ORNL-CF-57-4-27.txt

 
 

 
 

ORNL Central Files Number
57=4=27 (Revised)

C=-84 - Reactors-Special Features
of Aircraft Reactors

% /

b

Contract No. W-ThO5-eng-26

 
 

A PRELIMINARY STUDY OF MOLTEN SALT POWER REACTORS

H. G. MacPherson
L. G. Alexander
D. A. Carrison
J. Y. Estabrook
B. W. Kinyon

L. A, Mann

J. T. Roberts

F. E. Romie

F. C. Vonderlage

DATE ISSUED: April 29, 1957

DEC 31957

OAK RIDGE NATIONAL LABORATORY
Operated by
UNION CARBIIE NUCLEAR COMPANY
A Division of Union Carbide and Carbon Corporation
Post Office Box X
Oak Ridge National Laboratoxry

   

 
 

iim

 

Internal Distribution

 

1-20. ILaboratory Records

External Dlstrlbution

»-«m a'n Tyt F **‘flfl; '
21-23, Alr Fbrce nggggsggnflissile Division
2k-25. AFPR, Boeing, Seattle
26. AFPR, Boeing, Wichita
27. AFPR, Curtiss-Wright, Clifton
28. AFPR, Douglas, Long Beach
29-31. AFPR, Douglas, Santa Monics
52. AFPR, Lockheed, Burbank
33-34., AFPR, Lockheed, Marietta
52. AFPR, North American, Canoga Park
36. AFPR, North American Downey
37-38. Air Fbrce Special Weapons Center
39. Air Materiel Command
. Air Research and Development Command (RDGN)
L1. Air Research and Development Commend (RDTAPS)
42-55. Air Research and Development Command (RDZPSP)
56. Air Technical Intelligence Center
o>7=59. ANP Project Office, Convair, Fort Worth
60. Albuquerque Operations Office
61l. Argonne Natioanl Laboratory
62. Armed Forces Special Weapons Project, Sandia
63. Armed Forces Special Weapons Project, Washington
64. Assistant Secretary of the Air Force, R&D
65-70. Atomic Energy Commission, Washington
Tl. Atomics International
T2. Battelle Memorial Institute
T3=T4. Bettis Plant (WAPD)
T5. Bureau of Aeronautics
T76. Bureau of Aeronautics General Representative
. BAR, Aerojet-General, Azusa
78. BAR, Convair, San Diego
9. BAR, Gleann L. Martin, Baltimore
80. BAR, Grummen Aircraft, Bethpage
81. Bureau of Yards and Docks
82. Chicago Operations Qffice
83, Chicago Patent Group
84. Curtiss-Wright Corporation
85. Engineer Research and Development Laboratories
86-89. General Electric Company (ANPD)
90. General Nuclear Engineering Corporation
91. Hartford Area Office
92. Idsho QOperations Office
93. Knolls Atomic Power Laboratory
9k. Lockland Area Office
95. Los Alamos Scientific Laboratory
96. Margquardt Aircraft Company

Pimovaiggy-ny

 
  

 

 

 
 

g8.

99.

100.
101.
102,
103,
10L,
105.
106.
107.
108,
109.
110.
111-11k,
115,
116.
117.
118,
119,
120.
121,
122,
123-124,
125-1k2,
143-167.

 

Netional Advisory Committee for Aeronautics, Cleveland
National Advisory Committee for Aeronautics, Washington
Naval Air Develorment Center

Naval Air Material Center

Naval Air Turbine Test Station

Naval Research Laboratory

New York Operations Office

Nuclear Development Corporation of America

Nuclear Metals, Inc.

Office of Naval Research

Office of the Chief of Naval Operations (OP-361)
Patent Branch, Washington

Patterson-Moos

Pratt & Whitney Aircraft Division

San Francisco Operations Office

Sandia Corporation

School of Aviation Medicine

Sylvania-Corning Nuclear Corporation

Technical Research Group

USAF Headquarters

USAF Project RAND

U. S. Naval Radiological Defense Laboratory
University of California Radiation ILaboratory, Livermore
Wright Air Development Center (WCOSI-3)

Technical Information Service Extension, Osk Ridge
T

 

TABLE OF CONTENTS
Page
SECTION I - Summary, Recommendations and Acknowledgements eeeceeseses 1

SECTION II - Survey and Analysis of the State of Molten Salt
Power Reactor TechNOlOZY esesccecsccsssscsccssscscsscecnsa [

A.. Ma'terial.s UL L B AL B BN B BN BN BN BN B BN NN N BN BN BN BN BN R BE BN BN B BN BN RN BN BN BN BN BN RN NN NN OB B B N NN R NN NN BN RN NN BN 7

Fuel Carrier Evaluation cesesecescoccccsccccoscancnnans [
Blanket Material Evaluation .eeceseccccsccscevesscccaee 13
Intermediate COOLANtS cveceseccccccrsscsscsssssescecess Ll
Container Materials seeceevessssoscascassccssssacosnocese L1
MOQerator MEterials ceececceccecccsosrersssssscscsecoces 2l

\n-l-"ylml—'

B. Materia]—s S 5 560000000500 H S 0C S80S PO PO OISO 0SS SO SONE BSOS PCPOS SO 25

e« PUlDS ceeerceconscsaccsssssacsscssnssssosssssccscscescense 26
« Healt EXChBNgers .ceccecsscsscsssssssssssscssccscesssscasa 29
+« Reactor Vessels sceeesscevessscesssscssssscssssscscscoe 39
e Other VesSSelsS cescscessscsssssssccssssssssssssccaccssns 30
o JOInts and VaBlVeS ecceceososcessssssssscssssscnsscsnscsee 39
. Instrument and Control COMPONENtS ceecceescescoccsccsess HO

O\ AN O

C. Component SYSLEmMS eeeeecsccccosscccsssscsssccscsssssssscccnes 45

1. Salt and Liquid Metal Charging and Storage Systems .... 45
2, Off=0as HANATINg seeececcccccncccacosscosesssssonsscsce UO
3. Inert Gas SYStEM ceccecccccocrccnsoscoscaccccssocsnsoes U4S
4. Heating and Cooling of COMPONENLES +eesceccocconoossensa US

D. NuClear Considerations @ * 000 OO 0O O et PO IE DSBS OO0 OES RN SSasS 50

Previous Work and Early Consideration sececececcccecscss 50
One Region ReACtOTS sececccescococssssssscccscsscsccsaas 5
« TwO Reglon ReacClOI'S ceeecesscscccccssnsscsssscsssscssses 59
. Reactivity Effects in Typical ReQCLOT .cveecccsscecccess 69

W

E. Reactor Operation, Control and Safety ecceseccecssccccccaass 70

l. The Control Problem of Nuclear Power Reactors seceseeee 1
2. F‘lleling S & & 0 0O 0O 0PV B OB O PO e OO OO PAEARDE OO R O PRSP ESIPSOES 76
3. Criticality Staz't-up o2 0 60 0 000 00 O "B OO O S BSOS PO SE NSO eI e 77

F. Build-up of Nuclear Poisons and Chemical Processing s.ccce.. 80
1. Fission Product POiSORING ececcecescoccscncssscsssnsecse S0

2. Pa255’ NP237 and NP239 POisoning * 0 &9 0SB E SO S SO0 SEeD RS 82
3, Corrosion Product POLiSONING sececcacsscsccsascscccscccs 83

 

 
R UMENTED m*

 

k., Chemical Processing and Fuel Reconstitution .cecosecceeo 83
5. Build-U.P Of E'Ven-MaSS"NUInber Uranim ISOtOPes eeceeces o 88
6. Rwioa’ctive Waste Disposal 24 8 8 8 8 0 &9 0 00 "B OSSOSO S ODODOOSBSND 91

G. F‘llel CycleEcOnomics ® 0 8 & 88 85 0 08 900 S PSS OSSO SO S OO0 S SE S S S S DOOOCO 95

l e Cost mse s ® 0 0 O 8 0 00 0 02O SO 80O OO S OB OO eSO AE SN S SSSBSEESEOOO0DOODO0OO0OS 95
2. "Steady State" Neutron Balances and Comparative
Fuel costs O 0O PR OO O F PP OE PP EP S OO NP ® OO0 MO8 0D 914'

SECTION III - Reference mSi@ ReaCtor @0 600 00 0QC0OOSHDAISOIIODPDOROODODE O 98
A. IntrOduCtion 0SSP O8O 00N P EE OO 0P OSSOSO SO SDONO0EDGOS NS00SO RS 98
B, Heat Generation, Transfer, and Conversion Syste® .ccecsccoe 103

l. Reactor * 0 OO0 00RO ESECE OSSN OSSO0 000GS OO0 800000 105
2 @ Heat mchmlgers ® O 5 8 00 5 88508 PSS BSOSO NSO SDRPOSOS O OO0OO0OCSS OSSO loh
5 * stem cycle ® D O 8 8 8 80 8 8 8 50 800800 S PRS0 000 S 9S8O0 O0Aa S8 00008 DH lll

C. Components and Component SysStell scceecscessesssscosscesscocea L1l3

1. PUNDPS eecoscrssvsssscssssssassansoscssosssssncsssssoscses 113
2o VALVES vesecessesscsssscessossssscsesssossssassnccscnace LoD
3, Pipes and TUDES eeeeecesessossscossosconsoasaocsososos LLU
h, PFill-and-Drain Tanks .eesecccoescsccsocoasscssosaonooo LLI
5. Gas Supply Systems (Helium, Nitrogen, and Compressed .
O I
6. OFf=CaS SYSTLEM eeesercocccoooncsancnoossassocoscoosose LLT
7. Preheating and Temperature Maintenance .ccoecococecooscos 1LY

D. Pl‘allt Ial.yout OO0 6 8 00 000 ¢ PO P PD SO DT OO SO SS90 9P S BsCOEROCO0 0000000 118
E. Chemical Processing and Fuel Cycle EcCOROMICS soeees00000ess 122

1. Core ProceSSiNg ececsssossecscascsssosssscscsnsconcescns Lol
2. Blanket Processing cceevecsscsccssosncenssccooecconces 122
3, Chemical Processing COStS ccecesccscscsscosssocssccsoo L2k
k, TFuel Cycle ECONOMICS veeeescecesccascscsncoscocsacosse 125

F. COSt AnahlySiS ® 6 0 5 60 600 0 0" 0P PSSO S SO S0 SN CO0E0OCC0CSeE SO R 0000 6N 126

TNET0AUCHLION eeesevoesacoscecoessacssvsssssanssoncoosana 12O
Materials and Components Development COStS ccoecoccsos 127
Design and Construction COStS seccescnesccscseccessccs 129
Cost of Power from the Reference Desigh Reactor ccsceo 130

Fobr

o

Appendix 8 & 00 ¢ 90 ® 00 S 0O D OO OO O OO B BN OO PP RFOO S SEDeEP0R00000eeB000O0D0O0CSEe®BD 155

o
Fig.
No.

 

wVie

LIST OF FIGURES

Title

 

Fission Cross Sections and Eta for U233 and U235 in the
Ocusol=A Program.

Reference Design Reactor Heat Transfer Clrcuit Showing
Simulator Constants.

Change in Coolant Inlet Temperature for Intermediate Heat
Exchanger Due to Fuel Burn-up, for a Typical Fused Salt
Circulating Fuel Reactor at a Power Density of 200 watts/cm?.
Pused Salt-Fluoride Volatility Uranium Recovery Process.

UFg Reduction Process Flow Sheet.

Schematic Diagram of Heat Transfer System.

Reference Design Reactor.

Temperature-Heat Diagrams for Heat Exchangers.

Steam Cycle Diagram.

Plan View of Power Plant.

Section Through Reactor and Power Plant.

Page

56

5

86
89
101
105

108

120

121
-1 -

A PRELIMINARY STUDY OF MOLTEN SALT POWER REACTORS

SECTION T

Summary, Recommendations and Acknowledgments

Molten salts provide the basis of a new family of liquid fue1 power
reactors. The wide range of solubility of uranium, thorium and plutonium com-
pounds makes the system flexible, and allows the consideration of a variety of
reactors. Suitable salt mixtures have meltinglpoints in the 850-950°F range
and will probably prove to be sufficiently compatible with known alloys to pro-
vide long-lived components, if the temperature is kept below 13000F° Thus the
salt systems naturally tend to operate in a temperature region suitable for
mocern steam plants and achleve these temperatures in unpressurized systems.

The molten salt reactor system, for purposes other than electric power
generation, is not new. Intensive research and development over the past seven
Years under ANP sponsorship has provided reasonable answers to a majority of the
obvious difficulties. One of the most important of these is the ability to handle
liquids at high temperatures and to maintain them above their melting points. A
great deal of information on the chemical and physical properties of a wide variety
of molten salts has been obtained, and methods- are in operation for their manufac-
ture, purification and handling. It has been fOund that the simple ionic salts
are stable under radiation, and suffer no deterioration other than the build-up
of fission products.

The molten salt system has the usual benefits attributed to fluid fuel
systems. The principal advantages claimed over solid fuel elements are: (1) the

lack of radiation damage that can limit fuel burn-up; (2) the avoidance of the
-2 -

expense of fabricating new fuel elements; (3) the possibility (partially demon-
strated in the ARE) of continuous gaseous fission product removal; (4) a high
negative temperature coefficient of reactivity; and (5) the ability to add make-

up fuel as needed; so that provision of excess reactivity is unnecessary. The

latter two factors make possible & reactor without control rods, which automati-
cally adjusts its power in response to changes of the electrical load. The lack

of excess reactivity can lead to a reactor that is safe from nuclear power excursions.

In comparison with the aqueous systems, the molten salt system has three
outstanding advantages: it allows high temperature with low pressure; explosive
radioclytic gases are not formed; and it provides soluble thorium and plutonium
compounds. The compensating disadvantages, high melting point and basically
poorer neutron economy, are difficult to assess without further work.

Probably the most outstanding characteristics of the molten salt systems
is their chemical flexibility, i.e., the wide variety of molten salt solutions which
are of interest for reactor use. In this respect, the molten salt systems are prac-
tically unique; this is the essential advantage which they enjoy over the U-Bi
systems. Thus the molten salt systems are not to be thought of in terms of a
single reactor - rather, they are the basis for a new class of reactors. Included
in this class are all of the embodiments which comprise the whole of solid fuel
element technology: straight U235 or Pu burner, Th-U or Pu-U thermal converter or
breeder, Th-U or Pu-U fast converters or breedérsq Of possible short-term interest
is the U255 or Pu stralght burner: Dbecause of the inherently high temperatures and
because there are no fuel elements, the fuel cost in the salt system can be of the
order of 2 mills/kwh. Moreover, the molten salt system 1s, except for the molten
Pu alloy system, probably the only system which will allow plutonium to be burned

at high temperature in liquid form.

Ll
-3 -
The state of present technology suggests that homogeneous converters

using a base salt composed of BeF, and either Ii7F or NaF, and using UFh for

2
fuel and ThFh for a fertile material, are more suitable for early reactors than
are graphite moderated reactors or Pu fueled reactors. The conversion ratioc in
such an early system might reach 0.6. The chief virtues of this class of molten
salt reactor are that it is based on well explored principles and that the use of
g simple fuel cycle should lead to low fuel c¢cycle costs.

With further development, the same base salt (using Li7F) can be com-
bined with a graphite moderator in a heterogeneous arrangement to provide a

self-contained thorium-U233

system with a breeding ratio of about one. The ¢hief
advantage of the molten salt system over other liquid systems in pursuing this
objective is, as has been mentioned, that it is the only system in which a soluble
thorium compound can be used, and thus the problem of slurry handling is avoided.

Plutonium is an alternate fuel in the fluoride salt system. Only moderate
breeding ratics are expected in thermal or epithermal reactors; but a small, highly
concentrated flucride reactor may be fast enough to provide a breeding ratio of one.
Eventuzally the use of chloride salts might provide a fast plutonium reactor with
& breeding gain, although this would require use of separated 0137° The plutonium
system needs additional research to determine the stability of Pu compounds and
to provide a suitable chemical processing system.

The present report is primarily intended to summarize the state of the
molten salt art as applied to c¢ivilian power. The report is divided into three
parts. Section I is the summary and recommendations. Section II is a survey of
the state of molten salt technology. Section IITI is an analysis of one possible

T

molten salt reactor embodiment - a two reglon converter based on the fuel 69 Ii F -

30 BeF2 - 1 UFh and the blanket composition Tk LiTF - 26 ThFh. ~This embodiment,
-4 .

called the reference design, has been examined carefully, primarily to bring to

focus the problems which may arise if a full-scale molten salt system were to be

built soon.

The conclusion which we can draw from our study of the molten salt

situation is that a large-scale molten salt reactor - either a straight U burner

or a non-breeding converter - could be built, but that prior to building, two

important questions should be answered:

1.

Will any molten salt reactor produce economical power? Our study
shows the answer is probably yes, provided longevity of components

can be assured. Hence the issue depends on the second question:

From what we know about materials compatibility, how likely are we

to develop a salt and a container metal which will last for many

years of operation? This is the central issue in the civilian molten
salt nuclear power reactor program. The information gathered by the

ANP project, added to our general knowledge of the mechanism of attack
on metals (particularly INOR-8) by fluorides, suggests that the outlook
for a solution tc this problem is very good. However, very little long-
term testing at power reactor temperatures (e~ lEOOoF) has been done;
our recommendations, therefore, center around thé necessity for acquir-
ing this long-term data as soon as possible. Should these tests demon-
strate;the long-term compatibility of-materials, there will still be
required the development of reliable large-scale components. Experience
on ANP indicates that this part of the development should not present

major difficulties.
Recommendations

 

In view of the preceding, we recommend:

1. The long-term corrosion resistance of the proposed alloys in the
salts that could be used in a power reactor should be established. This will
involve the operation of a number of pumped loops incorporating a temperature
gradient, to be operated at the temperature of interest for periods of at least
a year.

2. The effects of radiation and fission product build-up on the com-
patibility of the salt and alloy should be thoroughly investigated. At least
two in-pile pumped loops simulating the condition in a molten salt reactor should
be operated for a long period of time. These in-pile loop tests should bhe supple-
mented by small-scale studies of the behavior of fission products.

3. It is recommended that a modest reactor study effort be maintained.
Different embodiments of molten salt reactors would be examined, so that if favor-
able results from Items (1) and (2) are obtained, it will be possible to recommend
a specific reactor, probably a burner or converter, for design and construction.
Also, the problem of remote maintenance, which 1s shared by all circulating fuel
reactors, could be examined in further detail.

Y}, Since there is always a time lag between the initiation of research
and the availability of practical developments, it is recommended that a modest
research program aimed at longer term possibilities be maintained. Objectives
would be (a) the incorporation of a solid moderator, (b) utilization of plutonium
in the molten salt system, (c) better alloys, and (d) improved fission product

removal systems,
Acknowledgements

 

Many members of ORNL and other organizetions have helped in the work
of the group and have shown great interest in its progress. It is difficult to
single out individuals for mention, but the following people have been serving
on a project steering committee:

A. M. Weinberg
J. A. Swartout
R. A. Charpie
S. J. Cromer
W. K. Ergen
W. Re Grimes
W. H. Jordan
W. D. Manly

Others who have heen especially close to the project are:

E. S. Bettis

E. A. Franco-Ferreira
J. L. Gregg

F. Kertesz

W. B. McDonald

E. R. Mann

P. Patriarca

M. T. Robinson

H. W. Savage
-7 -

SECTION II

Survey And Analysis Of The State Of Molten Salt Power Reactor Technology

In the research and development program cerried out by the ANP for
the construction of the ARE and future high performasnce reactors, much technical
information hfis been derived which is applicable to power reactors. The purpose
of this section 1s to abstract the information that is most pertinent to the con-
struction of power reactors, and to provide adequate references to document
properly the summaries given. This has been supplemented by studies of nuclear
characteristies of homogeneous one and two region power reactors. It will be
seen that most of the information required to design a practical power reactor
is available. However, long-term tests of materials and components are lacking,

and they must be supplied by & power reactor research and development program.
A. Msterials

l. Fuel Carrier Evaluation
The applicability of molten salts to nuclear reactors has been ably

reviewed by W. R. Grimes and others Y, g/, by Crooks et al 2/, and Schuman E/o

 

1/ Grimes, W. R., et al, "Molten Salt Solutions"”, Proceedings of the Second
Fluid Fuels Development Conference, ORNL-CF-52-4-197 (1952), p. 320 et seq.,
Secret _ ,

Grimes, W. R., et al, "Fused Salt Systems", The Reactor Handbook, Vol. II,
Engineering, RH-2 (1955) p. 799 et seq., Unclassified

3/ Crooks, R. C., et al, "Fused Salt Mixtures as Potential Liquid Fuels for
Nuclear Power Reactors", BMI-864 (1953), Secret

L

Schuman, R. P., "A Discussion of Possible Homogeneous Reactor Fuels",
KAPL-63k4 (1951)

&

 
-

The most promising systems are those comprising the fluorides and chlorides of
the alkali metals, zirconium, and beryllium. These appear to possess the most
desirable combination of low neutron absorption, high solvent power, and chemical
inertness. In general, the chlorides have lower melting points, but appear to
be less stable and more corrosive than the fluorides.  The use of chlorides in
& homogeneous fast reactor would be preferable except that the strong (n,p)
reaction exhibited by 0135 would necessitate the separation of the chlorine
isotopes.

The fluoride systems appear to be preferable for use in thermal and
epithermal reactors. Many mixtureé have been investigated, mainly at ORNL and
at Mound Iasboratory. The physical properties of these mixtures, in so far as
they are known, have been tabulated by Cohen et al 2/. Phase studies are exten-
sively reported é/o

117

has an attractively low capture cross section (0.0189 varns at
0.0759 ev); but Ii6, which eomprises'To5 percent of the natural mixture, has a
capture cross section of 542 barns at this energy. The cross sections for several

compositions are shown in Table I; also shown are the thermal cross sections of

Na, K, Rb, and Cs.

Table I

CAPTURE CROSS SECTIONS OF AIKALI METAIS AT 0.0759 ev (llBOOF)

 

Element Cross Section, barns
Lithium 6

0.1 % Ii 0.561

0.01 " 0.0731

0.001 " 0.0243%

0.0001 " 0,019k
Sodium 0.290
Potassium 1.13%0
Rubidium 0.401
Cesium 29

 

&2

Coben, S. I., et al, "A Physical Property Summary for ANP Fluoride Mixtures",
ORNL-2150 (1956), Secret C-84

§/ The Atomic Energy Commission, The Reactor Handbook, Vol 2, Engineering, RH-2
(1955), Secret, and ANP Quarterly Reports

 

 
-9 -

The capture cross sections at higher energies presumably stand in

7 has

approximately the same relation as at thermal. It is seen that purified Ii
an attractively low cross section in comparison to the other alkali metals, and
that sodium is the next bept alkali metal.

The fluorides of Ii, Ne, K, and Rb melt at 1550, 1820, 1560, and 1460°F,
respectively Z/° Binary mixtures of these salts with UFH form eutectics having

melting points and compositions shown in Table II.

Table II
BINARY EUTECTICS OF UFh AND AIKAILI FIUORIDES

Alkali Fluoride Mole % UFL in Eutectic Melting Point, °p

 

 

 

IiF 26 915
NaF 26 11%0
KF 14 1345
RbF 10 1330

With the possible exception of the first, these combinations are too
high melting to be attractive as fuels; however, the eutectics of UFh with IiF
and NaF might be suitable for use in the blanket of a two region plutonium breeder-
converter. DBinary mixtures containing less than 1.0 mole percent UFh do not exhibit
liquidus temperatures below 1450°F.

IAF and NaF form an eutectic melting at 1204°F 8/. Small adaitions of
UFh raise the liquidus temperature slightly. The ternary eutectic melts somewhat
below 8hO°F and contains approximately 30 mole percent UFh' This system is attrac-

tive only as a blanket material.

