1013
Average in graphite- 1.42 x 1013 2.43 x 1013 1.42 x 10%2
moderated regions
Average in circulating fuel 3.98 x 10%2 6.81 x 1012 3.98 x 10%?
Reactivity coefficientd
Fuel temperature, (°F)~! —3.03 x 107> —4.97 x 10~° —3.28 x 10~
Graphite temperature, —3.36 X 107> —4.91 x 10~° —3.68 x 103
(oF )-—1 -
Uranium concentration 0.2526 0.3028 0.1754°
0.2110%
13%Xe concentration in —1.28 x 108  —-2.04 x 108 ~L.33 x 108
core, (atom/barn-cm)—1
135%e poison fraction —0.746 —0.691 —0.752
Fuel salt density 0.190 0.345 0.182
Graphite density 0.755 0.735 0.767
Prompt-neutron lifetime, sec 2.29 X 10™% 3,47 x 10~4 2.40 x 1074

“Fuel loaded to compensate for 4% 5k/k in poisons.
PBased on 73.2 ft? of fuel salt at 1200°F.
€At operating fuel concentration, 10 Mw.

dat initial critical concentration.
ficients for variable x are of the form

are of the form (x/k)/(dk/dx).

®Based on uranium isotopic composition of clean, cr
enriched uranium.

TBased on highly (~93%)

Where units are shown, coef-
(1/k)/(3%/dx); other coefficients

itical reactor.
20

Table 5 summarizes the neutron balance for the reactor with fuel C
at the clean, critical condition. The large leakage values illustrate
the relatively small "nuclear size" of the MSRE. This factor has an im-
portant effect on the size of the temperature coefficients of reactivity
because changes in neutron leakage with temperature contribute heavily to
the coefficients in this reactor. '

Figures 3 and 4 show, respectively, the axial and radial distribu-
tions {at the position of maximum thermal flux) of the two-group neutron
fluxes at 10 Mw with fuel C. The zero position for the axial distribu-
tions is the bottom of the graphite matrix in the core. The thermal flux
distribution in the main portion of the core (0 to 65 in.) is very closely
approximated by a sine distribution with an effective core height of 78
in. The depression in the radial distribution of the thermal flux near
the axis is caused by the control rod thimbles and the extra fuel that
surrounds them.

Table 5. Neutron Balance for Fuel C at the Clean, Critical Condition
(per 10° neutrons produced)

Absorptions

Region '
235y 238y  galt® Graphite INOR Total
Main core 45,459 7252 4364 795 1380 59,250
Upper head 3,031 928 675 1 131 4,766
Lower head 1,337 449 294 0 1480 3,560
Downcomer 1,496 338 203 0 0 2,037
Core can 0 0 0 0 3635 3,635
Reactor vessel 0 0 0 0 3056 3,056
Total 51,323 8967 5536 796 2682 76,304
Surface Leakage

Fast ‘ Slow Total

Top 1,991 10 2,001

Sides 19,619 1004 20,623

Bottom 1,068 4 1,072

Total 22,678 1018 23,696

aConstituents other than <3°U and <38U.
91

Later Refinements?

Some additional calculations have been made in an effort to define
the basic nuclear properties of the reactor more precisely. An improved
version of the cross-section program, GAM-IIL, which includes lithium and
fluorine is now available. Calculations using this program indicate that
the critical concentrations are overestimated by about 14%. This differ-
ence is due primarily to the slowing-down effect of inelastic neutron
scattering by L19F which was not considered in the earlier calculations.

Some calculations were performed to evaluate the advantages of using
32 fast-neutron groups. Sets of multigroup cross sections with 32, 16, 8§,
and 4 fast groups were generated using the GAM-II program. These cross
sections were used in one-dimensional diffusion calculations (MDDRIC) of
the critical concentration. All the calculations agreed to within 1% on

UNCLASSIFIED
ORNL—DWG 63-8149A

(x 10'%) . (x 10"
14 T

FAST FLUX \

/

sec

i

e N\
/

/

FAST FLUX (neutrons cm~2 sec— 1)

THERMAL FLUX {neutrons cm™2

—-10 o] 10 20 - - 30 40 50 60 70 80
AXIAL POSITION (in.)

Fig. 3. Axial Distribution of Two-Group Fluxes 8.4 in. from Core
Center Line, TIuel C, '
92

UNCLASSIFIED
ORNL—DWG 63—8151A

( x 10" ( x 10")
16 . . 8
14 ~ 7

\FAST FLUX

12 A : 6
T
— Q
Q
| \ <
g 10 S o
e £
o [ ]
IE o
© o
» : -
g /\\ 4 8
g 8 7 \ -
£ >
3
S THERMAL FLUX T

|

L \ ' <_Zl'.I
— b 3 =
2 o
(W0
w . I
b

4 \ \ 2

2 N\ 1

0 0

0 5 10 15 20 25 30

' 'RADIUS (in.) '

Fig. 4. Radial Distribution of Two-Group Fluxes near Core Midplane,
Fuel C,
93

critical concentration, indicating the adequacy of the four-group treat-
ment for future work.

Bffects of Fuel Circulation

‘The fact that the fuel circulates through the MSRE core has a defi-
nite effect on some of the nuclear properties. SubJjects that have been
given separate consideration are the loss of some ¢of the delayed neutrons
and the effects of gas bubbles in the circulating stream.

Delayed Neutrons. Since the transit time through the active region
of the core (where the fission density is substantial) is only about one-
third the loop transit time, significant portions of the delayed neutrons
from the longer-lived precursors are emitted outside the core, where they
are lost to the chain reaction. Calculations have been made to evaluate
this loss under steady-state operating conditions. For this purpose, the
core was approximated as a single-region cylinder with uniform fuel flow
velocity. The effect of the precursor movement on the axial distribution
of the delayed-neutron emission densities is shown in Fig. 5. The source

UNCLASSIFIED
ORNL-LR-DWG 73620R2

(x1079)
2.8 T
GROUP #y, (sec)
| 56
2 23
2.4 3 6.2
q 2.2
5 0.6
50 6 0.2

-
Z.

7
Y | |

0 0.2 0.4 0.6 0.8 1.0
RELATIVE HEIGHT

RELATIVE DELAYED NEUTRON SOURCE DENSITY {cm~3)

Fig. 5. Axial Distribution of Delayed-Neutron Source Densities in
an MSRE Fuel Channel. Fuel circulating at 1200 gpm.
9%

of delayed neutrons from the shortest-lived group of precursors has es-
sentially the same distribution as the prompt-neutron flux. At the other
extreme the distribution from the longest-lived group is nearly flat.

The effects of the distorted source distributions and the lower average
energies of the delayed neutrons on their nonleakage probabilities were
included in the evaluation of the effective delayed-neutron fraction. The
total effective fraction of delayed neutrons is 0.00362 at a 1200-gpm
circulation rate and is 0.00666 in the static core. The total yield of
precursors is 0.00641l. The effective delayed-neutron fraction was used
in subsequent kinetics calculations instead of trying to account for
precursor transport during transient conditions. '

Gas Bubbles. The use of the xenon-stripping device in the fuel-pump
bowl results in the introduction of helium gas bubbles into the circulating
fuel stream. At operating conditions the fuel in the core may contain
about 1 vol % of helium bubbles. The presence of these bubbles makes the
fuel compressible, causing a pressure coefficient of reactivity. In addi-
tion, the presence of the gas modifies the fuel temperature coefficient
of reactivity because the density of the salt-gas mixture changes with
temperature at a rate different from the density of the salt alone.

For rapid changes in pressure the mass of helium in the core remains
essentially constant, and the reactor exhibits a positive pressure coef-
ficient of reactivity. However, for very slow changes, the gas volume in
the core is inversely proportional to the compression ratio between the
pump bowl . (where the gas is introduced) and the core. This leads to a
negative pressure coefficient. The presence of gas increases the temper-
ature coefficient of the mixture density; this leads to an increase in the
size of the negative temperature coefficient of reactivity.

The magnitudes of the pressure and temperature coefficients of re-
activity with entrained gas in the core are listed in Table 6 for a core
temperature of 1200°F and a pump-bowl pressure of 5 psig.

During normal operation, the presence of entrained gas introduces
additional power '"noise'" as fluctuations in core outlet pressure are con-
verted to reactivity perturbations. In power excursions the gas enhances
the negative temperature coefficient of reactivity of the fuel. At the
same time, the pressure coefficient makes a positive contribution to re-
activity during at least part of the power excursion. However, the
relation between pressure rise and temperature rise in any credible ex-
cursion is such that the net reactivity feedback from the combined pres-
sure and temperature effects is negative.

Control Rods

The MSRE contains three flexible control rods whose centers are lo-
cated at three corners of ‘a 4~in. square around the axis of the core.
The rods operate in re-entrant INOR-8 thimbles, 2 in. in diameter with
0.065-in. walls. The poison elements are hollow cylinders, 1.08 in. OD
by 0.84 in. ID and 1.3 in.long, containing 30 wt % Al,03 and 70 wt %
95

Table 6. Reactivity Coefficients with Entrained Gas in Core®

Tuel A Fuel B Fuel C

Fuel density coefficient of 0.190 0.345 0.182
reactivity

Fuel temperature coefficient of
reactivity, (°F)™r (x 10™?)

No gas or slow changes —-3.03 44.97 —3.28
with gas

Rapid changes with gas -3.14 —5.17 -3.39
Pressure coefficient of reac-
tivity, psi™t (x 10™7)
Slow changes with gas -3.8 =7.0 —3.7
Rapid changes with gas +6.3 +11.4 - +6.0

8Evaluated at T = 1200°F, P = 36.5 psia (pump-bowl pressure, 5 psig).

Gdp03. Thirty-six such elements, each clad in Inconel, are used for each
control rod.

Rod-Worth Calculations

The initial rod-worth calculations were done in two dimensions with
two neutron energy groups (fast and thermal) with EQUIPOISE-3. For this
purpose a horizontal cross section of the core was represented in X-Y
geometry, and an equivalent axial buckling was applied. The poison ele-
ments were regarded as ''black" to thermal neutrons and transparent to
fast neutrons. The control rod 'cells" were sized such that the perimeter
of the square representation of each rod thimble was equal to that of the
actual item. These calculations gave the reactivity effect associated
with the insertion of three control rods completely through the core for
each of the three fuels. The calculations were extended for fuel A to ob-
tain the effects of various combinations of fully inserted rods. Other
EQUIPOISE-3 calculations, in R-Z geometry with a single control cylinder,
were used to estimate the variation in worth with insertion distance.

Table 7 lists the total control rod worths obtained from the
EQUIPOISE-3 calculations. The worths of the various combinations listed
for fuel A indicate the existence of appreciable interaction or '"shadowing'
between the rods. All values are for a clean reactor with uranium con-
centrations corresponding to the "just critical" condition without rods.
Figure 6 shows the fractional worth of a rod or bank of rods as a function
of fractional distance inserted. .This curve is displaced somewhat to the
left of the idealized f sin® curve.

t
96

Subsequent calculations which considered epithermal absorptions in
the poison elements have been made using the EXTERMINATOR program in two
dimensions with the cross section of the core represented in R-0 geometry.
Four neutron energy groups were used: a fast group, a thermal group,
and two groups covering the important gadolinium resonances. Two calcu-
lations were made: one using the latest clean, critical uranium concen-
tration computed for fuel C with GAM-II cross sections and one with fuel
containing 1.9% Sk/k of fission product poisons compensated by additional
235U. The results were, respectively, 5.51 and 5.22% 8k/k for three rods.
In the clean case, epithermal absorptions accounted for 12% of the rod
worth.

Table 7. Control Rod Worths in the MSRE

Worth

Fuel Rod Configuration (% Bk/k)
A Three rods in 5.6
Rod 1 in, rods 2 and 3 out 2.4
Rod 2 in, rods 1 and 3 out 2.3
Rods 1 and 3 in, rod 2 out A
Rods 1 and 2 in, rod 3 out 4.l
Three rods in 7.6
C Three rods in 5.7
.UNCLASSIFIED
ORNL-LR-DWG 57687A
{.0 /(__fi
0.8 /
I
&
(@]
=
-?Ij 0.6
'.,_
2
5
S 04
E /
0.2 /;//
0

0 0.2 0.4 0.6 0.8 1.0
FRACTION OF LENGTH INSERTED

Fig. 6. Fractional Worth of a Rod or Bank of Rods as a Function of
Distance Inserted.
97

Control Requirements

Since the fuel in this reactor is fluid, it is not necessary for the
control rods to perform flux-flattening functions for optimization of
burnup and fuel-to-coolant heat transfer. In addition, the temperature
distributions in the core are mild enough that temperature gradients are
no problem. As a result the control rods can be operated in the simplest
manner possible, consistent with good reactivity control.

In general, the two shim rods will be kept at equal distances of
insertion, and the servo-operated regulating rod will be inserted deeper
to keep the tip out of the "shadow" of the shim rods. The capability of
adding uranium at any time to compensate for burnup and stable poison
buildup permits steady operation at full power with the rods almost com-
pletely withdrawn. The only requirement is that the rods be deep enough
so the insertion of negative reactivity is not delayed by travel time in
regions of low effectiveness.

The driven speed of all three rods is 0.5 in./sec in both directions.
This allows the regulating rod to change reactivity at 0.02 to 0.04%
8k/k per second, depending on its position within the allowed operating
range. Extensive analog computer studies indicated that this capability
provides adequate control for all normal operations. The maximum capa-
bility for reactivity addition with all three rods moving in the region
of maximum sensitivity is 0.08% 8k/k per second for fuels A and C and
0.1% per second for fuel B.

The shim requirements to compensate for transient reactivity effects
during operation are given in Table 8. These requirements, coupled with
the control rod worths, allow for a shutdown margin of at least 3% 8k/k at
1200°F for all operating conditions. The major uncertainty in the shim

Table 8. Rod Shim Requirements

Effect (% dk/k)

Cause

Fuel A Fuel B Fuel C

Loss of delayed neutrons 0.3 0.3 0.3

' Fntrained gas 0.2 0.4 0.2
Power {(0-10 Mw) 0.06 0.08 0.06

135%e (equilibrium at 10 Mw) 0.7 0.9 0.7

Samarium transient 0.1 0.1 0.1
Burnup (120 g of 23°U) 0.03 0.07 0.03

Total 1.% 1.9 1.4

98

requirements is that associated with 135¥e. Because of difficulties in
calculating this quantity, its evaluation will be the subject of extensive
experimental analysis. However, present indications are that the values
given are conservative.

Core Temperatures

The distributions of the fuel and graphite temperatures in the core
are of interest because of their effects on reactivity, particularly in
connection with power changes. Overall temperature distributions were
calculated for the nominal full-power conditions (i.e., a fuel flow rate
of 1200 gpm and a temperature rise of 50°F) in the reactor. Results are
presented for fuel C, but those for fuels A and B were practically the
same.

Fuel Temperatures

For burposes of these calculations, the contents of the reactor
vessel can be divided into a main core region, where most of the power

UNCLASSIFIED
ORNL-DWG 63-8165A

1290

1280 TN

1270 — ~\\\
1260 \

GRAPHITE \
1250

- w ) \\

0

o 1240 \

pun |

'_

- \

4230

w

a.

2 FUEL \\

M 1220 =\\\\ \\
1210 |— \\\\\ K
1200 \\\\\\\\

1
190 — k\\\
1480
0 5 0 .15 20 25 30
RADIUS (in.)

Fig. 7. Radial Temperature Profiles near the Core Midplane.
99

is generated, and a number of peripheral regions associated with the
inlet and outlet. The main core region in this context is that portion
that contains the graphite moderator.

Approximately 14% of the reactor power is produced in, or transferred
to, the fuel as it flows through the peripheral regions. At 10 Mw this
results in a 3.6°F temperature rise in the fuel before it enters the main
core region and another 3.4°F rise as it flows through the upper head to
leave the vessel.

The spatial distribution of the fuel temperature in the main core
region was calculated by combining the fission-density distribution
produced by EQUIPOISE-3 and the fuel-flow distribution predicted by
hydraulic studies on a full-scale model of the reactor vessel. Axial
conduction in the graphite was neglected, so heat produced in the graphite
was transferred to the fuel with approximately the same axial distribution
as that of the fuel heat source. The results of this calculation describe
only the distribution of the average temperatures in the fuel channels.
Variations in fuel temperature within individual channels are superimposed
on this distribution.

Figures 7 and 8 show the overall radial temperature distribution near
the core midplane and the axial distribution in the hottest fuel channel.
These results are based on average fuel inlet and outlet temperatures of
1175 and 1225°F respectively. The low temperatures near the axis in the
radial distribution result from the fact that the flow velocity in this
region is about three times the average for the core.

UNCLASSIFIED
ORNL-DWG €3-8164A

1300 (

P
GRAPHV -\\

1280 7

/ L

1260 / /

"
: / yd
w
1240
g / L
|_
<
E:J ) /
1220 /
< 4
w /
'_
4200
/ ,
uao,////
150
0 10 20 30 40 50 60 70

AXIAL POSITION (in.)

Fig. 8, Axial Temperature Profiles in Hottest Channel and Adjacent
Graphite Stringer.
100

Graphite Temperatures

Heat produced in the graphite by absorption of beta and gamma radia-
"tion and the elastic scattering of fast neutrons amounts to about 7% of
the total heat produced in the reactor. Figures 9 and 10 show the radial
and axial distributions of heat generation in the graphite. Since this
heat must be transferred to the fuel for removal, the temperature of the
graphite is, in general, higher than that of the fuel in the adjacent
channels. The temperature difference is enhanced by the fact that the
fuel near the channel walls is hotter than the average for the channel
(Poppendiek effect). Permeation of the graphite by fuel, if it occurs,
will also enhance the graphite-fuel temperature difference. Table 9
indicates the effect of fuel permeation on the maximum difference between
the average temperature of a graphite stringer and the average for an
adjacent fuel channel. The overall radial and axial temperature profiles
in the graphite are shown in Figs. 7 and 8.

Table 9. Maximum Values of Graphite-Fuel Temperature Difference
as a Function of Fuel Permeation

Fuel permeation, vol % of graphite 0 0.5 2.0

Graphite-fuel temperature difference, °F

Poppendiek effect in fuel 55.7 58.3 65.4
Graphite temperature drop 5.5 6.7 - 9.8
Total 61.2 65.0 75.2

Average Temperatures

The concept of average temperatures has a number of useful applica-
tions in operating and in describing and analyzing the operation of the
system. The bulk average temperature, particularly of the fuel, is
essential for all material balance and inventory calculations. The
nuclear average temperatures of the fuel and graphite, along with their
respective temperature coefficients of reactivity, provide a convenient
means of assessing the reactivity effects associated with temperature
changes.

The fuel and graphite bulk average temperatures computed for the
reference condition (1175°F inlet, 1225°F outlet) were 1199.5 and 1229°F
respectively. Fuel permeation of 2% of the graphite volume increases the
graphite average temperature by 4.4°F.

Nuclear average temperatures were obtained by weighting the tempera-
ture distributions with their nuclear importance functions. For the same
reference conditions, the nuclear average temperatures of the fuel and
101

UNCLASSIFIED
ORNL—DWG 63-—-8185A

0.9 /

\\\ifTAL

O
[0)]
\

AN \

/

HEATING ¢ wcHs/cm3)

0.4 \\ \
0.3 \\\' | \\
0.2 _—_-"“‘“~=-§§\\N <
\
\\
0.1 \\\\\\ \\

BETA \

] q 8 12 16 20 24 28
RADIUS {in.)}

Fig. 9. Heating in Graphite; Radial Distribution near Midplane at
10 Mw. .
102

UNCLASSIFIED
ORNL—-DWG 63— 8186A

0.8

0.7 //\\TOTAL

0.6 / \

0.5 \
/ —
X // GAMMA
2 0.4 / N
. \
=
=
<l
L
X

0.2'// ”””,;g——-———-—___ \\ \

,//////// NEUTRON EN\\\‘\\\
iy ’///////// | \\\\\\\\\¢\\\\\
| BE TA
0
0 10 20 30 40 50 60 70

AXIAL POSITION (in,)

Fig. 10. Heating in Graphite; Axial Distribution 8.4 in. from Core
Center Line at 10 Mw,
103

graphite were 1211 and 1255°F respectively. The effect of 2 vol % fuel
permeation in this case was an increase of 7°F in the graphite nuclear

average temperature.

Power Coefficient of Reactivity

-~

Whenever the reactor power is raised, temperatures of the fuel and
graphite throughout the core must diverge. However, the shape of the
temperature distributions at power and the relations between inlet, out-
let, and average temperatures are inherent characteristics of the system
which are not subject to external control. On the other hand, the position
of the temperature distribution at power, relative to the temperature of
the zero-power isothermal reactor, can be controlled by manipulation of

the control rods. The amount of control rod manipulation, or reactivity

change, required to maintain a particular temperature relation msy be re-

garded as a power coefficient. Once the temperature behavior has been

prescribed, it is possible to evaluate the power coefficient. However,
this coefficient is not a constant since it depends on the arbitrary

- prescription of the temperature behavior. Values of the power coefficient
of the MSRE and some pertinent temperatures for three fuels and three
different modes of temperature control are given in Table 10.

Fission Product Poisoning

/

Operation of the reactor at power and changes in power level will
produce changes in reactivity as fission products are bullt up and/or

Table 10. Power Coefficients of Reactivity and Temperatures at 10 Mw

- Mode of Control

Power Coefficient Temperature®
[ (% ok/k)/Mw] (°F)
Fuel A Fuel B TFuel C T(Out) T(in) T? T;

Constant T(out)

H(

—0.006 -0.008 -0.006 1200 1150 1186 1230

+ T,.
Constant —Out) > (in) 5,000 —0.033 —0.024 1225 1175 1211 1255

"Hands off"

Fuel A
Fuel B
Fuel C

0 | 1191 1141 1177 1221
0 1192 1142 1178 1222
0 1191 1141 1177 1221

aSystem isothermal at 1200°F at zero power.
104 -

destroyed. The negative reactivity effect due to the buildup of stable
fission products, which depends primarily on integrated power, will be
compensated by periodic additions of <3°U. However, the transient ef-
fects, such as those due to 135%e anad 1498m, will be compensated by
control rod manipulation.

Xenon-135

The reactivity effects associated with the buildup and decay (or
removal) of 135%e are expected to form a significant part of the shim
requirements in this reactor. Because of this and the uncertainties in
predicting the xenon behavior, this problem will be studied extensively
during the operation of the reactor. Some preliminary calculations have
been made, using rather gross assumptions, to estimate the xenon poisoning
at full power.

Production Mechanisms. Xenon-135 is produced directly from fission
of uvranium as well as from the radiocactive decay of 13371 circulating with
the fuel. In addition, there is a possibility that iodine, which pre- .
sumably will not be stripped out in the pump bowl, may be trapped either
in the graphite or on metal surfaces. Xenon produced from the decay of
fixed iodine may have significantly different residence times, inside or
outside of the core, than that produced in the circulating stream. The
possibility of Xenon production from fuel salt that is soaked into the
~graphite further increasés the complexity of the problem.

Destruction Mechanisms. There are two major competing mechanisms

for the removal ,of 13°Xe from the circulating fuel. These are stripping
in the pump bowl and migration to the unclad graphite moderator. Approxi-
mately 85% of the 13%Xe that gets to the graphite is destroyed by neutron
absorption, thus adding to the poisoning effect; the remainder decays to
13505, Both the xenon migration to the graphite and the stripping ef-
ficiency are highly uncertain, but it appears that 50% or more of the
total 127Xe produced may be removed at the pump bowl. In addition to the
major mechanisms, it is also necessary to consider the decay and burnup

of xenon in the fuel itself. Different destruction rates for xenon
produced from trapped iodine and from fuel soaked into the graphite should
also be considered in a complete analysis.

Poisoning Effect. Preliminary calculations of the xenon poiSoning
were made for a range of pump-bowl stripping efficiencies and three
assumed values of the xenon diffusion coefficient in graphite. In these
calculations it was assumed that all the iodine circulated freely with
the fuel and that all the xenon is produced in circulating fuel (no fuel
soakup in graphite). The results are shown in Table 11. The expected
values reflect a xenon diffusion coefficient in graphite of 1.5 x 10™°
£ft%/hr (4 x 107° cm?®/sec), and the maximum and minimum values show the
effect of increasing and decreasing the parameter by a factor of 100.

An important assumption in these calculations was that all the
circulating xenon was in solution in the salt. It has recently been
105

Table 11. Reactivity Effects of 13°Xe

Fuel A or C ~ Fuel B
Pump bowl stripping efficiency, % 25 50 100 50
Reactivity effect, % 8k/k
Expected -1.2 0.7 -0.5 -0.9
Minimum -1.0 -0.5 —0.3 —0.5

shown that the presence of circulating helium bubbles, introduced by the
stripper flow in the pump bowl, has a pronounced effect on the Xxenon be-
havior. Because of the extreme insolubility of xenon in molten salt, the
addition of 1 wvol % of circulating helium bubbles reduces the equilibrium
xenon partial pressure (in the salt and bubbles) by a factor of about 40,
for a given total amount of xenon in circulation. That is, about 98% of
the circulating xenon will be in the bubbles. Since the xenon partial
pressure provides the driving force for migration to the graphite, the
preliminary calculations may overestimate the contribution to the poison-
ing by xenon in the graphite.

Samarium

Assuming that all the samarium circulates with the fuel, the buildup
of this fission product will ultimately poison about 0.9% &k/k at 10 Mw.
This level will be approached after about 100 days' operation at full
power. This effect can be compensated by uranium addition to the fuel.
However, each time the power is reduced to zero, the samarium poisoning
will increase an additional 0.1% over a period of 7 days. This increase
- is burned out in about 7 days by operation at full power. Thus only the
0.1% Bk/k samarium transient need be compensated by the control rods.

Low-Cross~Section Poisons

The large majority of the fission products may be regarded as an
aggregate of low-cross-section nuclides. The poisoning effect of this
group increases initially at a rate of about 0.003% Bk/k per day at full
power.

Neutron Sources

The presence of a neutron source that is independent of the fission
chain reaction is essential to the safe and orderly operation of the re-
actor. Three classes of source neutrons will be available during the
106 ‘ .

operation of the MSRE. These include neutrons from the source that is
inherent in the fuel itself, photoneutrons produced by decay of fission
products, and the external source.

Inherent Source

The fact that the uranium is intimately mixed with the other con-
stituents of the fuel salt provides a substantial source of neutrons from
the interaction of alpha particles from the uranium with beryllium and
fluorine. Some neutrons are also produced by the action of alpha particles
on lithium, but this yield is negligible compared to that from fluorine
and beryllium. Diffusion-plant enrichment of uranium in 2357 also in-
creases the 234U content, and about 97% of the alpha activity that leads
to neutron production is due to 22%U. Over half of the O-n reactions take i
place in fluorine because of its high concentration in the fuel. The
total yield of &-n neutrons from the fuel in the core varies between 3
and 4 x 107 neutrons/sec, depending on the choice of fuel. There is also
an inherent source due. to spontaneous fission, primarily of 238 U, but this
amounts. to less than 1% of the G-n source, even in fuel C, Wthh contains .
about 0.6 mole % 238y, gince the inherent source is always present in
the reactor when the fuel is there, it provides an absolutely reliable
source whose presence need not be externally verified. The size of the
inherent source has been shown to be adequate from the standpoint of re-
actor safety.

Photoneutron Source

Once the reactor has been operated at power, the action of gamma
rays from the decay of fission products on the beryllium in the fuel pro-
vides an additional source of neutrons. This source provides a minimum
of 107 neutrons/sec in the core for 50 days after operation at full
power for 1 week.

External Source

From the stahdpoint of orderly operation, it is desirable to monitor .
the filling of the reactor (during routine startups) with the nuclear
instruments from the start of this operation. Since detection of the .

inherent source requires that some fuel be in the reactor, an external
source 1s required for this purpose. The external source will also serve
a number of useful purposes during the initial startup program.

A thimble is provided for the insertion of a neutron source in the
thermal shield which surrounds the reactor vessel. Because of the dis-
tance from the source thimble to the detectors on the opposite side of
the reactor and the low sensitivity of the chambers (0.026 count/nv), a
relatively large source will be required to produce a significant count
rate (>2 counts/sec) on the startup fission chambers when there is no
fuel in the core. Calculations using the two-group, two-dimensional
program EQUIPOISE BURNOUT indicate the need for a source of 4 X 107
neutrons/sec for this purpose.
107

The source requirement could be met with an antimony-beryllium
source, but the 60-day half-life of 1245, gictates a substantially greater
source to avoid frequent replacement. (The neutron flux in the thermal
shield at the source tube is too low to maintain the 124Sp activity, even
at full power.)

Because of uncertainties in the calculations of the source require-
ment, the final specification of the external source will be based on
measurements to be made with the reactor vessel installed. The con-
struction and startup schedule is such that there is time for procurement
of a source after these measurements and before the source is needed for
nuclear operation.

Kinetics

A number of studies, using both digital and analog computer tech-
niques, have been made to explore the kinetic behavior of the reactor.
These calculations can be divided into two categories: those dealing with
small disturbances such as random '"noise" and changes in load demand that
would be encountered in normal operation, and those dealing with major
disturbances that arise from abnormal situations.

Normal Operation

The réactor will be self-regulating during normal operation because
of the negative temperature coefficients of both the fuel and graphite.
The degree of self-regulation depends on the power level and is rather
sluggish even at full power. A number of factors contribute to the slow
response of the system. These include: '

. low power density in the core,
. high heat capacities of the fuel and graphite,
low heat-transfer rate between fuel and graphite,

. Jlow heat producfion in the graphite,

L B AR VS A N

long loop delay times between the heat sink at the radiator and the
heat source in the core.

Figures 11 and 12 show the results of analog calculations of the
system response to changes in load demand at the radiator, in the absence
of external reactivity control. 1In both cases, slow changes in demand
were started at zero time, and the temperature and power response of the
reactor were recorded. The slow decrease in fuel temperatures after the
increase in power and the increase after the power reduction reflect the
attainment of the steady-state temperature distribution in the graphite.
The slow power oscillation at low power was observed in all simulations
and appears to be an inherent characteristic of the system.

‘During routine operation, some external reactivity control will be
imposed through the use of a servo-operated regulating rod. This is ex-
108 .

pected to improve the response of the system and to minimize the
"wallowing" at low power. At powers above 1 Mw the servo system will be
used to regulate the reactor outlet temperature, with reset action based
on the nuclear power level. At lower powers the neutron flux will be con-
trolled directly, and the heat removal will be manually balanced against
the production.

UNCLASSIFIED
ORNL—-DWG 63—8228A

1240
1220 _\\‘ -
& (7) .
w 1200 W ~— " OouT
[v 4
=]
fd
Lt
cr [ ]
o
S 1180
Lt
3 \
\\(;l)-m\
1140
12
N /\\
/ \\\__//’?;;;;—
s e ] .
s
[0uy
LL' L ]
; -
< 6
>
] ™
|
[/
4
2 //
0

0] 100 200 300 400 500 600 700 800 S00 1000
‘ TIME (sec)

Fig. 11. Response of Nine-Region Model of MSRE to an Increase in
Power Demand.
109

UNCLASSIFIED
ORNL-DWG 63-8473A

1360

1320 ~

1280

1240 ) B

TEMPERATURE (°F )
<
=
C

1200 7

1160

8 .
\POWER

FLUX POWER ( Mw)

0 100 200 300 400 ‘500 600 700 800 900
TIME (sec)

Fig. 12, Response of Nine-Region Model of MSEE to a Decrease in
Power Demand.

Abnormal Situations

The consequences of several, highly abnormal reactivity incidents
have been examined in detail to characterize the inherent safety of the
reactor and to evaluate the instrumented safety system. The excursions
associated with reactivity accidents are inherently self-limiting by
virtue of the negative temperature coefficient of reactivity of the fuel.
110

(The negative temperature coefficient of the graphite is of little value
during a rapid excursion because of its long reaction time.) However, the
action of the safety system is required to prevent equipment damage in
the most severe cases considered.

One of the greatest possibilities for equipment damage in this system

is collapse of the control rod thimbles. However, the pressure excursions
“calculated for the core were very small, and they always occurred before
the temperature excursions reached the point where they might reduce the
allowable stress on the thimbles.

The reactivity incidents considered, along with a brief description
of how each might occur, are listed below. No attempt was made to evalu-
ate the credibility of the occurrences.

1. Uncontrolled rod withdrawal — Simultaneous withdrawal of all three
control rods, with criticality being attained while the rods are
moving in the region of maximum differential worth, is postulated.

2. Cold slug accident — The fuel circulating pump is started with the
core Jjust critical at 1200°F while the fuel in the external loop is
at 900°F.

3. Abnormal concentration of uranium during fuel addition — A slug of
120 g of 235U, added in the pump bowl, is assumed to go around the
loop without dilution and enter the core as a "front" uniformly
distributed to all fuel channels.

4. Displacement of graphite by fuel salt — Loss of an entire graphite
stringer (62 in. long) from the center of the core and replacement
by fuel salt is postulated.

5. Premature criticality during filling — The core is filled at the maxi-
mum possible rate with fuel salt in which the uranium concentration
has been increased 60% by selective freezing in the drain tank.
Criticality is achieved with the core 60% full.

6. Loss of fuel circulation — Failure of the power supply to the fuel
circulating pump and the onset of natural-convection circulation are
considered.

7. Loss of load — Instantaneous loss of all heat-removal capability at
the radiator is assumed.

In all the events considered, except the filling accident, the re-
duction in reactivity obtained by dropping any two of the three control
rods (with a O.l-sec time delay and an acceleration of 5 ft/secz) limited
the power, temperature, and pressure éxcursions to easily tolerable
values. In this context, tolerable values are those at which no damage
to equipment in the reactor cell occurs. In the case of the fuel-pump
power failure, an additional action, closure of the radiator doors, is
required to avoid freezing the coolant salt in the radiator.

Aside from the filling accident, the incident with the greatest
damage potential is the uncontrolled withdrawal of the control rods.
111

Figure 13 shows the results of a digital calculation of the transients
produced by the rod withdrawal with fuel C in the reactor. The power
at k = 1 was 2 mw, the minimum attainable with only the inherent G-n
neutron source present. If nothing were done to stop the rods, it is
clear that intolerable temperatures would be reached. A rod scram at

15 Mw can limit the temperature excursion to a tolerable value, even if
it is assumed that one of the three clutches fails to disengage. The
effect of dropping two control rods, while the third continues to with-
draw, is shown in Fig. l4. The actual response time of a prototype con-
trol rod was much shorter than that assumed for this analysis.

The filling accident represents a special case because the amount
of reactivity available is not limited to that associated with a normal
fuel loading. Prior to filling the reactor, the control rods will be
withdrawn so that the core is Just slightly subcritical when completely
full. This provides a capability for rapid insertion of negative reac-
tivity in the event of a power excursion during filling. However, the
reactivity that could be added in the postulated accident is greater

UNCLASSIFIED
ORNL-DWG 64-8177A

7

1900

1800 ////
AN /175 max

1600

Avims
LA
/

TEMPERATURE (°F)

1400

4300
J/ i R -

1200

500

400

POWER

300

POWER {(Mw}

200

100

J 1

Q 2 4 6 8 10 12 19 16
TIME (sec)

Fig. 13. Power and Temperature Transients Produced by Uncontrolled
Rod Withdrawal, Fuel C.
112

UNCLASSIFIED
ORNL-DWG 63-8478A
1320

1300

1280 (75 ) max

N Y

1240

1220 \\\

TEMPERATURE (°F )

1200

240

200 fi

160

POWER

120

POWER ( Mw)

80

40

JA

6 8 10 12 14 16
TIME (sec)

Fig. 14. Effect of Dropping Two Control Rods at 15 Mw During Un-
controlled Rod Withdrawal, Fuel C.
113

than that which can be compensated by dropping two control rods. As a
result it is necessary to stop fuel addition to limit this incident.

The occurrence of a high-flux signal causes, either directly or
indirectly, three automatic actions, any one of which will stop the fill
if it is in progress. (In the present design, this will occur at 20 kw
during filling, but the analysis is based on its occurrence at 15 Mw. )
These actions are closure of the helium supply valves to the drain tanks,
opening of the drain-tank vent valves, and opening of equalizing wvalves
between the drain tanks and the fuel loop. The least effective of these
actions is that of stopping helium addition because this allows a "coastup"
of the fuel level in the core and does not result in draining the fuel
from the core. In analyzing the accident, it was assumed that only this
one action actually occurred.

The calculated results of the postulated filling accident are shown
in Fig. 15. The initial power excursion was checked by dropping two con-
trol rods at 15 Mw. However, the fuel coastup, due to failure of the vent
and equalizing valves to open, resulted in a second critical event. The
second excursion was slow enough so that only minor temperature rises
were required, even with the reduced temperature coefficients applicable
to the partly full core, to check it. If no further action were taken to
insert the third control rod or to drain the fuel, the core would remain
critical at some low power level.

Biological Shielding

The biological shielding above the reactor and drain-tank cells
consists of 7 ft of concrete in two 42-in. layers with iron inserts to
fill the cracks between blocks. The bottom layer above the reactor cell
is barytes concrete, and the top layer is ordinary concrete. Both
“layers above the drain-tank cell are ordinary concrete. The shielding
around the sides of the reactor cell is, for the most part, 3 ft of
magnetite sand and water. Some additional shielding is provided by the
thermal shield around the reactor vessel. This shield is 16 in. thick
and is composed of about one-half iron and one-half water.

Extensive calculations have been made to evaluate the adequacy of
the biological shielding. In general, it appears that, within the limits
of the techniques available for evaluation, the biological shielding
should be adequate. A few potential weak spots were recognized during
the design. Provision has been made to stack additional concrete blocks
in these areas if surveys during the initial approach to power indicate
a need for them.

References

1. Most of this material has been reported in.detail previously. ©See:
P. N. Haubenreich et al., MSRE Design and Operations Report. Part
III. Nuclear Analysis, ORNL-TM-730 (Feb. 3, 1964).

2. MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, pp. 53-54.

114

UNCLASSIFIED
ORNL—DWG 63—8182A

1360 ’
TMAX
&
~ 1320
ul
xr
) /\ \
[
<1
z \
= 1280 ! e~
i /—""—___:
1240 /75/
TMAX _-_Z//
1200 [:il
|
T 294 Mw
4 “
3
E
s
x /\
ted
2
5 POWER //
a
!
I
1 !
[ N
4 \
________/
oL/
0 50 100 150 200 250 300 350
TIME (sec)
Fig. 15. Power and Temperature Transients Following Maximum Filling

Accident.
115

INSTRUMENTATTON AND CONTROL OF THE MSRE

Je. Re. Tallackson

Introduction

The purpose of the instrumentation and control system is to:

1. provide data to verify the designed performance,
2. provide information for reactor experiments,

3. provide safe, orderly operation of the reactor system.

I think you will conclude, as I proceed, that the MSRE is well
equipped with instruments to meet both operational and experimental re-
quirements., It is appropriate to point out that this is primarily an
experimental system, one of the first of its kind., Therefore, it would
be wrong to assume that the instrumentation and control equipment and its
application is completely representatiwve of any typical, future, molten-
salt power reactor system.,

Temperature

Temperature readings are the most important measurements in a system
like this. High salt temperatures, which are thermodynamically desirable,
must be limited and controlled lest they exceed limits imposed by mate-
rials and components; low salt temperature (freezing) can damage the
system; and finally, steep gradients must also be avoided. These, then,
are the reasons that the largest single concentration of instrumentation
and control equipment is concerned with temperature.

Virtually all temperature measurements of consequence are made with
Chromel-Alumel thermocouples. These are magnesium oxide insulated, In-
conel sheathed, units and are welded or clamped to pipes or vessels. Most
of the thermocouples are dual; that is, both the Chromel and Alumel wires
are contained in a single sheath. A few selected couples in critical ap-
plications are made from two separately sheathed, single wires. A ther-
mocouple failure, as by loss of a weld or clamp attachment, will then
produce an open circuit and thereby signal that a failure has occurred.

We have been liberal in our use of thermocouples, Sufficient couples,
with redundancy, have been installed, so that we can get temperature pro-
files of the tanks, the fuel and coolant salt piping, freeze valves, the
pumps, and the radiator. More than 50% of these thermocouples are in-
stalled as a substitute for experience with the system and in order to
obtain special data for a short time during early phases of the operation.
Most of the temperature information now being gathered should not be re-
quired to run the machine. ‘
116

Conventional methods for readout and data collecting from this number
of points, such as multipoint recorders, would be an exercise in futility.
We employ relatively new techniques to read temperatures and to use the
temperature data effectively. Figure 1 is a block diagram of this temper-
ature-measurement system.,

Almost all the thermocouples are routed to the patch panel, so that
any of the available readout methods can be applied to any given couple
which comes into the patch panel.

The Operations Monitor is used for control system interlocking and
to provide high or low limit alarms with a high degree of accuracy and
with good reliability. It is in use at all times during reactor operation
and not only indicates, with lamps, whether each temperature is within
limits, but alarms both visually and audibly if any temperature goes be-
yond its particular, preset high or low value. It employs modular con-
struction and is so designed that its capacity can be easily expanded or
contracted. Circuitry consists entirely of transistors and magnetic am-
plifiers and is expected to be considerably more reliable and maintenance-
free than comparable vacuum tube circuitry. The principal use of the
Operations Monitor will be to monitor important temperatures at freeze
valves in the fill and drain system and to provide control-system inter-

UNCLASSIFIED
ORNL-DWG 64-6476A

 SE— INDICATES NORMAL .
<l '—IN:D?JQTS-- N OR = HIGH ALARM
S — MONITO LOW AL ARM
I
o CONTINUOUS
: 400 THERMOCOUPLE [ :I'gg‘\ALLEESLAY
pr— ————
: PATCH INPUTS SCANNER | o | Ow ALARM
PANELS _J
[ [ = CAPACITY CONTINUOUS MONITOR
I 960 TC'S 70 DATA -——e AND PRINT-QUT
I INPUTS™ LOGGER e HIGH ALARM
I COMPUTER _ | LOW ALARM
|
I
| 48 PRECISION
| INPUTS™™]  INDICATOR
I
 S—
 —
L — 16 STRIP CHART
y INPUTS™™ RECORDER
18 STRIP CHART
INPUTS™ ™| RECORDER [ LOW ALARM
12 STRIP CHART oW ALAR
INPUTS™ ™|  RECORDER = LOW ALARM

STRIP CHART

— 24 oy L HIGH ALAR
INPUTS RECORDER M

-Fig. 1. Block Diagram of MSRE Temperature Measurement System.
117

lock signals based on these temperatures. The freeze flanges in the main
salt loops and at the steam domes in the after-heat removal system are-
also monitored by this equipment.

The Thermocouple Scanner gives the operator an immediate and contin-
uous picture of temperature profiles in the system, Figure 2 is a diagram
of the system. You will note that this device operates in a straight-
forward way; that is, a high speed selector switch is used to scan a group
of 100 thermocouples, each couple being monitored every 50 msec and com-
pared with a reference thermocouple. The difference is displayed on a
17-in. oscilloscope after suitable amplification. Temperature differences
of 5°F are immediately apparent, and, if any couple differs by 50°F from
the reference couple, it is annunciated. You will note that the scanner
does not read absolute temperatures, only differences. A second thermo-
couple, adjacent to the reference couple, is therefore used with a preci-
sion temperature indicator to obtain the reading of the reference temper-
ature.

The scanner will be most useful when the salt system, or a part of
the system, 1s undergoing an operator-produced change in temperature, or
when it becomes necessary to make certain that a particular piping loop

UNCLASSIFIED
CRNL-LR-DWG 58910R3A

¥*
THROW CHOPPERS, OPERATING REFERENCE

to-—0 ONE COUPLE ABOVE THE '

o REFERENCE THERMOCOUPLE~—1_

ANIS) : ] I_,—._.—\_,—\J
o5 | TWO DOUBLE-POLE, DOUBLE - A

SYNCHRONOQUSLY

[—

|
|
|
]
‘ — l
v O
i
2 | —— | i > o ONONONGNO)
a k—— |
! :
0C
g) D — ! AMPLIFIER 17-in. 0SCILLOSCOPE
O | '
< 0o | % REFERENCE APPROXIMATELY
x (Y7 i EQUAL TO SCANNED THERMO-
T ! COUPLE SIGNAL
= |
3% :
oy
S0
o <L
ww
T
=z
© ALARM —
MERCURY JET SWITCH -

DISCRIMINATOR ANNUNCIATOR

(100 CONTACTS )
ALARM SETPOINT :Q
+ 500F
REFERENCE THERMOCOQUPLE

ON REACTOR PIPE

7\

Fig. 2. Block Diagram of Temperature Scanning System.
‘118

or tank 1s at a uniform temperature, so that thermal gradients are mini-
mized to unquestionably safe values. A typical situation occurs just be-
fore and during the filling of either the fuel- or the coolant-salt loop.
The operator will be able to check temperatures around the loop and to
make adjustments on the electrical heaters, as indicated by deviations

on the oscilloscope.

Both the Operations Monitor and the Scanner hawve been thoroughly
tested on experimental salt systems and test setups.

The Logger-Computer, which I will describe in more detail a little
later, is capable of combining the operational features of both the Oper-
ations Monitor and the Scanner, but omitting the oscilloscope display, !
and, as we have 1t programmed, it is not used for control circuit inter-
locking.

More Important Process Measurements

Salt levels and flow rates are the most challenging measurement prob-
lems in the MSBRE. Salt level in the drain tanks is measured in two ways.
Pneumatic weigh cells are used to give a continuous indication of each
tank's contents, and a two-point level detector tells us when salt level
is at elther 10 or 90% of tank capacity and gives us reference points for
the weigh cell instrumentation. The level detector measures the voltage
drop through the salt from electrodes suspended inside the tank. The
weigh cell instrumentation is also used to provide control system inter-
locks when the reactor is being filled.

Weld-sealed pressure transmitters are used to measure helium pres-
sures in lines to the drain tanks, and in the fuel pump bowl and the over-
flow tank. Salt levels in the fuel pump bowl and overflow tank are meas-
ured with submerged bubbler tubes and weld-sealed differential pressure
transmitters. The output signal from all of these transmitters is elec-
trical, and problems of transmission through the containment are thereby
minimized.

.We have also developed a float-type level transmitter, which gives
a direct measurement of salt level without producing any major containment
problems (see Fig., 3). No new concepts are involved. The float simply
actuates the armature of a differential transformer which produces the
signal. The magnetic coupling of the armature to the transformer coils
is through the containmment walls. The developmental problems involved
pertain to environment, and results using this device are so encouraging
that we have installed one on the coolant-salt pump bowl and will put a
similar unit on the spare fuel-salt pump. We have operated one of these
transmitters for more than two years on a test loop., Ninety percent
(18,000 hr) of this time was at a temperature of 1250°F and 5% (1000 hr)
at 1300°F. During this test period, the calibration shifted 0.2 in. (4%).
A temper?tgfie change from 1000 to 1300°F causes a shift in indication of
0.3 in. (6
119

UNCLASSIFIED
ORNL-DWG 64-7151A

FIBERFRAX

ARMCO
IRON CORE

DIFFERENTIAL
" TRANSFORMER

’:>%( CORE TUBE
PUMP /A
BOWL
g FLOAT
5 " INOR-8 OR GRAPHITE
5 /

-——— FLOAT CHAMBER

INOR-8
CAN

Fig. 3, MSKRE Float-Type Level Instrument.

A number of very small helium flows in the system are monitored with
a flow element consisting of a packed bed of glass beads. Flow rate
through this element is directly proportional to pressure drop, and this
device is less subject to plugging and temperature effects than a capil-
lary type element is.

Coolant-salt flow rate is measured by a Venturi and a specially de-
signed differential pressure transmitter (Fig. 4). This transmitter is
unusual in that intermediate NaK-filled capillary tubes, isolated by slack
diaphragms, are used to transmit pressures from the molten salt to a sil-
icone-o0il-filled pneumatic transmitter of conventional design. A strain
gage transducer has been added, so the output signal is electrical rather
than pneumatic.

Nuclear Instruments

During normal operatioh from startup to full power, the MSRE will
require three types of neutron sensors, namely: (1) fission chambers,
(2) compensated ion chambers, and (3) safety chambers.
120

UNCLASSIFIED
ORNL-LR—DWG 78815

NaK FILLED @/pmcsss FLUID
ARMORED CAPILLARY

_________ RN =
___________________________ NN
\ N PROCESS
SEAL CONNECTION

ELEMENT (HIGH}

SILICONE
OlIL FlLLED—————

\\\\'/
;' 7 /
/ = ! S

DIFFERENTIAL- \/‘q N J e
PRESSURE-SENSING el _| 1 oy
DIAPHRAGM B P— T 7 et

\ \ Z j !

ZERO
ADJUSTOR

STRAIN GAGE
TRANSDUCER

FORCE BEAM
SEALING
BELLOWS
RANGE
ADJUSTOR

SEAL ELEMENT {LOW)

VENDOR: TAYLOR INSTRUMENT CO.
MATERIAL: INOR-8 AND

STAINLESS STEEL TR SR - PROCESS
RANGE: 0-50 TO 0-750 in. H.0 N CONNECTION
DESIGN TEMPERATURE: 1250°F
QUTPUT: 0-25mv

RN RN

Fig. 4. All-Welded, High-Temperature, NaK-Filled Differential-
Pressure Transmitter.

For the initial critical experiments, four temporary channels of BF,
counters will also be used. With the exception of two of the temporary
BF53 channels, all of the chambers are installed in a single penetration
(Fig. 5). This penetration is 3 ft in diameter, is water filled, and con-

tains a total of 10 internal guld.e tubes, which loosely retain and locate
the chambers.

Figure 6 is a plan view of the reactor and the penetration.

During the early stages of design we provided two wvertical holes in
the shield for BF3 counters and a third hole for the neutron source., Sub-
sequently, analysis by Haubenreich and Engel showed that the majority of
the neutrons at these locations will be source neutrons (as opposed to
fission neutrons) which will have gotten to the counters by traveling
around in the annular space between the shield and the core vessel. TFor

that reason, we are putting the other BF5 chambers directly opposite the
source tube in the penetration.
121

241t 2in,

._i
|

NEUTRON INSTRUMENT

JUNCTION BOX ]

HIGH BAY AREA

/WATER-SANO ANNULUS FACE

REMOVABLE CONCRETE SHIELDING

FINISHED FLOCOR
EL. 852 fTOin.\ 7

=]

UNCLASSIFIED
ORNL-DWG 64~-625

¢ FLANGE FACE

41° 50’

FACE OF INST.
" SHELL FLANGE
EL. 851 f Bin.

LEVEL
849 ft 8in.

REACTOR
THERMAL
SHIELD

CHAMBER GUIDE
TUBE (TYPICAL)

_ ¢ END OF 48-in. CD SLEEVE
EL. 841t 1 in.

ELEV.
8371t 7in.

48-in.OD OUTER SLEEVE
ASSEMBLY {WITH EXPANSION JOINT)

T~EXP JOINT

¢ CONTAINMENT 5
PENETRATION EL. 834 ft 5@ in.

¢ EL.830 ft 3in.

CHAMBER
(TYPICAL)

INSULATION
OQUTER STEEL
SHELL

REACTOR CONTAINMENT
CELL

Fig. 5. Elevation View of Nuclear Instrument Penetration.
UNCLASSIFIED
ORNL-DWG 64-984

UPPER END OF NEUTRON
INSTRUMENT PENETRATION
/
/ /
/

REACTOR
CELL
/
/ /
/

NEUTRON
INSTRUMENT

BF,
CHAMBER—__
BF5 CHAMBER

PENETRATION

TUBE
U/
— ! \ \
BF3

\\\\\\\i!llg'
REACTOR VESSEL - ~d
AP
4 - -~
| 7 Vv \\ ‘o
I SR /7 N \ CHAMBER TUBE
/ \
|
/
/

\
7 \
[ . |
| 4 !
\ \ !
N /
N \_______
N - P
\"——___

SOURCE TU BE—//O

THERMAL
SHIELD

@ ®

Plan View of Core Vessel, Showing BF3 Counters in Shield

. Fig. 6.
and Source Location.
123

The fission chambers are the input sensors for the wide range count-
ing system. This was described in an information meeting about two years
ago. (see Fig. 7.). The fission chamber is moved by a servomotor, sO
that it maintains a constant, preset count rate for all values of reactor
power sufficient to produce this count rate. If the attenuating medium
surrounding the chamber produced ideal exponential attenuation, the log
of reactor power would be proportional to the distance the chamber is
withdrawn from its innermost position and the reactor period would be
proportional to the velocity. ©Since this ideal situation is never attain-
ed, an adjustable trimming device, the function generator, is included
in the servo to compensate for the deviation from true exponential
neutron attenuation.

UNCLASSIFIED
ORNL-DWG 64-1218R
—_———

| FUNCTION ; ]
| GENERATOR
| |

POSITION | |
POTENTIOMETER - ]
CHAMBER POSITION
SERVO AMPLIFIER ég
MOTOR \\\‘J
TACH %gDEMAND
LINEAR
»| COUNT-RATE
> T-R PERIOD
METER AMPLIFIER
LOG
PULSE
- NT-RA
AMPLIFIER COUMETERTE
A.[ SUMMING
PREAMPLIFIER AMPLIFIER

+——FISSION CHAMBER
POWER PERIOD

Fig. 7. Block Diagram of a Wide-Range Counting Channel.

The actual physical arrangement as used in the MSRE is shown in Fig,
8. Two compensated ion chamber channels are used to provide the most ac-
curate flux information and for the input signal to the servo controller.
Figure 9 is a highly simplified block diagram of these two channels.

By means of range switching on-the picoammeters, an overall range
of seven decades can be covered without changing chamber position or am-
124 .

plifier gain, In contrast to the wide range channels previously dis-
cussed, output is linear, and each range setting covers only half a decade,
thereby giving good resolution. Output from either chamber and its range
setting are continuously indicated and recorded on the main control board.

A third group of three chambers provides flux information, used as
inputs to the safety system. These are uncompensated, neutron-sensitive
ion chambers and have no function other than safety. They are effective
in the power range from 1 kw to above 15 Mw,

UNCLASSIFIED

ORNL-DWG 64-$219 N o-

MOTOR DRIVE ASSEMBLY i
MOTOR

UPPER LIMIT SWITCH ' .

SLIP CLUTCH

LOWER LIMIT SWITCH

POSITION
SENSING TRAIN DRIVE TuBE
GEARS
WATER LEVEL

LEAD SCREW

GUIDE BLOCK

THRUST TUBE

CABLING RIG

PREAMPLIFIER AND -
FISSION~-CHAMBER ASSEMBLY \/

Fig. 8. Positioning Mechanism for Wide-Range Counting Channel.
125

UNCLASSIFIED
ORNL DWG 64-6480A

RANGE
CHANGER

POWER
SUPPLY (?

COMPENSATED
ION CHAMBER

RANGE

2-PEN
RANGE RECORDER

CHANGER

T

|

: CONSOLE CHANNEL
| SELECTOR SWITCH
: TO SHIM-REGULATING
|

POWER‘ )
SUPPLY

@ 0 FLUX ROD
= d \

GOMPENSATED . SERVO
-
ION CHAMBER CONTROLLER

Fig. 9. MSRE Nuclear Instrumentation — Linear Flux Ievel Channels.

Rod Drives

Figure 10 is a diagram of a control rod drive. The articulated con-
trol rod is suspended from a sprocket chain driven by a gear-reduced ac
servomotor. The three drives are identical; they include coarse and fine
synchros to transmit rod position, an irreversible worm and gear set, and
a potentiometer for the position input signal to the safety system. You
will note that two parallel power transmission paths hawve been provided
to the sprocket chain. One is via the electromagnetic clutch; the alter-
nate drive path is through an overrunning clutch. The latter always pro-
vides a positive drive in the "insert" direction, and also maintains a
positive connection to the drive, so that reverse torque at the output
sprocket cannot produce accidental rod withdrawal, Any one of the three
rods is capable of being used as the servo-controlled shim-regulating
rod, by merely changing the electrical interconnections.

The shock absorber (shown in Fig. 11) used with this drive unit is
somewhat unusual. It consists of a cylinder and a piston or plunger with
the unique feature that the working "fluid" is composed of 3/32-in.-diam
steel balls. The shock absorber assembly moves up and down with the rod.
When the rod is scrammed, the cylinder halts at the stop blocks at the
bottom of the housing. The rod and the plunger continue to move down-
wards, and the kinetic energy of the rod is absorbed as the steel balls
are forced through the narrow annulus around the plunger into the upper
part of the cylinder. This device is a development of the Vard Corpora-
tion, who are building the rod drives.
126 -

UNCLASSIFIED
ORNL-DWG 63-8394

~--— FINE

TO POSITION READOUTS SYNCRO NO.2
IN CONTROL ROOM -— COARSE 60° PER INCH
OF ROD MOTION

POSITION
POTENTIOMETER

ROD POSITION
INPUT TO SAFETY SYSTEM -=

TO SAFETY
STOTEM = - REDUCTION _ 2 -
GEARING _
FAN SYNCHRO NO. 1 -
MOTOR 5° PER INCH
OF ROC MOTION
TACH
SERVO
MOTOR
]
DRIVE
ELECTROMECHANICAL SPROCKET
CLUTCH .
| I
b —e—
n |
I
REDUCTION EQUAL RPM o L.— SPROCKET
GEARING F 1-TO-4 GEARS : CHAIN
INCLUDES il } :
REVERSE | e;’:[:
LOCKING T4
I
OVERRUNNING AIR FLOW g :
CLUTCH TO COOL ROD {f |
7 I
A F -
{
FLEXIBLE ||
TUBULAR ROD 1 .
SuPPORT-—"[| | .
ke
V=0.5in./sec PR .
POISON ELEMENTS 1 i
J\ l b
THIMBLE ~— ;
\ | )
HORIZONTAL U
GRAPHITE BARS J J | GRID PLATE
N
CORE VESSEL—*

Fig. 10. Electromechanical Diagram of Control Rod Drive.
127

UNCLASSIFIED
ORNL-DWG 64-983

L

CARRIAGE

AR INLET TUBE-— |

DRIVE CHAIN

BUFFER RETURN SPRING—__]

BALL RESET SPRING ~—— _|

CYLINDER

PLUNGER————— 1

Hp-in-DIAM

STEEL BALLS——— _ ]

CARRIAGE TRACK

ROD DRIVE HOUSING ——— |

TOW BLOCK

THIMBLE

ELEMENT

POISON

MSRE Control Rod Drive — Shock Absorber Assembly.

Fig. 11..
128

Scram is effected by opening the clutch coil circuit, which disen-
gages the clutch. :

The curves in Fig. 12 give a comparison of actual scram performance
with an arbitrary performance specification based on an acceleration of
5 ft/sec® and a maximum release time or delay of 100 msec between the ex-
istence of scram conditions and actual disengagement of the clutch. Anal-
ysis has shown this to be adequate for any contrived situation. You will
note that there is a wide margin of actual acceleration over that required.
Th7 acgual acceleration obtained from tests of a prototype unit was 13
ft/sec”,

UNCLASSIFIED
ORNL-DWG 64— 1109R

I I I A
) 1\ AN

X
40 o

DISTANCE (in.)

50 \ .

60

0 0.25 0.50 0.75 1.00 1.25 1.50
ELAPSED TIME (sec)

CURVE A-REFERENCE CURVE OF SATISFACTORY SCRAM PERFORMANCE;
BASED ON ACCELERATION OF 5 ft/sec?2 AND RELEASE TIME OF 0400 sec

CURVE 8-SCRAM PERFORMANCE FROM TESTS OF JAN. 27-28, 1964

Fig., 12. MSRE Control Rod Height vs Time During Scram. Curve A —
reference curve of satisfactory performance, based on acceleration of 5
Tt/sec® and release time of 0.100 sec. Curve B — scram performance from
tests of Jamuary 27-28, 1964,
129

The curves in PFig. 13 have been calculated from the insertion dis-
tance vs time curves seen in Fig. 12 and the reactivity worth of two
rods, starting from an initial withdrawal height of 51 in. Again we
have a substantial margin of performance over that deemed adequate.

UNCLASSIFIED
ORNL-DWG 64-2146

— —_—

—a ~ —] _REFERENCE CURVE PROVIDING
] | ADEQUATE PERFORMANCE
™ ~DURING SCRAM

0.01
N | ]
\’CALCULATED PERFORM- ]
ANCE BASED ON ROD

]
Ll
i 002 \GRO_TESTS OF PROTO- N
H TOTAL ROD WORTH: TYPE SYSTEM AND CAL- [
z 0041 Bk | \CULATED ROD WORTH DATA
<« 0.03 —NITIAL WITHDRAWAL :
P 51 in. \

0.04 AN

\--_
0.05
of 02 03 04 05 06 07 08 09 10

ELAPSED TIME AFTER SCRAM SIGNAL (sec)

Fig. 13. MSRE Control Rods — Reactivity Insertion vs Time After
Scram,

o Safety System

The MSRE safety system is composed of reliable and redundant in-
strumentation, designed to protect the system and to ensure containment.
Figure 14 is a block diagram of the system. We hawve extended the scope
of the MSRE safety system beyond direct considerations of conditions in
the core. The string.of blocks on the left side of the diagram are con-
ditions which initiate safety system action. Two of these, temperature
and flux, call for a rod scram. Most of the remaining inputs and safety
actions are concerned with various process instruments. The individual
instruments and related equipment, which provide the input-output func-
tions shown in the diagram, have been selected or designed with reliable
operation as the first consideration., Redundancy, in the form of either
two or three independent input channels, is used to produce the output
action required. Periodic, in-service testing of all or a major portion
of each channel is provided. For example, Fig. 15 shows three independent
channels of temperature input information as we use it to scram the con-
trol rods or to drop the radiator doors. The instrumentation used to
convert thermocouple emf to a usable signal is conventional, but for in-
service testing we have added an additional pair of thermocouple junctions
in series with the measuring junction on the pipe. During normal opera-
tion, both the added junctions are at the same temperature and hence
produce no net signal. We can test the rod scram input circuitry by ap-
plying electrical heat to one of the added junctions, thus simulating a
temperature rise at the reactor outlet. A temperature drop can be simu-
lated by reversing the polarity of the artifically heated junction.
CONDITION

FLUX

130

/

UNCLASSIFIED

ORNL-DWG 64-637

CORRECTIVE ACTION

>15 Mw

- SCRAM

REACTOR OQUTLET
TEMPERATURE > 1300°F

* THE BYPASS VALVES AND
THE VENT VALVES ARE
REDUNDANT SETS, EITHER
WILL PROVIDE PRESSURE
RELIEF IN DRAIN TANK
OURING A FUEL DRAIN

! ]

OPEN x
BYPASS VALVES

THAW

ALL RODS

DRAIN

FREEZE VALVES

OPEN x
VENT VALVES

f

ANY ROD NOT ABOVE

REACTOR

CLOSE MAIN He SUPPLY

FILL POSITION

FUEL PUMP BOWL

VALVE TO DRAIN TANKS

" GLOSE He SUPPLY VALVES

PRESSURE > 2 psig

FUEL PUMP BOWL He

USED TO FILL REACTOR

OPEN FUEL PUMP BOWL

PRESSURE >50 psig

FUEL PUMP OVERFLOW

TANK LEVEL >20% FS

ACTIVITY IN REACTOR
CELL AIR-

COOLANT PUMP OFF-GAS

ACTIVITY =20 mr/hr

VENT VALVE

CLOSE He SUPPLY VALVE TO

FUEL STORAGE TANK

CLOSE BLOCK VALVES
He LINES

He SUPPLY PRESSURE

CLOSE BLOCK VALVE IN CELL
EVACUATION LINE

< 30 psig

—

ACTIVITY IN

GLOSE He SUPPLY LINES TO IN-CELL
LEVEL INSTRUMENT BUBBLERS

OFF-GAS

TEMPERATURE AT RADIATOR

OUTLET < 900°F

LOSS OF COOLANT

CLOSE BLOCK VALVE IN MAIN
OFF-GAS LINE TO STACK

CLOSE LUBE OIL SYSTEMS
VENT TO OFF-GAS VALVES

SALT FLOW

ACTIVITY IN He LINE TO

CLOSE RADIATOR DOORS

REACTOR CELL

IN-CELL BUBBLERS >

CLOSE LIQUID WASTE
SYSTEM BLOCK VALVES

PRESSURE >2psig

ACTIVITY IN IN-CELL

CLOSE INSTRUMENT AIR
LINE BLOCK VALVES

COOLING WATER SYSTEM

?  Fig.