 

7/ The Atomic Energy Commission, The Reactor Handbook, Vol. 2, Engineering,
RHE-2 (1955), Secret

8/ TIbid., p. 948

 
~ 10 -

The Na-Zr fluoride system has been extensively studied at ORNL and a
phase diagram published 2/, An eutectic containing about 42 mole percent Zth
melts at 9lOOF° Small additions of UFh lower the melting point appreciably.

A fuel of this type was successfully used in the Aircraft Reactor Experiment.
Inconel is reasonably resistant to corrosion by this system at 1500°F° Although
long~term data are lacking, there is theoretical reason to expect the corrosion
rate at 1200°F to be sufficiently low that Inconel equipment would last several
years.

However; in relation to its use in a central station power reactor,
the Na-Zr fluoride system has several serious disadvantages. The Na capture
cross section is less favorable than the 117 cross section. More important,
recent data 19/ indicate that the capture cross section of Zr is intolerably
high in the epithermal and intermediate neutron energy ranges. In addition,
there is the sc-called "snow" problem, iaené ZrF) tends to evaporate from the
fuel and crystallize on surfaces exposed to the vapor. In comparison to the
Ii-Be system discussed below, the Na-Zr system has inferior heat transfer and
cooling effectiveness. Finally, the expectation at Oak Ridge is that the INOR-8
alloys will prove to be as resistant to the Be salts as to the Zr salts, and that
there is, therefore, no compelling reason for selecting the Na-Zr system.

The capture cross section of beryllium appears to be satisfactorily
low at all enefgies° A new phase diagram for the system IiF-BeF. has recently

2
been published ll/. A mixture containing 31 mole percent BeF,, (Mixture Th)

 

9/ The Atomic Energy Commission, The Reactor Handbook, Vol. 2, Engineering,
RH-2 (1955), p. 952, Secret

Macklin, R. L., Private Communication, ORNL (1957)

&k

Eichelberger, J. F. and Jones, L. V., "Iiquid Cycle Reactors, Fused Salts
Research Project - Report”, ML-CF-57-1-10 (1957), p. 3, Secret (Supersedes
Figure 6.2.26, p. 950 of The Reactor Handbook)
- 11 -

reportedly liquifies at 968°F; however, Cohen et al lg/give 941°F as the 1iquidus
temperature of Mixture Th. Other physical properties are listed by Cohen, who
gives 7.5 cp for the viscosity at 11120F. Further additions of BeF2 increase the
viscosity lé/n A new ternary diagram for the system LiF-BeFE-UFh has recently
been published l&/. Additions of UFh to the compound LieBth (1iquidus tempera-
ture BTBOF) lower the liquidus temperafure appreciably. A mixture melting somewhat
below 840°F (possibly as low as 820°F) can be obtained, having about 5 mole percent
UFho The ternary eutectic melts at 805°F and contains about 8 percent UFh’ 22 per-
cent BeFé, and 7O percent LiF. The system LiF-BeFé is attractive as a fuel carrier.

Substantial concentrations of ThFu in the core fluid may be obtained by
blending Mixture T4 with 3 IiF - ThFh, and a liquidus temperature diagram for the
ternary system has been determined li/_ The liquidus temperfiture along the join
between Mixture T4 and 3 IiF ° ThFh appears to lie below 950°F for mixtures con-
taining up to 10 mole percent ThFh. The liquidus temperature thereafter rises
slowly at first, and then more rapidly. ©Small additions of UFh to any of these
mixtures should lower the liquidus temperature somewhat.

No data on the system NaF-BeFE-ThFh are available; however, tpe solubility

of ThFh and other physical properties are expected to be nearly as good as for the

Li-Be system.

 

12/ Cohen, S. I., et al, "A Physical Property Summary for ANP Fluoride
Mixtures", ORNL 2150 (1956), Secret C-8k4

Barton, C. J., Private Communication, ORNL (1957)

N

Eichelberger, J. F. and Jones, L. V., "Liquid Cycle Reactors, Fused Salts
Reactor Project - Report", ML-CF-57-1-10 (1957), p. 6, Secret (Supersedes
Figure 6.2.2, p. 930, Vol. 2 of The Reactor Handbook)

&

The Atomic Energy Commission, The Reactor Handbook, Vol. 2, Engineering,
RH-2 (1955), p. 962, Secret

 

 
- 12 -

Mixture T4 has moderating power substantially less than beryllium or
carbon; gzt stands in the relation 0.176, 0.064, and 0.037 for beryllium, graphite |
and Mixture Th, respectively.

Nuclear calculations on these systems were performed by means of the
Univac program Ocusol lé/, The ages from fission to various energies for Mix-
ture 74 were computed and listed in Table ITI, together with the corresponding

capture-escape probabilities.

Table III

NUCIEAR PROPERTIES OF MIXTURE Th-A
(69% IiF,* 31% BeF,)

 

o 2 Fission Neutrons
Ener ev e, cm Capture-Escape Probability
1234 207 0.973
112 298 0,971
10,16 396 0.96k
0,0759 591 0,848

* 11 isotopic composition: 99.99% 117
Cohen et al 17/ give 1.3 x IO'u/oF for the mean volumetric coefficient
of thermal expansion for Mixture T4 in the liquid state, presumably in the range
from 1100 to lSOOOF. This may be compared to the coefficient of Mixture %0
(50 NaF, 46 ZrF) , L UFh), which is 1.58 x 10‘”/°F. The heat capacity of the

liquid is given as 0.67 Btu/1b-CF and the density as 120 1b/ft° at 1150°F.

 

16/ Alexander, L. G., Carrison, D. A. and Roberts, J. T., "An Operating Manual
for the Univac Program Ocusol-A, A Modification of Eyewash", ORNL-CF- (in
preparation)

17/ Cohen, S. I., et al, "A Physical Property Summary for ANP Fluoride Mixtures",
ORNL-2150 (1956), Secret C-8k
- 13 -

The stability of alkali fluorides and zirconium fluoride toward heat
and radiation seems to be well established by the work at Oak Ridge. Beryllium
fluoride is thermally stable at temperatures of interest; preliminary in-pile
tests lé/ indicate that BeF2 is as stable toward radiation, including fission
fragments, as Ztho

The compatibility of the systems under consideration with container
materials and adjacent fluids is dealt with in later sections, as is also the
problem of processing irradiated fuels. Costs are listed in Section II-F.

On the basis of presently available information, the fuel carrier
salts which have been considered appear to stand in the following order of pre-
ference: IiF-BeF,; NaF-BeF.; LiF-NaF-BeF,. The LiF-BeF. system has slightly

2 2 2
7

better moderating power, lower parasitic absorption (if high purity Ii' can be
obtained), and adequate solubility for ThF) and UF). It may prove to be more

corrcsive than the NaF-BeFé system, and the cost is greater.

2. Blanket Material Evaluation

 

The Ii-Be-Th fluoride mixtures recommended above as fuel carrier appear
to be suitable for use in the blanket of a two region reactor. There is evidenceig/
that these mixtures when containing no UFh are much less corrosive than fuel bear-
ing mixtures. As mentioned above, a mixture containing 10 mole percent ThFh has
(according to the diagram on p. 962, Vol. 2 of The Reactor Handbook) a liquidus
tempersture of 9520F° If a safety margin of 100°F 1s specified, the minimum

blanket inlet temperature would be 1032°F.

 

18/ Keilholtz, G. W., et al, "Solid State Division Quarterly Progress Report
Ending May 10, 1952", ORNL-1301 (1952), Secret

19/ Blakely, J. P., "Corrosion Results of Be Salts in Thermal Convection Loops”,
Memo of April 6, 1956, to C. J. Barton, ORNL
 

- 14 -

It might be possible to dispense with the BeFé and use a mixture of
IiF and ThFho A phase diagram for this system is given 29/, The compound
53 LiF - ThFL melts at 107OOF, and may possibly be a satisfactory blanket fluid.
The density was estimated by the method of Cohen gl/ to be 4.55 g/ce at 11120F,
and the melt conteins about 2700 grams of thorium per liter of solution. The
viscosity has not been reported, but is not expected to be greater than 7 cp at
1100°F., The corrosion rate in Inconel is low gg/. Additions of NaF to this com-

pound should lower the liquidus temperature appreciably, perhaps as much as IOOOF.

3. Intermediate Coolants

From the standpoint of simplicity, it would be desirable to transfer the
reactor heat directly from the circulating fuel to the steam. This, however, has
several serious disadvantages, among them being the induction of radiocactivity in
the steam by delayed neutrons; the danger of contamination of the power-producing
equipment by leakage of fuel into the power loop, and the danger of nuclear or
other accidents in case of leakage of water into the core system. It therefore
seems desirable to employ intermediate coolants.

Among the intermediate coolants considered were water, organic liquids,
liquid metals, and molten salts. High pressure, and nuclear and chemical compati -
bility with fuel eliminate water. The organic liquids have poor thermal stability

above '1100°F. Among liquid metals, sodium (or NaK), mercury, lead, and bismuth

 

20/ Cuneo, D. R., "ANP Chemistry Section Progress Report for October 9-22, 1957",
ORNL-CF-56-10-121 (1956), Secret (Supersedes Fig. 6.2.31, p. 958 of The
Reactor Handbook)

21/ Cohen, S. I. and Jones, T. N., "A Summary of Density Measurements on Molten
Fluoride Mixtures and a Correlation Useful for Predicting Densities of
Fluoride Mixtures"”, ORNL-1702 (1954), Secret

22/ Doss, F. A., "Supplement to WR Salt Mixtures in Thermal Convection Loops",
Memo of October 5, 1956, to W. R. Grimes
- 15 -

were considered. JIead and bismuth appear to be excessively corrosive (mass

transfer effects). Mercury has poor heat transfer characteristices and has

special problems of containment. |
Sodium has relatively good heat transfer characteristics, can be

readily pumped, but is chemically incompatible with both UFh bearing salts

and water. The reaction of sodium with a Zr based fuel in a pump loop with

a simulated leak was investigated by L. A. Mann gé/’ 22/. It appears that slow

addition of sodium to the system IiF-BeFé—UFh would result first in the reduc-

tion of the UFh to UF This would probably not result in the formation of a

3°
precipitate at ¢oncentrations of UFh under consideration. The UF3 and ThFh
would be reduced next, and then the BeFe. Solid phases containing uranium
metal would probably be formed shortly after the reduction of the thorium begins.
Molten salts considered for intermediate coolants include Mixtures T4
(three variations), 12 and 84. A study of these, together with metallic sodium,
was performed by means of a simplified systems analysis. The results, together
with relevant physical properties, are listed in Table IV. It is seen that Mix-
ture TU-A, which is the base recommended for the fuel mixture, has a melting
point too high for safety, being only 3hoF less than the proposed intermediate
coolant inlet temperature (IOOOOF). Mixture T4-C has a satisfactorily low
melting point, but the viscosity (14.0 cp) seems excessive. Mixture T4-B appears
to be suitable from standpoint of both melting point and viscosity. It would
also be completely compatible with a ffiel based on Mixture T4. Ieakage of Mix-

ture TL-B into the fuel circuit would not result in the formation of precipitates,

 

23/ Mann, L. A., Private Communication, ORNL (1957)

2/ Grimes, W. R. and Mann, L. A., "Reactions of Fluoride Mixtures with
Reducing Agents", ORNI-1439 (1952), p. 118
- 16 -

would not contaminate the fuel provided the lithium were of the same isotopic »
composition as that in the fuel, and could only decrease the reactivity by dilu-
tion of the fuel. It should be possible to tolerate small, continuous leaks in

normal operation.

 

On the other hand, salts containing beryllium are incompatible with
sodium metal, which displaces beryllium from the fluoride compound. The reaction
is expected to be energetic and rapid, but not explosive, since no gases are
formed 22/. It was estimated from the heat of formation data given by Quill et
al Eé/ that the addition of one mole of sodium to BeFé would release about 30 Kecal
of heat at 1000°F. This is sufficient to raise the temperature of the stoichio-
metric mixture about 13000F above the initial temperature. In addition, the
beryllium metal formed would deposit throughout the system and might lead to
embrittlement of the material of construction. The consequences of the leak-
age of sodium metal into IiF-BeFé thus could be serious. .
Mixture 12 (a Flinak) appears to be completely inert toward sodium.
From a heat transfer standpoint, it appears to have a slight advantage over Mix-
ture T4-B, as shown in Table IV, where the required heat transfer areas are compared.
Mixture 12 may be slightly more corrosive than Mixture T4 22/, but the corrosion in
the intermediate coolant loop is not expected to be critical because of the lower
temperatures prevailing there. It has fairly good compatibility with a fuel based
on Mixture T4 (LiF-BeFe),_ Small leaks of Mixture 12 into the core system probably

would not result in the formation of precipitates. The potassium would poison the

nuclear reaction, as would also any Ii6 present. Iarger lesks might lead to the

 

25/ Grimes, W. R., Private Commumnication (1957)
26/ Quill, L. L. (editor), "Chemistry and Metallurgy of Miscellaneous Materials-

Thermodynamics”, National Nuclear Energy Series IV-19B, MeGraw-Hill Book Co.,
New York, N. Y. (1950)

W _

 
 

- 17 -

precipitation of binary compounds of XF with UFh and ThFh. Precipitation of
UFh outside the core would tend to decrease the reactivity in the core; the
precipitation of ThFh would have the opposite effect. The precise course of
events cannot at present be predicted, but it seems doubtful that the reactivity
increase due to the precipitation of ThFh could override the decreases due to
the precipitation of UFh and the addition of X and 1160

On the basis of these considerations, Mixture 12 would appear the
safest choice of intermediate coolant.

By comparing the results listed for metallic sodium in Teble IV with
those for the salts; the penalty imposed by the use of salts as intermediate
coolants can be assessed. The heat transfer areas and fuel holdup volumes are
significantly less with sodium. The power consumption for pumping fuel and
sodium is excessive for the case where the tube pitch is the minimum allowable.
Doubling the tube pitch reduces the pumping power to a negligibly low level

without increasing the heat transfer area or fuel volume excessively.

4., Container Materials

 

The feasibility of the molten salt reactor system depends in large
measure c¢n the existence of a suitable container material. Any corrosion of
the container metal must be slow enough so that components will be long-lived.
The container material must be obtainable in sufficient quality and quantities
from commercial vendors, must be fabricable into suitable shapes, and must have
satisfactory strength, creep characteristics; and other physical properties at
the operating temperatures to be encountered.

Several hundred high temperature static and dynamic tests have been
carried out since 1950 to determine which materials were most practical for con-

tainment of the fluoride salts EZ/. Of the pure metals, molybdenum; columbium

 

27 ORNL-1491
- 18 -

Table IV

COMPARISON OF INTERMEDIATE COOLANTS

Fuel Inlet - 1100°F
Coolant Inlet - 1000°F
0.378" x 0.039" Tubes on minimum allowable pitech (0.495 in.)

Basgis:

Mixture Number

Composition, mole %
IiF
NaF
KF

BeF2

Melting Point, °F

Density; p o
1b/ft5 at 110073

Viscosity, p
cp at 11000F

Heat Capacity, Cp
Btu/1b-OF at 1292°F

Thermal Conductivity,
K, Btu/hr-ft-°F

Molecular Weight, M
gm/mole

Relative Heat Transfer
Surface

Relative Pumping Power,

g *

Fuel Holdup Volume, ft5
Compatibility with

Fuel

Compatibility with
Na

Induced Radiocactivity

* Percent of heat transferred

Th-A
69.0

31.0
966

120
T3
0.67
4.2

2.4

Good

Poor

Low

T4-B
62.7

37.3
842

120
9.0

(0.67)

(4.2)
33.7
1.09
0.068
125
Good
Poor

Low

Th-C
56.8
43.2
797

124

14.0

Good
Poor

Low

%
(1) Flow area on shell side same as for salts

Fuel Outlet - 1200°F

12

46.5
11.5
42.0

849

131

5.0

0.45

2.6

41,2

1.00

0.076

11k

Fair

Good

Moderate

(2) Flow area on shell side twice as great as for salts

8L

35.0
27.0
38.0
640
125

8.1

3.2
38.3
1.15
0.076
151
Poor
Poor

Moderate

Coolant Outlet - 1125°F

Na

0.30

57

High
- 19 -

and nickel were outstanding in corrosion resistance, but were eliminated for

reasons of fabrication difficulties and/or physical property shortcomings. From

the numerous alloys tested, Inconel was selected for extensive additional testing
and study in both thermal convection and pumped loops containing large temperature
gradients. Iater work has indicated the greater desirability of a nickel-molybdenum
alloy, INOR-8, and ORNL at present is active in trying to bring it into status as

a commercial alloy.

Both Inconel and INOR-8 alloys contain chromium. Their mechanism of
corrosion §§/ in NaF-Zth salts has been determined to be the diffusion of chromium
to the hottest metal surface, its solution in the salt there, and the diffusion of
chromium into the colder metal surfaces. In this way, subsurface voids ar: created
in the hottest region, from which the chromium is removed. The usual mechanism of
the corrosion is the same for both Inconel and INOR-8 alloy, and it is therefore
presumed that the general relstionships of corrosion rate to time, temperature,
and other test conditions found by analysis of hundreds of thermal convection
and pumped loop tests for Inconel in the NaF-Zth salt will also hold for INOR-8.
With cther salts there is the possibility of the deposition of free Cr metal in
the cold region, thus changing the rate limiting mechanism.

In tests of Inconel against Fuel-30 (50% NaF, 46% ZrF) 4, UFh)’

. approximately the same corrosion rates were found for both thermal convection
loops and pumped loops in which a 200°F to 3000F temperature difference was
maintained between the hot and cold sections. The depth of corrosion was found

to depend primarily on the metal wall temperatures 22/. Most tests were carried

 

28/ ORNL-2106, Parts 1-5, p. 96

29/ ORNL-2217

 
- 20 -

out at maximum metal wall temperatures of 1600--170001"‘° 1000-hour pumped loop
tests at 16OOOF wall temperature typically glve corrosion depths of 5 to 7
mils, and this increases to 9 to 10 mils at l'TOOOFo Thermal convection loops
with maximum wall temperatures estimated to be as low as 15500F give 2000 hours
corrosion of 7 mils, as compared to corrosion of 13 mils at 1640°F. Thus the
existence of a temperature coefficient is known, but its magnitude is uncertain
for extrapolating down to wall temperatures of lEOOOF.

With & maximum wall temperature of about 1600°F, the corrosion in the
usual pumped loops is about 3 mils in the first 15 hours, and 3 to 4 mils per
1000 hours after that. The longest tests were run for 3000 hours (1li4-mil attack)
and 8300 hours or sbout one year (25-mil attack). The time dependence is consis-
tent with the theory that in the first 15 hours, the salt becomes saturated with
chromium, and that after this initial period, the rate limiting step is that of
diffusion of the chromium into the walls at the cold temperature. Attempts to
show an increase of corrosion rate with an increase in area of the cold wall
section were inconclusive, however.

Early thermal convection loops 5—0-/ operated at 15OOOF showed a depend-
ence of corrosion on the difference in temperature between hot and cold legs, with
corrosion being decreased by a factor of 2 by lowering the temperature difference
to 15OOF. Tests with pumped loops él/, while not conclusive, suggest that tempera-
tu:e drops of greater than 200°F noticeably increase the attack, but that decreasing

the temperature drop from 200°F to 100°F does not correspondingly reduce the attack.

 

30/ ORNL-1729
31/ ORNL-2221

 
- 2] -

The UFL content of the salt definitely contributes to the corrosion é@{

probably because of the reduction of UFh to UF, by oxidation of chromium. A

3
reduction of uranium concentration from 4 mole percent to the less than one per-
cent suitable for a power reactor will substantially reduce the corrosion rate.

These data hafie been interpreted to indicate that with the 1200°F peak
temperature contemplated for a power reactor, the maximum depth of void formation
would not exceed 10 mils in one year or 20 mils in three years with Inconel. The
thinnest hot metal section will occur at the heat exchanger tubes, and these could
be designed to allow for a 3-year operation, as far as weakening due to corrosion
is concerned.

Tests on salts other than Composition-30 against Inconel have in most
cases yielded higher corrosion rates. In particular, lithium fluoride salts
are about twice as corrosive as Fuel-30 when tested against Inconel at a metal
temperature of 1600°F 22/. Under these conditions the IiF salts form a dendritic
growth of Cr in the cold leg.

A great many tests have been run on low chromium, low iron, nickel-
molybdenum alloys, but most of these have been at 1600°F wall temperature or
hi gher, and have been for fixed periods of 500 to 1000 hours. These nickel-
molybdenum alloys uniformly give low corrosion depths of the order of 1/2 to
2 mils under these conditions, and are thus estimated to be at least 5 times as
registant to corrosion as Inconel is to Fuel-30. Most of the tests on the nickel-
molybdenum alloys have been with the more corrosive salts containing lithium

fluoride; within the limits of the low corrosion rates, there does not seem to

be much dependence on the particular molten salts used.

 

32/  ORNL-1864
33/  ORNL-2217
- 22 -

If this superiority of performance can be extrapolated to lower tempera-
tures and longer times, the nickel-molybdenum alloy INOR-8 can be expected to
produce reactor components which should last many years before failure due to
corrosion. Obviously, what is needed here is some long-term testing under the
temperature and radiation cbnditions desired for a power reactor.

Other alloys now being examined by ORNL may prove even better for the
power reactor than INOR-8. These alloys do not contain chromium, and the ~mission
of chromium may result in complete thermodynamic stebility of the container =:d
salt. There remains the question of mass transfer attack as a result of tempera.-
ture gradients. One of the alloys being tested substitutes niobium for chromium
in the INOR-8 type alloy and is currently being called INOR-9. It is not clear,
however, just how much work on new alloys will be undertaken without active
sponsorship from a power reactor program.

Other Properties - Inconel has the following nominal composition:

 

Ni - TT% C - 0.08%
Cr - 15% Mn - 0.25%
Fe - 7% S - 0.007%
Si - 0.25% Cu - 0.2%

Its melting point is 25&0-2600°F, Some of its properties 25/ in the range of
temperature suitable for a power reactor are given in Table V. Inconel is a
commercial alloy and fabrication techniques are well established.

In the experimental program on INOR-8 carried out at Osk Ridge, a range
of composition was considered. Table VI gives the properties of INOR-8 as of

January 15, 1957 22/@

 

34/  PWAC-56L
35/  ORNL-CF-57-1-108
- 23 -

Table V
REPRESENTATIVE PROPERTIES OF INCONEL

Ultimate strength:

Strain Ultimate strength at 1350°F
0.04"/" /min 37,500 psi
0.2 "/"/min 45,000 psi

Stress to rupture:

1000 hour stress rupture strength

1200°F 14,000 psi
1350°F 6,000 psi

Other physical properties:
elastic modulus 25.5 x 106 psl at 12009F
Poisson ratio 0.32 at 1200°F
thermal conductivity 12 Btu/hr-£t-°F at 1200°F

mean thermal expansion coefficient 8.8 x 10°°/°F 32%F - 1300°F

- mn e Emam Ay amomm

Table VI

PROPERTIES OF INOR-8
(This table is based on experience with the following compositions)

Mo  Cr Fe Other oM
Composition Range 10-24 3-10 4-10 0.5 Al; 0.5 Mn; 0.06 C Bal.
(Wt percent)
Number of Compositions 18
Number of Heats 25

Fabricability: Between Inconel and Hastelloy B, depending on the composition.

Oxidation Resistance: Rates for 6 percent Cr alloy O.4 to 1.6 mg/cm?/lOO hr at
15000F. Oxide scale is borderline with respect to stability (7 percent
results in a stable scale).

Joining: Weldability between Inconel and Hastelloy B. Brazeable in dry H2.

 

Stress Rupture: 11 tests, 3 stress levels, argon, Fuel No. 107, and 2 conditions.
Rupture life at 8000 psi and 1500°F: 242 - 900 hr.

Corrosion: Fuel No. 30, 1000 hr, 1500°F, 1 mil
Fuel No. 107, 1000 hr, 1500°F, 1/2 - 2 mils
Type of attack in molten salts - subsurface voids. Mass transfer
detected in sodium loops.

 
- 24 -
(Table VI - continued) .