14,

Block Diagram of

CLOSE IN-CELL COOLING

WATER BLOCK VALVES

MSRE Safety

oy stem.
131

UNCLASSIFIED
ORNL-DWG 64-626

CURRENT
ACTUATED SWITCHES

~ =

TEST ASSEMBLY: CONTROL ROD
0. ELECTRICAL Tosfs"}zfl* ALARM  REVERSE
HEATER
al SEE FIG. 2.0 0
I b. TEST THERMO- DATA
| COUPLE LOGGER
{SEE NOTE 1.} T
| - CURRENT
! = |_, " ACTUATED L‘| }_' L‘I l—,
R N R SWITCH
2 s o )
T it ISOLATION 7
N | AMPLIFIER 00 0
MEASURING CONVERTER L
THERMOCOUPLE L o ol
> g o o 400 Q .
1 o o
< o o )l [ ] 10-50 ma
N 10 TC 50 ma METER

CHANNEL NO.2
SAME AS ABOVE

CHANNEL NO.3
NOTE:
{. POLARITY SHOWN FOR TEST
THERMOCOUPLE 1S THAT USED TO
SPARE TEST SYSTEM RESPONSE TO A
TEMPERATURE INCREASE

NN

———— REACTOR OUTLET PIPE
LINE NO. 100 AND RADIATOR
OUTLET PIPE, LINE 202

Fig. 15, Typical Temperature Measuring Channels Used in MSRE Safety
System,

The three channel systems used to ensure containment are worth noting.
Figure 16 shows the instrument setup used to close the block wvalves in
the instrument air lines to in-cell pneumatic instruments when cell pres-
sure exceeds 2 psig.

This 2 out of 3 system uses three independent pressure switches with
diverse power sources to signal high pressure in the cell. Agreement of
any two will initiate wvalve closure. This allows one nonoperating channel
for either maintenance or testing. In service, routine testing is accom-
plished by manually opening, one at a time, the hand valves which lead to
the nitrogen bottle and observing the pressure gages in the solenoid ma-
trix. A similar technique is used with the three instrument channels used
to measure the pressure of the inlet helium which enters the containment.

Rod scram is produced either by excess reactor outlet temperature
or by excess flux. Figure 17 is a dliagram, somewhat simplified, of the
rod scram system. Each channel is independent. The neutron flux channels
are continuously monitored to verify that the chambers are supplied with
the correct voltage. The monitor unit provides means for testing channel
response by simulating an excess flux signal. This is a 2 out of 3 coin-
cidence system, and a rod scram will also be produced if any two input
channels malfunction., This would occur if there were a loss of voltage
to any two ion chambers.
132

UNCLASSIFIED
ORNL-DWG 64-6477A

EMERGENGY EMERGENCY
PRESSURE 2 AC +TVA T 0C
SWITCH — - .
\\’/, . I// \\ ‘/ \\
——t[::] T PS-l ] T,ps2 —\m T PS-3
N N L ~

VALVE

SOLENOIDS\ 13 sv SV Sv SV Sv SV
A A2 B1 B2 cl c2
AC NEUTRAL

DC NEGATIVE

PRESSURE
C? GAGE

INSTRUMENT
Al

. R PRESSURE
\ 25 ) : GAGE
SINTERED gy v Al TVA
DISK SNUBBER ) B>

REACTOR

CELL \ E'g—"O‘ _ L~ VENT
P
TVA ad +DC
v _5 - SV
5 E@L% tor gy
25
PI
+DC AS kva sv
SV =3 L1 A2
AIR SUPPLY Ct \
TO IN-CELL
INSTRUMENTATION
‘ “'WALVE SOLENOIDS
\_HEADER

Fig. 16. MSRE Redundant Instrumentation for Block Valves.

The electronic instrumentation making up this system is hewly de-
signed. It uses solid state modular construction and was not designed
specifically for the MSRE or any other particular reactor. Rather, it
is highly versatile in its capability of being adapted to a wide variety
of design situations and is suitable for use in a wide variety of system
configurations. For example, both the MSRE safety system and that of the
High Flux Isotope Reactor use the same nuclear instrument components.

Figure 18 diagrams the system used for emergency closure of the ra-
diator doors. A conservative analysis has shown that the coolant salt
in the radiator can freeze in less than a minute in the worst situation.
Since freezing and subsequent thawing may damage the radiator, a reliable
system is employed to close the doors if the radiator outlet temperature
becomes too low or if the coolant salt ceases to flow. The decision to
scram the doors does not involve safety or containment; it was based
solely on economic considerations.

(L
TYPICAL FLUX
AND
TEMPERATURE <
SAFETY INPUT
CHANNELS

133

————— —} ALARM
| I
CHAMBER
|__| POWER f VOLTAGE M,
| [SUPPLY] | |mMONITOR AND TEST
NEUTRON | | :
SENSITIVE |
ION CHAMBER | | .
C——_ | FLux| | TRIP s _
" I TRIP | AMPLIFIER 1
i |
[
T'I
-—ne
THERMOCOQUPLE FROM | — M=
EMF TO
<] CURRENT owmon [ onanngrs| e
\ CONVERTER 0.2 | —T,
~—
FRoM | —Mz™™
TAND ¢ o
CHANNELS 3"

MHg. 17.

UNCLASSIFIED

ORNL-DWG 64-6479A

COINCIDENCE
RELAY MATRIX
(2 OUT OF 3)

~ St
i L COINGIDENGE
LN S2—=| RELAY MATRIX
c3_a] (2 0UTOF3)
N Y
; GCOINGIDENGE
I 52— RELAY MATRIX
F
s3] (2 0UT OF3)

Diagram of Rod Scram System.

ROD DRIVE
CLUTCHES

il
UNCLASSIFIED
ORNL~ DWG 63-8389
COOLANT
PUMP | SAFETY
MOTOR CIRCUITRY
MAGNETIC PICKUPS PUMP SPEED | Sy 5y OR Sz Sy OR Sz
INSTALLED N »7 [ - MONITOR 3 |NDEPENDENT )
SPARE (o] - o
+—fr] [y : R Fy RELAY  —
— {(a)2 FLOW —  MATRIX
PUMP SPEED | S 5 (b) SPEED F> N
MONI TOR - > -
MOTOR TO PUMP : L :
FT :
_— SHAFT COUPLING VENTURI _ o ool X ‘
¢ HEAT 'DIFFERENTIAL Eb
PUMP BOWL EXCHANGER PRESSURE TRANSMITTERS E
T COINGCIDENCE
A—P—= | 3INDEPENDENT [ 7| ST or3) [ N
| —————e > TEMPERATURE —_—
: M T3 RELAY
1 ) SIGNALS -
m | RADIATOR MATRIX 7
CLUTCH BRAKE
THERMOGOUPLE UPSTREAM DOOR
CIRCUITRY
; NOTE:
| EMERGENCY CLOSURE PRODUGED BY INPUT
f SIGNAL COMBINATIONS AS FOLLOWS: .
- {a) S; OR S, + F, E]nfl
' ‘ (6) S, OR S+ Fy E
CURRENT (c) F+ Fy —
THERMOCOUPLES —____| ACTUATED (d) ANY TWO OF THE THREE TEMPERATURE
(LINE 202 : T SWITCHES ——, INPUT SIGNALS E .
. 7 }
o \\ ? CLUTCH BRAKE
ALARM . .
DOWNSTREAM DOOR
= 10 READOUTS
CONVERTERS '
. o ALARM
!
Fig. 18. Diagram of System Used for Emergency Closure of the Radiator Doors.

134

135

In addition to the safety instrumentation, there are several impor-
tant control system interlocks which are not designed to as high a degree
of reliability as the safety-grade protection. For example, there is a
derivative signal from the drain tank weigh cells which is used to limit
the reactor fill rate. We automatically insert or reverse the control
rods if the reactor outlet temperature exceeds 1275°. This temperature
is halfway between the design-point full-power temperature and the scram
temperature., We also get an automatic reverse if power exceeds 12 Mw,
if fuel level is too high in the overflow tank, or if, during startup,
we are on a period of less than 10 sec or power exceeds 1.5 Mw,

Reactor System Control

Operation of the MSRE can be logically divided into two principal
modes, and we use the control system to make the division. The first,
or "prefill," mode includes those operations which are preparatory to nu-
clear operation, such as transfers of salt from the storage tank to the
drain tanks, transfers among the drain tanks, circulation of helium in
the fuel and coolant salt loops, and testing of all the wvarious elements
of the system. The last includes testing of pumps, cooling air and water
systems, radiator doors and blowers, etc. The second mode, which we call
"operate,"” includes nuclear operation of the system and aspects of opera-
tion, such as filling the system, which can lead directly to nuclear op-
eration., :

Figure 19 is a block diagram which shows,-in a very broad sense,
the more important restrictive and permissive interlocks in the control
system.

The reactor may be operated either manually or with automatic con-
trol of the shim regulating rod. During the early phases of the design,
we held the hope that the negative temperature coefficient would supply
all the control required when the reactor was in the power range above
1 Mw. Analog simulation showed us that this would not be the case and
that supplementary, external control would be required. This is the price
we must pay to produce low-power-density design and to use an existing
facility which imposes unfavorable piping geometry from the standpoint
of control.

This reactor system is characterized by high heat capacity, long
transport lags, low power density, and, consequently, a very sluggish
temperature response to load and power changes. About half of the temper-
ature coefficient is in the graphite moderator, which occupies nearly 80%
of the core volume, It takes the graphite between 12 and 20 min to reach
temperature equilibrium following a step change in reactor fuel-salt tem-
perature. The transport lag from the radiator to the core is nearly 1/2
min in each direction; that is, it takes 25 sec for a change in radiator
load to show up as a temperature change in the core, and the same length
of time for a temperature change in the core to affect the loading at the
radiator.
136

UNCLASSIFIED
ORNL-DWG 64-6478A

OPERATE
f. PROHIBITS FUEL TRANSFERS

PREFILL AMONG DRAIN TANKS
i, REACTOR IS EMPTY 2. ALLOWS REACTOR FILL
2. REACTOR FILL AND DRAIN 3. PROHIBITS JUMPERED
VALVES FROZEN SAFETY SYSTEM CONTACTS

3. PERMITS SALT TRANSFERS
AMCNG DRAIN TANKS

4. PERMITS HELIUM CIRCULATION
!N FUEL AND COOLANT LOOPS

5., PERMITS JUMPER BOARD ———0 o—
BY-PASSING OF SAFETY
SYSTEM FOR TESTING

START RUN

. DRAIN VALVE FROZEN . DRAIN VALVE FROZEN

1 1

2. REACTCR FULL 2. REACTOR FULL

3. LOW POWER, <! Mw 3. POWER > 1 Mw

4. SERVO CONTROLS 4. SERVO CONTROLS CUTLET
FLUX ONLY SALT TEMPERATURE

Fig. 19. MSRE Control System Modes.

If there were no external control system, the analog simulation
showed that in some situations the reactor would oscillate with very long
periods around the average power demand level., These oscillations, while
not dangerous, would be intolerable from an operating standpoint.

The controller which was developed as a result of these analog
studies is diagrammed in Fig. 20. When the reactor is operating as a
load~-following power generator, the servo controller is used to hold the
core outlet temperature constant. This mode of control is optimum ther-
modynamically and is what we would do if we were using the reactor to
generate steam.

The controller is composed of four operational amplifiers; three are
used for summing and for sign inversion, and the fourth is used as an in-
tegrator. Four inputs are used to attain control: flux, inlet tempera-
ture, outlet temperature, and outlet temperature set point. These inputs
are combined. The controller can thus compare the reactor power, repre-
sented by flux, with the load, represented by the temperature drop across
the core, and compute an error signal, calling for regulating rod action,
should the two be different. The curves at the bottom of Fig. 20 show
the open loop responses of the controller to step changes in flux, inlet
temperature, and outlet temperature set point. This response curve for
set point temperature is academic, since we use a low speed, motor driven
potentiometer to make certain that we can't make step changes in tempera-
ture demand. The integrator produces, on a long time scale, a subtractiwve
correction to the initial step inputs and gradually causes a reduction in
the error signal. This scheme, using a long time constant integrator to
reduce the error signal, also makes the controller insensitive to slow
drifts in the inlet temperature and flux sensors.
137

UNCLASSIFIED
ORNL-DWG 64-6206

=7
SUMMING AMPLIFIER (TYPICAL),
OUTPUT IS NEGATIVE SUM OF
INPUTS
T - INTEGRATOR
~¢ S
%],
€> +3, INSERT ROD
€< —§, WITHDRAW ROD
z €
T = FUEL SALT TEMPERATURE, CORE INLET
7 = FUEL SALT TEMPERATURE, CORE OUTLET
¢ =NEUTRON FLUX
[@ 45 = SET POINT, FUEL SALT OUTLET TEMPERATURE
zl,, |
0 = TIME
INSERT
€ 0 -
WITHDRAW |
7
A
INSERT [““~———-___________
€ 0O oo
WITHDRAW
O E o
INSERT r\
€0 -
WITHDRAW

Fig. 20. Block Diagram of Servo Controller, Open Loop Responses
Step Inputs. _
138

Below approximately 1 Mw, the controller is used as a simple flux
comparator, and temperature control is by manual adjustment of system
heaters., The results of an analog simulation of the reactor system using -
this controller are presented in Fig. 21.

Each regulating rod is capable of producing a total change of 2 to
3% in reactivity — considerably more than we care to put at the disposal
of this or any other rod control servo. We have added a limit switching
mechanism which reduces the effective stroke of the servo-controlled rod
to an acceptable value.

UNCLASSIFIED

ORNL-DWG 64—-620 -
4250 T T T T T T T -
///REACTOR OUTLET - DEMAND INCREASED
TEMPERATURE L— TO 10 Mw
1225 bl : I s A\
= \\ -
1200 - A
e
= [Aé—fl—RAmATORINLET_SAU' \
o 1475 g TEMPERATURE °
g l<2£~"——RADIATOR OUTLET SALT \
150 TEMPERATURE
= // ==\
a \
IflLJ 41425 / \ AN
1400 / \\
1075 l 7 \\\
4050J <
1025
10 I —
I\ | yd
8 | -
3 — DEMAND - (RADIATOR HEAT ! 4
= . . REJECTION)- — / y
p \ | l | | .
5 I \(fRESPONSE—REACTOR | //
= o Ul POWER (¢) !
a
N\ | :
2 i (| -
| N I |
o | \ I/

0 50 100 450 200 250 300 350 400 450 500 550 600 650
TIME (sec)

Fig. 21. Results of Analog Simulation of the Reactor System Using
the Servo Controller.

The diagram on the left of Fig., 22 is strictly conceptual and shows
what the limit switches accomplish, The motion of the control rod ac-
tually is remotely reproduced in the control room by a servomotor-driven
cam which actuates the 1limit switches. A second motor is also coupled
to the cam by means of a differential gear and locates the span of the
limit switches with respect to the rod position, when shimming with the
regulating rod.
ROD DRIVE MOTOR,
SERVO OR MANUALLY
CONTROLLED

GEARBOX /"q

UPPER
LIMIT
SWITCH | |
TOTAL
ROD
STROKE
«fl[ [}
= 2 \
LOWER
LIMIT SWITC

CONTROL ROD

SHIM LOCATING
MOTOR, OPERATOR
CONTROLLED

GEARBOX

NOTE: AS REDUCED TO PRACTICE THE
CONTROL ROD MOTION IS
REPRODUCED IN THE CONTROL ROOM
AND LIMIT SWITCHES ARE ACTUATED
BY THE REGULATING ROD LIMIT
SWITCH ASSEMBLY. SEE DIAGRAM.

REGULATING ROD
SPAN Ay,

REGULATING ROD
LIMIT SWITCHES

SYNCHRO
POSITION
TRANSMITTERS

SCRAM
CLUTCH

UNCLASSIFIED
ORNL-DWG 84-622

REGULATING ROD L!MIT SWITCH ASSEMBLY
IN INSTRUMENT CABINET

SPROCKET
GEARBOX

QRIVE MOTOR:

OPERATED BY ROD CONTROLLER
ONLY WHEN IN SERVO OPERATION
AND BY ROD ACTUATCOR SWITCH
IN MANUAL OPERATION

ROD DRIVE UNIT
SEE FIG. 2.3

SHIM LOCATING MOTOR
OPERATOR CONTROLLED
BY CONTROL ROD ACTUATOR
SWITCH. USED DURING

SERVO OPERATION ONLY

ROD POSITION
TRANSMITTER

SYNCHRO
CONTROL BALANCE
TRANSFORMER MOTOR

MECHANICAL
DIFFERENTIAL

CLUTCH-BRAKE:

CALIBRATION
a. IN SERVO OPERATION THE DIAL
CLUTCH IS ENGAGED AND

THE CAM IS COUPLED TO SLIP CLUTCH

THE BALANCE MOTOR

b. IN MANUAL OPERATION
THE OUTPUT SHAFT IS
LOCKED AND THE
BALANCE MQTOR
DECOUPLED

GEAR BOX FOR
SPAN CHANGE GEARS

%——REGULATING ROD

LIMIT SWITCH
CAM AND LIMIT SWITCHES LOWER
WITH MECHANICAL STOP
REGULATING ROD UMIT

SWITCH UPPER REGULATING ROD

POSITION TRANSMITTER
CLOCK CAM ROTATION (SYNCHRO])
EQUIV. TO:
a. WITHDRAW ROD NO. 1
b. INSERT LIMIT SWITCH
SPAN {DECREASE ¥,)

|
|
/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/
!
|

DO Ot

REG. ROD POSITION
INDICATOR (RELATIVE
TO SPAN LIMIT
SWITCHES)

CONTROL CONSOLE

3% AVAILABLE TO SERVO
£ CONTROLLER WITH
REG. ROD SWITCHES
OKE AT ¥,
COARSE  FINE
- . SHIM ROD
POSITION
INDICATORS
kLS
P’
Fig. 22. Diagram of Regulating Rod Limit Switching Assembly.

6ct
140

A similar device has been operating very satisfactorily in the con-
trol system of the ORR for several years, and we have no reservations as
to its use in the MSRE, '

Load Control

Load, or reactor demand, is determined by the radiator door opening,
by the by-pass damper position, and by the blowers. These components are
all independently capable of affecting the load, and the manipulation of
all of the pushbuttons, switches, etc., required to make load changes in
a consistent and orderly fashion would be a burdensome task for a busy
operator. These switching operations are programmed so that load changes
may be accomplished by a single control switch at the console. Manual
control of the wvarious components responsible for the load is optional.

Circuit Jumpering

System checkout and tests of wvarious parts of the reactor system,
and the nonroutine operations required to conduct experiments, inevitably
create conflicts and inconsistencies between operational requirements and
prohibitions imposed by the control and safety system. An obvious example
in the MSRE is the checking of the operation of the radiator doors when
the system is empty and shut down. The safety system keeps the doors
closed if salt (or pipe) temperature is below 900°F or if there is no
coolant-salt circulation., Obviously, it is good practice to periodically
check the operation of these doors. It isn't good practice to do this
inspection by using clip leads to bypass interlocks or by temporarily re-
arranging the electrical interconnections on an informal basis.

A jumper board has therefore been mounted on the main control board
to allow formal, administratively controlled bypassing of interlock and
safety circuit contacts. Figure 23 is a diagram showing typical circuitry
required to jumper a safety system contact. Actual contact bypassing is
accomplished by relays which are energized when plugs are inserted in the
board. Consider the situation in which there is a string of contacts
ahead of a safety relay with several of these jumpers for bypassing. An
open neutral could, conceivably, produce a sneak circuit via the lamps
which might carry enough current to keep the safety relay energized, re-
gardless of whether the safety contacts were open or closed. This is
prevented by the silicon diodes which, in this situation, are back-to-back
and prevent alternating current flow. The use of a bypassing jumper re-
quires permission of the Operations Supervisor. Any jumper in the safety
system circuitry prohibits switching into the "operate" mode. The jumper
itself is visible from the control console, and each jumper or closed
contact is indicated by a light on the Jjumper board. If any safety system
contact is Jjumpered, an annunciator is actuated. The Jjumper board is de-
signed as a graphic panel display of the control system circuitry and,
in conjunction with the indicating lamps, gives the operator an immediate
graphic display of the circuit condition of all vital control circuits.
141

UNCLASSIFIED
CRNL-DWG 64-6080

SAFETY SYSTEM
RELAY CABINET

JUMPER BOARD
ON MAIN CONTROL BOARD

VL ,////////////////j/////ll///L///
2 7/ ?
LINE—F 5V
/ n ;
[/ 7 % ONE 4
NEUTRAL—F1— ZERY CHANNEL OF ]
¢ 77 SILICON A TYPICAL [
; f 9 R DIODE SAFETY g
] /]
4 ,_Q v SUB-SYSTEM [/
JACK, % 2ERY
CIRCUIT IS j)}@) 77 4
JUMPERED BY /] ; ]
INSERTING PLUG —T/] By-PasSING_L.  _L_ SAFETY 9
% % CONTACT =T CONTACT 7
. (] Ll Y ! %
o ] 1 1 /
% Ll /
[ f g {
1| 7
INDICATING [/ | 2 / / 4
(TbfifiiL)—~—é"" 707 / g
% 4 ,
/] ; Ir
L | /]
7 i g
2 N 7 ,’ 'é
% i 47 r SAFETY [/
4 7 / SYSTEM
7 ¢ BY-PASSING RELAY /
4 /Y RELAY %
4 ¢ %
] g ¢ V]
g < . %
] U / ]
[ 7 9
(777 rr 7 77 777N T T T 7777777777777

NOTE: THE JUMPERS AND THE INDICATING LAMPS ARE ON THE MAIN
CONTROL PANEL AND VISIBLE FROM THE OPERATOR'S CONSOLE.

Fig. 23. Diagram of Safety System Bypassing with Jumper Board.

Process Radiation Monitoring

The principal function of our process radiation monitors is to ensure
"containment; none of these monitors is used for routine operational con-
trol purposes.

We do monitor, with redundant input systems, the process lines from
the containment which can provide a credible avenue of escape for con-
taminating radiation. These lines are, of course, equipped with block
valves, which are closed automatically by the radiation monitoring equip-
ment., ‘

Line monitors are installed on the inlet helium supply lines to the
fuel-salt loop and on lines from the containment in the off-gas system.
These off-gas system monitors look at the cooling air flow in the sec-
ondary containment cell and at helium flows from the fuel- and coolant-
salt loops, from the drain tanks, from the pump lube 0il systems, and
from the sampler-enricher, The off-gas is monitored again by the stack
142

gas monitor, just before being discharged to the atmosphere. The coolant
pump off-gas should not contain any activity whatsoever. The origin of
any activity in this line would be a leak in the heat exchanger, causing
contamination of the coolant salt. Similarly, leakage of salt or fission
product gases from the fuel-salt loop into the secondary containment would
be detected by the monitors on the cell evacuation air line., Activity

in either cell air or coolant pump off-gas initiates a drain and closes
block valves in the lines involved.

The in-cell cooling water is also monitored for activity at the pump
inlet pipe. This cooling water system circulates water to the fuel- and
coolant-salt pumps, the in-cell air coolers, and the thermal shield. These
monitors also produce automatic block wvalwve closure.

Where radiation monitoring instruments are used to close block valves
or to initiate an automatic drain, the input channels are redundant and
the installation includes provisions for periodic testing with a source.
In some cases, the redundancy is of the 2 out of 3 coincidence type for
in-service testing. ‘

We have also installed gamma chambers in the secondary containment
cell around the reactor vessel, in the drain tank cell, and in the radi-
ator pit. The fuel sampler-enricher and the lube oil systems are also
equipped with individual radiation monitors.

Health Physics Monitoring

With respect to the individual instruments used, the MSRE health
physics monitoring instruments are typical. That is, at strategic points
in the building are located seven Constant Air Monitors to detect airborne
contamination, and seven Beta-Gamma Monitors and seven Monitrons to detect
sources of radiation. Readout from this scattered array of instruments
is obtained locally at the instruments and centrally in the control room.

The control room alarm and annunciator instrumentation has been de-
signed to meet ORNL requirements for health physics monitoring installa-
tions. Figure 24 is a block diagram of the health physics monitoring and
alarm system. The system

1. indicates some types of instrument malfunction (i.e., an instrument
failure evidenced by zero output produces an audible alarm and turns
up a lamp showing which instrument is not functioning),

2+ 1ndicates by lamps and the audible alarm any instrument that exceeds
the permissible health physics radiation level and provides a second
indication at the high alarm level (3 times that of the health physics
level for the Monitrons),

3. provides a control signal at any reasonable degree of coincidence.
In the MSRE we have selected two groups of monitors so that if any
two Monitrons out of six or any two Constant Air Monitors out of four
are at the "high" alarm level, an automatic evacuation signal results.

. 143

UNCLASSIFIED
ORNL-DWG 64-9703

* F;commm_ ROOM
- ALARM MODULES
INTERMED. o,
TROUBLE l [(
X8 ; M
TO EMERGENCY
[no.z} [eleie: .
| COINCIDENCE CONTROL CENTER EVACUATION
constant | [No-3] feisisd, MATRIX [ HORN
AR (ANY 2 OF 4)
moniTorRs | LNO-4] gt
7t o3 (G
NO. 6 [lsle: L
o
. MOBILE UNIT B
[ L
. (INOJ} } [e7ers] ,
‘ COINCIDENCE [~ ALARM
NO. 2 fofste] MATRIX ———
MONITRONS 1 ,:J = (ANY 2 OF 4)—m=  TO BEACONS
' l EMERGENCY {1
- Noe roloje} CONTROL —{3
| : CENTER I
- @
NO. 1 1 E—c
“ —{
By ) [no2 (]
MONITORS i
! @
NO.7 1|

Fig. 24. Block Diagram of MSRE Health Physics Monitoring Systen.

This annunciation and alarm instrumentation, which ties the indi-
vidual monitors together as a system, is of modular construction. The
design of a health physics monitoring and alarm system is accomplished

. by the interconnection of a sufficient number of modules to meet design
criteria for a particular facility.

- All evacuvation signals, whether produced automatically or adminis-
tratively by the reactor operator or other responsible person on the site,
are transmitted to the ORNL Emergency Control Center, where appropriate
emergency procedures are set in motion. The automatic alarm can be
blocked out by a key switch administratively controlled by the Chief of
Operations when system testing or special maintenance is required.

The health physics instrumentation is not intended to be used for
automatic control or system shutdown. Its main purpose is to provide
reliable information to supervision on the site and to the Emergency Con-
trol Center.
144

Beryllium Monitoring

We also monitor for beryllium contamination. The air passing out
the radiator stack is sampled continuously with filter paper which is
spectrographically analyzed once a minute. This operation is automatic,
and the results are recorded. The reactor bullding contains 15 air sam-
pling stations which draw building air through filter papers which are
analyzed on a routine basis,

This beryllium monitoring operation is carried out by the Industrial
Hygiene Department of the ORNL Medical Division.

Data Logger-Computer

The MSRE instrumentation system is capable of generating a forbidding
quantity of data. Our previous experience with large experimental systems
equipped with typical strip chart recorders, indicators, etc., has shown
that the recording, reduction, and evaluation of system data necegsary
for both routine operation and experimental purposes is a task of major
proportions. In addition, conventional multi-point recorders, having
long scanning periods, are inherently incapable of recording the frequency
structure of all but wvery slow changes. Storage space for chart rolls
must be provided, and truly accurate time referencing of all these rolls
of data is always questionable, The problems of data reduction become
particularly onerous and expensive when the design of a particular test
is guided or determined by the results of the preceding test. There is,
in these circumstances, a really large incentive to reduce the data re-
duction time which determines the nonproductiwve time between tests. An
operation like the MSRE may cost $4,000/day to run, and it is obvious
that reductions in down time or nonproductive time will produce substan-
tial savings.

For these reasons, we have added a logger-computer to the MSRE in-
strument system. Since our experience with systems of this type has been
limited, we are proceeding with the logger-computer installation on a
test and evaluation basis. Therefore, the design and operation of the
MSRE is such that no vital functions involving data gathering, safety,
or control are wvested solely in the logger-computer. This enables us to
rent the device on a month-to-month basis with credit toward ultimate
purchase at our option. Figure 25 is a block diagram of the system,

The machine is capable of accepting a total of 350 analog plus 112
on-off contact inputs. These may have ranges from millivolts to volts
of either ac or dc. During routine operation, the continuous scan rate
is 90 points/sec, giving a maximum complete cycle time of 5 sec; this in-
cludes time for other computer functions during the scan. Data storage,
either programmed or by operator command, is accomplished on either the
magnetic or the paper tape units, or both, and in a format acceptable to
our other large general purpose computers, The X-Y plotter can be pro-
grammed to automatically produce curves based on logged or computed data.
(g

145

UNCLASSIFIED
ORNL-DWG 64-6481A

ANALOG SIGNALS
TO REMOTE
INDICATORS

CONTRCOLLERS ETC.

L

Xy

PLOTTER

L

3 LOGGING
TYPEWRITERS

DIGITIZED
QUTPUTS

N
ANALOG
OUTPUTS

OPERATOR'S CONSOLE

INPUT

KEYBOARD

PROGRAM CONTROL,
INDICATING LIGHTS
AND ALARMS

PAPER TAPE
PUNCH AND
READER

INPUT-
CUTPUT
TYPEWRITER

CONTROL
AND
PROGRAM

—————-
L - INPUT-OUTPUT INSTRUCTIONS |
™ | 350 ANALOG EQUIPMENT AND CONTROL
i INPUT SIGNALS DIGITAL
REACTOR | | _ {(MULTIPLEXER, DIGITAL INPUT COMPUTER
SYSTEM | VOLTAGE DIVIDERS, |,
INSTRUMENTS FILTERS, AND DIGITAL OUTPUT
N AMPLIFIERS)
> | 112 "ON-OFF"
, -
T o [ siGNaLsS CONTROL
e 1
1 DIGITAL
( INFORMATION
MAGNETIC

TAPE UNITS

Fig. 25. Block Diagram of MSRE Data Logger-Computer.

The scanned data are printed out routinely on the automatic type-
writers in engineering units once an hour. They can be printed out at any
time by operator request. In addition, there are printouts every 8 and
24 hr for summary and daily reports. The frequency of these routine
printouts is at our option. The time of the log is printed and the in-
struments or sensors which produce the data are positively identified on
the log sheet.

The machine continuously monitors the system operation. As the pro-
gram now stands, approximately 150 input signals are compared with preset
high and low limits. In the event that any of these input signals ex-
ceeds its limits, the equipment produces an alarm signal and prints out
the off-limit signal. If a certain number of these 150 input signals
reach the alarm limits, the system will pass into its fast scan mode.

In the fast scan mode the system scans 64 preselected inputs continuously,
taking about 1/4 sec for the complete scan cycle. The initiating signal
is logged by the alarm typewriter, and the data are stored on magnetic
tape which can subsequently be processed by the data system computer or,
if desired, by one of our large, general purpose computer facilities.
146

The unit is also capable of producing 32 digitized outpuf signals,
readout, etc., should we so desire.

Typical calculations for which the machine is now being programmed
are:

1. heat balance,

2. reactivity balance,

3. 1integrated power,

4. nuclear average temperature,

5. average temperatures and thermal cycling records of the salt pumps,
which will be used to program laboratory tests by the metallurgy

group,

6. a statistical survey of heat-exchanger temperatures, which will pro-
vide an evaluation of heat-exchanger performance.

The question of computer control of the reactor system inevitably
arises when we discuss the application of this equipment. Existing pro-
gramming plans do not include reactor or process control as part of its
functions. The first reason, of course, is that we did not commit our-~
selves to making the unit an indispensable part of the design. Secondly,
we want to gain experience with both the computer-logger and with the re-
actor before embarking on such a venture. I can, however, assure you that
there is a substantial number of people who are eager and anxious to ex-
tend the use of the computer into the area of system control,

The initial programming is a joint effort by ORNL and the seller.
We are dewveloping a cadre of gqualified personnel, drawn from the Instru-
mentation and Controls, the Reactor, and the Mathematics Divisions, to
alter the machine operation and computational functions to suit the chang-
ing needs of the experimental program.

We are not installing this equipment on the assumption that it will
effect immediate and obvious reductions in the cost of operating the MSRE,
We do expect that, once we have learned to use it, we will extract more
and better 1nformat10n in less time from the reactor system. This is a
somevhat intangible cost saving, but it can be a wvery substantial one.
Once experience and confidence in applying and using such equipment has
been gained, I am confident that this or similar equipment will assume a
greatly expanded role in large systems such as the MSRE.

L]
147
PUMP DEVELOPMENT

A, G. Grindell P, G. Smith

Introduction

The purpose of this report is to review the work done on producing
the high-temperature salt pumps used in the Molten-Salt Reactor Experiment
(MSRE) and the present status of that pump technology. It provides a
brief description of the pumps, discusses in more detail certain distin-
guishing features of the pumps which hawve not been previously fully re-
ported, describes noteworthy problems which were resolved during their
design and testing, and gives a summary of the status of their technology.

Two of these pumps are required to circulate the salts in the MSRE,
one in the fuel-salt and one in the coolant-salt circuits. Each pump must
provide for the thermal expansion of the salt in its circuit and for the
sampling of that salt. In addition, the fuel-salt pump must provide for
the removal of 13°Xe from the fuel salt and for the addition of fuel con-
centrate to the fuel salt as burnup progresses.

The detailed design of these pumps is based on precepts taken from
many disciplines, including metallurgy, heat transfer, thermal stress
analysis, hydraulic design of centrifugal pumps, mechanics, electrical
machinery, and Instrumentation and controls, and the Boiler and Pressure
Vessel Code of the American Society of Mechanical Engineers. While we
gratefully acknowledge the contributions from each of these disciplines
to the production of satisfactory and reliable molten-salt pumps, we per-
force limit ourselves to the discussion of noteworthy pump features and
problems in this paper.

Salt Pump Design

General Description

The fuel=-salt pump will be used for purposes of demonstration in this
presentation, and the few differences between it and the coolant-salt pump
will be pointed out during the exposition. The fuel-salt pump, a hermetic
sump-type centrifugal pump, consists of three main components: the pump
tank, the rotary assembly, and the drive motor (see Fig. 1). These are
supplemented with associated lubrication and purge-gas systems and instru-
mentation and controls, All the salt-wetted parts are constructed of
Hastelloy N,

The pump tank is a semipermanent component in the fuel-salt circuit
and is provided with two freeze flanges for convenient replacement. The
motor and rotary assembly may be removed from the pump tank together or
singly. The three main components are made hermetic by ring-joint flanges
which are monitored by the leak-detection system,
22002272225

UNCLASSIFIED

ORNL-LR-DWG-56043-8R
I I [ .hv;\ Y A
.! §
9 —
G _— —3= J
SHAFT G — — WATER N o=
COUPLING G— 7 PR O COOLED \r 7
';Li__:__’_’,‘_——,_:_:) MOTOR §
SHAFT SEAL @) INY 4
(See Inset) \r S ey ) e At § Pt
; NI/ AR &
l\ WShr e
LEAK i Q o\ T
DETECTOR LUBE OIL BREATHER N / ~
LUBE OIL IN - 7
> =7 T/ #_ SN L
BALL BEARINGS
(Face to Face) BEARING HOUSING
BALL BEARINGS GAS PURGE IN
(Bock to Back) SHAFT SEAL (See Inset)
LUBE OIL ouT SHIELD COOLANT PASSAGES
NREE (In Paraltel With Lube Qil)
SEAL OIL LEAKAGE :
DRAIN SHIELD PLUG
LEAK DETECTOR GAS PURGE OUT (See Inset)
SAMPLER ENRICHER X GAS FILLED EXPANSION

(Out of Section)
(See Inset)

BUBBLE TYPE
LEVEL INDICATOR

\ OPERATING | i

LEVEL ! -

SPACE
XENCN STRIPPER

(Spray Ring)
SPRAY

|
o |

L

-

To Overflow Tank

Fig. 1.

|

!

; A“““\

MSRE Fuel Pump.

8¥T
149

The pump tank contains the pump volute. During operation it houses
the impeller and the shield plug, which are supported from the rotary
assembly. It also contains the system liquid-gas interface, the spray
device and gas purge for the removal of 12°Xe, and the ligquid-level in-
strumentation. The rotary assembly consists of the bearing housing, a
pump shaft mounted on commercially available bearings, and upper and
lower shaft seals. The shaft seals confine the circulated oil, which
lubricates and cools the assembly, to the bearing housing.

The drive motor is of the squirrel-cage induction type, housed in a
hermetic vessel. The electrical insulation system and the grease lubri-
cant for the motor bearings are radiation resistant., A more complete and
detailed description of this type of pump may be found in papers by Savage
and Cobb! and Grindell, Boudreau, and Savage.?®

Distinguishing Design Features Not Previously Reported

The MSRE salt pumps require certain features not found in conven-
tional sump-type pumps. The pump tank contains a salt spray device to
remove 135Xe from the circulating fuel salt and two elevated-temperature
liquid-level indicators for molten-salt operation. A split purge-gas
flow in the shaft annulus keeps o0il vapcrs from the salt system and fis-
sion gases from the region of the lower shaft seal; the down-shaft purge
also strips, dilutes, and removes '2°Xe from the salt spray. Since the
reactor vessel is used as the system anchor, the fuel pump is mounted on
flexible supports to accommodate thermal expansions. A flow of nitrogen
(cell air) removes nuclear-deposited heat from the upper section of the
pump tank. The drive motor is specially enclosed. These features and
the differences between the fuel- and coolant-salt pump designs are dis-
cussed in this section.

Nuclear-Poison-Removal Device. Some of the 13?Xe poison is removed
from the fuel salt by contacting a spray of the salt with helium in the
fuel-pump tank. Approximately 50 gpm of salt i1s introduced through 291
small jets at a spray velocity of approximately 15 fps into the gas-filled
- portion of the fuel-pump tank. Here, the helium which enters the pump
tank from the shaft annulus purge and that which is associated with the
operation of the bubble-type liquid-level indicators remove the nuclear
poison from the salt spray, dilute it, and carry it to charcoal traps.

Tt should be noted that this device is not included in the design
of the coolant-salt pump.

Molten-Salt Liquid-Level Indicators. Two bubble-type liquid-level
indicators are used to show the liquid level of fuel salt. They are
identical except that one reaches to a 2-in. greater depth than the other.
In this type of indicator a flow of helium passes through a tube inserted
vertically into the salt from above the salt level. The difference in
pressure between the gas supply to the level indicator and the gas space
in the pump tank is a function of the depth of the tube in the salt. The
gas flow from the indicators exits through the gas purge nozzle, helping
to dilute and carry off fission gases, as mentioned above.

150

The coolant-salt pump also contains a buoyancy level indicator, which
uses the output from a calibrated differential transformer to indicate -
level.

Gas Purge System. Helium enters the pump shaft annulus below the
shaft lower seal. A portion of the helium (0.1 liter/min) flows upward
in the annulus to reduce the migration of oil vapors from the lower seal
leakage catch basin to the pump tank. A drain passage for the oil which
leaks past the lower shaft seal conducts the flow of helium from the pump -
to the charcoal traps.

The remainder of the helium purge (2.3 liters/min) flows down the
shaft annulus, reducing the migration of fission gases from the pump tank
to the region of the lower shaft seal. This flow enters the pump tank gas
space and dilutes and transports fission gases to charcoal traps.

Flexible Supports for the Fuel-Salt Pump. The fuel-salt pump 1is
mounted on a horizontal plate which is supported at each end by the ar-
rangement shown in Fig. 2. The end supports are identical and provide .
for three degrees of translational displacement. These supports are shown
in Fig. 3. Roller bearings on the top surface of item 1 permit sideways
translation of the pump support plate. Bearings on the upper side of item
1 mate with raceways in item 2 to permit forward and backward displacements
of the pump. Item 2 is in turn supported by inclined linkages, attached to
the base, which provide for vertical displacements. The weight of the pump
system is counterbalanced with springs as shown and, as the salt piping
system expands or contracts, the pump moves accordingly.

The coolant-salt pump is mounted in a rigid support.

Provisions for Cooling the Gas-Bounded Portion of the Pump Tank. The
outside surface of the upper section of the fuel pump tank is gas cooled
to remove nuclear-deposited heat. This is shown in Fig. 4. A distributor
is fitted around the neck of the pump tank to distribute the flow of cool-
ing gas (approximately 200 cfm of nitrogen) across the outer surface of
the tank. .

This cooling is not required on the coolant-salt pump tank.

. Drive Motor. The pump drive motor is in a hermetic containment ves-
sel, fabricated to meet the requirements of a Primary Nuclear Vessel as
specified in the ASME Boiler and Pressure Vessel Code and appropriate
Code Case Interpretations 1270N-2 and 1272N-4. Two grease-lubricated
angular contact bearings, designed to operate continuously for 10% hr
without maintenance or relubrication, support the shaft. The grease,
Nuclear Radiation Resistant Grease-159, is reported® to be capable of
withstanding 3.4 X 102 rads total radiation dosage. The basic insula-
tion system used in the motor windings consists of silicone-glass-mica
laminates, silicone rubber over glass and type "ML" enamel, and multi-
dip-impregnations of silicone varnish.  These materials should be capable
of withstanding an integrated radiation dose in excess of 10° rads.

UNCLASSIFIED
PHOTO 63422

Fig. 2. MSRE Fuel Pump Support.

(CLASSIFIED

NCLASS
ORNL-OWG 64-6625

3. TFuel Pump Support.

152

UNCLASSIFIED
ORNL-DWG 64-66249

COOLING
INLET
COOLING
CooL It . COOLING EXIT
2 PUMP TANK % DISCHARGE

INLET

A~ HEATERS AND INSULATION |~

Fig. 4. Pump Tank Upper Shell Cooling.

Each of the salt pump motors is 75 hp, but the fuel-salt pump motor
is 1200 rpm synchronous speed and the coolant pump motor is 1800 rpm.

Noteworthy Problems Resolved During Design and
' Development of Salt Pumps

During the design and development of these pumps, a number of note-
worthy problems were resolved. The ability of the pump tanks to withstand
the strain fatigue associated with thermal cycling from room to operating
temperatures and from zero to full reactor power was investigated. The
effectiveness of the purge-gas flow in the shaft annulus in limiting the
back diffusion of fission gases was studied with the prototype pump using
85Ky, The need for maintaining this purge-gas flow to prevent plugging
of the lower end of the pump shaft annulus with solidified salt was dem-
onstrated. Attempts were made to determine the effectiveness of the de-
vice in the fuel-pump tank for removing fission gases. Adequate shaft
running clearances to provide for off-design pump operation were deduced.

Pump Tank Thermal Stresses

The ability of the fuel- and coolant-salt pump tanks to sustain 100
reactor startup cycles from room temperature to 1200°F and 500 cycles from
zero to full reactor power was studied. Values of the maximum thermal
stress and the resultant strain fatigue associated with these cycles were
computed and were compared with values of the thermal fatigue strength of
Hastelloy N, deduced from metallurgical experiments. It was found that
the upper section of the fuel pump tank required cooling, but the coolant
pump tank did not require this cooling.
153

In analyzing the thermal stresses,4 the configuration (Fig. 5) was
considered to be composed of: (1) an internal cylinder, cylinder "A,"
extending from the volute to the junction with the spherical shell; (2)
an external cylinder, cylinder "B," extending from the junction with the
spherical shell to the top flange; and (3) the upper spherical. shell of
the pump tank. '

Internal heat generation, conductive heat flow, convective and ra-
diative heat transfer with the salt, and cooling of the shield plug and
upper pump tank surfaces were the bases for calculating temperature dis-
tributions in the pump tank for various operating conditions with the
generalized heat conducticn code (GHT Code).? The axial temperature dis-
tributions of the volute support cylinder at variocus operating conditions
and the effect of the gas cooling on the temperature distribution are
shown in Fig. 6. The temperature distributions along the pump tank spher-
ical shell (i.e., the upper section) for various operating conditions in-
cluding cooling are shown in Fig. 7.

UNCLASSIFIED
ORNL—LR—-DWG 64491R

I\\\\ TOP FLANGE
' ; VOLUTE SUPPORT
CYLINDER /rr//f””,_, CYLINDER
"B" /
g PUMP TANK
’ C SPHERICAL
% SHELL
/
CYLINDER [
"A"
LIQUID [/
LEVEL /
]
N
~
~
~

LIQUID LEVEL

Fig. 5. Pump Tank and Volute Support Cylinder Geometry.
154

UNCLASSIFIED
ORNL-LR-DWG 644992R

1400 i
VoLUTE SPHERE
ot * -‘r-:::_.\ JUNCTION
" "
1200 - NN
= A\ N\
[ . \\\
\\ e N\
A
\\ \ \ FUEL PUMP, 10-Mw POWER,

1000 —_— N\ N \“/ NO EXTERNAL COOLING _
= S~ N % FUEL PUMP, 10-Mw POWER, 200 -ctm
5 ~ N\ \ AIR COOLING
" \\ a \\
x ~ N FUEL AND COOLANT PUMP, ZERO
£ 800 ———— tUE pUMP. ZERO POWER, [\ \ POWER. NO EXTERNAL COOLING|
@ 200-cfm AIR COOLING N L
¢ | R\ N\ N
] W\
= COOLANT PUMP, 10-Mw \

600 POWER, NO EXTERNAL COOLING — N

® AND A INDICATE TEMPERATURES PREDICTED

J‘ \\Oi '
AR
N

400 | BY THE TEMPERATURE EQUATIONS FOR THE 2 L rop
O-AND 10-Mw POWER CASES WITH 200-cfm SRy FLANGE
COOLING AIR FLOW. : ~
| SSE
A A
200
0 2 a 6 8 10 12 14 6

AXIAL POSITION (in.)

Fig. 6. Axial Temperature Distribution of Volute Support Cyllnder
at Various Operating Conditions.

UNCLASSIFIED
ORNL-LR-DWG 64493R

0% | | | | |

o AND e INDICATE TEMPERATURES PREDICTED BY THE
TEMPERATURE EQUATIONS FOR THE O AND 10-Mw
POWER CASES WITH 20C-cfm COCLING AIR FLOW

1800

| | | |
CYLINDER
JUNCTION FUEL PUMP, 10-Mw POWER, NO EXTERNAL COOLING
1600 PY .l e
P -~ ~d

- ~
1400 ) g
// ' \\
OLING
0 P POWER, NO EXTERNAL €O o
—

' ZERC
1200 /’ CUEL AND COOLANT 7! PUMPS,

——r NG
. {R Coou!
/ =1 \p. 10-Mw POWER, 2007 -ctm A
-~ L PU 1)
- FuE :

M 10-MW
P

TEMPERATURE (°F}
~

W

\

-—__.'__—-1

800 I/ -| [ ] .

e
t F PUMP, ZERQ POWER, 200 -cfm AIR COOLING

600

o] 2 4 6 8 10 12 14 16
MERIDIONAL POSITION {in.)

Fig. 7« Meridional Temperature Distribution of the Torispherical
Shell at Various Operating Conditions.

155

An Oracle program was used to obtain the pressure stresses on the
cylinder, and an IBM 7090 program was used to obtain the thermal stresses
in the two cylinders and the sphere. The principal thermal stresses in
the fuel pump at cylinders "A" and "B" for zero power operation with and
without cooling are shown in Fig. 8. The stresses at cylinders "A" and
"B" for 10-Mw power operation with and without cooling are shown in Fig.
9, The thermal stresses in the spherical shell for zeroc power operation
with and without cooling are shown in Fig. 10. The stresses in the spher-
ical shell for 10-Mw power operation with and without cooling are shown in
Fig. 11.

The results of the strain-fatigue analysis are presented in Tables
1 and 2. The last column in each table, often referred to as a usage
factor, is the decimal fraction of the strain-fatigue limit of Hastel-
loy N used up. By combining the usage factors in Tables 1 and 2 the
data in Table 3 are obtained.

A usage factor of 0.8 or less indicates a safe operating condition.®
At cooling air flow rates from 100-300 cfm the usage factor is less than
0.8, and a cooling air flow rate of 200 cfm has accordingly been selected
for the MSRE.

A detailed report of the thermal-stress and strain-fatigue analyses
for both the fuel- and coolant-salt pumps has been published.*

Back Diffusion of Fission Gases in the Pump Shaft Annulus

It has long been recognized that the migration of fission gases from
the pump tank to the region of the shaft lower seal via the shaft annulus
can polymerize leakage seal-oil and lead to plugging of the drain line for
that leakage. A flow of purge gas down the shaft annulus was introduced
to minimize this migration.

Back-diffusion tests,7 using 85Kr, on the prototype pump measured
the effectiveness of the shaft annulus purge. Figure 12 shows a sche-
matic of the test setup. The upper end of the annulus through which the
"down the shaft" purge flows contained a labyrinth seal with a 0.005-in.
diametral clearance with the shaft. Krypton-85 was injected into the pump
tank as shown. Count rate meters determined the concentrations of 85kr
in the pump tank off-gas and in the leakage oil catch basin at various
flow rates of purge gas. Table 4 shows the data obtained during the tests.
A count rate meter, capable of detecting a concentration as small as 0.95 x
10710 curie/cm?, detected no 89Kr in the leakage oil purge gas. These
data indicate the effectiveness of the purge gas to reduce the concentra-
tion of &%Kr in the catch basin by a factor of at least 20,000, compared
with the concentration in the pump tank.

The radiation tolerance of the oil used with the MSRE pumps is 107
rads/g. With this amount of irradiation, the oil will flow and drain from
the catch basin. The permissible source in the catch basin for the MSRE
fuel pump for a dose of 107 rads per g of oil with 1-Mev beta radiation
156

UNCLASSIFIED
ORNL-DWG 64-6819

40
]
30 }
I,.’—“NO COOLING
/
20 /l
3 /
a /— / \
8 Lo} < < 1
] N~ \
= \
3 o A\
& AR
E 200 cfm COOLING AIR /’\
AT
-10 \‘~—./ \\\
N\
\\
_20 \
\
\
-30
-8 -6 -4 -2 (0] 2 4 6 8

AXIAL PCSITION (in.}

Fig. 8. TFuel Pump Principal Thermal Stresses at Cylinders "A" and
"B" for Operation at Zero Power.

UNCLASSIFIED
ORNL -DWG 64-6820

80
l
60 |
|
J=——NO COOLING
/
40 f
/
= [\
S 20 L
S /7\\
- / \
9 / \\
o 0 I/ \
7S /// \\_
‘ \
\/"—\\
-20 /\)\
200 cfm COOLING AIR \
-40
-60
-8 -4 0 4 8 2

AXIAL POSITION (in.}

Fig. 9. Fuel Pump Principal Thermal Stresses at Cylinders "A" and
"B" for Operation at 10 Mw.,
157

UNCLASSIFIED
ORNL-DWG 64-6818

- 60
\\
. 40
- \\\
a
o .\NO COOLING
2 N
< 20 \\
= N|
h ‘\\\
s ~
= N~
v O - — =
/-—-_- ‘-..___\
500 ctm COOLING AIR ]
| B
. 0 1 2 3 4 5 6
. MERIDIONAL POSITION (in.}
. Fig., 10, Fuel Pump Principal Thermal Stresses at Spherical Shell

for Operation at Zero Power.

UNCLASSIFIED
ORNL-DWG 64-6821

120
100 \\
\
80 X
\NO CooLING
—~ ©0 - \
o \
T 40 \
a % \\
L
. E N\
w20 N
- N
200 cfm \\
COOLING AIR
0 \.
~
~
\\
-20 \\h""'--.
-40
0 1 2 3 4q 5 ©

MERIDIONAL POSITION (in.)

Fig. 11, Fuel Pump Principal Thermal Stresses at Spherical Shell
for Operation at 10 Mw.
158

Table 1. Fuel Pump Strain Data for Heating Cycle (to 1200°F)

. Maximum Cycle Cycle -
Alr Stress Str?ss Allowable Fraction Fraction in
Flow . Amplitude
(cfm) Intensity (psi) Cycles Per 100 Cycles, -
(psi) p Cycle Py /Ny
50 31,124 15,562 700 0.00143 0.143
100 14,400 7,200 2,500 0.00040 0.040
150 14,143 7,072 2,500 0.00040 0.040
200 16,095 8,048 2,100 0,00047 0.047
250 21,760 10,880 1,300 0.00077 0.077 _
300 26,955 13,477 880 0.00114 0.114 -

Table 2. Fuel Pump Strain Data for Power-Change

Cycle from O to 10 Mw .
. Maxcimum fl Cycle  Cycle
Alr Stress Str§ss Allowable  Fraction Fraction in
Flow . Amplitude 00
(cfm) Intensity (psi) Cycles Per 500 Cycles,
(psi) : Cycle P, /N, |
50 37,971 18,985 520 0.00192 0.961
100 24,763 12,382 - 1,000 0.001 0.500
150 18,930 9,465 1,600 0.000625 0.312
200 18,814 9,407 1,600 0.000625 0.312 -
250 18,775 9,388 1,600 0.000625 0.312
300 18,639 9,320 1,600 0.000625 0.312
Table 3. Usage Factor as a Function of Air Flow Rate .
Aj Cycle Cycle Total : y
Flgz Fraction in Fraction in Usage
(cfm) 100 Cycles, 500 Cycles, Factor,
Py /N; P, /Ny b Pi/Ni
50 0.143 0.961 1.10
100 0.040 0.500 0.54
150 0.040 0.312 0.352
200 0.047 0.312 0.359
250 0.077 0.312 0.389

300 0.114 0.212 0.426

159

UNCLASSIFIED
ORNL-DWG 63-6482

LOWER SHAFT
SEAL

PUMP
SHAFT
—

OFF-GAS TO

\ NSO RADIATION
catcH ’////////"EOUNT METER
\—-

] 777777 TOTAL PURGE iN
-

N

N

NSNS NN

&

I PUMP TANK
OFF-GAS TG
RADIATION
\ COUNT METER

SHAFT ANNULUS
PURGE TO PUMP TANK

2

- Fig. 1l2. Diagram for Purge Gas Flow in Shaft Annulus of MSRE Pro-
totype Pump.

VL
L

K8 INPUT

160

Table 4, Back-Diffusion Tests

a
85xr Concentration

Titere am)  iteneamyy in Pup ek

x 107®
406 487 0.89
406 487 0.89
430 497 2.06
382 ' 481 2.77
366 354 3.28
215 360 1.85
108 360 1.21
173 673 1.65
109 360 0.75
109 681 0.90
104 681 3.54

aConcentration values for catch basin were below the limit of detec-
tion.

is 0.13 curie. This is based on the conservative agsumption that the
leakage through the catch basin is 1 g/day and that all the energy of the
beta emitters is absorbed by the oil. Since the volume of the catch basin
is 423 cm®, the maximum permissible concentration in it is 3.1 x 10™
curie/cm®. This concentration should not be exceeded, according to the
data in Table 4, if the concentration in the MSRE fuel pump tank is main-
tained at less than 11 curies/cm®. The concentration of fission gas in
the MSRE fuel pump is related to purge-gas flow as shown in Fig. 13. The
data indicate that a flow of approximately 1400 liters/day will ensure
that the concentration in the pump tank will be less than 11 curies/cm?.
Purge-gas flows of 3300 liters/day down the shaft annulus and 1300 liters/
day from the level indicators are available in the MSRE fuel pump.

shaft Annulus Plugging

One 700-hr test® with molten salt at 1200°F was terminated when fluc-
tuations in the power required by the motor led us to conclude that the
shaft was rubbing, possibly in the shield plug. Inspection revealed the
presence of solidified salt in the pump shaft annulus. Subsequent anal-
ysis of the prevailing operating conditions revealed that the pump tank
pressure had exceeded the supply pressure of the gas to the shaft annulus
by 5 psi and had caused the gas flow in the shaft purge annulus to flow
upward instead of downward. The chemical composition of a sample of the
material taken from the shaft annulus was very close to that of the salt
being circulated. The results of x-ray and petrographic examinations of
the sample indicated that the material was carried to the annulus as an
entrained mist.
. 1631

UNCLASSIFIED
ORNL-DWG 63-6483

. 20 T r T
= |, 10 Mw POWER
E e 100 % STRIPPING EFFICIENCY
e \ 50 gpm BYPASS FLOW ]
T \\\ 5 psig PUMP TANK PRESSURE
O
512 AN
= N
= 3
-
=
- \\\\\\\\»
Z g
8 \
o \
. g \.

. 5 4
[7)]
7]
w

0
0 1000 2000 3000 4000 5000

PUMP TANK PURGE (liters /day)

Fig. 13. Fission Gas Concentration in Pump Bowl vs Purge Flow Rate.

To prevent reoccurrence of this incident and to provide protection
against the back diffusion discussed in the previous section, the gas
flows to and from the pump tank are monitored to assure that gas flows
down the shaft annulus, The flow rate of gas from the pump tank must
equal the flow rate of the purge down the shaft annulus plus the flow
rate of gas to the bubble-type level indicators.

Nuclear Poison Removal from the Fuel Sait

Simulating the removal of the 135¥e poison by helium contact with salt
spray in the fuel pump tank, tests were made in the water test facility”
with water (salt), dissolved carbon dioxide (xenon), and air (helium
. purge). However, since the applicability of these tests to the molten

salt, xenon, and helium system in the reactor is doubtful, the results
are not included here.

During the back-diffusion tests on the prototype pump, several at-
tempts to determine the effectiveness of the spray devicel® using 85%y
failed. When the pump tank gas space was charged with 85Kr at a concen-
tration of 3.5 x 107° curie/ecm® and while the salt was circulating, at-
tempts were made to detect the activity in the salt through the system
piping and to determine the activity decrease with time. However, the
instrumentation was unable to detect or measure any activity through the
salt piping.
162 .

Running Clearances to Accommodate Hydraulic Unbalance

Centrifugal pumps have a best-efficiency operating condition at which
the various losses in the impeller and volute are minimum and the pressure
distribution in the volute is uniform. During operation at this condition
(see the balance line shown in Fig. 14), the net radial force on the im-
peller is nearly zero. During operation at higher and lower flows, the
volute pressure distribution becomes non-uniform and a net radial force
appears on the impeller. This force increases as the operating condition
departs farther from the balance line, and operation at conditions which
are off-design thus produces shaft deflections.

During the prototype pump tests with the 13-in.-diam impeller, a shaft
seizure occurred.!! Operation at off-design conditions produced a deflec- .
tion of the shaft sufficiently large to cause rubbing of the hot, dry shaft
against the bore of the shield plug at its lower end. The shaft was fric-
tion-welded to the bore of the shield plug over a length of about 4 in. .
(Fig. 15). The running clearances between the shaft and the shield plug
were increased to 0.045 in. to avoid shaft seizure at anticipated off-
design operating conditions.

UNCLASSIFIED
ORNL-LR-DWG 72413

CONSTANT FLOW @
RESISTANCE LINES

MSRE FUEL
CIRCUIT {DESIGN}

CONSTANT _
SPEED

INCREASING
LOAD

HEAD —>

INCREASING
LOAD

7 e

/‘\PUMP HYDRAULIC BALANCE LINE
MINIMUM RADIAL FORCE ON IMPELLER

PROTOTYPE PUMP OPERATICN WITH 13-in. IMPELLER

FLOW —>

Fig. 14. Hydraulic Radial Unbalance of Centrifugal Pump Impeller.
UNCLASSIFIED
PHOTO 38873

Fig. 15. Results of MSRE Prototype Pump Shaft Seizure.

Results of Tests with Prototype Pumps

Initial testing of the salt pumps was performed in a water test fa-
cility to verify the hydraulic performance characteristics and to develop
the necessary baffles and guides to obtain satisfactory behavior of the
liquid in the pump tank. Hot testing on the prototype pump followed.

The hot tests were conducted to further verify the hydraulic performance
during molten-salt operation and to ascertain the adequacy of the design
to withstand the temperature and functional requirements.

Water Tests

Water tests® were performed with each of the salt pumps to obtain
their hydraulic performance data for comparison with the manufacturers'
reported performance. The pump characteristics were selected from the
manufacturers' reported performances to fit the estimated system require-
ments in the MSRE. There was little difference between the performance
obtained at ORNL and the menufacturers' performance.

164

Prototype Pump Hot Tests

A large portion of the MSRE pump development consisted of work with
the prototype pump. Table 5 enumerates the principal tests and gives the
operating conditions, the impeller. diameter, the test duration, the pri-
mary purposes of tests, and the reason for their termination. The problem
of undissolved gas in the circulating salt was investigated in the proto-
type pump test facility. Measurements were made with a radiation densi-
tometer to determine the content of undissolved gas in the circulating
salt. The results are not reported, because the accuracy of the measure-
ments is not yet certain.

Status of Molten-Salt Pumps

Prototype-Pump

Remaining Development Tests. Further measurements will be made of
the concentration of undissolved gas in the circulating salt. Measurements
will be continued with the radiation densitometer, and the effects of
questionable variables on the results will be evaluated.

Because of shaft-rubbing incidents, the clearance between the shaft
and the purge-gas labyrinth was increased from 0.005 in. to 0.010 in. An
additional back-diffusion test with 85Kr, at various purge flow rates,
will therefore be made, using the 0.010-in. clearance.

The pump and test facility will be used when required to investigate
problems which may arise throughout the operation of the reactor.

Endurance Operation. As time permits, the prototype pump will be op-
erated at 1200°F to investigate the endurance and reliability of the unit.

MK-1 Pumps

Each of the two salt pump rotary assemblies for the MSRE was operated
satisfactorily at 1200°F, circulating salt, for approximately 100 hr in the
prototype pump test facility prior to delivery to the MSRE. The MSRE pump
drive motors were used to rotate the pumps for these tests. General be-
havior, performance of the bearings and seals, and hydraulic performance
were observed. Adequacy of the running clearances was verified, The hy-
draulic performance data obtained with the coolant-salt pump has only
minor value, however, since the impeller was matched with the fuel pump
volute in the test facility.

The MK-1 fuel and coolant pumps, including their drive motors and
lubrication systems, have been delivered to the MSRE.

—
Table 5.

Prototype Pump Tests

Molten- Pump Molten- ' N
Impeller Test
Test Salt Shaft Salt . . . . .
Diameter  Duration  Primary Purposes of Test Reason for Termination
No. Temp Speed Flow (in.) (hr)
(°F) ( rpm) (epm) '
1 80-1200 1150 13 118 Shakedown
1 1200 700-1030  750-1600 13 217 Hydraulic performance Shaft seizure
2 1200 700-1030  750-1200 13 96 Hydraulic performance Scheduled
3 1200 600-1150 400-1500 11-1/2 1968 Hydraulic performance Variable frequency
motor-generator
failure
4 1200 1185 1100 11-1/2 1848 Back-diffusion tests Variable frequency
motor-generator
_ failure
5 1200-1320. 1185 1100 11-1/2 792 Gas concentration tests  Scheduled
in circulating salt,
back-diffusion tests
6 1200 1150 1070 11-1/2 335 Gas concentration tests
in circulating salt
1000-1400  600-1150  500-1070 11-1/2 368 Hydraulic performance Shaft annulus plugging
1200 1150 1070 11-1/2 120 Proof test of MSRE fuel
pump lubrication stand
and the fuel-pump sup-
ports
7 1100-1300 700-1150 600-1070 11-1/2 409 Scheduled
8 1100-1300 1150 1070 11-1/2 168 Proof test of coolant Scheduled
pump lubrication
stand and the fuel
pump supports
Total operating time: 6439

GOT
166

MK-2 Fuel-Salt Pump

A modification of the MK-1 fuel-salt pump is under development. It
contains the same impeller and volute in a pump tank which will provide
an additional 6 ft3 for thermal expansion of the system salt. The 36-
in. diam of the tank was retained, but its height was increased approxi-
mately 9-3/4 in. Increasing the shaft length and thus the overhang of
the impeller below the lower shaft bearing was necessitated. It was pos-
sible, however, to retain the MK-1 pump shaft diameter.

A mockup of the pump tank alone is being used in water tests to de-
velop the hardware necessary to control the concentration of undissolved
gas in the circulating liquid and to measure the rates of removal of carbon
dioxide dissolved in the circulating liquid. The resulting tank configu-
ration will be incorporated into the water test pump for further water
testing., A hot salt test will be made with the final tank configuration.

MK-3 Fuel-Salt Pump

In the planning of the MK-3, an advanced salt pump design of larger
capacity, advantage will be taken of the performance of MK-1 and MK-Z2 de-
signs, the experience in the AEC program for the development of large
sodium pumps, and the in-salt bearing experience at ORNL.1%.,13 At the
present time, a vertical sump-type pump, the upper end of the shaft sup-
ported by oil- or grease-lubricated radial and thrust bearings and the
lower end supported by a salt-lubricated journal bearing, is the pre-
ferred design for the MK-3.

References

1. H. W. Savage and W. G. Cobb, Chem, Eng. Progr. 50, 445 (1954).

2. A, G, Grindell, W. F. Boudreau, and H. W. Savage, Nucl. Sci. Eng.
7, 83-91 (1960).

v

3. J. G. Carroll et al., Lubrication Eng. 18, 64 (1962).

4. C. H. Gebbard, Thermal-Stress and Strain-Fatigue Analysis of the MSRE
Fuel and Coolant Pump Tanks, ORNL-TM-78 (Oct. 3, 1962).

5. T. B. Fowler, Generalized Heat Conduction Code for the IBM 704 Com-
puter, ORNL-2734 (Oct. 14, 1959), and Supplement ORNL-CF-61-2-33

(Feb. 9, 1961).

6. Tentative Structural Design Basis for Reactor Pressure Vessels and
Directly Associated Components (Pressurized, Water Cooled Systems)
{esp. p. 31), PR 151987 (Dec. 1, 1958}, U.S. Dept. of Commerce, Office
of Technical Services.

7. MSRP Semiann. Progr. Rept. July 31, 1963, ORNL-3529, pp. 47-49.

8. Letter from P. G. Smith to R. B. Briggs, Shaft Plugging Incident;
MSRE Prototype Fuel Pump, MSR-63-29 (Nov. 12, 1963) {internal use
only).

167

9. P. G. Smith, Water Test Development of the Fuel Pump for the MSRE,
ORNL-TM-79 (Mar. 27, 1962).

10. MSRP Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 44.

11. TLetter from A. G. Grindell to R. B. Briggs, MSRE Prototype Pump Shaft
Seizure, MSR-62-67 (Aug. 27, 1962) (internal use only).

12. P. G. Smith, ASLE (Am. Soc., Lubrication Engrs.) Trans, 4(2), 263-74
(1961). -

13, MSRP Semiann. Progr. Rept. Feb. 28, 1962, ORNL-3282, p. 56.

COMPONENT DEVELOPMENT IN SUPPORT OF THE MSRE
Dunlap Scott, Jr.

The reliable performance of components and auxiliaries used in the
circulation of molten salts was established by over 200,000 hr of accu-
mulated loop operations and formed the basis for the specifications of
components for the MSRE.

As an added ensurance of reliability and safety, prototypes of crit-
ical MSRE components were operated out-of-pile under conditions resembling
those of the reactor.

In general, there are two phases of development work done on MSRE
components. The first, in areas not well defined, consists in evolving
component and system design specifications through experimentation. The
second consists in operating prototypes of well-defined components for
use in the MSRE to evaluate the adequacy of performance and safety. It
is the purpose of this paper to discuss the effect of the development
work in establishing the reliability of the MSRE components, and, mostly,
it will cover only the second phase.

Core Hydraulics and Heat Transfer

First, I would like to discuss the core hydraulics and heat transfer.
The first phase of development consisted in building and operating with
water a 1/5 linearly scaled model as a rough, preliminary design check in
establishing the acceptability of design concepts. The scale factor of
1/5 was chosen because, when the model was operated with water at 95°F
with fluid velocities egual to the salt velocities in the reactor, the
168

Fig. 1. Full-Scale Core Model.

169

model was fluid-dynamically similar to the reactor, and certain measur-
able parameters were related by a simple proportionality constant. This
constant was 5 for the linear dimensions and fluid age, 1.38 for the tur-
bulent heat transfer coefficient, and 1 for fluid velocity, Reynolds
number, and relative fluid pressure gradients. The model was used prin-
cipally to design the entrance volute, the flow distribution orifice as-
sembly, and the antiswirl vane configuration in the bottom head.

The second phase of development used a full-scale core model, which
is almost an exact duplicate of the reactor. TFigure 1 shows this model.
Tt is constructed of carbon steel, except for the core blocks and part
of the moderator support grid, which are aluminum. Other differences,
such as increased tolerances, were made for convenience and economics,
Note the plastic windows, probe access holes, and the girth flange. All
the initial data were taken with water as the fluid. This results in a
Reynolds number for the model that is about four times that of the re-
actor. Later in the program, a thickening agent { JAGUAR-508, by Stein
Hall and Co.) was added in order to more closely simulate the Reynoclds
numbers, and measurements which were suspected of having a strong depend-
ence on the Reynolds number were repeated.

The areas in which measurements were made may be seen in Fig. 6,
p.18. I shall quickly run through the measurement in the same order as
the flow of fluid through the vessel: first, into the entrance volute,
then through the flow distribution corifices, down the shell cooling wall
annulus, intc the lower head, through the core support lattice and core,
into the upper head, and out the exit nozzle. Figure 2 gives the center-
line velocity of the volute at three flow rates, The linear decrease in
velocity indicates that the angular distribution of flow into the core-
wall cooling annulus is uniform. Since it is desirable to maintain high
velocities and high heat transfer coefficients through the cooling an--
nulus, the orifice holes were drilled at an angle to retain in the an-
nulus the rotational flow established in the volute. Figure 3 is a plot
of the resultant center-line wvelocity vs elevation in the annulus. ' The.
decrease in velocity results from the alteration of the tangential com~-
ponent, Measurements of the center-line velocity distribution around the
core at the bottom of the annulus indic¢ated that the flow to the lOWerj
head is uniform. s

If the tangential component of the flow velocity were not removed by
the antiswirl vanes in the lower head, it would produce an unfavorable .
vortex-type pressure distribution at the entrance into the moderator
blocks. The antiswirl vanes performed well, as can be seen in Fig. 4.
Due to the radial pressure distribution at the wall of the lower head,
the pressure is a little higher in the center. This is in the right di-
rection to promote higher flow rates through the center of the core. An
additional attractive feature of this entrance condition in the lower
head is the production of a Jet at the wall as ‘'shown: in Fig. 5. This
velocity profile was measured at a radius of 17 in. and a flow rate of
1200 gpm. A heat transfer coefficient of 1200 Btu hr? ft™ (°F)™! was
calculated from these data. This Jet did not persist much past this
point, and, in fact, the flow over the vessel center line is gusty and
170

UNCLASSIFIED
ORNL-DWG 64-6721A

24 T
,/‘. : ]
[ I
Vs . \ /
7 \0\. ~ !
20 7 bt 1 ¢
' N /
@ S MEAN INLET "\ /
= PIPE VELOCITY I’
o 16 ® ~a.
= PN /
3 ® . /
)
3 S1200 gpm /
12 {
|<(>_- R R SN \“\0 ° Il /
S n L] o\“__. \f /
1 8 \\ /
L;‘.J .-;..._. /
‘\ /
= ) 600 gpm
aj .N.QP //
i 4 e e - ‘T .‘.‘!" w
0
0 45 90 135 180 225 270 315 360

8, ANGLE FROM INLET TANGENT (deg)

Fig. 2. Center-Line Velocity Distribution in Volute.

VERTICAL DISTANCE UP CORE WALL COOLING ANNULUS (in.)

TG

50

40

30

20

UNCLASSIFIED
ORNL-DWG 64-6723A

ELEVATION

OF VOLUTE

300 gpm/ /

J

/600 gpm
®

/
./
|

!

|

/
]
./
|

I

l

2

4

6

8 10 12

SALT VELOCITY AT ANNULUS ¢ (fps)

Fig. 3. Center-Line Velocity Distribution vs Elevation in Cooling

Annulus.

8
171

UNCLASSIFIED
ORNL-DWG 64-67224

'/ ANNULUS

7
| .

MODERATOR CAN

-

VANE

PRESSURE TAPS

=0
%
]

=
Bo .
g

od — ____
._c_’$0.2

» > |
== —
ST -~
2E . !
83(/) /
3t . '
s |

-02 —

Fige 4. Radial Pressure Gradient at Wall of Lower Head.

i

UNCLASSIFIED
ORNL-DWG 64-6727A

2.5

%]
(@]
%/

e N
s \
-
<L A
. $ 15 BN
= \
(@]
. o
v '8
w DO Wk
=
g
-
: ©
)
0.5 A

A (o] -

O et

0 0.5 1.0 1.5 2.0 2.5 3.0
FUEL SALT VELOCITY (fps}

ks

Fig. 5. Velocity Profile of Wall Jet in Lower Vessel Head at a Ra-
dius of 17 in. and a Flow Rate of 1200 gpm. Measurements were made at
90° to each other.
172

not predominant in any one direction. The results of measurements made
with a heat meter in this region are shown in Fig. 6. There are some un-
certainties in correlating the data at 4 in.; however, the results are
believed to be within 50% and may be somewhat conservative, when thermal
convention is included. Both curves indicate turbulence in this area.

UNCLASSIFIED
ORNL-DWG 64-6726A

IR | [ [ []

e RADIUS=17 in.- DETERMINED WITH HEAT METER
- &4 RADIUS =17 in.-DETERMINED FROM VELOCITY PROFILE 4
4 RADIUS = 4in.-DETERMINED WITH HEAT METER

2, SLOPE FOR
"/ TURBULENT

w

n

'y ~ Re0.8
103 // /FLOW (#~Re )
A ry.d
] P
// //
) 1
5 v A
7/ //fi
e

HEAT TRANSFER COEFFICIENT [Btu hr™" 172 (oF) ™" |

100 200 500 1000 2000 5000
SALT FLOW RATE (gpm)

Fig. 6. Heat Transfer Coeffic¢ients in Lower Head of Reactor Vessel.

We now come to the moderator assembly, where the graphite support
lattice and the moderator core blocks are. The support lattice consists
of two layers of rectangular bars, one layer on top of, and perpendicular
to, the other. The resultant square fuel passage holes are small, and
the high fuel velocity through them produces an orificing effect. The
velocity through the core blocks is much lower; consequently, the bulk .
of moderator assembly frictional head loss is taken in the lattice bars,
The data in Fig. 7 were taken with water and later verified with a thick-
ening agent, added for Reynolds number similarity. The total assembly
pressure drop 1s shown along with the pressure drop across the lattice
support measured in a separate mockup. These curves indicate turbulent
flow in the moderator fuel channels, although it should be laminar since
the Reynolds number in the channels is about 1000, based on the equivalent
diameter. This disagreement only indicates that the entrance conditions
persist well up into the channel., The flow distributions among the
graphite channels are shown in Fig. 8. The two curves represent data
taken in'mutually perpendicular channels having different inlet condi-
tions due to this support lattice, as shown in the sketch. To correct
this deficiency, l/lO-in.-diam'holes were drilled in the upper graphite
support bars directly under the starved channels., This was the only
change indicated necessary by the measurements. Note also that the flow
decreases slightly with increasing radius, which reflects the radial pres-
sure distribution measured in the lower head., Measurements made in the
upper head indicated that no vortexing existed there. The overall pres-
sure drop through the core, from the 5-in. inlet pipe to this 5-in. exit
173

UNCLASSIFIED
ORNL-DWG 64-67244A

1.0 I I I
——
7 —
F}?SLOPE 2.0
o
0.5 3
y/
o /
o
= J
- 02 Py
0
8 / PRESSURE DROP RESULTING
-~ TOTAL .,/ FROM SUPPORT LATTICE
2 0. //
] 7
T )
[
e 7
0.02
100 200 500 1000 2000 5000 10,000

SALT FLOW RATE (gpm)

Fig. 7. Fuel Frictional Pressure Drop Across Moderator Assembly,

UNCLASSIFIED
ORNL-DWG 64-6T19A

2‘8 I }
CHANNELS DESIGNATED BY @
CHANNELS DESIGNATED BY A
2.4 | i
. PLAN VIEW OF CORE SUPPORT BARS
— WITH FUEL CHANNELS SUPERIMPOSED
E MoNoNEl | |
(e
f’ 2.0 S ‘
Z 000 | |
2 . —— TYPICAL HOLE THROUGH SUPPORT
I ~ STRUCTURE TO GET MORE FUEL TO
16 . 0 THE STARVED CHANNEL | z
L I
2 REACTOR CORE CAN~ 7/
z ¢
" %
g 1.2 7
q .
: 7
7
: z
| é
(VI
H o.8
: . .
@ LEAST SQUARE LINES
.
v
0.4 ‘ &
B DATA TAKEN BY DIFFERENT TECHNIQUE IN ANNULUS Z
BETWEEN GRAPHITE AND MODERATOR SHELL 7
é
0 v,
0 6 12 18 24 30

RADIUS OF CHANNEL FROM CORE CENTER LINE (in.)

Fig. 8. Flow Distribution of Fuel Among Core Channels.
174

pipe, is given in Fig. 9. Note that the line has a slope of 2; which in-
dicates that the pressure drop is independent of Reynolds number.

Experiments were run to study the tendency of solids to settle out in
the lower head. It was found that solids with a specific gravity of 10
and less than 300 p in size would continue through the lower head and up
through the moderator, causing no problem.

In summary, the results of these tests indicate that the flow dis-
tribution within the reactor vessel assembly will provide adequate cool-
ing to all parts of the assembly.

UNCLASSIFIED
ORNL -DWG 64-6720A

20

~ea

FLUID FRICTIONAL HEAD LOSS (ft fluid)
i

2 /

100 200 500 1000 2000 5000 10,000
FLOW RATE (gpm)

Fig. 9. Owverall Fluid Frictional Head Loss Across MSRE Core.

Heat Exchanger

- Hydrautic tests, similar to those just described for the reactor,
were run on the primary heat exchanger. Figure 10 shows this test being
performed last December, The initial tests indicated that the pressure
drop through the tube side agreed with the calculations, but that the
pressure drop on the shell side was excessive and that the tubes vibrated
excessively at flow rates above 800 gpm in that side. The INOR-8 shell
was removed, and a special stainless steel test shell was installed. As
a result of subsequent tests, the tube vibrations were eliminated by at-
taching tight-fitting spacer bars to the baffle plates between the rows
of tubes, and the pressure drop through the shell side was reduced by re-
moving four tubes and increasing the size of the outlet nozzle. TFigure
11 shows the fluid frictional. head loss, as measured in the heat exchanger
ONCLASSIFIED
PHOTO 71340

Fig. 10. Heat Exchanger Test.

UNCLASSIFIED
ORNL-DWG 64-67254

H

40

20 |—

FLUID FRICTIONAL HEAD LOSS (ft fluid)

200 500 1000 2000 5000
FLOW RATE (gpm)

Fig. 11. Fluid Friction Head Loss in Primary Heat Exchanger.

UNCLASSIFIED]
PHOTO 365514)

176
DISCONNECT IS

Core Heater.

Fig. 12,
177

after the alterations were completed. The slopes of the two curves in-
dicate that the pressure drop on the shell side is mostly due to entrance
and exit losses and that the pressure drop on the tube side results mainly
from frictional losses. We believe that the heat exchanger will now per-
form as originally planned,

Heaters

The next development area that I would like to discuss is concerned
with the wvessel and line heater and insulation packages. These units
provide a controlled temperature distribution during heatup and ccoldown
as well as reduce the heat loss to the reactor and drain tank cell during
high-temperature operation. Figure 12 shows a full-scale mockup of a
section comprising 15% of the heater for the reactor furnace which was op-
erated for 20,000 hr at 1250°F without incident.

A prototype of a single heater for the drain tank was operated at
1200°F over 10,000 hr without difficulty. Figure 13 shows the heater
during assembly. The leads to the ceramic elements are interconnected
by welding, and one of these welds failed early in the test operation.

The design and inspection of the weld was improved, and no further trouble
- was encountered. -

The heaters for the 5-in. pipes are enclosed in all-metal reflective
insulation which was chosen because of the dusting problem normally en-
countered with refractory materials when much handling is required. Heater-
insulation units of the type shown in Fig. 14 have operated with the in-
terior above 1200°F for more than 5000 hr without difficulty or apparent
deterioration, as would be indicated by a change in heat loss.

Another unit associated with the control of temperature of the ves-
sels is the drain tank cooler, which is used to remove the afterheat fol-
lowing extended operation at full power. Figure 15 shows the parts for
the test of a three-tube mockup of the original design of this cooler.
The reentrant thimbles shown were immersed in a salt which was maintained
between 1300 and 1400°F. The Inconel cooling bayonets, which include.
smaller concentric tubes for cooling water flow, were inserted into the.
thimble and cycled by allowing water to flow into the bayonets. Since
the temperature of bayonets had been above 1300°F, the thermal shock from
the water quench was terrific. Figure 16 shows a portion of the inner -
tube which had experienced more than 2600 such thermal cycles. ' The de-
sign of these tubes was changed in the area of the welds at the spacer
fins, and new test units were constructed from)INOR-8. These units have
operated through more than 1000 thermal cycles with no evidence of crack-
ing, even at the surface. On the basis of these tests, the drain tank
cooler should operate without trouble through at least 100 reactor shut-
downs from full power.
178

UNCLASSIFIED
PHOTO 39334

Rie

Fig. 13. Drain Tank Heater During Assembly.

179

UNCLASSIFIED
PHOTO 70759

Fig. 14. Pipe Heater.
180

Freeze Flanges

Mechanical-type joints are provided in the 5-in. piping in the fuel-
and coolant-salt system inside the reactor cell to permit the major equip-
ment to be disconnected for removal and replacement. The so-called "freeze
flange" type of joint was chosen because of its proven reliability in pro-
viding tight connections with zero salt leakage and insignificant gas leak-
age under all anticipated thermal-cycling conditions. Freeze flanges have
been a part of molten-salt engineering development for several years.
Flanges with integral cooling were incorporated in the Remote Maintenance
Demonstrations Facility, where it was shown that they could withstand re-
peated thermal cycles and could be broken and reassembled. As a result
of these preliminary tests, the design of the flange was improved to pro-
vide greater strength toward axial loads, to reduce the thermal stress, to
improve the tightness of the buffer seal, and to remove the requirement
of integral cooling. TFlanges of the improved design (shown in Fig. 17)
were thermal cycled between room temperature and 1300°F more than 100
times without significant changes which would indicate excessive thermal

ncasrieo
¥ horo e

1-in. COOLING BAYONETS
CONTAINING Yo-in. WATER
INLET TUBES

1%3-in. CONTAINMENT THIMBLES

Fig. 15. Drain Tank Cooler Test Components.

181

UNCLASSIFIED
PHOTO 398044

THERMAL
STRESS
CRACKS

»
z
25
=

Fig. 16. Drain Tank Cooler 1/2-in. Tube.

182

stress. During all these tests the gas leakage from the buffer zone re-
mained less than 5 x 107 gtd cm® of helium per second. It was found that
a nev flange would make up and seal satisfactorily with a flange which
had experienced several thermal cycles from room temperature to 1300°F.

In addition, there was a flange pair of this type incorporated in the
prototype pump test loop, described in the section on pump development,
and it has operated satisfactorily throughout that test. We have found
nothing, from these or other tests, which would indicate problems with
the use of these freeze flanges in the MSRE.

In addition to the work on freeze-flange disconnects, small, remotely
maintainable buffered seal disconnects were developed for use in the
sampler-enricher and the various gas lines of the MSRE., These disconnects
satisfactorily demonstrated their reliability under extreme temperature
conditions of operation.

UNCLASSIFIED
ORNL-LR-DWG 63248

FLANGE
CLAMP ——
2-in. TAPER
0400in./in.
5 (TYP) n.
BUFFER / %
CONNECTION 1'%4-in. TAPER
0.050 in./in.
MODIFIED R-68 (TYP)
RING GASKET - T
SCREEN ;
INSERT H¥%g-in. R
|
| L
4/2-in.-fi (TYP)
i
/ SLOPE {:4
4% in.
I (TYP)
5-in. SCHED-40 PIPE SiZE

Fig. 17. Freeze Flange.
183

Freeze Valves

A reliable high-temperature mechanical valve was not available at
the start of this project. However, there was much experience with
frozen plugs of salt in controlling the transfer of salt, It was nec-
essary to develop and demonstrate a valve system which would operate
reliably and give consistent indications of its condition. Many vari-
ations of freeze valves were built and tested; Fig. 18 shows the arrange-
ment chosen for use in the reactor drain line at the drain tank.

This valve consists of a flattened section of pipe surrounded by an
insulated air shroud, all of which is enclosed in a special removable pipe
heater. It is characteristic of this wvalve that the frozen plug is formed
within the insulated air shroud and that thawing must proceed from the
ends of the plug toward the center. Since the energy for the thawing
operation comes primarily from that stored in the removable pipe heater,
the time required to open the valve is changed only slightly by a power
failure. When it is required that a valve always remain frozen, the
gspecial pipe heater is allowed to cool to below the melting point of the
salt. Valves of this type have operated through more than 100 freeze-thaw
cycles without dimensional change, and require approximately 20 min to
freeze and 15 min to thaw. They can be frozen even with salt flowing
through them at rates of 300 in.’/hr. Results of these and other tests
indicate to us that these freeze valves are satisfactory for the MSRE.

Control Rod System

The control rod drive is shown in Fig. 19 along with the drive housing.
The development problems with this control rod system for the MSRE resulted
from the facts that the control rod must be flexible enough to go around
an offset in the rod's path and that oil bath lubrication of the rotary
parts of the drive mechanism is undesirable. The flexibility requirement
created problems ‘related to the effect of the stretch of the rod on posi-
tion indication and wear at the contact surfaces in the path offset. A
structural mockup of the control rod is shown in Fig. 20. Note the bellows
inside the metal braid in the upper section and the wire cable inside the
spiral hose in the lower sectlion. This construction method was necessary
to reduce the stretch of the rod, while permitting cooling air flow through
the center. A method was developed which remotely calibrates the position
of the bottom of the control rod with the remotely indicated position of
*the top of the rod by using this air flow. The wear problem was solved
by incorporating graphite-bushed rollers into the guide thimble at the
critical locations in the offset. These and other small changes were in-
corporated into the final design, and a prototype rod and guide thimble
was then successfully operated at 1200°F through 38,000 full cycles of
102 in. of travel per cycle.

A radiation-resistant grease was used to lubricate the rotary ele-
ments of the control rod drive and proved satisfactory except at the
worm drive section. The wiping action at the contact surface removed
the grease there, causing overheating and destruction of the aluminum
Fig. 18.

184

AIR INPUT

Freeze Valve.

UNCLASSIFIED
i 7HoTo 70156

185

UNCLASSIFIED
PHOTO 62053

DISCONNECT

GEAR BOX

HOUSING

Control Rod Drive.

Fig. 19.
186

UNCLASSIFIED
PHGTO w06zs

Gd POISON ELEMENT
ASSEMBLED ELEMENT
WIRE CABLE
CONVOLUTED TUBE

BRAIDED HOSE

SPIRAL -WOUND HOSE

AIR OUTLET $.0.3 2

Fig. 20. Structural Mockup of MSRE Control Rod.

bronze worm wheel after 16,000 full stroke cycles. While this perfor-
mance was considered adequate for the MSRE, testing of other materials

has continued. A preliminary test using & hardened-steel pair of gears
operated in excess of 29,000 full stroke cycles without significant wear.
Gears of this type have been specified for use in the reactor. There

have been some other changes as a result of the development testing, but
they are in the category of optimization and do not affect the reliability.

Cover-Gas System

Another area examined for problems was the cover-gas system. This
system consists of the inert-gas supply section, the distribution and
control section, and the off-gas disposal sections. Of these, only the
purification portion of the inert-gas supply section required experimental
study, since the other sections are based on what has become a well-es-
tablished technology. The inert-gas supply section consists of the supply
and storage reservoirs, the oxygen- and water-removal units, and the on-
line gas analysis section. On the basis of preliminary experiments, full-
scale oxygen- and water-removal units were constructed and tested; the
oxygen-removal unit is shown in Fig. 21. The test results indicated that,
with a titanium bed temperature of 1200°F and an inlet oxygen concentra-
tion of 100 ppm, an outlet concentration of less than 1 ppm was maintained
until 58% of the titanium sponge had been consumed. Since the average
oxygen concentration at the inlet to the purifier for the MSRE is not ex-
pected to exceed 50 ppm, the life of a titanium charge should be six months
or longer with a helium flow of 6000 liters/day. The performance of a
commercial oxygen analyzer was checked during the helium purifier experi-
ment and was deemed satisfactory after modification by the manufacturer.

The helium dryers were designed using the manufacturer's design pa-
rameter, and performance checks were limited to an occasional check of
the water concentration at the dryer outlet. We determined that the dryer
bed would require regeneration after 30 days of operation with an inlet
moisture concentration of 50 ppm of H,O. Two dryers, installed in par-
allel for alternate operation in the MSRE, provide for the regeneration.

: 187

UNCLASSIFIED

4 CRNL—LR-DWG 68585
. THERMOCOUPLE
\
O ~
= — (1) GETTER TUBE
| THERMOCOUPLE —_| | (2) HEATER, 1000 w
v (3) HIGH-TEMPERATURE INSULATION
(@) PIPE, 4-in, SCHED-40 SS
@ (5) FLANGES, 4-in. 1500-b SS,
26%in. ] WELDING NECK, RING JOINT
& (&) INSULATION
2tin. ©) (7) REFLECTOR
THERMOCOUPLE ~ |— THERMOCOUPLE
3 THERMOCOUPLE ~ | THERMOCOUPLE
. Q
d THERMOCOUPLE
h %
ol h—O®
vl ’
1
5 g | THERMOCOUPLE
3 LA LA
rll'

Fig. 21. Oxygen-Removal Unit.

While no reliable method for the continuous detection of water in helium is
avallable at this time, some work is continuing toward that goal, and in
the meantime occasional checks will be made of the output of the purifi-
cation section. The results of our tests indicate that the capacity and
- performance of the purification sections are more than adequate for use
] in the MSRE.

Sampler-Enricher

The major problem in the sampling and enriching of liquid-fuel re-
actors is the required penetration of the double contaimment. In general,
the system developed for the MSRE uses a series of containment areas through
which the sample moves after it is isolated from the circulating stream.
Figure 22 shows the component arrangement for the sampler-enricher. Basi-
cally, it is a power-driven cable which lowers a small bucket into the
liquid of the pump bowl. The cable then raises the bucket into a manipu-
lator area for transfer into a shielded container for transport to the
analytical chemistry hot cell., Figure 23 shows a sample and an enriching
capsule of the type tested. Prototypes of critical components were as-
sembled into a.mockup of the sampler-enricher system, which was used to
remove some 56 salt samples and to add 11 capsules of enriching salt to
188 A

UNCLASSIFIED
ORNL-DWG 63-5848%

—~REMOVAL VALVE AND -
SHAFT SEAL

CAPSULE DRIVE UNIT

LATCH-
ACCESS PORT—1.
V/

A 4 ¢
(PRIMARY COMTAINMENT) —

H[ \?'fir

SAMPLE CAPSULE — ‘/77" \_ MANIPULATOR

—AREA 3A (SECONDARY
CONTAINMENT)

“— SAMPLE_TRANSPORT
CONTAINER

OPERATIONAL AND

MAINTENANCE VALVES < ~ LEAD SHIELDING £

SPRING CLAMP

== AREA 28 (SECONDARY
DISCONNECT

CONTAINMENT) s

TRANSFER TUBE
(PRIMARY CONTAINMENT)

LATCH STOP
PUMPS
/

,/ CRITICAL CLOSURES

REQUIRING A BUFFERED

. s
y& wisT sHeLo

2 SRR U I
""‘ CAPSULE GLIDE

FEET

Fig. 22. Component Arrangement for Sampler-Enricher.

UNCLASSIFIED
PHOTO 713334

~=— LATCH PIN—=—

ENRICHING CAPSULE
SAMPLE CAPSULE

Fig. 23. Sample Capsule and Enriching Capsule.

189

a circulating loop. After some adjustments and minor alterations, the
system operated reliably and safely. The results of the chemical analyses
of the samples agreed within 1—1/2% with the results from samples taken
with another proven method. These and other tests indicate that there
should be noc major difficulty with the use of these sampler systems in
the MSRE.

Engineering Test Loop

The final area I would like to discuss concerns a collection of simu-
lated MSRE components, the Engineering Test Loop, shown in Fig. 24. 1In
this loop satisfactory performance of the freeze valves, level indicators,
sampler-enricher, frozen salt-graphite access joint, and gas-handling
system was demonstrated under simulated reactor operating conditions. A
method of gas purging and salt flushing the system to remove oxide before
operations with fuel was demonstrated. A method was demonstrated for the
removal of oxide from the salt by treating with HF and hydrogen in the
drain tank. The Engineering Test Loop was operated in excess of 15,400

UNCLASSIFIED
ORNL-LR-DWG 344924

~——SAMPLER-ENRICHER MOCK UP

SAMPLER~ENRICHER TRANSFER TUBE

DANA PUMP
GAS

—a—GRAPHITE HANDLING DRY BOX

MANUAL
SAMPLER

GRAPHITE ACCESS
(FREEZE SEAL)

~—8-in, PIPE (GRAPHITE CONTAINER) T

FREEZE VALVES MANUAL SAMPLER

Mg, 24, Engineering Test Loop.
190

hr at 1200°F., The last 7000 hr of the test was conducted with LiF-BeF,-
ZrF,-UF, salt to test the sampler-enricher and to follow the inventory
of uranium added to the loop. The uranium concentration was varied from
0.1 mole % at the beginning of the period to about 0.7 mole % at the end,
The results of the chemical analyses agreed with the book inventory of
the additions within the experimental uncertainties of 3%. Metal and
graphite samples were removed periodically for examinations, and no un-
usual changes were noticed.

Summary

In summary, the performance of the components tested has exceeded
the requirements for the MSRE. 1In addition, the components needing main-
tenance have been made easily replaceable, Finally, the components and

systems have been shown to be compatible when operated as a loop. This

experience leads us to believe that the MSRE should have a long and un-
eventful life.

REMOTE MAINTENANCE OF THE MSRE

Robert Blumberg

To be attractive as a power producer, this reactor must possess a

. measure of serviceability., To this end the maintenance criteria influ-
enced not only the system design but also the entire facility. The major
requirement, stated simply, is that all the fuel-salt-carrying equipment
and all other equipment that may fail in the reactor and the drain cell
must be remotely replaceable.

Two structural types were considered. In the first, a large vessel
with complex intermals housed all the important elements of the systen,
such as the core, pump, heat exchanger, etc. Maintenance involved opening
a large vessel closure and then working down, layer by layer, until the
failed element was reached. This was discarded in favor of its counter-
part, a system composed of a number of separate pieces of equipment joined
by piping. This arrangement led to this set of guides or rules to make
the MSRE capable of being easily maintained:

1. The system should be composed of several separate units joined by
suitable disconnects.
2. Each unit should be accessible in the plan view of the reactor.

3. Removal and replacement should be considered in the design of each
unit.

4. Maintenance equipment should be provided to handle the equipment from
| above, |

8
191

It was originally planned to maintain the MSEE with a bridge-mounted
manipulator, remotely controlled, using television for auxiliary viewing.
However, because the MSRE was fitted into an existing building, there was
not enough head room for this. The system that was adopted is a combina-
tion of the totally remote and the semidirect (long-handled) approaches.
This combination of techniques stems from experience gained in operating
the molten-salt remote maintenance demonstration facilityl and in main-
taining the Homogeneous Reactor Test (HRT).

Schematically, the method of maintenance is as follows., For small
pieces of equipment such as heaters and valves, a portable maintenance
shield, similar to the one developed for the HRT,2 is set up over the
opening left by the removal of two adjacent shield blocks. Long-handled
tools are inserted through the portable shield (Fig. 1) to disconnect at-
tachments and 1ift the unit out of the cell. Direct viewing is provided
by built-in windows and lighting and by periscopes inserted through the
shield. TFor large pieces of equipment, the portable shield and long-
handled tools are used to prepare the item for a direct 1ift. This may
involve discomnecting attachments, displacing the unit to provide clear-
ance, and installing a 1ift fixture. The rest of the procedure is carried
on by personnel from inside a shielded control room which overlooks both
cells. From there, using remotely controlled cranes and a variety of ac-
cessories, sufficient shielding is remowved from the cell roof to gain ac-
cess to the item to be handled. The crane is then used to remove the item
from the cell, put it into the storage pit, and replace the shielding
blocks. Viewing is provided by windows in the wall of the control room
and a pair of portable television cameras.

In summary, the MSRE, from a maintenance viewpoint, is a collection
of replaceable parts joined by disconnects. Small parts will be main-
tained by long-handled tools inserted through a portable shield, and large
parts will be prepared for removal with the long-handled tools and then
removed by remote equipment. There are three general divisions that com-
prise the equipment to be used; the shielding, the long-handled tools, and
the remote equipment.

Shielding

The reactor and drain cells are shielded by two layers of concrete
blocks., For maintenance, the top layer is taken out of the way, and the
upper surface of the lower layer then becomes the working floor for the
maintenance crew. The lower layer has been designed to meet various needs
of remote maintenance. First, the width of the blocks provides, as far
as possible, a 2-ft modular grid to coincide with the 4-ft opening of the
portable shield. Second, the blocks are laid out to allow convenient ac-
cess to the major components in the cell. Third, penetrations through the
block are provided where necessary, allowing access for tools without re-
moving the shield block. ¥Finally, the blocks are equipped with pickup
sockets, which allow them to be handled remotely.

The portable maintenance shield is a device which is set up over any
4-ft opening in the shielding. It provides 12 in. of steel shielding for
the personnel who will insert long-handled tools through bushed holes.
192 -

UNCLASSIFIED
PHOTO 63913A

ROTATING WORK PLUG

Fig. 1. MSRE Maintenance Shield.
193

For convenience, a total of 17 holes are provided at strategic locations.
A combination of linear and rotary motions provides coverage over the en-
tire 4-ft opening. Larger pieces of equipment, such as heaters and valves,
are passed through the shield by opening the joint between two adjacent
slabs of the portable shielding. A hole on the edge of one side of the
Joint is provided so the handling tool can support the equipment while it
is made ready for removal. A motor provides the linear motion of the
slide and can be operated from the control room. This permits evacuation
of personnel when large openings are made in the shielding. Floodlights
and windows are provided as plug-in inserts in the shield slabs and can
be moved as required. Special cutouts and track sections have been pro-
vided for areas where interferences of the building facilities exist.

Long-Handled Tools

Most of the in-cell operations will be done with long-handled tools.
Some typical examples are shown in Fig. 2. These tools have been made
simple and sturdy, to the extent that whenever a choice exists between
two designs, the factors of strength, reliability, and simplicity are

o
¥ PHOTO 70460

PIPE ALIGNMENT FOR FF-400

*‘/““ S

Pt
—
S—
-2 HEATER AND DISCONNECT
' ! ~ LONG-DISTANCE "C" CLAMP

THERMOCOUPLE DISCONNECT

Fig. 2. Lower Ends of Five Miscellaneous Long-Handled Tools.
194

considered more important than appearance, ease of operation, and versa-
tility. Experience has indicated that simple tools are more reliable than
complex ones. Also, when they become contaminated, they are easier to
clean, store, and maintain and are less expensive to replace. In gen-
eral, the tools consist of a pipe mast with rods or cables inside to ac-
tuate movable parts at the working end. In some cases, they are merely
extensions of simple tools such as socket wrenches, hooks, clamps, or
pliers. In some cases, however, the basic character of the tool is masked
in order to fulfill special requirements. An example of this is the ther-
mocouple-disconnect tool. This is basically a hook that can be drawn up
tight against a stop to give a solid grip on the disconnect handling bail.
However, the alignment guides make the tool look somewhat more complex
than a simple hook. An example of a more complex tool is the pipe align-
ment tool. It provides both vertical and horizontal motion for proper
positioning of mating flanges. The heart of this device is a scissors
Jjack and a lead screw, There are approximately 40 different long-handled
tools.

The in-cell operations can be viewed through the lead glass windows
in the portable shield (with or without binoculars) and also through a
periscope which is inserted in the same manner as a long-handled tool.
The periscope is a commercially available, 1,625-in,-diam instrument with
radiation-resistant optics and both right-angle and straight-down objec-
tive lenses. It is sheathed in a 2-in.-O0D stainless steel tube which
protects it from contamination, Lighting is provided by special drop
lights and built-in floodlights. In doing the maintenance, one individual
observes the working end of the tool down in the cell with the scope and
relays instructions to the tool operator. A third man will, perhaps, ob-
serve through the windows. Thus, the operation is a team effort and a
high degree of skill is not as important as a good knowledge of the tools
and the in-cell equipment. '

Remote Equipment

When large openings are made in the cell (two lower shield blocks
or more), high radiation levels make the high bay area uninhabitable for
personnel. This situation will occur when handling blocks in preparation
for semidirect maintenance, when the portable shield is spread apart to
pass a large plece of equipment, or when handling large components such
as the fuel pump bowl, heat exchanger, drain line, etc, During these
times the reactor building is evacuated and the maintenance crew go into
the shielded control room. The key piece of equipment that will be oper-
ated from there is the 30-ton crane. The equipment in the control room
consists of the following:

1. two shield windows filled with zinc bromide, located to permit direct
viewing of the high bay, the reactor, and the drain cells,

2. controls and monitors for three closed-circuit television cameras,

3. control boxes for operating both the 10- and 30-ton bridge cranes,
including a switch to rotate the 30-ton crane hook,
195

4e a weigh cell readout indicating (in pounds) the load on the 30-ton
crane,

5. switches to operate the slide of the portable shield.
Located outside the control room (in the high bay area) are a number

of accessories to be used with the crane. They are all set up or sup-
ported so as to be available for pickup by the crane hook. These include:

l. a shield-block support beam pickup device,

2. two shield-block pickup tongs (one of which is shown in Fig. 3)
equipped with cam devices which alternately open and close the tongs,

3. tTwo portable camera stands with large handling bails, camera plat~
forms, height adjustments, and heavy base plates for stability,

4, an assortment of 1lift fixtures to engage the crane hooks with the
component to be handled.

UNCLASSIFIED
ORNL-LR-DWG 61193

) / MOTORIZED CRANE HOOK

SHIELDING PLUG—""

Fig. 3. Remotely Operated Lifting Tongs for Shielding Plugs.
196

At the present time, control room operations are limited to what can
be done with the crane and its accessories. The complexity of the task
arises in the differences between operating a crane remotely from the con-
trol room and operating it directly with a rigger crew. These differences
are basically those of viewing, feel, and method of engaging the hook.
Whereas a rigger crew will casually use slings, bars, shackles, etc., we
have had to preplan all operations in order to hawve a suitable handling
bail or socket available for the crane. The restricted viewing and feel
of the operator inside the control room are augmented by binoculars, tel-
evision, and the weigh cell. The crane controls feature extremely slow
speeds. Further, specialized practice, familiarity with the equipment,
and the elimination of speed and haste combine to make our capablllty ap-
proach that of ordlnary crane usage.

The Development Effort

Starting from the idea of unit replacement and ending with a system
capable of maintaining the MSRE required going through a series of activi-
ties which included design, development, testing, and the recording of in-
formation. Development programs were started as soon as problefi areas
were identified. The end products of these programs were actual working
tools,. operating procedures, and proven equipment.

Remote Operation of the 5-in. Freeze Flange

One major objective was to establish the technique of making and
breaking the five special flanges which connect the primary loop compo-
nents. By use of a full-scale mockup, where the sight, space, and dis-
tance restrictions were similar to those of the reactor, tools and tech-
niques were worked out and refined. The overall procedure was separated
as follows: ‘

i. driving the flange clamps onto and off of the flanges with a design
- load of 40 tons (Fig. 4 shows the tools in place in the mockup),

2o storing the heavy clamp assembly on an in-cell bracket,

3. replacing a ring gasket without damaging its sealing surface,

.4; 'dovering the open piping to reduce the spread of contamination,

5. making up the flanges with sufficient force and precision to assure

- a good seal.

Six different tools are used to accomplish this manipulation, which can
be considered to be a small extrapolation of existing long-~handled tool
experience.
UNCLASSIFIED
PHOTO 39120

HYDRAULIC CLAMP

OPERATOR

Fig. 4. Remote Assembly of Freeze-Flange Clamps.

198

Remote Cutting and Brazing

Besides the 5-in. piping of the primary system, the only other fuel-
carrying lines that we must be able to be disconnect are the 1-1/2-in.
pipes which interconnect the drain tanks, flush tank, and reactor vessel.
A method of remotely cutting these lines and brazing them back together
again has been developed for replacement of these components. The proce-
dure is to lower a positioning vise into the cell onto a built-in plat-
form. A pipe cutter, mounted on the vise, severs the pipe so that the
failed component may be removed. Next, a pipe-tapering device, bolted
to the platform and facing the pipe in the vise, machines the outside of
the pipe, producing the male member of the joint. The furnace assembly,
complete with induction heating coils, inert-gas purge, water cooling,
thermocouples, and insulation, is slipped over the male joint. The re-
placement component, with the female portion of the joint and the end
closure of the furnace, is lowered into place. A stationary vise then
clamps onto the new piece. Figure 5 shows this part of the procedure.
The furnace is closed, operated, and, when the melting temperature of the
braze metal is reached, the positioning vise forces the joint together
for optimum configuration while the metallurgical bonding takes place.
After pressure testing, the furnace and the machinery are taken out of
the cell, and the heaters and insulation are reinstalled. Long-handled

) T s
PURGE LINE = THERMOCOUPLE
FEED SCREW LEADS
ORIVE VISE - JAW

f& INDUGTION
HEATING
LEADS

FIXED VISE

Fig. 5. Joint and Furnace Ready for Assembly.

199

tools, operated through the portable shield, are used to handle the ma-
chinery described. The process and equipment are described more com-
pletely in applicable sections of the MSRE design report and in the open
literature.? Many Jjoints have been produced this way under simulated re-
actor conditions, and all have successfully passed both destructive and
nondestructive tests of their integrity. This process is a step toward
the ultimate goal of highly reliable, easily maintained nuclear systems.
Furthermore, it gives the MSRE the bonus of starting with an all-welded
drain system.

Miscellaneous Development

Besides the above-mentioned problem areas, an effort was made to
unify the design of all the equipment in the reactor that had to be main-
tained and to standardize it wherever possible.

The basic procedures for maintenance were evolved as information be-
came available on radiation levels, head room requirements, component
layout, and costs. The reactor design itself was continuously monitored
to see that it reflected the needs of maintenance. Items whose function
was principally maintenance were designed under the management of main-
tenance engineers. These items included the reactor access stand pipe,
the in-cell jacking and aligning equipment, the portable maintenance
shield, the long-handled tools, and the lift fixtures. A program of test-
ing and demonstration was included in the construction schedule, and is
now partially complete. The results of work in the drain cell are very
encouraging.4 Finally, procedures are being prepared® that will describe
the maintenance equipment and how it is to be used at the MSRE.

Present Status

The status of the maintenance effort is summarized below. The work
remaining is noted in parentheses, and the percentages are approximate.
We estimate that four months of effort remain for one engineer and one
technician., ~

ety zeomt
Developmental testing (set up mockup of 98
graphite sample handling) :
Writing procedures 50
Monitoring reactor design and construction 85
(reactor cell assembly)
Design of long-handled tools 92
Fabrication and testing of long-handled tools 85
Testing and demonstration at the reactor : 30

(reactor cell components, freeze flanges,
and graphite sampler)
200

Conclusions

From the beginning, the management of the MSRE considered the attain-

ment of the maintenance objective important enough and difficult enough
to warrant the investment of considerable time and effort. As a result
of this preparation, we look forward to power operation with confidence
in our capability. This is based on the following points.

1.

2e

3.

and.

every effort was made during the design and detailing stages to re-
flect the needs of maintenance;

we are using simple, straightforward methods that hawve been demon-
strated before;

by the time the reactor goes to power, we will have tested working
tools on actual reactor components;

tests at the site have shown that the tools work well and that we can
improvise solutions in unanticipated situations.

We believe, therefore, that the MSRE can be maintained efficiently
safely with the equipment described.

References

W. B. McDonald and C. K. McGlothlan, Trans. Am., Nucl. Soc. 3(1), 191
(June 1960). - B

P. P. Holz, Dry Maintenance Facility for the HRT, ORNL-CF-60-10-85
(October 1960),

E. Co Hise, Fo. W. CoOke, and R. G. Donnelly, Remote Fabrication of
Brazed Structural Joints in Radioactive Piping, ASME-63-WA-53.

R. Blumberg, internal memorandum, July 27, 1964.

R. Blumberg, Remote Maintenance Procedure, MSR-64-13 (Mar. 23, 1964)
(internal use only).

201

FUEL PROCESSING FACILITY

R. B. Lindauer

A fuel processing facility for the MSRE is under construction, with
completion scheduled for February 1965. The facility will be located in
Building 7503 in a cell adjacent to the drain tank cell and will be op-
erated from a separate panelboard in the high-bay area above the cell.

The facility has two functions. The first is to remove any oxide
accumulation in the molten salt by sparging the salt with a mixture of
hydrogen and hydrogen fluoride. The second is to recover uranium from
the molten salt by sparging with fluorine and volatilizing the uranium
as uranium hexafluoride. This second function should not be required
until the program using the 30%-enriched fuel ends. At that time, both
the flush and fuel salts will be fluorinated to remove the uranium of low
enrichment. For the next phase of the program, 93%-enriched uranium will
be added to the fuel salt.

The facility will handle 75-ft> batches of salt, which is about the
volume of blanket salt which would be processed daily in a 1000-Mw (elec-
trical) thermal breeder station on an 80-day cycle. The cost of the fa-
cility will be about $350,000, This appears to be considerably less than
comparable costs for large-scale, on-site processing plants which do not
take advantage of the sharing of facilities with the reactor.

Flowsheet Description: Hp-HF Treatment

Salt will be treated with Hy-HF whenever it is suspected that oxide
contamination has occurred. The flush salt will be treated after the
initial flushing and shakedown operations and after flushing if the system
has been opened for maintenance. TFuel salt could become contaminated
through a leak in the system or by difficulties with the helium blanket
system. Since operation of the reactor at power with a precipitate of
zirconium oxide is not planned, processing will be required before the
oxide solubility limit is exceeded. For fuel salt, this limit is about
80 ppm of oxide. During shakedown operations, however, the solubility
limit of 210 ppm of BeO in flush salt may be exceeded.

A simplified flowsheet of the facility is shown in Fig. 1. The fuel
processing facility is separated from the drain tanks by freeze valves.
The fuel or flush salt is pressured by dry helium gas from the drain or
the flush-salt tank to the fuel storage tank, where processing takes
place. If the salt is fully irradiated fuel salt, a decay time of four
days 1s required before transfer to allow for xenon decay, since the fuel
processing facility is not wvented through the large charcoal beds. If
transfer of a fully irradiated fuel batch is desirable before 26 days'
decay, it will be necessary to cool the batch in the fuel storage tank.
This can be done by circulating the cell wventilation air through a space
around the tank.
202

SALT 200°F NaF ABSORBERS iN CUBICLE UNCLASSIFIED
SAMPLER ORNL—DWG 63-3{23AR

SALT W

CHARGING

HIGH BAY AREA J

0

oy R HF OR F,
” FvV
- _
SALT TO OR
FROM DRAIN Fv

AND FLUSH TANKS

ACTIVATED
CHARCOAL
TRAP

[
|
|
WASTE | FUEL
SALT { STORAGE
TANK I TANK

ABSOLUTE
FILTER

T50°F
NaF BED

CAUSTIC g&g§g&

NEUTRALIZER

Fig. 1. MSRE Fuel Processing System.

The salt will be sparged with a gas mixture of probably 10 volumes
of hydrogen to 1 volume of hydrogen fluoride at a total gas flow rate of
100 liters/min. The hydrogen is required in order to maintain a reducing
condition in the melt and thereby minimize corrosion. The off-gas stream,
consisting of hydrogen, helium, water, and excess hydrogen fluoride,
passes through heated lines to a heated bed of sodium fluoride pellets.
This bed will decontaminate the off-gas stream of volatilized corrosion
and fission products. The gas stream then passes through the absorber
cubicle, which is in the high-bay area above the cell. In this cubicle,
the gas stream can either be sent through the absorber train for uranium
recovery (as will be discussed later), or it can be sent directly back
to the cell. For the hydrogen—hydrogen fluoride treatment, the gas stream
next passes through a cold trap where water and part of the hydrogen flu-
oride are condensed. The oxide-removal rate is followed by observing the
automatic siphoning of the small pot below the baffled vapor separator.
The siphoned liquid, as well as the off-gas stream, is then bubbled
through a 10% KOH solution to neutralize the unreacted HF, Off-gas from
the caustic scrubber is passed through two successive activated-charcoal
canisters, where any volatilized iodine will be removed, and then through
a flame arrester before entering the cell ventilation duct. Air flow in
this duct will be kept well above the 100 cfm required to keep the hy-
drogen content below the explosive limit of 4%. The caustic solution
mist be replaced after about two days of processing. The caustic solution
is jetted to the liquid waste tank, from which it is pumped to the Melton
Valley intermediate-level waste tanks.
‘|

203

Ilowsheet Description: Fluorination

Since some fission product volatilization is expected during fluori-
nation, the fuel salt should be allowed to decay for as long as the re-
actor operating schedule will allow. Fission product volatilization is
minimized also by operating at as low a temperature as possible, about
50 to 100°F gbove the melting point or around 900°F, The most important
fission products that form volatile fluorides are iodine, tellurium, ni-
obium, ruthenium, and antimony. Experience has shown that most of the
volatilized fission products are reduced to the metallic state and plate
out on the. equipment surfaces.

The molten salt is sparged with fluorine at a flow rate of 100
liters/min., About 2 hr is required for the conversion to UF5 of all the
UF, in a salt "C" (0.9 mole % UF,;) fuel batch. At this time, evolution
of UFg will begin, and the fluorine disposal system will be put in opera-
tion, although experience has shown that fluorine breakthrough will prob-
ably not occur until 1 mole of fluorine per mole of uranium has been added
(about 4 hr at 100 liters/min). The fluorine disposal system reacts SO,
preheated to 400°F, with preheated fluorine in a preheated reactor to form
inert SO0,Fs. When fluorine breakthrough occurs, as evidenced by an in-
crease in the load on the preheater, the fluorine stream can be diluted
with 50% helium, since utilization will be quite low by this time. Cor-
rosion is expected to be high., In the Volatility Pilot Plant fluorinator,
corrosion averaged about 1/2 mil/hr in a nickel vessel. Corrosion tests
at Battelle Memorial Institute indicate the corrosion of INOR-8 to be
somewhat less.

The off-gas from the fuel storage tank passes through the heated
NaF bed, where considerable decontamination occurs both of volatilized
fission and corrosion products., The UFg is then absorbed by low-temper-
ature sodium fluoride in a series of absorbers in the cubicle in the high-
bay area. With salt "C" it will be necessary to stop the fluorination
twice to replace the absorbers. The loading of the absorbers is followed
by observing the temperature of the bed. The bed is cooled by air flow
around the absorbers, and the temperature should be kept below 300°F, At
this temperature the UFg has a vapor pressure of 0,1 mm over the UFge2NaF
complex, and the uranium loss would be 190 g, or slightly less than 0.1%.
The loaded absorbers will be sent to the Volatility Pilot Plant for de-
sorption and cold trapping of the UFg. The radiation level at the surface
of the unshielded absorbers should be less than 100 mr/hr and should pose
no handling problem.

The SOpF,; from the fluorine disposal system passes through the caus-
tic scrubber. The excess S0, and the S02F, will be partially neutralized
by the caustic, and some fission product removal will be achieved. As
mentioned before, the off-gas then passes through the charcoal trap be-
fore passing through the absolute filter, the containment filters, and
the stack.

The heated line with a freeze valve is provided from the processing
tank to outside the cell for removal of waste salt. This may not be
204

needed until operation of the MSRE has been terminated. At the present

time the exact method of removal has not been determined; the waste salt

will either be transported in cans and buried, or will be stored in an i
underground tank at the reactor site. |

Equipment Description

In general, the fuel processing system is designed for direct main-
tenance., If maintenance is required after irradiated fuel salt has been
processed, aqueous decontamination will be required. Exceptions to this
are: (1) the sodium fluoride trap has been designed for remote replace-
ment since plugging of this unit during fluorination is possible; (2) the
two air-operated valves in the cell can be remotely replaced; and (3) all
heaters on lines and equipment in the cell have duplicate spares installed.
The roof plugs have been designed to permit use of the remote maintenance
shield designed for the reactor, and the trap and valves are located under -
these removable roof plugs.

o

The fuel storage tank is an INOR-8 vessel 50 in, in diameter and -
116 in. high. The height of the tank has been increased 30 in. over the
height of the drain and flush tanks to minimize salt carry-over due to
sparging. The tank is heated by four sets of tubular heaters which can
be controlled independently. A l-in., air space is provided around the
tank through which the cell wventilation air can be diverted if necessary
for cooling of an irradiated fuel salt batch. ©Salt inventory in the tank
is determined by weigh cells similar to those in the drain tanks.

The sodium fluoride trap is a 20-in.-diam Inconel vessel, 14 in.
high, designed to operate at 750°F, Tubular heaters are provided both
on the outside and in a 4-in.-diam center pipe. The gas enters at the
outside of a cylindrical baffle and flows down through the bed and up
the inside of the baffle. The bed is loaded with 70 kg of 1/8-in. right-
circular cylinders of NaF., This vessel is installed with ring-joint
flanges which can be disconmnected remotely, and it has remote dlsconnects. g
for the electrical leads and thermocouples.

The sodium fluoride absorbers are constructed of carbon steel which "
is sufficiently resistant to fluorine at the temperatures involwved and
for the short duration of operation. These absorbers will be discarded
after one use. ~The absorbers have a l4-in, OD with an internal cylin-
drical baffle similar to the NaF trap. They can be loaded to a bed depth
of 6 to 10 in. IKach absorber is mounted in an open-top container with
an air distributor pipe at the bottom, The cooling air can flow up the
outside of the absorbers and through a 2-in. center pipe. A thermowell
is provided near the gas inlet for detecting UFg breakthrough from the
preceding absorber and for regulating cooling air,.

The caustic scrubber is a 42-in,-diam, 7-ft-high Inconel tank with
a gas distributor. The tank has cooling-water coils on the outside to
dissipate the heat from HF neutralization. A steam jet transfers the
solution to the liquid waste tank. '
- 205

: r

The fluorine disposal system uses similar electrical prehesaters for
the S0; and the fluorine streams. The SO, heater is stainless steel; the
fluorine heater is Monel. The heated streams are combined in the fluorine
reactor, which is a 5-in, by 8-ft Monel pipe with a steam coil around the
inlet half. ©GSome heat is required to initiate the reaction, but once
started, the reaction is exothermic, and the steam then functions as a
coolant to keep the temperature around 400°F,

The salt sampler will be used mainly to determine if salt is satis-
factory for return to the reactor after processing. After Hp-HF treat-
ment, the iron and HF content must be low. After fluorination, the sample
should verify tThat all the uranium has been removed. The sampler is sim-
ilar to the sampler-enricher for the reactor and is described in the

. chapter on "Component Development.” It is actually the original mockup
of the sampler-enricher with slight modifications and can be used for the
e chemical plant because the contaimment and pressure reguirements are con-
y siderably less severe.

PLANS FOR OPERATION OF THE MSRE

P. N. Haubenreich

Introduction
. Ob jectives
. | The purpose of the Molten-Salt Reactor Experiment, stated broadly,
- is to demonstrate that many of the desirable features of molten-salt re-
actors can presently be embodied in a practical reactor which can be op-
. erated safely and reliably and can be serviced without undue difficulty.

The program which has been laid out for the MSRE is intended to provide
that demonstration in a safe, efficient, and conclusive manner.

Although the complete success of the MSRE depends in part cn the re-
actor being able to operate for long periods at full power, the test pro-
gram recognizes that the success of a reactor experiment is not measured
solely in megawatt-days. The tests and the experiments are designed to
be penetrating and thorough, so that, when the experiment is concluded,
not only will reliable operation and reasonable maintenance have been
demonstrated, but there will be as many conclusive answers as possible
to the important questions pertaining to the practicability of molten-
salt reactors of this general type.
206

Areas to be Investigated

Chemistry and Materials. Some of the most important questions to
be answered by the MSRE have to do with the behavior and interactions
of the fuel salt, the graphite, and the container material in the reactor
environment. The major points to be investigated in this area are:

. fuel stability,

. Ppenetration of the graphite by the fuel salt,
. graphite damage,

xenon retention and removal,

. corrosion, and

o v W

. behavior of corrosion procducts and nonvolatile fission products.

During operation, the principal means of investigation will be reg-
ular sampling and chemical analysis of the fuel salt, analysis of the
long-term reactivity behavior, and determination of the isotopic com-
position of the xenon in the off-gas. Periodically, during shutdowns,
specimens of graphite and of INOR will be removed from the core for ex-
amination.

Engineering. The MSRE incorporates some novel features and compo-
nents which have been developed and designed specifically for molten-
salt reactors. The test program will obtain performance data on these
items, permitting evaluation of ideas and principles which could be em-
ployed in future reactors.

The broad heading of engineering also covers the extensive startup
program which must be devoted to the checkout, calibration, and prelimi-
nary testing of the many more or less conventional devices and systems
in the MSRE.

Reactor Physics. From the standpoint. of reactor physics, the MSRE
core is unique, but its nuclear design posed no serious problems. One
reason for this apparent paradox is the simplicity of the core, which
makes simple spatial approximations valid. Another reason is that the
demands for accuracy in the predictions of the design are not severe.
This situation arises because the fuel is fluid, permitting easy adjust-
ment of the uranium loading, and eliminating hot spot problems associated
with heat transfer from fuel to core coolant. For these reasons, the de-
sign did not employ extremely sophisticated calculational procedures, and
there were no preliminary critical experiments. Instead, the physics part
of the reactor test program is relied on to provide such accurate infor-
mation on nuclear characteristics as may be required.

lThe program of reactor physics measurements begins with the experi-
mental loading of uranium to attain criticality. Following this, experi-
ments will be conducted to determine whether the system is stable and
207

safe. Accurate measurements of rod worth and of reactivity coefficients
will be made to facilitate later analysis of the reactivity behavior dur-
ing power operation. This analysis will be concerned with, among other
things, the transient behavior of 12°Xe. The reactivity behavior will
also be scrutinized for anomalies which might indicate changes in condi-
tions within the core.

Phases of Program

The testing and experimental operation of the MSRE are subdivided
into phases which must follow in a natural sequence. They are:
. DPrecritical testing,
. 1nitial critical measurements,

low-povwer measurements, and

N W oo

. reactor capability investigations.

The precritical testing phase begins with part of the training of
new operators; this consists of their checking the location of equipment
and comparing the installed piping against the flowsheets. As systems
are completed, leak testing, purging, filling, calibrating, and test op-
eration are started. The precritical testing culminates in shakedown
operation of the entire reactor system, with flush salt in the fuel sys-
tem.

In the initial critical experiments, fuel salt will be loaded and
enriched uranium will be added in increments to bring the concentration
up to that required for operation, During this phase, the control rods
will be calibrated and fuel concentration, temperature, and pressure co-
efficients of reactivity will be measured. Baseline data on the fuel
chemistry and corrosion will also be obtained during this period.

After the critical experiments have been conducted at a few watts
of nuclear power, the power will be raised to permit certain tests. These
will include tests of the nuclear power servo-control system, tests of
the automatic load control system, calibration of power indicators, and
surveys of the biological shielding. The nuclear power will be less than
2 Mw during this period.

Capability investigations consist of two parts: the first, a step-
wise approach to full power; second, extended operation. During the ap-
proach to full power, temperatures, radiation levels, and the nuclear
pover noise will be observed to determine whether any unforeseen condi-
tion exists which would restrict the attainable power level. Extended
operation will test the reliability of equipment and long-term corrosion
and fission product behavior, Malntenance will be carried out as re-
gquired, and the reactor will be shut down periodically for removal of
samples from the core. Long-range plans include chemical processing of
the fuel salt and operation with different fuel salt compositions.
208

Staff

Organization

When the MSRE begins operation, about 43 persons will be employed
at the reactor site. This staff is organized, as shown in Fig. 1, to
perform the three primary functions: operations, analysis, and main-
tenance., |

UNCLASSIFIED
ORNL—DWG 64-9884

SUPERINTENDENT

| | 1

COORCINATOR HEALTH PHYSICIST INDUSTRIAL HYGIENIST

| L

ANALYSIS GROUP OPERATIONS GROUP

ENGINEERING AND
© MAINTENANCE GROUP

Fige 1. Organization of MSRE Staff.

Operations. A senior engineer, called the Operations Chief, is in
charge of the Operations Group and is responsible for the training, quali-
fication, and performance of the men who actually handle the controls of
the reactor. Another engineer assists him in these duties and also serves
as relief Shift Supervisor. '

There are four crews that operate the reactor on a rotating shift
basis. Fach crew consists of two engineers and three technicians. One
engineer is the Shift Supervisor, who is responsible for all operations
in the reactor area while he is on duty. The other Shift Engineer may
operate any part of the reactor system or may direct the technicians in
operations. A senior technician on each crew is designated Control Room
Supervisor. He is the person who ordinarily operates the reactivity and
load controls and maintains the control room log. The other two techni-
cians attend to the operation of the auxiliary systems and to the manual
collection of data.

Two additional technicians are assigned to the day shift, to take
samples and to assist in special operations. They may also serve as
relief for shift technicians who are sick or on vacatlon.

The organization of the Operations Group is shown in Fig. 2. All
the personnel shown are members of the MSRE Operations Department of the
ORNL: Reactor Division.
209

UNCLASSIFIED
ORNL-DWG 64-—9885

OPERATIONS CHIEF I

|

ASSISTANT CHIEF
TECHNICIAN
TECHNICIAN
CREW A CREW B CREW C CREW D
SHIFT SUPERVISOR SHIFT SUPERVISOR SHIFT SUPERVISOR SHIFT SUPERVISOR
SHIFT ENGINEER SHIFT ENGINEER SHIFT ENGINEER SHIFT ENGINEER
CR SUPERVISOR CR SUPERVISOR CR SUPERVISOR CR SUPERVISOR
TECHNICIAN TECHNICIAN TECHNICIAN TECHNICIAN
TECHNICIAN TECHNICIAN TECHNICIAN TECHNICIAN

Fig. 2. Organization of Operations Group.

During the first several months of operation, while the reactor equip-
ment and systems are being started and checked, each crew will have crafts-
men attached to it to make any adjustments or minor repairs which may be
required. A pipefitter, an electrician, and an instrument mechanic may

be required on each shift at first.

Analysis. The group which 1s assigned to the analysis of the MSRE op-
eration is represented in Fig. 3. The personnel assigned to nuclear and
mechanical analysis are from the MSRE Operations Department of the Reactor
Division. This group will also include loanees from Euratom (not shown).
The other personnel are from the Reactor Chemistry, Metals and Ceramics,
Chemical Technology, and Instrumentation and Controls Divisions.

UNCLASSIFIED
ORNL-DWG 64-9886

ANALYSIS
NUCLEAR AND MECHANICAL CHEMISTRY CHEMICAL PROCESS
GROUP LEADER CHEMIST
ENGINEER
ENGINEER TECHNICIAN
ENGINEER
TECHNICIAN
MATERIALS COMPUTER
METALLURGIST PROGRAMMER

Fig. 3. Organization of Analysis Group.
210

Analysis personnel plan the experiments required in each field,
prepare detailed procedures, and monitor and participate in the execution
of the experiments. They then digest and evaluate the information and
report the findings. An experienced computer programmer who participated
in the initial programming of the on-line computer works with the other
analysis personnel in collecting and reducing data.

Maintenance. The maintenance function includes semiremote mainte-
nance of equipment in high radiation fields, normal maintenance of other
equipment, design work associated with modifications, and procurement of
parts and supplies. The personnel permanently assigned to this function
are indicated in Fig. 4. The Instrumentation and Controls engineer and
the Plant and Equipment engineer are from those respective ORNL divisions;
the others are from the Reactor Division. Pipefitters, electricians, mill-
wrights, riggers, and instrument mechanics will be assigned from the ORNL
service organizations as required. The number of such craftsmen at the
reactor site will fluctuate, depending on the status of the reactor.

When remote maintenance work is undertaken, specialists in this field will
be called in from the Development Department of the Reactor Division.

UNCLASSIFIED
ORNL-DWG €4-9887

ENGINEERING AND
MAINTENANCE

MAINTENANCE COORDINATION DESIGN LIAISON
*DUAL CAPACITY

ENGINEER * ENGINEER

INSTRUMENTS AND CONTROLS PLANMNING AND PROCUREMENT PLANT AND EQUIPMENT
ENGINEER *

ENGINEER ENGINEER

TECHNICIAN 6
TECHNICIAN

Fig. 4. Organization of Engineering and Maintenance Group.

Other Staff Members. Because the MSRE is an experiment, the operat-
ing conditions may be changed — within limits — and new phases may be
added to the program, when analysis shows such changes and additions to
be desirable, This flexibility requires that the various experiments be
combined in an efficient manner and that the operations group be supplied
with up-to-date instructions. This is the function of the coordinator
(who also spends part of his time as a member of the nuclear and mechanical
analysis group). ' '

211

The operation and maintenance of the MSRE will be carefully monitored
to avoid overexposure of personnel to direct radiation, radiocactive con-
tamination, or ingestion of beryllium. This will involve area monitoring,
area surveys, and personnel monitoring. The Health Physics Division will
monitor the radiation exposure and contamination; the Industrial Hygiene
Department will monitor for beryllium. Representatives of these groups
will work closely with the MSRE staff as indicated on the organlzatlon
chart, Fig. 1.

Qualification and Training

The collective background of the MSRE staff includes many years of
experience in reactor development, design, and operation. 8Six of the
senior personnel participated in the operation of the Homogeneous Reactor
Experiment No. 2, with duties similar to their present assignments, and
six are graduates of ORSORT.

All the senior technical personnel were involved in the design and
development of the MSRE for two to four years before joining the Operations
Department. Their experience and special knowledge of the MSRE systems
were put to use in the preparation of the operating procedures and other
parts of the Design and Operations Report, in the writing of test proce-
dures, and in the operator training program.

All the operating group, both engineers and technicians, were given
an intensive two-week training program covering the theory, design, and
layout of the entire reactor system. On-the-job training then began with
detailed checking of the installation against flowsheets and drawings.
This Jjob thoroughly familiarized the operators with the physical layout
of the MSRE. Visits to the ORR and to the molten-salt Engineering Test
Loop were arranged in order to enable the operators to observe and par-
ticipate in reactor operation and molten-salt sampling and transferring.
One important purpose of the flush-salt operation is to give the operators
experience in every phase of the operation except the nuclear and heat-
removal aspects, which will be simulated insofar as possible.