Tensile Properties: Alloys containing 20 percent Mo or less show no tendencies

 

 

to be brittle; 9 tests, > temperatures, and 3 conditions. >
Test Temp. Yield Point, 0.2% Ultimate Elonagation
F Offset, psi T.S5. psi percent
RT 46 ,000-47,000 114,000-13%0,000 y7-51
1300 31,000-32,000 63,000-T73,000 13-21

As a result of the data leading to this tabulation, the following is chosen as
the nominal composition of INOR-8 for larger scale tests:

Alloy INOR-8 (nominal composition, weight percent)

 

Mo: 15-17 Mn: 0.8 max
Fe: 4-5 Si: 0.5 mex
Cr: 6-8 Ni: Balance

c: Oool‘l"‘o 008

— mm mWe mm wma vemm sems

5. Moderator Materials

 

The highest breeding ratios for molten salt thermal reactors will be
obtained with the use of & supplementary moderator. Although the salt Mixture-Ti4 .
has a slowing down power of 60 percent that of normal reactor graphite, its slow
neutron absorption is about 4 times that of graphite, and the lower neutron absorp-
tion of the mixed moderator and salt is definitely beneficial in breeding ratio.

Moderator materials_that can most readily be considered in this high
temperature system are beryllium metal, beryllium oxide, and graphite. The former
two require canning for protection from the salt, and since the canning materials
have rather high neutron cross sections, their value in obtaining an improved
neutron economy is questionable. Uncenned graphite is the best hope for a moderator
that will provide good neutron economy.

Several graphite samples have been tested which show very little pene-

tration by the molten salt in short-term tests éé/, It is ressonable to expect

 

36/ ORNL-2221, p. 158

O

 
 

- 25 -

that one or more of these grades could be made commercially if longer term tests
confirm their penetration resistance, particularly under the effects of radiation.
The other problem associated with the use of graphite in direct contact
with the salt is whether or not the metals of the system will become carburized.
Preliminary tests with Inconel and Fuel-30 show no detrimental effects in 100

hours 21/, Further testing with other metals is underway, and there is hope that

 

a suitable metal system for use with bare graphite and a molten salt will be
developed. However, much more experimental work is required to demonstrate the

practicality of a molten salt-unclad graphite system.

B. Materials

This section is a survey of the factors involved in selecting com-
ponents, including an appraisal of the experience in actual operations.

All equipment and controls which are parts of, or directly affect, the
reactor: molten salt systems, liquid metal systems, and radicactive gas systems,
require components of much higher quality and greater dependability than has been
generally accepted as "commercial practice" in steam-electric plants. It should
be strongly emphasized that the high quality demanded of these systems is actually
necessary, and must be achieved because of the difficulty of maintenance of highly
radioactive systems. ZExperience at the development laboratory level has shown
that, with proper attention to detail, the standards of quality required are prac-
tical of attainment. Specifications and procedures have been rather thoroughly
developed and detailed. The importance of the strictest possible quality control

and inspection of absolutely every step can hardly be overemphasized.

 

37/ ORNL-2221, p. 187
- D6 -

All items not involved in handling molten salts, molten metals, or
radioactive solids, liquids; or gases require only the same standards of quality
as conventional installations. This includes turbines, condensers, feed-water
heaters, water pumps; generators, and electrical gear, together with their
instrumentation, controls, and auxiliary equipment, and most of the building

and building services.

1. Pumps

The choice of pumps for liquid metals is among three proven types:

(1) electromagnetic; (2) canned rotor; and (%) gas-sealed. The frozen seal 38/
type may be sufficiently proven within the next two or three years to add to
the list, when more actusl operating data are available.

Electrcmagnetic pumps of several different designs have beefi success-
fully operated with little or no trouble I>r several years. Among the larger
users are: General Electric Company at KAI'[; Aircraft Reactor Engineering Divi-
sion (ARED) at ORNL; North American Aviation Company (Atomics International);
Westinghouse; Mine Safety Appliances Company, Callery, Pennsylvania; Brookhaven
National Iaboratory; and Argonne National Iaboratory. Among other users are
Battelle Memorial Institute; Allis-Chalmers Company, and Babcock and Wilcox
Company. Most applications have been at temperatures below 900°F, and at rather
low capacities of only a few gallons per minute. Some experience exists at ORNL
and other places at temperatures up to lSOOOF and higher. EM pumps of 5000 gpm
at 100 psi and even higher capacities are advertised for sale by General Electric
Company, Westinghouse Electric Company, Allis-Chalmers Company, and perhaps others.
Very recently, EM pumps with efficiencies greater than 20 percent are reported

to have been developed 22/.

 

o

38/  NAA-SR-1804 (TID-7525); ANP Quarterly Reports

39/ Information from Broockhaven National Iaboratory, etc.
- 27 -

Canned rotor pumps for liquid metals have been developed in the past
three or four years by Westinghouse, Allis-Chalmers, and others for high perform-
ance operation. Some difficulties inherent in canned rotor pumps are: friction
problems in bearings during start-up and shutdown operations; low efficiencies;
thin and relatively fraglle walls between stator and rotor, and the requirement
for keeping the temperature of rotor and stator low.

Both EM pumps and canned rotor pumps have the advantage of being
"sealless", that is, there is no communicsation via packing or gas passages to
the surrounding atmosphere. This is a major engineering advantage in pumping
oxygen-~-gsensitive liquid metals, obviating any need for more or less elaborate
sealing mechanisms. Each has the disadvantage of low efficiency and lack of
long~time proof testing with liquid metals at temperatures above 700 or 800°Fo

Gas-sealed centrifugal pumps &9/ are at present the most adequately
proven pumps for moving either liquid metal or molten salts at temperatures above
8OO°F° Efficiencies are good; no different than for centrifugal pumps in general. t
The limits of head and capaclty are similar to those of the ordinary centrifugal
pump. More than usual "overhang" (distance from nearest bearing to impeller) is
required, so that the seals and bearings can be maintained at a low temperature.
Gas-sealed pumps were used exclusively for molten salt and liquid metals in the
ARE operation, and in the scores of high perfofmance, high temperature (lOOOoF
to 1700°F) heat exchanger tests, pump tests, pumped system tests, in-pile loop

tests, etc., operated by ARED for pumping both molten salts and liquid metals Ei/o

 

40/ Complete details of design and performance are available from ORNL-ANP

41/ See ANP Quarterly Reports
- 28 -

The largest capacity high temperature pumps used to date by ANP are about 1200
gpm at 370 feet of head at 12000F and higher temperatures. Pumps of this type
are advertised commercially in capacities to 20,000 gpm. Their degree of
reliability is not known at ORNL.

The two most vulnerable parts of gas-sealed centrifugal pumps are
(1) the seal and (2) the bearings. To date, at ORNL, bearings are lubricated
by cireulating oil, & small part of which is also used to lubricate the seal.
Both the seal and the nearest bearing are usually within 18 inches of the pumped
liquid. Therefore, pumps for radiocactive fluids require shielding of the bear-
ing and sesl region and provision for replacement of the lubricant as it gradually
becomes damaged by radiation. Provisions for such replacement are included in
the ART pumps and lubricant systems designs Eg/,

Gas-sealed centrifugal pumps have been operated by ANP at 1200°F
for durations up to 8000 hours without bearing, seal, or other pump mainten-
ance. The operation of such pumps is now considered by ANP to be routine and
trouble-free. While some work is continuing there on the further refinement of
seals and shielding, the chief present use of such pumps in the Experimental
Engineering Division is to serve tests of other components requiring pumped hot
liquids {e.g., heat exchanger and radiator tests).

Possible future improvements in gas-sealed pumps for high temperature
liquids include the use of hydrodynamic bearings and liquid centrifugal seals.

Hydrodynamic bearings could reduce or remove the need for compromising
between "overhang” from the nearest oil-lubricated bearing and radiation damage
to the lubricant. Before using in a reactor system, a development program involv-
ing tests at operating temperatures, flows and heads would be required. A
difficulty is the close clearance, without rubbing, required between rotating

and static parts.

 

42/  "Design Report on the ART", ORNI-2095, p. 29 et seq.

o A
- 29 -

Iiquid seals similar to mercury seals sometimes used in Laboratory
bench tests would allow relaxation of tolerances now required in the mechanical
seals employed. A problem is the length of available sealing annulus required
for handling possible pressure surges. The combination of a liquid seal with
& hydrodynamic bearing would appear a good future prospect for power reactors.

Where two or more gas-sealed pumps operate in parallel;, free flow
between the liquid-to-gas interfaces in the pumps must be provided in order to
prevent surging of the level in one pump significantly above the levels in the
other pumps. The close coupling can be accomplished by providing large connect-
ing flow channels to maintain approximately equal levels in all the parallel
pumps of any one system, or, preferably, by containing the parallel pumps in
one single reservoir.

Pump design is well enough understood that larger pumps of a proven
type can be designed so that they can be confidently expected to deliver the
flows and heads calculated. However, a modest development program should be
expected and a series of thorough prove-in tests will be required before pro-

curing larger pumps on a routine basis.

s

2. Heat Exchangers

Heat exchangers for a power feactor will differ greatly in general
design from those used in the ANP reactors. The removal of space and weight
limitations allows the heat exchangers to be designed to minimize stresses and
reduce construction difficulties. On the other hand, the ANP date on heat
transfer coefficients and fluid flow, the metallurgy involved in fabrication,
and the chemistry of corrosion are invalusble in the design of a power reactor

heat transfer system.

 
- 30 -

To minimize thermal stresses, the heat exchangers should normelly be
of U-shaped construction. The two tube sheets are then side by side, and the
individual tubes are free to expand longitudinally with very little consequent
stress in the tube-to-tube sheet joints.

Countercurrent flow, which has two advantages, is possible with the
U-shaped configuration. Counter flow exchangers, in any given application,
require less heat transfer area, and have lower maximum local temperature dif-
ference between the two fluids than for any other configuration. The maximum
temperature drop through the tube wall;, and thus the tube wall thermal stress,
is consequently lower.

Although thermal stresses due to temperature differences between the
internal and external walls of the exchanger tubing tend to be relieved by creep
under steady operating conditions; the unavoidability of some thermal cycling
requires that such stresses hbe considered. They can be minimized by using thin
tube walls which yield a small temperature difference between the inner and outer
surface. The small temperature drop through the wall is desirable in itself,
because it allows higher steam temperatures for the same fuel temperature. These
factors must be balanced against choosing the tube wall thickness for maximum
reliability. Tube walls greater than 20 percent of the diameter are poor from
the tube fabrication standpoint, but for small diameter tubes, this 20 percent
limit should probably be approached closely to minimize weakening effects asso-
ciated with high temperature immersion in the molten salt.

Sample calculations have shown that where minimum fluid volume has
great importance; the tube diameter should be as small as the technology will
allow for economical and reliable manufacture. The ease and reliability of
manufacture as a function of heat exchanger tube diameter have not been accu-

rately assessed. However, ANP experience with Inconel indicates that simple

-

-

 
- 31 -

fusion welds can be made for tubes between one-quarter inch and one-half inch
in diameter, and that for appreciably larger tubes, filler metal will have to
be added while welding. The present indications are that 5/16-inch and 3/8-inch
diameter tubing may be desirable.

ANP experience indicates that the weakest places in a heat exchanger
are in the neighborhood of tube-to-tube sheet joints, where stresses tend to be
a meximum. Thus special attention must be given to the design and metallurgical
aspects of this problem.. For the molten salt heat exchangers, the following types
of jolnt are all possibilities, and the one chosen wi;l be determined by further

development.

a. Face welded joint

 

The tube sheet is machined so that through most of its thickness, the
hole diameter is appreciably larger than the tube diameter. The remaining thick-
ness of the tube sheet is drilled to match the tube 0.D. closely, and a trepan
is machined to leave a thin section of the tube sheet to be welded to the tube
end. This type joint can be fusion welded for tube wall thicknesses no greater
than 0.060 inch. The presence of a crevice on the outside of the tube, where
it enters the tube sheet, precludes the use of such a Jjoint with water or steam
on the shell side of the exchanger, since chlorides and other impurities will
collect in such cracks and cause embrittlement. No comparable phenomenon has
been observed with sodium or molten salt heat exchangers, however, and such a .

crevice may not be detrimental in them.

b. DBack brazed joint
Brazing, or seal welding with back brazing, has been used extensively
in the ANP program. This avolds stress concentrations in the joints which might

lead to cracking, but the metallurgy of the braze may cause difficulty for

 
- 32 -

long-life applications. Diffusion of braze metal constituents into the tube
wall may cause embrittlement. For the salts, 100-hour tests on gold-nickel
alloy brazing have indicated corrosion resistance and little tendency to

diffusion Eé/o

¢c. Butt-welded joint

 

In this joint, the tubing is butt-welded to a nipple which may be
forged on the tube side of the tube sheet. The resulting joint is perhaps the
most satisfactory of all, but the fabrication is the most difficult, particularly
in small size tubes. Further testing is required before this type joint can be
considered as proven technology for the desired tube sizes.

All salt-to-salt heat exchangers should be constructed entirely of
INOR-8 for resistance to corrosion. It is probable that salt-to-sodium heat
exchangers can be made from INOR-8 also; however, this is subject to confirma-
tion of low mass transfer of nickel in sodium circuits by tests over long periods
of time. In the event that excessive mass transfer rates are encountered,; a
duplex construction, with INOR-8 on the salt side and 316 stainless steel on the
sodium side, can be used. Construction with duplex tubing is more difficult and
probably less reliable, but has been used in other applications.

The situation in regard to sodium-to-water or steam boilers or super-
heaters is being intensively studied by several organizations at the present
time, and it is anticipated that considerable progress in defining the best
solution will be made in the next two years. The basic difficulty is that leaks
develop, probably starting as small stress-corrosion cracks originating from the

water side. Both impure water and impure sodium are very corrosive, so that a

 

43/ ORNL-1934
-33-

small leak quickly spreads. The problem is either to achieve perfection in

leak-proof construction, or to design the system so that small leaks can be

detected quickly. One way of accomplishing leak detection is to have a double

tube construction, with a leak detecting fluid between the tubes. In one case,

mercury is used as the intermediate fluid; others are trying an inert gas be-

tween tubes that meke a partial metallic contact. Either of these solutions

requires large area boilers and superheaters because of the decreased heat

transfer coefficients.

Another approach is to use single walled tubes, but to use double

tube sheets, with the tubes sealed to each tube sheet. The idea behind this is

that leakage will be most probable into the space between the two headers and

can be detected there. This approach has been used in a steam generator on the

HRT project, and can be adapted to sodium-to-water boilers and superheaters.

Inconel has been suggested as a good material for the boilers and

superheaters, since it is not ordinarily subject to stress-corrosion cracking,

and the consequences of a small leak should be considerably lessened. Alternately,

ferritic steel, such as 2 1/4 Croloy, could be used in the boiler and either a

ferritic or an austenitic stainless steel, such as type 316, could be used in

the

high temperature service of the superheater. This is standard power plant

practice and can be adapted directly, since both metals are compatible with

gsodium. If the austenitic steel is used for the superheater, it must be pro-

tected from any but dry steam. For this reason, a once-through boiler, delivering

slightly superheated steam to the superheater, is indicated.

3,

are:

Reactor Vessels

 

The design requirements of the core vessel of a circulating fuel reactor

(1) geometry to fulfill nuclear requirements; (2) compatibility of container
- 34 -

material(s) and liquid(s); {(3) a fluid flow pattern which will, without excessive
fluid pressure drop, prevent local stagnation or eddies that might cause harmful
temperature transients; (4) low neutron poisoning by materials of construction;
and (5) adequate safety against core shell failure from mechanical or thermal
stress level or stress cycling.

Fortunately, criteria (1) and (2) are relatively easily fulfilled in
a circulating homogeneous molten salt fuel reactor system. Criterion (3) can
be fulfilled for low power density cores (less than 250 w/cc) by employing the
data and experience of the aqueous homogeneous reactor project and by tests in
transparent prototype mockup of actual geometry and fluid flow types. Criterion
(5) can be fulfilled by good choice of basic geometry and competent stress analysis
and adequate design safety factors. Criterion (h) can be satisfied more easily
for single region than for two region reactors because little or no material of
construction is needed in the region of high flux. Possibly further improvements
in this respect may be made in the future (e.g., if improvements in technology
allow for use of graphite as the core shell material). One of the main advantages
of circulating molten salt systems is the low pressure level and the consequent
low mechanical stress.

It should be noted here that the extreme complication of design of the
interior of reactor cores which has already become traditional with solid fuel
elements; moderators, control rods, etc., is virtually eliminated by the use of

& homogeneous fuel, moderator, and, to some extent at least, reflector.

a. Choice of number of regions

 

Involved in the choice of number of regions are: (1) transmutation
ratio potentiality; (2) ability to manufacture "pure" v or Pu; (3) flexi-

bility of operation and experimentation (different composition fluids in core

 

 
- 35 -
and blanket); and (4) cost of inventory, fabrication, and processing. These
aspects are discussed elsewhere in this report. More auxiliary equipment
(storage tanks, fill-and-drain tanks, pumps, heat exchangers, instruments,

ete.) will, of course, be required for a two reglon system.

b. Single region reactor vessels

 

For low neutron leaksges, single region reactors are large. In terms
of spheres, calculations indicate that the optimum diameter is between 10 and 20
feet, probably about 1k feet &E/. Uranium concentration decreases with diameter
increase and increases with increasing concentration of thorium. Note that the
external volume of fuel is independent of reactor size, but external holdup of
uranium is linear with concentration of uranium ifi the fuel.

Container shape is an important fabrication parameter. Precise, final
geometry will depend on optimized compromises determined by studies of: (1) nuclear
characteristics; (2) fluid dynamics, including temperature transient studies;

(3) stress analysis; (4) availability and fabricability of materials of construc-
tion; and (5) costs. In general, spheres are optimum from nuclear considerations,
cylinders for fabricability, spheres or spheroids for low stress, cylinders for
flow pattern optimization. Stress considerations include weight, pressure, flow,
and wall temperature patterns, and must therefore result in a compromise. Possibly
the best geometrical compromise is approximately a right circular cylinder with
ellipsoidal heads.

One important design decision is the choice of relative locations of

the entry and exit passages for the fuel. They may be placed at the same end

 

44/ The volume of & lh-foot sphere is 1437 ft2. Welght of fuel at 135 1b/ft” =
194,000 1b. Weight of l-inch shell at 500 1b/ft° = 26,000 1b. Power density
at 600 Mw = 17.5 w/cc
- % -

of the reactor by making use of concentric pipes, or they may be at opposite
ends for "straight-through" flow. The former appears to be better for avoid-
ing stresses between core shell and blanket shell in a two region reactor. For

a single region reactor, the latter appears to offer fewer complications.

c. Two region reactors

 

Because the transmutation ratio is such an important factor in the
mills/kwh cost of electrical povwer, and because in two region reactors the core
sfiell thickness has an important influence on the transmutation ratio (since it
is a neutron absorber), it is very desirable to design for the thinnest core
shell compatible with safety against failure. It is not believed that thermal
stress will be a significant factor in determining the thickness of the core
shell 52/ in low power density homogeneous reactors. Pressure stresses will
probably determine the thickness required. These matters must, however, be
investigated when firm data on heat flows and neutron and gamma fluxes become
available. If the core and blanket fluids are kept at low pressures by appro-
priate flow passage geometries and appropriate placement of the system pumps,
the thickness of the core shell can be kept as low as 5/16 inch.

At this thickness, the highest stress under normal operating condi-
tions for the Reference Design Reactor described in Section TII would be a
compressive stress of from 500 to 850 psi, depending on core shape. This is to
be compared with estimated long-time creep strengths of several thousand psi
at 1200°F. (Inconel, a weaker material than INOR-8 at these temperatures, has
a 1000-hour stress-rupture strength at 1200°F of about 14,000 psi.) Accidental

45/ Personal communication from H. F. Poppendiek, formerly Chief, Heat Transfer
and Physical Properties Section, ORNL
- 37 -~

core or shell drainage could prodfice short-time stresses of from TOO to 2000 psi,
depending on shape and condition. These stresses are to be compared with yield
strengths of the order of 30,000 psi at 1300°F. Against collapsing tendencies,
a 6-foot diemeter core with a 5/16-inch thick wall has a factor of safety of
from 7 to 100, depending on whether the shape is.cylindrical or spherical.

In two region reactors, it would probably be impractical to maintain
the core shell and the blanket shell at the same temperature level at all times.
Therefore, to prevent stress from differential thermel expansion of the two
shells, they must be designed to be free to expand and contract independently
of each other. This can be done by (1) employing straight-through flow with
one or more flexible connections between the two shells, or (2) by connecting
the two shells at only one end and directing the core fluid both in and out
of the same end. Since flexible connections such as bellows or diaphragms are
necessarily thin to allow flexibility, they are unavoidably subject to corrosion
or mechanical damage. It is therefore preferable to avoid using them by direct-
ing the core fluid into and out of the same end of the core, with no rigid
restraint on thermsal expansion and contraétion of the core shell by the blanket
shell. For thermal symmetry and resultant minimum stress concentration; the
core fluld should enter and leave the core via concentric passages. Studies
based on experience with aqueous homogeneous reactor flow tests, and considera-
tions of design of the expansion chamber, pumps, and flow passages will determine
whether the fuel should enter the core through the central tube and exit via the
annulus, or vice versa.

Blanket shells are not restricted in thickness by nuclear considera-
tions and may therefore be designed to any thickness and geometry found desirable

from stress, fluid dynamics, and support considerations. Because of the very low

 
- 38 -

power densities, restrictions on flow pattern can be relaxed considerably from
those required in the core. Somewhat more freedom is allowable in the use of
flow directing vanes near the outside of the blanket, where neutron flux and,
consequently, poisoning effects are low. Appropriate blanket geometry can
materially lessen shielding requirements.

Twc region reactors may be supported from either top or bottom.
Since core and blanket fluid pumps will probably be above the top of the reactor,
it is anticipated that the main supports should be near the top of the reactor

to minimize both vibration stresses and problems of thermal expansion.

k., Other Vessels

Vessels other than the reactor will be required, including: (1)
special storage vessels from which each of the salt and liquid metal systems
will be filled and to which they will drain; (2) storage tanks in which new
and used salts and liquid metals will be stored; and (3) one or more tanks for
the boiler make-up water, which will probably be of special purity and/or
composition.

Fill-and-drain tanks - Vessels in which fused salt for the fuel cir-

 

cuit or blanket circuit will be stored must be designed for both heating and
cooling. To minimize thermal shock, they should be preheated before receiving
molten salt; and it will be desirable to maintain them at temperatures above
the melting point of the salt whenever salt is in them. It must also be possi-
ble to prevent undesirably high temperatures from rising in them from "after-
heat” generated by beta and gamma decay of fission products. Preheating and
cooling may be provided in a number of different ways as described later in

Section II-C-L4.
- 39 -
Tanks for filling and draining non-radiocactive salts and liquid
metal systems will require provisions for heating, but not for cooling, and
may therefore be compact (right circular cylinders).
Transfer of molten salt or metal between containers through pipes or
tubes is usually accomplished by pressurization of the vessel to be emptied with
inert gas. Pumps are almost never used for this purpose. Simple gravity drains

may also be used, of course, where appropriate.

5. Jolnts and Valves

 

Joints for contalning liquid metals, molten salts, or radiocactive gas
require welding or brazing for satisfactory security against leakage. For long
life in high temperature joints, no brazing meterial has yet been proven adequately
safe; however, basic investigation of brazing is continuing and shows some promise
for these applications. Welding has proven to be very satisfactory for both leak-
tightness and long life. The high degree of competence that has been developed
in welding and the adequacy of the welded joints of Inconel are evident in the
lack of weld failures in the ARE and in the large number of circulating molten
salt and liquid metal systems operated by ANP. Design and procedure specifica-
tions for welds are spelled out in detail in the ORNL Welding Code (ORNL Metallurgy
Division). A few vendors have already been trained in, and have met the require-
ments of the code.