At the conclusion of the precritical testing, there will be another
period of intensive instruction, with special emphasis on nuclear experi-
ments and operation. The operators will then be examined to ensure that
they are thoroughly prepared and qualified to operate the MSRE safely and
efficiently.

Use of Computer

An unusual feature of the MSRE operation will ‘be the use of a high-
speed digital computer, installed in the reactor building and dlrectly
connected to the reactor by sensing elements.

The Bunker-Ramo model 340 computer which will be installed uses 28-
bit words, with 12,288 words in a magnetic core memory and 28,672 words
212

in a magnetic drum memory. It is designed for computer control applica-
tions and completes typical programs at a rate of over 70,000 operations/
sec,

In the MSRE application, the computer will not exercise any direct
control, and the reactor can be operated with the computer out of service,
The functions of the computer include collecting and displaying data and
performing on-line calculations for immediate and long-range purposes.

Computer Functions

A total of 273 signals from the reactor system are connected to the
computer, Of these, 177 are temperatures; the remainder include pressures,
flow rates, liquid levels, weights, fluxes, radiation levels, and control
rod and radiator door positions. All of these are normally scanned once.
every 5 sec. Flux signals are repeated in the scan sequence so that the
nuclear power is read once every 0.2 sec. A total of about 200 alarm
points are entered in the computer, and any signal going beyond these
limits produces an audible alarm and a printout on an alarm typewriter
in the reactor control room. If any of 24 important variables go out of
limits, the scanning is switched to a list of 64 selected variables, which
is recorded on a magnetic tape once every 0.25 sec. Every 10 min the en-
tire list of readings is stored on magnetic tape. The purpose of this
storage is the retention of data for future analysis. All data on the
tape are identified by point and time, so that they can be retrieved for
- use in tabulations, plots, or calculations.

The status of the reactor system and an up-to-date record of its op-
erating history are typed out routinely by the computer on a preprinted
sheet which is changed once a day. Fifty-four signals are typed once an
hour, 186 signals are entered at 8-hr intervals, and 21 items, including
calculated quantities, are entered on a daily report section of the sheet.
Also on the preprinted sheet are the results of routine heat balances,
taken at 4-hr intervals, and salt inventories, taken once each 8-hr shift.
A reactivity balance which will be described later, is typed out cnce an
hour.

Most of the scheduled calculations can be initiated at any time by

entering a simple demand signal through the computer console. Results
of these calculations are typed out on another typewriter.

The Reactivity Balance

The reactivity balance is an important part of the analysis of the
core behavior. Briefly, it consists of calculating the cumulative changes
in reactivity produced by changes in fuel-uranium concentration, concen-
trations of 1?°Xe ‘and other fission products in the core, core tempera—
tures, and control rod positions. At the beginning of operation, when
pover levels are moderate, it will be assumed that there are no unforeseen
mechanisms at work in the core, and the balance will be used to measure
the reactivity which must be ascribed to xenon to make the net equal zero,
213

Analysis of the transient behavior of this apparent xenon effect will be
used to help determine values for parameters affecting the xenon retention
and removal. These quantities will then be put into a program which pre-
dicts xenon poisoning from the power history. After an adequate descrip-
tion of the 125Xe behavior becomes available, the reactivity balance will
be used to detect anomalous developments. The reactivity balance is per-
formed routinely every 5 min and is typed out once an hour. It may also
be computed and printed out on demand.

Schedule

Completion Dates

The assembly of the MSRE staff at the reactor site was completed at
the first of July. (Most of the engineers had already been there for one
to two months, working on the operating procedures and preparing training

lectures.) The first two weeks in July were devoted to the initial train-
ing program. Following this, the operating crews began detailed checking
of the installed piping and equipment, while continuing to receive class-
room instruction. Visits to the ORR and the Engineering Test L.oop were
also made at this time.

The program of testing and calibration will be completed in the fall
of 1964, and operation of the entire system (with flush salt) will begin.
After the system has been shown to operate satisfactorily, it will be
shut down, and final preparations will be made for nuclear operation.

Fuel carrier salt will be charged into the reactor, and uranium will be
added in fthe critical experiments. These experiments should begin in early
spring, 1965, and full power be attained during the summer of 1965.

Uranium Content

The uranium used in the initial operation will contain only about
40% #3%U. This is to be used instead of highly enriched material so that
the total uranium concentration will be high enough to provide a large
tolerance for UF;. (If fluorine were lost from the fuel, UF; would be
formed, and if the UF,/UF, ratio got high enough, uranium might precip-
itate.) Experimental operation with the original fuel composition may
last for about a year, If fluorine loss proves to be low, as expected,
it is proposed to then strip the uranium from the salt (by fluorination
in the storage tank) and add highly enriched uranium. Operation with the
lower total uranium concentration would last perhaps another year. A
year's operation with fuel of a third composition — one containing thorium
— is contemplated. This would provide more experience with salt similar
to that which might be used in a breeder reactor. We have not made plans
for operation of the reactor beyond this point. Favorable experience and
acceptance of molten-salt reactors as promising power breeders would prob-
ably lead to modification of parts of the plant and continuance of experi-
ments important to the development of breeder reactors.
214

CHEMICAL BASIS FOR MOLTEN-SALT REACTORS

W. R. Grimes

Introduction

Nuclear reactors with liquid fuel and blanket systems have, in

- principle, advantages and disadvantages in comparison with more conven-
tional reactor types. Ease in reactor control, ease of preparation of
fuel and blanket, and the possibility of continuous side-stream process-
ing.of fuel and blanket are probably the most important advantages. Of
the disadvantages, the increase in inventory imposed by the external
circuitry and the necessity to achieve and to maintain the integrity of
the complex and radioactive plumbing are the most obvious.

Molten-salt mixtures are versatile; they can serve as fuels, blan-
kets, and coolants. With them a family of reactors, operating at low
pressure and at usefully high temperature, can be provided. Whether such
reactors can compete in the power market with other types depends upon
the extent to which designers can optimize the advantages and minimize
the disadvantages of each type. The intrinsic merit of the materials
that must be used will, finally, control the extent to which such optimi-
zation can be accomplished. For liquid-fueled reactors such control will
almost certainly lie in the nuclear properties and the chemical behavior
of the materials. The thermodynamic behavior of molten-salt systems for
use 1in nuclear reactors, with special emphasis on those for use in MSRE,
is surveyed briefly in the following.

General Requirements for the Fluids?t

Molten-salt reactors cannot be operated at all unless the minimum —
though stringent — demands that the reactor makes upon its fluid fuel can
be met. These minimum demands include the following. The fuel must dis-
solve more than the critical concentration of fissionable material at
temperatures safely below the temperature of exit from the heat exchanger.
The fuel must consist of elements of low (and preferably very low) capture
cross section for neutrons of the energy spectrum typical of the chosen
design. The mixture must be thermally stable, and its vapor pressure
must be low over the temperature range proposed for operation. The fuel
mixture must possess heat transfer and hydrodynamic properties adequate
for its service as a heat-exchange fluid. It must be relatively non-
aggressive toward some (otherwise suitable) material of construction,
presumably a metal, and toward some suitable moderator material. The
fuel must be stable toward reactor radiation, must be able to survive
fission of the uranium (or other fissionable material), and must tolerate
fission product accumulation without serious deterioration of its useful
properties.
215

If molten-salt reactors are to be competitive, we must add to this
impressive list the requirement of genuinely low fuel cycle cost; this
clearly presupposes a cheap fuel and an effective turnaround and reuse
of the unburned fissionable material, or an effective and economical
decontamination and reprocessing scheme.

Blankets and coolants face requirements which differ in obvious
ways. The radiation intensity in the blanket will be considerably less —
in the coolant markedly less — than that in the fuel. Efficiency of the
blanket as a heat transfer fluid may be relatively unimportant; but a
high concentration of fertile isotope is essential, and an effective re-
covery of bred material is likely to be vital.

Choice of Salt Composition

General Considerations

The compounds which are desirable constituents of fuels for thermal
breeders are those that can be prepared from beryllium, bismuth, boron-11,
carbon, deuterium, fluorine, lithium-7, nitrogen-l15, oxygen, and the
fissionable and fertile isotopes. For an epithermal reactor this list
can probably be augmented somewhat, and a thermal reactor intended to
burn 235U and to obtain a modest conversion of 32Th to 233U might use
(as diluents) the compounds prepared from the elements listed in Table 1.

Of the known compounds containing useful concentrations of hydrogen
(or deuterium), only the hydroxides of the alkali metals, the saline
hydrides of lithium and calcium, and certain interstitial hydrides (e.g.,
zirconium hydride) show adequate thermal stability in the 1000 to 1300°F
temperature range. [Acid fluorides (e.g., KHF, ) might be permissible in
low concentrations at lower temperatures.] The hydrides are very strong
reducing agents and are most unlikely to be useful components of any
uraniferous liquid fuel system. Alkali hydroxides dissolve extremely
small quantities of uranium compounds at useful reactor temperatures and
are very corrosive at such temperatures to virtually all useful metals.
Hydrogen-rich compounds, which might provide self-moderation to molten
fuels, do not seem capable of use as fuels or coolants.

The nonmetals carbon, nitrogen, silicon, sulfur, phosphorus, and
oxygen each form only high melting and generally unsuitable binary com-
pounds with the metals of Table 1. From these nonmetals, however, a
wide variety of oxygenated anions are available. Nitrates, nitrites,
sulfates, and sulfites can generally be dismissed as lacking adequate
thermal stability, and silicates as having undesirably high viscosities.
Phosphates, borates, and carbonates are not so easy to dismiss without
study, and phosphates have, in fact, received some attention. The
several problems of thermal stability, corrosion, solubility of uranium
and thorium compounds, and, especially, radiation stability would seem
to make their use doubtful.
216 o .

Table 1. Elements or Isotopes Which May Be
Tolerable in High-Temperature

Reactor Tuels

Material

Absorption Cross Section

(barns)
Nitrogen-15 0.000024
Oxygen 0.0002 .
Deuterium 0.00057 .
Carbon 0.0033
Fluorine 0.009
Beryllium 0.010
Bismuth 0.032
Lithium-7 0.033
Boron-11 0.05
Magnesium 0.063
Silicon 0.13
Lead 0.17
Zirconium 0.18
Phosphorus 0.21
Aluminum 0.23
Hydrogen 0.33
Calcium 0.43
Sulfur 0.49
Sodium 0.53
Chlorine-37 0.56
Tin 0.6 .
Cerium 0.7
Rubidium 0.7 .
Zine 1.06 -
Niobium 1.1
Strontium 1.16 -
Barium 1.17
Yttrium 1.27
Nitrogen 1.88
Potassium 1.97

217

If the oxygenated anions are eliminated, only fluorides and chlo-
rides remain. Chlorides offer mixtures that are, in general, lower
melting than fluorides; in addition, UCls3 is probably more stable than
UF3 with respect to the analogous tetravalent compounds. For thermal
reactors, fluorides appear much more suitable for reasons which include
(1) usefulness of the element without isotope separation, (2) better
neutron economy, (3) higher chemical stability, (4) lower vapor pressure,
and (5) higher specific heat per unit weight or volume.

Fluoride mixtures are, accordingly,. preferred as fuel, blanket, and
coolant mixtures for thermal reactors. The fluoride ion is insufficient
to moderate the neutrons in cores of reasonable size; additional moderator
is, accordingly, required. The fluoride ion, however, moderates too well
to permit its use in genuinely fast reactors. 1In such reactors chloride
mixtures, using 35Cl, will certainly be preferred.

Choice of Active Material

Uranium Floride. Uranium hexafluoride is a highly volatile compound
clearly unsuited as a component of high-temperature liquids. The com-
pound UOzF,, though relatively nonvolatile, is a strong oxidant which
should prove very difficult to contain. Fluorides of pentavalent uranium
(UF5, UyFy, etc.) are not thermally stable and should prove prohibitively
corrosive 1f they could be stabilized in solution.

Uranium tetrafluoride is a relatively stable, nonvolatile, and non-
hygroscopic material, which is readily prepared in high purity. It melts
at 1036°C, but this very high freezing point is markedly depressed by
several useful diluent fluorides. Pure, crystalline uranium trifluoride
is stable, under inert atmospheres, to temperatures above 1000°C, but it
disproportionates at higher temperatures by the reaction

4UF3 = 3UF, + U° . (1)

Uranium trifluoride is appreciably less stable in molten fluoride solu-
tions than in the crystalline state. It is tolerable in reactor fuels
only insofar as the equilibrium activity of U° which results is suffi-
ciently low to avoid reaction with the moderator graphite or alloying
with the container metal.? Appreciable concentrations of UFj (see below)
are tolerable in LiF-BeF», mixtures such as those proposed for MSRE. In
general, however, uranium tetrafluoride must be the major uraniferous
-compound in the fuel.

Thorium Fluoride. All the normal compounds of thorium are quadri-
valent; ThF, (mglting at 1115°C) must be used in any thorium-bearing
fluoride melt. ~

Choice of Fluoride Diluents

The fuel system for thermal reactors of the types given most atten-
tion to date will require low concentrations (0.2 to 2 mole %) of uranium.
218 o .

The properties (especially the melting temperature) of such fuels will be

controlled by the composition of the diluent mixture. Blanket mixtures -
(and perhaps fuel systems for one-region converters or breeders) will

require considerable concentrations of high-melting ThF,. These fluids

must, if they are to be compatible with the temperature requirements of

large steam turbines and are to retain a reasonable margin of safety be-

tween the minimum reactor operating temperature and the freezing point

of the fuel, be completely molten at 975°F (525°C). It is clear that the

diluent fluoride system must be low melting and must be capable of large

depressions of the freezing points of the active fluorides.

Simple consideration of the nuclear properties leads one to prefer
as diluents the fluorides of Be, Bi, ‘Li, Mg, Pb, Zr, Ca, Na, Sn, and Zn
(in that order). Equally simple considerations (Table 2) of the stability
‘of diluent fluorides toward reduction by common structural metals, how=-
ever, serve to eliminate BiF,, PbF,, SnF,, and ZnF, from consideration.

No single fluoride can serve as a useful diluent for the actiVe
fluoride. The compound BeF, is the only stable one listed whose melting
point is close to the required level; this compound is too viscous for
use in the pure state.

The very stable and high melting fluorides of the alkaline earths
and of yttrium and cerium do not seem to be useful major constituents of
low-melting mixtures. Mixtures containing about 10 mole % of alkaline
earth fluoride with BeFp melt below 500°C, but the viscosity of such
melts is almost certainly too high for serious consideration.

Some of the possible combinations of alkali fluorides have suitable
freezing points. Equimolar mixtures of LiF and KF melt at 490°C, and
mixtures with 40 mole % LiF and 60 mole % RbF melt at 470°C. The ternary
systems LiF-NaF-KF (see Fig. 1) and LiF-NaF-RbF have lower melting regions
than do these binaries. All these systems will dissolve UF, at concen-
trations up to several mole % at temperatures below 525°C. They might
well prove useful as reactor fuels if no mixtures with more attractive
properties were available.

Mixtures with useful melting points over relatively wide ranges of
composition are. available if ZrF, is a component of the system. Phase -
relationships in the NaF-ZrF, system (Flg 2) show low melting points over
the interval 40 to 55 mole % ZrF,. A mixture of UF, with NaF and ZrF,
served as fuel for the Aircraft Reactor Experiment. The phase diagram3
of this ternary system is shown as Fig. 3. '

The compounds ZrF, and UF, have very similar unit cell parameters
and are isomorphous. They form a continuous series of solid solutions
with a minimum melting point of 765°C for the solution containing 23
mole % UF,. This minimum is responsible for a broad, shallow trough which
penetrates the ternary diagram to sbout the 45 mole % NaF composition.
A continuous series of solid solutions without a maximum or a minimum
exists between G<3NaF+UF, and 3NaF+ZrF,; in this solution series the tem-
perature drops sharply with decreasing ZrF, concentration. A continuous .
219

Table 2. Properties of Fluorides for Use in
' High-Temperature Reactors

I'ree Energy . Absorption Cross
Compound of Formation M;i;iig Section® for
at 1000°K (°c) Thermal Neutrons
(kcal/F atom) (barns)
Structural metal
fluorides
CrF,- =74 1100 3.1
FeF, —566.,5 930 . 245
NiF»o ' —-58 1330 4.6
Diluent fluorides
CaFs | —-125 1330 0.43
LiF —125 870 0.033P
Bal's —124 | 1280 1.17
Sri's =123 1400 1.16
CeF, -118 1324 0.7
YFq —-113 1144 1.27
MgF'» —-113 . .1270 0.063
RbF —-112 790 0.70
~ NaF _ =112 1000 0.53
KF —-109 880 1.97
BeF's -104 545 0.010
ZrF, —94 912 0.180
AlF4 -90 1040 0.23
ZnF's =71 872 1.06
SnFo —62 213 0.6 .
PoFo : —62 850 0.17
BiFs -50 727 0.032
Active fluorides
ThF, —-101 . 1115
UF, - -95.3 1035
CUF3 —-100.4 1495

80f metallic ion.

Boross section for 7Li.
TEMPERATURE IN °C
COMPOCSITION IN mole %

220

UNCLASSIFIED
ORNL- LR~DWG 1199A

NaF
990

900
850
800
£ " N\
750
710 652
3 700
650 A
600 A
550 o
3 IS
' 3
A
o
% 8 §
KF \/\ \/ I i/ LiF
856 4547 492 845
The System LiF-NaF-KF,

Fig. 1.

-....;—«-—”“...—._‘\
221

UNCLASSIFIED
ORNL-LR-DWG-22105

1000

900 AN : —

=

800 \ ‘
\ \
|
l
S |
e 700 J /
o | \
o
>
}_
<
o
i N\
= i
l'._ 2 /
N \
500 i :”?
] \ I’
|l uY \ R IS Y
N N Vs 3
Il N Y |~ Ve <
400 W 1 (i i (ol N
O (= [=] ~N o [=]
zZ |l =z z o~ 2 z
| L
' o
' =
300
NaF 10 20 30 40 50 60 70 80 90 ZrF,
ZrF, {mole %)

Fig. 2. The System NaF-ZrF,.

solid~solution series without a maximum or a minimum also exists between
the isomorphous congruent compounds 7NaF.6UF, and 7NaF.6ZrF,; the liquidus
decreases with increasing ZrF, content. The ligquidus surface over the
area below 8 mole % UF, and between 60 and 40 mole % NaF is relatively
flat. All fuel compositions within this region have acceptable melting
points. Minor advantages in physical and thermal properties accrue from
choosing mixtures with minimum ZrF, content in this composition range.

The lowest-melting binary systems of the usable diluent fluorides
are those containing BeF, with NaF (Fig. 4) or LiF. (The ternary system
LiF-NalF-Bel'; has been examined in some detail, but it seems to have no
important advantage over either binary.) Since beryllium offers the best
cross section of the diluents (and i ranks very high), fuels based on
the LiF-Bel; diluent system have been chosen for MSRE.

The binary system LiF-BeF; has melting points below 500°C over the
concentration range from 33 to 80 mole % BeF,. The presently accepted
LiF-Bel'; system diagram, presented in Fig. 5, is characterized by a
single eutectic* between BeF, and 2LiF-BeF, that freezes at 356°C and
222

UNCLAPSIFIED
ORNL—LR-DWG 19886R
UF,
1035

ALL TEMPERATURES ARE IN°C

3NaF-UF,

4 #—

3 NaF-2 ZrF, 3NaF-4Zrf,

I‘BNaF-ZrF4 ’
5NaF.-2 ZrF,

4 2NaF-Zrf;

TNaF-6 ZrF4 NaF-ZrF,

Fig. 3. The System NaF-ZrF,-UF,.
223

UNCLASSIFIED
ORNL-LR-DWG 16425R
900 4

800 \

NaF + LIQUID

o® = ORNL DATA
700 HQ = HIGH QUARTZ
LQ = LOW QUARTZ

a-2NaF-BeF,+LIQUID -

o
<~ 600 fl
w AN
% L J L J .
E ’—_Ic
B-J /
- S 500 -~
w
[

\ /

a - 2NaF - BeF, +NaF B~ NaF-BeF, BeF,(HQ) + LIQUID

= : \ (+ LIQUID _ |

- 400 BeF,{HQ)} + '~ NaF-BeF,

\.///“k\_/ P 2
\

Id

a-2NoF-BeF,+ 8 -NaF -BeF;

2"
. : { PR \B—NQF-BEF2+LIQUID
300 f——— 4 _onaF- - NaF -BeF, | \
a’'~2NaF -Bef, +NaF + | BeF,(HQ) + B- NaF-BeF,
| [ a'-2NaF-BeF,
B - NaF-BeF, +y ~2NaF - Bef, | BeF,{LQ) + 3 - NaF-BeF.
y -2NaF - BeF,+ NoF | / I 1
200 .
NaF 10 20 30 40 50 60 70 80 90 BeF,
BeF, (mole %)
Fig. 4. The System NaF-BeF,.
UNCLASSIFIED
ORNL—LR—DWG 16426R
900
800 \\
. 700 N
”
i o
- 2 600 |[—— LiF + LIQUID
w \
14
=
[,
5 [
o
2 \ '
< 500
. N / BeFp + LIQUID
400 2LiF - wf?_\\ y
+
Y LIQUID \/
LiF + 2LiF - BeF,
Ty 2LiF-BeFp + BeFp (HIGH QUARTZ TYPE)
300 @ I [
5 ! |
3 i o
N ZL'F+BGF2 S LiF - BeFp + BeFp (HIGH QUARTZ TYPE)
LiF-BeF, & ' ' ]
200 ‘rrBete 3 LiF - BeF2+ BeFp (LOW QUARTZ TYPE)

s LiF 10 20 30 40 50 60 70 80 90 Befy _

. BeF, (mole %)

Pig. 5. The System LiF-BeF,.

&
224

contains 52 mole % BeFs. The compound 2LiF-BeF, melts incongruently to
LiF and liquid at 460°C; LiF:BeF, is formed by the reaction of solid BeF,
and solid 2LiF-BeF, below 274°C.

LiF-BeF, Systems with Active Fluorides

The phase diagram of the BeF,-UF, system (Fig. 6) shows a single
eutectic containing very little UF,;. That of the IiF-UF,; system (Fig. 7)
shows three compounds, none of which melts congruently and one of which
shows a low temperature limit of stability. The eutectic mixture of
4LiF«UF,; and 7LiF-6UF, occurs at 27 mole % UF, and melts at 490°C. The
ternary system’ LiF-BeF,-UF,, of primary importance in reactor fuels, is
shown as Fig. 8. The system shows two eutectics. These are at 1 mole %
UF, and 52 mole % BeF, and at & mole % UF, and 26 mole % BeF,; they melt
at 350 and 435°C respectively. Moreover, the system shows a very wide
range of compositions melting below 525°C.

The system BeF,-ThF, is very similar to the analogous UF, system.
The LiF-ThF, system (Flg 9) contains four compounds. The compound '
3LiF*ThF, melts congruently at 580°C and forms eutectics at 570°C and 22
mole % ThF, and 560°C and 29 mole % ThF, with LiF and with 7LiF-6ThF,
respectively. The compounds 7LiF:6ThF,, LiF.2ThF,, and LiF«4ThF, melt
incongruently at 595 and 890°C. The ternary system LiF-BeF,-ThF, (see
Fig. 10) shows only a single eutectic® with the composition 47.0 mole %
LiF and 1.5 mole % ThF, melting at 356°C. In spite of small differences
due to the phase fields of LiF-2ThF,, 3LiF:ThF,, and 4LiF.UF,, the sys-
tems represented by Figs. 8 and 10 are very similar.

Moreover, ThF, and UF, form a continuous series of solid solutions
with neither maximum nor minimum, and the LiF-ThF,-UF, system (see Fig.
11) shows no ternary compounds and a single eutectic which contains 1.5
mole % ThF, with 26.5 mole % UF, and freezes at 488°C. Most of the area

UNCLASSIFIED
ORNL-LR-DWG 285%8A

{100

1
_-®

1000 L1QuiD P
ll/

200 — "
_,u".
et

800 _—
‘18‘

o~ UF, + LIQuID
700 &

" TEMPERATURE {°C)

[
/
[ ]
{

[

600 H

500 >y erBeFs +LIQUID

‘ QH|GHBGF2+UF4

400 L1 |

BeF2 10 20 30 40 50 60 TO0 80 20 UF4
UF, (mole %)

Fige 6. The System Bel,-UF,.

[L )

¢
@

TEMPERATURE (°C)

225

UNCLASSIFIED
ORNL ~LR-DWG 17457

Rlele

/

{000 /

900 v

N i

600 \

-

-
500 V// w w
vl @ +
. [V .
4LIF- UF, = w
P~ -
400
LiF 10 20 30 40 50 60 70 80 20 UF,
UF, (mole %)

Fig. 7. The System LiF-UF,.

on the diagram is occupied by primary phase fields of the solid solutions
UF,-Th¥,, LiF.4UF,-LiF-+4Th¥,, and 7LiF-UF;-7LiF-ThF,. ILiquidus tempera-
tures decrease, generally, to the LiF-UF, edge of the diagram.

It is clear from examination of the diagrams shown that fuel systems
melting below 500°C are available over a wide range of compositions in the
LiF-BeF,-UF,; system. Since up to 28 mole % of ThF, can be melted at tem-
peratures below 1100°F (see Fig. 10), blanket systems with very large
ThF, concentrations can be obtained. Moreover, the very great similarity
in behavior of Th¥, and UF, permits fractional.replacement of ThF, by UF,
with little effect on freezing temperature over the composition range of
interest as fuel. Fuels for single-region thermal converters are, ac-
cordingly, available in the LiF-BeF,;-ThF,-UF, quaternary system. ’

Oxide-Fluoride Equilibria. When a molten mixture containing only
LiF, BeF,, and UF, is treated with appreciable quantities of a reactive
oxide (i.e., HyO0, 0032—, FeO), precipitation of UO, results. The U0, so
produced is stoichiometric, and if it is maintained in contact with the
melt for sufficient time, it forms transparent ruby crystals of U0z, q0-

226

UNCLASSIFIED
MOUND LAB. NO.
1035 ) 56-141-29 (REV)

ALL TEMPERATURES ARE IN °¢

E = EUTECTIC
P = PERITECTIC LiF - 4UF,
UFy| = PRIMARY PHASE FIELD
TLiF - 6UF,
E
P K
ALiF -UF, o\

. —— - — —— T———
LiF 2LiF-BeF,” P 400 J£__400 BeF,
B45 548

NG

Fig. 8. The System LiF-BeF,-UF,.

[

—~ ‘—-’w\
-

227

UNCLASSIFIED
ORNL-LR-DWG 26535A

/

1050 - el

—

TEMPERATURE (°C)

\,hjf

3LiF-Thi—] 7LiF -6ThE—] LiF-2ThF,—

LiF-4ThE

450 | | |
LiF 10 20 30 40 50 60 70 8C 90 Thip

ThE, {mole %)

Fig. 9. The System LiF-ThF,.

UNCLASSIFIED
ORNL-LR-DWG 374204

548

Thi, 1414
S
TEMPERATURE IN °C
COMPOSITION IN mole Yo
1050
LiF-4ThE,
7LiF 6ThE—m . o}
4 AN LiF-2ThF, 100
3LIF-ThE, ss F48 4
433 LiF-2ThF,
950
P 897
2LiF BefF,
900
LiF-6ThE,
P 762 850
GOO
P 597
£ 565
3LIF-ThE & N, 750
£ 568
?oo
650
6e %O ; 50
’?5‘ ’)OOSO (&) 6“5 0
e, O 00 350 £ 526
OO \ (@)
LiF VA BeF
AEQ 2
845 2LiF-BeF; 500?450 400 400 450 500
P 465 £ 370

Fig. 10. The System LiF-BeF,-ThF,.
228

’

ThF,
LT UNCLASSIFIED
ORNL-LR-DWG 28245AR2 .
PRIMARY-PHASE AREAS
TEMPERATURE IN °C (a)
COMPOSITION IN mole % ) URy=ThR (ss)
(b} LiF-4UF,~LiF-4ThF, (ss)
{¢) LiF- 2Th(U)F {ss)
LiF‘4ThF;‘ {d) 7LiF- 6UF TLlF 6ThF4 (SS)
{e) 3LiF- Th(U)F (ss)
(f) LiF
LiF-2ThF,
P 897 A
P
AN %b "
7LiF-6ThE, ) .
o)
(o) s
P762 .%\
‘ (9] v,
o)
°, (a) )
P 597 4 R
£ 565
3LiF-ThF,
£ 568
£ P609
o q‘,\o
(f) %5 A\
N
%%‘{‘,\ PSOO
xD
S \ %(\ W
£488
LiF ] \/ \ \ \/ U,
845 ALiF- UFy - P500” E490 P 610 PT75 LiF-4UF, 1035
TLiF-6UF,
Fig. 11l. The System LiF-ThF4-UF4.
UNCLASSIFIED
s ORNL-DWG 64-1999 "
m \ <«
I [N
= .
\ \\
T3 . -
he 2 500°C
K \ \.\
A N, 600°C
= 700°c\\\\\
= 2 ~ \\\
= \\\‘\_ - mw\
8 \\‘__\r‘—'—__\
————] [ —
=} ]
I
w |
1
0 0.2 0.4 0.6 0.8 1.0
[Zr4+] (mole /kg)
Fig. 12. Ratio of [Zr%t] /[U4+] at U0, and Zr0, Saturation of

2LiFeBeF, as a Function of [Zr%t].
UNCLASSIFIED

PHOTO 62267

“".":.':!' T B J’b_
(] ‘ ar - o
e

Q
ald e

A 8w
(c)

Fig. 13. Photomicrographs of (a) UOz,1» Starting Material, (b) ZrO,
Starting Material, and (c) Mixture of Well-Crystallized UO, and ZrO, Re-
sulting from Equilibration with 2LiFeBeF,-UF,-ZrF, at 500-700°C.
230

Precipitation of U0, has been suggested as a separations process of pos=-
sible value.® However, such precipitation constitutes a possible danger
for the MSRE since slow precipitation of UO, followed by a sudden en-
trance of the material into the core would permit uncontrolled increases
in reactivity. While precautions will be taken to assure cleanliness of
the system, the fuel mixture, and the cover gas, it must be anticipated
that some inadvertent contamination of the system will occur. Accord-
ingly, an additional safeguard against precipitation of UO; was developed.

If ZrF, is added in reasonable concentration to the fuel mixture,
then contamination of the melt by oxide ion (from Hy0 or BeO) results in
precipitation of ZrO, (Fig. 12). Continued addition of oxide ion leads
to continued precipitation of ZrOp until the Zr**/U** ratio falls to a

value slightly below 2; additional oxide leads to precipitation of both - .

7r0, and UO, as separate species (Fig. 13).° Recent studies suggest that

the relationship between these M*T concentrations and 02~ concentration s

follows the mass action behavior expected for the reaction s
MOz y = MY+ 20%7 . (11) )

The Zr0O, so precipitated from solution shows the x-ray pattern and
the optical properties expected for pure Zr0O,. Solid solutions of U0,
in Zr0, do not form even when both solids are equilibrated at 700°C for
weeks with a fluoride melt under conditions such that gross crystal growth
of both species occurs. Since this behavior appears to contradict pub-
lished phase diagrams of the Zr0;-U0,; system and since it may be important
to the MSRE, it has been demonstrated repeatedly. As a further step it
has recently been established that solid solutions of UOp; and ZrO, pre-
pared by heating the mixed oxides to high temperatures are decomposed into
ZrO, and UOp; by equilibration in a molten Li,BeF, mixture.

Use of ZrF, in the MSRE fuel at concentrations such as five times
that of UF, should effectively prevent precipitation of UO,; and should
completely avoid criticality hazards from this source. The Zr4+/U4+ v
ratio has, accordingly, been set arbitarily at a value of 5 or more.

Addition of sufficient oxide ion to LiF-BelF, mixtures results in =
precipitation of BeO, although the solubility of BeO in such melts is
considerably larger than the solubilities of U0, and ZrO,. No inherent
hazard 1s apparent from the precipitation of small quantities of BeO.
The precautions of system cleanliness and original melt purity are ex-
rected to suffice for the coolant circuit. PR N

MSRE Fuel and Coolant Composition

The final fuel composition for the MSRE will, in fact, depend on-the
amount of uranium required to bring the system to the critical, and then
to the operating, condition. At present it is anticipated that the
composition will be (in mole %) 65 LiF, 29.1 BeF,, 5 ZrF,, and 0.9 UF,.
The pertinent physical properties for this mixture are shown in Table 3.
231

Table 3. Composition and Properties of Fuel and Coolant Selts

Fuel Coolant
Composition, mole %  LiF 65. 66
BeF', 29.1 34
ZrF, 5
UF, 0.9
Liquidus temperature, °F 842 851
Physical properties (at 1200°F) {at 1062°F)
Density, lb/ft> 146 120.5
Heat capacity, Btu 1b™% (°F)~% 0.455 0.526
Viscosity, centipoises 7.6 ‘ 8.3
Thermal conductivity, 3.2 3.5
" Btu £t72 hr™t (°F)7T £t
Vapor pressure Negligible Negligible

Introduction of ZrF, as an important constituent of the MSRE fuel
adds, of course, to the complexity of the phase behavior of the system.
No complete diagram exists at present for the quaternary system so
produced. Figure 14 shows the {essentially completed) diagram for the
LiF-BelF,-ZrF, ternary system.lo A single ternary compound (6LiF- *Bel's -
ZrF,) and two ternary eutectics (at 3 mole % ZrF, and 54 mole % BeF,, mp
370°C, and at 25 mole % ZrF, and 48 mole % BeF,, mp 470°C) occur in the
system.

Careful study of phase behavior of the complex quaternary system in
the region chosen as fuel reveals no phase-instability problems at or
near the reactor temperature. Figure 15 shows, as an approximation of
behavior to be expected of the MSRE charge, an appropriate section across
the more complex quinary system, with ThF, as an additional minor
component.ll

The final MSRE fuel will be obtained by deliberate and controlled
addition of molten LiF-UF, eutectic (27 mole % UF,, mp 490°C) to the
appropriate uranium-free LiF-BeF,-ZrF, solution. From Figs. 14 and 15,
it is apparent that, if the reactor is maintained at temperatures above
500°C, addition of the LiF-UF, material will not produce separation of
solid phases at intermediate or at extreme concentrations.

Choice of secondary coolant is, of course, a more simple matter. We
decided to use a ‘LiF- -BeF, mixture as coolant to avold energetic reactions
with the fuel (as with a NaK coolant) or contamination of the fuel if a
leak developed between fuel and coolant.

Secondary coolants that have extremely low ligquidus temperatures
could be chosen (Fig. 5) from mixtures containing 40 to 60 mole % of BeF,.
232

UNCLASSIFIED
ORNL-DWG 64-1992

Zrfy
@ LiF 903
Lig BeFy ZrFg A
, 900
@© Li, BeF,

COMPOSITIONS IN mole %
© Lip zrFg TEMPERATURES IN °C
® 3LiF-ZrF,

. 850
® 3LiF-azrF,
@ BeF,
@ zrF, 80Q SA
A 750
3LiF- 42¢F, ®
700
P- 5204
£-507

AN
2LiF- Ze R,
596
E-570
3LIF-2rF,

£-598—4 /‘

. %50\ 8s0 \ 450 " ,
LiF aoo \TOO\/VOO \Q © \/ \Zfl/ \‘.\/ N \/ BeF,

845 2LiF - BeF,” ‘P-453 - 548

Fig. 14. The System ILiF-BeF,-ZrF,.

UNGLASSIFIED
ORNL~LR-DWG 688384

500
o
SOLID PHASES:
280 \ A=4LiF-UR
AN B=T7LiF 6UF, OR 7LiF- 6 (U,Th}F, ]|
A+B+L '\ C=LiF
D =6LIiF- BeFs- ZrF,
® 2 4
T—
~ £ = 2LiF - BeF,
5 \.\\ , 2
460 .
w .
:, ~
= B+CHL \ L
o
51
@ ]
= . ¥ o+
TN G UHEXL®
o~ TTVBIDTETL
420 B+CHOHL B¥D+E
o e BICHOFEFL
B+CH+O+E
400
25 20 i5 10 5 0
UF, (mole%)

Fig. 15. The Section LiF-UF, (73-27 mole %)-LiF-BeF,-ZrF,-ThF,;-UF,
(70-23-5-1-1 mole %). :
.

233

Safety considerations, however, gave a mild advantage to a coolant freez-
ing at a temperature slightly above that of the fuel. This fact, plus
the advantage in viscosity gained by diminishing the BeF, content, led
to choice of the mixture with 66 mole % LiF (mp 851°F) as the secondary
coolant. Values for certain pertinent physical properties of this ma-
terial are included in Table 3.

Prenuclear operation of the reactor will use a uranium-free LiF-BeFj
mixture essentially identical with the coolant to test the circulating
system and to remove some or all of the oxides and oxygen-bearing species
from the MSRE system. This material, commonly referred to as the flush
salt, will also be used to flush and clean the reactor circuitry when
necessary maintenance is required after nuclear operation.

MSRE operation will require about 75 £t of fuel mixture, 75 £t of
flush salt, and about 44 ft> of coolant. A total of nearly 25,000 1b of
material must be prepared and transferred in pure form into the fuel,
coolant, and flush tanks of the MSRE. Details of the preparation, puri-
fication, and handling operations on these materials are described in
the following section.

Physical Properties

The physical properties of the MSRE fuel and coolant salts are listed
in Table 3, taken from a recent review of molten salts as reactor ma-
terials.l?

The vapor pressures of the coolant and fuel salt were stated to be
negligible. This conclusion was established by extrapolation to lower
temperaturest® of measured vapor pressures in the temperature range 1000
to 1300°C. TFigure 16 illustrates the effect of doubling the concentration

UNCLASSIFIED
ORNL-DWG 63-486
TEMPERATURE (°C)
1200 {4150 {1100 {050 1000
200 ] | T 1 | |

100 N

N
NN

50 NN

AN

Fig. 16. Vapor Pressure of Two
™ Mixtures in the System LiF-BeF,-ZrF,-

COMPOSITION (mole %) \\\ Uk, .
™.

LiF BeF, 2rfy UR >

10 |- e 86.52 29.38 395 044 \ N
@

4 6467 27.014 843 O.16 ~

- EXTRAPOLATED VAPOR PRESSURE FCR MSRE £

- FUEL AT DESIGN TEMPERATURE (704 °C) =

i 0.02 mm )

.
N

20

PRESSURE {mm Hg}

w

6.6 6.8 7.0 7.2 7.4 7.6 7.8
10,000/ oy
of ZrF, in a fuel salt mixture, while Fig. 17 shows the effects of changes
The possible exploitation of vacuum

234

in the lithium to beryllium ratio.

distillation as a means of chemical reprocessing of irradiated MSRE fuel
(in which the volatile BeF, and LiF are removed, leaving UF, and non-
volatile rare-earth fluorides behind) will call for additional study of

the vapor pressure and vapor composition of fluoride mixtures.

In addition to the data on viscosities of molten fluorides previously

available,:L an experimental program is currently devoted to the systematic
UNCLASSIFIED
ORNL-LR-DWG 78444
100 \\
80 .
\\
40 \‘
N\ \_\ . .
L J A
\ A \ r’
20 \ \1 A
Ne N
. No N
E .\ \ -\ \.
E % N\ .
; RN
’ ;5) 10 \ \\ \
o A N N
: NN N G N
s 6 Ld Y .L_. \ .\\
\\ \{\
A
.4 .\‘\\\ \\
\\ \ . 4
N
COMPOSITION (mole T) \ \\ \
2/ e 100 (THIS STUDY) N
——= 100 {CANTOR, REF. 43) \‘\
A 90 10 \
s 80 20 :\

64

66

€8

10,000/7_(

7o

OK)

72

74

75

Fig. 17. Vapor Pressures in the LiF-BeF, System.

T T
235

UNCLASSIFIED
ORNL-DWG 64-2004

TEMPERATURE (°C}

106 800 700 600 500 475

T 7z T 1 1

/ d ‘
Ay

~ 7
/ A

10% =4 Z

L1
“\ v .
}(/ //
AN
/ AX
4 ALIF BeF, (2.24-97.76 mole %) |
104 LiF-BeF, (4.39-95.61 mole %)

7 7 LiF-BeF, (6.71-93.29 mole %)

103

n (centipoises}

102

FUEL SALT : LiF-BeF,-ZrFy- UFy —
(66.5—28/1’-4.0-0.8 mole %)

"\

101 — )

1
I
COOLANT: LiF—BeF2 (66.0-34.0 mole %)

0.90 0.98 1.06 114 1.22 1.30 1.38

1000/7 (o)

Fig. 18. Viscosity-Temperature Relation for LiF-BeF, Mixtures and
for an MSRE-Type Fuel Mixture,

measurement of viscosities of molten fluoride salts over ranges of temper-
ature and com.position.14 Figure 18 illustrates the tremendous influence
of LiF in lowering the viscosity of BeF, as well as the comfortably low
viscosity of the compositions used as fuel and coolant in the MSRE.

otability of UF, and Ul

Uranium tetrafluoride is the least-stable necessary component of the
MSRE fuel. ©GSince some reduction of UF, to UF3 must be anticipated through
corrosion reactions (as described below) and since UF3 is known to dis- °
proportionate to UF, and metallic uranium under some conditions, the
stability of these materials both as solids and in solution has been
examined.
236

The standard free energy of formation (AF%) of UF, appears accu-
rately known. Rand and Kubachewski give the value —95.3 keal/FO atom
for AF} at 1000°K for the crystalline solid.l? |

Baes has used an LiF-BeF, mixture (67 mole % LiF) in a study of the
reaction

UF4(d) + 2H20(g) ;:_UO2(c) + 4HF(g) , (111)

where g and ¢ represent gaseous and crystalline solid states, and d in-
dicates that the species is dissolved in the molten solvent.i6 He has
shovn that the activity coefficient (y) for UF, in this solution, based
upon the crystalline solid as reference state, has the value 0.55 at
1000°K. 1In solutions like the MSRE fuel, therefore, the activity (a) of
UF, is approximated by half its mole fraction (N).

Long has studied the reduction of UF, to UFj3 by h;ydrogen.l7 In his
work equilibrium values were defined both with UF, and UF3; as crystalline

solids and with these materials dissolved in a molten LiF-BeF, (67 mole %
LiF) mixture. For the reaction

UFa(e) ¥ %HZ(g) = UFso) * H(y) | (Tv)

at 1000°K he obtains

HE _ _HE _ oo 104 atmd/2 . (1)
a  plj2 plf2
H» Ho

From this value, and from the AFQ values of —65.8 kcal/mole for HF and
—381.2 kcal/mole for UF,, he obtains AF% = —301.2 kcal/mole for UFs; (all

at 1000°K). This latter figure agrees well with that estimated by Brewer™$8
in 1945.

When the reduction is accomplished at 1000°K in the LiF-BeF, solu-
tion,* Long has shown that for

1 —
Faa) * F2(g) T @) T F(a) (V)
N P
Ky = s Pf€2 = 8 x 107° . (2)
Nor, TS

*These experiments were done with about 5 mole % UF, in solution. The
v;lues should approximate very well those in MSRE fuel with 1% UF,; and
5 4 ZI‘F4-
237

Since the fugacities (f) and the pressures (P) of the gases are
identical and since (after Baes) NUF4 = aUF4’ these values of Ké and KN

can be combined to yield the activity coefficient (y) of UFs with the
crystalline solid as reference state; this combination yields-yUF = 50.
3

From these data the extent of the disproportionation reaction may be
assessed. For the reaction

UF3(d) = %U + 2UF4( (V)

A a)

the equilibrium quotient (KN) is given by

N3/ 4
}%:——&UF a. . (3)

NUF3 U

From the free-energy change the value Ka for Reaction V with products and
reagents in their standard (crystalline solid) states is known to be

K = ____UE& a. = 4-7 X 10“4 . (4)

From thé activity coefficients given above for UF3; and UF45 the value of
Ky [identified in Eq. (3)] may be computed to be 3.5 x 10™°. From this

value the ratio of UF, to UF; or the extent of reduction of UF, which is
in equilibrium with any designated activity of U° can be estimated.

In the MSR the activity of o might be controlled by formation of a
uranium carbide or by alloying of U° with the INOR-8. The values at

which aU would be controlled may be calculated from values of AFg of the

Table 4. Calculated Values of the Fraction of the
Total Uranium in Solution Present in the
Trivalent State (UF;/Total U) in Equilibrium
at 1000°K with Various Phases

(total uranium in solution = 1 mole %)

Phase U°® Activity UF3/Total U (%)
U metal 1.0 >99
uc 3 x 1073 89
UC» 5 x 10~7 68
Ni alloy ~10"8 49
Ni alloy ~2 x 1010 20

Ni alloy ~1 x 10715 1

238

appropriate carbide and may be estimated (aU = 108 seems pessimistic)

for the alloys with Ni®. Table 4 shows the calculated ratios of UFs/total
uranium for several such uranium activities.

These numbers suggest strongly that quite high ratios of UF¥j3 to
total uranium could be tolerated in MSRE before precipitation of uranium
or its reaction with other MSRE components became a problem. Conversely,
if, by any process, as much as 70% of the uranium were reduced to UFj,
then one could expect the formation of UC, in the presence of graphite.

Chemical Compatibility of MSRE Materials

Successful operation of the MSRE requires compatibility of the
molten fuel mixture with unclad graphite and INOR-8 during years of
rapid circulation of the fuel through an appreciable temperature gradi-
ent. Such compatibility must, moreover, be assured while the fission
process proceeds with its consequent intense radiation field and the
buildup of fission product species. The evaluation of these implied
problems has required a large research and development program in which
many tests have been conducted over a period of several years. Details
and specific findings of the large program of corrosion testing are
presented as a separate chapter ("Metallurgical Development"”) in this
report. The chemical properties of the materials and the chemical nature
of their several interactions are described briefly in the following
section. The effect of irradiation on compatibility of the fuel-graphite-
INOR system is described below.

Corrosion of INOR-8 by Fluorides

Fluids that are circulated in metal circuits at rapid rates through
pronounced temperature gradients frequently attack the metal by very
complex processes. Such corrosion, or its prevention, often depends upon
the stability and self-healing quality of insoluble corrosion product
films; the corrosion resistance of aluminum to hot water (in spite of the
thermodynamic instability of aluminum to water) is a familiar example of
such behavior. '

Molten fluorides such as those proposed for use in MSRE react with

" (and generally serve as fluxes for) oxides of the common metals. Ac-
cordingly, protection of the metal by an oxide or other film seems to be
of no value in minimizing corrosion by fluorides. Corrosion behavior
seems to be governed by the thermodynamic stability of the systems in-
volved and by rates of migration of the reactive species within the
alloy.

‘The corrosive attack (shown in detail in the chapter "Metallurgical
Development"), which is extremely slight, occurs by selective oxidation
of the chromium in INOR-8. Selective removal of chromium from the sur-
face of the alloy results in diffusion of chromium atoms from interior
layers toward the surface.:? The process, accordingly, leads to an
239

excess of vacant sites in the solid scolution; these vacancies agglomerate
in areas of disregistry, principally at grain boundaries and at the sites
of impurity atoms in the lattice, to form voids. These voids tend to
agglomerate and to grow in size with increasing time and temperature.

The subsurface voids are not interconnected with each other or with the
surface. Under MSRE conditions subsurface void formation may be expected
to depths of perhaps 0.5 mil/year.

A considerable fraction of the slight corrosive attack occurs
shortly after startup of the test assembly. However, the evidence (and
especially the older evidence obtained with Inconel, which is more se-
verely attacked than is INOR—S) suggests that attack continues at a slow
rate through some sustained-corrosion mechanism.

Nature of the Chemical Reactions. Selective attack upon the chromium
in INOR-8 by oxidizing impurities in the molten fluorides is not sur-
prising. Casual inspection of the thermochemical data of Table 2 sug-
gests that both FeFy; and NiF,; should oxidize Cr to CrFs. Blood?Y has
made a careful study of the reaction

MFg(d) + Hz(g) = M(c) + 2HF(g) , (VI)

where M represents either Cr or Fe, and c, g, and d indicate that the .
species are crystalline solid (at unit activity), gaseous, or dissolved
in a molten LiF-BeF, mixture. In these experiments the solvent contained
62 mole % of LiF. Results of his studies show (see Table 5) that at
1000°K the experimentally determined equilibrium constants (KN) have the

values 8 X 10™! and 1.3 x 10™* for FeF, and CrF, respectively. The
equation for KN is

K =w X P, (5)

where P and N indicate partial pressure and mole fraction of the desig-
nated species. The reaction of FelF, dissolved in molten LiF-Bel,; with
pure chromium should proceed (producing pure Fe and dissolved CrF,) until
the ratio NCng/NFng = 6000.
Attack of dissolved FeFp; upon chromium in INOR-8 will, of course,
be somewhat different. INOR-8 is a solid solution alloy with 0.083 atom
fraction of chromium and 0.055 atom fraction of iron. Activities of the
two metals should be approximated by these atom fractions. The reaction
of chromium in the alloy with dissolved FeF, to produce iron in the alloy
should be slightly more complete than indicated above. Another likely
reaction,

Fez (q) + CT(g5) = Fe(qy * CrF2 , (V1)
240

Table 5. Experimentally Determined Equilibrium Constants
for the Reaction MF + H = M + 2HF in
2(a) 7 "2 (e) (c) (g)

LiF-BeF, Mixture Containing 62 mole % LiF

(a)
fu

Temperature
(°c)
For CrF, For Fel's For NikF,

800 4o X 1074 1.9

727 1.3 x 10™% 0.80

700 7.5 x 1077 0.53 7 X 10°
600 1.2 X 107 0.13 1.5 x 10%
a 2

Where KN = PHF

NMF2 X PH2

where Cr(ss) is in solid solution in INOR-8 and Fe(c) is crystalline iron

* - - - » g OO. -
at unit activity, should proceed until the ratio NCng/NFng 5 De

pletion of chromium in the surface layers of INOR-8 (to be replenished
slowly by diffusion from the interior) would, of course, lower this ratio.

Selective attack upon the chromium can occur by any of several re-

actions. These include:

1.

Reactions with Dissolved Impurities in Melt — The fused fluorides
are purified (see the chapter "Preparation of MSRE Fuel, Coolant,
and Flush Salt" for details) by successive treatments with HF-I,
mixtures and with H,. Incomplete removal of HF or of any easily
reducible fluoride will subsequently lead to corrosion of the
INOR-8. Reactions such as

PHF + Cr = CrF, + Ho (VIII)

and
NiF, + Cr = CrF, + Ni (IX)
are examples. Available free-energy data and direct experimentation

confirm that these reactions are even more complete than Reaction
(viz).

Reactions Due to Oxides on the Metal — Oxide films on the INOR-8
are attacked by the molten fluoride by reactions such as
241

2CI‘203 + BZI'F4 - BZI‘Oz + 4CI'F3 P (X)
2Fe0 + ZrF, - Zr0, + 2FeF, . (X1)

The resulting Zr0O, appears to be of little consequencé in corrosion,
but the CrFs; and FeF, are strong oxidants to chromium in INOR-8.

Reactions of the type shown under 1 and 2 above can, if the fuel and
the container are not carefully prepared, lead to serious initial
corrosion. Such reactions can be minimized (or even avoided) by
careful purification and manufacturing procedures. It is likely that
these two classes of reactions will produce CrF, in the fuel equiva-
lent to about 250 parts of chromium per million parts of fuel. The
CrF, from these sources should, therefore, be at concentration
N = 2.5 X 1074,
Crls

3. Reactions with Necessary Constituents of Melt — Attempts to disclose
the existence of Zr>t and Bet species in molten fluoride solution
lead to the conclusion that reactions such as

37rF, + 7Zr = 4ZrFi (XII)

do not occur; Zr3+ species do not exceed concentrations of 10 ppm if
they are present at all. Accordingly, reactions such as

Cr + 27rF,; = CrF, + 27ZrFs; (XIII)

apparently need not be considered.

Uranium trifluoride is, however, a well-known compound that is
relatively stable in molten fluorides of the type to be used in MSRE.
The extent of reaction is relatively small; the values in Table 2 in-
dicate a free-energy change of +12 kcal for oxidation of Cr by UF4. In
spite of this, the reaction is an important contributor to the corrosion
process; an evaluation of its contribution is examined in detail in the
following.

Extent of Chemical Reactions. The data obtained from studies of
hydrogen reduction equilibria by Long and by Blood (described above) can
be used to yield reasonable (and consistent) estimates of the activity
coefficients of the several species in melts similar to the MSRE fuel.
These activity coefficients, based on the crystalline solid as reference
state, are shown in Table 6.

From these values and the values (Table 2) of AF% for the several

compounds, we may assess the extent of the reaction

2UF4(d) + Cr(SS) = Cng(d) + 2UF3(d) , , (XIV)
242

where the UF,, Crl's, and UFi3 are in solution in MSRE fuel and the Cr is
at the relatively low activity it has in unaltered INOR-8. TFor this
reaction,

2 2
. a (y ) (M )
K = 2.4 x 107 - s " Poxry | TV I0, TV erm, (6)

2 . 2
ur, * %or (% )gp, 2oy

If, initially, the salt were completely pure and the metal contained
no oxide {so that all CrF, was generated by this reaction), then

Yo, = Mopr, © (7)

For reaction of a fuel with 1 mole % UF, (I\TUF4 = ayp, = 0.01) with INOR-8

(acr = 0.083), the equilibrium indicated in Eq. (6) is satisfied at

_ —
I\ICrF2 = 1.1 x 10

and

N 2.2 x 1074 .

UF3

Accordingly, less than 3% of the UF, would be reduced to UF3, and the
chromium fluoride concentration of the melt would be 130 ppm (as Cr). In
a more realistic case if CrFy equivalent to some 250 ppm of chromium

arose from other sources, then the total CrF; concentration at equilibrium
would rise to slightly more than 300 ppm (as Cr), and N would be about
1 x 1074, ¥

Table 6. Activity Coefficients® Estimated for Various Fluorides
in Molten LiF-BeF, Solutions at 1000°K ‘

AFO Approximate
Species f Equilibrium Studied Concentration 04
(kcal/F atom) (mole fraction)
UF, —95.3 UF, + 2Ho0 = U0, + 4HF 0.02 0.55
UF5 ~101.4 UF, + %Hg — UFs; + HF 0.0001 50
CrFs, —74 © CrFy + Hy = Cr + 2UF 0.001 0.5

®Based on the crystalline solid as reference state.
243

It is easy to show that if, as a consequence of FeF, impurity, the
INOR-8 surface were severely depleted of chromium and if iron at unit
activity were available within the system, the reaction would yield at

equilibrium NUF = 3 x 10™° (corresponding to reduction of only 0.3% of
3

the available UF,) and NFng

15 ppm as Fe). It is highly unlikely that any tests have been run under
such circumstances.

= 1.5 x 10~° (corresponding to FeF, at about

This reaction will certainly not produce sufficient UF3 to cause
trouble from disproportionation of this material. Moreover, the total
quantity of chromium removed from the metal should prove trivial in any
uniform temperature system.

However, in a nonuniform temperature system such as MSRE the small,
but real, temperature dependence of reactions such as (XIV) can result
in removal of chromium from the metal in the hotter regions and its
deposition in the colder regions of the circuit. However, since the tem-
perature dependence of the reaction is so slight, the deposition can re-
sult only in a slight increase in chromium activity (concentration) in
the INOR-8 in the cooler regions.

If nickel, iron, and molybdenum are assumed (as is approximately
true) to be completely inert diluents for chromium, and if the circulation
rate of 1200 gpm for MSRE can be considered infinitely rapid, the process
can be simply described. Under such conditions uniform concentrations of
UF3 and CrF, are maintained in the fluid; these concentrations satisfy
(at some intermediate temperature) the equilibrium constant for the re-
action shown. Under these steady-state conditions, there exists a tem-
perature (T), intermediate between the maximum and minimum temperatures
of the loop, at which the initial composition of the structural metal is
at equilibrium with the fused salt. Since the equilibrium constant for
the chemical reaction increases with increasing temperature, the chromium
concentration in the alloy surface is diminished at temperatures higher
than T and is augmented at temperatures lower than T. In some melts
(NaF-LiF-KF-UF4, e.g.) the equilibrium constant of reaction changes
sufficiently under extreme temperature conditions to cause formation of
nearly pure chromium crystals in the cold zone. In other melts (LiF-
BeF,-UF;, e.g.) the temperature dependence of the corrosion equilibrium
is small, and the equilibrium is satisfied at all useful temperatures
without the formation of crystalline chromium. In the LiF-BeF,-UF, system
the rate of chromium removal from the salt stream at cold-leg regions is
dependent on the rate at which chromium can diffuse into the cold-leg
wall. If the chromium concentration gradient tends to be small, or if
the bulk of the cold-leg surface is held at a relatively low temperature,
the corrosion rate in such systems is almost negligible.

As the alloy surface in the hot region at the circuilt becomes de-
pleted in chromium, chromium from the interior diffuses down the concen-
tration gradient toward the surface. Since diffusion occurs by a vacancy
process and in this particular situation is essentially monodirectional,
it is possible, as previously mentioned, to build up an excess number of
s

vacancies in the metal. These precipitate in areas of disregistry,
principally at grain boundaries and impurities, to form voids. These
voids tend to agglomerate and grow in size with increasing time and tem-
perature. The subsurface voids are not interconnected with each other
or with the surface.

The mechanisms described above seem to explain such observations as
the complete independence of corrosion rate and flow rate for a given
system, and the increase in corrosion with increase in tenperature drop
as well as with increase in mean temperature within a system.

The results of many, long-term tests of corrosion of INOR-8 and
Inconel by MSRE fuel mixtures and other salt combinations (see the chapter
"Metallurgical Development"”) seem to substantiate very well the expecta-
tions from these thermodynamic considerations.

Compatibility of Graphite with Fluorides

Graphite does not react chemically with molten fluoride mixtures of
the type to be used in the MSRE. Available thermodynamic data suggest
that the most likely reaction:

LUF, + C = CFy; + 4UF; (xv)

should come to equilibrium at a concentration of CF,; which is below the
detection limit (about 1 ppm) of this compound by mass spectrometry.
Moreover, graphite has been used as a container material for many Nai'-
ZrF,;-Uf,;, LiF-BeF,-UF,, and other salt mixtures, with no evidence of
chemical instability.

The MSRE will contain about 70 ft> (nearly 8000 1b) of graphite.
Even though chemical stability is assured, several possible problems re-
main. These include (1) hazardous increase in uranium content of core
through permeation of the graphite by fuel, (2) reaction of fuel material
with oxygenated gaseous species desorbed from the graphite, and (3) car-
burization of the INOR-8 structure by carbon dissolved, suspended, or
otherwise carried in the circulating salt. These possibilities have been
studied experimentally and found to be inconsequential or to have practi-
cable solutions. :

Graphite is not wetted by MSRE fuel mixtures (or by other similar
mixtures) at elevated temperatures. The extent to which graphite is
permeated by the fuel is, accordingly, defined by well-known relationships
among applied pressure, surface tension of the nonwetting liquid, and the
pore-size gpectrum of the graphite specimen. Typical tests?! with MSRE
graphite have exposed evacuated specimens to MSRE fuel mixtures at 1300°F;
applied pressures were set at 150 1lb, a value three times the reactor
design pressure. The observed permeation did not change with time after
a few hours. In these tests 0.3% of the graphite bulk volume was per-
meated by the salt; such permeation is well within that considered
tolerable during MSRE operation. Specimens permeated to this extent have
245

been given 100 cycles between 390 and 1300°F without detectable change
in properties or appearance.

Graphites contain appreciable quantities of gases that are not re-
moved by pumping at room temperature but that are desorbed on heating in
vacuum to high temperatures. These gases typicallg consist of Hy0, CO,,
CO, Np, Hp, and small quantities of hydrocarbons .? The MSRE graphite
was specified to contain less than 30 STP em?® of CO (or equivalent 0) per
100 em? of graphite after pumping at 30°C to a pressure of 10™2 mm HgP;
subsequent tests?®t have shown the graphite to contain about 6 cm® of these
gases per 100 cm®. The fuel mixture could, if necessary, withstand much
larger quantities of active oxide without difficulty from precipitation
of Zr0,. In addition, it has been shown that a helium purge, maintained

- while the graphite is slowly heated to 400°C, effectively removes the
small quantity of physically absorbed and chemisorbed water from the MSRE

: specimens.?? Carbon monoxide and CO, react slowly if at all with the
molten fluoride mixtures, and most of the water should be readily re-
movable during preliminary heatup of the reactor. It is unlikely that
desorption of gases from the moderator will be of consequence in MSRE
operation.

UNCLASSIFIED
ORNL-LR-DWG 27902

SAMPLE INSERT

Fig. 19. Diagram of Forced-Circulation Loop for Weight-Loss Studies
of INOR-8 Exposed to Circulating Fused Salts.
246

Carburization of INOR-8 by graphite in contact with molten fluorides
has been studied in static capsule tests?% and in a long-lived forced-
circulation :Loop.25

The static capsules were run at 1300°F with INOR-8 specimens, graph-
ite, and an LiF-BeF,-UF, fuel mixture contained in capsules of INOR-8.
Exposures were for multiples of 2000 hr up to 14,000 hr. All tests indi-
cated no carburization and no appreciable alteration of the tensile
properties of the metal. The longest test (14,000 hr) produced a slight
roughening (less than 0.5 mil) of the INOR-8 specimens.

The forced-circulation loop?® of INOR-8 was operated at 1300°F for 1
year with an LiF-BeF,-UF, (62-37-1 mole %) mixture as the circulating
medium. An INOR-8 box, located at the heater outlet of a standard loop
(see Fig. 19), contained 32 rods (11 X 1/2 in.) and 18 rods (11 X 3/16
in.) of a special experimental graphite from National Carbon Company;
this graphite was generally similar to the proposed MSRE moderator ma-
terial. Each of the larger rods was fitted with four spacer rings of
INOR-8 wire to obtain the desired flow characteristics and to provide
samples for metallurgical tests for carburization. The graphite speci-
mens after test are shown in Fig. 20. Weight changes, dimensional
changes, and chemical analyses of machinings from the graphite26 revealed
no detectable erosion of the specimens, and permeation of the graphite by
only traces of fuel. Carburization of the INOR-8 appeared to be absent.

No difficulties with out-of-radiation compatibility of the fluoride-
graphite-metal MSRE system have so far been found.

UNCLASSIFIED
PHOTO 71408

°
o

L n 1 1
INCHES

Fig. 20. Graphite Specimens After Exposure to Molten Fluoride.
/Hflw—'\—u\'

247

Behavior of Fission Products

When fission occurs, the fragments originate in energy states and
ionization levels very far from those normally encountered. When fission
occurs in a well-mixed liquid medium, these fragments must, as they
quickly lose energy through collisions in the liquid, come to a steady
state {made very complicated by the radioactive decay of many species) as
common chemical entities. The valence states which they assume in the
liquid are, presumably, defined by the requirements that cation-anion
equivalence be maintained in the liquid and that redox equilibria be
established both among the components of the melt and between the melt
and the surface layers of the container metal.

When fission occurs in aqueous solution, the requirement of cation-
anion equivalence can, if necessary, be satisfied by HY or OH™ ions
supplied by the solvent. No such mechanism is available to the MSRE fuel.
In these media, since Fp and valence states of uranium higher than 4 are
unstable toward reduction by the container, the fission product cations
must neutralize the fission product anions plus the 4 fluoride ions as-
sociated with the fissioned atom. If the fission product cations are
inadequate, or if they become adequate only by assuming oxidation states

 incompatible with INOR-8, then the container metal must supply the

deficiency.

Thermochemical data, from which the stability of fission product
fluorides in complex dilute solutions can be predicted, are lacking in
many cases. No precise definition of the valence states of all fission
products can be given, and the complex situation in the MSRE fuel cannot
be defined in real detail. Such information as seems definite is briefly
described in the following. The data obtained in many in-pile tests (see
below) suggest strongly that the fuel mixture survives the fission process
without requiring substantial contributions from the container metal.

Rare Gases

The fission products krypton and xenon are volatilized from the high-
temperature melts as elements. The solubilities of these gases in molten
fluoride mixtures obey Henry's law, increase with increasing temperature,
decrease with increasing atomic weight of the gas, and vary somewhat with
composition of the solvent.?” Henry's law constants and heats of solution
for the rare gases in NaF-ZrF, and LiF-BeF, mixtures are shown in Table 7.
The positive heat of solution ensures that blanketing or sparging of the
fuel with helium or argon in a low-temperature region of the reactor can-
not lead to difficulty due_ to decreased solubility and bubble formation
in higher-temperature regions of the system.