For containment of radiocactive gases at low temperatures, both welding
and brazing have been found to be satisfactory (zero leakage over long life). No
other methods of fabricating joints have proven to be thoroughly reliable over
long periods. Although many tests of API ring gasketed flanged joints and of
compression fitting joints such as the Swagelok have shown leak-tightness over
fairly long periods of non-cyclic operation, they are not considered to be depend-

able enough for this type of service.
 

- 40 -

Valves for handling radioactive gases in small (less than 1/2-inch)
lines do not present any unusual problems in attaining the required freedom
from leaks to or from the surrounding atmosphere. This requirement is met satis-
factorily by employing only bellows-sealed, welded or brazed connection valves.
Fully adequate valves for these small sizes for gases with limited radiocactivity
are available for "off-the-shelf" purchase. Valves for controlling gases with
high beta decay activity may require redesign to reduce internal free volume,
thereby reducing the amount of heat removal required.

Flow control valves which are required only to control flow rates
without entirely stopping flow in circuits of molten salt or liquid metal are
ifems of semi-routine design and fabrication, except for the requirement of
extreme leak-tightness to the atmosphere. Because of the requirement for leak-
tightness, bellows-sealed, welded connection type valves are specified. Maximum
inspection and acceptance testing are required, but no particular difficulty is
encountered in meeting the specifications. Good mechanical design and the care-
ful following of the weld design and procedure code are required.

Valves for stopping flow of hot salts or liquid metals, with zero or
almost zero through-flow required, are more difficult because of the requirement
for seat materials which will neither stick nor be attacked by the fluid at high
temperatures. This problem has been essentially solved by the use of cermet
valve seats and the accurate alignment of parts, but the solution is still too
new to be regarded as fully proven for temperastures of 12000F or higher until
the results of the additional tests now in progress are known. Stop valves
can be avoided in the hot molten salt and liquid metal circuits without serious
difficulty (for example, barometric legs or "freeze valves" can be employed in-
stead of mechanical valves in the sodium and salt drain lines). There are no

significant difficulties involved in the use of stop valves at temperatures

 
- 41 -

lower than about llOOoF; therefore, such valves can be freely used in low
temperature molten salt and in liquid metal drain lines.

ORNL has developed and is successfully using stop valves in molten
salt lines as large as 2-inch iron pipe size. Work toward further improvement

is continuing, with emphasis on optimizing seat materials and design.

6. Instrument and Control Components

 

Assessment of the adequacy of an existing instrument or control com-
ponent for a reactor starts from consideration of the function of the system
for which the component need be a part. Instruments are here evaluated in terms
of thelr usefulness in systems which fall into four broad categories, according
to their functions: those required for safe start-up, those vital for safe
control of the reactor once it is in operation, those needed for monitoring and
control of reactor auxiliary systems, and those installed for purposes of
evaluating performance and gathering data for experimental or test purposes.
Attention is focused here, in the main, only on sensory devices which are
needed to operate at temperatures up to 1300°F in fused salt or liquid metal
gystems. These include sensory devices for the measurement of temperature,
liquid level, fluid flow, rotation speeds, neutron flux, pressure, and for

the detection of lesks.

a. Temperature sensors

 

For sensing of high temperatures of interest, the thermocouple is
the device upon which principal reliance has thus far been placed. Present
ANP designs include a rugged chromel-alumel thermocouple of wire size B and
S No. 8 or larger, coupled with magnetic amplifiers of reliability equal to

that of A.C. transformers. Using fabrication and installation techniques

developed in the ANP program, these thermocouples have demonstrated no

 
- 4o .

detectable drift in a number of 3000-hour tests conducted at 1200°F, Enough
experience has been gained to Justify confidence that these temperature sensing
systems will operate in the 1000°F range with a plus or minus SOF accuracy
reliably over a 20-year period, in the absence of nuclear radiation.

The calibration stability of these instruments over extended times
in nuclear radiation fields has not been adequately tested. Consequently, at
the present state of the art, there is no assurance that thermocouples installed
in the core or blanket system would remain calibrated. Provision can be made in
reactor design for insertion of a calibrated thermocouple to recalibrate installed
thermocouples during periods when the reactor is isothermal, as, for example,
during shutdown.

Work has been done on & constant volume sodium or rubidium vapor
thermometer operating in the form of a bulb or thimble which is inserted into
the hot fluid. This device appears to be promising for measurement of the

temperatures of molten salts subject to a radiation field.

b. Iiquid level devices

 

At least two satisfactory liquid level sensors are presently avail-
able. These two have been tested for use in the ANP program. One is a resis-
tance probe for use with liquid metals, in which the resistance varies with
the level around the probe. The other is a float type &é/o Difficulties in
installation of these sensors have been experienced in highly compact reactors
due to access limitation. It should be possible to avoid this problem where

space is not at a premium.

 

46/  Southern, A. L., "Closed-Trap Ievel Indicator for Corrosive ILiquids
Operating at High Tbmperatures", Instrumentation, ORNL-2093

 
- L3 -

These instruments have performed satisfactorily and reliably in 3000-
hour duration tests, and there was no indication that they would not perform
for a much longer period. They are not vulnerable to nuclear rediation damage.
The equipment is sound in principle and promises to be satisfactory for many

years of service.

¢. Fluid flow measuring devices

 

Tests indicate the electromagnetic flow meter to be a dependable
device for measurement of high temperature liquid metal flow. Care in design,
fabrication, installation and calibration is needed, but the device is rugged
and durable and accuracy of 5 percent should be maintained.

The art of measurement of flow of molten salt at elevated temperatures
is not as well developed. Iimited tests have been made on venturi and rotating
vane types with varying degrees of success. Such devices, if needed in a power
reactor; will require further testing. The ANP program has under development a
rotating vane type which shows promise of reliable operation, and is similar to
a8 type used successfully in the ARE. The application of venturi type flow meters

depends only on successful application of pressure sensors.

d. Pump speed indicators
Pump speed measuring devices or tachometers which are durable and
reliable are presently available. Access for 1nstallati$n on rotating machinery
has proven the only difficulty encountered in the ART project. With restricted
access, the more elaborate, but nevertheless reliable, pulse type tachometer

must sometimes be used in preference to the simpler D.C. type tachometer.

e. Nuclear sensors

 

Nuclear sensors in molten salt reactors pose no problems not shared

with other reactors. Existing and well tested fission, lonization, and boron

 

 
- 4 -

trifluoride chambers are available for installation at all points essential to
the reactor. Their disadvantages of limited 1life can be countered only by
duplication or replacement, and provision can be made for this; however, for
circulating fuel reactors,; these instruments are not essential to the routine
operation of the reactor. Nuclear sensors to withstand high temperatures are

not needed.

f. Pressure sensors and pressure controlling devices

 

There exists a variety of pressure sensors of the diaphragm or mano-
meter type which will almost certainly operate wherever needed in a reactor
system. Among these, devices are known that will operate reliably over long
periods of time in the absence of radiation; there are no tests which have
demonstrated that they will operate reliably over long times when subjected to.
radiation. There is no reason to believe that all of these pressure sensors
will fail to operate satisfactorily in a radiation field. %

Gas pressure controllers are required in molten salt reactor systems.
Except for extended use in radicactive gas systems, adequate, well tested pres-
sure controllers are available commercially for all applications. Under strong
irradiation, valve seats or seals made of organic materials are suspect. The
requirement tfiat the valve make a perfect seal is not essential in a design
which incorporates pressure relief valves; metallic seats will probably give

valves which will serve the purpose. The construction of such a valve seems

straightforward, but a long-life test under radiation conditions will be necessary.

g. leak detectors
Leaks between a fluid system and its external surroundings, or between
two fluid systems, tend to grow in size. To minimize serious damage which might

result in a reactor, simple leak detectors capable of detecting very small leaks

 
- 45 -

are desirable so that action may be initiated in time to 1limit the size of the
leak. ILittle progress has thus far been made in the development of leak detectors
which will reliably meet these exacting requirements. The proper approach in
design is to avoid relying on sensitive leak detectors, to design for maximum
reliability against leaks and for minimum or no damage should leaks occur. Means

of detecting larger leaks are of course available.

C. Component Systems

 

l. BSalt and Iiquid Metal Charging and Storage Systems

Methods of handling molten salts and liquid metals in atomic energy
installations have been developed from methods used by manufacturers of sodium,
gasoline, and other chemically active or dangerous liquids. It has become
almost universal practice to make batch transfers by either inert gas pressuri-
zation or by gravity.

Both sodium and salts are at present purchased less pure than required.
Commercial sodium purity is such that, as & rule, only oxide removal is needed.
This is accomplished by filtering the sodium at temperatures as little above its
melting point as is convenient, and by "cold trapping" the oxide in operating
systems (precipitating the Naeo on a cold wall at a selected point in the system).

Salt purification is also accomplished in the molten state. After the
desired proportions of salts have been mixed and melted under a protective atmos-
phere, successive purifying and purging gases (helium, HF, hydrogen) are bubbled
through the melt EZ/. The melt is then transferred into storage containers under

inert gas and stored either molten or frozen until needed.

 

47/  ORNL-57-5k-6-126
- 46 -

The liquid salt or sodium may be charged into the preheated and pre-
cleaned systems either directly or, as is nearly always done, via fill-and-drsin
tanks provided in each liquid system. The mechanics of such transfers is routine "
at ORNL, KAPL, MSA, and a number of other installations.

Treatments, transfers, and storage should, of course, be carried out
in container materials compatible with the liquid used.

A closed circuit of especially purified water, extremely low in
chlorine and oxygen content, should be used in the water-steam system Eg/’ &2/.
One or more storage tanks should be provided for make-up water.

Considerations in locating storage or fill-and-drain vessels are:

(1) radiocactivity, and required shielding and off-gas handling facilities; (2)
accountability for 11255 and other accountable materials; (3) convenience in
storing and handling, including ease of access, mobility, length of transfer
lines, cost of heating, etc.; (4) hazards other than radiocactivity, such as

biological poison; high temperature, etc.

2. O0Off-Gas Handling

 

A major advantage of circulating fuel reactors is the ability to keep
the concentration of xenon and krypton at a low level by continuously removing

them from the fuel system zg/o Their removal greatly reduces: nuclear poisoning

 

48/ Williams, W. Lee and Eckel, John F., "Stress-Corrosion of Austenitic
Stainless Steels in High Temperature Waters", Journal of The American
Society of Naval Engineers, Inc. (February 1956)

 

 

49/ Wilson, R. M. and Burchfield, W. F., "Nickel and High Nickel Alloys for
Pressure Vessels", No. 24, Welding Research Council Bulletin Series
(Janvary 1956)

€

ORNL-CF-57-1-8, "ANP Chemical Section Progress Report", p. 8; ORNL~192k;
ORNIL-2116; ORNL-2095
 

- 47 -

of the reactor by these gases (and by thelir descendents which would have appeared
bad the gases not been removed); the difficulties and hazards of re-starting after
shutdown; and the biologlcal hazard of any leak or other accident which might open
the system and allow leakage of fission gases to the surrounding atmosphere.

Removal of gases can be accomplished by by-passing part of the fuel
flow through a compartment having a liquid-gas interface, accompanied by aglta-
tlon of the liquid in the compartment, by bubbling helium through the liquid,
or both. There must be at least one such compartment in any liquid fuel circuit
to allow for thermsl expansion of the liquid.

The gases so removed will in general be too radiocactive to discharge
directly into the air, and must be "stored" until their activity has decayed
to a level acceptable for discharge into the air. Decay holdup time may be
provided by directing the gases through a large volume of appropriate geometry
(e.g., a very long pipe), or by directing them through appropriately designed
charcoal packing, in which the xenon and krypton will be held up by absorption
on the charcoal. Experiments and designs El/ to date indicate that initial
holdup in a gas volume to allow decay of short half-life nuclides, followed
by holdup in charcoal, is the preferable method. These studies, together with
engineering anslyses 22/, have spelled out the required parameter relations
and design criteria.

It may prove to be feasible to recirculate the helium instead of dis-
charging it to the atmosphere, if the gas is found in operation to be clean and
pure enough after the holdup operation. This is, however, only a future possi-

bility, as yet unproved as to feasibility.

 

51/ HRE, HRT and ART

52/  ORNL-2116; ORNL-192k
3. Inert Gas System

 

The molten salts and liquid metals under consideration require com-
plete protection from contact with oxygen and water vapor. All liquid-gas
interfaces must therefore be protected by inert gas blankets. Systéfis of
high pressure gas stdrage containers; pressure-reducing valves, check valves,
flow control valves, instruments, and tubings and fittings are required to
direct the gas to the required locations at the desired pressures and flow
rates. A very considerable amount of experience in deéigfiing and operating
experiments with precisely the same requirements has been accumulated at
ORNL 22/ and BMI.

Extremely high purity EE/ helium, argon and nitrogen are the most
frequently used gases for these purposes. At ANP, helium is the gas nearly
always used, primarily because it is the cheapest inert gas of the required

purity.

4, Heating and Cooling of Components

 

Since the melting points of the salts and sodium are above ambient
temperatures, it will be necessary to provide heating for all parts of the
system which will contain either sodium or salt. It will also be necessary
to provide heating for the water-steam system. Components subject to after-
heat (beta and gamma decay heat) will in general require provision for heat
removal.

Heating and cooling design should, of course, be such as to minimize

rapld transients or large gradients in temperature where high thermal stress

 

53/  ORNL-2095

54/ <10 ppm oxygen and < minus T0° dew point
- 49 <

would result. This restriction is common to all conventional high temperature
component design. In the few locations where beta and gamms heating may be
large enough to have significant effect on temperature gradients, it should be
taken into account.

Several methods of preheating have been used successfully, including:
(1) use of electrical strip heaters, heat tubes, ceramic-protected hot wire type
heaters, etc., in which the heaters are clamped or otherwise fastened to the
parts to be heated; (2) direct resistance heat, in which the components to be
heated are made a part of an electrical circuit; (3) gas heating; (4) induction
heating 22/; and (5) steam heating (used in England for sodium system heating).
In most high temperature salt and sodium systems at ORNL, adequate design and
fabrication for preheating using Methods (1) and (2) are now semi-routine. No
significant difficulties in preheating or maintaining temperature have been
encountered in systems in which the entire system was carefully engineered (e.g.,
the scores of large and small circulating salt and liquid metal systems success-
fully operated at ORNL in the past three or four years). The most common
difficulty has been failure to take account of local heat sinks, such as support
connections, flanges partly exposed to ambient atmosphere, etc. Good engineering
of thermal insulation is clearly required.

Cooling radioactive molten salt is not routine because a volume heat
source is involved. The use of small diameter vessels such as tubes eases the
problem because it provides a short distance for heat to go from the center of
the liquid to the cooled surface, and because it increases the surface area of

4

the vessel. When appreciable after-heat is being generated, the vessels should

 

55/ Used to a considerable extent in the SRE (NAA - Atomics International)

¢

 

 

 
- 50 =

be cooled with forced circulation, and for reliability, auxiliary power sources .
should be available for this purpose. It is desirable to design the system so

that maximum use can be made of naturel air convection, and thus minimize the

time required for use of the forced circulation and cooling system. Simple

means of biological protection are also desirable. Both of the latter aims

could be implemented by use of an underground tunnel with a stack at the end

of the tunnel.

D. Nuclear Considerations

 

1. Previous Work and Early Consideration

 

Burners - Molten salt 0235 burner resctors for mobile power have been
extensively investigated at Oask Ridge on the ANP program, and these studies pro-
vide the foundation for the present investigation. In 1953 a'group of students
under the leadership of T. Jarvis §§/ at ORSORT investigated the applicability .
of molten salts to package reactors. More recently (1956), another ORSORT group ¢
led by R. W. Davies has prepared a valuable study of the feasibility of molten
salt U255 burners for central station power production 21/,

Fast Breeders - Fast reactors based on the U258-Pu cycle were studied
by J. N. Addoms et al of MIT §§/’ and, more recently, by yet another ORSORT group

led by J. Bulmer 22/. Both groups concluded that it would be preferable to use

 

Jarvis, T., et al, "Fused Salt Package', ORNL-CF-53-10-26, Secret

Davies, R. W., et al, "A 600 MW Fused Salt Homogeneous Reactor Power
Plant", ORNL-CF-56-8-208, Secret

Addoms, J. N., "Engineering Analysis of Non-Aqueous Fluid Fuel Reactors"”,
MIT-5002 (1953)

€ I8

Bulmer, J., et al, "Fused Salt Fueled Breeder Reactor", ORNL-CF-56-8-20k,
Secret .
- 51 -

molten chlorides rather than the fluorides on account of the relatively high
moderating power of the fluorine nucleus, although it was recognized that the
chlorides are probably inferior in respect to corrosion and radiation stability.
Furthermore, Bulmer pointed out that it would be necessary to use purified Cl57
on account of the (n,p) reaction exhibited by c1”°,

In view of these disadvantages of the chloride systems, and, further,
in view of the fact that the technology of handling and utilizing Np and Pu
bearing salts is largely unknown, it was decided to postpone consideration of
fast chloride salt reactors.

Epithermal Breeders - In 1953 an ORSORT group led by D. B. Wehmeyer §9/
analyzed many of the problems presently under study. Many of the proposals set
forth in that report have been adopted in the present program. A study by
J. K. Davidson and W. L. Robb of KAPL él/ has been most helpful, also. Both
this and the Wehmeyer study concerned the possibility of using thorium in a
U253 conversion-breeding cycle at thermal or near thermal energles.

A consideration of molten fluoride reactors based on the Th--U255 cycle
points up the fact that U255 is not available in sufficient quantity to provide
the initial charge for a breeder reactor. While conceivably the U233 required
could be made in a production reactor, the uncertainties in cost and time of
availability militate against designing a reactor to be started up with U253
in the near future. It would appear that, whatever system was selected, the

initial charge would be composed of 93 percent enriched U255.

 

60/ Wehmeyer, D. B., et al, "A Study of a Fused Salt Breeder Reactor for
Power Production", ORNL-CF-53-10-25, Secret

61/ Davidson, J. K. and Robb, W. L., "A Molten Salt Thorium Converter for
Power Production", KAPL-M-JKD-10 (1956), Confidential

 

 
- 52 -

It is well known that the variation of 73 for U235 with energy impairs ¢
the éonversion ratio of a reactor utilizing U235 and Th, and operating in‘the
epithermal neutron energy range. On the other hand, the low moderating power
of the fluoride salts (including IiF and BeF2) makes it impossible to design
a high performance, homogeneous converter in which, say, 90 percent of the
fissions in U235 are caused by thermal neutrons. Parasitic capture in the
fuel carrier, etc., would be excessive. It was concluded that heterégeneous
cores would be required in thermal converters or breeders to obtain breeding
ratios approaching 1.0.

None of the container materials under consideration for use with
molten fluorides would be satisfactory for the canning of moderators for a
thermal breeder reactor. The parasitic absorptions would be intolerably high.
At present; graphite is the only suitable moderator that éhows promise of being
compatible with the salt, and considerable development work will be required to
establish its usefulness. Therefore, consideration of heterogeneous,; thermal,
molten salt reactors has been postponed.

It was decided to investigate the nuclear properties of homogeneous,

233

epithermal, one and two region, molten salt, U converter-breeders. The in-
vestigation to date has been exploratory in nature and most of the work has been
centered on two region systems. The calculations were handicapped by a lack of
data on nuclear cross sections in the epithe¥mal range and lack of a computational
method that would take into account inelastic scattering, resonance saturation,
and Doppler broadening. The first calculations were performed on one region

reactors by hand,; using cross sections then available in the Univac program

Eyewash ég/, which was originally written for the analysis of aqueous homogeneous -

 

62/ Alexander, J. H. and Given, N. D., "A Machine Multigroup Calculation--
The Eyewash Program for Univac", ORNL-1925 (1955) ¢

 
- 53 -

reactors. The first calculations using the computer were for simple, optimistic
cases; i.e., presence of U238, fission products, protactinium, etc., were neg-
lected, as well as the presence of 116 in the carrier. Iater, the cross sections
were revised on the basis of latest information, including the effects of
resonance saturation and Doppler broadening. The lethargy intervals were modi-
fied, and other changes were made to increase the amount of information to be
obtained from the computer (Ocusol-A program) éé/. These facts should be kept
in mind when comparisons of Univac results for the various cases are made.

Tfie advantages of a U235 burner versus breeder-converters were con-
sidered briefly. Devies et al éfl/ proposed to operate the burner without
reprocessing of the irradiated fuel. Builld-up of fission products and other

35

parasitic materials was to be overcome by the addition of U2 in excess of
that rzquired to replace the U235 consumed in the reaction. Depending on the
assumptions made regarding the cross sections of poisons in the intermediate
range, it was found that the reactor would operate from 5 to 20 years before
it would be economically advantageous to dump the spent fuel and recharge.

A study of the effects of adding fertile material to the Davies'
system disclosed that the performance would be improved by the addition of
even moderate amounts of thorium to the core. The amount of U235 required to
replace U235 destroyed and to override nuclear poisons is reduced because the

0233 formed not only replaces U235, but also has a higher fission cross section

and lower absorption cross section in the intermediate range. It was concluded

 

63/ Alexander, L. G., Roberts, J. T. and Carrison, D. A., "The Univac Program
Ocusol-A, A Modification of Eyewash", ORNL-CF- (in preparation)

64/ Davies, R. W., et al, "A 600 MW Fused Salt Homogeneous Reactor Powver Plant"
ORNL-CF-56-8-208 (1956), Secret
 

- 54 -

that conversion-breeding is desirable. It remained to be determined whether the .
ultimate power cost can be further reduced by reprocessing the spent fuel.
Some effort was expended in determining whether it is possible to
obtain & breeding ratio of 1.0 in a homogeneous reactor employing pure U233.
Although the results are not definitive, it appears that it may barely be
possible. It seems fairly certain that a breeding ratio of 0.9 can be obtained,
and it is felt that it would be economically feasible to compensate for the

breeding deficiency by purchasing U253

at a premium price.
Since, however, U233 1s not presently available in amounts suffi-
cient to fuel a full-scale power reactor, attention has been concentrated
mainly on U233--U255 breeder-converters, in which the deficiency in production
of fissile material is compensated by the addition of U235.
It is clear that the optimum conditions will depend to a large extent
on the cross sections of U235, U233, and Th in the epithermal range. The U255 *
cross sections are reasonably well known. In the hand calculations, and in
the first series of Eyewash calculations, an n of 2.28 for U253 was used ’
uniformly from thermal to 0.2 mev. This was consistent with the then latest
published data of Spivak et al éz/’ éé/. More recent data by Magleby et al éZ/,

and Moore et al é@/ for energies ranging up to 8 ev disclose a complicated

 

65/ Spivak, P. E., et al, "Measurement of 7 for U-233, U~235 and Pu-239
with Epithermal Neutrons", Sov. J. of Atomic Energy, 1, 13 (1956)

 

Spivak, P. E., et al, "Measurement of 7 for U-233, U-235 and Pu-239
with Neutrons in the Energy Range from 30 to 900 kev", Sov. J, of

Atomic Energy, 1, 18 (1956)

Magleby et al, "Energy Dependence of n for U-233 in the Region 0.1
to 8.0 ev", PTR-142 (1956), Unclassified

&

S

Moore, M. S., et al, "Uranium-233 Resonance Parameters for Neutron
Energies Below 4.0 ev", PTR-141 (1956), Unclassified

&
- 55 -
resonance structure in the range above 2 ev, and it appears that the avérage n
in this range cannot exceed 2.1. Recent KAPI data §2/ indicate that even at
30 ev, n does not exceed 2.21. On the basis of conservative estimates for the
ranges where data are lacking, new values of the absorption cross section of

U253 were computed.

Table VII

VAIUES OF n FOR U>> USED IN THE UNIVAC CALCULATIONS

 

Neutron Energy, ev Eyewash Ocusol-A
0.076-0.125 2.28 2.28
0.125-0,.186 2.28 2.08
0.186-0.92 2.28 2.28
0.921-10.2 2.28 2.08
10.2-33.7 2,28 | 2.13
33.7-67,000 . 2.28 2.21
67,000-183,000 2.28 2,25
183,000-820,000 2,49 | 2.4k
820,000-10" 2.46-2.52 2.52

—. S Sem SEm e G eemm ame

In Figure 1 are graphed the fission cross sections and n's used in the
Ocusol-A program for U‘Q53 and U235. It 1s seen that, with two exceptions, the
average fission cross section of U253 is substantially greater than that of U235
in all lethargy intervals. Also n is substantially greater at nearly all
lethargies. It seems clear that reactors fueled largely with U253 will give
the best neutron economy when the majority of the fissions are caused by neutrons
in the epithermal range, where parasitic absorptions in fuel carrier and other

materials are relatively less important.

 

69/ Knolls Atomic Power laboratory, "Report of the Physics Section for June,
July end August 1956", KAPL-1611 (1956), Confidential

 
56
UNCLASSIFIED
ORNL~LR-~DWG 20325

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

233
2.5 = —H
.} BEs=————"" =TT !
& 2.0 — L n
U235/|__ r
s ] |
[
]
200 [;
u233\
£ 100 nl_ 1 235
< i
~
i L
O - -
0 {0 20
LETHARGY

FIGURE 1 Fission Cross Sections and Eta for U233 and U235

in the Ocusol- A Program.

 
- 57 -

The average energy of neutrons causing fission can be increased by

35

increasing the U2 concentration, which increases the probability for a neu-
tron to cause fission before it gets slowed down very much. The resulting
increase in reactivity can be compensated for by adding thorium to the core

or by using smaller cores. The improvement in breeding ratio must be bvais.ced

against increased inventory charges.

2, One Region Reactors

 

The critical mass of a bare reactor fiasses through a minimum with
increasing diameter at the point where the effect of diminished leakage on
critical concentration is just compensated by the volume increase. The minimum
inventory for the entire system occurs at a somewhat larger diameter which
depends on the ratio of total fuel volume to core volume. By means of preliminary
hand calculations, a reactor having a diameter of 10 feet was selected as provid-
ing a reasonable compromise between the demands for low inventory and reascnable
size. The power density in the four cases studied was set nominally at 100 w/ce

4

of core volume. The results are summarized in Table VIII, and discussed in

 

detail by Carrison and Alexander 19/.
Pure U233 was used and neutron losses to fission fragments, protactinium,

uranium isotopes, and Lié were ignored. Further, the optimistic U233 cross sec-

tions of the Eyewash program were used. Thus the results are only qualitative.

However, it does not seem probable that a breeding ratio of 1.0 can be obtained

in a practical, bare, one region system. On the basis of the relative cross

gsections and from the comparison of certain two region reactors discussed below,

—

IQ/ Carrison, D. A. and Alexander, L. G., "Nuclear Characteristics of Fused
I1-Be Fluoride Reactors", ORNL-CF-57-1-141 (1957), Secret
- 58 -

Table VIII
U255 ONE REGION, MOLTEN SALT, BREEDER REACTORS o
Fllel Carrier Ose LB O BB SLERBTES Id.F"BeFE
Fertile Material ...cccceee ThFh
Nominal Power Density ..... 100 w/cc
Temperature ..ccecssececnes 1200°F
Computational Program ..... Eyewash
Case S1%* S2 S3 Sk
Core
Diameter, ft 9.34 10 10 10
Nominal power, Mw 1150 1400 1400 1400
IiF, mole % 51.0 50.8 63.0 63.0
BeF,, mole % 47.8 5.0 26.0 15.0
ThF) , mole % 1.0 4.0 10.0 20.0
Fuel U-233 U-233 U-233 U-233
Results .
Critical concentration, mole % UF,, 0.0323 0.243 1.02 2.16
Critical mass, kg, U-233 K 53.7 388 2010 37%0 .
Specific power 21,400 3,600 700 375
Transmutation ratio ** 0.585 0.848 0.986 1.05
Mean energy of neutrons causing fission, ev ~0.1 ~n2 AJMOB mlol‘L

¥ 1l-inch reactor vessel of Ni-Mo alloy

*¥* Combined breeding and conversion

it is estimated that if U255 were substituted for U233 in these reactors, the
critical masses would be increased by a factor of 2 or 3 and the conversion
ratios would decrease markedly for s given thorium concentration.
The use of reflectors should improve the performance of the one region -

reactors somewhat, particularly in smaller sizes. This matter hag riot been explored

 
- 59 -
fully because it was felt that a reflector containing fertile material (a
blanket) would in evéry case give better neutron economy for a given fuel
inventory than a non-fertile reflector.

As expected the mean energy of the neutrons causing fission (bottom
line, Table VIII) increased sharply with thorium loading, and a reasonably
fast spectrum (th ev) was obtained in Case Sk. This hardvspectrum was obtained,
however, at the cost of a very large critical mass of U233. The critical masses
of U235 or Pu would be even larger. However, these results do offer some hope

that a truly fast, homogeneous, molten fluoride reactor based on the plutonium

cycle can be achieved.

5. Two Region Reactors

 

Emphasis has been placed on the study of two region rather than o:o
region reactors for severai reasons. First, higher specific powers, and thus
lower inventory costs at a given breeding ratio, can be obtained with smaller
reactors. Second, in single region reactors there is no convenient way of
separating the U255 or protactinium from the U235 énd U258. Third, at power
densities of 250 w/cc or below, all the power that can be utilized by one
turbogenerator can be obtained from 4 to 6-foot cores. To save neutrons, one
places a blanket of fertile material around the core. |

Preliminary calculations were performed by means of the Eyewash pro-
gram using the nuclear cross sections then available on the Eyewash "sigma”
tape. It was felt that the calculations would at least disclose the area where
the most favorable combination of specifications would likely be found.

Spherical reactors having core-diameters of 4, 5 and 6 feet were
studied. The basic fuel carrier was a mixture of IiF and BeF2 with zero (four

cases) and 4 mole percent ThF) . The blanket (28 inches thick) consisted of the

 
- 60 -

compound Ii_ThF._, and the core vessel was 1/3-inch of INOR-8. The core and

57T
lanket densities were computed for a temperature of lQOOOF, but the nearest
thermal neutron temperature available in the Eyewash program was 12800F.
Cases 21, 22 and 23 of Table IX comprise a series in which the effect
of varying the core radius on conversion ratio and critical mass was investigated

235

for reactors employing U as fuel and having thorium in the blanket only. The
conversion ratios (line 17) are of the order of 0.6 and are relatively insensi-
tive to changes in core size. The critical mass (line 14) increases sharply
with decreasing core diameter in the range of 6 to 4 feet, as does the mean
energy of the neutrons causing fission (line 35). This hardening of the spec-
trum results in a decrease in parasitic absorptions in Ii, Be and F (lines 28
and 29) from about 0.15 to 0.08 neutrons, and a decrease in the absorptions in
the core vessel (line 31) from 0.08 to 0.03 neutrons. However, this decrease

235

is offset in great part by an increase in the parasitic captures in U from
0.10 to 0.16, i.e., the mean value of 1 (line 34) decreased from2.0to 1.80
(n at 0.08 ev = 2.08).

2355

In the Eyewash program, mn of U is taken uniformly as 2.28 (v = 2.52)

in the energy range from 0.025 ev to 0.18 mev. The result of using this 7 is

233

shown in Case 25, which is similar to Case 23, except that the fuel is U and
the Ni-Mo core vessel is 1 inch instead of 1/3-inch thick. The breeding ratio
(line 16) is only 0.53%0; however, it is observed that the absorptions in the

core vessel amounted to 0.206 neutrons. For a comparison with Case 22, the
excess absorption (0.206 - 0.056 = 0.150) was prorated among the blanket com-
ponents, and a breeding ratio of 0.78 was estimated for a 1/3-inch core vessel
thickness on the basis of the increased absorptions in the thorium. The critical

233

mass for Case 25 was gratifying low, amounting to only ~ 13 kg of U .
- 61 -

The principal avoidable losses of neutrons in Case 23 were those due
to absorptions in fuel carrier and core vessel, the sum amounting to about 0.15
neutrons. It was conjectured that these losses could be decreased by adding
thorium to the core. Case 26 shows the result of adding ~/4 mole percent ThFh
to the core of the reactor of Case.22. The thermal flux was entirely suppressed
(1ine 36). There was a substantial increase in the absorptions in thorium, over
half of these taking place in the core, and a corresponding increase in the con-
version ratio (line 16). Absorptions in carrier salt were substantially reduced
and the absorptions in the core vessel were sharply reduced from v 0.06 to~s0.01
neutrons. This last effect was due in part to the increased transparency of the
core vessel at higher neutron energies, and in part to the fact that fewer neu-
trons leaked from the core in Case 26 (0.210 neutrons vs. 0.385 in Case 22).

It was noticed that n (line 34) for Case 26 was slightly greater than for Case 23.

even though the mean energy of the neutrons causing fissions was higher in Case 26.

This circumstance results from the fact that n for U235 passes through a minimum
with increasing neutron energy.

There are a number of small discrepancies between Cases 22 and 26 which
resulted from the premature termination of the calculation in Case 26 and a print-
out for a subcritical condition. This shows up principally in the neutrons
absorbed in fission reactions (line 24), where the absorptions in Case 26 corre-
spond to a v of 2.56. The breeding ratios and the critical masses for all cases
were corrected by linear extrapolations for these departures from criticality,
but the neutron balances were not corrected.

Case 27 shows the édvantage of using thorium in the core with U233 as
the fuel, provided the capture cross sections of U253 in the epithermal range
are as low as those used in the Eyewash program. Parasitic absorptions are re-

duced to a very low value and the breeding ratio (line 16) is quite favorable;

 

 

 

 
- 62 -
Table IX

EYEWASH STUDIES OF TWO REGION MOLTEN SALT REACTORS

Fertile Carrier .c.oeececeess LiF-BeFo Core Vessel ..... 1/3-inch Ni-Mo alloy
Fertile Material cccce. ss00ce THF) Reactor Shell ... l-inch Ni-Mo alloy
Nominal Power Density ....... 200 w/cc Temperature ..... 1280°F

 

 

Iine Case 21 22 23 25% 26 27
1 Core
2 Diameter, ft 6 5 Y 5 5 5
3 Nominal power, Mw 638 268 188 268 268 368
4 IiF, mole % 65.4 65.3 64.6 65 .4 50.6 52,8
5 BeFp, mole % 34,5 34,3 34,7 34,5 b, 3 Lh.7
6 ThFY, mole % - - -- -- 3.9 3.9
7 Fuel U-235 U-235 U-235 U-233% U-235 U-233%
8 Blanket
9 Thickness, inches 28 28 28 28 28 28
10 IiF, mole % 4.5 4.5 4.5 4.5 4.5 4.5
11 ThF, mole % 25.5 25.5 25.5 25.5 25.5 25.5
12 Results
13 Crit. Conc., mole % 0.063% 0,160 0.728 0,056 1.30 0,397
1% Crit. Mass, kg 7.7 4.2 95.6 12.7 303 95.4
15 Sp. Power, kw/kg 11,000 4,300 950 14,000 580 1,850
16 Transmutation Ratio¥* 0.550 0.610 0.618 0.530% 0.734 1.13
17 Core Loading, tons
18 UFY, 0.05 0.07 0.1% 0.02 0.4k 0,139
19 ThF), -- -- - - 1.36 1.36
20 BeFo 3.50 2.0k 1.01 2.05 2,38 2.40
21 IiF 3,60 2.01 1.0k 2.03 1.50 1.52
22 Total 7.15 h,12 2.19 4.10 5.68 5.42
23 Neutron Balance
2L U-fissions 0.397 0.398 0.397 0.393 0.390 0.392
25 U-captures 0.101 0.123 0.160 0.039 0.141 0.037
26 Th in core - -- -- —- 0.211 0,276
27 Th in blanket 0.276 0.321 0.347 0.230 0.193 0.218
28 F 0.078 o 102) 0.075 0.069 0.026 0.030
29 11 and Be 0.068 *T) 0 0.063 0.027 0,029
30 Total Absorptions 0.920 0.943 0.972 0.794 0.988 0.982
31 Core Vessel 0.080 0.056 0.029 0.206% 0.012 0.017
32 Leakage from blanket 0,000 0.000 0.000 0.000 0.000 0.000
33 Balance 1.000 0.999 1.000 1.000 1.000 0.999
3k n 2,00 1.91 1.80 2.29 1.84 2.30
35 E}, mean energy of ~ 0,09 ~n0.6 ~ 50 ~0.12 ~200 ~s150
neutrons causing
fission
36 Thermal Flux 39x100°  1x10%7 0.6x10%7  28x10%7 o ~0

* l-inch Ni-Mo core vessel. For l/B-inch shell, T.R. was estimated to be ~ 0.78.

**% Combined breeding and conversion

*

 

 
- 63 -

however, no account was taken of fission product poisoning or captures in Pa,
uranium isotopes, 116, etc. Thus refinements in the calculations can be
expected only to reduce the estimate of the breeding ratio.

Compared to Case 25, Case 27 exhibits a rather large increase in
critical mass. The optimum concentration of thorium in the core can only be
determined by an analysis of the sum of fuel costs and inventory charges as a
function of thorium concentration.

Detailed neutron balances for Cases 22, 26 and 27 are shown i:
Tables X, XI, and XII. The thermal and epithermal absorptions in the various
components of the core and the blanket and absorptions in the core vesscl are
given.

From an examination of the flux plots, it was estimated that a blanket
only 12 inches thick wofild have captured substantially all of the neutrons leak-
ing from the core vessel.

The results listed in Table IX are, of course, quite optimistic and
are of the nature of an upper limit on performance. rUse of corrected cross
sections and inclusion of the various poisons is expected to reduce the trans-
mutation ratios sharply and to increase the critical masses stbstantially.
However, 1t is felt that thesg systems show promise of being able to produce
povwer at an attractively low cost, mainly through reduction of fuel costs, and
it was decided to refine the calculations and to extend the investigation.
Accordingly, new and additional cross sections were computed and the program

itself was modified in several respects ll/ and renamed Ocusol-A.

 

71/ Alexander, L. G., Roberts, J. T. and Carrison, D. A., "The Univac Program
Ocusol-~A, A Modification of Eyewash", ORNL-CF- (in preparation)

 
- 64 -

 

 

 

Table X .
NEUTRON- BALANCE
Case No. 22 .
Epithermal Thermal
Core Blanket Core Blanket Total
25-Fissions 0.2617 —— 0.1365 ——- 0.3982
25-Captures 0.0953 - 0.0273 - 0.1226
Thorium -—-- 0.3019 --- 0.0187 0.3206
Fluorine ) ) 0.0041  0.0003
Be 0.0848)  0.0084) 0.0005 -—- 0.1019)
Td ) ) 0.00%3  ).0002 )
Total Absorptions 0.4418 0.3103 0.1720 0,0192 0.9433
Absorptions in 0.0435 0.0128 0.0562
core vessel
Ieakage from Blanket 0.0003% 0.0000 0.0003%
Balance 0.9998
Table XI.
NEUTRON BALANCE "
Case No. 26
Epithermal Thermal
Core Blanket Core Blanket Total
25-Fissions 0.3897 ——— 0.0002 - 0.3899
25-Captures 0.1408 ——— 0.0000 _—— 0.,1408
Thorium 0,.2105 0.1923 0.,0000 0.0015 0.4043
Fluorine 0.0195 0.0064 00,0000 0.0000 0.0259
Ii, Be 0,0268 0,0000 0,0000  0,0000 0.0268
Total Absorptions 0.7873 0.1987 0.0002 00,0015 0.9877
Absorptions in 0.0122 0.0000 0.0122
core vessel
Leakage from blanket 0.0007 0.0004 0.0011
Balance 1.0010

;
- 65 -

Table XII

NEUTRON BALANCE

 

 

 

 

Case No. 27

Epithermal . Thermal
Core Blanket Core Blanket Total
23.Fissions 0.387 ——- 0.006 - 0.392
23.Captures 0.0%6 - 0.0006 —_— . 0,037
Thorium 0.276 0.217 0.0008 0.001 0.495
Fluorine 0.022  0.008 0.000  0.000 0.030
11, Be 0.029 0.000 0,000 0,000 0.029
Total Absorptions 0.751 0.225 . 0.007 0.001 0,984
Absorptions in 0.017 0.000 0,017

core vessel

| Leakage from Blanket : 0.000

Balance 1.001

Five reactor cases have heen computed by the Ocusol-A program to date.

The results have not heen completely analyzed; however, the critical loadings,

breeding ratios, and simplified neutron balances are given in Table XIII. |
These reactors have cores 6 feet in diameter. For a total power of

600 Mw, the required average power density in the core is 187 w/cca Thé tempera.-

ThF_ as before; and the core

377
vessgel is l/5~inch of Ni-Mo alloy. The core fluid consists of 69 mole percent

ture is llSOoF. The blanket is composed of Ii

IiF and 31 mole percent BeFe,

indicated. The lithium contains 0.0l percent 116. The thorium cross sections

together with additions of ThFh and UFh as

were modified appropriately to account for saturation of the resonances and

Doppler broadening. The U255 was assumed to be contaminated with T mole per-
cent U258.

 

 
- 66 -

Case 38 has a thorium concentration in the core of 1.0 mole percent
and is fueled with U255. Token amounts (10 ppm) of °> and fission fragments
were added in order to determine the mean effective cross sections of these
materials in the neutron energy spectrum pertinent to each case. The critical
mass was found to be 273 kg of U235, and for an assumed ratio of total fuel
volume to core volume of 4, the inventory would be 1100 kg. The conversion
ratio is 0.610. The effects of build-up of fission fragments, Uejh, U236,
and U238 on this reactor have not been investigated. This reactor is reason-
ably attractive, but the inventory is large.

A reactor having a critical mass of 125 kg of U235 (500 kg inventory)
was then selected for study. The clean reactor (Case 44) was found to be critical
at a concentration of thorium in the core of 0.284 mole percent. The conversion
ratio was 0.614. This improvement, in the face of decreased thorium loading, was
surprising, and was ascribed to the fact that the second reactor was appreciably
more thermal than the first and, consequently, profited from an improvement in
the mean n for U235° By including token amounts of fission products and U233,
data were obtained from which the rate of build-up of 0233 3 U23h, U256 » Tission
fragments, and protactinium could be estimated. The concentrqtions of various
poisons corresponding to about 2UO days of operation at 600 Mw (30 percent fission
burn-up of original charge of 500 kg U235) were computed, and the thorium con-
centration required for criticality was determined (Case 418)., It was found
necessary to remove nearly all the thorium from the core to maintain criticality;
the transmutation ratio decreased to 0.567. Note that in this period nearly 6 kg
of U253 accumulated in the core, while about 50 kg was produced in the blanket
(some present as Pa). The U253 in the blanket system was not considered to con-

tribute to the reactivity of the core.

 

 
- 67 -

Since it would have been necessary to remove nearly all of the thorium
from the core during the first 240 days of operation, it was clear that the build-
up of U235 had been overestimated. On the other hand, there was still some U233
available from the decay of Pa present in the core. In order to avoid repeating
the calculation, it was assumed that the decay of the Pa would approximately com-
pensate in the next 240 days of operation for the overestimate of the U255 made
for the first 240 days. All thorium was removed from the cdre and the 6 kg of
U235 was allowed to burn out. The concentrations of fission fragments, Ugjh,
etc., corresponding to another 240 days of operation, were computed, and the
critical mass of U255 was found to be 154 kg (Case 50). The breeding ratio
decreased to 0.480, and about 100 kg of Pa and U235 had accumulated in the
blanket system.

At some as yet undetermined point, it would become profitable to begin
processing the blanket for the recovery of U253 and thé core for the removal of
fission fragments. The processing cycle chosen consists of'recovéring the uraniumfi
isotopes by the fluoride volatility process and, in the case of the core fluid,
throwing away the contaminated salt. The steady state concentrations of the
various materials corresponding to a processing rate of 600 ftj/yr (compared to
a total fuel volume of about U450 ft5) were estimated. The system was then tested
for eriticality. On the basis of the results, the estimates were corrected; the
results are listed under Case 49 in Table XIII. The transmutation ratio for this
case was found to be 0.56. The U--” feed rate is approximately 0.5 g/Mw-day of
heat produced.

The concentration of fission fragments in the fuel will, of course,
vary with the rate of processing the core. The transmutation ratio will vary

correspondingly, and the eritical loading will change. The conversion ratio can

 

 
_ 68 -
Table XIII

OCUSOL-A STUDIES OF MOLTEN SALT REACTORS

 

Case 38 by 48 50 k9
Fission burn-up, % of 0 0 30 60 0
initial inventory
Thorium in core, mole % 1.0 0.284 0.010 0.000 0,000
Critical mass
kg of U-235 273 125 125 154 99
kg of U-233 0 0 5.9 1.0 37
Neutron balance
U-235 0.573% 0.549 0.493 0.540 0.281
U-233 -- - 0.052 0,008 0.23) *
U-234 -- -- neg. neg. 0.030
U-236 -- -- 0.018 0.030 0.082
U-238 0.018 0.022 0.026 0.031 0.017
Fission products and -- -- 0.055 0.097 0.039
neptunium
Salt, core vessel, 0.075 0.110 0.099 0.09k 0.087
and leakage
Thorium in core 0.1L45 0,090 0.0%2 - --
Thorium in blanket 0,189 0.229 0.225 0.200 0.231 -

Transmutation ratio
(including absorptions in
U-238 and U-235)

Core 0.28 0.19 0.16 0.11 0.10
Blanket 0.3% 0.ho 0.41 0.37 0.h6
Total 0.61 0.61 0.57 0.48 0.56
Accumulation of U-233 in - 0 ~ 50 ~ 100 ~100
blanket, kg

* Includes 0.011 absorptions in U~233 in blanket

be increased by adding thorium to the core fluid. This will,of course, increase
the inventory of fissionable material. The best combination of thorium loading
and processing rates can be determined only by an economic analysis. These matters

are discussed in Section II-G.

 

 
- 69 -

These reactors appear to be reasonably attractive. It is planned
: to perform a parametric study of the effect of variation of core diameter and
thorium loading on the transmutation ratio. Initial, transient, and steady
state performance will be investigated. These studies will form the basis for
a detailed, parametric study of the economics of power production in reactors

of this type.

L. Reactivity Effects in Typical Reactor

 

Reactor Case U4 of Table XIII was selected for a study of reactivity
effects of changes in temperature, fuel concentration, and thorium concentration.
In the Ocusol-A program, it was possible to change the thermal neutron tempera-
ture and to change the atom densities of the materials in the cbre, but it was
not possible -~ short of modifying the program - to make changes that would
. reflect the effect of the shift in the thermodynamic temperature on the Doppler
broadening of the resonances in the various cross sections. The contribution
of the change in thermel neutron temperature was found to be -0.3 x 10'5/°F,
and the change in salt density, (1/k)(0 k/Jp)(dp/dT), was found to contribute
-6.x 1077 /°F.
The reactivity effects of adding U‘?35 and Th were evaluated independ-

ently: (N25/k)(a k/9 N>°) = 0.4585 and (1°2/x) (0 /3 B°2) = -0.116. It follows

 

that the ratio of the U‘?55 addition required for criticality to the thorium

 

addition, d N°°/3 N°2, is 0.26, or one must add one atom of U->? for every 4
atoms of Th added.
The reactivities determined for the U235 and Th were used to break
the term (1/k)(2k/Jdp)(dp/dT) into its component parts, thus:
(1/%)(2 %/ 3 0)(dp/aT) = (1/K)( 2%/ N)(aN"/aT) + (1/k)(3k/IN°)
: (ar°2/aT) + (1/x)(2 k/d °)(an®/aT)
- 70 -

where NS represents the atom density of the salt carrier. The left-hand
member is -6.3 X 10-5, as mentioned above. The first two terms on the right

are, -5.4 x 1077 and +1.k4 x 10_5, respectively. Thus, by difference,
(1/%) (3 k/ 2 ¥°)(an®/aT) = -2.3 x 10~ /°F

This reactivity effect associated with the carrier is due to an increase in
fast leakage as the density of the salt decreases. On the other hand, it was
found, by averaging the macroscopic transport cross sections over the flux
gpectrum, that changes in U255 concentration had little effect on the leakage,
and that the reactivity effect of changes in 1\125 was due almost entirely to

. changes in the macroscopic fission cross section.