The very low solubilities of these gases suggest that they should be
readily removed from reactor systems. Only a small fraction of the calcu-
lated xenon poisoning was cbserved during operation of the Aircraft Re-
actor Experiment, where the only mechanism for xenon removal was the
helium purge of the pump bowl. Removal of xenon from the MSEE will be
248

Table 7. Solubilities at 600°C and Heats of Solution for

Noble Gases

in Molten Fluorides

NaF-ZrF, LiF-Bel's
- mole o - mole 7
(53-47 mole %) (64-36 mole %)
Gas
' v g Heat of Solution g8 Heat of Solution
kx x 10 (kcal/mole) K* X 10 (kcal/mole)

He 21.6 £ 1 - 6.2 11.55 = 0.4 5.2
Ne 11.3 £ 0.3 7.8 4.63 £ 0.2 5.9
A 5.1 £ 0.15 8.2 0.98 + 0.02 8.6
Xe 1.94 £ 0.2 11.1 0.233 + 0.01 12.1

K = moles of gas/(cm® of

solvent) (atm).

UNCLASSIFIED

ORNL-LR-DWG 56202A

5 .
‘ ‘KHIe=1x1O_6 CEY AGOT |
r=0% *
— et ] gt
ag ) // : -1 g
> p / /,/ /
z LA =
o e / d e L~
E’ 3 // / / // /
- 4 /] / 4 e @ AT
P / / / / // /
g 5 L / / 104 / L/
o e
VANV
e /| 122 / /
5 1 /// A /////
504
/// / € =10.52 %
///::100’ D)= Dx10_5)1/2 cm/sec
oE==—rT111! LTt L
10-7 10°8 1075 1074

Ox (em/sec)

Fig. 21. DPossible Xenon Poisoning in Molten-Salt Reactors.

- ——
5

249

complicated by holdup of this gas in the pores of the graphite and by the
relatively small fraction (5 to 10%) of the fuel flow which is bypassed
through the pump bowl. Figure 21 shows a family of curves which relate
poison fraction to be expected as a function of recycle rate (percentage
of fuel flow equilibrated with a stripping gas) and of permeability of

the moderator graphite. It is clear from this figure that very marked
improvement in tightness of graphite will be required to solve the problem
at low recycle rates, but that a simple gas-stripping system operating on
the entire fuel flow should solve the problem for a reactor with any
reasonable moderator graphite.

Elements of Groups I-A, II-A, ITI-B, and IV-B

Rubidium, cesium, strontium, barium, zirconium, yttrium, and the
lanthanides form very stable fluorides. These fission products should,
accordingly, exist in the molten fuel in their ordinary valence states.
Many studies show that large amounts of ZrF,, the alkali fluorides, and
the alkaline earth fluorides can be dissolved in the MSRE fuel mixture
at operating temperatures. ©Since the trifluorides are less soluble, the
solubility behavior of the fluorides of yttrium and the rare earths, and
of plutonium have been examined in some detail.?8:29 mpe saturating phase
from scolutions in LiF-Bel'; and related mixtures is the simple trifluoride;
when more than one rare earth is present, the saturating phase is a
(nearly ideal) solid solution of the trifluorides. The solubilities of
these materials depend strongly on composition of the melt and may be near
the minimum in compositions close to that of the MSRE fuel. Even in this
region, however, the solubility (near 0.5 mole % at MSRE operating tem-
peratures) is such that many years would be required for the reactor to
saturate its fuel with any of these fission products.

The statements regarding rubidium and cesium above do not apply to
those fractions of these elements which originate in the graphite as
daughters of the rare gases which have permeated the moderator. These
alkali metals form compounds with graphite at high temperature; the
quantities involved, however, are so small that no damage to the graphite
seems possible.

Other IFission Products

The fission of the uranium atom of UF, would yield more cation equiva-
lents than anion equivalents if the noble-gas isotopes of half-life
greater than 10 min were lost from the fuel and if each of the other fis-
sion products formed a fluoride of its lowest recognized valence state;
if this were the case, the system would retain cation-anion equivalence
by depositing the most noble fission products as metal and by reduction,
if necessary, of some UF, to UF3. The sparse thermochemical data suggest
that the fluorides of germanium, arsenic, niobium, molybdenum, ruthenium,
rhodium, palladium, silver, cadmium, tin, and antimony should be reduced
by chromium at its activity in INOR-8. In dilute solution in the MSRE
fuel, not all these reductions would be complete, but some of these ele-
ments (and perhaps part of all of them) must be expected to exist as
250

metals. Only about 3.2 cation equivalents result from fission in UF, if
all these fission products exist entirely as metals; the fission process
would in that event result in oxidation of INOR-8. Considerable -qualita-
tive information from in-pile capsules and in-pile loops suggests that
the true situation is comfortably between these extremes and in a region
where the cation-anion equivalence is satisfied at a redox potential near
to that of INOR-8. ,

References

1. W. R. Grimes et al., "Chemical Aspects of Molten-Fluoride-Salt Re-
actor Fuels," in Fluid Fuel Reactors, ed. by J. A. Lane, H. G.
MacPherson, and Frank Maslan, Addison-Wesley, Reading, Mass., 1958;
and W. R. Grimes and D. R. Cuneo, "Molten Salts as Reactor Fuels,"
in Reactor Handbook, 24 ed., vol. I, Materials, ed. by C. R.

. Tipton, Jr., Interscience, New York, 1960.

2. W. R. Grimes and D. R. Cuneo, "Molten Salts as Reactor Fuels," p.
' 457 in Reactor Handbook, 24 ed., vol. I, Materials, ed. by C. R.
Tipton, Jr., Interscience, New York, 19€0.

3. R. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materials,
" ORNL-2548 (Nov. 2, 1959), p. 116.

4. Toid., p. 33.

5. L, V. Jones et al., J. Am. Ceram. Soc. 45(2), 79-83 (1962).

6. R. E. Thoma et al., J. Phys. Chem. 64, 865 (1960).

7. C. F. Weaver ¢t al., Phase Equilibria in Molten Salt Breeder Reactor
Fuels. I. The System LiF-BeF,-UF,~ThF,, ORNL-2896 (Dec. 27, 1960).

8. J. H. Shaffer et al., Nucl. Sci. Eng. 18, 177-81 (1964).

9. Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1963, ORNL-3417, p. 38;
C. F. Baes, Jr., J. H. Shaffer, and H. F. McDuffie, Trans. Am. Nucl.
Soc. 6(2), 393 (1963); Reactor Chem. Div. Ann. Progr. Rept. Jan. 31,
1964, ORNL-3591, p. 45.

10. Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591, p. 4.

11. Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1963, ORNL-3417, p. 5.

12. W. R. Grimes, Nucl. News 7(5), 3-8 (1964).

13. MSRP Semiann. Progr. Rept. Jan. 31, 1963, ORNL-3419, pp. 116-19.

14. Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591,
pp. 52=53.

15.

l6.

17.

18.

19.

20.

21.

22

23.

24 .

25.

26 .

27.

28.

29.

251

M. H. Rand and O. KhbaSéhewski, The Thermochemical Properties of
Uranium Compounds, Interscience, New York, 1963.

Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591,

pp. 46-50.

MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, pp. 119-20.

L. Brewer et al., The Thermodynamic Properties and Equilibria at
High Temperatures of Uranium Halides, Oxides, Nitrides, and

Carbides, MDDC-1543 (1945).

W. R. Grimes et al., "Radiotracer Techniques in the Study of Cor-
rosion by Molten Fluorides," pp. 55974 in Proceedings of Conference
on the Uses of Radioisotopes in the Physical Sciences and Industry,

Copenhagen, Sept. 16—18, 1960, vol. 3, IAEA, Vienna, 1962.

C. M. Blood, Solubility and Stability of Structural Metal Difluorides

in Molten Fluoride Mixtures, ORNL-CF-61-5-4 (Sept. 2L, 1961); and
C. M. Blood et al., "Activities of Some Transition Metal Fluorides
in Molten Fluoride Mixtures," in Proceedings of the International
Conference on Coordination Chemistry, 7th, Stockholm and Uppsala,

June 25-29, 1962, Butterworths, London, 1963.

W. H. Cook, "Evaluation of MSRE Graphite," Metals and Ceramics Div.
Ann. Progr. Rept. July 1, 1963, ORNL-3420, pp. 165-67.

L. G. Overholser and J. P. Blakely, "The Degassing Behavior of
Commercial Graphites," in Proceedings of the Fifth Conference on

.Carbon, Pergamon; 1962.

MSRP Semiann. Progr. Rept. Jan. 31, 1963, ORNL-3419, pp. 122-27.

MSRP Semiann. Progr. Rept. Aug. 31, 1961, ORNL-3215, p. 93.

J. L. Crowley, MSRP Semiann. Progr. Rept. Jan. 31, 1958, ORNL-2474,
p. 31.

Reactor Chemistry Div. Ann. Progr. Rept. Jan. 31, 1960, ORNL-2931,

pp. 69-70.

W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys. Chem. 62, 862
(1958); M. Blander et al., J. Phys. Chem. 63, 1164 (1959); . M.

Watson et al., J. Chem. Eng. Data 7(2), 285 (1962).

W. T. Ward et al., Solubility Relations Among Rare-Earth Fluorides,
ORNL~2749 (Oct. 13, 1959).

A. J. Singh, G. D. Brunton, and R. E. Thoma, Zone Melting of In-
organic Fluorides, ORNL-3658 (to be published).

252

EFFECTS OF RADIATION ON THE COMPATIBILITY OF MSRE MATERIALS

A

. . Blankenship

Summary

Postirradiation examination of the several capsules from three in-
pile tests of MSRE fuel-graphite-metal compatibility continues to furnish
evidence of evolution of ¥, and CF,, graphite damage, and other potentially
unfavorable phenomena. All cases of unusual behavior, however, seem to be
consequences of F, generation at low temperatures under conditions which
will not normally obtain in MSRE.

Radiolytic formation of F, at temperatures below 100°C was observed
over a 5-month interval in two capsules that were fitted for sampling of
the evolved gases. Fluorine release proceeded at a maximum rate near room
temperature (G = 0.01 to 0.02 molecule of F, per 100 ev absorbed) and de-
creased to zero rate at —70°C and +80°C. At the low temperature, slow
diffusion in the crystals presumably limited the evolution rate, while at
high temperatures the back reaction of radiolytic products appeared to be
controlling. Deliberate addition of F, produced results which confirmed
this picture and which demonstrated that F, reacts more rapidly with salt
which has previously lost most of its F, radiolysis. These observations
conclusively eliminated the specter of fluorine generation at reactor op-
erating temperatures. In reactor components such as freeze valves or
freeze flanges and in tanks for storage of irradiated salt, however, the
temperature must be kept high enough (200°C seems sufficient) to prevent
salt radiolysis.

Chemical analyses of graphite cores from the large capsules of one
assembly revealed the presence of appreciable (0.1 to 0.5 wt %) quantities
of uranium, as well as molybdenum, lithium, and container metals. Auto-.
radiography, before and after neutron activation, and x radiography of
transverse sections of the cores showed that the uranium was present in
a 2-mil-thick layer on the surface of the specimens. Neutron activation
analyses gave slightly lower results for uranium than those obtained by
chemical analyses. The deposition of uranium is believed to have been a
consequence of fuel radiolysis during reactor shutdowns in the course of .
the in-pile exposure. To test this conjecture, the design of the last
in-pile experiment (47-6) provided for heaters on each capsule to prevent
radiolysis during shutdown. The performance of this experiment supported,
in every way tested, the conclusion that the MSRE components are compatible
for use as designed, and the gross visual appearance of the components on
disassembly of the experiment has given further support to this conclusion.
The components (INOR-8, graphite, salt, and molybdenum specimens) have been
. separated and packaged for return to ORNL and detailed hot-cell examination.

T S
253

Introduction

Many familiar examples of systems displaying stability at high tem-
peratures depend on the formation of protective films; but the MSRE fuel-
metal-graphite compatibility relationships are not among these. As indi-
cated in the preceding chapter, the components of the fluoride melts used
as a fuel are chosen with a view toward complete thermodynamic stability
with respect to nickel alloys and graphite, and the resulting combination
is substantially corrosion free, isothermally at least, without the inter-
mediation of interfacial films. Pertinent details of corrosion-related
behavior of the fuel that have been described previously present no phe-
nomena which obviously render them inherently susceptible to adverse ef-
fects from fissioning. The possibility that off-equilibrium species might
cause undesirable effects has been examined experimentally and found to be
vanishingly small or beyond the limit of detection. However, the history
of these experiments provides some interesting sidelights, particularly
in connection with the difference in behavior of crystalline and molten
fuel.

There exists, as a legacy from the ANP Program, an impressive and
reassuring collection of irradiation experiments involving fissioning
molten fluorides in contact with nickel-based alloys — more than 100
exposures to reactor irradiation in various fluoride mixtures in Inconel
capsules and three in-pile forced-circulation loop experiments. The con-
clusion, as stated on May 1, 1958, was as followss:?!

"Within the obvious limitaticns of the experience up to the
present time, there is no effect of radiation on the fuel and
no acceleration of corrosion by the radiation field."

But, less was known about the compatibility of graphite with fissioning
fuel. Owing to the chemical inertness of graphite with respect to the
fuel, there was little reason to suspect gross corrosive effects, but
the interfacial behavior was a matter of concern,

The fuel, because of its high surface tension, dces not soak into
the pores of graphite. It was of great importance to ensure that such
a favorable interfacial behavior persisted in spite of irradiation ef-
fects, since about 10% of the volume in the core is comprised of the
voids in graphite, '

Another point of interest for long-term operation is the nature of
the deposition of the noble metals produced as fission products. When a
uranium cation fissions, the resulting fission fragments correspond, on
the average, to cations having a combined valence of about +3.4 in place |
of the +4 of the uranium cation. The exact replacement valence deficiency
depends on the rate of removal of short-lived xenon ‘and krypton that de-
cay to cation-forming elements and also, more importantly, on the fact
that the wvalences adopted by the fission products will be those for equi-
librium with the chromium metal at the chemical activity chromium pos-
sesses in the INOR-8 container. The net effect is a gradual chemical
oxidation at a rate which would increase the Cr?* concentration in the
254

MSRE by about 50 ppm[year. Concurrently, there is a deposition of fissicn
products such as molybdenum and ruthenium as elemental metals. In practice,
the change in oxidation level of the fuel (i.e., in the UF, to UFs ratio)
is readily adjusted by scheduled reprocessing treatments, and any mismatch
with the corrosion equilibrium

PUF, + Cr® = 2UF3 + CrF;

is spontaneously corrected by a shift in concentrations toward values cor-
responding to equilibrium.

An underlying premise of the foregoing considerations was that the
chromium in the INOR-8 container remains available as a reducing agent,
and that this is the main factor influencing the deposition of noble
metals. Should the INOR-8 become plated with noble metals, then occa-
sional reprocessing may be necessary to prevent the slow oxidizing effect
of fissioning from tending to release iodine and tellurium from the fis-
sioning fuel.

Preliminary Experiments

Beginning in 1960, a sequence of in-pile experiments was directed
toward resolving uncertainties regarding the compatibility, under irra-
diation, of MSRE components. :

The first two of these, Experiments 47-1 and 47-2, were essentially
duplicates and suffered from an experimental shortcoming such that the
results were quite limited and need only brief mention. The experimental
difficulty derived from an attempt to maintain pressure on the irradiated
system by using a bellows completely filled with salt as the capsule wall
submerged in a pressurized bath of liquid sodium. F'reeze-thaw cycles dur-
ing operation of the pile led to rupture of the bellows and to the loss
of seven out of eight capsules. The tendency to rupture was probably
aggravated by fluorine gas that might have been present in accord with
phenomena to be described below.

The single surviving capsule gave generally reassuring results in
that the surface of a graphite core appeared unaffected by exposure to
fissioning fuel. Numerous observations with a microscope of metallo-
graphically mounted sections of the graphite indicated that no significant
penetration of the graphite by salt had occurred and that the graphite
appeared unchanged.

Autoradiographs and radiochemical analyses showed minor intrusions
of fission products into the graphite by two mechanisms. There were a
few fissures or faults in the graphite that were repositories for fis-
sion product radioactivity resembling that in the fuel. Otherwise the
activity in the interior of the graphite was quite low. The other mode
of pickup of activity may have been augmented by diffusion of gases; this
was in the form of an annulus of activity at the circumference of the
graphite core. The inner fringe of this annulus, as studied by drilling

255

with a hypodermic needle and gamma spectroscopy of the drillings, showed
only one major gamma activity, that of cesium. Approximately two-thirds
was 34Cs and one-third was 127Cs; the decrease in concentration with in-
creasing penetration suggested that the cesium was depocsited by xenon that
diffused into the graphite, but there were discrepancies that prevented
establishment of this mechanism with certainty.

Experiment 47-3

While it appeared from the preliminary trials that nothing drastic
happened to graphite when irradiated in the presence of fuel, there were
still several questions on which further information was desired. For
example, the matter of the penetration of the moderator by fuel was suf-
ficiently important to require further confirmation.

A1l to 2% permeation of the graphite by fuel could be tolerated in
reactor operation. But, should penetration attain or exceed the 4% ac-
cessible voids in the graphite, as might occur if the wetting character-
istics changed during operation, serious difficulties with reactor control
could be anticipated., Initially, this problem was thought to be the most
serious one the MSRE faced; several design features such as an increased
ratio of core volume occupied by fuel to that occupied by moderator, di-
Juting the 2357 content of the salt, and increasing the capacity of the
control elements were invoked to reduce the seriousness of a change in
wetting behavior to manageable proportions. But there was still a ques-
tion regarding the extent of fission fragment damage that could cccur in
the matrix of permeated graphite,

As far as was known at the time, radiation-induced wetting might
have been caused by two possible paths: (1) changes in the surface ten-
sion of the molten salt by fission products and (2) alteration of the
surface energy of the graphite by either radiation damage or chemical
interaction such as the formation of carbides or interlamellar compounds,
although most of such mechanisms appeared unlikely. Since graphites of
varying degrees of graphitization had been shown to exhibit no wetting
by the fuel, there was little reason to suspect that damaged graphite
would show an enhanced tendency to wet.

In evaluating the integrity of the moderator, one factor which was
considered was the stripping of the surface of carbon atoms by the energy
deposited in the surface layer by fission fragments. If it were assumed
that each fission fragment striking the surface sputters 20 carbon atoms,
and that the mean fragment range normal to the surface was 10 u, then for
1.0° fragments/sec incident on 1 cm? of surface, 2 x 10'0 carbon atoms
would be sputtered per cm? per second. This amounts to &bout one atom
layer per day, a relatively negligible amount of graphite.

Another potential problem centered on the possibility that a portion
of voids near the surface of the graphite might become occupied by salt
and that fission heat produced in these regions might develop more rapidly
than it could be removed by thermal transport. This could lead to frag-
mentation of the surface layer adjacent to the void, but probably the salt
would merely expand intc interconnecting voids.
256

The fission fragment range in graphite is about 20 i, and graphite
within this distance of the fuel could be severely damaged. Little is
known about the nature of fission fragment damage in graphite; but, if
it is an accentuation of the effect of fast neutrons, it could conceiv-
ably cause contraction to such a degree that microscopic checking would
occur. Progressive and cumulative fission fragment damage could then
proceed, with a resulting spalling off of surface material.

Still another problem was the possibility of chemical attack by re-
active fission products. The two most likely agents, because of both
chemical behavior and fission yield, were the halogens bromine and iodine.
It is well known that halogens form intercalation compounds with graphite.
Extensive compound formation leads to a separation of the graphite layers
and eventual breakup of the material. In the fuel the halogens would exist
in the atomic state only during an exceedingly brief transitory period,
and their concentration would probably be too low to give a detectable
amount of reaction. :

During steady irradiation, iodine reaches a maximum yield of 0.13
atom/fission at 10% sec, from which it declines to a small value at
longer times. At the end of 10 days' operation, the surviving yield is
only about 0.03 atom of iodine per fission, and this would be in the form
of iodide under reactor operating conditions. The bromine accumulation:
is lower than this value by an order of magnitude. Therefore, chemical
attack is not expected to be a serious factor for moderator integrity.

The extent of carburization of the reactor vessel or the metal com-
ponents is another detail that requires consideration. Mild carburization
is of little concern, but enough carburization to cause a hard case is
another matter. Sacrificial pieces of INOR were used to prevent the
graphite pieces from contacting structural members to avoid damage from
carburization in the MSRE.

Description of Experiment

Relatively extreme exposure conditions were used to accentuate the
effects of fissioning in Experiment 47-3. A power density of 200 w/cm3
was developed by adjusting the concentration of 235UF4 to 1.5 mole % in-
stead of the 0.3 mole % to be used in the MSRE. Consequently, the four
3-week cycles in the MIR, accumulating 1594 hr at power, gave a burnup
of 8.5% in comparison with the 6%/year anticipated for the MSRE at 10 Mw,
The contact angle of the fuel meniscus was chosen as a convenient and re-
liable index of possible changes in wetting.

To make the contact angle manifest, a vertical blade of graphite was
~arranged to dip into a pocl of fuel (11.4 g or 5 ml) in a graphite boat,
as shown in Fig. 1. A step along a portion of the length of the bottom
or submerged edge of the blade extended to within 1/16 in. of the floor
of the pool of salt; molten fuel, with a surface tension of about 200
dynes/cm, does not normally penetrate a 1/16-in. crevice without pres-
surization, and thus another device for detecting wetting behavior was
incorporated in the capsule design.
257

UNCLASSIFIED
ORNL-LR-DWG 56754

METAL LOW PERMEABILITY
SPECIMENS GRAPHITE

SALT FUEL GRAPHITE BODY
0 Ya Vo 34 1

Fige 1. MSRE Graphite-Fuel Capsule Test, ORNL-MTR-47-3.

There was room in the irradiated assembly for four capsules hcelding
boats 3 in. long by 1—1/4 in. wide. To provide a contrast in behavior
and to answer questions regarding the effects of penetrated fuel, two of
the boats were prepermeated with fuel, This was accomplished by trans-
ferring fuel into a previously evacuated container holding AGOT graphite
boats and using an overpressure of 60 psig of helium above the submerged
boats. AGOT graphite is sufficiently permeable that this allowed 9 g
of fuel to penetrate the boat (66 g). The fuel was LiF-BeF,-ZrF, -ThF, -
UF, (69.5-23-5-1-1.5 mole %) purified by treatment at 800°C with HF and
H,. The other two boats, as well as all the blades (3/16 in. wide), were
R-0025 graphite, a relatively impermeable grade,

To gain further information from the irradiation, coupons of INOR-8,
pyrolytic graphite, and molybdenum were attached to the blade in a manner
such that the liquid level of the fuel extended half way up the coupons.
This was to provide a contrast between vapor and liquid contact.

The capsules were arranged in the sodium bath so that capsules 3

and 8 were at the top and bottom of a diamond array, and capsules 15 and

16 were at the midplane. The upper capsule ran hotter than the lower one
T because of thermal convection in the sodium. Capsules with preimpregnated
fuel in the graphite boats (3 and &) ran hotter than the capsules with R-
0025 graphite boats (15 and 16)., The resulting temperatures, as given in
Table 1, were relatively high in comparison with MSRE operating tempera-

tures.
258

Table 1. Time-Averaged Operating Temperatures and Identification of Irradiated Capsules

Average Operating Temperature (°C)

Capsule ., a
No. Graphite Pretreatment Fugl, Blade Boatb Thermocouple
Maximum Fuel Fuel in Graphite
(estimated) (estimated) (estimated) P
15, 16 R-0025  2000°C in vacuum 835 790 730 - 710
8 AGOT Preimpregnated with 850 810 750 730
' 9 g of fuel
3 AGOT Preimpregnated with 945 900 850 825
9 g of fuel

-~ 8R_0025 graphite is relatively impervious compared tc AGOT.

bInterface temperatures; no allowance for films at the fuel interface was made.

Results of Experiment

’

After irradiation was completed, on July 27, 1961, the in-pile as-
sembly was dismantled. The individual capsules were punctured by a her-
metically sealed drilling rig so that the released cover gas could be
collected by a Toeppler pump for analyses.

Although the main results of the experiment were that no change in
wetting behavior had occurred, the most surprising results were found in
the gas analyses shown in Table 2. As could be later surmised, in light
of subsequent experiments, the CF, that was found came from F, evolved
from radiation damage- to the fuel at room temperature (during intermediate
shutdown periods of the MIR). The much lower yield of CF, from capsules
3 and 8 as contrasted with 15 and 16 is believed to be a consequence of
higher F, yield in capsules 15 and 16. This is consistent with the fact
that the expected amount of xenon is found in capsules 3 and 8 but is

/
Table 2. O0ff-Gas Compositions

(averaged values from mass spectrometric results obtained at ORNL and BMI)

Capsule Stizgard Components (vol %)
No. sampled®  Air” He CF, Xe Kr 0, Ar®
15 10 5.8, 80.6 9.85 0.012 1.40  0.04  2.29
16 19 22,75 61.45 8.73 0.003 1.14  4.73  1.62
8 7 5.36  76.9 2.32 11.82 1.96  0.11  4.20
3 17 4,57 79.6 0.67 11.43  1.78  0.10 0.9
Gontrol® 10 7.38  87.0  <0.0003 0.17  5.54

aA.pparent volume of gas transferred from gas-release chamber.

bMainly inleakage to the evacuated release chamber while drilling the capsules.

cArgon was supplied as a blanket gas for the welding arc used in a helium-
atmosphere glove box to seal the capsules.

dThe control was heated through cyecles roughly corresponding to the thermo-
couple reading vs time for capsules 15 and 16. )
LI
kX

259

missing from capsules 15 and 16. A lack of xenon is now associated with
the presence of sufficient F; to form the relatively nonvolatile xenon
fluorides, although, indeed, the existence of such compounds was not dis-
covered until after the low xXenon yields had remained a perplexing problem
for months. Needless to say, the formation of CF, had also been quite
perplexing until it was realized that F, which had evolved at room tem-
perature had been responsible, The difference in the F, yields in the
two types of capsules has not been explained but is thought to be related
to the freezing history of the crystals. Slower freezing, according to
one hypothesis, produces crystals that are both larger and less strained
and, consequently, less vulnerable to radiation damage. This could con-
ceivably account for the low F, yields in capsules 3 and 8, which con-
tained prepermeated graphite and could have cooled more slowly than
capsules 15 and 16.

There was sufficient potential cause for alarm in the results of the
gas analyses that the gas behavior tended to eclipse the initial purpose
of the tests and the important favorable results in regard to the persist-
ence of nonwetting behavior. The meniscus of the irradiated fuel was the
same as that of the control specimens and showed definitely nonwetting be-
havior toward graphite; the metal coupons were wetted. Fuel was not found
in the 1/16-in. clearance between the blade and boat but did fill the region
under the blade, where the clearance was 1/8 in. The nonwetting behavior
was confirmed by metallography of the graphlte, which showed no evidence of
penetration by salt. :

Dimensions of the graphite parts did not change within the precision
that measurements in the hot cell were made; changes greater than 0.1%
should have been detectable, Weight changes were not very meaningful
since they represented the combined effect of broken or lost graphite, of
distilled salt condensed in the pores, and of some sticking salt that was
difficult to remove. The information from weight changes was at least
indicative of no pronounced permeation of the graphite by fuel.

The electrical resistance of the graphite approximately doubled as a
result of the irradiation; such increases have been attributed to trapping
conductance electrons in defects induced by radiation.

In looking for evidence of damage to the surface of the graphite,
particular attention was paid to hardness tests. Rockwell hardness meas-
urements on the R-0025 graphite increased from about 94 to 108 as a result
of the exposure. The hardness of the AGOT graphite boats increased from
about 56 to about 80 as a result of preimpregnation with salt, and the
readings after radiation were not significantly higher. Most importantly,
no differences were found between regions under the salt pool and chambers
in the same boat; this was indicative that no measurable sloughing of the
surface was occurring in response to fission fragment damage.

Autoradiography and gamma-ray spectrometry were used to study the dis-
tribution of activity in the capsule; no consistent patterns of unusual
interest were found. In general, it was noted that the prinecipal activity
was Zr-°°Nb and that 103-106gy tended to plate out at fuel interfaces.
260

Visually and photographically, the INOR-8 coupons initially appeared
undamaged, although weight changes and metallography showed that the ex-
treme conditions encountered by coupons had indeed caused appreciable
amounts of corrosion. The extent to which the effects of F,; are shown in
the results is difficult to assess but is probably large. The INOR-8
specimens lost about 1.5 mils in thickness and were carburized to a further
depth of 1 to 2 mils., Molybdenum specimens from capsules 3 and -8 (low
fluorine yield) appeared visibly much less corroded than those from cap-
sules 15 and 16, in which the submerged molybdenum edges were corroded to
resemble a knife blade. The molybdenum sample from capsule 15 lost 39
mg, or 15%, of its weight, and the thickness of molybdenum from capsule
16, originally 20 mils, varied from 7 5 to 13 mils.

Three of the pyrolytic—carbon samples showed no abnormalities in
the metallographs cr in dimensions, with the exception of a thin, un-
identified black film on the specimen from capsule 16. The sample from
capsule 15, however, showed a thickness decrease of 7 to 12 mils, and
one spot of the metallographic cross section showed a leafing open of
several layers of pyrolytic carbon at the surface. The rest (more than
95%) of the section appeared normal metallographically, with the usual
incidence of cracks parallel to the carbon layers. The observed damage
in this carbon may be related to the high-yield CF, in the cover gas from
this capsule; however, the possibility of mechanical damage during dis-
mantling may not be excluded.

In general, the coupons underwent worse corrcsion than was encountered
in other capsule parts of this or subsequent experiments that presumably
involved comparable amounts of F,; the extreme temperatures were the most
obvious difference and must therefore have been primarily responsible for
the accentuated corrcsion.

Experiment 47-4

At the time the CF, was encountered during the postirradiation ex-
amination of Experiment 47-3, the origin of the CF, was not known except
that it was coming from graphite and fuel. The F, that undoubtedliy ac-
companied the CF, escaped detection in the gas train used for the col-
lection and analysis of gas; the F, reacted before it could be detected.
Since fuel and graphite were not thermodynamically expected to react,
the origin of the CF, was very perplexing. It was assumed that CF,
formed at the fuel-graphite interface could somehow make its way into
the voids in the graphite and there be preserved from reacting with the
fuel in the manner that would have been expected .for thermodynamic equi-
librium. The configuration in the capsules of Experiment 47-3, which
was a pool of fuel in a graphite boat, would have favored isclation of
the CF,. Accordingly, a new experiment was designed, in which central
graphite cores were submerged in vertical cylinders of fuel. This con-
figuration was pictured as perhaps accumulating CF, in the voids in
graphite; but because of the submersion of the graphite, the CF, could
not make its way to the capsule vapor space without a good opportunity
to react with the fuel.
el

261

Description of Experiment

The 47-4 assembly contained six capsules constructed of 1/16-in.-
walled INOR-8 tubing. Each of four large capsules, Fig. 2, was 1 in. in
diameter, 2.25 in. tall, and contained a CGB graphite core (1/2 in. in
diameter by 1 in. long) submerged approximately 0.3 in. in about 25 g of
fuel, The fuel generated from 67 to 117 W/cmB, depending upon capsule
location. The temperature at the graphite-fuel interface was 730°C.

Two smaller capsules of the type shown in Fig. 3 were designed to
study the effect of power density and temperature; they contained fuel
to operate at 43 and 260 w/cm3 at 720 and 900°C respectively; and, be-
cause they had graphite liners that extended into the gas space, they
vere thought to provide a favorable configuration for the accumulation of
CF,. The experiment was irradiated in the MIR from March 12 to June 4,
1962, again in a molten-sodium heat transfer medium.

The large capsules were filled by a transfer of molten fuel of the
nominal composition LiF-BeF,-ZrF,-ThF,-UF, (70-23.3-5-1-0.7 mole %). The
small capsules were loaded with molded ingots of about 10 g of fuel, and
in one of them a UF, concentration of 1.4 mole % was used.

Exposure data for assembly 47-4 are summarized in Table 3.

UNCLASSIFIED
ORNL-LR-DWG 67715

Q'rNICKEL POSITIONING LUG
<

UNCLASSIFIED
ORNL-LR-DWG 67744R

2 / 4
Cr—Al THERMOCOUPLE NICKEL THERMOGOUPLE Aj‘
i , WELL :

NICKEL POSITIONING LUGS {2}

.- A
NICKEL FILL LINE ( D K/NICKEL VENT LINE
. ~
- ./u /
¥ lw—PUNCTURE AREA FOR
: > ,_/ A INCR -8 CAP ? 4 HELIUM COVER GAS,| I GAS SAMPLING
_f . . {24cm3) .
7 Sy Y
/% //Jr’ o, [
HELIUM COVER GAS ! [ s A |~ MOLTEN-SALT FUEL
{3.5cm3) el }" _ (10 g)
/] | N<—mor-8 can / y
PUNCTURE AREA ; - - :
FOR GAS SAMPLING P == HELIUM GAS cap—T7 A/CSF?UETBALPEH”E
N it : s y
— KXXe
oo MOLTEN SALT FUEL g v
N K205 N (25q)
[605¢%
—- [R5 N 4 V]
eleds
I %% % B N
[65¢%
- B i N ——cr—ai
oSe K| THERMOCOUPLE
- R - N /1
K
K — INOR- 6 CAN
K5 - i

CGB GRAPHITE
7/ 12.1 cm? INTERFACE

N 2
-
e
O
>
0 XN
0. 0‘0.

FLRLLL,
Y X

INOR—8 CENTERING PIN

NICKEL POSITIONING LUG NICKEL POSITIONING LUG

Fig. 2. Experiment MTR-47-4, Fig. 3. Experiment MTR-47-4,
Large Capsule. Small Capsule.
Table 3.

Exposure Data for Capsules in Assembly 47-4

Thermal.-Neutron

Fast-Neutron Temperature (°C)

Capsule Uranium Flux® Based on (>3-Mev) Flux Power Calculated
P Content 60 . . Based on °8Co Graphite-to-Salt INOR-8-to-Salt Density Burnup
No. Co Activation . . c 3 235

(mole %) (neutrons om—? sec—l) Activation Interface Interface (w/cm?) (% U)
(neutrons cm™ sec™t)
x 1013 x 1012

Large
12 0.7 2.10 2.1 680 610 67 5.5
24 0.7 2.70 3.2 . 760 605 83 7.0
36 0.7 2,71 3.3 ' 710 595 85 7.0
45 0.7 3.85 3.8 710 - 600 117 9.7

Central Reglon
(°c)

Small
4 1.47 | 4,79 5.2 895 260 11,4
1.3 715 43 1.3

6 0.7 1.31

aAverage external neutron flux.

Estimated temperatures.

-

cThermocou.ple readings prior to termination of final irradiation cycle,

coc
Jes 2
L0

263

Results of Experiment

After 1553 hr of exposure between March 12 and June 4, 1962, the ex-
periment was returned for disassembly in ORNL hot cells. The capsules
again were punctured to obtain gas for analysis. The gas was delivered
into a calibrated collection system that had been evacuated and checked
for leaks. A Toeppler pump was intended for use in transferring gases
collected at low pressures. The analyses were made by mass spectrometry.

Capsule 6, one of the crucible types, which had received the minimum
exposure in the assembly, was the first to be punctured; its cover gas
was found to contain 4.2 vol %, or 0.07 cm®, of CF,;, which did not seem
surprising in view of the fact that CF, had been found in the 47-3 series.

The next capsule to be punctured was No. 24, and considerable amaze-
ment prevailed when the cover gas was found to be predominantly F, gas.
The I, was first noticed when the mercury in the Toeppler pump was black-
ened by the transferred gas.

It was necessary to rebuild the gas-collection system of nickel and
to pretreat the nickel with fluorine in order to provide the capability
of measuring fluorine. When this was done, the gas from the remainder
of the capsules was analyzed; the results are shown in Table 4. In the
extreme case (capsule 45), about 3.5% of the total fluoride in the capsule
was converted to F, or CF, in the gas phase; the resulting pressures were
50 atm of F, and a tenth that much CF,. Fortunately, these figures were ,
high enough that it appeared very implausible that the F, could have been .
present at operating temperatures. All the evidence combined to support
the suggestion that the fluorine was evolved from low-temperature radia-
tion damage to the frozen fuel rather than high-temperature radiolysis
of molten fissioning fuel. The CF, could then be explained as being de-
rived from the subsequent interaction of the fluorine with graphite (per-
haps in the brief period of temperature rise after the MIR shutdown was
concluded). Two other factors were associated with the presence of Fs.
In the Tirst place, whenever F, was found, xenon was absent. This was
recognized to be a consequence of the then newly discovered compounds of
xenon and fluorine. In the second place, a prominent radiocactivity due
to 129Te accompanied the ¥, and hampered studies of the gas. This af-
forded a contrast between the mildly reducing conditions anticipated at
operating temperatures where no 129qe activity was expected and conditions
such that F, was being produced, in which case 1?%Te activity was volatil-
ized from the salt as TeF, or TeFg.

At these temperatures, fluorine at any appreciable pressure should
have reacted violently with the graphite and noticeably with the INOR-8.
A longitudinal section through capsule 24 (Fig. 4) did not reveal any
visual evidence of corrosion of metal or damage to graphite. A metal-
lographic examination of a section of the wall of this capsule (Fig. 5)
showed no evidence of damage.

Assuming that half of the decay energy released during the 95-day
cooling period was absorbed in the fuel, a G value of 0,035 (molecules
Table 4. Data on F, and CF, Produced Following an Exposure of Experimental Assembly
ORNL-MTR-47-4 to an Integrated Flux of 1029 neutrons/cm?

Capsule F;iiign Krypton | Xenon Mii%i;q?ivgiznts Decgy ?otal F Total F
Tdentification Density Found . Found . per:Millimole of Period  In Gas Converted
(v/cm3) (% of theoretical) (% of theoretical) U Burned (dayg) (meq) to CF, (%)
Small ca.psulesb _
6 43 145 140 0.2 63 0.012 100
4 260 85 0 ‘ 23.4 137 8.6 8.5
Large ca.psulesb
24 83 140 0 28.0 71 8.2° 31.4
36 85 125% 0 59,6 95  17.2 18.2
45 117 65 0 6l.5 134 25.3 15.5
12 67 130 0 72.9 141 16.7° 57.2

aCalculated from cobalt activity in stainless steel dosimeter wire,
bCapsules listed in order of sampling.
cFlOurine analyses were corrected to include fluorine reaction products formed during sampling.

dAn average'%ased on Kr-to-He and Kr-to-CF, ratios in three confirmatory samples; direct analyses from the first
samples were higher by a factor of 2.

79¢
265

B UNCLASSIFIED
R11213

Fig. 4. Cross Section of Capsule 24 Embedded in Resin.

UNCLASSIFIED
R-11624

T
100X

Fig. 5. Longitudinal Section at Bottom of Wall of Capsule 24, MTR-
4T=be

of F, per 100 ev of energy absorbed) would have been required to produce
the fluorine found in capsule 36. However, a wide and puzzling variation
in the amount of F, production from different capsules was apparent.

Subsequently, all the capsules were opened for examination. Although
the fuel was darkened by radiation damage to the point that it appeared
black, the fuel showed no unusual tendency to crumble or shatter. One
bit of curious behavior, not found in the out-of-pile controls, was the
occurrence of bubbles in the fuel. The bubbles were attached to the metal
at the bottom hemispherical section of the capsules and appeared to be
associated with nonwetting of the metal by fuel. The bubbles are thought
to have been formed in response to the difference in helium solubility
along the temperature gradient in the capsule.

Metallographs of the metal and of the graphite were prepared. There
was no evidence of attack on the INOR-8 and, with one exception, the same
was true of the graphite. Capsule 4 contained graphite in the form of a
liner. The portion of the liner that projected above the liquid level
appeared to have been attacked by fluorine.

//’é—“w’-\

267

Small samples of salt from the irradiated capsules have been examined
carefully with the optical microscope to determine the nature of the crys-
talline material., These materials are, in general, and as is .frequently
the case for rapidly cooled specimens, too fine grained to be positively
identified. However, a sample from near the graphite at the center of
capsule 24 has been shown to contain the normally expected phases, though
some are discolored and blackened; these phases included 7LiF+6ThF,, which
was shown to contain UF, (green in color) in solid solution. It.seems
certain, therefore, that the uranium was not markedly reduced at the time
the fuel was frozen.

An attempt was made, with capsule 35, to remove the salt by melting
under an inert atmosphere to provide representative samples for chemical
analysis and, especially, for a determination of the reducing power of
the fluorine-deficient salt., This melting operation was not successful
since only about one-half of the salt was readily removed in this way.
That portion which melted and flowed from the capsule was green in color
and obviously contained some tetravelent uranium. Since the salt was
highly reduced — 2% of the fluorine, approximately equivalent to that as
UF,, had been lost — melting of the material would be expected to result
in deposition of some metal; this metal may have resulted in a high-melting
heel of uncertain composition.

A variety of chemical analyses were performed on materials from these
capsules. No evidence of unusual corrosion behavior was observed in any
of the analyses.

Salt specimens have been examined by gamma-ray spectrometry to eval-
uate gross behavior of fission product elements., In general, the ruthenium
activity appears to be concentrated near the bottom of the capsules; this
may indicate its existence largely in the metallic state so that a gravity
separation could occur. Cerium as well as zirconium-niobium activity ap-
pears, as expected, to be fairly evenly distributed-in the fuel. §Some
cesium, with some zirconium-niobium activity, is found on metal surfaces
exposed to the gas phase, :

After the graphite cores were removed from capsules 45 and 12 and
examined, they were stored in screw-cap plastic bottles for 5 months. The
necessary manipulation involved many hours of exposure to hot-cell air.

On removal from storage, the graphite was heated to 1000°C under vacuum,
and the evolved gases were collected for analysis by mass spectrometry.
Xenon was recovered in an amount of 7.5% of the originally anticipated
yield on the basis of the burnup. '

The xenon fluorides should not have remained as such in the graphite
during the long exposure to air because of both their sublimation pressure
and their reactivity to form the nonvolatile XeO3. The XeO3; molecule has
a dipole moment and should adsorb on graphite much more readily than the
symmetrical XeF,; however, the dipole moment of XeO3 is probably not large
because of the compensating effect of the unshared electrons on the xenon
atom at the apex of a triangular pyramid with three oxygens at the base
at Xe-0 distances of 1.76 A and 0-Xe-O angles of 103°. Oxygen that might
268

have resulted from decomposition of XeO; during heating was apparently
converted to CO.

- Large pieces of sample (0.8 g or more), for which.the ratio of mass-
to-surface area originally exposed to the salt was similar to that for
the entire core of the MSRE, were analyzed chemically to produce the data
shown in Table 5. These data show that of all constituents of the fuel,
uranium, lithium, and molybdenum were selectively concentrated in the
graphite; analyses from capsule 36, whose fuel was removed by melting,
were not notably different from the others. The mechanism by which dep-
osition in the graphite occurred is not known; but subsequent examinations,
including metallography, agreed with the chemical analyses that it was not
by penetration as fuel. Tests by chemical leaching and by metallographic
examination showed that there was no deposition of uranium in or on the .
metal (INOR-8) capsule walls.

Circular cross sections of the graphite cores were activated in the -
ORNL Graphite Reactor at a flux of 5 x 10! neutrons cm™ sec™ for & hr.
Contact autoradiography before and after the activation showed that the
uranium was in the outer rim, and gamma spectrometry provided a measure -
of the amount present. The results by activation analyses are compared
with those by chemical analyses in Table 6. The chemical analyses sug-
gest a correlation of deposited uranium with power level {(or perhaps with
the consequent rate of F, generation after irradiation). Analysis by
neutron activation, which should probably be preferred because of the
simplicity and directness of the technique, does not support such a cor-
relation. Both analyses, however, show the deposition to be substantialj
if the graphite contains 1 mg of uranium per gram, it has taken up about
0.5% of all uranium in the capsule.

Contact autoradiography showed a high concentration of radioactivity
from fission products (presumably mostly from daughters of gaseous prod-
ucts) to a depth of about 0.2 mm in the periphery of the graphite cross
sections. Neutron activation did not appreciably affect this radiocac-
tivity; this indicated that the uranium was in the same region as the -
fission products.

A closer resolution of the uranium distribution was obtained from
x radiography by a technique (recently developed in the ORNI, Metals and
Ceramics Division) in which the interference from fission product activity
is negligible. The main deposit of uranium penetrated the graphite to a
depth of less than 0.05 mm; however, there were also fringed proliferations
that extended perhaps three times as deep. Facilities for attempted iden- _
tification by x-ray diffraction are nearing completion, but nothing is -
yet known of the chemical form of the uranium in the graphite. ‘

It seems clear that uranium is deposited in the outer layers of the
graphite but not on or in the metal wall and that uranium, lithium, molyb-
denum, and other metallic constituents of INOR-8 are deposited in the
graphite in preference to beryllium, zirconium, and thorium. Mechanisms,
such as oxidation to UFg or reduction to UF; and to uranium and formation
of UC,, for which the situations resulting from F, release might provide
Table 5. Chemical Analyses on Samples from ORNL-MIR-47-4 Graphite-Fuel Compatibility Test

Composition (wt %)

Li Zr Be Th Ni Cr Fe Mn Mo
Original fuel 3.89 11.6 10.1 4.8 5.8 0.0025 0.0024 0.0096
Irradiated fuel
Capsule 12 3.36 <0.018 <0,004 <0.004 <0,001 <0.013
4,08
Capsule 45 3.96 0,016 <0,007 <0, 007 <0.007 <0.025
3.81
Graphite
Capsule 12  0.136 0.04 <0.017 0.0008 D™ 0.013 0.014 o) 0.019 0.062
Capsule 24 0.214 0.052 0.0054 0.003 0.015 <0,012 <0.015 <0,005 0.089
Capsule 36 0.13 0,06 0.004 0.0003 <0,03 0.01 <0.006 0.04
Capsule 45 0.523 0.062  <0.016 0.002 > 0.01 0.04 N> 0.01 0.062
Capsule 3Ab 0.00088 <0.001 0.00003 0,012 0.004 0.067 <0,014 <0,035

“ND, not detected.

bUnirradiated control capsule,

69cC
270

Table 6. Uranium in ORWL-MTR-47-4 Graphite

Power Chemical Activation

Cafisule Density Liizfiign Analysis Analysis
o, (v/cn?) (mg of U per g)  (mg of U per g)

12 6'7 D 1.3

o4 83 B 0.6
. 1.5
D 2.1
C 104
D 5.2
B 1.2
D - 1.3

aNominal power density in fuel surrcunding gfaphite.'

bRelative locations along core, top to bottom, are designated A-D.

appropriate conditions can be imagined. None of these mechanisms, how-
ever, predicts in detail the observed behavior, and they all appear even
less applicable at high (reactor-operating) temperatures where no off-
equilibrium behavior has been demonstrated., Additional experiments were
necessary to decide whether these deposition phenomena can be a problem
in the MSRE or are an artifact of our present experimental procedures.

Conclusions from MIR-47-4

Of the six capsules in assembly 47-4, five were found to contain
large quantities of ¥; and CF, while the sixth, which had received a
considerably smaller radiation dose, showed a small quantity of CF, but
no Fpy. This latter capsule alone yielded the expected quantity of xenon
while all others retained this element nearly quantitatively. The quan-
Tity of F~ released from the fuel melt generally increased with increase
in uranium fissioned (or any of the consequences of this), but no quan-
titative correlation was found. Neither the quantity of CF, nor the
ratio CF,:F, showed an obvious correlation with fission rate, burnup, or
time of cooling before examination.

The appearance of the opened capsules and all completed examination
of components from these seem to preclude the possibility that the F,
was present for any appreciable period during the high-temperature op-
eration of this assembly., Accordingly, the generation of F, from the
salt at low temperatures during the long cooling period must have oc-
curred.
271

Considerable energy is, of course, available from fission product

'decay to produce radiolytic reactiocns in the frozen salt, and the quantity

of energy increases with burnup of uranium in the fuel salt. Capsules as
small as those can absorb only a small fraction of the gamma energy re-
leased by fission product decay. On the other hand, a substantial frac-
tion (perhaps as much as one-half) of the beta energy would be absorbed.

A yield (G value) of less than 0.04 fluorine atom per 100 ev absorbed
would suffice to produce the fluorine observed in any capsule. Production
of I, with such a low G value seems a reasonable hypothesis, though no ob-
servations of fluorine production from salts on bombardment by beta irra-
diation had been published.

Experiment 47-5

Assembly ORNL-MTR-47-5 was designed, after the observations of CFy
in the gas space in capsules from 47-3 referred to above, to include six
capsules of INOR-8; four of these capsules contained graphite specimens
such that a large variation in salt-graphite interface area and in the
ratio of graphite area to metal area in contact with salt was provided.
The other two capsules, whose behavior is described below, were generally
similar to the large capsules of assembly 47-4. However, these capsules

UNCLASSIFIED
ORNL-LR—DWG 71570R

INOR—8 THERMOCOUPLE WELL

INOR-8
COVER GAS
PURGE LINE

INOR--8
COVER GAS
PURGE LINE

—— —-HELIUM COVER GAS

PUNCTURE AREA
FOR {3 em?)

GAS SAMPLING

INOR-8 L7 s i =
INTERFACE AREA- P -G 1 eSS = | MOLTEN-SALT FUEL
(35 cm?) [ 4 I (25 g}
55 g
—- ’:‘:u
K555
- - —-
INOR-8 CAN——" [}~ - ---
'''' <]

CHROMEL-ALUMEL : CGB GRAPHITE
THERMOCOUPLE - | INTERFACE AREA

{12.5 cm?)

INOR—8 CENTERING PIN

INCH NICKEL POSITIONING LUG

Fig. 6. Assembly MTR-47-5 Purged Capsule,
272

included additional vapor space above the molten salt, and each was pro-
vided with inlet and outlet gas lines extending to manifold systems be-
'yond the reactor. It was possible, accordingly, to sample the cover gas
above the melt during reactor operation as well as during and after re-
actor shutdown, '

The two purged capsules, referred to as 3 and 4 in the following and
illustrated in Fig. 6, were 0.050-in.-thick INOR-8 vertical cylinders, 1
in. in diameter by 2.25 in. long, with hemispherical end caps. FEach cap-
sule contained a core of CGB graphite, 1/2 in. in diameter by 1 in. long,
submerged to a depth of approximately 0.3 in. in about 25 g of salt. The
top cap contained two 1/4-in.-diam purge lines, a 1/8-in.-diam thermocouple
well, and a projection to keep the graphite submerged. The two capsules
differed only in the composition (primarily in the uranium concentration)
of the fuel salt. Table 7 shows the nominal composition of the salt mix-
tures in each capsule.

Table 7. Fuel Composition

Specific gravity at 1200°F: 2.13

Liquidus, °F: 842 :
Thermal conductivity at 1200°F, Btu hr™t ft72 (°F)™' ft: 3.21
Specific heat at 1200°F, Btu 1b™* (°F)~1: 0.455

Composition (mole %)

Component
Capsule 3 Capsule 4
LiF 67.19 : 67.36
BeF», 27.96 27.73
ZrF, 4.5 4426
UF,, 0.34 ) 0.66

Preparation of the salt and handling and filling operations were
very similsr to those described for assembly 47-4 above, with the ex-
ception that the filling was accomplished with frozen ingots of salt
under a helium atmosphere,

The six capsules were, as was the case in previous experiments, sus-
pended in a tank filled with sodium, which served to transfer the thermal
energy from the capsules to the tank wall. Heat was removed from the so-
dium tank through a variable annulus filled with helium to cooling water
flowing in an external jacket. As in past experiments, no auxiliary heat-
ing was provided. The capsule wall attained temperatures of about 600°C
during high-level power operation.

The twin manifolds which comprise the pressure monitoring and gas
collection systems were moderately complex, A schematic diagram of one
273

UNCLASSIFIED
ORNL-LR-DWG 77838

/7

N
il
~—p4—p4 XK XK X DRIER [><] = HELIUM .

—_—
P OO X

Ll |

_’.' B
P XXX TO MTR

- ;,L‘;&;
e }
@ E®)

[ W COUPLING

- |

L |

) ~—— MOLTEN SALT
N

— SUBMERGED GRAPHITE

\9} E

SAMPLE BOTTLES
CARBON TRAPS

xxxxxxx CAPILLARY TUBING
PR PRESSURE RECORDER

Fig. 7. ORNL-MTR-47-5 Gas-Collection System.

of the assemblies is shown as Fig. 7. The equipment was_carefully cali-
brated to ensure that the samples drawn as desired into 200- em? metal
sampling bulbs included a large and known fraction of the gas contained
in the capsule.

Behavior Under JIrradiation

Assembly MTR-47-5 was carried through five MIR cycles during the
4-1./2 months ending January 23, 1963. Operating conditions for this
assembly were varied as desired throughout the test to include condi-
tions anticipated in different regions of the MSRE. Temperature and
power level in capsules 3 and 4 were varied independently over a con-
siderable interval but were, insofar as possible, held constant during
periods represented by accumulation of the gas samples. Fuel tempera-
tures during reactor operation varied from about 190 to 1500°F, with
power levels ranging from 3 to 80 w/em® within the fuel.

Gas samples were taken by isolating the capsule for a known and pre-
determined interval (6 to 96 hr) and then purging the accumulated gases
into the sample bulb. Pressure was carefully monitored (to the nearest
0.1 psi) while the capsule was isolated. Fifty-nine gas samples were
taken from the two capsules during the first four MIR cycles under five
distinct sets of conditions within the capsules. These sets of condi-
tions (and numbers of samples) are: |
. Treactor
. Teactor
reactor

. periods

o~ W

. periods

Partly

214

at full power (40 Mw) with fuel molten (34),

at intermediate power (5 to 20 Mw) with fuel frozen (8),
shut down with the fuel at about 90°F (9), )
spanning fuel melting at reactor startup (4),

spanning freezing of fuel following reactor shutdown (4).

because the amounts of CF, produced were generally too low

to measure accurately, there was much scatter in the results on any of
the bases on which correlations were attempted. The most enlightening
information was obtained from plots such as Figs. 8 and 9, wherein the

Fig.

UNCLASSIFIED
ORNL-DWG 63-662

1 ! 1 1 I ]
2 PALANCE WITH FLUCRIDE EXCESS FROM FISSIONING |

05 o Ch/ RATIO BY MASS =
SPECTROMETRY _
0.2 ® ESTIMATED LIMIT OF DETECTION| —|

FOR NEGATIVE SAMPLE
{AS RATIO)

1

C'Z/XG iN GAS SAMPLE

]
0.005 ——ANTICIPATED SENSITIVITY OF ANALYSES

0 0.01 0.02 003 004 005 006 007
Xe IN SAMPLE (%)

8. Production of CF, from Fuel and Graphite at MSRE Power

Densities (Although Slightly Lower Temperatures).

Fige 9. Production of CF, from Fuel and Graphite at MSRE Tempera-

UNCLASSIFIED
ORNL-DWG 63-664

» | BALANCE WITH FLUCRIDE EXCESS FROM FISSIONING _
. ' 7 :

)
4
0.5 o Cfa/y. RATIO BY MASS —
SPECTROMETRY )
0.2 ® ESTIMATED LIMIT OF DETECTION| |
FOR NEGATIVE SAMPLES
04 |— fo (AS RATIO) _

“fa/xe IN GAS SAMPLE

0.005
I [

0002 ANTICIPATED SENSITW
OF ANALYSES

0.001 | | |

0.02 0.04 0.06 0.08 Q.40 Q.42 0.44
Xe IN SAMPLE (%)

tures (Although Slightly Higher Power Densities).
275

amount of xenon (long lived) in a sample served as an internal dosimeter
and measured the exposure in terms of the relative amounts of fission
energy that were released while the samples accumulated. The amounts of
CF, in a sample were presented as ratios of CF, to xenon in the same sample
for the following reasons: (1) the primary generation of CF, is probably
strongly influenced by the energy, as is of course certainly true of xenon;
(2) compariscn of mass spectrometric peaks on the same sample eliminates
many possible discrepancies that might stem from unrecognized differences
in sampling; and (3) also a convenient reference factor is available for
assessing the chemical effects on fuels of any given loss of fluoride as
CF,. The excess of fluoride ions arising from fissioning UF,, because

the fission product cation valences for equilibrium with the container

do not accommodate the available anions from the fissioned UF,, gives a
small increase in oxidizing power of the fuel with increased burnup. Or-
dinarily, this oxidation should in principle be manifested as a relatively
insignificant increase in corrosion product concentration., However, if
the moles of CF, evolved from a fuel were approximately equal to the gram-
atoms of stable xenon produced during the same interval, the oxidation-
reduction level would tend to remain favorably balanced. The exact ratio
for balance varies with conditions of operation but for purposes of esti-
mation can be considered as unity or a little higher. Thus, the cross-
hatched bands at the top of Figs. 8 and 9 denote the upper limit of the
CF4/Xe ratios below which the removal of CF, is a corrosion problem of
negligible proportions for the MSRE. Higher rates of fluoride removal
could conceivably require occasional re-treatment of the fuel with HF,

but such a chemical tolerance limit (due to the approach to sufficient
reduction to deposit uranium) would be encountered only after an appreci-
able reduction of UF, to UFj.

Estimated limits of sensitivity for each of the samples that gave
negative results have been plotted as filled circles in Figs. & and 9
for use as upper limits. The shaded area shows the locus of points cor-
responding to about 5 ppm CF,, and the open circles correspond to samples
for which definite CF, contents were found. Taking into account both posi-
tive and negative samples, if CF, were present in amounts proportional to
xenon production, the quantity was too small to be measured with certainty.
On the other hand, there was a pessibility that CF, underwent decomposition
and/or reaction with the wall because of the radiation field in the capsule;
thus the amount present in samples collected over long periods could repre-
sent a steady state between formation and disappearance. The straight
dashed lines in Figs. 8 and 9 represent trial extrapolations to zero time,
effectively at least, and thus show a possible relative primary rate of
generation of CF, compared to xenon in the vapor space in the capsule, In
principle, a curve for CF, concentration should pass through all open
circles and beneath all filled ones; but, because of the scatter of the
data, the curves remained indeterminate. However, the trial values ob-
tained from the dashed extrapolations were considered representative of
the rate of production of CF, that could have prevailed. Thus the con-
clusion would be reached that CF, was produced in the vapor space at about
6% of the rate of xenon and that a steady-state concentration at, or Jjust
under, 5 ppm CF, was approached. A simple basis, though not a rigorous
or necessarily correct one, for scaling such results from capsules to a
276

reactor could be the assumption that for given power density in the fuel
the CF, production was proportional to the area of the interface between
graphite and fuel. Then, since xenon production varies with the fuel
volume in the core and the MSRE core contains 7 x 10° em® of fuel in con-
tact with about 10® cm? of graphite, compared to the capsules with 10 cm?®
of fuel in contact with 12 cm® of graphite, the scaling factor would be

10° cm? " 10 cm®
12 em= 7 7 x 10° cm?

or roughly unity. Thus, to the extent that the assumptions involved in
the foregoing consideration are justified, the loss of CF, from the MSRE
is not expected to cause difficulty..

It was during the reactor shutdowns, which periodically interrupted
the collection of gas at operating conditions, that the true nature of
the evolution of F, came to light. At shutdowns the capsules were promptly
cooled to about 40°C, and, after variable induction periods, an accumula-
tion of gas in the capsules was noted by pressure measurements. The ac-
cumulating gas was found to be F, and to be released from the frozen fuel

at GF values (molecules of F, per 100 ev of absorbed energy) from about
2

0.005 to 0.031, although on occasion the induction period persisted through-
out the shutdown. ©No correlation between capsule power density Jjust before

shutdown and either GF values or duration of induction periods was ob-
2

gserved. There were seven shutdowns of sufficient duration for the effects
to be noted.

Short periods of low-power operation (4 hr) were arranged during MTR
startups in order to observe the behavior of the solid fuel at wvaried
levels of fissioning which covered the range from 2 to 10 w/cm3 and tem-
peratures from 85 to 325°C. Except for some suspicious indications at
the lower temperatures, no accumulation of gas was detectable; this indi-
cated, as was later confirmed, that the susceptibility of the solid fuel
to radiolytic decomposition diminishes greatly or disappears at relatively
low annealing temperatures (e.g., 200°C). An accelerated rate of recombi-
nation with increasing temperatures was found in subsequent postirradiation
studies, and this is undoubtedly related to the pronounced and favorable
temperature effect. At any rate, the operating ranges in which frozen
salt in freeze valves and freeze flanges can be maintained without evolving
F, appear to be ample for practical purposes.

Evolution of F, During Postirradiation Decay

On January 23, 1963, after 2000 hr of irradiation at an average
thermal flux of 2 x 10'? neutrons em™ sec™, the assembly was withdrawn
from the reactor; 11 days later the gas evolution from the two capsules
was being monitored in an ORNL hot cell. The F, evolved into an evacuated
gas space in which the rate of accumulation was followed by pressure meas-
urements precise to 0.0l psia or 0.1l std em?® of F>. The gas was occasion-
‘ally collected and analyzed; this restored the vacuum and provided a meas-
ure of the leak rate, which was generally quite small. The loss of F, for
T

277

the first 95 days of the cooling period is shown in Fig. 10, where the
loss is expressed in terms of the percent of the fluoride ions removed
from the fuel.

The effect of temperature was found to be quite pronocunced, as shown
in Fig. 11, which is a differential form of the data in Fig. 10. The cal-
culated yield curve in Fig., 11, which approximates the initial ambient-
temperature fluorine-generation curve for the higher-power capsule, rep-

resents a GF value of 0.02. A revision of this value may be necessary
2
when better estimates of capsule power densities are available from uranium

isotopic analyses of the irradiated fuels. The Jjagged rate curve is a
consequence of equilibration periocds at various temperatures, Either cold
(—70°C) or warm (80°C) temperatures sufficed to suppress the evolution.
Intermediate behavior found between these extremes implied that a maximum,
which shifted somewhat with time, occurred near 35°C or higher. Bursts

of Fp occurred immediately after a temperature rise, and several days were
sometimes required for a steady state after a drop in temperature.

Beginning on day 113, the effect of various partial pressures of added
F, gas was explored. Above 50°C there was a pronounced pressure effect in
Tthe expected direction, that is, decreasing loss at increasing pressure,
but since the actual rates seemed to depend markedly on the history of the
fuel (probably on the state of chemical reduction in the surface layers),
it was not possible to determine a quantitative expression for the effect
of pressure. Qualitatively the reaction order with respect to F, pressure
appears to have been initially near unity and to have decreased to 0.5 or
lower. Below 50°C, evolution continued even at 1 atm of F,. The results
are depicted in integral form for the higher-powered capsule in Fig. 12.
Consumption of F,; was clearly evident at increasing rates as the tempera-
ture was increased. As the fuel consumed more fluorine, higher tempera-
tures were required to give further consumption at measurable rates.

These data suggest that the F,; rate of evolution is controlled at
low temperatures by rates of diffusion of radiolytic species within the
solid, and at higher temperatures by a back reaction whose rate is strongly
temperature dependent.

Analyses of Gas from Sealed Capsules

For the first 5 months following the in-pile irradiation, the assembly
could not be dismantled without disturbing the investigation of F, evolu-
tion described above. As soon as the rate of F, evolution in the swept
capsules diminished to the point that it was difficult to measure changes,
the four other sealed capsules in the assembly were punctured and their
cover gases were analyzed.

Two. of the sealed capsules had been designed to provide extremes in
the ratio of graphite to metal areas to which the fuel was exposed. Cap-
sule 1. (Fig. 13) had a graphite liner and a graphite core with a total
salt-graphite interface area of 27 cmz; the liquid extended above the liner
and flooded 7 em? of INOR-8 wall. Capsule 2 (see Fig. 14) had only a small
278

UNCLASSIFIED
CRNL -DWG 63-599R

COOLING TIME (days)

10 20 30 40 50 . 60 70 80 90
> | | I I | | I I |
4 65 w/cm® CAPSULE\ — e B
see IN-PILE
< 3 ] -——IN TRANSIT
5 / ——OQUT-OF-PILE
0 /]
a /
1—1 2 //
W 4 ]
l/ /<—_——
1 V // 35 w/cm® CAPSULE
//
* rd
. -
o :”.u—/
36 50 {86 |3al-70| 3 1.7 29 69 [87 28
TEMPERATURE (°C)
Fig. 10, TLoss of Radiolytic Fluorine from MIR-47-5 Capsules.
UNCLASSIFIED
CRMNL-DWG 63—-598R
14 :
I | I
2 __\ _
10 —
\ CALCULATED YIELD FOR HIGH-POWER
\ CAPSULE WITH 6=0.02
- 2
3 8 — _|
s
[«
LI.N
o & 65 w/cm> ]
E CAPSULE
[a]
54— S M~ —
- T
\\ S e
35 wiem® T
CAPSULE ~~
2 - — ]
\ |
\
N |1
\ |
N\
0 1
| 36°c| 50° | 86°|34°|-70°| 31° | 1.7° 290 | 69 I87° 28°
-5 I | | | | | | I |
10 20 30 40 50 60 70 80 90 100

Fig. 11.

COOLING TIME (days)

Postirradiation Fluorine Release in MIR-47-5 Capsules.