These preliminary results seem reasonable, and indicate that control-

wise the reactors under consideration are well-behaved and can undoubtedly be

controlled entirely by changes in heat load and adjustment of the U255 concentration.

E. Reacltor Operation, Control and Safety

The master-slave relationship which exists between power demand and
power production in a circulating fuel reactor has been demonstrated conclusively
in both the HRE and the ARE. Once the reactor is adjusted to produce power at
the design point; load adjustment to meet fluctuations in power demand is achieved
automatically without actuation of control rods or control equipment of any kind.
Rupture in the reactor cooling system or malfunctioning of any of its components
leading to increased temperature in the reactor core, autometically reduces the
nuclear reactivity to a subcritical value. The master-slave feature makes possi-
ble, in & practical reactor design, complete elimination of control rods and all
attendant equipment and instrumentation for imposing direct and prompt control

of nuclear reactivity. Reliability, safety, and great simplification of design
 

 

- 71 -

is achieved. Instrumentation and control equipment is assigned a secondary
role: fo function in start-up; to monitor and adjust fuel make-up to main-
tain design point power; and to monitor and indicate performance of auxiliary
equipment for operational or experimental purposes. This section focuses
principally on the stability characteristics of molten salt reactors, and the

controls and indicators necessary for start-up and adjustment of power level.

1. The Control Problem of Nuclear Power Reactors Zg/

For nuclear pover reactors, control and safety problems are directly
related to the control of temperatures and the possible rapid variations there-
of. Temperature excursions can be induced by a number of events. Basically,
they occur whenever the power generated is not in balance with the power removed.
This relationship is expressed by the following equatipn giving the time rate
of change of the temperature:

ar = E(P - Pc) (1)
at

 

Ig/ The kinetics of circulating fuel reactors have been studied and reported
in & number of papers. Among them are: "Current Status of the Theory
of Reactor Dynamics”, W. K. Ergen, ORNL-53-7-137 (1953); “Kinetics of the
Circulating Fuel Nuclear Reactor”, W. K. Ergen, Phys. Rev. 25, 702 (June
1954); "Stability of Solutions of the Reactor Equations"”, John A. Nohel,
ORNL report; "Some Aspects of Non-linear Reactor Dynamics", W. K. Ergen
and A. M. Weinberg, Physics XX, 413 (195k4); "A Theorem on Rearrangements
and Its Application to Certain Delay Differential Equations", F. H. Brownell
and W. K. Ergen, Journal of Rational Mech. and Analysis, 3, 565 (195k)

 
- 72 -

Im is the mean core temperature, E is the reciprocal of the heat capacity,
P 1is the power generated, and Pc is the power removed.

During operation of a reactor, the balance of power can be upset either
by an excursion in the power generated or by a change in power removed, or
both. Variation in generation may come about by misoperation or failure of
a component either in the reactor itself or in the control equipment, if there
is any. Change in power demand can occur from two distinct causes: by changes
in the electric load or by a breakdown of the heat transfer system between the
reactor core and the load. Removal of the electric load as well as pump failure
or rupture in the fluid heat transfer system results in stopped, or reduced,
heat flow. Should any of these breakdowns occur, prompt and reliable control
schemes must be provided in the reactor design unless, as in the molten salt
reactor, there is inherent fundamental protection against these dangers. Pro-
vision must be made for the removal of after-heat when there is no load. Strictly
speaking, this is not, however, a reactor control problem.

To see the full consequences of an unbalance between the power generated
and the power removed, one must consider the relationship between the rate of
change of powef generated, the temperature coefficient of reactivity @, and the

mean core temperature Tho It is,

d (InP) =-a T, (2)
dt T

where T 1is the mean neutron generation time of both prompt and delayed
neutrons. For a given concentration of fuel and poison, the reactor is critical
at only one value of Tm, which is taken as the zero of fhe temperature scale.
Molten salt circulating fuel reactors have large and promptly acting
negative temperature coefficients of reactivity, and according to Equations (1)

and (2), this feature provides great stability for the mean core fuel temperature.
- 73 -
For example, when the reactor is at full power, a stoppage of the power removal,
accomplished either by loss of electrical load or by failure of the heat trans-
fer system, will have little effect despite the sharp rate of rise of temperature
predicted by Equation (1), for even a slight rise in temperature will shut off
the power generation, as indicated in Equation (2). A partial loss of power
load does not cause more than a temporary perturbation in the mean core temperature.

The only possible difficulty of automatic control of temperature per-
turbation occurs at low power levels. For extremely low power densities such
as one finds during start-up of the reactor, these temperature perturbations
can be large. This comes about because any increase in load appears in the
reactor as a decrease in the fuel temperature. Since the power level is so
low, this decrease in fuel temperature increases the reactivity without at
first raising the temperature appreciably. Consequently, the reactor may go
on a shcrter and shorter period until the power rises sufficiently to heat up
the fuel and cancel out the excess reactivity which was introduced by cooling
the fuél initially. The reactor power level at which the fuel temperature rise
cancels the excess reactivity will then be considerably higher than the load.
The mean temperature could then continue to rise and overshoot the steady state
by a wide margin.

In a molten salt circulating fuel reactor, one can expect temperature
stability without significant overshoot even at low power densities. A typical
value for o is 5 x 1077 pk/k/°F. Based on electronic simulator methods which

accurately predicted the ARE performance, E. R. Mann Ié/ has determined for the

 

73/ Mann, E. R., Private Communication, ORNL
- T4 -

heat transfer circuit, sketched in Figure 2, that a power density of 4 watts
per cubic centimeter is an adequately low limit, above which the reactor will
be stable against significant temperature overshoot. This means, for example,
that the reference design reactor with a temperature coefficient of reactivity
of -5 x lO_S/OF and starting with a power density as low as 4 watts per cubic
centimeter can take up load increases at a rate equivalent to placing the
reactor on a 10-second period without any appreciable perturbation of the mean
fuel temperature. For a design point of 187 w/cc, a reactor of this design can
therefore be brought from a power output of about 13 megawatts to 600 megawatts
in 43 seconds without serious perturbation in the mean fuel temperature by
merely increasing the load demand on the reactor. Thus the reactor will re-
spand completely, safely, and automatically to changes in load demand above

15 megawatts without significant changes in mean core temperature and without
control equipment of any kind. Of course, this automatic reactivity control
occurs as a result of density changes in the core, so that an expansion chamber
must be provided, and the passageways to it must be large enough to allow for
expansion without pressure shock in the core.

At power levels of less than 4 w/cc, the only denger that can arise
is that of suddenly injecting into the core fuel which is too cold. In normal
operation, this would not occur since the start-up times for turbogenerators
far exceed limits below which power increases in the core are not safe. One
unusual event that could cause trouble without proper safety precaution is that
of stopping momentarily the fuel pumps in all of the circulating fuel circuits,
cooling the fuel to a lower than normal temperature, and then restarting all
bumps simultaneously. During stoppage, the reactor core would become isothermal

above its critical temperature and the reactivity would drop to a low value.
UNCLASSIFIED
ORNL-LR-DWG 20468

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

REACTOR CORE FUEL SYSTEM FLINAK SYSTEM SODIUM SYSTEM BOILER
o sec
3 sec
>
O T o~
w9 Mmoo "o noo Mmoo » L‘E'JZ L ©
S s 3 o =9 S5 9eZ23d o
o N O o n © 0 © o~ - F U) @) It
Y I o . © ™ o0 waela S
< o - N D202 <
© o -~ S QL =
a © o0 O
’_
3 sec
FIRST HEAT EXCHANGER {0 sec

TIME CONSTANT = 0.53 sec

SECOND HEAT EXCHANGER
TIME CONSTANT = 0.67 sec

TEMPERATURE COEFFICIENT = 5 x40™ %/°F

Fig. 2. Reference Design Reactor Heat Transfer Circuit Showing Simulator Constants.

 
- 76 -

The sudden insertion of cold fuel could then lead to the difficulty previously
described. Accidents of this kind can be prevented by suitable interlocks. A
more thorough investigation of such unlikely contingencies will be made, but
no difficulties that can not be handled by proper design are anticipated.

An increased load on the reactor system is met by an increased AT
across the core and heat exchangers, with the mean temperature of the core
remaining the same. Simulator studies demonstrate that even with sudden
failure of coupling, such as might conceivably occur in one or more of ti:e
multiple heat transfer systems, the transients of inlet and outlet tempera-
tures occur smoothly and in intervals of time which preclude thermsal or

pressure shocks.

2., Fueling

As fuel is depleted and fission product poisoning builds up, the
mean core temperature at which the reactor is critical sinks lower and lower.
At constant power load, this depression of temperature is reflected at all
roints in the heat exchange system and, in particular, the sodium return tem-
perature from the boiler is depressed. This temperature must be kept abave a
minimum set by the requirement that the sodium return temperature should be
maintained above the melting point of the salts in the salt-to-sodium heat
exchangers. This is done by increasing the concentration of fuel in the core
as needed.

The relation which gives the mass, AM, of fissionable material to be
added to the reactor for a given increase in the steady state mean core tempera-

ture, A&m, is given by the expression,

M=af M Amh
where, B = 6Mc -A_k
M /k
- 77 -
Here & is the temperature coefficient of reactivity, k is the effective multi-
plication constant, and ME is the critical mess. For epithermal reactors,
the ratio B ranges from two to six, usually greater than four, and it can be
obtained from criticality experiments or by cowuputation. The graph, Figure 3,
shows the reduction in sodium return temperature during the course of time
resulting from burn-up when the constant power reneration is 200 w/cc. For
example, 1f the fuel inventory is 1000 pounds o U?jfi and B is four,'and if
the sodium return temperature can be allowed to drop SOOF, then the reactor
must be refveled at intervals no greater than every nine and one-half days. Wheu
it is so fu=led, the amount of U255 that is added is 10 pounds.

The mean core temperature will rise with addition of fuel. The chance
for conceivably dangerous temperature transients due %o too rapid feed can be
climinated completely by adopting a design which limi*ts the rate of feed. For
examnle, the rate of feed can be limited by designing r:fueling equipment to
inject fuel salt in properly sized pellet form only at safe minimum intervals
of time. The pellets would be held under the top surface of the fuel in a
heavy mesh wire screen until dissolved. The slugging >7fect, caused by tempo-
rary inhomogenelty of the concentration of fuel in the e¢irculating salt, is

reduced by feeding fuel in solid form due to the time required for complete

melting of pellets.

3. Criticality Start-up

 

Before adding fuel to the reactor core system, the core, blanket and
all of the fluid heat transfer systems will have been heated to temperature,
checked for gas leaks, filled with their respective fluids, all liquid systems
checked for full operation, and the systems again checked for leaks. With no

heat generation in the reactor and the pumps running, the whole system‘will be

>t

 
Na OUTLET TEMPERATURE FROM BOILER (°F)

920

880

840

80O

760

720 ¢t

680

UNCLASSIFIED
ORNL —LR-DWG 20467

 

 

    

 

 

 

 

 

 

 

INVENTORY |

 

 

 
 

 

 

 

 

 

 

 

 

{000 Ib

 

 

 

   

TEMPERATURE COEFFICIENT

 

 

 

 

 

   

 

o 85X 10—5/ of 7: e

_ 1 |
| | |
\ .- POWER DENSITY = 200 wotts /em3 — -

 

 

 

 

 

 

 

 

 

 

 

 

 

25 30 35 40
DAYS

Fig. 3. Change in Coolant Inlet Temperature for Intermediate Heat Exchanger due to Fuel

Burn —-up, for a Typical Fused Salt Circulating Fuel Reactor at a Power Density of 200 wofls/cmz’.

8L
 

 

- 79 -
isothermal and the temperature can then be controlled in a sensitive way by
steam supplied to, or removed from, the boller. With the reactor in this
condition, it can be brought to eriticality by the safe procedure outlined
below. Nuclear instrumentation will be required for this initial fueiing
operation.

With no fuel in the reactor, the counting rate from neutrons origina-
ting i1n the source is accurately determined and its reciprocal value is used
later in the familiar reciprocal counting rate versus fuel mass plot. With
fuel pumps operating, concentrated fuel salt is added until the fuel load
reachegs 80 to 90 percent of the value which makes the reactor criticsl, as
previously determined from hot criticallity experiments. Then a neutron count
rate is measured, after which carefully limited amounts of concentrated fuel
salt are added step-wise with intervening counting and plotting of points.

When the reactor is near critical, as determine? from the plot, the reactor

core temperature is slowly lowered until the neutron counter indicates eriti-
cality, at which point the temperature of criticality is determined. The zcoler
is turned off, the temperature of the core rises to design point, and the adding-
fuel, counting, determining-criticality-temperature cycle is repeated until
criticality at core design point temperature is reached. Experience from the
ARE has demonstrated that a well planned and deliberately executed procedure

such as this 1s a simple, safe, and reliable one for achieving criticality in
the reactor.

Reactor constants, important for future fueling and other operations,

can be determined from data recorded in the reactor start-up log.

 
- 80 -

F. Build-up of Nuclear Poisons and Chemical Processing

 

l. Fission Product Poisoning

 

A 600 Mw reactor will produce about 190 kg/yr of fission nroducts.
Abou* 23 atom percent of the fission fragments have_a decay chain such that
tey appear as an inert gas - Xe or Kr - with a half-life greater than one
hour and thus are subject to removal by purging with He or N2 gas. These re-
movable isotopes contribute about 25 percent of the total fission product
poisoning in the 100 ev resonance region. This percentage is higher in a
thermal) reactor because of the very large thermal neutron cross section of

.
ie‘fis, bt burn-out limits the Xe155

poisoning to a maximum of about 5 percent.
I the resonance region, however, adjoining nuclei do not have great differ-
ences i.. cross section, and burn-out is relatively ineffective in limiting
poison. Thus, to a first approximation, poisoning increases almost linearly
with time if fission products are not removed.

About 26 atom percent of the fission products are rare earths. In
a 100 ev resonance reactor, they contribute about 40 percent of the fission pro-
duct poisoning. The remaining non-rare-ges and non-rare-earth fission products
include a wide variety of elements, fio one of which is outstanding from the
poisoning point of view.

In a thermal U255 burner reactor operated at constent power and at
constant U255 inventory, but with no fission product removal, the fission pro-
duet poisoring, szth Zifth’ 135
poisocaing (0-5%) plus equilibrium Sm

is approximately equal to the equilibrium Xe

149 poisoning ( ~v1.2%) plus the contribu-

tion from "all other fission products”. According to Blomeke and Todd IE/,

 

T4/  ORNL-2127

 

 
- 81 -

the "all other fission product” poisoning totals about 3 percent when tb~ total
amount of U255 burned is equal to the U255 inventory (100% burn-up), increases

to about 19 percent at 1000 percent burn-up, and to about 51 percent at 10,000
percent burn-up. Thus it is possible, although not necessarily economical, to

run a thermal, fiuid fuel reactor for many years without being forced to vrocess
to remove fission product poisons., One would, of course, pay for not processing
by higher inventory charges for U255 and by lower breeding-conversion ratios. A
600 Mw thermal reactor burns about 230 kg U255 per year, so that with 460 k- 102
inventory, the fission product poisoning would increase from O-5 percent initially
to 20-25 percent after 20 years.

Even in thermal reactors, resonance captures in fission products meke
the poisoning somewhat worse than the numbers given above. The magnitude of
the extra poisoning depends on the ratio of the neutron flux at resonanée energle
to the thermal neutron flux, which is determined in part by the effectiveness of
the moderator. In resonance reactors, the fission product poisofiing is consider -
ably worse due to the higher average fission product absorption cross sections
relative to U235. In a 100 ev reactor with a 530 kg U255 inventory, the total
fission product poisoning 12/ would increase approximately linearly from zero
percent initially to ~ 48 percent after 2 years at 600 Mw.

For U255 fueled reactors, the fission product polsoning is about the
same as for U255 at thermal neutron energles, but in the resonance region, the
higher U255 cross section reduces the poisoning effect of the fission products
by a factor of two over U235. Thus a resonance breeder-converter burning half-
and-half U253 and U235 would have s fission product poison level of ~s9 percent

if processed twice per 100 percent burn-up.

 

75/ Greebler and Hurwitz, KAPL-1L40

 

 

 
- 80 -

233 237 239

2. Pa and Np Poisoning .

233

, Np

Neutron capture in Pa or Np239 has the same result as a non-fission

capture in U233 or Pu259, i.e., a fissionable atom is effectively lost as well
257

as a neutron. Although neutron loss to Np does not involve loss of a fission-
able atom, the total poison can be worse when U255 is used as make-up fuel.
Although neutron capture by any of the three yields a fertile atom, at present
relative prices for fertile and fissionable materials, the gain is negligible
compared to the loss.

235

The average ratio of neutron captures to B-decays by Pa in a reactor

is given to a good approximation by,

— Pa

0.046 P (1 + Q) [B'R'é[ca
Th - Th

M 9y

 

where, P = reactor power level, thermal Mw
(1 + @) = ratio of absorptions to 7tssions in fuel :
B.R. = breeding or conversion ratio '
M?h = kg of thorium in system .

The P and « refer to the whole system. The other parameters can refer

either to the whole system or to the core and blanket separately. 1In molten
233

salt power reactor studies to date, neutron captures in Pa have been negli-

glble, due primarily to the large thorium inventories considered.

In a U233-U235 breeder-converter using highly enriched UE35 make-up,

239

Np poisoning is relatively unimportant, but if the breeding-conversion ratio

2
5T poisoning can become objectionably high if it is allowed to

is poor, Np
reach its equilibrium value with no removal by chemical processing. This is
éspecially true in resonance reactors, in which U236 (and hence Np257 at
equilibrium) ylelds may be twice as high as in thermal reactors. Chemical

processing to remove Np is discussed in a following section. *

 
- 83 -

3. Corrosion Product Poisoning

 

Corrosion product concentrations in molten fluoride salts in INOR-8
loops at 15000F have not been observed to exceed 100-300 ppm for Fe, Cr and
Al, or 100 ppm for Ni, Mo and other alloy constituents (with ~ 20 ppm being
typical for Ni and Mo). At lower temperatures, these numbers are smaller, but
precise values at 1100—1200°F are not yet available. These corrosion products
will be removed from the core system, along with the fission products, by the
chenical processing system. They will build up to equilibrium in the blanke:
system (where they are relatively less objectionable than in the core since
blanket poisons have to compete with such a very high concentration of thorium).

The Fe, Cr, Ni and Al are relatively light elements and thus should
have lower capture cross sections than typical fission products. Molybdenum
is typicai of the lighter group of fission products (about 18 percent of
fissione z-entually yield a stable Mo isotope). Since the chemical process-
ing will probably be at such a rate that the steady state fission product
concentration in core salt will be several thousand ppm, the corrosion product

poisoning hias been neglected.

4. Chemical Processing and Fuel Reconstitution
The "ideal™ reactor chemical processing scheme would remove £ission

Pa255 239 as soon as %“hey were formed. After

products, corrosion products, and Np
=3 239 .
the latter two had decayed to and Pu ”7, it would then return them %o the
reactor, along with the U and Pu passing through the process, in the desired form.
This ideal chemical plant would have low capital and operating costs, would hold
up only small amounts of fissionable and other high-priced materials, and would
discharge its waste streams in forms that could be inexpensively disposed of or
sold for a profit. Present technology, however, does not offer such an ideal

process for any reactor.

 
- 84 -

More practical short-term goals for processing a molten salt reactor
might be (a) continuous removal of most of the gaseous fission products by purg-
ing with §e or N, gas, (b) an "in-line" removal of rare earth, noble metal or
other fission products by "freezing out" of part of the salt stream, "plating
out” of fission products on metallic surface either naturally or electrolyti-
cally, “"salting out", e.g., of rare earths by keeping the salt saturated with
Ce, or "slagging out" of & solid carrier phase by adding oxygen or oxides, and
(¢) continuous or batch removal of salt fuel from the reactor at an economically
optimum rate to separate the U, Pu and salt from the remaining fission products
and corrosion products by the least expensive method available. Present tech-
nology does not offer all of this for a molten salt reactor which has to be
designed now, although there is reason to expect that an adequate development
program would enable it to be approached more closely in the "second reactor”
design. Operaticn of the ARE and of ANP in-pile loops indicates that gaseous
fission product removal can be achieved and that Ru, Rh, and Pd plate out on
metal surfaces. The fluoride volatility process achieves one of the most impor-
tant goals in separating U from salt and fission products. Scouting experiments
in ORNL Chemistry and Chemical TECHnology laboratories have indicated a basis
for optimism that further development will yield methods of separating salt from
fission products. Adequate technology already exists for the preparation of
fresh fuel starting with non-radioactive UF6 and fluoride salts, but further
development is required to demonstrate "hot" (and hence remotely operated)
reconstitution of fuel from recycled uranium (and salt if possible) if long
cooling times are to be avoided.

Fluoride Volatility Process - Processes for the decontamination of
uranium by fluorination to produce volatile UF6 are under active development

at the Argonne and Ozk Ridge National laboratories. Both sites have studied
- 85 -

the dissolution of solid fuel elements (e.g., STR) in molten fluoride salts
as a preparatory step for fluorination, and ORNL has also studied the fluori-
nation of molten salt reactor (ARE and ART) fuels as well. The ANL process
fluorinates with BrF_, and distills the UFE product to complete the decontami-

5

nation. The ORNL process fluorinates with F, and completes the UF6 decontami -

2
nation with a sorption-desorpti«: cvecie on a NaF pellet bed.

The ORM'. volatility rosram is currently running at approximately a
Sl,OO0,000-a-year level and is a% an =ariy pilot plant stage. The pilct plant
is now being broken in with nor-radioactive feed. This cold stage wili be
followed by processing of warm (long-cooled) ARE fuel and then hot STR (10-30
day cocled) fuel, over fiscal 1958 and fiscal 1959 according to present plans.
The chemistry of the process has been described by Cathérs Zé/, and the pilot
plant by Milford ZZ/. Figure 4 is a biock flow sheet of the proc=ss. For pro-
cessing ARE type fuel, the uranium bearing molten salt is transferred to the
fluorinator in 1.k ft5 batches. Fluorine, diluted with N2, is bubbled through
the salt at 600-650°C until its U content is ~ 10 ppm, and UFg, N, and excess
Fé pass out of the fluorinator through a 100°C NaF pellet bed (capacity of
10 kg U) wiiich removes the UF6 from the gas stream. At present, the excess
Fé is scrubbed out of the gas stream with a reducing KOH solution. In vola-
tility laboratory studies, F, recirculation with a K-25 B-4 pump has been
used and may be installed later in the pilo* plant, thus reducing F2 consump-

tion and simplifying the disposal problem. The UF6 is desorbed from the NaF

by raising the temperature to 400°C and sweeping with F_-N

5~Nye The UFB passes

 

76/  ORNL-CF-56-9-21, American Nuclear Society Meeting, December 1956

77/ ORNL-CF-57-%-18, American Chemical Society Meeting, April 1957

 

 
N(JF-ZrF4

Zr-U FUEL ELEMENTS

 

 

 

 

 

 

 

 

UNCLASSIFIED
ORNL-LR-DWG 19090

 

F, DISPOSAL =

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

" Ru |
F> — COLD |
(DESORPTION) L TRAP |
UFg
ANHYDROUS F, XA coLD |-
HF TRAP
X .
Hy + HF NaF NaF
BED BED
Y XB UFg
UFg +F5 PRODUCT
s———
HYDRO- | NoF- 7, -UF, FUSED SALT
FLUORINATION | FLUORINATION
650°C 600-650°C
— e
ABSORBE RS
WASTE NaF-ZrF Y WASTE NaoF

Fig.4. Fused Salt-Fluoride Volatility Uranium Recovery Process

&
- 87 -

through & second NaF bed and is then enllected in cold traps at -40 to -6000.
The plant product is liquid UFB, or;tained by isolating the cold traps from the
rest of the system, heating to above the triple point and draining into the
product receiver.

Most of the decontamination is achieved in the fluorination since mocs®
of the fission and corrosion products remain in the salt. The volatile contani -
nants (%s, Kr, I, Te, and Mo, most of the Ku, »nd part of the Nb and Zr) ei“her
pass through the NeF bed while the UFy is retained (Xe, ur, I, Te, Mo, Ru) or
remain ou the bed when the UFg is desorbed (Mo, Zr). The I, Te, Mo, and Ru
are most.iy removed by a :cid trap, the remainder being scrubbed out of the gas
system aicng with the Fé. The Xe and Kr follow the NQ to the plant off-gas
system. The Nb and Zr slowly build up on the laF bed, which is replaced when
poisoned. laboratory development indicates tha% a lOOOC micrometallic nickel
filter betwszen fluorinator ard NaF bed removes most of the Nb and Zr, and this,
too, may be added to the pilot plant. This addition would greatly extend the
life of the NaF bed.

According to ORNL =xperience, only part »f the Pu follows the U. The
behavior of lip and Pa has not beeu ztudied. A molten salt power reactor develop-
ment program should include studies of these three elements. Volatility processing
of LiF-BeF2 and Na.F-BeFEj salts has not been demonstrated, since the present process
is based on NaF-Zth salts, but no difficulties are expected. 1Iu fact, the Ii salt

complexes UFh-UF -UF6 less strongly than Na salts and hence should require less

5

excess F..
2

Reduction of UFE to UFh - The continuous reduction of highly enriched

 

UFE to UFL with hydrogen is well proved as & non-radioactive process. The process
- 88 -

developed and used at K-25 Ié/ is indicated in Figure 5. The reduction takes s
place in a UF|6~F2~H2 flame in a Y-shaped tower reactor. The F2 is added to
give a hotter flame. The reaction products are UFu powder and HF-H2 gas.
Micrometallic filters recover any UFh which is entrained in the exit gas. A
vibrator is used to shake free any UFL vhich clings to the filter or tower
walls. A chemical trap using a CaSOh or NaF pellet bed recovers any unreacted
UF6 in the exit gases; although the amount so collected is negligibly small
in normal operation. The exit gases are scrubbed free of HF with either a
KOH solution spray or a NaF bed.

This process has Dot been operated at a high level of radioactivity.
The UFB from the power reactor volatility process would be somewhat radiocactive,
with the major activity probably T, 4 "hgt" pilot plant demonstration should
be provided in a fused salt power reactor development program. No unusual diffi-
culties are anticipated, however, since the present continuous process is smooth- :
running and practically automatic.

For molten salt reactors using sodium rather than lithium based fuels,

the UFB to UF& recycle may well be greatly simplified by reducing the UF6 with

H2 on the NaF bed and using the NaF-UFu pellets for fuel make-up.

5. Build-up of Even-Mass-Number Uranium Isotopes

The build-up of U2, 1PoF, 1P, and 1°Pas non-fissionable isotopic
diluents in U233 and 0235 plays an important part in fuel cycle economics.
Although U252 does not build up enough to affect the neutron balance signifi-
cantly, its hard-gamma-emitting daughters grow fast enough to be a biological

hazard in the handling of U233, and thus adversely affect the resale value of

 

78/ Smiley and Brater, TID-7518, Part 2, p. 156-210

 

 
 

 

SURGE DRUM C.V.

 

i-l E - —D4
ORIFICE
a
o
'.—
<{
H, CYLINDERS g:)
>
Fz'——%—"{_—__“‘—*“—-%—’ .@1_
SINTERED METAL
UFs DRAIN FILTERS
CYLINDERS
UFG - UF4
REACTOR

r—(%-——:,l | » 1'% >

 

UFs VAPORIZERS

 

  

 

 

 

 

 

 

 

 

 

 

 

=

 

  
   

 

POWDER REMOVAL
SYSTEM

 

 

 

i

 

 

]

e

r -~ ———

 

 

FLEXIBLE CONNECTION

UFs REDUCTION PROCESS

FLOW SHEET

FIGURE 5

[

 

 

 

CHEMICAL

 

 

 
- 9 -
the U233o It has been assumed that the molten salt power reactor will process

and burn all the U23§ it produces, hence the U252 problem has not been con-

sidered in detail.

233 233

Radiative captures in Pa and U lead to isotopic contamination

of the U233 with U23h. With no processing to separate these isotopes, and none
seems feasible, the U23h builds up until it is being produced and burned at the
same rate. At equilibrium, in thermal reactors, there is ~/ 57 percent as much
U23h as U253, with U25h capturing ~» 9 percent as many neutrons as U235o In

fused salt resonance reactors with higher capture to fission ratios, at equili-

234 233

brium, U and is present in

amount equal to ~35 percent of the U255, Nevtron capture in U23h produces

captures ~ 13 percent as many neutrons as U

fisgionable U255, but a capture in thorium would be preferabl: since U235 is
a better fuel than U255°

Neutron capture in U256 results in an isotopic poison,; since U257,
with a 6.75 day half-1ife, decays to Np257 too fast to permit useful amounts
of the fissionable U237 to build up. In completely thermal reacors fiith no

36 builds up until it is ~ 18 times as abundant as U235,'

isotopic processing, Hg
and captures ~ 16 percent as many neutrons. Normally, in any real thermal
reactor, resonance capture by U236 will reduce the steady state ratio to less
than 18. Isotopic separation «f U235 and U256 may be feasible because of the
large amounts involved an: because it is important in a breeder-converter
economy. On its own merits in a separate cascade, it should cost at least
9 times as much as separation of U235 from U258, but by feeding it into exist-
ing cascades, either by adding top stages or accepting lower production rates,

less expensive processing can probably be achieved. The K-2§ Operations Analysis «

division is studying the gaseous diffusion problem, and ORNL is studying the

 

 
 

- 91 -

over-all problem. At present, AEC buy-back prices for U235 do not penalize

236 any more than they do dilution with U258°

isotopic dilution with U23u and U
Despite this present buy-back policy, this molten salt reactor study has
assumed tha*t, in the 1dng run, equilibrium U256 polsoninz would simply have to
be toleratcd. This is pessimistic if the U235 is kept separate from the U235
in a two region converter, or if a resale market to the military or to other
reactors is available at prices near the present AEC buy-back level.

For a steady state reactor using highly enriched U255 feed (93% 25,
6% 28, 1% 24) and with no iscvtopic processing, the U238 at equilibrium will
capture 6/93 as many neutrons as the U fed. In a "e-r:.pletely thermal”
reactor, at equilibrium there would be 16 times as much U238 as U255° In
fused sait resorance reactors, the U238 would build up to only about 10 percent
~ ¢ the U235 fed. Thus at isotopic equilibrium, U256 is a worse contaminant
than U238e Neutron cAapture in U258 produces fissionable Pu239° For this study,

it has been assumed that equilibrium concentrations of and captures in U238

235 chain is kept separate from the U233

would have to be accepted. If the U
chain, this is a more pessimistic assumption thar is usually made. Even with
mixed chains, it is a more pessimistic assumption than is often made. For
example, present buy-back prices pefmit the Consolidated Edison reactor to use
a mixed U255, thorium fuel and sell the resulting mixture of U«~232-233-234-235-

236-238 for $15 per gram of the amount of U797 and U°2° that it contains.

6. Radioactive Waste Disposal

 

At 600 Mw of heat for TO00 hrs/yr, a reactor produces ~ 190 kg/yr
of fission products. In molten salt reactors, perhaps 24 weight percent of
the fission products can be removed as Xe-Kr gases, leaving 145 kg to be removed

by chemical processing. The proposed chemical processing flow sheet waste streams

 
- 92 -

include fused salt, NaF pellets, F2-N2 and HF-H2 gases. Most of the non-gaseous
fission products remain in the core salt residue and may be stored in this form.
Most of the remaining fission products are removed by periodically flushing out
the micrometallic filter between the fluorinator and the NaF bed and the cold
trap between the NaF bed and the Fé disposal unit. The NaF bed 1s replaced
occasionally when it becomes poisoned with niobium and zirconium. Any remain-
ing fission products in the gas streams are serubbed out with the F2 and HF

or vented to the reactor off-gas system.

For optimum costs, high power molten salt reactors shoul. probably
have relatively large inventories of enriched uranium and be processed sbout
twice per fuel-inventory burn-up, a compromise between the cost of salt replace-
ment and the savings due to improvgd breeding-converzion ratio. The core salt
discard rate indicated in the follewing section is 500-1000 ft3/yr. This volume
corresponds to a farily high fission product concentration, comparing favorably
with any other type of power reactor processing. The disposal problem is not
" a small one, however. The high concentration of fission products makes the heat
dissipation a problem. The high value which the salt would have if the fission
products were removed ( SleO/ft3 for L17Be, perhaps half that much for Na-Be),
makes it desirable to store it in an easlly recoverable form until means of re-
claiming it economically can be developed. Underground storage tanks cooled
by natural circulatioi of air, with the salt kept molten, appear to he an

acceptable answer. These might well be built one at a time, each with a one-

year storage capacity, until a final decision on ultimate disposal is made.
..95..

G. Fuel Cycle Economics

 

1. Cost Bases

Fissionable isotopes have been valued at $17/g and inventory charges
calculated at 4%/yr. Thorium cost in fluoride salts has been taken as $30/kg,
an arithmetic average of published prices of $17/kg as nitrate and g43/kg as

metal. A price of $91/kg for 99.99 percent 117

in large quantities, an authori-
tative estimate from Y-12, has been assumed. A fluoride salt cost of #5/1b,
plus the cost of the special materials uranium, thorium, lithium-7, has been
used 18/ . This leads to $l1l.1-/1b for 69 IiF-31 BeF, and 315.4%0/1b for 75 LiF-
25 ThFh. About one-ialf of the latter is for thorium. The original salt
inventories have been capitalized with a 16 percent annual charge corresponding
to a 20-year depreciation period. It is assumed that new core salt will be
purchased to replace the amounts rrocessed. The blanket salt is used over the
20-year reactor life without excessive fission product poisoning build-up.
Fissionable material consumption cost is based on feeding 93 percent

enriched U235

to the core system to compensate for a breeding conversion ratio
of less than unity. It is assumed that U233 is not available for purchase at
an economic price, although it would be worth more than U235 in a molten salt
resctor due to its better nuclear characteristics. It is also assumed that

53

isotopic re-enrichment of U235 or U2 or resale of even isotopic contaminated
uranium is not economical and, therefore, that the molten salt power reactor
must burn out the non-fissionable isotopes and pay for it in lower breeding-

conversion ratio.

 

79/ Private estimate of W. R. Grimes, ORNL

 
 

- 94 -

2. "Steady State" Neutron Balances and Comparative Fuel Costs

 

The neutron balances given in this section are not the actual ones
given by the Univac; but; rather, modifications or corrections of the original
balances as required to make the reactors both "eritical" and "steady state".

The Univac results were used to obtain flux-averaged microscopic cross sections
for the various elements, and these were assumed to remain constant as con-
centrations of the individual absorbers were changed provided the total macro-
scopic absorption cross section was held constant. The results are believed
to be fairly good for comparison purposes, but no better on an absolute basis
than the basic cross sections themselves.

The fuel costs compared below include only fissionable material rental,
fissionable material purchased for burn-up, and core salt purchased to replace
that processed. These are the "rapidly variable" fuel cycle costs, i.e., those
which vary sharply in "reference design" type.reactors (Section III), depending
on choice of operating conditions. Although the chemical plant capital and
operating costs would also vary to some extent with processing rate, their varia-
tlon is not fast enough to affect the optimization appreciably.

Three "different” reactor cases are compared below, five versions of
a "reference" case and one version each of "low inventory" and "high inventory"
cases. The first two cases are for 6-foot diameter (113 ft5) cores with 337 ft5
external holdup. The last case is for a 5-foot diameter core (65 ftB) with the

same external holdup. The values of v were taken from BNL~325 to be:
U-23%3: 2.54 U«235: 2,46

Flux-averaged a's and u's were chosen separately for each case from the Univac

calculations on which they were based.

o
-95_

Reference Case - The columns headed A, B, C, and D in Table XIV
describe steady state reactors based on Ocusol-A Case 49 for various chemical
processing rates and using 9% percent U255 make-up. The reactor of Column E
is like that of D except that U235 make-up is assumed. The flux-averaged

values of o for these reactors were:
U-233: 0.15 U=-235: O.41

The flux-averaged microscopic absorption cross sections were in the follow-

ing ratios:

U-233: 1.00

U-234-6-8: Q.37
FP: 0.22

Th (blanket): 0.036
Th {core): 0.125

Column A gives the fissionable material inventory, neutron balances and fuel
costs for an "infinite processing rate", i.e., no uranium or fission products
in blanket and no fission products in core. In the other columns the blanket
is assumed to be processed for uranium removal at a once-per-year rate (with
an average-over-thedife-of-the-reactor amount of fission products included in
the blanket) and the core processing rate is varied. Corrosion product and
Np237 polsoning is not included as such, but it is felt that this is compensa-
ted for by taking no credit for separate removal of gaseous products or for the
production of, or fissions by, plutonium. The costs in Column E assume that
U235 could be bought for $17/g. They may be interpreted to indicate that, by

comparison with Column D, the fused salt reactor could pay twice as much for
U233 as for U235.
- % -

Table XIV

NEUTRON BALANCES AND FUEL COSTS FOR DIFFERENT INVENTORIES AND PROCESSING CYCIES

I Column

IT Processing Cycle, yrs
Core
Blanket

IIT Inventory, kg
33, Blanket
23, Core
25, Core

IV Neutron Calance
02,
o2,
23,
23,
2k,
25,

2
28,
FP,
FP,
Other

Totals
Effective B.R.

v Fuel Costs, 3/yr
23 + 25 rental
23 or 25 make-up
Salt make-up

QuoaaaaaaaaQbQw

 

 

 

A B C D E F G
0 0.25 0.50 0.75 0.75 0.54 0.97
0 1.00 1.00 1.00 1.00 1.00 1.00
——- 105 105 105 105 100 100
269 228 202 175 318 ok 400
192 258 313 369 69 288 700
471 591 620 649 492 482 1200
0.2446 0.2312 0.2312 0.2312 0.2312 0.2033 0.1854
0.0926 0.066L4 0.0333 —— 0.1099 - 0.0959
——— 0.0110 0.0110 0.0110 0.0110 0.0211 0.0200
- 0.3372 0.2866 0.253%5 0.2202 0.3993 0.1822 0.261h
0.0433 0.0382 00,0430 0.0297 0.0527 0.0278 0.0373
0.1457 0.1960 0.2382 0.2806 0.0527 0.3126 0.2186
0.0k426 0.05Th 0.0698 0.0821 0.0155 0.0873 0.0641
0.0072 0.0110 0.0143 0.0175 —— 0.0199 0.0127
_— 0.0024 0.0024 0.0024 0.0024 0.004kT 0.0050
_— 0.0129 0.0257 0.03%86 0.03%86 0.0405 0.0271
0.0868 0.0869 0.0866 0.0867 0.0867 0.1006 0.0725
1.0000 1.0000 1.,0000 1.0000 1.0000 1.0000 1.0000
0.82 0.73 0.64 0.56 0.83 0.50 0.68
320,000 402,000 422,000 141,000 535,000 328,000 820,000
815,000 1,260,000 1,630,000 2,030,000 774,000 2,275,000 1,450,000
® 2,430,000 1,220,000 810,000 810,000 1,116,000 670,000
@ 4,092,000 3,272,000 3,281,000 1,919,000 3,719,000 2,940,000
-97_

Low Inventory Case -~ The above-discussed reference case had its median

 

fission in the 34-61 ev energy group, with U233-U235 inventories of ~ 600 kg.
Column F describes a reactor (based on Ocusol-A Case L4i4) with a lower uranium
inventory. The medilan fission was in the 10-15 ev group. The flux-averaged

a values were:
U-233: 0.16 U-235: 0.39
The relative abso~:tinm: cross sections were:

U=-233: 1.00
U=-23%5: 0.58
U-234-236-238: 0.31
Th (bleanket): 0.028
Th (core): 0.097
A3
Fuel make-up with i 77 55 assumed. This case does not appear to offer as
economical a fuel cycle as the reference case. The biggest difference is in
the increased fractios: of neutrons lost to salt, shell and leakage ("other"

in the table).

High Inventory Case - Since it appeared from the above two cases

 

that an even higher inventory might be still mo¥e economical, another case

was examined and is listed in Column G (based on Eyewash Cases 26 and 27, with
Ocusol~A a's inserted). Altfiough this case differs from the preceding two in

core size as well as inventory, it does appear that the higher inventory "pays

off" in increased breeding conversion ratio and reduced salt replacement costs.
- 98 -

SECTION ITI

Reference Design Reactor

 

A, Introduction

To allow a realistic estimate of the performancg, safety, economics
and cost of construction of a typical molten salt reactor, a specific reactor
type and size was chosen for detailed study. The principal items considered
in making this choice were:

1. The reactor should be capable of relatively early construction.

No component, material, or process is used that is not elther already avallable,
or tested or under test in at least pilot plant scale equipment. The assumption
is madg{ however, that equipment such as pumps and heat exchangers can be de-
signed and built successfully in larger sizeg than those tested to date., Devel-
opment and testing charges will be associated with this scale-up,

2. Safety in an early reactor is of paramount importance. Great
emphasis is placed on avoiding any reasonable possibility of chemical accident
i£volving radioactive comporents,

3. Long life of components is esgential for long-term eeconomy. On
this basis the alloy INOR-8 was chosen as a material of construction; however,
1ts reliability has not been demonstrated in long term tests, although short
term tests suggest that it should last for many years.

L, The réactor should possess the ability to serve as a prototype
for central station power generation.

>. The possibility that this may be the first power reactor of its
kind requires that as.much information as possible be derived from it. Thus
the reactor should be as versatile as possible without incorporating experimental

facilities that seriously affect the economics or continuity of power production,

 
 

- 99 -

While it was not expected that this "reference design reactor" would
necessarily turn out to be the best one possible, it was chosen after considera-
tion of a large variety of factors and should be representative of good molten
salt power reactors that can be built in the near future.

The reference design reactor is a two region homogeneous converter
with a core approximately 6 feet in diameter and a blanket 2 feet thick,
Moderation is prov;ded by the salt itself so there is no need for a moderator
or other structure inside the reactor. The core, with its volume of 113 cubic
feet or 3200 liters, is capable of generatin 600 megawatts of heat at a power
density of only 187 watts/cc. This rate of power generation is well within
safe 1limits, and the total power output of 600 thermal megawatts from the
core, plus additional heat from the blanket, allows a net electric power genera-
tion of approximately 240 megawatts,

The basic core salt is a mixture of about 7O mole percent LiTF and
30 percent BeFe. Additions of thorium fluoride can be made if desired, an?
enough UFh will be added to make it critical. The blanket will consist of
the eutectic of LiF and ThFh, or mixtures of it with the basic core salt. The
melting point of the core salt is 867°F and that of the blanket salt is lOBOoF,
or lower, |

Both the core fuel and the blanket salt will be circulated to external
primary heat exchangers, six in parallel for the core and two in parallel for
the blanket. The numbers of parallel circuits were chosen so that useful opera-
tion could still be obtained if any one circuit should fail. It also turns out
that reasonable sizgs of pumps and heat exchangers result from the 100 Mw heat
removal requirement for each of the six core circuits.

Flinek, a mixture of alkali fluorides (mixture 12), was chosen as

the intermediate coolant because it has reasonable chemical compatibility with

 
 

- 100 -

the core and blanket salts and with sodium, and, of the salts, it has good heat
transfer characteristics, It will become only slightly radioactive by exposure
to the delayed neutrons from the core fuel and will serve as an isolating link
to keep all radioactivity from contact with the steam system. The melting
point of the Flinak is 850°F.

Secondary heat exchangers will transfer heat from the Flinak to the
sodium, which is used directly to heat the bollers, superheaters and reheaters,
Details of a reasonaflle heat transfer system have been worked out, so that with
a fuel temperature of lQOOOF, a steam temperature of lOQOOF can be achieved.
Figure 6 gives a block diagram of the gross features of one of the heat trans-
fer systems of the core circuit. The blanket cooling systems are similar.

After a careful consideration of the problem of control of the reactor,
it has been decided that there is nc need for any control rods. Reactor con-
trol is automatically maintained by the negati#e temperature coefficient°
Uranium fluoride fuel or thorium fluoride as a poison will be added when the
need arises, as indicated by temperature measurements in the non-radioactive
sodium system., Thus, this reactor can be properly called a nuclear furnace,
requiring fuel additions to maintain temperature.

The basic hardware of the system lends itself to a number of different
fuels and fuel cycles, For example, the core can be operated with UEBSFh or

UzBBFh, or, perhaps later, with PuF i Different inventories of uranium age

50
possible, depending on the emount of fertile material introduced, and the
amount of fisslion products and heavy elements that are allowed to build up.
One indicated mode of operatlion is to start up with 0235 in the core, together

with a little thorium, and to process the core on such a cycle that a complete

reprocessing would be carried out appro;imately every year, The processing
 

 

 

 

 

 

 

 

 

 

 

UNCLASSIFIED
ORNL-LR-DWG 19890

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1100 F
12.55C
1000F 36.8C
@ 1000F
___B890F 5.2C
FUEL SALT T 311C < -
TO FLINAK FLINAK
HEAT TO SODIUM TO 1ROE61-(I)EFAT
EXCHANGER HEAT X 6 ac —————— - 060 X
EXCHANGER : :
1930F SUPERHEATER— |
12.61C —]
H25F
1.4 C \V/ \
1090F Na TO No
BLENDER —— o] - HEAT EXCHANGER
i 1200F
. 940F
650F
860 F
FROM REHEAT
REACTOR 800 F
36C
770F
9.1C
BLENDER— I
670 F 26.9¢C ec
91¢C
BOILER ‘
LEGEND
F = TEMPERATURE, °F 515 F
C = FLOW, ft¥/sec 1.8C

Fig. 6. Schematic Diagram of Heat Transfer System.

 

101

 
- 102 -

would consist of removing the uranium from the core fluid as UF6 by the volatility
process, storing the core fluid with its contaminating fission products for later
processing, and making up fresh fuel from the reduced_UF6, with new UFh added
to compensate for fuel burn-up and for heavy element build-up. U235 would build
up from whatever thorium is included in the core. The 0235 extracted from the
blanket by the volatility process would either be stored until enough is cbtained
for a pure U233 core loading, or it could be added to the core as soon as it 1is
extracted, In either case, the uranium could be recirculated indefinitely and
economical operation would be achieved, even after equilibrium concentrations
of U25h, U256, and U258 arz built up. |

A factor in the selection of the multiple stage heat transfer system .
involving two intermediate Tlulds was the desire to have only compatible fluids
in adjoining volumes where leaks could involve radioactive materials. This
avoids the possibility of liberation of fission products as the result of a
chemical accident. As the system is now designed, the pressures developed by
the pumps, by difference in density of the fluids, and by over pressure, are
such that any failure producing mixing of fluids would tend to produce flow
toward the reactor core, rather than from it;, tending furthgr to confine the
fission products.

An exothermic chemical reaction would result if the sodium and water
or steam were mixed, However, this would not involve radioactive materials
and would pose only the same danger problems as in any conventional plant han-
dling quantities of chemically active materials whgre no biological poisons
are involved.,

The components of the reference design reactor are relatively simple;

the apparent complexity stems mostly from the number of components rather than

 
 

- 103 -

their type. The large number of components is required because of the large
power output contemplated, The basic simplicity of the components together
with the high thermal efficiency and low fuel turnover cost encourages the
belief that the future cost of power from reactor plants of this type might

be fairly low,

B. Heat Generation, Transfer, and Conversion System

 

l. Reactor

The heat transfer system for the reference design reactor has six
parallel, independent paths for heat flow from the reactor core fuel to the
steam- system, and two similer paths for heat from the blanket fluid. Figure
7 shows section and plan views of the reactor vessel, gilving the general
arrangement of six parallel core pumps and return lines. The reactor vessel,
piping, and pump housings and impellers will be made of INOR-8.