279

UNCLASSIFIED

ORNL-DWG 63-3423
°C

83°C 54 ° 88
a5 33°C | | s88°c | 34°C | 74°C | | 38°C |
B esesssscesettelioge
2 -4.0 "cAPSULE NO. 4 e
o : ®
3 .
E (chm ) .........oooOioo.......
- M LL T PYR esee- IG5
o _3.5 oe0d
2 { N "'--\ —
> FLUORINE PRESSURE, ~~SAMPLED |
= CAPSULE NO.4 a
g 730 F» ADDED : l é
1 S — B
2 _|
w —2.5 {atm O Ll
F» EVACUATED = 4
o NORMAL F, N 2 L a4 a
a ACCUMULATION I
S -2.0 Lo S 0
3 oooooooooooooo0000000000oooooooooooooooooooooooooooooooooooooooo
o CAPSULE NO. 3 {35 w/cm3)
_,1_5 | |
95 105 1s 125 135 145 155
COOLING TIME (days)
Fig. 12, Effect of Pressure and Temperature on Recomposition of

Radiatively Reduced Fuel.

UNCLASSIFIED
ORNL-LR—~DWG T1568R2

NICKEL POSITIONING LUGS (2) INOR=8 THERMOCOUPLE WELL

/— INOR—8 CAP

%

&

L A <9
I
HELIUM COVER GAS — | :
{3cm®) !
N\ Vg || 7 N ™~ inor-s can
PUNCTURE AREA b ‘i
FOR GAS SAMPLING == || BN
S i KX —— MOLTEN-SALT FUEL
INOR—8 R b2 (19 g)
INTERFACE AREA . | E

(7 cm?) Bl N ¢4 N

HELIUM GAS GAP/ < " CHROMEL-ALUMEL
e THERMOCOUPLE

Y W4

GRAPHITE
INTERFACE AREA
(27 cm?)
0 Yo 1 i NICKEL POSITIONING LUG
== b
INCH ' O

13. Capsule Design with Large Area of Graphite-Fuel Interface,
280

UNCLASSIFIED
ORNL—LR—-DWG 7156T7R2

INOR-8 THERMOCOUPLE WELL

NICKEL PCSITIONING LUGS (2) e,
lj ‘
INOR-8 CAP —\\ i ii:

&7
\_""_\

HELIUM COVER GAS
/ (3cmd)

/ INOR—8 CAN

PUNCTURE AREA
FOR GAS SAMPLING

4
: =aA
MOLTEN-SALT FUEL—— XA
{25 g) . INOR-8
o INTERFACE AREA
—__'/ (46 cm?)
CG8 GRAPHITE N o
INTERFACE AREA———"R"~ 1o ALl 4
{1 cm?) - 3

CHROMEL-ALUMEL
THERMOCOUPLE ——

INOR—8 CENTERING PiN

NICKEL POSITIONING LUG

0 Y 1
e ——
INCH

Fig. 14. Capsule Design with Small Area of Graphite-Fuel Interface.

UNCLASSIFIED
ORNL~-LR—-DWG 71569

NICKEL POSITIONING LUG

q PUNCTURE AREA
/FOR GAS SAMPLING

,..w
R
—

<]

2

HELIUM COVER GASA—/"E'
(6 em®)

5%
SRS

o
=

2
a

v.
2
ol
4
&

)
%

X

3
0.‘
o
¥
Q.A

(R
A
®

5%
53
4?
35
XX

3
o,
ot
55
oo
e}
55

Pa$

S
Selels)
255
by
55

.,..,
3
5%

%5
55005

255
9%

3%
CRS0RLRS
KK
S5,
KRR
1

AGOT GRAPHITE
IMPREGNATED WITH
MOLTEN- SALT FUEL/

(Yo g}

<
55

/— INOR—8 CAN

4%%?’ KRR
el teted
XS
S0,

L2

s,
>
&2
35
355
.0
20

7S
S
&
<5
s
2L
e

S
Yalil
>

e

| go—
FUEL RESERVOIR ——— 1

0 l/2 B

' ;

INCH

NICKEL POSITIONING LUG
rd

Fig. 15. Capsule Design for Graphite Impregnated with Fuel.
281

disk of graphite (1 cm? of graphite-fuel interface) and provided 46 cm?e

of metal-fuel interface. The other two capsules (Fig. 15) were of smaller
diameter; they each contained a mounted graphite (AGOT) bar (3 g) that

had been prepermeated with fuel. No other fuel was added; therefore, these
capsules, initially at least, exposed no metallic surface to the salt.

Analyses of the gas drawn from these four capsules are shown in Table
8 along with some details of the in-pile exposures. It is clear that the
three capsule geometries yielded very different quantities of F, and CFy;
capsule 2 yielded about 39 cm® of F, and 70 cm® of CF,, capsule 1 yielded
about 2.5 cm® of F, and 0.6 cm?® of CF,, and the small capsules yielded
none of either gas. The notion that the salt-metal interface area af-
fects radiolytic generation of F, from solid salt now appears absurd.
The possibility that graphite surfaces somehow catalyze back reactions or
otherwise slow the generation process now seems almost equally unlikely.
It appears more plausible to ascribe the pronounced difference in behavior
to large differences in cooling rates of the fuel and to consequent dif-
ferences in crystallite size and degree of stress introduced into the
crystallites, The small capsules, with the gas envelope around the fuel,
cooled most slowly and, perhaps, produced least readily damaged fuel.
Capsule 2 clearly should have cooled most rapidly.

The particle size of the fuel dispersed through the graphite bars
in the rear and front capsules must have been very small, The fact that
no detectable amount of F, was generated in these capsules suggests that
small particle size alone is not a sufficient factor to induce F, gener-
ation.

No evidence of decomposed CF, residues or products, nor even deposits
of a suspicious nature in this regard, was found on visual inspection and
attempted sampling of the interior of gas lines from Experiment 47-5.

Metallographic results from this experiment have not yet been re-
ported, but no evidence of attack on the INOR-8 is expected to be found.
The loss of fluorine from the swept capsules during the in-pile portion
of the test might be expected to contribute toward additional uranium
deposition in the graphite in these capsules.

Experiment 47-6

Description of Experiment

This experiment was designed after low-temperature fluorine genera-
tion had been established and after its significance with respect to CFy
production, metal corrosion, and uranium deposition in graphite had been
considered. Based on these considerations, some questions to which the
experiment was addressed, in order of importance in complicating the
structural array of the assembly and accessories, are as follows:
Table E.

282

ORNL-MIR-47-5 Gas Analyses and Exposure Conditions

Capsule Type

Capsule Designation 1 2 Rear Front
Weight of graphite, g 14,362 0.536 3.6655 3.6353
Weight of fuel, g 18.928 25.240 0.6465 0.5037
U in fuel, mole % 0.659 0.659 0.339 0.659
Weight of 93.26% 23°U, g 0.770 1.027 0.01345 0.0205
Neutron flux,a neutrons em™2 sec™l  1.94 x 103 2,24 x 10%3 1.91 x 10*2 3,87 x 10%3
Fission power density,’ w/cm3 65.0 75.0 32,5° 130.0°
Burnup, " % 7.7 8.8 7.6 14,7
Volume of cover gas, ml (STP) . 5.67 115 2.63 2.76
Fs,” % ‘ 43 34 0 0
CF,, % 10 61 0 0
He + Ar, % 41 2.0 99.6 98.2
kr,t % 2.7 (73%)  0.11 (40%)  0.055 (40%) 0.20 (51%)
xe,’ ¢ <0.,01 0.024 (1.7%) 0.32 (46%)  1.57 (813)

(<0.05%)

Air + COp, % 3.5 2.7 0 0

. ®alculated from 60Co aralysis of capsule metal. ,
bCalculated from the tabulated neutron fluxes and the known fuel densities at 1300°F.

CCalculated for fuel itself; overall power densities in the impregnated graphite are

15% of these.

dCalCulated from the tabulated neutron fluxes.

®Corrected to include 5i¥,, OF,, COF,, etc.

i

fNumbers in parentheses are percentages observed of yields calculated from burnups.
283

1. Can a more sensitive measurement of the production of CF, be obtained?

2. Can the conjectured absence of deposition of uranium in graphite under
operating conditions be more clearly confirmed?

3. Does the concentration of U*% cations in the fuel, through catalytic-
type mechanisms involving the variable valence of the uranium, have a
perceptible effect on the recombination of nonequilibrium oxidizing
and reducing species that are produced by radiation?

4, Was a previously noted augmented corrosion of molybdenum metal (Ex-
periment 47-3) due to attack at operating conditions or was it a con-
sequence of gaseous F, released by radiation damage from decaying fis-
sion products in the frozen fuel?

5. Can-the radiolysis of added CF, be detected?

Foremost was the provision for heating the four irradiated test cap-
sules to avoid freezing during reactor shutdowns. This was to avoid re-
lease of F, and much of the ensuing obfuscation encountered in recent
postirradiation examinations. But this alsc entailed a problem in opening
and dissecting the capsules without allowing F, tc intrude again after
removal from the MTR. The remedy involved a prompt dissection in MIR hot
cells before removal to ORNL hot cells.

Figure 16 shows one of the four test capsfiles in cross section. All
four INOR-8 capsules are vertical right cylinders (~2 in. in height) con-
taining a 1/4-in. annulus of fuel (25 g) that surrounds and covers a
central core of graphite (1/2 in. diam), leaving a gas space of about 3
cm® above the liquid level. Central thermocouple wells penetrated the
graphite. Two capsules were swept by helium, at measured flow rates
variable to 5000 cm?/hr, which passed to a sampling assembly thet in-
cluded a molecular sieve column for ccncentration of CF,. The other
two tapsules were sealed.

The question of CF, evolution was approached by having the two gas-
swept capsules identical in all respects except graphite area (and depth
of submersion). An INOR-8 core section replaced the upper half of one
of the graphite cores and reduced the graphite area from approximately
14 cm® to approximately 7 cm®. Both swept capsules contained 0.9 mole %
UF, which included a sufficient enrichment of 235U to give 14 w/cm’ at
minimum insertion; these values were chosen in conformance with MSRE op-
erating conditions and provided 52 w/cm3 as a maximum.

The gas flow system was designed to sweep any CF, from the gas space
fagster than it could be radiolytically decomposed. Additionally, there
were facilities, including thermal conductivity meters, for measuring the
rate of radiolysis of CF, as prepared mixtures of CF, and helium were
passed through the capsule. Also, sensitive analyses at low concentra-
tions of CF, were available from a "helium breakdown" sensor being de-
veloped by the Analytical Chemistry Division. Previous results implied
a steady-state concentration of CF, so low that the rate of loss of added
CF, might be very much easier to measure than the rate of appearance of
evolved CF, in the capsule gas space.
284

UNCLASSIFIED
ORNL-DWG 64-4010

THERMOCOUPLE

/'/,—PURGE TUSE

vl

e A A N N N NN

N N N e .Y

o
[T > .|
<3 " 8 =
W_w o ) _.|__._ fe3) z
> S ) w g
z2 w J S
<2 woooo 2 S
_
o2 T i o o
o a " « |
[ <l > o [
5 L ? o
2 3 5 g
©
%) = 2 2
s N NN NN NSNS NSNS NN SN NN NSNS NN SN N NNNNNN r

'\

S SRR RO RIS
ooooooooowooonooooowou%on R oooooo %% QQuonon%ooouon wouowoo %
oo .00. N,
9:0:0:0:0:0: 9
.ooooooo oooooo 000\000000 00%00
ogetees 0 00 XXX ooo 00000 oo & 0 o o X 0 RRAHRS
P2 .IQPQ ’P .}"0 258 .90 P” XXX . ’ .9, . XXX ).)

/Illlm\\\\\\\

T,

THERMOCOUPLE

.

L

LLL

7 : /
/////////////W////,////w///////////////////
o o o I
N o X _m (L]
S I L @
x M T o
5w w T
(&)
5 3 2
=z
L4

-

HEATER LEADS

5 Y

GAS CAP PURGE SUPPLY TUBE

Typical Capsule Assembly Equipped for Purging, for Experi-

Fig. 1l6.
ment ORNL-MTR-47-6,
285

The question of uranium in or on the graphite was apptroached by de-
signing flat areas on the graphite cores; this should greatly Tacilitate
examination by x-ray diffraction. Further, the two sealed capsules are
addressed primarily to comparing minimum with large concentrations of UF,
in the fuel; one has 4 mole % UF,, with the maximum enrichment consistent
with safe temperature, while the other contains highly enriched. uranium,

- but in the minimum amount sufficient to dupllcate the power density in

the swept capsules.

On the one hand there is the possibility, if deposition does indeed
occur, that the higher the uranium concentration and the higher the power
density, the more uranium might be deposited on the graphite. On the other
hand, there is a thecry that uranium cations might serve as a recombination
catalyst for off-equilibrium radiolytic "free radicals,” and that the
steady-state concentration of these might be higher with lower concentra-
tions of UF,, and hence conceivably induce greater deposition in low- |
concentration fuels. With the available neutron flux for the experiment,
the lowest concentration of fully enriched UF, consistent with power den-
sities of interest is about 0.5 mcle %. This precludes testing at this
time with quite low concentrations that presumably would be of greatest
interest in checking for catalytic effects and that would represent a
relatively unexplored concentration range, but neither have the higher
concentration ranges been explored. Thus, as far as uranium content is
concerned, the experiment carries one low, two medium (swept), and one
high concentratlon capsule.

Both sealed capsules contain coupons of molybdenum which are expected,
on the basis of corrosion encountered in 47-3, possibly to be more sensi-
tive than INOR-8 as indicators of corrosive influences. Also, the com-
parison of uranium behavior in the presence of pure molybdenum, as con-
trasted with the presence of only 17 at, % of molybdenum in INOR-8 in the
swept capsules, may prove to be of interest.

The question of the corrosion of molybdenum seemed to be one that

-~ could probably be answered without interfering with the other main pur-

poses of the experiment. Admittedly, there was risk of confusion from
superposition of variables, but the advantages were deemed to outweigh
the risks. In the absence of radiation, molybdenum and INOR-8 form a
bimetallic couple, and electrochemical transfer of metal is expected;

in practice the rate is alleged to be so slow that no clear evidence of
transfer has been obtained. In the absence of radiation, molybdenum is
carburized when in contact with graphite; in this experiment the molyb-
denum is in submerged radial sheets or slivers that touch INOR-8 on one
edge and do not touch graphite. Molybdenum and carbon both give rise to
volatile fluorides under oxidizing conditions; this could conceivably be
the basis for a response (different than nickel, e.g.) to the off-equi-
librium species produced by radiation in the fuel. In the absence of
irradiation, molybdenum behaves as rather more noble than nickel toward

fluoride melts.
286

Results of BExperiment

The experiment was exposed to radiation in the MIR from June 8 to
August 10, 1964. During the first two 3-week cycles of the MIR, 26 gas
samples were swept from INOR-8 capsules containing MSRE fuel and graphite.
A variety of sweep rates, temperatures, and power densities were explored,
but no CF, has been found in the samples that have been analyzed thus far,
nor has there been any detectable radioactivity other than that of Xenon
and krypton. This indicates that neither F, nor CF, was produced at rates
that would be of significance to the MSRE. -

For the first 24 hr following the final MTR shutdown, the assembly
was maintained at a temperature above the salt liquidus temperature in
order to prevent fluorine generation during the period of most rapid ra-
diocactive decay of the fission products in the salt. Then the salt was
allowed to freeze at a relatively slow rate (10°C/hr) through the tem-
perature range 450 to 400°C, in which the primary, secondary, and ter-
tiary phases are found to appear. Through this slow freezing it was
hoped that the fuel would be deposited in a relatively good state of
crystallization, permitting petrographic confirmation of the expected
phases and further minimizing fluorine evolution. Having cooled the
salt to a solid, the assembly was removed from the reactor as quickly
as possible and disassembled in the hot cells at the MIR. The capsules
were opened and the components separated to prevent any postirradiation
effects which might obscure the in-pile behavior of the system.

Visual examination of the components during disassembly did not re-
veal any unexpected or disturbing information. The salt was eilther green,
gray-green, or black, as frequently seen in the past. There were small
droplets of salt components on the metal above the liquid phase. The
salt broke cleanly away from the graphite with no evidence that the graph-
ite had been wetted, though the metal was apparently wetted by the salt.
The molybdenum coupons appeared not to have been attacked. In the swept
capsules the INOR-8 surfaces above the liquid appeared to be covered with
a dirty gray film or series of mottled spots similar to the observations
in other experiments; this is currently attributed to the effects of traces
of water or other reactive components in the helium sweep gas. In the un-
swept capsule (No. 2) the salt was green, the graphite looked very clean,
and all metal below the salt was shiny. The thermocouple well was also
shiny. The contact angle of the salt on metal was about 90°, and there
were a few beads of salt above the liguid line.

A1l the components from the experiments have been packed for ship-
ment to ORNL and detailed examination in our hot laboratory facilities.
More satisfactory statements confirming the excellent results suggested
by the gross visual observations will have to await this examination.

It should be recalled that provisions have been made in the MSRE for
the direct exposure of graphite and metal test specimens in the very
center of the reactor. An access port has been provided through which
these can be removed for observation and replaced by additional specimens
as required to confirm and extend the results of the small-scale irradia-
tion test program which has been discussed. '
287

OQut-of-Pile Experiments

Sallient features of the fluorine generation and recombination proc-
esses have been verified with simulated MSRE fuels, and with some of the
component fluorides, on exposure to pure gamma, beta, and x radiation.

An INOR-8 capsule containing MSRE salt and graphite was exposed to
gamma radiation from a ©%Co source for ~7400 hr at temperatures ranging
from 38 to 150°C. The salt received a calculated 0.45 x 1020 ev hr—1 g1,
After an induction pericd of about 600 hr, fluorine gas was generated at
rates (G values) varying between 0.03 to 0.07 molecule of fluorine per
100 ev absorbed at temperatures up to 110°C. At 150°C any fluorine gas
vhich had been generated was recombined. At 130°C recombination rates
roughly equaled generation rates, No evidence of any appreciable re-
action of the F, gas with the graphite to generate CF, was obtained.?

Experiments to investigate fluorine evolution from studies of be-
havior of solid MSRE fuel mixtures under fast electron bombardment have

yielded values of GF as a function of dose up to a total dose of about
2

1.6 x 108 Mev per g of salt. At doses below about 8 x 1017 Mev/g, maxi-
mums in the fluorine production rates were observed or indicated in each
experiment. In one of the experiments exposed to a greater dose, a second
increase in fluorine evolution rate was observed after a total dose of
about 9 to 10 x 1017 Mev/g. It 1s observed that reduction of the dose
rate by a factor of about 2 causes an increase in the initial rates of

fluorine evolution per unit dose (GF ). The maximum Gp_ observed was 0.02.
2 2

Similar experiments with solid LiF showed that fluorine was not evolved
as a result of bombardment with fast electrons but was taken up in small
amounts.?> '

Soft x-ray irradiation of a solid mixture similar to the MSRE fuel
gave evidence of radiolytic decomposition with yields of volatile fluorine
compounds equivalent to GF (moles of F» per 100 ev absorbed) values ranging
2

from 0.0006 to 0.04. 1In some cases elemental fluorine was identified, but,
generally, the products were carbon tetrafluoride and carbonyl fluoride.
Prefluorination of the salt removed carbon-containing impurities and en-
couraged liberation of elemental fluorine., Fine particles liberated vola-
tile fluorine-containing products at about ten times the rate from coarse
particles. The compound 6LiF+BeF,+ZrF,, one of the complex compounds which

crystallize from the MSRE ffiel, decomposed with an equivalent GF of about
2
0.02. Prefluorinated thorium fluoride liberated elemental fluorine at

rates of about 0,005 molecule per 100 ev. Neither lithium nor zirconium
fluorides gave any evidence of radiolysis.4

References

- MSRP Status Report, May 1, 1958, ORNL-2634.
. MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL—3626, pp. 101-5,
MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, pp. 105-10.
. MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, pp. 110-11.

W

288
PREPARATION OF MSEE FUEL, COOLANT, AND FLUSH SALTS

Je. He Shaffer

The initial operation of the Molten-Salt Reactor Experiment (MSRE)
will utilize about 26,560 1b of fused fluorides as fuel, coolant, and
flush salt mixtures. The reactor fuel mixture will have the composition
LiF-BeFy-ZrF,;-UF, (65.0-29.1-5,0-0,9 mole % respectively). Fissionable
2357 will comprise about one-third of the uranium inventory; the balance,
as nonfissionable uranium, has been included for chemical purposes. An
initial loading of about 11,260 1b of fuel salt mixture will be required
for a desired fill volume of approximately 73 2 in the reactor assembly.
The coolant and flush salt mixtures will be chemically identical and will
have the composition LiF-BeF, (66-34 mole %). Approximately 15,300 1b
of this mixture is needed for a coolant salt volume of 42 ft> and a flush
salt volume equivalent to that of the fuel mixture.

Techniques for preparing fused fluoride mixtures at ORNL were de-
veloped to support the Aircraft Nuclear Propulsion Program and were suc-
cessfully demonstrated during the preparation of fluoride mixtures for
the Aircraft Reactor Exr_perim_en.t.1 The production facility has since be-
come an integral part of the Molten-Salt Reactor Program to provide fused
fluoride mixtures for its chemical and engineering test programs and for
other related projects of the Léboratory and the USAEC. Since March 1964,
the facility has been operated at capacity for the production of MSEE
fluoride mixtures. Sufficient quantities of LiF-Bels (66-34 mole %) for
the coolant salt mixture and about 50% of that required as the flush salt
have been prepared to date. Production of the fuel salt mixture will
follow the loading of coolant and flush salt into the reactor facility.

Process Chemistry

Although starting materials of very high purity are used in produc-
tion of fused fluoride mixtures, further purification is needed before
they are acceptable for use in the MSRE, The removal of a limited number
of impurity species during production operations is achieved by treatment
of the fluoride melts with anhydrous HF, hydrogen, and, in some instances,
by the use of beryllium metal as a strong reducing agent. Impurities
which can be volatilized are removed in the process gas effluent stream;
those which can be rendered as insoluble particles are removed by filtra-
tion.

Oxide Removal

On heating the starting materials above the liquidus temperature of
the salt mixture, oxides are formed by pyrohydrolysis of the fluoride
salts with their absorbed water. Although oxide impurities, in them-
selves, are probably not detrimental, their presence in the fluoride mix-
tures of the MSRE could result in the deposition of solid oxide particles
or scale. The resulting heterogeneous system could alter heat transfer
properties of the reactor components and might also create localized heat
sources in the reactor core by the deposition of uranium dioxide.

[ 4
289

Oxides are removed during the initial gas sparging of the fluoride
melt at 600°C with an anhydrous mixture of HF in hydrogen. They react
directly with hydrogen fluoride by the reaction '

0= + ZHF = 2F + H,0

and are conveniently removed from the process as water vapor. Although
precise analytical methods for determining the oxide content of fluoride
melts have not been suitably dewveloped, recent measurements of equilibrium
- quotients for this reaction show that the efficiency of oxide removal by
this process should be high.2 In the production process, treatment with
HF is continued to a practical reaction completion which should provide.

a suitable "oxide capacity" of these fluoride mixtures for inadvertent
contamination during reactor operations. :

Sulfur Removal

Sulfur impurities need also to be removed because of their corrosive
attack on nickel-base alloys at elevated temperatures. In the starting
materials these impurities are found primarily as sulfates and require
multiple treatment to effect their removal from the fluoride mixtures.
During a preliminary treatment, sulfates are reduced to sulfides by the
addition of beryllium metal turnings to the molten raw material mixture.
Hydrogen sulfide is produced simultaneously with oxide removal by the re-
action

3

82— 4 ZHF — 2F + H,S

and is removed from the system in the gas effluent. However, the rate

at which H»oS can be removed is low and is proportional to the hydrogen

flow rate. This condition may result from an intermediate reaction of
H>5 with the nickel salt container; sulfur removal would be controlled

by the reaction

Hp + Nis = Ni® + HpS .

Hydrogen flow rates of about 10 liters/min are used in the production
process; the maximum sulfur-removal rate has been approximately 0.7 g/hr.

Because of these difficulties in removing sulfur impurities from
molten fluoride mixtures, efforts are made to obtain sulfur-free starting
materials, Of the fluoride salts that have been purchased for the pro-
duction of fluoride mixtures for the MSRE, only beryllium fluoride has
been found to contain sulfur impurities; howewver, through cooperative
efforts with the wvendors, these impurities have been virtually eliminated.
Approximately two-thirds of the BeF; required by the MSRE is sulfur-free;
the balance contains concentrations of 10 to 1000 ppm.
290

Removal of Structural-Metal Impurities

The INOR-8 alloy (Hastelloy N) used as the structural metal container
in the MSRE contains 6 to 8% chromium as a constituent. It is expected”
that the chromium activity in the surface layer of metal in contact with
MSRE fuel will be depleted until the following equilibrium is established:

cr® + 2UF, = 2UF; + CrF, .

If the molten fluoride mixtures introduced into the MSRE contain nonequi-
librium concentrations of structural-metal fluorides more easily reduced
than UF,; (e.g., NiF, or FeF,), the INOR-8 container would corrode.

Structural-metal fluorides are present as impurities in the fluoride
raw materials and may also be introduced by corrosion of the process
equipment during production operations. Thus, the control of structural-
metal fluoride concentrations in the purified fluoride mixtures is an im-
portant process consideration.

Hydrogen fluoride will readily attack structural metals and alloys
that are suitable as salt containers at the operating temperatures of the
production process by reactions of the type

M0 + CHF = MF, + Hp .

This reaction is arrested in the gas phase of the treatment vessel by the
formation of a rather impervious layer of the structural-metal fluoride
on the metal surfaces. However, metal surfaces which are in contact with
the fluoride mixture are continually renewed by the dissolution of the
structural metal fluorides into the melt. Experimental studies of the
thermodynamics of corrosion processes in molten‘:f‘luorides6 have shown
that the corrosion of nickel could be controlled by admixing HF with hy-
drogen in concentrations which would also be effective for oxide and
sulfur removal. Concentrations of HF in hydrogen which would be needed
to control the corrosion of iron and chromium are too low to be of prac-
tical wvalue by current production techniques. Accordingly, nickel metal
is used as the primary salt containment material in the process equipment.
‘A gas mixture containing approximately 1/10 mole fraction of HF in H, has
been used effectively in the fluoride purification process. As suggested
by studies of high-temperature thermocouple research,7 the presence of
hydrogen should also reduce the corrosiveness of the HF-H,0 effluent gas
mixture which accompanies the conversion of oxides to fluorides.

Gas sparging of the melt at 700°C with hydrogen alone is used as a
final phase of the purification process to reduce FeF, and remaining NikF,
concentrations to low values. The reduction of CrF, by hydrogen is too
slow to be effective at process temperatures. However, chromium concen-
trations of 25 to 50 ppm that are present in the starting materials are
not prohibitive for reactor applications and remain essentially unchanged
by the production process. When structural-metal fluoride concentrations
are exceedingly high, beryllium metal, a strong reducing agent, is added
to the melt. This treatment is followed by a mild hydrofluorination to

’
291

convert the unreacted excess of beryllium metal to its fluoride salt.
Insoluble metallic impurities are separated from the purified fluoride
mixture by decantation and by filtration through sintered nickel during
transfer of the molten mixture to its storage container.

Production Operations

Fused fluoride mixtures are prepared in two batch processing units
from fluoride salts that are normally purchased from commercial sources,
Although the units are operated independently of each other, both are
charged with molten raw materials from a single meltdown furnace assembly.
Figure 1 shows the floor plan of the main operating level of the produc-
tion facility and indicates the flow of materials through the processing
units., A1l operations are coordinated so that the batch processing units
will operate on a semicontinuous schedule.

Starting Materials

Fluoride salts that are used as starting materials for the produc-
tion of fused fluoride mixtures for the MSKE were purchased, when pos-
sible, from commercial sources on a competitive bid arrangement, Since
further purification of these materials during the production process is
limited, only those of highest purity, within the competitive market, were
accepted; chemical specifications are shown in Table 1. Although all
materials obtained for use in preparing MSRE fluoride mixtures were gen-
erally within these specified limits, iron concentrations of 250 and 500
prm were allowed in BeF; and LiF respectively. Some carbonaceous impuri-
ties were also allowed since they are readily removed as carbon by gas
sparging and are inherent in some manufacturing processes. Approximately
12,000 1b of beryllium fluoride was purchased for preparing the first
reactor loading of fluoride mixtures. Zirconium tetrafluoride that is
relatively free of hafnium (<100 ppm) and otherwise of high purity was
also found to be readily available from commercial sources. Approximately
2300 1b of this material was purchased for use in the MSRE fuel mixture.

For neutron-absorption cross-section considerations, all lithium
fluoride used in salt mixtures for the MSRE must be essentially free of
61i. Sufficient "LiF (22,000 1b) was obtained from the USAEC for one
complete reactor loading and the planned replacement of the flush and
fuel salt mixtures. The isotopic assay of this material is at least
99.99% 7Li; however, those batches having the highest ®Li content have
been used in preparing the coolant salt mixture.

Although the #3°U enrichment in the MSEE fuel mixture during power
operation will be approximately BO‘ZO5 all ?3°U will be obtained for proc-
essing as highly enriched (92% in <2°U) uranium tetrafluoride. About
90 kg of 2357 will be required for reactor tests currently scheduled.

The balance of uranium inventory in the reactor fuel system will be made
up of UF, that is depleted of 23577, These materials are readily available
from USAEC.
292

UNCLASSIFIED
ORNL —DWG 64—-6998

DOWN 7T %B/
SHOWER =
PROTECTIVE 7, —
- CLOTHING
STORAGE —T™

l:fi_/scm_Es
LOADING ROOM | Y{&—LOADING HOPPER

,
!y

et {OH-EHD)

MELT DOWN
SALT

RECEIVER \/— @

\CONTROL

PANELS

TTTTTT
COWN
—yllll

Fig., 1. Fluoride Production Facility — Layout of Operating Level,

Table 1., Fluoride Production for MSRE -- General Chemical
Specifications for Starting Materials

Allowgble Concentration
Impurity (wt %)
(1 ppm = 0,0001 wt %)

Water : 0.1

Cu , 0.005
Te | 0,01 ¢

Ni 0.0025

S 0.025
Cr 0.0025,
A 0.015"
si 0.01¢

B . 0.0005,
Na 0.05¢

Ca 0.01¢ - B}
Mg . 0.01 ¢

K 0.01 [
i (natural) 0,005

Zr (natural) ' - 0,025
cd 0.001

Rare earths (total) 0,001,

293

7LiF Densification

With the exception of 7LiF, all raw materials obtained for preparing
the MSRE fluoride mixtures can be charged directly to the meltdown fur-
nace, The lithium fluoride, as received, was found to have a very low
bulk density. Subsequent examination showed that it also contained sub-
stantial amounts of water. Since the direct use of this material would
result in loss of production capacity and excessive oxide contamination,
pretreatment of the "LiF to improve these properties was desgired.

Densification of LiF by heating to 650°C had appeared promising in
laboratory experiments.8 " Intermediate-scale tests were made preparatory
to production-scale operations to develop a processing procedure and to
examine the feasibility of pretreating the 22,000 1b of material that was
on hand. These tests showed that periodic agitation of the LiF while
heating to 650°C was necessary to produce a free-flowing granular product.
Anhydrous HF was mixed with the helium sweep gas while the charge was

being heated to about 400°C to convert LiOH, either initially present or

formed by pyrohydrolysis, to LiF; otherwise, heating to 650°C would permit
the LiOH to be fused with the LiF to give an intractable mass.

‘The production-scale eqfiipment is a horizontal Monel reaction vessel
(17 in. in diameter by 8 ft long) equipped with a full-length agitator
and a heating Jjacket. The apparatus had been used during the ANP Program
in the conversion of solid ZrCl, to ZrF, by treatment with anhydrous HF
at elevated temperatures. This equipment is shown in Fig. 2. Up to 300
1b of 7LiF has been processed in a single batch; approximatel¥ 11,500 1b
has been processed to date. The average bulk density of the ‘LiF has been
increased from about 0.6 g/em® to 1.1 g/cm® by this operation.

Raw-Materials Charge and Meltdown

Because of the toxicity of fluoride salts (beryllium fluoride in par-
ticular), the raw materials are handled by operating personnel in a load-
ing room isolated from other areas of the production plant by shower fa-
cilities and air locks. Personnel working within the loading room wear
fully protective, plastic, fresh-air suits to avoid exposure to these
hazardous chemicals, Raw materials are assembled in this area, weighed
into appropriate batch sizes in a well-ventilated hood enclosure, and

“transferred to the meltdown furnace by a vibratory conveyer. A part of

this operation is illustrated in Fig. 3.

The meltdown furnace assembly shown in Fig. 4 adjoins the raw-ma-
terials loading room and is operated to provide a molten raw-material
charge to each of the two adjacent batch processing units. After melt-
down the molten fluoride mixture is sparged with hydrogen and helium at
relatively high flow rates to remove insoluble carbon by entrainment.
Beryllium metal turnings[are also added to the molten charge to reduce
sulfate impurities to sulfides and to reduce structural-metal impurities
to their insoluble metallic states., During subsequent transfer of ma-
terial to a batch processing unit, some of these insoluble impurities are
separated by decantation.
Fig. 2. Fluoride Production Facility — Horizontal Densification
Kiln.

Fig. 3. Fluoride Production Facility — Raw Materials Handling Area.
295

UNCLASSIFIED
PHOTO 70838

Fig. 4. Fluoride Production Facility — Premelting Furnace Assembly.

296

Melt Purification

Each of the two processing units has a capacity of 2 ft? of salt per-
batch. Thus, the production of all fluoride mixtures for the MSRE will
require a minimum of 94 batch preparations. A 10% production excess is
planned to allow for usual contingencies. Although the production rate
varies with the quality of the starting materials, the average batch cycle
encountered thus far, including maintenance shutdown periods, is 164 hr.

_ Salt treatment wvessels used in the process units as well as in the
meltdown furnace are constructed of 12-in. iron pipe size 304L stainless
steel pipe and are approximately ¢ ft long. An inner liner fabricated
from l/8—in. grade A nickel sheet provides primary salt containment. Ex-
cept for conventional loading ports and gas-line connections, the vessel
is of welded construction. Salt storage contalners are of the same ap-
proximate diameter but are only 3 £t in length. These vessels are made
of nickel and are used only to store and dispense purified salt mixtures.
The treatment vessels and salt storage vessels are heated by commercial
resistance furnaces of 50 and 25 kw respectively. Figure 5 shows the ar-
rangement of equipment in one of the two identical batch processing units.

A simplified schematic flow diagram of the unit process is shown in
Fig. 6. The salt treatment vessel connects directly to the receiver
vessel (salt storage container) through a small-diameter tube which func-
tions as the process gas entry line during purification of the fluoride
mixture and as the salt transfer line at the completion of the process.
These vessels are also connected through the gas manifold system to pro-
vide centralized control over process conditions during salt purification
and regulation of the differential gas pressure on the system during
transfer of the molten charge to its storage container. :

The gas effluent from the process is passed through a bed of sodium
fluoride pellets to reduce HF concentrations to low values before dis-
charge to the atmosphere. Since large quantities of hydrogen are used,
exhaust gases are bubbled through an inert-liquid trap to isolate the
process equipment from the atmosphere.

At the completion of the production cycle, the purified fluoride mix-
ture is transferred to the salt storage container and allowed to cool.
Interim storage under a static helium cover gas is provided within the
production facility. The containers will be shipped to the reactor site,
as they are needed, for remelting and transfer into the fluoride drain
tanks of* the MSRE.

Process Control

To ensure delivery of acceptable fluoride mixtures to the MSRE, ma-
terial balances and process conditions must be controlled during each
phase of the production operation. 1In addition to accounts that show the
source, quality, and amounts of raw materials used in each production
batch, various measures are exercised during the processing cycle to pro-
vide quality control of the finished product. Primary process control is
UNCLASSIFIED
S PHOTO 70834

Fig. 5. Fluoride Production Facility — Batch Processing Unit.

298

v UNCLASSIFIED
ORNL-DWG 64-6993

HF SUPPLY AND

METERING SYSTEM GAS MIXING CHAMBER
. __FILTER
___________ i |
: HELIUM : [ T§
| HYDROGEN figfi__sow GAS - \RESSé'Ii\-IrOIR
| VACUUM | M ANALYSES : .
|
L] I—— Se-FURNACE
NN
et
CARBON TRAP
CoLD 7 [_
EXHAUST TO TRAP !
_ATMOSPHERE - —
$ ‘-'-J-‘ GAS : TREATMENT
" VESSEL
//" ANALYSES I :
ANALYSES PE ETS - i N\ Furnace
"" CONDENSATE . !
HF REMOVAL DRAIN LINE
WATER
REMOVAL

Fig. 6., Fluoride Production Facility — Simplified Schematic Process
Diagramn,

achieved by frequent analyses of the process gas streams. These analyses
are used as guldes for alteration or termination of process conditions
and are, therefore, directly related to production efficiency and quality
control. ©Secondary control of product quality is derived from chemical
analyses of salt samples that are withdrawn from melts during processing
operations. These results are used to determine the acceptability of the
finished batches.,

Process Gas analyses

The HF-Hp concentration ratio in the gas influent to the salt treat-
-ment vessel 1s regulated according to results obtained by direct titration
of a side stream with a standard KOH solution. A similar titration of the
gas effluent stream is used to determine HF utilization. Typical results
obtained during preparation of the MSRE coolant salt mixture are illus-
trated in Fig. 7 and show the high degree of HF utilization that is char-
acteristic during initial treatment of the molten raw materials.

Hydrogen sulfide is collected in an ammoniacal cadmium chloride so-
lution from the gas effluent stream and is titrated with a standard iodine
solution. Material balances on sulfur evolved as H»S from the treatment
vessel as compared with the quantity of sulfur believed to be present in
the raw-materials charge are often in disagreement and probably result
from the precipitation of sulfur compounds in the meltdown furnace. How-
ever, experience obtained thus far shows that the rate of H,S evolution
is indicative of the concentration of sullfur remaining in the melt., This
condition appears wvalid only when all sulfates have been reduced to sul-

- fides. Removal of sulfur from the melt, therefore, is not considered com-
plete until a subsequent addition of beryllium metal fails to result in
299

an increase in the HoS evolution rate. Results shown in Fig. 8 illustrate
the sensitivity of this control measure.

Water vapor which results from the reaction of HF with oxide ion in

the melt 1s condensed, with some HF, from the gas effluent stream in a

UNCLASSIFIED
ORNL-DWG 646996

HF INPUT CONCENTRATION
BATCH NO. C-116 /(/O-
31 . WEIGHT: 120 kg

O FROM ANALYSIS OF
_© GAS EFFLUENT

(s /
/

0 5 10 15 20 25 30
PROCESS TIME {hr)

HF CONCENTRATION (meq/liter Hy)
ro

o

0

Fig. 7. Fluoride Production for MSEE — Utilization of HF During
Purification of "LiF-BeFs (66-34 mole %) at 600°C.

UNCLASSIFIED
ORNL-DWG 64-6995

0.7 I | T
BATCH NO. C-116
WEIGHT: {20 kg

0.6 |—- HF CONCENTRATION: 4 megq /liter Hy
INITIAL SULFUR: 85ppm=10.2 g

0.5

0.4 \

S -

0.2 b,

/ AREA = 6.59 SULFUR
e 0.1 /

RATE OF SULFUR REMOVED (g/hr)

0 5 10 15 20 25 30
PROCESS TIME (hr)

Fig. 8. TFluoride Production for MSRE — Removal of Sulfur Impurities-
from "LiF-BeF, (66-34 mole %) During HF-H, Treatment at 600°C,
300

cold trap. The condensation chamber is maintained at about —12°C by re-
frigerating a surrounding ethylene glycol—water bath. Dewvelopment tests
have shown that essentially all water wvapor can be condensed from an HF-H»
stream at this temperature. The contents of the cold trap are drained
reriodically and analyzed for water content. The reproducibility of re-
sults obtained by this process control method on a number of salt prepa-
rations is shown in Fig. 9. Oxide removal is considered complete when
the water-removal rate becomes negligible.

During the final hydrogen sparging treatment, the reduction of struc-
tural metals results in the evolution of HF. The rate of HF evolution
is indicatiwve of the reduction rate and is also a function of the struc-
tural-metal ion concentration of the melt. Therefore, the HF content of
the gas effluent is monitored periodically by direct titration with a
standard KOH solution. Values obtained by this procedure are used to de-
termine process completion.

UNCLASSIFIED
ORNL-DWG 64-6994

i

400
l l

[ ]
BATCH a® a
= gol___ 0 C-134 °©
o ‘. FAY
2 A C-133 o
.z o C-132 R
(o]
o 60 ® C-130 ‘D
z & C-128 8
[}
z |
>
S 40 o
ul
[+ d 8]
[veg AJ
, L 20 .
L=
= 0
f“T
0

0 5 10 15 20 25 30 35
PROCESS TIME (hr)

Fig. 9. Fluoride Production for MSRE — Removel of Oxides from 7LiF-
BeFo (66-34 mole %) During HF-H, Treatment at 600°C.

Salt Analyses

'Chemical analyses are obtained on salt samples that are withdrawn
from the melts into copper filter tubes during processing operations,
These tubes are previously fired in an atmosphere of flowing hydrogen and
installed in a sampler device., This assembly is fitted onto a wvalved port
of the treatment vessel and permits sampling under existing process con-
ditions without extraneous contamination to the sample. Salt is crushed
from these sample tubes in a glove box and stored in glass sample bottles.,

- The concentrations of impurities found by chemical analyses of final
salt samples taken for production thus far are shown in Table 2. Values
for "oxides removed" were determined from quantities of water collected

301

Table 2. Fluoride Production for MSRE — Results of Chemical
Analyses of Batch Preparations of 7LiF-BeF, (66-34 mole %)@

Average Concentrations of
Tmpurities (ppm)

Batch Nos. Oxides Removed
Cr Ni Fe S
101-105 13 31 183 <5 1391
106—110 11 26 161 <5 1110
111-115 10 19 180 <5 1828
116~120 15 20 164 <5 1473
121-125 50 21 115 <5 1465
126—130 14 22 127 <7b 2301

®Batch size, 120 kg of salt mixture.
bBatch 126 at 37 ppm of sulfur, all others at <5 ppm.

in the effluent gas cold trap. The values for chromium concentrations
are low but significant because they reflect the integrity of the treat-
ment vessel, Should the nickel liner in this wvessel fail, a rapid in-
crease in chromium concentrations would occur by corrosion' of the stain-
less steel outer vessel.

Reactor Loading

The fluoride production method is generally independent of fluoride
mixture composition provided that the liquidus temperature of the mixture
is within the capability of the process equipment. The production of mul-
ticomponent mixtures, however, is sometimes facilitated by the preparation
and subsequent combination of simpler binary or ternary mixtures. As ap-
plied to the production of the fuel salt mixture for the MSRKE, this tech-
nique is advantageous to both production and reactor operations.

To provide for conservation of fissionable material, for nuclear
safety, and for planned reactor operations, the MSRE fuel salt will be
prepared as a fuel concentrate mixture and as a barren fuel solvent mix-
ture. The fuel concentrate mixture will consist of the binary eutectic
mixture ‘LiF-UF, (73-27 mole %); the barren fuel solvent mixture will con-
sist of the balance of fluoride salts required to achieve the overall
fuel composition LiF-BeF,-ZrF,-UF,; (65.0-29.1-5.0-0.9 mole % respectively).

The fuel concentrate mixture will be further segregated as the en-
riched fuel concentrate mixture, which will contain all 22°U as highly
enriched 235UF4, and as the depleted fuel concentrate mixture, which will
302

contain the balance of nonfissionable uranium required for the fuel salt
mixture. This depleted material will be prepared in two batches by con-
ventional processing techniques. '

The enriched fuel concentrate mixture will be produced in six small
batch preparations, BEach batch will contain approximately 60 1lb of salt '
mixture, of which 33 1b will be #?°U. Although the processing technique
will be essentially the same as that used for conventional fluoride pro-
duction, the operation will be carefully regulated according to estab-
lished criticality-control procedures.

Before critical experiments are conducted in the reactor, uranium
analyses and inventory procedures will be checked out on a mixture of the
depleted fuel concentrate and the barren fuel solvent. Approach to the
critical and then to the operating conditions of the MSEE will be achieved
by incremental additions of enriched fuel concentrate mixture to the re-
actor fuel.

Production Economics

During the course of fused fluoride production at ORNL, modifications
and revisions have been made on the facility to meet specific safety re-
quirements and to incorporate procedural changes which have resulted from
process development programs., When the use of beryllium fluoride in fused
fluoride mixtures became attractive to the Molten-5alt Reactor Program,
extensive modifications in the physical plant were made to cope with
health hazards which accompany the handling of beryllium compounds. The
required preparation of large quantities of identical fluoride mixtures
for the MSRE also made possible the effective use of a single meltdown
furnace assembly. Other production features developed for the process
have reguired only minor revision of the facility. The replacement wvalue
of the production facility is probably in the range of $300,000 to $500,000.

Present operation of the production facility for the preparation of
- MSRE materials is conducted on a seven-day, three-shift schedule. Each
shift requires a technical operator and an assistant for routine mainte-
nance. Other supporting crafts are employed for nonroutine maintenance
as required. Operation, maintenance, and equipment costs are budgeted
at an average of $20,000/month.

Data on the fluoride starting materials that hawve been acquired for
the production of MSRE materials are shown in Table 3. The cost of the
raw materials for the coolant and flush salt mixture, ‘LiF-BeF, (66-34
mole %), is $11.29/1b, and that of the fuel salt, excluding *3°U, is
$10.13/1b. As calculated from operating and raw-materials costs, the
coolant and flush salt mixture has an estimated value of $19.7l/lb. A
similar cost for the fuel salt mixture, excluding °2°U, is estimated at
$17.33/1b. Thus, the initial charge of coolant, flush, and fuel salt mix-
tures for the MSRE will have an estimated value of $496,400, exclusive
of plant amortization and uranium costs.
303

Table 3. Costs of Raw Materials Used in MSRE Fluoride Production

Material Quantity Uni%$§ost Tota%$§ost
iF 22,000 1b 16.502 363,000
BeFs, 12,000 1b 5,70 68. 400
ZrF, 2,300 1b 8.00 18,400
UF,, 90 kg "~ 12,000.00 1,080,000

(435U basis) |
Total 1,529,800

“Includes $1.82/1b for preparation as fluoride salt.

References

G. J. Nessle and W. R. Grimes, Chem. Eng. Progr., Symp. Ser. 56(28),
51 (1960). -

MSRP Semiann. Progr. Rept. Jan. 31, 1964, ONRL-3626, p. 137.
MSRP Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 117.

C. Fo Baes, Jr., and H. H. Stone, private communication to H. F.
McDuffie, July 27, 1964.

Fluid Fuel Reactors, ed. by J. A. Lane, H. G, MacPherson, and F,
Maslan, p. 599, Addison-Wesley, Reading, Mass., 1958.

C. M, Blood, Solubility and Stability  of Structural Metal Difluorides
in Molten Fluoride Mixtures, ORNL-CF-61-5-4 (Sept. 21, 1961).

G. W. Keilholtz et al., Reactor Chem, Div., Ann. Progr. Rept. Jan. 31,

1961, ORNL-3127, p. 133.

B. J. Sturm, A Method for Densifying Lithium Fluoride Powder, MSR-
62-94 (Nov. 20, 1962) (internal use only).

304

FUTURE CHEMICAL DEVELOPMENT

H. F. McDuffie

The MSRE illustrates a reactor concept — the fluid fuel concept —
which has been espoused and under development in two different embodi-
ments by ORNL since 1949. The major advantage of this concept, and the
one which supports continuing chemical research and development, is that
changes in the fuel or blanket compositions can be made without the re-
quirement of metallurgical or ceramic refabrication. Fertile material
can be added, bred fissionable material and fission products can be re-
moved, and samples can be taken for analysis, all without interrupting
the operation of the reactor. The accompanying maJjor disadvantage is
that the radioactive fission products are dispersed throughout the sys-
tem, complicating the containment and maintenance problems.

As a consequence of the research and development work done for the
ARE and the MSRE, and also the fundamental research studies of molten
salts supported by the AEC Division of Research throughout the last 15
years, there are a number of chemical developments which permit a tenta-
tive look ahead to the extension of the molten-salt concept to economical
operation and to a broader range of reactors. Reactors closely related
in type to the MSRE will be discussed first, with other types of thermal
and fast reactors discussed later.

One- and Two-Region Molten Fluoride Reactors

Fuels and Blankets

The importance of using thorium and the existence of possibly suitable
vehicles have been discussed in previous sections, but a restatement will
be made for emphasis. Figure 1 shows the system LiF-BeF,-ThF, as studied
by Thoma et al.;l a maximum of 24 mole % ThF, can be accommodated in so-
lution at 550°C. This is equivalent to 78 wt % Th¥,, or 2400 g of thorium
per liter. The system LiF-BeF,-UF, (ref. 2) is quite similar to the
thorium system, with even a slight lowering of the liquidus temperatures
upon the fractional substitution of the uranium for thorium in the qua-
ternary system LiF-BeF,-UF,-ThF, (ref. 3). Thus both thorium blankets
for two-region breeder reactors and thorium-uranium fuels for single-
region converters are potentially available.

Barton and co-workers have also measured the solubility of PuF; in
alkali fluoride-beryllium fluoride mixtures® and found values of 0.2 to
1.0 mole % in the temperature range 500 to 600°C. TFigure 2 illustrates
the effect of temperature and shows, also, that there is a substantial
effect of solvent composition on plutonium solubility. Thus, molten
fluoride reactors should be operable with plutonium as the fissionable
fuel component instead of uranium.
305

UNCLASSIFIED
ORNL-LR-DWG 37420AR2

ThE, 4414
W00
TEMPERATURE IN °C
COMPOSITION IN mole %
1050
LiF-4The,
LiF-2ThF, 1000
LiF-2ThF,
950
P 897
2LiFBeF,
200
7LiF-6ThE
76
prez 850
800
P 597
£ 568
3LIF-ThE1 Y 750
£ 565
65
s Os '
’3‘ )OOSO O, 6-5 0o
8, 0 00 950 £ 526
] o0/ \ Q

LiF \/ > BeF,
845 2LiF-BeF; 500%50 400 400 450 500 548

P 465 £ 370

Fig. 1. The System LiF-BeF;-ThF,.

Consideration has also been given to reactors in which a fast re-
actor core can be surrounded by a static molten-salt blanket. (A1l
moderator materials would have to be absent from the core; and lithium
and beryllium could not be used in the blanket, but fluorine could be
present.) The system NaF-KF-ThF, (ref. 5) provides molten mixtures
which appear suitable for such a blanket; the lowest melting eutectic in
the system, as shown in Fig. 3, contains 26 mole % ThF, and melts at
about 562°C. This mixture contains 65 to 70 wt % ThF,; and about 2400 g
of thorium per liter. |

Materials Compatibility

The compatibility of molten fluorides with INOR-8 (or‘Hastelloy N)
and graphite is apparently excellent, as indicated by the chapters on
chemistry, radiation effects, graphite, and metallurgy, as well as by
the long history of development work with molten fluorides in both the
MSR and the ANP programs. Nevertheless, many questions of long-term
306

UNCLASSIFIED
ORNL—LR—-DWG 36820

TEMPERATURE (°C)

6 600 550 500
10.0 ?O ? T ]
5.0 mole % LiF  mole % BeF,
@ o 7.3 28.7
@ a 68.1 31.9
@ o 63.0 37.0
@ v 56.3 437
G = 51,7 48.3
~ 20
§ \\‘.
L) N
o
E 1.0 “oa¥ \\
e “\ \I\\
& b N N
“&5\‘ g
05 N “~
. & \ ~d
e N S . N
[ ]
S
Vv \
@\“~ ®
0.2 ‘x\\\‘\ \\
\ y
@ R .
0.1
105 1.0 1.5 120 12.5 3.0 135 14.0
10,000/7 (°K)

Fig. 2. ©Solubility of PuF3 as a Function of Temperature for LiF-
BeFs> Solventse.

compatibility remain to be determined. The chemical effects of the
accumulation of fission products will certainly require study, and the
irradiation tests should be extended to cover progressively more severe
conditions which ultimately exceed by a substantial margin the projected
reactor operating conditions. |

Chemical Reprocessing

The main advantage of the fluid fuel concept, the ready availability
of the fuel for whatever physical or chemical treatment is desired,
furnishes the main source of future chemical development — laying a basis
for ‘selective chemical reprocessing. The addition of components to a
molten-salt mixture is a relatively simple operation, but the removal of
either dissolved or suspended materials is considerably more difficult.

A number of fuel-reprocessing schemes have been proposed, and some
have been developed. Dawson and Sowden,6 in their excellent recent book,
discuss these under the following headings:

1. Fluorination — to remove volatile uranium hexafluoride (described in
a previous section), : :
307

UNCLASSIFIED
ORNL—DWG 64-4014

PRIMARY PHASE FIELDS

® NaF
4Naf - Thf,

(© TNoF:2ThF,
2Naf - ThF,
(® Nof-KF-Thfy
®

7{Na,K)F- 6ThF,

COMPOSITIONS IN moie %
TEMPERATURE IN °C
+HHH INDICATES SCLID SOLUTION

£-690
NoF: KF-ThF,
2NoF-ThFy firdh FHHHHHHHHHHHHHHHH
o
30 30 @ .3
& ®
éu
E£-562
A < 600
£-618,
(B
7NaF - 2T hFy ' pet3 © €90 4
P-627

4NoF -Th¥;

NoF \/ \/ \/ \/ \VA \/ A\V4 A\V4 \/
10 20 30 40 )
KF {mole %)

Fig. 3., The System NaF-KF-ThF,.

2. TFluoride precipitation - including the exchange of rare earths with
cerium trifluoride,

3. Electrolysis — which would remove elements whose decomposition TO-
tential is lower than that of the main melt constituents (leaving

lanthanides in solution),

4. Oxide precipitation — including the use of beryllium oxide and water
vapor to precipitate uranium dioxide.

- Fission Product Rare Gases. Of the fission products produced in a
molten fluoride fuel, the noble gases should most readily escape as a
consequenice of their low Solubility.7 Figure 4 illustrates the low
solubility of helium, neon, argon, and Xenon in molten LiF-BeF, over a
temperature range of interest to reactor operation. Future chemical
development in this field will be limited to confirmatory measurements
of solubility in new fuel or blanket compositions. The factors governing
the diffusion of gases through graphite and other porous media have been
thoroughly studied® in connection with the development of gas-cooled re-
actors. . For application to the unclad graphite in molten-salt reactors,
it may be important in the future to determine the diffusive behavior of

308

UNCLASSIFIED -
ORNL-LR-DWG 41908

TEMPERATURE {°C)

7 800 700 650 600 500
5X40
f I | |
46/__ :
, \ :.5200 .
Ca,l/
2%
4,9{_ e
167 fsooc
— <% ~—~
£ ko iy, —
- T
> 5 ‘\___. \
E ™.
8 s
:, 2 \CQ// Y
io’ \. O/ \
£ <y
S o® ~:’e p \.\
X v 0, L
N o, 3 \\
N0, ~e.
5 G ™
AN \1{6‘ 9
(X \ ]
N, N
2 \\"’é‘,,,
\3(\
= Ll |
- 8 9 10 1 12 13 {4 15
10.000/T (°K)

Fig. 4. Temperature Dependences of Solubilities of Noble Gases in
LiF-BeF, (64-36 mole %).

UNCLASSIFIED
ORNL-LR-DWG 31423A

2 ] T . ;
\ (k‘\\PROBABLY UNSATURATED
\\ N\
1 /A.:‘ \O\\
~L0F3 "\\ \

(@]

o
O—.—
[¢.]

g
(1]
4}5
/){

3
-n
o

SOLUBILITY { mole % )

o SmF \\ IN LiF-BeF; - UFy
3
r{62.8-36.4-0.8
0.2 ACeF, ( —

mole % )
ALaFy \\

A

04

9 10 M 12 13 14 15 16
10,000/ 7 (°K)

Fig. 5. Solubility of CeF; and LaF; in LiF-BeF,-UF, (62.8-36.4-0.8
mole %).
309

the fission product gases in the very impermeable graphites which are
suitable for use with molten fluorides.

Rare Earths. The long-lived rare-earth fission products form very
stable trifluorides and contribute a large proportion of the total fission
product poisoning. Thelr removal is very necessary for economic opera-
tion. Figure 5 illustrates the solubility of CeFs, LaFg, and SmF5; in a
LiF-Bel's mixture of the same ratio as the MSRE solvent.

The work of Ward et al.l® showed that exchange reactions between
solid cerium trifluoride and dissolved rare-earth fluorides of higher
cross section would occur, with the molten fluoride becoming saturated
with cerium fluoride and the solid phase being a solid solution of a small
amount of rare-earth fluoride in cerium fluoride. The solubility of
cerium fluoride has been shown to be extremely dependent on the compo-
sition of binary fluoride solvent mixtures such as LiF-BeF, and NaF-BeF,,
as i1llustrated by Fig. 6; strong dependence on temperature is also shown
by these studies. Although it is not attractive to consider use of a
fuel saturated with even ~0.1 mole % of cerium, it seems quite likely
that more study of the solubility and composition relationships involving
the rare-earth fluorides may be profitable.

A technique, which from recent results appears simple and feasible,
involves the vacuum distillation of an irradiated fluoride fuel to re-
cover 7L1F BeF,, and ZrF,, leaving the uranium and rare- earth fission
products behlnd Vapor phase transport of salt components is, of course,
well known. Kirslis, Savolainen, and Blankenshlp noted distillation as
a consequence of the temperature gradients present in capsules under ir-
radiation. Their evidence suggested that the distillate had a compo-
sition near that of 2LiF-BeF,. Kreyger in studies of the behavior of
sessile drops of molten 2LiF-BeF,;, noted that the surface became rich in
LiF and that distillate drops were 95% BeF,. Thoma, Singh, and Ross
studied the purification of LiF by vacuum'distillation;l they reported
that the impurities and the LiF distilled in the order of the vapor
pressures of the components, with NaF and MnF, coming over ahead of the
bulk of the LiF and with CaF; and Mgk, being left behind. In very recent
studies, Kelly has examined the vacuum distillation of MSRE fuel, to
which he had added europium to represent the rare-earth fission product
poisons. A charge of 57.2 g of fuel was distilled from a stainless steel
pot at temperatures in the range of 1000 to 1050°C. Samples of approxi-
mately 5 g each were taken serially and analyzed for the fuel components
Li, Be, Zr, F, U, and (by counting techniques) Eu. Figure 7 shows the
smooth change in Li to Be ratio of the distillate, the inclusion of the
zirconium along with the beryllium, and the very low concentration of
europium. Actually, the relatively larger concentration of europium ob-
served in the initial stages of the experiment is attributed entirely
to the result of dusting, as the initial charge was powdered and the
vacuum was applied before the charge was first melted. If these results
are confirmed by tests with other rare earths, a "brute-force" technique
for recovering the lithium and beryllium fluorides seems assured, with
the residue containing the rare earths plus the uranium being a much more
desirable charge for the fluoride volatility process than the entire fuel
volume would have been.
310

. UNCLASSIFIED
" ORNL—LR - DWG 39658

2.4 \ I
® LiF- Ber SOLVENT ¢ NaF -BeF, SCLVENT

2.2 \

2.0

\7’OO°C
1.8
700°C
1.6
\. 7
o
/ ¢
1.4
\ /.l /
A ® /

CeF; IN FILTRATES (mole %)
)

0.8 \ T [ T

0.6 ‘
\\\\.,g——acr’zr’» ;o0

0.4

| a i \y‘-"’(/a /
o) 0/
0.2 Qe 00°C y{
°C
o 500
o]
20 30 : 40 50 20 30 40 50 60
BeF, IN SOLVENT (mole 7s) . BeF, IN SOLVENT (mole %)

Fig. 6. Solubility of CeF; in LiF-BeF, and NaF-BeFa.

Zone melting is also being explored as a potentlal basis for purifi-
cation of irradiated fuel. Singh, Brunton, and Thoma'* have reported
that 12 passes of a 3-cm zone at 1 cm/hr removed an initial concentration
of 1000 ppm each of CeF3, GAF3;, and LuF3 from a 30-cm ingot of LiF. The
choice of the three rare earths was such as to provide representatives of
the three classes of phase behavior observed in MF-MF3; systems — no inter-
mediate compound, an incongruently melting intermediate compound and a
congruently melting intermediate compound.
311

UNCLASSIFIED
ORNL-DWG 64 -7169

100 [ \ | ! 1 | 10° #
. - ORIGINAL CHARGE: 57.16 g i
COMPOSITION: LiF-BeFp-ZrFy-UF, et
90 . mole %o 640-29.17-50- 083 = = / 9
EUROPIUM TRACER: 3000 ppm ‘
“\ POT TEMPERATURE: ~1000°C /

!

BRI S

]

()] -l

Q Q

"]
o

0

~—

(o))

/
A
[ ]
)?Q
\<_
i e

AN
\
/.

EUROPIUM CONTENT OF DISTILLATE {ppm)

COMPOSITION OF DISTILLATE {mole %)
()]
o
S
A
L]

30 — / 3
’ SN LY N
10 T ZrF; 7 s 1
» " S
(- . RJ\-‘———-—-—__./I\.___I'}-:—'—
- ~ S o
e} - N
o - 10 20 30 40 50 60 70 80 90 100

CUMULATIVE WEIGHT PERCENT OF CHARGE DISTILLED

Fig. 7. Distillation of MSRE Fuel,

The extraction of rare-earth fission products from liquid-metal re-
actor fuels into molten salts was explored at Brookhaven National Labo-
ratory in connection with the LMFR.1” Favorable distribution coefficients
were found for rare-earth removal when the composition and redox potential
of the salt were properly controlled. The converse extraction, from
molten fluorides into liquid metals, would be necessary 1f rare~earth
fission products were to be removed from the fuel of the MSRE or similar

. reactors. Baest® has recently considered the application of this process
to molten fluoride reactors in the light of available thermodynamic

. knowledge. The higher stability of the rare-earth trifluorides as com-

. pared with beryllium fluoride would require that an extremely stable

intermetallic compound of the rare earth and a reduced metal be formed in
order for the extraction to be successful. Recent information on the
activity coefficients of .cerium in molten zinc and leadt? suggests that
these metals would not be suitable; further consideration of tin and
bismuth may be worthwhile. A current paperl8 indicates that, at 600°C, in
the system Mg-40% Th/KC1-LiC1l-25% MgCl,/Zn—7% Mg, the three solutes
/ yttrium, cerium, and neodymium were each extracted from the Mg—40% Th
T through the salt and into the Zn—7% Mg alloy — to the extent of over 95% in
2 hr. Modifications of the composition of the molten-salt portion of a
metal-salt extraction system are known to have profound effects upon the
distribution of solutes; Moorel? has shown that changes in the A1Cl;:KCl
mole ratio can change the distribution coefficient for uranium between
the molten chlorides and molten aluminum at 725°C by a factor of 50 or
more. In future chemical development, considerable emphasis in fundamental
312

studies will be placed on the determination of the activities of various
solutes in molten fluorides, as a function of solvent composition, redox
‘potential, and temperature. The techniques of electrochemistry, liquid-
liquid extraction into molten metals, and control of gas-phase composition
will be used in these studies.

Removal of Uranium and Protactinium. Shaffer et al.?0 have reported
the recovery of protactinium and uranium from molten fluoride systems by
precipitation as oxides. Initially, trace quantities (1L to 2 ppb of
233Pa) were removed from the molten solvent mixture LiF-BelFs-Thiy (67-18-
15 mole %) by the addition of powdered BeO or Ca0, as shown by Fig. 8.

In subsequent tests more realistic concentrations (with respect to reactor
applications) of protactinium in the fluoride solvent were obtained by
dissolving 221Pa to the extent of 50 to 75 ppm in the fluoride melt. The
removal of this larger concentration of protactinium from solution by the
addition of powdered BeO or ThO, is shown in Fig. 9. When both protac-
tinium and uranium were present initially in solution, the addition of BeO
appeared to precipitate both, with the protactinium removed first and
preferentially, as shown in Fig. 10, Even UO; was able to remove the
23lpg from solution, as shown in Fig. 11. After each removal of protac-
tinium from solution, the precipitated protactinium was returned to solu-
tion easily when the contents of the pot (melt and precipitate) were
treated with gaseous HF. These results suggest a number of ways in which
selective precipitation could be used in connection with reactors which
depend on the conversion of thorium to 233U,

Studies of the crystallization path of MSRE fuel revealed that uranium
was not present in the first two phases to appear (2LiF-BeF, at 434°C and

UNCLASSIFIED
ORNL~-DWG 63-5368

100 @

® B0
O Cal

0]
(@]

INITIAL PROTACTINIUM CONlCENTRATIONS
10 counts/min—-g Salt = { ppb

<2}
@]

S
Q
r/

/

PROTACTINIUM FOUND IN SOLUTION (% REMAINING)

<100 y counts/min-g SALT

oy J———

0 { 2 3 ' 4 5
OXIDE PRECIPITANT ADDED (g)

Fig, 8. Removal of Protactinium (?33Pa) from Solution in LiF-BeF,-
ThF, (67-18-15 mole %) by Reaction with BeO and Ca0 at 650°C.

\,

LY

Yo lahmN ,/‘:.‘._\
R

313

UNCLASSIFIED
ORNL-DWG 63-4394

100 ’

INITIAL CONCENTRATION OF
Paay c~60ppm
80 .. OXIDE ADDED:
A ThO,

® BeO
60 \ -

ol AN
; <

0 i 2 3 4 5
OXIDE PRECIPITANT ADDED (g}

PROTACTINIUM FOUND IN SOLUTION (% remaining)

Fig. 9. Removal of Protactinium (231Pa Labeled with 23°Pa) from So-
lution in LiF-BeF,-ThF, .(67-18-15 mole %) by Reactions with BeO and ThO»
at 650°C.

UNCLASSIFIED
ORNL-LR-DOWG 52581

100 | |
O PROTACTINIUM
® URANIUM

g | |
z 80 'y INITIAL CONCENTRATIONS
<§I Pa 2.96 x10% ¢/min- g ~ 5ppb
W U=-1335ppm
[
> 60
j®)
=
o
—
Q
W
£ 40
(]
=
2 .
[P
.|
o \\\

20
J
= \

0 o o

2 3 4 5
BeOQ ADDED (grams}

Fig. 10, Reaction of BeO with Protactinium and Uranium Dissolved
in LiF-BeF,-ThF,; (67-18-15 mole %) at 650°C.

2LiF+7zrF, at 431°C). Zone-melting experiments with a vertical tube and

upward passes failed to cause any unusual concentration of uranium, but,
as illustrated by Fig. 12, 20 downward passes gave a very substantial con-
centration of uranium. An effect of gravity has been reported in zone
melting but only for insoluble trash in the material; this new effect in-

- volving soluble uranium should prove very interesting and rewarding to
314

UNCLASSIFIED
ORNL-DWG 63-4396

100 | ‘
[ ]

- INITIAL CONCENTRATIONS

£ OF ADDED REAGENTS

5 80 PaZ3! ~ 50 ppm

5 U~ qwr

(>

=

(@] L J

E 60 i

=2

-J

(@]

w

z

S

84o \

o L 4

=

=)

= \\

-

20

S

& \t\\\

& .
\.-.-.-

0 .---'-'_-—-
0 4 8 12 16 20

UO, ADDED (g}

Fig. 11. Removal of Protactinium (?31Pa Labeled with *?3Pa) from
Solution in LiF-BeF,-ThF, (67-18-15 mole %) by Reaction with U0, at 650°C,

exploit. There 1s a report from previous work that stratification some-

times resulted from thermal cycling melts containing uranium or thorium.?2t

Other Types of Molten-Salt Reactor Systems

Direct Heat Exchange

Direct heat exchange between molten salts (fluorides or chlorides)
and liquid lead has been given consideration as an innovation which would
permit the use of a nearly static core and blanket system, with the liquid
lead circulated through the molten salt and an external heat exchanger.
Such a system would greatly lower the fuel inventory and might offer
better heat transfer characteristics. Blankenship, Kreyger, and Kirslis??
found that molten lead gave good phase separation after being dispersed
in molten chlorides or fluorides, even in the presence of solid oxides.

In an NaCl-KCl melt at about 500°C, sparging a suspension of lead in the
melt with an oXygen-containing gas caused the development of a red color
(from dissolved lead oxide); subsequent sparging with hydrogen reduced
the lead salt to finely divided lead, which coalesced into liquid lead
within 5 min. Many chemical studies remain to be made in the development
of direct heat exchange.

Alternative Coolants

The possibility of using a mixture of LiF, NaF, and KF as the coolant
for a molten-salt reactor has been discussed in the chapter on chemistry.

i»
315

UNCL ASSIFIED
ORNL --DWG 64 —3551

100 T
80 gi—g i |
- CE e E— j——8—_g_ " LOURINE
60 ——e———¢——¢ o——o—
40
/
20
. ° ZIRCONIUM 9
ol LITHIUM A4 O o N\ /
A ——A———— s\ O i
10 das : _._o”‘\/'-\ ) ;/
A N ..—-"l
8 ‘h.-e'-'-'— /A
. u/ Ty / A A —
‘T_-fi-/ A
[2] 6 L 4} L ] a
& / . BERYLLIUM
E
~ 4
AN /
'_.
& /
z /
o 2
5
st /.IRANIUM
»
1 /
II //
0.8
[ s’
o6 L /
¢
0.4 »
/ ZONE TRAVEL RATE: t cm/hr
74 ZONE MOVEMENT -
- . . /
0.2 =
- L
]
[
. 0.1
0 0.2 0.4 0.6 0.8 1.0

x, FRACTIONAL DISTANCE ACROSS INGOT

Fig. 12. Element Concentration (wt %) in LiF-BeF,-ZrF,-UF, Mixture,
vs Fractional Distance (x) Across Ingot, After 20 Downward Zone Passes.
316

This would require that the entire coolant system be resistant to the
action of molten fluorides. It has been recently suggested that molten
carbonates be used as the coolant fluid.?? Owing largely to the work of
Janz and his assoclates at RPI,24 a number of properties of molten
carbonates have been established. Figure 13 presents a phase diagram of
the ternary system Li,C03-Nap,C03-K;CO5. The eutectic of the composition
Li-Na-K (43.5-31.5-25.0 mole %), melts just below 400°C and should require
less than 1 atm of CO, cover gas for stability in the temperature range
of interest. The mixture has been used for a number of years as a heat-
treating medium and is known to be noncorrosive to steel at 1400°F over
many months of exposure. No obvious corrosion has been observed after
about 4000 hr exposure at 1200°F to INOR-8. In recent tests at ORNL,
Bettis found that molten carbonate is apparently quite stable toward
molten lead at temperatures of 900 to 1000°F. Development of molten
carbonate fluids will require study of radiation stability, corrosion,
thermal. conductivity and viscosity, and compatibility with fuel salts
(the consequence of a leak).

UNCLASSIFIED
ORNL-DWG 64-7146

NepCOa 858

COMPQOSITIONS IN male 7%
TEMPERATURES IN °C
MIN. 705°

£-49T Ay

[o}

Py
LisCO3 C5\/ \ \
726 £-482

LIKCOs \ NE-498 901
mp 5045 °C

Fig. 13. Ternary System LizCO;-Na;C0;-K,CO3.

Molten Salts as Fast Reactor Fuels

The availability of large amounts of plutonium and the potential
advantages in the use of fast neutrons for breeding have prompted con-
sideration of molten chloride fuels. GSeparated isotopic 27C). would be
used (at 98% enrichment) but would have to be conserved. Some informa-
tion is available on the binary systems of alkali metal chlorides with
UCls3, the valence state of uranium which, for compatibility reasons,
would be used in a chloride system. Additional study of the system
KC1-UCl3 is in progress at ORNL. From the available information on

"l

»
317

binary mixtures,25 Thoma has predicted the behavior of relatively simple
ternary systems.26 Figure 14 presents a predicted diagram for the sys-
tem NaCl-KCl-PuClz. Very substantial solubility of plutonium is pre-
dicted at temperatures between 500 and 600°C. Figure 15 presents a
similar diagram for the system NaClL-KC1-UCls, and, again, very substantial
solubilities of uranium are predicted for the temperature range 500 to
600°C. The similarities in the uranium and plutonium systems suggest

that it may be feasible to achieve the desired plutonium to uranium ratios
of 5:1, a heavy metal concentration of 30 to 55 mole %, and a liquidus
temperature below 550°C. Testing of these predictions will be included
in future chemical development.

The thermodynamic stability of metals and their chlorides in the
presence of UClsz and PuCls; has been the subject of calculations by

UNCLASSIFIED
ORNL-DWG 64—1994

PuClz~769

KClI
77

NaCl
802 ~ 658 :

Fig. 1l4. Predicted Phase Behavior for the System NaCl-KC1l-PuCls.
318

UNCLASSIFIED
ORNL-DWG 64-7147
UCl3 840

COMPOSITIONS IN mole %
TEMPERATURES IN °C
NaCi-KCi: AKAPQV
NoCl-UCIz: BARTON
KCI-UCl3: BARTON

NaCl
802 . . ~65 774

Fig. 15. System NaCl-KC1-UClj;, Phase Diagram Predicted on Basis of
Limiting Binary Systems.

rd

Newton?? and by Long.28 From the point of view of corrosion by the salt,
chromium is considered to be somewhat worse than iron, nickel is better,
and tungsten and molybdenum are much better; iron is considered to be
probably good enough. The combination of an iron—molten chloride—molten
lead system, based on available thermodynamic information, appears to be
acceptable chemically with respect to corrosion of the iron by lead, to
contamination of the melt by FeCl,, and to uptake of uranium by the
molten lead. Future chemical development will require laboratory and
intermediate-scale loop testing of these tentative conclusions.

References

1. R. E. Thoma et al., J. Phys. Chem. 64, 865 (1960).

2. L. V. Jones et al., Phase Equilibria in the LiF-BeF,-UF, Ternary
Fused Salt System, MIM-1080 (1959); J. Am. Ceram. Soc. 45, 79 (1962).

3. C. F. Weaver et al., Phase Equilibria in Molten Salt Breeder Reactor
Fuels. 1. The System LiF-BeF,-UF,-ThF,, ORNL-2896 (Dec. 27, 1960).

4. C. J. Barton, J. Phys. Chem. 64, 306 (1960).

5. MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 125.

6. J. K. Dawson and R. G. Sowden, Chemical Aspects of Nuclear Reactors,
vol. 3, Miscellaneous Topics, D. 103, Butterworths, London, 1963.

7.

14.
15.
16.
17.

18.

19.

20.

2.

319

W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys. Chem. 62, 862
(1958); M. Blander et al., Ibid., 63, 1164 (1959); G. M. Watson
et al., J. Chem. Eng. Data 7, 285 (1962).

R. B. Evans III, G. M. Watson, and E. A. Mason, J. Chem. Phys. 35(6),
2076 (1961); Ibid., 36(7), 1894 (1962); Ibid., 38(8), 1808 (19653;

R. B. Evans IIT, JaéifTruitt, and G. M. Watson,ng. Chem. Eng. Data
6(4), 522 (1961); R. B. Evans III, G. M. Watson, and J. Truitt, J.
Eépl. Phys. gg(é), 2682 (1962); Ibid., 34(7), 2020 (1963).

W. T. Ward et al., Solubility Relations Among Rare-Earth Fluorides
in Selected Molten Fluoride Solvents, ORNL-2749 (Oct. 13, 1959).

W. T. Ward et al., J. Chem. Eng. Data 5(2), 137—42 (1960).

Reactor Chem. Div. Ann. Progr. Rept., Jan. 31, 1962, ORNL-3262, p. 21.

P. J. Kreyger, Euratom, private communication, intralaboratory
memorandum to F. F. Blankenship, June 25, 1963.

A. J. Singh, R. G. Ross, and R. E. Thoma, "Vacuum Distillation of
LiF," Journal of Applied Physics (submitted for publlcatlon)

A. J. Singh, G. D. Brunton, and R. E. Thoma, Zone MEltlng of In-
organic Fluorides, ORNL-3658 (to be published).

Fluid Fuel Reactors, ed. by J. A. Lane, H. G. MacPherson, and Frank

Maslan, Chap. 22, p. 791, Addison-Wesley, Reading, Mass., 1958.

C. F. Baes, private communication, intralaboratory memorandum to
W. R. Grimes, June 4, 1964.

F. A.Cafasso, Harold M. Feder, and Irving Johnson, J. Phys. Chem.
68(7), 194448 (1964).

P. Chiotti and J. S. Klepfer, Transport of Solutes Between Liquid
Alloys in Mutual Contact with a Fused Salt — Application to Fuel
Reprocessing, 148th American Chemical Society Meeting, Chicago, Ill.,
Aug. 31-Sept. 3, 1964, Abstract 39 (to be presented).

R. H. Moore, Cation Effects on the Distribution of Uranium in the
System: Aluminum—Aluminum Chloride—Alkali Chloride, HW-67574 (Nov.
28, 1960).

J. H. Shaffer et al., Nucl. Sci. Eng. 18(2), 177-81 (1964).

H. G. MacPherson, MSRP Quart. Progr. Rept. Jan. 31, 1959, ORNIL-2684,
p. 111; G. J. Nessle and J. Truitt, Reactor Chem. Div. Ann. Progr.
Rept. Jan. 31, 1960, ORNL-2931, p. 17.

22 .

23.

T

25.

26.

27 .

28.

320

MSRP Semiann. Progr. Rept. Jan. 31, 1962, ORNL-3626, p. 126.

H. F. Bauman and E. S. Bettis, Selection of the Heat Exchange System
for Advanced Molten Salt Reactors (to be published).

George J. Janz and Mex R. Lorenz, J. Chem. Eng. Data 6(3), 32123
(1961). =

J. A. Leary, Temperatfire-Composition Diagrams of Pseudo-Binary Sys-
tems Containing Plutonium(IIT) Halides, LA-2661 (Apr. 9, 1962).

R. E. Thoma, Predicted Phase Behavior in Ternary Systems of Uranium
and Plutonium Chlorides, MSR-63-52 (June 6, 1963) (internal use
only).

R. F. Newton, Thermodynamic Stability of Metals and Their Chlorides
in the Presence of UCls and PuCls;, MSR-63-53 (Oct. 31, 1963) (internal
use only).

G. Long, private communication, intralaboratory memorandum to W. R.
Grimes, May 26, 1964.

ANATYTTCAL CHEMISTRY FOR THE MOLTEN-SATLT REACTOR

J. C. White

The program being carried out by the Analytical Chemistry Division

for the MSRE can be classified into two categories: (1) development and
evaluation of the methods required for the analysis of the MSRE fuel and
related problems and (2) investigation of the feasibility of various tech-
niques for in-line monitoring of the fuel. ‘

Development and evaluation of methods have occupied the major portion

of the program. This phase should be essentially completed by the end
of this period. In-line analysis is in the initial phases of investiga-
tion. Among the analytical techniques being considered for in-line moni-
toring are: electrochemical analyses of the molten state, electron spin
resonance, and x-ray measurements. It is not contemplated to use these
in-line measurements on the MSRE but to evaluate them for possible future
use on molten-salt reactors.
321

Methods Development

Sample Preparation

Since the MSRE fuel samples will be extremely radioactive, consid-
erable effort has been given to sampling, transfer, and handling of the
fuel sample. The maximum expected activity of the sample taken for anal-
ysis (at 10 Mw power level) will be of the order of 15 curies. The fuel
salt is also quite hygroscopic and consequently subject to hydrolysis.
Thus chemical as well as activity considerations are of paramount impor-
tance., The actual operation of sample preparation will be performed in
the High Radiation Level Analytical Laboratory hot cells using tools and
apparatus controlled by réemote master slave manipulators. This equipment
has been especially designed for the MSEE.

The sample will be obtained by dipping a copper ladle into the molten
fuel. The ladle is removed from the fuel, which is allowed to solidifly,
and. then placed: into a lead-shielded container for transport to the ana-
lytical examination cell. The transport container decoupling device is
shown in Fig. 1. The copper ladle, filled with approximately 10 g of salt
(Fig. 2), is then placed into the pulverizer-mixer and agitated for 45
sec on the mixer mill to crush and mix the salt. The top of the ladle
is removed prior to the operation by cutting with a specially built tubing.
cutter. The powdered salt 1s then removed from the pulverizer-mixer and
transferred to a polyethylene bottle by means of the equipment“shown in

Fig. 3. Portions of the salt can now be taken for the approprlate anal -
ysis. The elapsed time from receipt of sample to transfer: of prepared
salt sample is about 2 hr. :

Dissolution of Sample

The salt will be dissolved in a mixture of boric, nitric, and sul-
furic acids. It is then diluted with water to ensure an acidity of 0.5 M
H»S04. The approximate concentrations of the major constituents of the B
fuel in the diluted solution (1 g of salt per 100 ml of 0.5 M H,S0,) ares
1.2 Li, 0.7 Be, 0.1 U, and 1.0 Zr (these are expressed in mg/ml). Appro-
priate size aliquots are then taken for specific determinations. Approx-
imately 3 hr is required to weigh, dissolve, and dilute the sample.

Separate samples will be taken for reducing power, fluoride, and
oxide analyses. In order to avoid possible evolution of fluorine from
the salt sample upon prolonged standing due to radiation damage, the re-
maining salt will be weighed and dissolved in a like manner and set aside
for possible future use. '

Uranium

The determination of uranium is perhaps the most important analysis
to be made., A number of well-evaluated methods that are compatible with
hot-cell operation are available. Controlled-potential coulometryl will
be used as the primary method of analysis because of its inherent high
Fig. 1.

322

UNCLASSIFIED
PHOTO 64234

Transport Container Decoupling Device for MSRE Fuel.
323

UNCLASSIFIED
PHOTO €394

Fig. 2. Container for Sampling MSRE Salt.

UNCLASSIFIED
PHOTO 2746

Fig. 3. Apparatus for Removing MSRE Salt from Pulverizer-Mixer to
Polyethylene Sample Bottle.

UNCLASSIFIED
PHOTO 62742

Fig. 4. Cell Assembly for Coulometric and Amperometric Remote Ti-
trations of Uranium, Zirconium, and Chromium.
325

!

order of precision, easy adaptability to remote operation, and applica-
bility to sulfuric acid solutions. A titration cell (Fig. 4) which in-
cludes an S.C.E. reference electrode, a separated electrode of coiled
platinum wire contained in O.5 M H»80,, and a Teflon-sleeved platinum
wire to comnect to the mercury was designed for this method. A test so-
lution representing approximately 500 pg of uranium is added to the cell
which contains 3 ml of mercury and 4 ml of 0.5 M HpS04. The solution is
deaerated for 10 min while stirring. Essentially complete elimination
of interference is accomplished by a prereduction at +0.075 v. The U(VI)
is then reduced to U(IV) at —0.325 v. The progress of the titration at
the 10— eguivalent range is followed by means of an ORNL model Q-2514
high-sensitivity coulometer and continued until the potential is reduced
to approximately 5 pamp. The readout voltage due to the titration is
read from an attached potentiometer. From 50 to 500 pg of uranium can
be titrated with a precision of the order of less than 1%. The method
is essentially free of interferences under normal circumstances of MSRE
operation. An extraction procedure is available to separate uranium quan-
titatively in case a bias in the method is found.

Zirconium

An amperometric method? for the titration of zirconium with cupferron
has been developed for the determination of zirconium. A titration as-
sembly (Fig. 4) was fabricated in order to adapt this method to remote
manipulation. Zirconium forms a complex with cupferron in 0.5 M HpS04;
the concentration of this complex is measured electrically. The progress
of the titration is followed by an ORNL polarograph model Q-1160 with the
mercury pool at —0.5 v respective to the platinum. The titration is con-
tinued until a sharp, vertical inflection in the titration curve is ob-
served. The method can be performed without any prior separations with a
precision of about 1.5%.

Beryllium

Beryllium will be determined by gamma activation’ using a 124gh
source., The neutrons liberated by the 7y,n reaction are counted by means
of BF; tubes. Instrument settings are established at which the net sample
counting rate is at a maximum and is independent of high voltage and
pulse~height discriminator settings. A test aliquot, usually 5 mg of
beryllium in O.5 M Hp304, is exposed to the 124gY source, and neutrons
are counted until 10,000 counts are collected. A calibration factor, ob-
tained from standard solutions of beryllium, is established to permit
final calculations of concentration. Large (2-in.) 10BF3 neutron counter
tubes are used which increase the neutron detection efficiency of the in-
strument so that a gamma source activity of only about 300 me of 124gb
is necessary. Interference caused by elements which absorb neutrons at
this energy level can be eliminated by the cadmium shield technigue. The
precision of the method is about 2.5%.
326

Li

ium

Lithium will be analyzed, when required, by flame photometry. Since
this determination is relatively unimportant, no estimate of its precision
has been made as yet.

Fluoride

The pyrolytic determination of fluoride* is carried out in an appa-
ratus designed for remote operation (Fig. 5). A nickel reaction tube is
first heated to 1000°C, and moist oxygen is passed through it at a rate of
about 2 liters/min. A test portion, usually 100 mg of the salt, is placed
in a nickel boat and mixed with about 3 g of powdered U3Og catalyst. The
boat is inserted in the heated reaction tube, which is immediately sealed
to prevent loss of fluoride. The evolved fluoride is trapped in a known
quantity of sodium hydroxide. After about 30 min the absorber solution
is removed from the apparatus, and the excess caustic is titrated with
standard hydrochloric acid. The fluoride is equivalent to the sodium hy-
droxide neutralized by the absorbed hydrogen fluoride. The precision is
of the order of 1%.

NCLASSIFIED
PHoTO 41000

Fig. 5. Apparatus for Pyrolysis of Fluoride for Hot-Cell Operation.
327

Chromium

Since chromium serwves as a means of monitoring the corrosion by the
fuel, a reliable and accurate method of analysis is required. This has
been achieved? by development of an amperometric titration with Felt, A
titration cell assembly which includes a pyrolytic-graphite electrode and
an S.C.E. was designed for this method (Fig. 4). Chromium, in 5- to 50-ug
quantities, is oxidized to Cr(VI) with argentic oxide in 0.5 M H2804. The
Cr(VI) is subsequently reduced to Cr(III) by titration with Fe®*t; this
reduction causes a decrease in diffusion current, which is measured am-
perometrically. The progress of the titration is followed by an ORNL po-
larograph, model Q-1160, with the pyrolytic-graphite electrode at +1.0
respective to the S.C.,E. The titration is continued until a sharp de-
flection in the curve is observed. The titration of chromium can be per-
formed without any prior separations, and the precision is of the order
of 1.5%.

Iron, Nickel, and Molybdenum

Thesé corrosion products will be determined by emission spectroscopy
or by spectrophotometric procedures. ©Standard methods are available for
this application.

Fission Products

The fission products will be determined by standard radiochemical
procedures.

Oxide

Knowledge of the oxide concentration of the MSRE fuel is highly de-
sirable in the course of operation of the reactor. The determination of
small amounts of oxide in the MSRE fuel is, unfortunately, extremely dif-
ficult and complicated due to problems imposed in handling highly radio-
active materials. The high-pressure high-temperature fluorination method
used on nonradioactive salts is not applicable because of hot-cell limi-
tations. As a result, most effort has been devoted to the inert-gas fu-
sion techniq_ue,7 in which the salt is confined in an enclosed graphite
cup and heated by induction to about 2400°C. Under these conditions the
oxides react with the graphite to form CO, which is then measured gas
chromatographically after oxidation to COz. This method is still being
studied; pure ZrO, has been used as a standard source of oxygen since
much, if not all, of the oxide in the MSRE fuel will probably be present
in this form.

Reducing Power

The reducing power of the fuel is considered to be an indication of
its trivalent uranium concentration. The method used is not specific for
U3*; any element or compound capable of reacting with the hydrogen ion
to liberate hydrogen will be included in the reducing-power value. The
328

method currently under investigation is that of reacting the salt with
tritiated HCl to liberate tritium as well as hydrogen. The two gases are
converted to water and collected, and the tritium is counted by scintil-
lation techniques. The method is capable of the great sensitivity which
1s needed to detect trace quantities of trivalent uranium, '

In all these procedures, such factors as ease of installation, ma-
nipulation, and replacement of equipment and apparatus have been given
foremost consideration. The size of equipment and apparatus is held to
a minimum since hot-cell space is drastically limited. Manual operations
that must be performed with master slave manipulators are also reduced to
a minimum,

Reactor Off-Gas Analysis

A gas chromatograph featuring the high-sensitivity helium breakdown
detector was specially designed for installation in the sample station
of ORNL-MTR-47-6. The objectives of this instrument were (1) to determine
sub-ppm concentrations of CF, together with ppm concentrations of krypton
and xenon emerging from the capsules; (2) to detect traces of permanent
gas impurities in the helium purge gas; and (3) to obtain data essential
for the design of a chromatograph for the analysis of off-gas from the
MSEE reactor.

The essential feature of this chromatograph is the helium discharge
detector in which the voltage that is developed across the gas gap be-
tween two closely spaced electrodes reflects the gas composition., With
pure helium in the detector and currents in the microampere range, a po-
tential of about 500 v is developed across the electrodes. This potential
is reduced by as much as several hundred volts by the passage of contami-
nants eluted from a chromatographic column. The sensitivity of the de-
tector is limited primarily by the noise level so that under optimum con-
- ditions certain permanent gas impurities can be detected at concentrations
as low as 5 ppb. During periods of stable operation, detection limits
are about 3 ppb for krypton, 5 ppb for xenon, and 30 ppb for CF,.

The following data were obtained with the instrument from the in-pile
test:

l. Permanent gas contaminants were present in the purge gas in much
greater than anticipated concentrations. Hydrogen, oxygen, nitrogen,
methane, and occasionally CO were observed, frequently in concentra-
tions above the range of the instrument (estimated >200 ppm). The
purification system removed at least 99% of the oxygen and about 90%
of the nitrogen. Oxygen, when present, was almost quantitatively re-
moved on passage through the capsules.

2. The concentration of fission gases in the capsule purge was subject
to rapid fluctuations, sometimes severalfold, apparently either the
result of bubbling of the fuel or of flow perturbations introduced
by check valves. The concentration of krypton paralleled that of
329

xenon so that an Xe/Kr ratio of 4.2 was maintained in agreement with
- mass spectrographic measurements.

3. No CF,; was detected in the purge gas (<250 ppb) at maximum insertion
and low purge rates {400 cm3/hr). Moreover, CF, was not detected in
capsule effluent when the capsules were purged with helium containing
1 ppm of CF,.

4. No evidence of adverse effects on the chromatograph or changes in
sensitivity of the detector resulting from radioactivity were ob-
served. Standard additions of krypton and xenon through the capsules
were recovered within the reproducibility of the fission-gas-genera-
tion rate.

5. PFor a particular current the response for krypton and xenon remained
_constant within about 10% despite variation in temperature and radio-
activity. The response for CF,; was variable over a factor of 2 and
depended on the immediate history of samples analyzed.

6. As had been noted in laboratory tests, reducing gases such as hydrogen
and methane frequently introduced violent base-line noise, while CF,
diminished this noise.

On the basis of this experience it is concluded that this type of
instrument offers considerable promise for applications to the MSRE re-
actor. The detector should be modified to provide improved stability
for CF,; analysis. (Tests of various electrode materials are in progress., )
In addition, a multicolumn instrument, which would permit the determina-
tion of €O, and possible resolution and identification of the unknown
contaminants, should be considered. An all-metal sampling valve should
be dewveloped to provide long-term service at higher activity lewvels.

In-Line Analysis

The advantages of in-line analysis and control of the fuel chemistry
are so obvious and appealing that a portion of the analytical effort has
been turned toward this objective. DNot the least of such advantages is
the immediate knowledge of chemical behavior of the fuel. Unfortunately,
the magnitude of the task is as great as its potential. Much basic in-
formation is required before adequate assessments of any of the techniques
can be made. FElectrochemical analysis is, in theory, adaptable to in-
line control and is especially attractive with regard to direct analysis
of corrosion products and other electroactive species in the molten MSRE
fuel. The electrochemical behavior of iron8 and nickel? in fluorides has
been reported. Investigations are under way on the characteristics of
chromium, Oxide analysis is also possible, and efforts are being made
in this direction also. Solid electrodes made from such materials as
pyrolytic graphite, "glassy" carbon, and pyrolytic boron nitride are be-
ing used in molten-salt solutions and offer much potential also. In this
area lie the most promising prospects for in-line analysis.

Feasibility studies of the possibie determination of oxidation states
of uranium in the MSRHE fuel by electron-spin-resonance measurements have
330

been carried out. Initial results are mildly encouraging. Adequate sen-
sitivity is only achiewved at very low temperatures (77°K) which poses ob-
vious difficulties when applied to a fuel at high temperature., X-ray
fluorescence and absorbance measurement techniques are also under study.

References

l. L. G. Farrar, P, F., Thomason, and M. T. Kelley, Anal. Chem. 30, 1511
(1958). —

2. Hisashi Kubota and J. G. Surak, Anal. Chem. 35, 1715 (1963).
3. G. Goldstein, Anal. Chem. 35, 1620 (1963).

4e R. F. Apple, Fluoride in MSRE Fuel, Pyrolysis Method, Method No.
9 021203, July 1964, ORNL Master Analytical Manual.

5. R. F. Apple and H. E. Zittel, Anal. Chem. 36, 983 (1964).

6. G. Goldberg, A. S. Meyer, Jr., and J. C. White, Anal. Chem, 32, 314
(1960). —

7. E. J. Beck, Determination of Oxygen by the Inert Gas Fusion Method
Using Graphite Capsules, Parma Research Center, Report URS-29 (Feb.
1, 1961).

8. D. L. Manning and Gleb Mamantov, J. Electroanal, Chem, 7, 102-8
(1964). ~

9. D. L. Manning, J. Electroanal. Chem., 7, 302-6 (1964).

METALLURGICAL DEVELOPMENTS

A, Tabdada

Introduction

The development of molten-salt reactors that circulate fluoride
mixtures depends on the availability of a material of construction that
will contain the mixtures for the life of the reactors and also afford
the required structural properties at operating temperatures. In the
MSRE, the material must be compatible with a graphite-fluoride system
and resistant to oxidation in air or air-nitrogen mixtures as well as
resistant to corrosion of fluorides; all while in the presence of a
neutron flux. Since the reactor is a dynamic system that operates at
about 1200°F, the material should have good elevated-temperature strength
L 1]

331

and ductility. It is further required that the material be metallurgi-
cally stable in this environment of high temperatures, corrosive fluids,
and irradiation. Finally, such a material must be Tfabricable into large
and complex engineering structures, such as pumps and heat exchangers.
Thus, it must be easily welded, cast, and worked into such forms as tubes,
plates, forgings, -etc.

The container material® selected for the MSRE is the alloy INOR-8,
also known as Hastelloy N and INCO. This alloy was developed at ORNL in
the Aircraft Nuclear Propulsion Program and was viewed as the most
promising container material for molten fluorides exposed to the severe
(1500°F) ANP conditions. The properties of this alloy are considered to
be more than adequate for MSRE application. Considerable effort and ex-
pense were devoted by the ANP Program to determine general metallurgical
and fabrication data for INOR-8. Although several changes in the chemistry
of the alloy appeared desirable in order to make it more economical and
give it improved properties under MSRE conditions, these changes were
precluded by the adequacy of the material, the availability of INOR-8
data, and the cost and time required to obtain metallurgical data for a
new alloy. INOR-8 has therefore been selected as the structural material
for all parts of the MSRE exposed to molten salts.

The metallurgical developments in support of the MSRE have been
limited to establishing the compatibility of INOR-8 in a molten-salt—
graphite system at MSRE conditions and to generating the metallurgical
information necessary to design, fabricate, and operate the reactor.
Included in this work was the preparation of standards for materials
procurement and for the fabrication of components. A description)of these
metallurgical achievements is presented.

Description of INOR-8

INOR-8 is a nickel-base alloy (see Table 1) that is solution-
strengthened with molybdenum and has sufficient chromium to be oxidation
resistant; however, the chemistry has been controlled to preclude aging
embrittlement. The aluminum, titanium, and carbon contents are limited
to minimize severe fabrication and corrosion problems, and the boron
content is limited to prevent weld cracking. Iron is included to allow
flexibility in starting materials during melting. The extreme examples of
permissible combinations of elements allowed by the chemistry specification
were studied, and in no case did any undesirable brittle phases develop.
Carbides of the form Ms3Cg and MgC exist in the alloy and are stable to
at least 1800°F.

Physical properties of INOR-8 are shown in Table 2. Specific heat,
electrical resistivity, thermal expansion, and thermal conductivity data
all show inflections in temperature curves, indicating that an order-
disorder reaction may be occurring in this alloy at approximately 1200°F.
332

Table 1. Chemical Composition Requirements for the
Ni-Mo-Cr Alloy — INOR-8

Element Percent®
Nickel Remainder
~ Molybdenum 15.00-18.000
- Chromium 6.00-8.00
" Iron 5.00
Carbon 0.04-0.08
Manganese 1..00
Silicon 1.00
Tungsten 0.50
Aluminum + titanium 0.50
Copper 0.35
Cobalt : 0.20
Phosphorus - 0.015
sulfur 0.020
Boron | ‘ 0.010
Others, total 0.50

aSingle values are maximum percentages unless
otherwise specified.
333

Table 2. Physical Properties of INOR-8

Density, Electrical Resistivity, Thermal Conductivity,
o P k
3 .. 3 Temperature P Temperature —2 1wl fomye—l
Temperature  (g/em®)  (1b/in.?) (°5) (ohmecm) (°7) [Btu £t72 hr™* (°F)* rt]
RT 8.93 0.320 . RT 120.5 300 6.94
| 1300 126.0 575 8.21
1500 124.1 825 9.25
985 10.40
1165 11.56
1475 13.87
Specific Heat Coefficient of Thermal Expansicn
- Temperature o y—1 Temperature . . =1 fooy—l
(°F) [Btu 1b (°F)~t] (°F) [in. in. (°F)~t]
, 140 0.098 212-752 7.0 x 10™®
572 0.10% 752-1112 8.4 x 1078
1000 0.115 1112-1832 9.9 x 107°
1292 0.138 212-1832 8.6 x 107°
Modulus of Elasticity, E
TemPSTATTS p (mm/in.? x 109)  TRSESTERC g /in 2y 109)
55 31.5 1475 23.7
425 29.0 1575 22.7
775 28.0 1650 21.9
925 27.0 1750 20.7
‘1075 26.3 1825 - 9.1
1175 26.0 1925 17.7
) 1300 24,8
. Corrosion Resistance of INOR-8 to Molten Salts

Corrosion of metals in salt mixtures occurs by oxidation of the
metal components to their fluoride salts, which are readily dissolved by
the molten fluorides. This precludes any possibility of protection by
passivation of the metal surface by corrosion products, and the extent of
corrosion is then controlled by the thermodynamic driving force of the
oxidation reaction. Oxidation reactions stem from three sources, which
are indicated in Fig. L. These reactions involve (1) impurities in the
salt, (2) impurities on the metal, and (3) components of the salt (UF,).

In the case of INOR-8 and similar multicomponent systems, corrosion
is manifested by the selective oxidation and removal of the least-noble
constituent. In INOR-8 this element is chromium. When it becomes
depleted, subsurface voids form in the depletion zone.
334

UNCLASSIFIED
ORNL-DWG 64—9678

1. Impurities in the Melt:

Example:

FeF2 +Cr —> CrF2+ Fe

Other Impurities;

N|F2, CrF CrF4, FeF3, HF, and UF

3’
2. Oxide Films on Metal Surfaces:

5

Example:

2Fe203+ 3 CrF4 —> 3Cr0‘,2 + 4 FeF3

Importance:

FeF:3 now reacts with Cr by 1

3. Constituents in the fuel;

Example:

Cr+ 2UF, —> 2 UF:3 + CrF2

4

Fig. 1. Types of Reactions Associated with Chromium Removal.

An extensive corrosion program was conducted to determine the extent
to which nickel alloys would be attacked by molten salts in polythermal
systems. The program was conducted in three sequential phases. In the
first phase, the corrosion properties of 13 fluoride salt mixtures were
compared in INOR-8 thermal-convection loops operated for 1000 hr. Specific
salt mixtures, whose compositions are listed in Table 3, were selected to
provide an evaluation of (1) the corrosion properties of beryllium-bearing
fuels, as compared with earlier property data amassed for zirconium-base
fuels; (2) the corrosion properties of beryllium-fluoride mixtures con-
taining large quantities of thorium; and (3) the corrosion properties of
non-fuel-bearing fluoride coolant salts.

The second phase of testing, which again used thermal-convection
loops, subJjected systems that demonstrated their compatibility in phase 1
tests to more extensive investigations for longer time periods and at two
temperature levels. The third phase of out-of-pile testing was conducted
in forced-circulation 100ps2 at flow rates and temperature conditions
simulating those of an operating reactor system. The temperature condi-
tions employed in each test phase are shown in Table 4.
335

Table 3. Compositions of Molten-Salt Mixtures To Be Tested for
Corrosiveness in Inconel and INOR-8 Thermal-Convection Loops

Composition (mole %)

Salt
Designation NaF LiF KF 2rF, BeF, UF,  ThF,
Fuel and Blanket Salts

122 57 42 1
129 55.3 40.7 4
123 53 46 1
124 58 35 7
125 53 46 0.5 0.5
126 53 46 1
127 58 35 7
128 71 29
130 62 37 1
131 60 36 4
133 71 16 13
134 62 36.5 0.5 1
135 53 45.5 0.5 1
136 70 10 20

BULT-14 + 0.5U0 67 18.5 ‘0.5 14

Cooclant Salts

12 11.5 42
84 27 38

Table 4. Operating Temperatures To Be Used in Various Phases
of Corrosion Testing

Operating Temperatures (°F)

Difference Between

Maximum Hot and Cold
Sections
Phase 1
Fuel mixtures 1250 170
Coolant salts 1125 125
Phase 2
Fuel mixtures 1250-1350 170
Coolant salts 1050-1250 170
Phase 3
Fuel mixtures 1300 200

Coolant salts 1200 200

336

Initial screening studies of INOR-8fluoride salt systems produced
no measurable attack during the 1000-hr test interval. Results of these
tests are summarized in Table 5. Accordingly, the majority of salt sys-
tems developed for phase I testing were included in the phase II corrosion
program. Results of phase II tests, which were conducted for 8760 hr,
are summarized in Table 6. The maximum attack in phase II tests was
limited to surface roughening and pitting to a depth of 1/2 mil. Figure 2
is representative of the attack found in these loops. Attack in most
cases was accompanied by the formation of a thin surface layer which is
discussed below in conjunction with phase III tests. Examination of cold
regions of these loops revealed no deposits or other evidence of mass
transfer.

Chemical analyses of salt samples removed from hot- and cold-leg
sections of phase I and II tests showed increases in chromium concentra-
tion of 50-100 ppm. Nickel and iron concentrations tended to either hold
steady or drop slightly  during loop operation. .

In the third phase of the program, 15 of the forced-convection loops
were made of INOR-8 and 9 of Inconel. Table 7 gives a summary of the
duration of these tests. Tubular inserts in the heated sections of some
of the loops provided information about the weight losses occurring during
the tests; salt samples taken from the pump bowls provided a semicontinu-
ous indication of corrosion product concentration in the circulating
system.

A summary of the operating conditions and the results of metallo-
graphic examination of the INOR-8 loops are presented in Table 8. Metal-
lographic examination of inserts removed after 5000 hr of loop operation
showed no evidence of surface attack. When tests exceeded 5000 hr, how-
ever, a thin, continuous surface layer was evident along exposed sur-
faces, as shown in Fig. 3. In the most severe case of attack found in a
20,000-hr test, this layer reached a maximum depth of 2 mils and was
found to contain subsurface voids (see Fig. 4). No transition or diffusion
zone was apparent between the layer and the base metal, and analysis of
the layer showed it to be composed predominantly of nickel with smaller
amounts of iron, chromium, and molybdenum. The layer had a composition
similar to the intermetallic phase Ni/Mo.

3

Increasing the maximum temperature of the forced-convection loop
from 1300°F to 1400°F did not vary the results. However, in loops with -
1500°F maximum temperature, surface attack was manifested by the appear-
ance of subsurface voids to a depth of 4 mils, as shown in Fig. 5, and
the surface layer was not observed. '

. The weight losses of corrosion inserts contained in the hot legs of
loops at 1300, 1400, and 1500°F are shown in Table 9. At MSRE tempera-
tures, weight losses were 2 to 5 mg/cm2 and showed little change after
5000 hr of loop operation. No measurable changes were found in the wall
thicknesses; however, if uniform attack is assumed, weight losses indicate
wall reductions-of 2 to 12 u. ‘
337

Table 5. Operating Conditions and Results of-Phase I INOR-8

Thermal-Convection Loop Tests

(operating time: 1000 hr)

Chromium Concentration

Temperature Depth of of Salt Mixture

fig?p ;z%t of Ho%ogiction ?EE?S% | (ppm) |
| Before After
Test Test .
1203 125 1250 0 72 235
1204 131 1250 0 30 79
1221 128 11250 0 37 70
1164 124 1250 0 200 300
1165 12 1250 0 |
1179 130 1250 0 120 10(?)
1197 126 1250 0 15 150
1205 129 1250 0 20 163
1227 134 1250 0 265 315
1228 133 1250 0 % 78
1229 135 1250 0 190 193
1241 136 1250 0 90 49
1242 136 1350 ~1/2 80 209
1195 84 . 1125 0 30 30
1194 12 1125 0 68 284

338

Table 6. Results of Metallographic Examination of Phase Il INOR-8 Thermul-Co_nvecfion-' Loops

Maximum . . e
Loop Te.sf Fluid-Metal Salt Metallographic Examination
Period
No. (hr) Interface No. Hot-Lea Appearance Cold=-Leg
Temperature (°F) or=Leg App Appearance
1162 6360 1250 123 Moderate surface roughening and pitting No attack
to 3/4 mil '
1224 8760 1350 123 Moderate surface roughening and pitting No attack
to 1 mil
1226 8760 1250 131 Moderate surface roughening, film 1/2 Light surface
to | mil thick roughening, with
" 1/2-mil corrosion
film
1231 8760 1250 134 Light surface roughening and pitting to No attack
< 1/2 mil
1238 8760 1350 134 Light surface roughening No attack, very
‘ shallow corrosion
film present
1244 8760 1250 135 Light surface roughening, moderate voids No attack
and pits to < 1 mil
1246 8760 1350 135 Light surface roughening No attack
1233 8760 1250 133 No attack; very shallow corrosion film No attack
1240 8760 1350 133 Light surface roughening No attack
1216 8760 1350 127 Moderate surface roughening and pitting No attack
to < 1/2 mil, light voids to <2 mils
1219 8760 1350 122 Light surface roughening No attack °
1200 8760 1250 130 Light surface roughening, pits less than Light surface
1 mil roughening
1196 = 8760 1350 130 Light surface roughening No attack
1185 8760 1250 126 Light surface roughening No attack
1206 8760 1350 126 Light surface pitting No attack
1212 8760 1250 125 Moderate pits < 1 mil in'depth No attack
1215 8760 1350 125 Moderate surface roughening Light surface
. roughening
1209 8760 1350 128 Heavy surface roughening No attack -
1190 8760 1250 127 Light surface roughening, pits less than No attack
1 mil
1208 8760 1250 12 Moderate surface roughening and pits up No attack

to 1 mil deep

Fig. 2.

339

UNCLASSIFIED

T-18961

Typical Appearance of Metallographic Specimen at Hot Le,
g

of INOR-8 Thermal-Convection Loop in Phase II Tests.

Table 7.

Summary of Forced-Convection Loop Operation

INOR-8 Inconel
Length of Test
(hr) Number of Total Hours Number of Total Hours

Tests Operated Tests Operated
3,000~ 5,000 2 6,440
6,000-10,000 6 52,680 5 44,760
11,000-15,000 4 54,370 2 28,190
16,000-20,000 5 99,950
Total hours accumulated 207,000 79,390

Table 8. INOR-8 Operating Conditions of Forced-Convection Loops and Results of Metallographic
Examinations of Loop Materials
Maximum
| Duration Fluid-Metal Flow :
Egop of Test Salt Mixture Interface (fg) Reyfiglds Rate Results Eiafiizziigiraphlc
. (hr) Tempsrature . (gpm)
(°F)
9354-1 14,563 126 1300 200 2000 2.5 Heavy surface roughening and pitting
| to 1-1/2 mils '
9354-3 19,942 84 1200 100 3000 2.0 No attack, slight trace of metallic
o deposit in cooler coil
9354=-4 15,140 130 1300 200 3000 2.5 No attack
9354-5 14,503 130 1300 200 3000 2.5 No attack
MSRP-6 20,000 134 1300 200 2300 1.5 Pitted surface layer to 2 mils
MSRP-7 20,000 133 1300 200 3100 1.8 Pitted surface layer to 1 mil
MSRP-8 9,633 124 1300 200 4000 2.0 No attack
MSRP-9 9,687 134 1300 200 2300 1.8 No attack
MSRP-10 20,000 135 1300 200 3400 2.0  Pitted surface layer to 1/2 mil
MSRP-11 20,000 123 1300 200 3200 2.0 Pitted surface layer to 1 mil
MSRP-12 14,498 134 1300 200 2300 1.8  No attack -
MSRP-13 8,085 136 1300 200 3900 2.0 Heavy surface roughening and
pritting _
MSRP-14 9,800 BULT-14 + 0.5U 1300 200 Pitted surface layer to 1/2 mil
MSRP-15 10,200 BULT-14 + 0.5U 1400 200 Pitted surface layer to 2/3 mil
MSRP-16 6,500 BULT-14 + 0.5U 1500 200 Moderate subsurface void formation

to 4 mils

O%7¢
UNCLASSIFIED
T-19555

Fig. 3. Typical Appearance of Metallographic Specimen at Point of
Maximum Wall mperature of INOR Forced-Convection Loop.

UNCLASSIFIED
T-20038

Fig. 4. Metallographic Specimen, Showing Maximum Corrosion Observed
in Forced-Convection Loop Operating at 1300°F for 20,000 hr.

342

Fig. 5. Typical Appearance of Corrosion at Point of Maximum Temper-
ature of Forced-Convection Loop Operating at 1500°F,

Equivalent

1300 1300

1300

343

Chemical analysis of salt samples that were periodically withdrawn
from the loop showed a slight increase in chromium concentration, while
concentrations of other metals remained unchanged. During a typical run,
the chromium concentration reached an asymptotic limit after about 5000
hr of operation, as shown in Fig. 6. This limit was between 300 and 500
ppm at 1300°F and between 600 and 800 ppm at 1400 and 1500°F.

The corrosion data for INOR-8, exposed in the pump loops, indicate
that corrosion reactions with fluoride salts at MSRE temperatures are
essentially complete within the first 5000 hr of operation, do not
generally exceed 1 mil even after 20,000 hr, and are probably associated
only with impurity reactions. No significant increase in corrosion
occurs due to temperature increase until about 1500°F. At this tempera-
ture, noticeable depletion of chromium occurs, as indicated by subsurface
voids, and the corrosion mechanism associated with components of the salt
is probably effective. The safety factor of the corrosion resistance of
INOR-8 appears to be ample, even for the temperature excursions or hot
spots that could occur during the MSRE reactor operation.

Compatibility of Graphite and INOR-8 in Molten Salt

Since the MSRE uses a graphite moderator in direct contact with
molten-salt fuel, the compatibility of graphite and INOR-8 in a fluoride
fuel was investigated.

The tendency for INOR-8 to be carburized was investigated in static
pots containing LiF-BeF,-UF, (67-32-1 mole %) and graphite at 1300°F for
times ranging from 2000 to 12,000 hr.

UNCLASSIFIED
ORNL-LR-DWG S58961R

T SALT COMPOSITION: LiF -BeFp-ThFg-UF4 (67-18.5-14-0.5 mole %)
S MAXIMUM METAL INTERFACE TEMPERATURE: 760°C
— LOOP AT: 200°F
z | | | |
=) | | ® Ni ) I
= i 2310 hr,LOOP 8770 hr,LOOP i
s 290 hr,LOOP (SOTHERMAL 4 Cr DOWN 195 hr TO  LOOP TERMINATED
b ISOTHERMAL FOR 29 hr ® Fe REMOVE INSERT  AFTER 10,878 hr
o FOR 14 hr I l OPERATION
S 800 ] | ' : i
9 : 12400 hr,LOOP A t I 4 4
© ! | SOTHERMAL ! }
3 ; | I} 3e50nrL00P I |
& i ,, ISOTHERMAL | !

4 I 1 FOR 11 hr I I
z 400np% |
5 i I !
o = & I | | |
w [ | |
g 200“ l 11 { I 3 B

L)
o i I i a s | |
© tibo | |
0" . ¢ |l 4 . ) I > ¢
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 40,000 11,000 12,000
TIME {hr)

Fig. 6. Concentration of Corrosion Products in Fluoride Salt Cir-
culated in an INOR-8 Forced-Convection Loop.
344,

- No carburization was metallographically detected on specimens from
any of the above tests. Tensile properties were determined on specimens
included in each test, and these properties are listed in Table 10. A
comparison of the tensile strengths and elongation values of the tested
specimens with those of the control specimens indicates that no signifi-
cant change occurred.

These data indicate that unstressed INOR-8 is not detectably carbu-
rized in the fuel-graphite system at the conditions described above. .

For the purpose of evaluating the compatibility of graphite and
INOR-8 in a circulating fluoride fuel, an INOR-8 forced-convection loop
of the design shown in Fig. 7 was operated* for 8850 hr. The loop oper-
ated at a maximum temperature of 1300°F and circulated a fluoride mixture

of LiF-BeF,-UFy.

/ Upon completion of the test, components from the loop were checked
metallographically and chemically, and specimens were checked for dimen-
sional changes and weight changes. The tests indicated that

Table 10. Tensile-Test Results on INOR-8 Specimens Exposed to
Fuel-Graphite Systems Compared with Control Specimens
Exposed to Argon for Various Times

Control Specimens® Test SpecimensP
Time of Test Tensile . Tensil
Strength Elongation Stre;gih Elongation
(psi x 10~2) (% in 2 in.) (psi x 10=2) (% in 2 in.)

Room-Temperature Tests

2,000 123.2 42.5 124.8 41.0
4,000 123.0 42.0 124.5 43.0

6,000 124.7 39.0 1242 36.0

g,000 127.7 43.5 128.2 36.5
10,000 130.8 43.0 130.3 41.0
12,000 130.6 36.5 126.7 36.3

1250°F Tests

2,000  74.6 18.0 75.6 . 18.5

4,000 75.9, 76.0 19.0, 20.0 76.2, Th.3 18.5, 18.5
6,000 2.8 16.5 7405 16.0 J
g ,000 76 .7 17.5 76.6 19.0
10,000 78.9 15.0 80.8 17.5
12,000  72.8 15.5 75.8 17.0

a
Control specimens were exposed to argon at 1300°F for the times in-
dicated. '

b ) '
Test specimens were exposed to LiF-BeF,-UF, (62-37-1 mole %) and
graphite at 1300°F for the times indicated.

b
UNCLASSIFIED
CRNL-LR- DWG 39365R

GRAPHITE CONTAINER

DETAIL "AM

SAMPLE LEG
FREEZE VALVE 17

DRAIN TANK 15

Fig. 7. Molten-Salt Corrosion Loop 9354-5 Containing Graphite Spec-
imens, Showing Thermocouple and Specimen Locations.,

43
346

1. There was no corrosion or erosion of the graphite by the flowing
salt.

2. There was very little permeation of the graphite by the salt, and the
permeation that occurred was uniform throughout the graphite rods.

3. The various INOR-8 loop components exposed to the salt were not
carburized.

4. The INOR-8 components exposed to the salt and graphite were negli-
gibly attacked..

5. With the possible exception of oxygen contamination, the salt appeared .
to have undergone no chemical changes as a result of exposure to the
graphite test specimens.

Oxidation Resistance of INOR-8

The oxidation resistance of nickel-molybdenum alloys depends on the
service temperature, the temperature cycle, the molybdenum content, and
the chromium content. The oxidation rate of the binary nickel-molybdenum
alloy passes through a maximum for the alloy containing 15% Mo, and the
scale formed by the oxidation is NiMoO, and NiO. Upon thermal cycling
from above 1400°F to below 660°F, the NiMoO, undergoes a phase transfor-
mation which causes the protective scale on the oxidized metal to spall.
Subsequent temperature cycles then result in an accelerated oxidation
rate. Similarly, the oxidation rate of nickel-molybdenum alloys contain-
ing chromium passes through a maximum for alloys containing between 2 and
6% Cr. Alloys containing more than 6% Cr are insensitive to thermal
cycling and to the molybdenum content because the oxide scale is predomi-
nantly stable Crp03. An abrupt decrease (by a factor of about 40) in the
oxidation rate at 1800°F is observed; then the chromium content is in-
creased from 5.9% to 6.2%.

The oxidation resistance of INOR-8 is excellent, and continuous oper-
ation at temperatures up to 1800°F is feasible. Intermittent use at tem-
peratures as high as 1900°F could be tolerated. For temperatures up to
1200°F, the oxidation rate is not measurable; it is essentially zero after
1000 hr of exposure in static air, as well as in nitrogen containing small
quantities of air (the MSRE cell environment). It is estimated that oxi-
dation of 0.00L to 0.002 in. would occur in 100,000 hr of operation at
1200°F. The effect of temperature on the oxidation rate of the alloy is
shown in Fig. 8.

Mechanical Properties of INOR-8

Mechanical properties tests were performed on eight commercial-size
heats of INOR-8. Tests were conducted at ORNL,5 the Haynes Stellite
Company,® the Battelle Memorial Institute,” and the University of
Alsbama.® Résults indicated that the strength of INOR-8 compares favor-
ably with austenitic stainless steels.
347

UNCLASSIFIED
ORNL -DWG 64-2855R

5.0 |
: P
4.5
INOR -8
Fay
. 4.0 /
fa
- 35 1900°F,
NI—-
o
- g 3.0
~ /
o -
E /
8. 25 .
z /
< .
o
- /
I
L 2.0
(V0 -
E s
15 b
faY
1.0 A
fay
/"
o 1800:!:_._—--'_“’.
0.5 g A O
- . .—.’
,—.__—«/
—T" 1600°F~ o
- ’_.__.—0"’ \ 1700°F o
ol_——"’-'—;——-e—-fi—'*?—'_—— o - A 4
N 0 50 100 150 200 250 300 350 400
TIME (hr)

Fig. 8. Effect of Temperature on Oxidation of INOR-8.
348

Five of the heats were experimental ones and were used to establish
design data for the alloy. These data were reviewed by the ASME Boiler
and Pressure Vessel Code Committee, and INOR-8 was approved for code
construction in Code Case 1315, subject to the allowable stresses cited
in Table 11. The original design values used for the MSRE were approxi-
mately 30% below these values to allow an added safety factor for possible
unknown irradiation effects.

~ The remaining three heats, randomly selected from material used to
fabricate MSRE components, were tested to evaluate the effects of large-
scale production-and improved quality requirements. In general, the
mechanical properties of these three heats were significantly better than
those of the earlier experimental heats, from which design data were
established.

Tensile Properties

Tensile tests were performed in air using sheet- and rod-type speci-
mens. Figure 9 is a summary of the ultimate tensile strengths at temper-
atures from ambient to about 1800°F. The minimum specified tensile
strength for purchase specifications is included. The 0.2% offset yield
strength over the same temperature range is shown in Fig. 10. The values
for elongation and for reduction in area are illustrated in Figs. 11 and
12 respectively. Metallographic data indicated that heats with very low
carbon and, consequently, large grain size tend to exhibit lower tensile
and yield strengths. Heats with similar grain size showed comparable
values at all temperatures.

Tensile tests of notched specimens were performed using a notch

radius of 0.005 in. The notched-to-unnotched strength ratios varied from
1.07 to 1.38 at test temperatures from ambient to 1500°F.

Creep Properties

Creep tests were performed on sheet and rod épecimens in both air
and molten salts. Most of the testing was confined to the 1100 to 1300°F
range, but a few tests were conducted at temperatures up to 1700°F.
Summary curves representing stress vs minimum creep rate for the MSRE
heats of INOR-8 are shown in Fig. 13. These data for heat 5055 are
plotted to show the time to various strains at 1300°F in Fig. 14. The
rupture life of INOR-8, plotted as a function of stress, is shown in
Fig. 15. :

Fatigue Properties

Rotating-beam fatigue tests were conducted on INOR-8 at 1100 and
1300°F by Battelle Memorial Institute.” Grain size and freguency were
the major variables studied. The results are shown in Figs. 16 and 17.
349

Table 11. Maximum Allowable Stresses for INOR-8 Reported by
ASME Boiler and Pressure Vessel Code

Maximum Allowable Stress

Temperature (P s1)
(°F) Material Other Bolti
Than Bolting +1g
100 25,000 10,000
200 - 24,000 9,300
300 23,000 8,600
400 21,000 8,000
500 20,000 7,700
600 20,000 7,500
700 19,000 7,200
800 18,000 . 7,000
900 18,000 6,800
1000 17,000 6,600
1100 13,000 6,000
1200 6,000 3,500
1300 3,500 1,600
140 ORNL-DWG 844444
120 '//g,//
s
Q 2?2 27 D/y
3 0
T a0 |- Yinere STECRED ‘ \“\///// _
= GTH ~ N
g N
n I I I ~ //
w N
= HEATASP%?;RSALLEL TO R.D. \\//%
% 40 — ®NORMAL TO R.D. Z /;}7/
- HEAT 5081 N
a PARALLEL TO R.D. N7/
VNORMITQLL I_To R.D. \////
20 — \\‘
o)

0 200 400 600 800 1000 1200 1400 1600 180OC
TEMPERATURE {°F)

Fig. 9. Ultimate Tensile Strength for INOR-8 to 1800°F.
350

- UNCLASSIFIED
ORNL-DWG 64—44145

70
60
a2 ) .
o Z -
O 50
g ?/
T 7 / SCATTER BAND FOR
= ) EXPERIMENTAL HEATS
o 40 ’( —_
5 \\ ////7 ’W/ ‘\‘ ./
o £ / Ly
o 20 ’ ~ | "---..._// %%%E V,B// i
L 7 ¥ 7
9 ’ ’ { ~-‘._/<///
w | R
> t 2 N
2 20 ™ HEAT 5075 \ | aza
~ 4 PARALLEL TO R.D. MINIMUM: SPECIFIED
o] * NORMAL TO R.D. : YIELD STRENGTH
10 |— HEAT 5084
o PARALLEL TO R.D.
v NORMAL TO R.D.
o \ | |
0 200 400 600 800  100C 1200 1400 1600 1800
TEMPERATURE (°F)
Fig. 10, Yield Strength for INOR-8 to 1800°F.
UNCLASSIFIED
CRNL-DWG 64-~4446
. - | |
| _——SCATTER BAND FOR
// EXPERIMENTAL HEATS
60 Z
/6;;//7
g i
50 77 : /:
2 p
o)
= 40 7
&
S| ——rcF———— 7
C
o 30 \MiNIMUM SPECIFIED —— —
ELCNGATION
| N
N
20 — HEAT 5075 —
a PARALLEL TO R.D.
¢ NORMAL TO R.D.
10 —— HEAT 5084 -
o PARALLEL TO R.D.
v NORMAL TO R.D.
o I

0 200 400 600 800 {000 1200 4400 1600 1800
TEMPERATURE (°F)

Fig. 11. Percent Elongation for INOR-8 to 1800°F,
351

UNCLASSIFIED
ORNL-DWG 64-4417
" R

60 /////
=

.
WL o xo ///%/%%//

HEAT 5081 447i§2;§?

o PARALLEL TO R.D.
v NCRMAL TO R.D.

-—-SCATTER BAND FOR
EXPERIMENTAL HEATS |

Fa)

.
|

<P-@

REDUCTION IN AREA (%)

0 200 400 600 800 1000 1200 1400 1600 1800
TEMPERATURE (°F)

Fig. 12. Reduction in Area for INOR-8 to 1800°F,

UNCLASSIFIED

(x 103) QORNL-DWG 64-4434
80 =1
_——I’
hfl—-—"”ls
60 A U_n ..——'/r //'
100 L;,;o, =" o o
40 a7 Pres
—-—‘A”;J d‘/
L—‘/
| b
Qlr
A L~
= 3/}0/ /0'/
- o I
£ 20 yf’:j. /.V
o "5 C |
17 d s
w 1300°F _+"a| P o
[ ,A
o 7 I‘//
10 e INOR-8 !
8 it AS RECEIVED-TESTED IN AIR L
.
o] MATERIAL FROM MSRE PLATE {1
6 b = o HEAT 5055 LY
Jllo A HEAT 5075
1500°F {7 =1
L 0 HEAT 5084
4 OPEN SYMBOLS -PARALLEL TO R.D.
CLOSED SYMBOLS -~TRANSVERSE TO R.D.
3 | | Lt ] [
- 1078 1075 1074 1073 1072 1071

in.
in.—hr )

MINIMUM CREEP RATE (

Fig, 13, Minimum Creep Rate for INOR-8 at 1100, 1300, and 1500°F,
352

UNCLASSIFIED
ORNL-DWG 64-4438

100 Il -
80 1300°F T
o HEAT 5055 7]
AIR T
.’;_; o () o~ \‘|(3~O)
- \\.
= a0 = q TR (e |
3 S LT ~STL =L o Jues
g Il v i B o‘\..{,_h"-\\ "~\.‘£38) (29)
0 N .'“"}\3:":: i \Om\I {50
0 ~ e "y » ~ o
W 20 T 8 \}Q_'\q_\o \L ‘*(27)
— ~< T B (25) 11
5 v | [ TN T~ T
0.2 % 0.5 1.0 2.0 50 RUPTURE
} i
0.1 1.0 10 100 1000 10,000
TIME (hr) :

Fig. 14, Stress vs Time to Produce Various Strains for INOR-8.

UNCLASSIFIED
ORNL-DWG 64-4418

100

I I i
| N {
o | I D — § A I i
1100°F Ha— ol L
50 + R e
t —— S —— —
1300°F L~ ] | | ~el el Tro—— L
i e SR e L B U Bl = b
< ] L
oty :TT‘\Q.‘-—- = i [T 4 “-UB“-EM
a 20 1500°F ==aliri— =4 R
8 (I =t ey s
Q \--"‘z-:o\\ 4 =
= 10 o HEAT 5055 i i =0
o & HEAT 5075 D R i
IEJEJ i a HEAT 5081 B ey e B K oy
" | OPEN SYMBOLS - PARALLEL TO R.D. 1T
CLOSED SYMBOLS-TRANSVERSE TO R.D.
2
i
! : 10 100 1000 10,000

TIME TO RUPTURE (hr)

Fig. 15. Creep Rupture Curves for INOR-8 at 1100, 1300, and 1500°F.
353

UNCLASSIFIED
ORNL-LR-DWG 34253

90 T T T T T
‘ ] ‘ H’ 100cpm 3000cpm
S.P.19 COARSE GRAINED o© a
80 S.P. 19 FINE GRAINED . .
1!\ l
—~ 70 : >
@ i el
= >l ¢ T~ C
L ™~ el |ie
S 60 3 =2 SR .
g E‘--.l:\ al |1 TTH
-} m o
CEL . "'--.."--.'--.._____--
a4 50 ‘ e
W -___'_—'D-—--—_._
wy
w
1
40
30
20 E
103 104 10° 108 107 10

CYCLES TQ FAILURE

Fig. 16. Fatigue Properties of INOR-8 at 1100°F; Rotating Beams.

UNCLASSIFIED
ORNL~LR-DWG 54252

" I 1T
100 cpm 3000¢cpm
80 S.P.19 COARSE GRAINED o a
S.P 19 FINE GRAINED .
—~ 70
Z
L ~
I:—) 60 ‘.\‘- ‘\""-j:
E \\0:_ .\--..\___- [ ]
b= ¢ "“WE‘.1 o il
< 50 :Fm:r-..{ —
[a] [m] """-..-
:m'j \3"'---.._& e T—
o --\--..-_-..
b 40 PR f»-.._—_L__..
30
20
103 104 10° 108 107 Texd

CYCLES TO FAILURE

Fig. 17, Fatigue Properties of INOR-8 at 1300°F; Rotating Beams,
354

The thermal-fatigue behavior of INOR-8 was investigated at the
University of Alabama. Figure 18 presents a graph of the plastic strain
range vs cycles to failure, and results are seen to obey a Coffin-type
relation. The tests involved several maximum temperatures, as noted, as
well as rapid cycling and hold-time cycling. Furthermore, the data were
obtained from two specimen geometries. These data show good agreement
with isothermal strain-fatigue data on this alloy.

An analysis of these same tests, based on a plastic strain energy
criterion, indicates that the total plastic work for failure of INOR-8
by fatigue is constant in this temperature range. The data have been
plotted in Fig. 19 as plastic strain energy vs cycles to failure. It is
seen that the data for 1300 and 1600°F fit curves having the same slope
(~ —1) but different intercepts. The intercept values (at N. = 1/2) are
in fair agreement with plastic strain energy values derived Trom tensile
tests at the appropriate temperatures.

Mechanical Properties of INOR-8 Irradiated at Elevated Temperatfires

Subsize tensile specimens of INOR-8 have been irradiated in the B-8
lattice position of the ORR. The specimens (heat 5081) were given a
solid-solution heat treatment at 1175°F prior to irradiation. They were
irradiated at 1300°F for approximately 2000 hr to a dose of 5 to 12 X 1020
nvt (fast and thermal).

The tensile properties of irradiated and unirradiated alloys with
similar thermal histories are compared in Table 12 as functions of defor-
mation temperatures for given strain rates. No significant differences
in yield stresses were observed, although the yield stress behaviors were
slightly different at 600 to 750°F. However, irradiation markedly re-
duced the ultimate and true tensile stresses for deformation temperatures
of 1112°F and above. The ductility was also affected by irradiation, and
again the effect was confined to deformation at elevated temperatures.
The true fracture strain was approximately equal to the true uniform
elongation for the irradiated specimens. Therefore, the fracture strain,
uniform strain, and total elongation were all reduced by irradiation.

The influence of strain rate at elevated temperatures was investi-
gated, and these data are given in Table 13. The effect of irradiation
increases with decreasing strain rate. Figure 20 indicates that the
effect .of irradiation on the uniform elongation and true fracture strain
may reach a maximum at a temperature that is strain-rate sensitive.

The effect of postirradiation annealing on the elevated-temperature
properties has been investigated. Numerous investigations have shown
that postirradiation heat treatments at 1750 to 2000°F anneal the damage
that causes irradiated stainless steel to be brittle when deformed at
ambient temperature to 212°F. Irradiated INOR-8 samples were heated for
0.75 hr at 2150°F prior to deforming at 1300 and 1650°F. The properties
of the irradiated alloy given the postirradiation heat treatment are com-
pared to those of the irradiated alloy not heat treated in Table 14. The
355

UNCLASSIFIED
ORNL-DWG 64-4006

-1 .

10 . A —
- . “ —_ae,n,288=0.53
£ A N\F

Ty
& AN
= N
H 2 a
< ™
= N
Y 1072 PNA
z =
w ™
(L]
z 5 4
< R
= g
< N
p 2 B
P 0 1600°F * N
o . _3 4 1500°F N
e 10 ® 1300°F N
@ ~
3 ~Me
o 5 Y
‘dfq \\
< [ ]
Nk
2 N
h.
T ~
10° 2 5 To L 5 102 2 5 102 2 5 100 2 5 109

My, CYCLES TO FAILURE

Fig. 18. Effects of Strain Range on the Thermal-Fatigue Life of

UNCLASSIFIED
ORNL-DWG 64-4007

104
N
;)‘. 103 E\\
=
3
) ~N
g 0 EN
St \\
W ]
o 4 9
> — N
© 1600°F \\ \:"_1300"':
(]

& 102 NS
> ~o
3 %
y s AN
v \\\\
g 2 \ \\
|— N ~N
w \ :l\\ .
Lot NN o
= : SIS
w . AN
I ™
T 5 EAN
& ]
Q

2

10° L_

100 2 5 10! 2 5 102 2 5 103 2 5 104 2 5 105

N, ,"CYCLES TO FAILURE

Fig. 19. Relation of Plastic Strain Energy Absorbed per Cycle to
the Fatigue Life of INOR-8. ‘ |
Table 12. Tensile Strength and Ductility of Irradiated and Unirradiated INOR-8 (Heat 5081)

Stress (ksi) Ductility (%)

Peformation

Tempfrature Yield Strength True Tensile Strength  True Uniform Strain  True Fracture Strain
e) Irrad. Unirrad. Irrad. Unirrad. Irrad. Unirrad. Irrad. Unirrad.
RT 46 .3 45.5 168.6 166.5 42.3 40.6 42.5 39.0
100 43.9 43.9 159.5 161.0 40.1 40.3 44 .6 37.2
200 38.4 40.7 150.6 157.5 40.3 41.9 4245 50.7
300 36.0 40.7 154.3 147.0 42.2 37.9 4é .6 4l.4
400 35.0 40.7 146.9 153.0 40.2 39.3 42.5 46 .9
500 35.8 35.8 129.5 144.0 35.3 42 Wb
600 32.5 36.2 82 .4 109.0 11.8 26.7 21.9 31.6
700 31.0 34.1 53.4 102.8 .0 30.8 11.6 42.1
800 28.5 30.9 38.4 59.9 .7 12.2 6.9 86.6

96¢
Table 13. Strain-Rate Sensitivity of Irradiated and Unirradiated INOR-8 (Heat 5081)

Stress (ksi) Ductility (%)
remperatice  Tate | tield Stremgen TS Temeile  Dme laifom e Fracturs
(°c) (per min)
Irrad. Unirrad. Irrad. Unirrad. Irrad. Unirrad. Irrad. Unirrad.
500 0.2 32.7 34.9 136.1 145.6 41.2 42.3 46 .6 53.4
500 0.02 35.8 35.8 129.5 144.0 35.3 42 4 51.4
500 0.002 34.4 37.4 122.5 131.5 32.3 33.3 36.5 34.2
600 0.2 32.9 34.1 112.7 134.4 31.2 38.9 36.5 48.7
600 0.02 32.5 36.2 82.4 109.0 17.7 26.7 21.9 31.6
600 0.002 34,2 34.6 63.1 106.0 10.3 29.9 13.2 29.7
700 0.2 30.5 30.9 66.8 ~ 106.5 13.5 32.2 19.2 39.2
700 0.02 31.0 34.1 53.4  102.8 8.0 . 30.8 11.6 42.1
700 0.002 32.1 33.7 47.0 80.5 5.6 20.0 7.8 29.0
800 0.2 29.3 $29.3 45.7 - 79.8 6.7 11.6
800 0.02 28.5 30.9 38.4 59.9 3.7 12.2 6.9 86.6
800 0.002 29.3 32.5 32.2 42.9 1.8 6.5 4.9 93.7

LGE

358

UNCLASSIFIED
ORNL—-DWG €4-4008

—— UNIFORM STRAIN
A . —— FRACTURE STRAIN
10 | €=0.2 (min) ™! | HEAT 5084 °C
' ANNEALED FOR 1 hr AT 1175 °C
FOLLOWED BY 2000 hr AT 700 °C
DOSEas 7 x102° sy (FAST AND THERMAL)
é 0.8
|8 —
213 os 0.8 i3
N | =
1E X 2|8
| A 06 |5
| odl LLos | =
A \\\
€ Z| 2
Wiw o4 \— 0.4 ©|d
x|z \ 2=
= : g
. GX Olw
= N 02 513
0.2 0 SRR ]
500 700 900 | &
DEFORMATION TEMPERATURE (°C) @ | @
IRRADIATION | N TN
TEMPERATUREI SN T~
| B
0
400 500 600 700 800 900

DEFORMATION TEMPERATURE (°C)

Fig. 20, Effect of Irradiation on the Ductility of INOR-8 at Ele-
vated Temperatures.

Table l4. Influence of Postirradiation Heat Treatment
of 1 hr at 2150°F on the Tensile Properties of INOR-8
at a Strain Rate of 0.002 min™*

Tensile - Elongation
Deformation Postirradiation Yield Strength Measured
Temperature Stress (ksi) in 1 in. (%)
° Heat Treatment . ,
(°F) (ksi)
Ultimate True Uniform  Total
1.300 As irradiated 32.1 443 47.0 5.8 6.1
1650 As irradiated 23.2 23.2 23.3 0.4 0.4

1650 1 hr at 2150°F 23.5 23.5 23.6 0.6 0.7

359

stability of the irradiation-induced defect associated with the elevated
temperature problem is remarkable.

A comparison of the effects of irradiation at the lower and upper
limit of neutron exposure given the INOR-8 are shown below:

Neutron Dose True Elongation

Yield Ultimate .
(nvt) Stress Strength gi?:;leh (%) .
Fast Thermal (ksi) (ksi) (ksi%t Uniform  Total

5 x 1029 6 x 10720 33.3 44, .0 46.0 4.5 4.7

12 x 1022 1.2 x 102* 3.1 4ty .3 47.0 5.8 6.1

The differences observed in these data obtained at 1300°F and at a stfain
rate of 0.002 min~t are believed to be insignificant.

In summary, the ductility of INOR-8 in the absence of irradiation is
considered to be low at deformation temperatures of 1472 and 1650°F. The
low, uniform elongations decrease with decreasing strain rate. The de-
crease in uniform strain is accompanied by a large increase in fracture
strain and a reduction in true fracture stress. The alloy appears to
work-soften at elevated temperature at a specific temperature that is
strain rate sensitive.

The effect of irradiation at high temperatures on the mechanical
properties of INOR-8 is similar to the effect observed on austenitic
stainless steels and other nickel-base alloys. This effect is caused by
thermal neutrons that apparently produce helium when boron atoms in the
grain boundaries are irradiated. The deterioration in the ductile
properties of the irradiated metal occurs only at temperatures above
1112°F. The magnitude of the effect increases with decreasing strain
rate. The effect is primarily one of lowering the ductility. However,
the ultimate stress and the true tensile stress are markedly reduced for
those deformation temperatures at which the alloy's work-hardening
coefficient is high.

Studies are now in progress to determine the effects of irradiation

on the creep strength of INOR-8 which is the primary criterion for MSRE
designs. > ‘ : .

Fabrication of INOR-&

Melting and Casting

The melting and casting of INOR-8 can be carried out by using con-
ventional practices for nickel and its alloys. Air and vacuum induction,
360

electric arc, and consumable arc melting methods have been successfully
employed. The usual precautions must be taken to avoid gassy melts
together with sulfur contamination. Manganese, silicon, aluminum, and
titanium are added to deoxidize the metal. Care must be exercised in
the use of boron as a deoxidizer because of its adverse effects on
weldability.

The alloy has been cast in molds of water-cooled copper, graphite,
rammed magnesia, and cast iron. Both the fuel and coolant pump impellers
and volutes were made from INOR-8 sand castings. More than 15 sets of
these castings were made by the Cerami-Sand Process of General Alloys
Company. Chemistry requirements for the castings did not deviate from
the chemistry specification for wrought material. Radiographic quality
levels of the castings were class 1 (ASME Boiler and Pressure Vessel
Code) in areas of high stress and class 2 in other areas; however, con-
siderable weld repair was required to obtain this quality.

Mechanical properties tests were performed on specimens made from
sand-cast blanks as well as on investment cast material. Results indi-
cated that the tensile strength of sound cast metal is about 60% of that
for wrought metal and the 0.2% offset yield strength for cast material
is about 80% of that for wrought material. Most castings had elongations
of greater than 20% at all temperatures. Short-term stress rupture tests
for cast material at 1100 to 1700°F indicated that it was stronger than
wrought material, and the conventional MSRE design strengths are therefore
considered suitable for cast metal.

Forming

INOR-8 has been produced on a commercial basis in a variety of shapes,
such as plate, sheet, rod wire, and welded and seamless tubing, by con-
ventional methods of hot and cold forming. More than 27 heats of INOR-8
were made for the MSRE, amounting to almost 200,000 1b of material.

- Hot forging of INOR-8 ingots has been successfully carried out both
on press- and hammer-type forges between 1832 and 2250°F. Hot rolling
is readily accomplished between 2100 and 2200°F using reductions of about
10% per pass. The optimum extrusion temperature range is 2150 to 2200°F
for extrusion ratios of 4:1 to 12:1l. Extrusion rate is one of the most
critical variables in tube blank extrusion; it should be kept to about
1 in. of billet length per second in order to avoid hot cracking.

The ductility of INOR-8 at room temperature allows cold-working
operations such as rolling, swaging, tube reducing, and drawing to be
performed sucessfully. The alloy work-hardens rapidly, as shown in
Fig. 21, and therefore requires frequent annealing in order to prevent
the formation of splits and cracks. Between anneals a reduction in area
of about 50% is possible.

The forming of INOR-8 heads for the MSRE was attempted using hot
spinning techniques as well as hot- and cold-forming methods. Spinning
STRENGTH (psi)

361

3 UNCLASSIFIED
{(x107) ORNL-LR-DWG 24088
250 ‘

200 — P /g/fi

ULTIMATE TENSILE STRENGTH

/ 0.2 % OFFSET YIELD. STRENGTH

150 / / ‘ 60

100

40

_ELONGATION

50 : } 20
/ | — =

0 10 20 30 40 50 60 70 80 90
REDUCGTION IN THICKNESS {%)

Fig. 21. Work-Hardening Curves for INOR-8.

PER CENT ELONGATION (2-in. qage length)
362

was attempted at 2100 to 2150°F with the INOR-8 plate sandwiched between
steel plates. Cracking occurred in the flanges and radii of all the
shells. This was due to the material's cooling too rapidly into the
region of low ductility. As a result spinning was discontinued as a
method for producing INOR-8 heads.

Heads 3 to 4 ft in diameter were successfully made by the hot-
forming method using starting temperatures of 2000 to 2100°F. This ma-
terial cooled to temperatures as low as 1550°F during working but did not
crack. Cold forming was the most successful method used to fabricate
INOR-8 heads. Heads as small as 9 in. in diameter by 1/2 in. thick and
as large as 60 in. in diameter by 1-1/8 in. thick (reactor vessel) were
made in this manner without the use of intermediate anneals.

Machining

INOR-8 was found to machine in a manner similar to other nickel-base
alloys, such as the Hastelloy type. Standard cemented carbide tools,
cutting speeds of about 35 to 50 fpm and feeds of 0.015 in. or less, gave
best machining results. Depths of finishing cuts were kept greater than
0.005 in. in order to machine below the work-hardened surface that occurs
in this alloy. Rough machining using a cutting depth of 0.250 to 0.375
in. was accomplished at speeds up to 150 fpm.

Heat Treatment

INOR-8 can be fully annealed by treating it at 2150 * 25°F for ap-
proximately 1 hr per inch of thickness, or a minimum time of 10 to 15 min.
For bright annealing an atmosphere of dry hydrogen is satisfactory.

The approximate recrystallization temperatures for INOR-8 have been
determined to be 1350 and 1500°F for 80 and 10% cold-worked materials
respectively; this was with annealing times of 1 hr.

The chemical specification for INOR-8 is such that the alloy is not
able to be age hardened by the precipitation of NiXMoy intermetallic com-

pounds at 1100 to 1500°F. Carbides initially present in the alloy will
precipitate eutectically and will spheroidize as a result of the fabrica-
tion and annealing treatment. High-temperature amnealing of the wrought
material, for example, between 2100 and 2250°F, results in partial solu-
tion of the carbides which then precipitate as fine particles when the
temperature of material is lowered to, for example, 1500°F. A slight
aging response has been attributed to the reprecipitation of the carbides.

Heat-treatment requirements for INOR-8 parts in the MSRE call for
the material t0 be annealed after any forming or working and to be stress-
relieved after final fabrication. The temperature used for annealing is
2150 + 25°F, and the temperature used for stress-relieving is 1600 + 25°F.
363

The cooling rate at temperatures above 600°F for any complex com-
ponent is 400°/hr per inch of thickness. Where possible, stress-relieving
was done after welding and heavy machining in order to give dimensional
stability at operating temperatures and to improve ductility in the 1400
to 1800°F range.

Cleaning and Descaling of INOR-8

Special precautions were taken to keep INOR-8 components clean and
free of harmful materials. After machining, forming, etc., and prior to
any welding or heat treatment, the alloy was degreased using a solvent
such as perchloroethylene and then inspected for cleanliness. At no time
were sulfur-containing compounds allowed in contact with the alloy. Also,
aluminum, lead, and mercury were not allowed to contact the alloy in
order to~preclude their presence at elevated temperatures where undesira-
ble reactions occur between these materials and INOR-8.

Oxide removal from INOR-8 was generally done by the vapor blast
method because of the safety and convenience of this technique; however,
other methods were also successfully used, such as processing with sodium
hydride and also pickling in a 30% HNO3;—5% HF bath at 140°F.

Welding

INOR-8 is readily weldable by the inert-gas-shielded tungsten-arc
process using weldrod with no changes in chemistry from the wrought ma-
terial. Welding procedures and specifications for welding the alloy to
Inconel and to austenitic stainless steels, as well as to itself, are
listed in Table 15. These procedures are similar to the welding pro-
cedures for other nickel-base alloys, such as Inconel.

The mechanical properties of INOR-8 weld metal are very similar to
those of INOR-8 wrought metal. That no aging occurs is shown in Fig. 22,
which compares INOR-8 with age-hardenable Hastelloy B and Hastelloy W.

Early tests of some experimental heats of INOR-8 indicated a sus-
ceptibility to weld-metal cracking of the type shown in Fig. 23 when
these materials were subjected to welding under restraint. Consequent
studies showed that weld-metal cracking tendencies can be induced in
heats with an abnormally high boron content and that the cracking tenden-
cies can be eliminated by the addition of 2% Nb or by improving deoxida-
tion and purification procedures during melting. A change to the chemis-
try specification (see Table 16) and a weldability testing of all heats
of material purchased for the MSRE has resulted in weldments of the
guality shown in Fig. 24. A comparison of the mechanical properties of
the experimental heat that had cracking tendencies (INOR-8 SP-23) with
the properties of an INOR-8 having a revised chemistry is shown in
Table 17.
364

Table 15. Welding Specifications for INOR-8 Material
Specification Date .
No. Issued Title
MET-WR-2 8/23/63 Welding of High-Nickel Alloy Components for
Nuclear Service
MET-WR-3 6/23/61 Welding and Brazing Requirements of Nickel-
Molybdenum-Chromium Alloy Test for Plate
Qrs-23 1/1/58 Qualification Test Specification Inert-Gas
Shielded Tungsten Arc Welding of INOR-8
AMloy Tubing
PS-24 ORNL Procedure Specification D.C. Inert-Gas
Shielded Tungsten Arc Welding of INOR-8
Alloy
PS-25 1/5/60 ORNL Procedure Specification D.C. Inert-Gas
Shielded Tungsten Arc Welding of INOR-8
Alloy Pipe, Plate and Fittings
P3-26 3/15/60 ORNL Procedure Specification D.C. Inert-Gas
Shielded Tungsten Arc Welding of INOR-8&
Aoy Pipe, Plate and Fittings
MET-WS-1 12/20/61  Joint Design for Welding of INOR-8
MET-WS-2 11/21/61 Machining Details for Matching Single-
Welded Joints
MET-WS-3 2/16/62 Interpretation of Weld Symbols of Company
Drawings
MET-WR-200 11/6/63  Procedure for Inspection of Welding of High
Nickel Alloys
Table 16. Chemical Composition Requirements for the
Ni-Mo-Cr Alloy — INOR-8
a
Element Percent
Original Specification Revised Specification
Nickel Remainder Remainder
Molybdenum 15.00-18.00 15.00-18.00
Chromiun 6.00- 8.00 6.00- 8.00
Iron 5.00 5.00
Carbon 0.04— C.08 0.04— 0.08
Manganese 0.80 1.00
Silicon 0.50 1.00
Tungsten 0.50 C.50
Aluminum + titanium ' 0.50 .50
Copper 0.35 C.35
Cobalt 0.20 .20
Phosphorus 0.010 G.015
Sulfur 0.015 G.C20
Boron 0.010 0.010
Vanadium 0.50
Others 0.50

a'Single values are maximum percentages unless otherwise specified.
365

UNCLASSIFIED
ORNL-LR-DWG 34774
100

DUCTILITY

ELONGATION IN 4in. (%)
N
(@]

TENSILE STRENGTH

7 AS ~WELDED

AGED AT 1200°F |
FOR 200 hr

AGED AT 1200°F
FOR 500 hr

180,000

N

160,000

N

140,000

(LT T
A NTRTLRRERT SRR

120,000

100,000

80,000

TENSILE STRENGTH (psi}

60,000

40,000

20,000

HASTELLOY HASTELLOY INOR-8
ALLOY B ALLOY W

Fig. 22. Room-Temperature Mechanical Properties of Hastelloy B,
Hastelloy W, and INOR-8 Weld Metal in the As-Welded Condition and After
Aging at 1200°F,
& UNCLASSIFIED
g Y-42135
LAY AR

Fig. 23. Weld Metal Cracking in Heat SP-23 of INOR-8.
AN Pt (O W e Oy BV 22 O
v

o3 L

4 £, UNCLASSIFIED

’_ Y2136 =

=

To 100X

Fig. 24. Weld Metal Structure of Heat 5055 of INOR-8, Showing No
Cracking.

Table 17. Comparison of the Mechanical Properties of Transverse Specimens from High Restraint
INOR-8 Welds Before and After Material Composition Reviision
Room Temperature Tensile 1300°F Tensile \\ . Stress Rupture®
Material Rupture Total
Designation Elongation Tensile Yield Elongation Tensile Yield J Time ' Stras
| (%) Strength  Strength (%) Strength  Strength reln
| (hr) (%)
Original 28 105,600 67,800 8 59,600 43,100 7.5 1.7
composition ’ -
Revised 40 114,200 70,000 16 67,200 47,200 96.5 5.8
composition |

®Stress of 27,500 psi at 1300°F.

L9t
368

Component Fabrication Development

Heat Exchanger Tube-to-Tube-Sheet Attachment

The INOR-8 tube bundle of the primary heat exchanger of the MSRE,
shown in Fig. 25, contains 163 1/2-in.-diam X 0.42-in.-wall U-tubes
welded to a 1-1/2-in.-thick tube sheet 18 in. in diameter. Procedures
were successfully developed for welding and back brazing the 326 closely
spaced tube-to-tube sheet joints in the unit, and the actual tube bundle
was fabricated without incident. The welded and back-brazed desigh
shown in Fig. 26 provides a double seal between the fuel and coolant
salts and was used because of the necessity for high Jjoint integrity.

Grooves were trepanned in the weld side of the tube sheet so that
low-restraint edge-type welds could be made. This technique greatly re-
duces the cracking problems associated with welding thin-walled tubes
to thick tube sheets. Welding and assembly procedures were developed
that provided welds of high quality with a minimum of roll-over. Welding
conditions are given in Table 18.

The 82 Au—18 Ni (wt %) brazing alloy used in this application is
corrosion resistant, ductile, and produces joints exhibiting satisfactory
mechanidal“strength. A unique Jjoint design, incorporating a trepan
groove and feeder holes in the tube sheet, was used for the braze side.
The purpose was to prevent the brazing alloy from melting and flowing
along the thin-walled tubes before the heavy tube sheet reached the
brazing temperature. ' .

After the determination of optimum welding and brazing conditions on
several subsize and full-size mockup samples, the actual unit was con-
structed. Nondestructive inspection of the welds revealed no defects.

The general flow of the brazing alloy was good, and ultrasonic examination
of brazed joints showed only minor, scattered porosity. The completed
unit passed both the helium leak test and the 800-psi hydrostatic test.

Development and Manufacture of MSRE Control Rod Elements

A total of 160 MSRE control rod elements were manufactured by
Westinghouse Atomic Fuel Department. The elements are gadolinium oxide—
aluminum oxide bushings clad with Inconel, as shown in Fig. 27. They are
about 1-1/2 in. long with an ID slightly larger than 3/4 in. and a wall
thickness of about 3/8 in. Each element contains three bushings which
are made by pressing and sintering prereacted powder, starting as 70%
gadolinium oxide and 20% aluminum oxide. The use of prereacted powder was
developed at ORNL in order to minimize shrinkage and eliminate the possi-
bility of hydrolysis, with subsequent deterioration of the solid pellets.

The prereacted materials have densities as high as 6.0 g/cm3. This
compares with a theoretical maximum of 5.89 g/cm3 for a fully dense mix-
ture. X-ray diffraction analysis identified the reacted material to be
primarily a perovskite-type phase GdAl03 with an excess of G-Al,03.
Fig. 25.

Teble 18.

369

MSRE Primary Heat Exchanger Components.

Welding Conditions for MSRE Tube-to-Tube-Sheet Joints

Electrode material
Electrode diameter
Electrode taper

Electrode-to-work distance

Electrode position

Inert gas
Gas flow rate

Welding amperage

Electrode travel speed

Weld overlap (full amperage)

Weld overlap (amperage
taper)

Welding position

Tungsten + 2% thoria
3/32 in.
30° including angle

0.035 in. (determined by
feeler gage)

0.005 in. outside joint interface
for full 360° rotation (*0.002
in. concentric)

Argon (99.995% purity)
12-13 cfh

39-40

7.3 in./min

30-60°

120-150°

Flat

wNCLAssIFIED
s

370

UNCLASSIFIED
ORNL-LR-DWG 65682R3

/TUBE

TREPAN GROOVE WiTH

BRAZE SIDE BRAZING ALLOY RING

WELD SIDE

TREPAN

{g) BEFORE WELDING AND BRAZING

11

(&} AFTER WELDING AND BRAZING

YIS AT 7777577
A

Fig. 26. - Schematic Drawing of Welded and Back~-Brazed Joint Design.

UNCLASSIFIED
ORNL-DWG 64-1105R

TOP CLOSURE

| | Gd05—Al03

I ' :-' BUSHING

P Sy 80T TOM
W/‘ CLOSURE

INNER TUBE

i,
X

7l y

SN e T e W e sl
r,"ll’”"i’llIllillllllllllll’lll’

%

///

OUTER TUBE

Fig. 27. MSRE Control Rod Element.
371

The manufacturing process for the bushing included the mixing of
the Al,05 and Gd,0; powder and prereacting at 1700°F. The reaction
products were crushed, ball milled, sized, pressed into a bushing, and
sintered at 1450°F. Each bushing was then given a thermal shock test
by heating to 1400°F and quenching in water. After grinding, the bushings
were visually inspected for any chips or cracks.

The top closure was welded to the outer and inner walls by the auto-
matic tungsten-arc, inert-gas shield process (TIG). The bushings were
heated to 1500°F to drive off absorbed gases. This heat treatment was
found necessary because some of the first elements bulged owing to gas-
pressure buildup during heating after canning. The bushings were then
loaded into the can, the bottom closure was pressed on in a helium
atmosphere, and the final two bottom closure welds were made.

After a preliminary dimensional inspeétion, a helium leak check,
and a fluorescent penetrant inspection, the elements were heated to 1700°F
in air in order to preoxidize the Inconel and thus prevent self-welding.

Prototype control rod elements were subjected to repeated thermal
and mechanical shocks in the control-rod rig test from room temperature
to 1400°F. '

These were examined radiographically and dimensionally after 24 (1
cycle), 350, and 600 hr of operation that included approximately 11,000
cycles and 1700 simulated scrams. Holes were made in one can to expose
the Gdp03-A1,03 to air for the final 250 hr of testing.

No measurable dimensional changes in the metal cans were noted
throughout the test. Axial cracks were radiographically observed after
the first cycle, and severe cracking in both the axial and transverse
direction was observed at later stages. However, no crumbling or
ratchetting was evident, and no condition was observed that might affect
the nuclear or mechanical performance of the control-rod elements.

MSRE Material Surveillance Program

AY

Surveillance specimens of INOR-8 and type CGB graphite will be placed
in a central position of the reactor core to survey the effects of reactor
operations on these materials. The specimens will be made from the stock
used to fabricate the reactor primary system and moderator. Identical
control specimens will be exposed to a duplicate thermal history while
submerged in fuel salt in an INOR-8 container to differentiate between
the effects of temperature and the effects of irradiation. Specimens
will be removed from the reactor and examined after six months of oper-
ation and then again after periods of operation that should result in
meaningful data based on the results of the first set of specimens. New
specimens will replace the material removed.
372

The analysis of INOR-8 specimens will include metallographic exami-
nation for structural changes and corrosion effects, mechanical proper-
ties, and a general check for material integrity and dimensional changes.
The analysis of the graphite specimens will include metallographic and
radiographic examination for salt permeation and possible wetting effects,
mechanical properties tests, dimensional checks for shrinkage effects,
chemical analysis for deposits on the graphite, electrical resistivity
measurements, and a general inspection for material integrity.

Summary

The materials problems attendant to containing molten salts as the
fuel-bearing fluid for a power reactor have been studied extensively for
a number of years. The mechanisms by which alloys are attacked by these
salts have been defined, and an alloy inherently resistant to such cor-
rosion has been developed. Thermal- and forced-convection loop tests,
operating for periods of from one to two years, have demonstrated the
validity of the corrosion model and the corrosion resistance of the alloy.

Problems of melting, casting, forming, and joining have been investi-
gated, and it is noted that INOR-8 offers no unusual difficulties in any
of these categories. The physical and mechanical properties of the alloy
have been measured, and sufficient data are available to establish design
criteria. Irradiation effects are being studied and large reactor compo-

- nents have been made from this alloy. It is therefore concluded that
INOR-8 successfully meets the requirements of a construction material for
the MSRE. Ior additional assurance, surveillance specimens will be placed
in the core of the MSRE and examined periodically.

References

1. W. D. Manly et al., Progr. Nucl. Energy, Ser. IV 2, 16479 (1960).

2. dJd. L. Crowley, W. B. McDonald, and D. L. Clark, Trans. Am. Nucl. Soc.
2(1), 174 (June 1959).

3. Met. Div. Ann. Progr. Rept. July 1, 1960, ORNL-2988.

4« R. C. Schulze et al., INOR-8 Graphite-Fused Salt Compatibility Test,
ORNL-3124 (June 1, 1961).

5. R. W. Swindeman, The Mechanical Properties of INOR-8, ORNL-2780
(January 1961).

6. Development Data on Hastelloy Alloy N, Haynes Stellite Company
Brochure, May 1959.

7. R. G. Carlsen, Fatigue Studies of INOR-8, BMI-1354 (January 1959).

8. A. E. Carden, "Thermal Fatigue of a Nickel Base Alloy," to be
published in the Journal of Basic Engineering.

373

MSRE GRAPHITE

W. He Cook

Summazz

The graphite purchased for the MSRE, grade CGB, is a new
nuclear graphite that is basically an extruded petroleum coke
bonded with coal tar pitch heated to 2800°C. Low permeation
is obtained through a series of impregnations and heat treat-
ments. The final heat treatment was at 2800°C or higher. Ex-

~perimental equipment and processes were used on a commercial
scale for the first time., The graphite was produced as 2—1/2-
in.-square X 72-in.-long bars which were machined to the re-
gquired shapes. The average bulk density is 1.86 g/em®. Its
matrix is not permeated by molten salts under conditions more
severe than those expected in the MSRE. It exceeded all the
requirements specified for the MSEKE except that it had longi-
tudinal cracks. Tests indicated that the cracks should not
have any significant adverse effect on the operation of the
MSRE, The shrinkage of the graphite at 350 to 475°C under ex-
posures between 0.60 X 1029 to 1.40 x 1020 nvt (E > 2,9 Mev)
indicated that this should not create any important adverse
effects on the operation of the MSRE.

Major Requirements and Procurement

The major requirements for unclad graphite for the MSRE and for
large central-station power plants are low permeation by salt, low per-
meability to gas, and a fully graphitized structure that would be least
affected by irradiation. When we began looking for graphite for molten-
salt reactors, only small pieces of superior-grade graphite produced in
the laboratory appeared to have the desirable qualities. Therefore, the
specifications for the MSRE were written around the properties of these
laboratory-prepared superior grades of graphite, the requirements of
large-scale power reactors, and reasonable limits of current graphite
technology. The latter required that the gas permeability requirement
not be specified since it certainly would be better than the best com-
mercial grades of extruded and fully graphitized material. The technology
also dictated that a 2-in.-square extruded bar was about the maximum
practical size.

Fixed-fee bids were solicited. Only one vendor, the Carbon Products
Division of Union Carbide Corporation, bid to supply the graphite for the
MSRE.
374

Experimental equipment and processes were used on a commercial scale
for the first time to produce the MSRE graphite designated as grade CGB.
It was produced as 2-1/4-in.-square X 72-in.-long bars which were machined
by the vendor to the shapes required for the MSRE,

The graphite is basically an extruded petroleum coke bonded with
coal tar pitch heated to 2800°C, Low permeation is obtained through a
series of impregnations and heat treatments. All components of the ma-
terial are fired to 2800°C or higher. Since all components graphitize
readily, this means that the product should be essentially a well-graph-
itized material. It is highly anisotropic.

The properties of grade CGB graphite are summarized in Table 1. These
data are combinations of those obtained by the Carbon Products Division
as part of their fulfillment of the purchase contract plus those obtained
by ORNL as part of their studies. The discussion that follows is limited
to some of the major properties of the graphite as related to the require-
ments of the MSKE and large-scale power plants.

Chemical Composition and Oxygen Content

Chemical analyses made by the Carbon Products Division on samples
taken from 5% of the graphite supplied for the MSRE indicated that it
meets the chemical composition requirements. These are compared in Table
2 with the specified values and those of a spot-check analysis made by
ORNL. '

The boron level is moderate and was set to keep the nuclear cross
section of the graphite reasonably low. The sulfur was limited to avoid
the sulfur embrittlement of the INOR-8.

A composite sample from three bars was analyzed spectrochemically.
Calcium and magnesium were the only additional elements present, and
these were less than 10 and 100 ppm respectively.

Grades of graphite tested prior to the grade CGB graphite had rela-
tively high oxygen contamination, which tended to detrimentally contami-
nate the molten fluoride salts.l Because of this, the compositions of
salts were adJjusted to protect against uranium precipitation as an oxide.
However, the quantity of chemisorbed oxygen in grade CGB graphite is rel-
atively low, and it does not appear to present a problem to the MSRE.

The measurement of the chemisorbed oxygen was determined by placing
specimens in a closed gas system, evacuating to <102 torr at room tem-
perature, and measuring the STP volume of carbon monoxide evolved from
the graphite at 1800°C. The oxygen content was determined to be 4 and 9
cm’® of CO per 100 cm’® of graphite, respectively, at ORNL?»? and the Carbon
Products Division. These are well under the maximum of 30 cm? of CO per
100 cm? of graphite permitted by the specification.

375

Table 1. Properties of MSRE Core Graphite, Grade CGB

Physical properfies
Bulk density,® g/cm’

Average 1.86
Porosity,P %
Accessible ' e 0
Inaccessible : 13.7
Total 17.7
Thermal conductivity,8 Btu ft™1 hr~1 (°F)~1
With grain
At 90°F 112
At 1200°F
Normal to grain ’
At 90°F : : 63
At 1200°F 34
Temperature coefficient of expansion,¢ (°F)~1.
With grain ' ) ‘
At 68°F - 0.27 x 107°
Normal to grain ' o
At 68°F - , | 1.6 X 1076 -
Specific heat,d Btu 1v™1 (°F)~1 | |
At O°F 0.14
At 200°F | 0.22
At 600°F 0.33
At 1000°F | 0.3%
At 1200°F 0.42
Matrix coefficient of permeability® to helium - 3'x 1074
at 70°F, cm?/sec
Salt absorption at 150 psig,® vol % - 0,20

Mechanical strengthP at 68°F
Tensile strength, psi

With grain
Range ' 1500-6200
Average : ~ 1900

Normal to grain
Range - 11004500
Average . N 1400

Flexural strength, psi

With grain

Range 30005000

Average ’ 4300
376

Table 1 (continued)

Mechanical strengthP at 68°F
Flexural strength, psi
Normal to grain

Range
Average

Modulus of elasticity, psi
With grain
Normal to grain

Compressive strength,d psi

Chemical purityf
Ash, wt %
Boron, wt %
Vanadium, wt %
Sulfur, wt %

Oxygen, cm’ of CO per 100 cm® of graphite

Irradiation datad
(Exposure: 1.0 x 10?0 nvt, E > 2.9 Mev)
Shrinkage, % (350-475°C)
Ao
Co

22003650
3400

3.2 X 106
1.0 x 108

8600

0.0041

0, 00009
00,0005

0.0013

2.0

0.10
0.07

gMeasurements made by ORNL.

bBased on measurements made by the Carbon Products D1v181on and

ORNL.

“Measurements made by the Carbon Products Division.

dRepresentative data from the Carbon Products Division,

®Based on measurements made by the Carbon Products Division on

pilot-production MSRE graphite.

fData from the Carbon Products Division.,

377

Table 2. Chemical Composition of Grade CGB Graphite

Analysis
Specified Maximum (wt %)

(wt %) -

ORNL2 CPD

Ash 0.07 0.0005 0.0041
Boron 0.0001. 0.00008 - 0.000089

Vanadium 0,01 0.0009 ' 0.0005

Sulfur 0.001 <0.0005 0.0013

aAnalyses made by ORNL on a composite sample from three randomly
selected bars. '

bEach value is an average of values obtained on composite samples
from 4 bars of each of the 13 lots. :

Bulk Density

A minimum bulk density of 1.87 g/cm® was specified for the MSRE
graphite because the laboratory materials that were examined and found
to have low salt permeation had about this density. The bulk density of
grade CGB graphite ranges from 1.82 to 1.89 g/cm® and averages 1.86 g/cm3.
Calculations indicate that an average bulk density of 1.84 g/em® is ac-
ceptable for the MSRE if the graphite meets the salt permeation require-
ments.* The grade CGB graphite meets these conditions.

. Porositx

The graphite has a total void volume or porosity of ~18.0%, about
77 of which is inaccessible or closed. Therefore, the accessible wvoid
space should be ~4.0% of the bulk volume of the graphite. The pore en-
trance diameters to the accessible voids range approximately between the
two extremes shown in Fig. 1 for grades CGB and CGB-X (an experimental
grade). This means that the major portion of the accessible voids hawve
entrances less than 0.4 p in diameter. The molten fluoride salts do not
wet the graphite, and it has been calculated that a differential pressure
of 300 psia would be required to force salt into these voids. This is
supported in the impregnation tests, which are discussed later.

Microstructure

The attainment of the high density and low accessible wvoids with
small pore entrance diameters is reflected in the microstructure of the
graphite,” as shown in Fig. 2. Note that almost all the original voids
378

UNCLASSIFIED
ORNL-DWG 64-2445

100 psia

02 05 10

OPEN-PORE ENTRANCE DIAMETER (microns)

3 T
w | ‘
5
2 2

z
<2 8h
3
= g
2
v
&

e 1

£

g

B

€

5

g )
ol
0.0t 002 005 o4

Fig. 1.

Fig. 2.

and heat treatments. As-polished.

Pore Sizes of Accessible Voids in Graphite.

" UNCLASSIFIED
Y-52432

Some Microstructural Details of MSRE Graphite, Grade CGB.
(a) The major constituent, graphitized coke particles (graphite parti-
cles); and (b and c) types of structures developed through impregnations

500x.
379

UNCLASSIFIED
- Y-53767

UNCLASSIFIED |
Y-52431 ¢

0.02

Fig. 3. Variations of the General Microstructure of the MSRE Graph-
ite, Grade CGB. As-polished. 100x.

in the base stock have been impregnated, and these impregnants fill or
almost fill each void although they have been subjected to 2800°C heat
treatments. There is some structure variation in the graphite which is
illustrated in Fig. 3. In general, all the black areas are voids filled
with graphitized impregnants.

Cracks

The high density and small pore structure are desirable character-
istics for unclad graphite in molten-salt reactors. However, this par-
ticular structure was apparently so dense, or tight, that hydrocarbons
produced through the pyrolysis of the impregnants could not escape rapidly
enough through the natural pores of the graphite during the fabrication
processes. Consequently, the matrix of the graphite was cracked as these
gases forced their way out. The appearance and results of these cracks
on the more severely cracked bars are shown in Fig. 4.

The majority of the cracks were 0.001 in. (25 p) to 0.002 in. (50 p)
thick, 1/8 to 1/4 in. wide, and less than 3 in. long as determined by
visual, microscopic, and radiographic techniques.

380

UNCLASSIFIED
PHOTO 60233

040 IN\DIV.

(5)

Fig. 4. Appearance of Spalls and Cracks in Samples of Severely
Cracked Grade CGB Graphite in (a) Spindle End of a Machined Bar and (b)
Midplane of an As-Received Bar. Cracks are indicated by arrows or are
clearly visible as ragged white lines.
381

Most of the graphite had been manufactured when the cracking, which
occurred in the final processing step, was discovered. Additional process
development, production, and machining of replacement material would have
delayed thé reactor at least 1 year.

With the exception of the cracks, the material met the specification.
Studies were made on defects in representative bars. The graphite would
not be good enough for the core of a high-performance breeder, but our
analyses showed that sizable flaws should not adversely affect the opera-
tion of the MSRE,™ The specifications were revised to define the flaws
that could be accepted. In addition to the integrity tests in the orig-
inal specification and additional modified macroscopic examinations, the
vendor agreed to make a 100% radlographlc examination of all the bars,
Bach bar used in the MSKRE passed limits set for the results of two radio-
graphs made perpendicular to, and along the full length of, two adjoining
lateral surfaces of each bar.®

Thermal Cycling of Salt-Impregnated. Graphite - .-

Mechanical tests indicated that the cracks resisted propagatigh;7'8
However, since it is probable that some of the cracks in graphite. will
be filled with salt, tests were made to determine if repeated mélting
and freezing of the salt would propagate the cracks.

One'Hundred Cycies,

Laboratory thermal cycllng tests were made on specimens of grade CGB
graphlte with salt-impregnated . cracks to determlne whether damage would
occur under conditions simulating the dralnlng, coollng, and reheatlng
of the MSRE core. : ' - :

The specimens were 2-in.-long transverse séctlofis'CUt:ffbm a machined
CGB graphite bar in a zone having a relatlvely hlgh concentratlon of'.
cracks. . ‘

Specimens were impregnated with salt at 700°C (1300°F) during 100-hr
exposures at 50 or 150 psig. The lower pressuré was selected because it
is approximately the maximum pressure expected in the reactor. The higher
pressure was selected to ensure that the cracks would be deeply penetrated
by the salt. The average bulk volume of the specimens permeated by the
salt was 0.08% at 50 psig and 0.14% at 150 psig. The salt impregnation
appeared. to be limited to the fissures, as shown in Fig. 5. The specimens
were subjected to 100 thermal cycles in which the temperature was in-
creased. from 200°C to 700°C in the first 9 min of each cycle; this is more
than 100 times faster than the maximum electrical heating rate of the -
core, After this treatment there were no detectable changes in the ap-
pearance of the graphite or in the lengths of the cracks (Fig. 5).
382

UNCLASSIFIED
PHOTO 60329

Y-48960

(a)

|
|
|
|
|
|
|Y-48961

Fig. 5. GCB Graphite Specimens Impregnated with Molten Salt at 50

psig. (a,b,c) As-radiographed and (d) after 100 thermal cycles.
383

High-Temperature Cycle

For more severe, high-temperature tests of this kind, two additional
specimens were impregnated in the standard way; that is, they were evacu-
ated, and then were exposed for 100 hr to molten salt at 150 psig at 700°C
(1300 F). Radiography, as for the previously reported work, indicated
that the salt was primarily confined to the cracks in the graphite speci-
mens, It was equlvalent to 0.09% of the bulk volume of one specimen and
0.11% of the other.

Each impregnated’specimen was subjected to a single highetemperatfire
cycle from 200 to 1000°C. The rate of temperature rise for each is sum-
mariZed in'TablejB. :

There were no detectable :changes 1in the graphlte spec1mens or the1r
cracks as a result of these thermal cycles. These exceeded the maximum
temperature, 730°C (1350°F), expected in the MSRE: a :

The max1mum'electr1cal heating rate? of the core 1is expected to be
30°C/hr 54°F/hr ' On this basis the slowest of the high- temperature
thermal cycles was >250 times faster in reaching 700°C (1300°F) than the
maximum preheating rate. ' S o

The results of the preceding tests indicate'thetftfle;tfiermal5cjcling
of the MSRE probably would not cause the salt to expand and break up the
graphite.

Table 3. Rates of Temperature Rise for the Thermal Cycles of Two .
Salt- Impregnated Segments from a Machlned MSRE Graphlte Bar

Time to Indicated- Cycle Temperaturea _

Temperature 1 (min)
[°C (°F)] .
. Specimen 198-4 .’ Speeimen 198-5
~200 (~390) 0 T o
700 (1300) | 3.7 3.4

1000 (1830) 36.0 . 23.0

a'I'he differences in heating rates are due to different quantitiee
of argon cover gas used in each test,

bStarting temperature of cycle.
384

Tmpregnation Tests

MErcugz

It was established that mercury impregnation into graphite at room
temperature could be correlated with molten-salt permeation into graph-
ite L0 Therefore, a mercury-impregnation test of graphite specimens at
room temperature was specified to provide a convenient quality control
for the manufacturer. It was specified that 0,125 X 1.500 X 1.500 in,
evacuated specimens should not gain more than 3.5 wt % when submerged for
20 hr in mercury at 470 psig. The vendor was required to apply this test
to a sample;takefi'l in. from the end of every bar. The average weight
gain was l.8%. Most of this was due to mercury pickup in cracks in the
specimens., ‘The maximum permitted by the specification was 3.5%.

Salt

The moltén salts are used at ORNL to screen various grades of graph-
ite. An average of less than 0.2% of the bulk volumes of specimens (0.500
in, in diameter X 1.500 in. long) from grade CGB graphite were filled when
evacuated and then exposed for 100 hr to LiF-BeF,-ThF;-UF,; (67-18.5-14-0.5
mole %) salt-at- 700°C (1300°F) under 150 psig pressure, the standard
screening test. The design limit for the MSRE is 0.5% of the bulk volume
of the graphite. The maximum pressure expected in the MSRE is 50 psig,
which is one-third that of the standard screening pressure. The bulk
volume of the small specimens impregnated with salt ranged from 0.01 to
0.46%. It should be emphasized that a full-scale bar would have even.
less salt plckup. "The small screening spec1mens tend to magnify the.de-
gree of salt 1mpregnatlon.

The MSKE de51gners did not want the molten salt to penetrate into
the matrix of the graphite., This was essentially accomplished. The pat-
tern of the impregnating salt is illustrated in Fig. 6, which shows re-
productions of radiographs of thin sections cut from an unimpregnated
specimen anhd three salt-impregnated specimens. As was the case for the
large thermal cycling specimens, the salt is limited to small penetrations
at the surface-and along cracks where the cracks inteérsect the exterior
surfaces. . Note that the salt is confined to the llmlts of the crack and
does not permeate the matrix.

Mechanical Properties

The tensile and flexural strength measurements on grade CGB graphite
averaged 1860 and 4340 psi, respectively, despite the presence of cracks
in the graphite. There were no values less than the minimums of 1500 and
3000 psi specified for the tensile and flexural strengths respectively.
These strength values were specified primarily to assure a good quality
of graphite. The stresses expected in the reactor are relatively modest
in comparison. The stresses are all less than 100 psi in the absence of
irradiation.t! Under fast-neutron irradiation at MSRE temperatures, the
385

UNCLASSIFIED
PHOTO 66313

[T —

Fig. 6.

Typical Salt Distribution in Grade CGB Graphite Cylinders
as Shown by Radiographs of Their Thin Sections.

(a) Control (no salt
present) and (b,c,d) salt-impregnated specimens; white phase is salt.

386
UNCLASSIFIED %
Visoss
(a) () (¢)
N |

INCH

Fig. 7. Tensile Specimens of Grade CGB Graphite. (a) As-machined.
(b) Typical stair-step fracture. (c) Normal fracture.

UNCLASSIFIED
ORNL-DWG 63-3354R

6000
5000
4000 -
8 .
& 3000 |
&
|
2000 !
© NORMAL FRACTURE
© STAIR-STEP FRACTURE
1000 WITH A RISER 5 Y4 in.
i
° | SN 1 | =
o 0.05 040 045 020 0.25 0.30
STRAIN (%)
Fig. 8. Tensile Properties of Grade CGB Graphite Parallel with Grain y

(Axis of Extrusion) Bar 45, 2 X 2 in.

387

flux gradient across the bars will create a difference in shrinkage and
will cause most of them to bow away from the center of the core. The
stress in tension on a completelg restrained beam will increase at a rate
of 950 psi per full-power year.l?® .

To say that the mechanical properties of the graphite meet the ex-
pected requirements of the MSRE does not adequately describe the graphite
for use in the MSKE or future molten-salt reactors. The tensile strength
of the material was examined in zones with and without cracks with a
specimen as shown in Fig. 7a. A typical fracture in the presence of a
crack is shown in Fig. /b, and a normal fracture in the absence of cracks
is shown in Fig. 7c. 'The ultimate tensile strength values obtained by the
Mechanical Properties Laboratory of the Metals and Ceramics Division are
shown in Fig. 8 with the overall range of tensile strength values just
discussed for the batch of MSRE graphite superimposed with double cross-
hatching. This corresponds to the specimens that demonstrated definite
effects of cracks, the black points, that had a minimum strength of 1510
psi and an average strength of 2940 psi. Specimens that were not appre-
ciably affected by cracks, the open circles, yielded an average strength
of 5440 psi and a modulus of elasticity of 3.2 X 10° psi. This is an un-
usuvally strong graphite.

Stability Under Irradiation

The effects of irradiation on grade CGB graphite have been studied
by Kennedyl? with grade EGCR-AGOT (grade AGOT), a needle coke graghite,
at 350 to 475°C under exposures between 0.60 x 1029 to 1.40 x 102° nvt
(E > 2.9 Mev). He reports that contraction of the grade CGB graphite is
initiated at lower dosage in the direction perpendicular to the grain
than in the parallel direction, but the increase rates for both are al-
most identical. These contractions are shown graphically in Figs. 9 and
10, In the same studies the effect on the average modulus of elasticity
was reported as changing from 3.25 X 10® psi to 8.42 X 10% psi in the par-
allel direction and from 1.06 X 10° psi to 3.78 x 10% psi in the perpen-
dicular direction. Calculations using greater contraction values than

UNCLASSIFIED
ORNL-DWG 64-9677

(x107€) | l |
* GRADE €GB
T 1600/— AGRADE AGOT (LOT 4074 )
~
= /
> 1200 | v
Zz 1 T
2 PARALLEL TO :“/ t
Q EXTRUSION AXIS ‘/. )
x 800 - ey
=z |- /
8 : / .A"l :‘
400 <
/fl
ol A
02 04 06 08 1.0  12(x10%0)

EXPOSURE (avt £ > 2.9 Mev)

Fig. 9. Effect of Irradiation on Graphite Dimensional Changes Par-
allel to the Extrusion Axis,.
388

UNCLASSIFIED
ORNL-DWG 64- 7059R

(x1076) ] |
* GRADE CGB : I
— ool 4GRADE AGOT ¢« o
B (LOT 1074 ) P
{ L
= NORMAL TO . /
—~ 600 EXTRUSION AXIS —[A* =
3 -1
5 . ’ h‘//
2 400 + T
= * A/ ’
4 / A
3 ] pd
200
A
0 .
02z 04 06 08 1.0 12 (x1029)

EXPOSURE (avt £ >2.9 Mev)

Fig. 10. Effect of Irradiation on Graphite Dimensional Changes
Normal to the Extrusion Axis.

these indicate that these should not adwversely affect the operation of-
the MSRE.1% 15 The actual contraction rates should be even smaller at
the operating temperature of the MSRE, because at these higher tempera-
tures nuclear grades of graphite are approaching a minimum contraction
rate.r® This, plus the creep of the grade CGB graphite,'? provide addi-
tional safety factors for conditions that appear satisfactory without
themn. ' ' '

4
Lattice and Ring Bars

The preceding information has been for the MSRE core bars that are
vertical in the reactor. These core bars are supported on a lattice of
horizontal bars and held together at the top by a graphite retainer ring
(sections of graphite pieces joined with INOR-8 pins). The graphite fab-
ricated for the lattice bars and ring sections was fabricated with a
higher permeability to help eliminate cracks and to secure more structural
integrity for these pieces. Radiographic examination of this material
was the only technique that showed any flaws.t® There were a few iso-
lated, small cracks. These closely approach being structurally flaw-free
bars., Limited sampling indicates that the properties of this material
are similar to the core material except that it takes up an average of
4e67 Wt % in the mercury test, about 2.6 times that of the core bars.
However, this is equivalent to a molten-salt takeup in the graphite of
only 0.6% of the bulk volume of the graphite. This is tolerable for the
MSRE because the lattice bars and retainer ring are a small fraction of
the total graphite and are in low-neutron-flux zones that are not subject
to the more severe operating conditions of the MSRE,

Developmental studies on MSEE-type graphite are being continued by
the Carbon Products Division. They report that in recent work they have
produced core-quality graphite without flaws without resorting to this
minor sacrifice to permeation.
389

Conclusions

At this time all the graphite for the MSRE appears to meet all the

requirements., The grade CGB graphite appears to be a substantial step
forward in the development of a high-density, low-permeability nuclear
graphite, but further improvement is needed to obtain a material for
structural units in molten-salt power reactors.

2.
3.

1'7.
18.

References

MSRP Quart. Progr. Rept. Oct. 31, 1958, ORNL-2626, pp. 61—-64.
MSRP Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 75.
Unpublished work.

R. B. Briggs, W. H. Cook, and A. Taboada, internal memorandum, Feb.
14, 1963,

MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORWL-3626, pp. 67-69.

R. B. Briggs, W. H. Cook, and A. Taboada, internal memorandum, Feb.
14, 1963, ' ‘

Ibid.

C. R. Kennedy, GCRP Semiann. Progr. Rept. Mar. 31, 1963, ORNL-3445,
p. 221,

MSRP Semiann., Progr. Rept. Jan. 31, 1963, ORNL-3419, p. 123.
MSRP Semiann. Progr. Rept. Feb. 28, 1962, ORNL-3282, p. 92.

R. B. Briggs, W. H. Cook, and A. Taboada, internal memorandum, Feb.
14, 1963.

Ibid.

C. R. Kennedy, Metals and Ceramics Div. Ann. Progr. Rept. June 30,
1964, ORNL-3670 (to be published).

R. B. Briggs, W. H. Cook, and A. Taboada, 1nternal memorandum, Feb.
14, 1963.

P. N.Haubenreich et al., MSEE De81gn and Operations Report. Part
IIT, ORNL-TM-730 (Feb. 3, 1964), pp. 183-84.

J. M. Davidson and J. W. Helm, "Effect of Temperature on Radiation-
Induced Contraction of Reactor Graphite," p. 286 in Proceedings of
the Fifth Conference on Carbon, vol. I, Macmillan, New York, 1962.

GCRP Semiann, Progr. Rept. Mar. 31, 1964, ORNL-3619, pp. 151—54.
MSRP Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 70.

L

MSRE DESIGN AND CONSTRUCTION

W. B. McDONALD

DESIGN AND FABRICATION

C. K. McGLOTHLAN* R
MECHANICAL DESIGN
C. A MILLS* R
W. M. BROWN R
J. C, CABLE R
F. J. UNDERWOOD GE&C
SITE ENGINEERING
N. E. DUNWOODY ** GE&C
T. E. NORTHUP* GE&C
W. E. SALLEE* GE&C

g
MOLTEN-SALT REACTOR PROGRAM 4
. :
JULY 31, 1964
R. B. BRIGGS, DIRECTOR D
i
i
k
!
MSRE OPERATIONS COMPONENT DEVELGPMENT INSTRUMENT ATION AND CONTROLS REACTOR CHEMISTRY METALLURGY
R P. N. HAUBENREICH R I. SPIEWAK* R R. L. MOORE I&C W. R. GRIMES* RC A. TABOADA M&C
D. SCOTT R J.R. TALLACKSON 1&C H. F. McDUFFIE* RC W. H. CODK MaC
S. J. BALL* 13C F. F. BLANKENSHIP RC T.G. GODFREY* MAC
G. H. BURGER . 1&C E. G. BOHLMANN® RC C.R. KENNEDY* MBC
D. G, DAVIS 13C C. F. BAES* RC E. J. MANTHOS* MAC
s. J.DITTO 1&C M. J. KELLY* . RC W. R. MARTIN* MAC
E. N. FRAY 1&C €| H. C.sAVAGE* RC D, L. McELROY* MAC
P. G. HERNDON I&C [“J, H. SHAFFER RC J. P. MOORE* M&C
SCHEDULING COORDINATION NUCLEAR AND MECHANICAL ANALYSIS FLOW MODELS ). W. KREWSON iac RYE. THOMA™ RC G M. TOLSON® MEC
J.L. REDFORD** I&C G. DIRIAN*** J. T. VENARD* M&C
W. W. GOOLSBY* PLE C. H. GABBARD** R J. R. ENGEL R R. J. KEDL* R C. E. STEVENSON 18C G. LONG**** J.L. GRIFFITH M&C
’ C. H, GABB ARD"** R B. J. YOUNG R T. M. CATE 1&C G. BRUNTON* RC E. J. LAWRENCE MEC
J. J. GEIST*** B. J. JONES* 18C F. A. DOSS RC
A. HOUTZEEL*** R. WEIS 18C H. A. FREIDMAN~ RC
H. B. PIPER R : G. M. HEBERT RC
). A WATTS R S. 5. KIRSLIS RC
K. A. ROMBERGER* RC
INSTRUMENTATION AND CONTROLS OPERATION 3' S' 253:; :E
INSTALLATION ) || REMOTE MAINTENANCE W KR FINNELL RC
R. H. GUYMON R B F uiTcH ne
J- L. REDFORD™ 18C J. H. CROWLEY R R. BLUMBERG R W, JENNINGS RC
R. W. TUCKER 18C P. H. HARLEY R CHEM{STRY J. R. SHUGART R C. E. ROBERTS RC RADIATION TESTING
\TI.E :Slfigon g e o RC R. 1. ROSS RC
: « E. THOMA™ W. P. TEICHERT RC
A. |. KRAKOVIAK R NUCLEAR AMALYSIS J. A. CONLIN® R
H. R. PAYNE R W. R. MIXON R
H. C. ROLLER R B. E. PRINCE R B. H. MONTGOMERY R
R. STEFFY R T. W. KERLIN R )
W. C. ULRICH R
CONSTRUCTION T : COMPONENTS AND LOOPS
. W. H. DUCKWOR TH** "R
B.H. WEBSTER R 1. D. EMCH R MATERIALS D. $COTT R
T. E. DIGGS P&E - D. EMC
ot . . E. H. GUINN R R. B. GALLAHER R
N. E. DUNWOODY GE&C T G HILL " W. H. COOK** MAC M. RICHARDSON R
D. T. DICE P&E C. C‘ HURTT R A, N, SMITH R
H. E. FELTS P&E G R E. CARNES R FUEL PROCESSING ANALY TICAL CHEMISTRY
R. S. JACKSON P&E J. €. JORDAN R S e |
Ww. D. TODD F&E L. E. PENTON R C. A. GIFFORD R . " . .
T W. E. RAMSEY R C. JONES R R. B. LINDAUER cT L. T. CORBIN AC
J. . FREEL P&E R. SMITH. IR R W. H. DUCKWORTH** R J. C. WHITE* AC
L. P. PUGH** R ' IR,
™ . L. UNDERWOOD R R. F. APPLE aC
J. W. TEAGUE R J. L. UN
o 5. R.WEST R CHEMICAL PROCESSING F. K. HEACKER AC
J. E. WOLFE R AL PRQ W, L. MADD OX* AC
: R. B, LINDAUER"" R
PUMP DESIGN AND DEYELOPMENT AC ANALYTICAL CHEMISTRY DIVISION
MAINTENANCE CT CHEMICAL TECHNOLOGY DIVISION
:. g g::ll.lr:«lDELL* R D DIRECTOR'S DIVISION
. G. SMITH R
B. H. WEBSTER** K L v WILSON R GE&C GENERAL ENGINEERING AND CONSTRUC TION DIVISION
L. P. PUGH=* R I8C INSTRUMENTATION AND CONTROLS DIViSIUN
M&C METALS AND CERAMICS DIVISION
P&E PLANT AND EQUIPMENT DIVISION
R REACTOR DIVISION
RC REACTOR CHEMISTRY DI VISION
* PART TIM
DESIGN LIAISON TIME ON MSRP
** DUAL CAPACITY
) C. K. McGLOTHLAN®** R *** EURATOM EXCHANGE SCIENTIST
. wv+ AERE, HARWELL, CHEMISTRY DIVISION

St

17.
18-168.
169.
'170.
171.
172.
173,
174.
175.
176.
177,
178.
179.
180.
181.
182.
183.
184.
185.
186.
187.
188.
189.
190.
1°91.
192.
193.
194,
195.
196,
197.
198.
199,

O 0010 Ul I W o -

?:é.mjé?j?q?..z:S.G?:J.CD.H.U%P'?JMQI_—'EQOH':UUODQQWZ':UIIIQOMPEUDjfijUF:JZICDU)O!—II';UL—'Q

393

Internal Distribution

. Adamson
Alexander
Apple

nwine
Baes

. Ball

. Beall

Bettis

. Billington

. Blankenship

. Blizard

lumberg

Boch

Bohlmann

Borkowski

Boyd

Breeding

Briggs

. Brown

. Bruce

Burger

. Cable

antor

Cardwell

. Carnes

Cate

. Center

. Conlin

. Cook

Corbin

. Cristy

. Crowley

. Culler

. Davis

DeVan

. Dice

Diggs

. Ditto

. Donnelly

Douglas

Duckworth

. Dunwoody

. Emch

. Engel

. bpler

. Ergen

. Felts

. R. Finnell

“HEpmeR

HAEHARODITOHDDrrpQQEHESD QPPN rPEHZEREEHEZT QORI R dgEH QDY S LN HE

i
=
0
4y
"

200.
201.
202.
203,
204,
205.
206.
207.
208.
209.
210,
211,
212.
213.
214 .
215.
216.
217,
218.
219.
220.
221.
222
223,
224 .
225,
226,
227,
228,
229.
230.
231.
232.
233.
234,
235.
236.
_237.
238,
239.
240.
241,
242,
243.
44,
245.
2460,
247,
248.

ORNL-3708 -
UC-80 - Reactor Technology
TID-4500 (35th ed.)

ey btnrafd?:?:?jyiF:F:C)U1z:ziola PrHgrmtEdAERdOoR-bOdOEPE S SOOI QMg

éjh+m z:zjtlfi o O

. Fray

. Freels

. Freidman
. Frye, Jr.
Gabbard
Gall

. Gallaher
. Geist

. Gifford
. Godfrey
Goolsby
. Grimes
Griffith
. Grindell
. Guinn

. Guymon
Harley
Harrill
Haubenreich
Heacker
Hebert

. - Herndon
. Hill

. Hill
Hise
Hitch

. Hoffman
Hollaender
D. Holt

Ei?!?>¢>w QEN=znmpmrmopbdEsrgdomnxdes ==

. P. Holz

S. Householder

. Houtzeel

L. Hudson
C. Hurtt
S. Jackson
Jennings

. Jd. Jdones

Jones
Jordan
. Jordan
Kasten
Kedl
Kelley
Kelly

. Kennedy
. Kerlin
. Kirslis
Krakoviak
Krewson
249, C. E. Lamb - 298. R. J. Ross
250. J. A. Lane - ~ - 292, W. E. Sallee
251. C. E. Larson 300. H. W. Savage
252. E. J. Lawrence 301. A. W. Savolainen
253. T. A. Lincoln 302. D. Scott

254. S. C. Lind 303. H. E. Seagren
255, R. B. Lindauer 304. D. R. Sears
256. R. S. Livingston 305. J. H. Shaffer
257. M, I. Lundin 306. E. D. Shipley
258. H. G. MacPherson ' 307. J. R. Shugart
259. W. L. Maddox 308. M. J. Skinner
260. E. R. Mann : 309, G. M. Slaughter
261. E. J. Manthos : 310. A. N. Smith
262. W. R. Martin 311. P. G. Smith
263. H. C. McCurdy 312. R. Smith, Jr.
264. W. B. McDonald 313. A. H. Snell
265, H. F. McDuffie 314. I. Spiewak

206. D. L. McElroy , 315. R. Steffy

2607. C. K. McGlothlan ‘ 316. C. E. Stevenson
268, E. C. Miller 317. H. H. Stone
269. W. R. Mixon "318. C. D. Susano
270. B. H. Montgomery ' 319. J. A. Swartout
271. J. P. Moore 320. A. Taboada

272. R. L. Moore 321. J. R. Tallackson
273. K. Z. Morgan 322. E. H. Taylor
274. J. C. Moyers 323. J. W. Teague
275. M. L. Nelson 324. W. P. Teichert
276. T. E. Northup 325. R. E. Thoma
277. W. R. Osborn 326. W. D. Todd

278. R. B. Parker 327. G. M. Tolson
279. L. F. Parsly 328. D. B. Trauger
280, P. Patriarca 329. R. W. Tucker
28l. H. R. Payne 330. W. C. Ulrich
282. L. E. Penton ' 331. F. J. Underwood
283. A. M. Perry 332. J. L. Underwood
284, D. Phillips 333. J. A. Watts
285. W. B. Pike 334, J. T. Venard
286. H. B. Piper 335, D. C. Watkin
287. B. E. Prince 336. G. M. Watson
288. L. P. Pugh 337. B. H. Webster
282. W. E. Ramsey 338. A. M. Weinberg
290, J. L. Redford 339. R. Weis

291. M. Richardson f 340. S. R. West

292. C. E. Roberts 341. J. H. Westsik
293. R. C. Robertson 342. J. C. White
294. T. K. Roche 343. L. V. Wilson
295. H. C. Roller 344. C. H. Wodtke
296. K. A. Romberger 345. J. E. Wolfe
297. M. W. Rosenthal 346. B. J. Young
;b

y ;

1 ! ¥
;

, _f [ \
; }"3 (%5 \
i / H
SRR | L
: 347. Biology Library 353-355. Central Research Library
348-349. Reactor Division Library 356-385. Labcratory Records
350-352. ORNL-Y-12 Technical Department
. Library, Document 386. Laboratory Records,

Reference Section ORNL R. C.

External Distribution

387-388. D. F. Cope, AEC, ORO
389. R. W. Garrison, AEC, Washington
390. R. W. McNamee, Manager, Research Administration, UCC, New York
391. R. L. Philippone, AEC, ORO
392, W. L. Smalley, AEC, ORO

393, M. J. Whitman, AEC, Washington
394. Division of Research and Development, AEC, ORO
395-1012. Given distribution as shown in TID-4500 (35th ed.) under
Reactor Technology Category (75 copies - CFSTI)