As explained in the irntroduction of this section, the multiple heat
removal circuit system was selected as offering reliability and safety. The
symmetrical positioning of reasonably sized components allows low fuel holdup and
low thermal stresses in the reactor structure and its attendant piping.

The fuel will be enriched to the extent that the average temperature
of the core for criticality will be 1150°F. At the full rated power genera-
tion of 600 megawatts of heat in the core, the system is designed so that the
fuel salt will enter the core at 1100°F and leave it at 1200°F. With this
100°F temperature range, the total flow of fuel through the core must be 3k ,000
gallons per minute. At this rate, it will take the fuel an average of 1.5
- seconds to pass through the core and six seconds to make a camplete circuit
of core plus external system, Each of the six parallel core heat removal cir-
cuits will then handle 5,650 gallons per minyte of fuel, over a temperature

range of lOOOF, and transfer 100 megawatts of heat,

 

 
- 10k -

The multiple circuits, each independent, make unnecessary any valv-
ing or flow regulation in the fuel or primary circuit. Stopping one fuel pump
would result in slow back flow through that pump and its exchanger. With all
fuel pumps stopped, convection would provide sufficient flow to dispose of
after-heat, provided primary coolant flow is maintained.

Each blanket cooling circuit will be able to provide full blanket
cooling. Normally one circuit would be in full scale operation and the other
pumping slowly, on stand-by. The total amount of heat to be removgd from
the blanket will vary with the U253 content of the blanket. Though detalled
calculations of the blanket heat removal system have not been made, it is
estimateq that the total heat generation in thé blanket will not exceed 35
megawatts at any time. Hence, it is expected that heat exchangers in that

system will be similar to, but smaller than, those used in the core system.

2. Heat Exchangers

 

All heat exchangers, including those in the steam system, are of
shell and tube design, with counter flow of the fluids, U-shaped shells,
which minimize tube thermal stress, will be used in every case except in the
reheat circuit, where tube and shell temperature differences are low, Table
XV gives a summary of the characteristics of the principal heat ;xchangers
used in the entire sjstem, and Figure 8 shows diagramatically the temperature
conditions of each heat exchanger.

The primafy heat exchangers will transfer heat from the fuel
(composition T4) to the primary coolant (Flinak). The principal objectives
in the design of the primary heat exchangé?s are first, dependability, and
second, low fuel holdup and low pressure drop. Both of these objectives are

met best by having the fuel flow through the tubes rather than on the shell

 
105

UNCLASSIFIED
ORNL-LR-DWG 19330

 

PUMP OUTLET

 

-
W
-
=

PLAN

 

 

SECTION

Fig. 7. Reference Design Reactor.

 
 

- » ' . ¥ &
- 106 -
TABLE XV
SUMMARY OF HEAT EXCHANGERS

Exchanger Primary Secondary

Fluid Fuel Flinak Flinak - Ne

Location Tubes Shell Tubes Shell

Material Inor Inor

Shape U U U U
Fluid (Hot End, °F 1200 1125 1125 1090
Temperature (Cold End,°F 1100 1000 1000 890
Temperature change, CF 100 125 125 200
Temperaturs (Hot End, °p 75 35
Difference (Cold End,°F 100 110
Heat Transfer Capacity, Mw 100 100
' Heat Transfer Surface, ft 3040 3600
Avg. Heat Flux, Btu/hr-ft< 120,000 9%,000
Tube Data: |

Length, ft 16 31

Number 2k50 930

0.D., inches .380 .600

Wall Thickness, inches .0L0 .050

Pitch (A), inches .580 .825

Tube Bundle Dia., inches 26 27
Flow, ft2/sec 12.55 12,61 12.61 31.1
Flow, gpm 5650 5680 5680 14,000
Flow, 1b/hr - - - i
Fluid Velocity (inlet), ft/sec 10.5 7.4 10 15.3
' Reynolds No. (Nominal) 6,000~T,000 6,000-8,000  12,000-17,000 2.6--3.0.:;;105
Pressure Drop, psi Lo 19 3l 18
Power Consumed, . Kw
Heat Transferred Mw .98 AT Bk 93
Max. Heat Flux, Btu/hr-ft2 120,000 145,000

o v

Thermal Stress, (l?vx ATewa.ll ; psi 4,000 6,000

 

 
 

 

 

 

 

 

 

- w - e 4 a
- 107 -
TABLE XV (Continued)
Na-to-Water Na-to=-Steam " Na-to-Na Na-to-Steam
Exchanger Boiler Superheater for Reheat Reheater
Fluid Water Ne. Steam Na Reheat Na Main Na Steam Na
Location Tubes Shell  Tubes Shell  Tubes Shell Tubes Shell
Material 2 1/4 Croloy %16 SS 316 S8 316 SS
Shape U L U Straight Straight
Fluid (Hot End,°F 650 T70 1000 1090 1060 1090 960 1060
Temperature (Cold End,®F 515 670 650 800 860 9Lo 610 860
Temperature Change, F 135 100 350 290 200 150 350 200
Temperature (Hot End, °F 120 | 90 30 100
Difference '(Cold End, OF 155 150 80 250
Heat Transfer Capacity, Mv 61.7 ol 2 13.5 81.2
Heat Transfer Surface, ft° 2540 2000 450 7550
Avg. Beat Flux, Btylm-fte 82,700 4l ,100 103,000 37,000
Tube Data:
Length, ft L5 46 1k 30
Number 28l 220 300 1190
0.D., inches 1,000 1.000 .500 1,000
Wall Thickness, inches 120 .120 .0k9 .095
Pitch (A), inches 1,800 1,800 745 1.800
Tube Bundle Dia., inches 29 28 1k 65
Flow, ft7/sec 1.8 36 20.8 5.20 4,03 5.4 682 2l ,2
Flow, gpm 810 17,300 -- 2340 1810 2430 - 10, 900
Flow, lb/hr 317,000 -- 317,000 - - -- 1,546,800 --
Fluid Velocity (inlet) 2 8.75 30 1.7 15 9,15 160 | 1,46
ft/sec '
Reynolds No. (nominal) 3.5 x 10° 6.9x10° 4.5 x 10° 1.2-1.7x10° 9 x 10° 1.65x10°  3.25 x 10° L2x10
Pressure Drop, psi 5 '; 3 6 1 20 4,5 16 1
Power Consumed, Kw
Heat Transferred MW .05 3b 1.02 -Ob 1.17 e 35 26.5 .05
Max. Heat Flux, Btyhr-ft° 95,000 54,000 169,000 56, 000
Thermal Stress, (g E _ AT wa'll)p_si"'{OOO 4200 10,000 £500

(T->" 7 2 )

 
)
1200
CUEL SN‘T 75
125
100
“

100 gl

1000

PRIMARY EXCHANGER {FUEL SALT-FLINAK
OT\m =87

SECONDARY EXCHANGER {FLINAK-SCDIUM }

 

890
AT\ m =655

SODIUM TO WATER BOILER
763 _ 770

    

OWUN

   

g0
120

670
€650

 

155

515
A7’,m=}¢105+——— 100 130

ALL TEMPERATURES ARE IN °F

108

UNCLASSIFIED
ORNL -LR-DWG 19889

SODIUM TO STEAM SUPERHEATER

 

 

 

1090 »
1000
-
800
150
650
O Tim= U7
SODIUM TO SODIUM FOR REHEAT
1090
1060
940
»
80
B60O
A T|m = 51
o
SODIUM TO STEAM REHEATER
1060
100
vh 960
200
860
@
Q/V‘
&
250
"
610
ATim= 123
B

Fig. 8. Temperature-Heot Diagrams for Heat Exchangers.

 

 
- 109 -

slde of the heat exchangers. The temperature range of the fuel affects these
£wo objectives., Increasing thé temperature range decreases the external fuel
volume holdup, but at the cost of increasing the pressure drop in the exchgpger.

Moderate fuel holdup in the exchanger 1s obtained with O.3-inch ID
tubing. Although a wall thickness of 0.039 inches has been selected as offer-
ing reasonable life and temperature drop, it is possible that additional data
and analyses will make a wall thickness of 0.060 inches appear desirable.

The secondary heat exchanger, Flinak-to-sodium, has flow design
criteria similar to the primary exchanger, but the volume of salt in the ex-
changer is a minor consideration. Consequently, larger tubes than are used
in the primary can be used, which will reduce the number of tubes in a unit
and simplify fabrication., One-half inch 09 x 0.050-inch wall tubing has been
selected.

The Na-to-water boiler and Na-to-steam superheater and reheater are
designed with single wall tubes but double tube sheets., Easier contalnment of
high pressure steam is achieved by putting it inside the tubes, as is standard
practicé‘in ordinary boilers, ?he low pressure sodium is easily contained in
the shell, As the sodium tube sheet and the steam tube sheet are at different
temperatures, sufficient length of tubing must be provided between them to glve
low bending stresses in the tubes, To keep this parasitic length low, the
tubes should be as small in dilameter as possible, but reliable fabrication 1s
easier with large tubes and thicker walls. One«inch ID x 0.120-inch wall tub-
ing has been selected as a_practical solution for these conflicting demands.

The boiler is of the "once-through" variety and is designed £o
deliver steam superheated by about 50°F to the superheater. This assures that

no moisture will enter the superheater, The superheater is designed to deliver

 

 
- 110 -

steam to the turbine at 1000°F, and an attemperator will be placed in the steam
line ahead of the turbine to insure control of the maximum temperature.

Throttle valves in the feed water supply line will be used to main-
tain balance between the eight separate circuits, while throttle valves in the
sodium lines of each separate circuit will maintain the balance of the heat
supply among the boiler, superheater, and heat exchanger.“

A separate sodium circuit will be used for reheat, with the reheater
located near the turbines. A sodium-to-sodium heat exchanger in each of the
six main sodium circuits will supply heat to the reheat sodium circuit. The
insertion of this extra heat exchanger link, together with the lower heat trans-
fer properties of the lower-pressure reheat :team (350 psi) means thgfi reheat-
ing to 960°F 1s more economically accomplished than reheating to 1000°F, This
does pot-appfeciably reduce the turbine efficiency, but only requires that the
first expansion be carried to a lower pressure than for reheating tp_lOOOoF°
With the reheater located near the turbines, 5‘percent pressure dr;p is predicted
as compared to 10 percent when the steam is returned to a furnacgeég/

The reheater, and probably the Na-to-Na exchangers in the reheat cir-
cuit, can be of the straight tube design without causing undue thermal stress.,
As the reheat boller has, of necessity, a large number of short tubes, the U-
shaped configura@ion would be difficult to fabricate., Thus, it is fortunate
that low stress permits use of the straight tube dgsign; ‘

As shown in Table XV, the primary and secondary heat exchangers will

be made of INOR-8, the boilers of 2 1/4 Croloy, and the superheaters, reheaters,

and sodium-to-sodium heat exchangers of 316 stainless steel, Although these

 

§Q/ Shannon, R, H., and Selby, J. B., "Double Reheat Cycles--The Next Step?"
.Power, February 1953

 

 
- 111 =-

are suitable choices at present, it is indicated in Section II that future ex-
perimental work may lead to the liberal use of high nickel alloys such as

Inconel in the sodium, water and steam systems,

3. Steam Cycle
The steam cycle selected for design consideraticn uses 1800 psia,

1000°F steam, with one reheat to 960°F in 3600-1800 rpm cross-compound tur-
bines, Figure 9 gives a diagram of the steam cycle, showing pressures, tem-
peratures, and mass flow rates of the steam. Seven bleed-offs are used for
heating feed water to 515°F, With a condenser pressure of 1 1/2 inches Hg
(920F), a turbine heat rate of 8070 Btu/kwh is achieved. This corresponds
to a steam cycle efficiency of 42.3 percent.

Coal-fired central stations use six percent of the gross power out-
put in.auxiliaries; seven percent appears to be a reasomable figure for this
reactor power plant. This latter figure is reached by subtracting one per-
cent for draft fans, coal pulverizers, ash handling, etc., connected with a
fossil fuel furnace, and adding two percent for pumping fuel and coolants
associated with the nuclear furnace. Net station efficlency is 39 percent
or 8675 Btu/kwh delivered at the bus-bars. For comparison, an efficient
coal-fired unit, the Commonwéalth Edison, Chicago, State Line Plant, reports
8550 Btu/kwh sent out from the 191 Mw Unit No. 3.

At this preliminary state, no attempt has been made to optimize
the steam cycle on a cost basis, or to select an extremely efficient
cycle. The conditions selected--1800 psi, 1000°F, reheat to 960°F--are
in line with proven industrial practice, and require that the.design of
the reactor system face up to problems of pressure, temperature, and han-

dling of reheat. The steam cycle proposed is based on the cycle used in

the Astoria Station of Consolidated Edison Company, New York, which

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     
   

   
 

  
 

  
  

UNCLASSIFIED

     
  

   

       
     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

340A 960F 1,546,746W ORNL-LR-DWG 19879
,! 1800A 1000F 1,851,060W
‘ REEAT 90A 648F 1,281,674W
INTERMEDIATE PRESSURE
TURBINE
SUPER HEATERS
[ T A o
LOW LOW
NER R TURBINE GENERATOR
[j GENERATO PRESSURE PRESSURE B
Pl
w Y =
o z 3
o oJ H')P
2 0 w - L ) g 1%-inHg 92F
1t o 4 7 7 3 1,132,479 W
O
@
I
o o
m m
a
T 59,789 W 55,632 W -
T
Y
| ] | ] d
BOILERS
S50F
58,308 W z
® =
5 [ S ?
© 0
o 3 \EVAPORATOR o
w o - r~
= g o~ ) w
o - L g = " 0
3 L e L - o = < =
- < @O @< =z P~ oJ a
W N I 5 ~ N ~
Q N~ |8 & o s - ©w =z
0, - O 0
- <1 8 8 | - w|o
L o M= I ek |0
n ® - 0|0
n ¥ 7 o
' ' 502F 304F 1 ! ‘ i
395,386 W
429F 373F 260F 210F 160 F 100 F
NN\ NN\ AN\ - NN AVAVAV: L AN N AN
449F 393F \DEAERATOR 250F 200F 140F a7F
304,314 W 129,019 W 190,104 W 263776 W
LEGEND
A = PRESSURE, psia 312F
F = TEMPERATURE, °F
w = FLOW, ib/hr
- # - | ] < .

¢l
- 11% -

was the result of an extensive study éi/’ Qg/, Modifications were made to
meet the special conditions of the heat supply and cooling water temperature
available.

A feed water temperature about lOOQF below the saturation témpera-
ture of the boiler seems to be as low as is desirable for a sodium boiler,
based on thermal stress considerations. This 1s higher than is normally used
with & fossil fuel boiler, where lower tefiperatures allow more heat extraction
from the flue gases, In compensation for the additional feed-water heaters,

however, the higher feed water temperature increases the steam cycle efficiency

 

slightly.
C. Compgpents‘and Component System
l. Pumps

The molten salt and liquid metal pumps will be of the same general
design as those now in use in ANP, except for differences of size and capacity,
and refinements of details which may develop before the time for procurement
arrives., One pump of each new size (capacity and head) will be given thorough
tests before additional pumps of the same size are procured, and each pump
procured will be proof-tested before use in a power plant. The pump sizes re-
quired are indlcated as to flow and head by the data in Table XV describing
the hea@ exchanger, Pumps, compressors, fans, etc.,, for all purposes, except

for the movement of molten salts and liquid metals, will be conventional,

2. Valves

Salt and sodium mechanical stop valves will be bellows sealed,

 

Ql/ Milne, G. R.,, "Basic Study for a Generating Station", ASME Transactions

82/ Milne, G. R., "Cost Amalysis for 160 Mw Units Favor 1800.ps1--1000°F. Reheat",
Electrical World, February 26, 1951

 
- 114 -

cermet-seat type. No salt or sodium stop valves will be used on other than

fill-and-drain lines, and these will be 2-inch IPS.

 

No flow control valves are required in molten salt lines,

The flow control valves in sodium lines will be bellows sealed., They
will not be required to stop flow, but dnly to control it.

All molten salts and sodium valves will follow the most recent ANP
design for such valves, except for size and possible modifications for diminish-

ing fluid head loss.

3., Pipes and Tubes

 

Pipes and tubes must be designed to absorb all temperature transients

and differences without exceeding specified stress levels. 1In general, they

 

will be anchored at walls and at heavy pleces of equipment, and appropriately
shaped bends between anchor points will be provided to absorb all expansions,
contractions, and twists. The main piping in the fuel salt circuits leading
to the primary heat exchangers will be 12 inches in diameter., The blankep cir-
cuits will use lO-inch . pipe. The coolant salt systems will use lk-inch .
pipe in the core heat.removal system and 10-inch . pipe 1n the blanket heat

removal system. All these pipes will be made of INOR-8.,

L, Fill-and-Drain Tanks

a, Fuel salt fill-and-drain tanks

 

Two tanks will be required, each with sufficient capacity to contain
all the fuel salt in the circulating fuel system (approximetely 500 ft5)° One
tank will be available for temporary storage of used fuel salt while the othepy
tank is being used to serve the continued dperation of the power plant.

Each of the tanks will be designed so that it can be heated to 1200°F,

and so that maximum after heat (approximately seven Mw total) can be removed

,
- 115 -

with the hottest part of the shell cooler than 1500°F and the axial salt tem-
perature considerably less than the fuel boiling point. It is estimated that
this criterion can be met by fabridating the tanks from a number of 12-inch
I1.D. pipe, aggregating to a total length of 600 feet. Shell temperatures under
after-heat conditions will be maintained by recirculating air, in turn cooled
by water-cooled radiators, Provision will be made for powering the cooling
gystem from emergency power (diesel or battery) if the regular power supply
should fail. Heating will be done by electrical resistance,

Tank fittings required will include the following: (1) salt charge
lines; (2) salt drain line; (3) line for transferring molten salt to and from
the circulating-fuel system; (4) inert-gas supply line; and (5) inert-gas

discharge (off-gas) 1line,

b. Blanket salt fill-and-drain tanks

 

Two tanks will be required, each with sufficient capacity to contain

all the blanket salt in the circulating blanket system (approximately 600 ft5).

One tank will be available for storage of used blanket salt while the cther is

being used to serve the_cpntinued operation of the power plant.

Each of the two tanks will be designed with provisions for heating
to l}OOoF. No special provision need be made for removal of decay heat other
than including capacity in the storage room atmosphere cooling equipment to
remove this additional heat from the room.

Tank fittings requipred will be the same as for the fuel salt tanks,

¢, Intermediate hHeat transfer salt fill-and-drain tanks
Four tanks are required: (1) two duplicate tanks, each capable of
containing all the salt in all the Intermediate heat transfer circuits; and

(2) two duplicate tanks, each capable of containing the salt for the largest
- 116 -

intermediate heat transfer circuit. Each of the four tanks is to be connected
for filling and draining each intermediate transfer circult separately. Each
of the four tanks is to be provided for heating to lEOOOF,

Tank fittings required are the same as for the fuel salt fill-and-

drain tanks,

d. Sodium fill-and-drain tanks

 

Four tanks are required: (1) two duplicate tanks, each capable of
containing all the sodium in all the circulating systems; and (2) two dupli-
cate tanks, each capable of containing the sodium for the largest single
sodium circuit. Each of the four tanks is to be connected for filling or
draining each of the sodium circuits separately. Each of the four tanks is
to be provided for heating to 1200°F.

Tank fittings required are the same as for the fuel salt fill-and-

drain tanks.

5. Gas Supply Systems (Helium, Nitrogen, and Compressed Air)

Helium gas, contalning less than 10 ppm.oxygen and having 4 dew
point less thanm minus TOOF, wlll be the only gas allowed to come in contact
with finy salt or sodium, except during chemical processing. Supply lines from
the helium storasge banks will contain pressure reducing valves, shutoff valves,
flow and pressure instrumentation, and safety devices as required for safety
and dependability.

The consumption of helium will depend on & number of design details,
including: (1) dilution of off-gas; (2) use of helium in instrumentation;
and (3) pump design for pumps serving barren salts and liquid metal systems.

Nitrogen may be used as an atmosphere around some assemblies afid*aub—

assemblies of equipment. If so, the handling equipment will be similar to the

 
- 117 -

 

heligm’handling equ%pment, but with less strict purity requirements. Whether
or not nitroggn will be used anywhere in the plant will depend on future deci-
sions as to ambient atmosphere requiréments.

Compressed alr will probably be requifed for some instruments, but

is a comparatively minor item.

6. Off-Gas System

The xenon and krypton evolving from the fuel at the fuel-to-helium
interface in the fuel expansion chambe; will be (1) diluted with helium in
that chamber, (2) removed through a tube to a holdup tank to allow time for
short half-life decay, thence (3) passed through a charcoal bed for further
holdup and decay, then (4) discharged to the atmosphere through a stack,

The bases for design of ;he system will be the safie as for the HRE,
tPe HRT, and the ART 92/. Parameter relations are becoming well established QE/
and continued experimentation is refining the quantitative knowledge still
further, The present design basis (for later refinement) is the ART off-gas

X

system, modified to ten times the ART off-gas production,

7. Preheating and Temperature Maintenance

Provision will be made for preheating the following items to 1200°F
in 48 hours, although a longer tifie will normally be used for preheating: |
reactor, salt expansion chamber-pump units, salt and salt-to-sodium heat ex-
changers, connecting pipes and tubes, salt and sodium filltgnd-d;ain tanks
and fill-and drain lines, salt fill-and-drain tank charging lines, new;éalt

storage vessels (from which molten salts are charged into salt fill-and-drain

 

83/ Reference Quarterly Reports

84/ ORNL-2116
- 118 -

tanks), and off-gas lines to off-gas system. Provision will be made for pre-
heating the following items to 1000°F in 48 hours: superheaters, boilers, re-
heaters, sodium lines, and sodium stérage vessels gé/o

The boilers will have provision for additional heat input for start-
up, both in the boiler and upstream of the boiler, so that excessive thermal
stresses will be avoided during start-up. The general pattern of boller start-
up operation will be very similar to that in a conventional power plant.

It is anticipated that most of the heating of reactor system com-
ronents will be done with heaters consisting of electrical resistance wire
embedded in shaped ceramic matripes.. More detailed engineering of components
may show that it is more convenient, more dependable, or cheaper té heat some
parts of the system with direct electrical resistance, electrical induction §§/,
gas heat, or auxiliary steam, | |

Preheaters will be so arranged as to allow separate temperature con-

trol of individualjsubassembliese

D. Plant Layout

Figfi;gs 10 and 11 show a workable disposition of the major power
plant components. “

Figure 10 is a plan view and shows the reactor surrounded by six
primary (Tuel salt tolFlinak) heat exchangers and two blanket-cooling heat
excgangers. All of these components should be as closely gfouped as remote
maintenance will allow, in order to minimize the volume of fuel salt.

Directly above the reactor and ips primary heat exchangers is a

shield containing removable plugs which allow access from above. (See

 

85/ May be designed for slower preheat rate

86/ Used by Atomics International on the SRE
- 119 -

Figure 11.) This shielded region above the reactor is allocated to tools for
remote disassembly &nd reassembly of fuel salt pumps or primary heat exchangers.
It may also be used as a temporary storage room for radicactive parts,

o Epclosing the reactor primary complex and the shieldedGSPace above
is a 1arge”steel shell to contain fission fragment gases should they escape
from the primary system. It also will allo# for maintaining a relatively
inert atmosphere should repairs to the system become necessary. The lock lead-
ing to the vessel makes it péssible to take parts or toq;s fn or out of the
vessel without appreciably affecting its atmosphere, Personnel may, if nec-
essary, enter the shielded space above the reactor to service or alter. the re-
mote handling tools.

The primary heat exchangers are showp_in“a vertical position and
are connected to the main piping py welded flange Jbints., This arrangement
is believed to provide forrthe least difficulty in thegremote operations of
removing and replacing heat exchangers. The primary pumps will be of the tap
access variety, to make the replacement of the motor-impeller portion less
difficult by remote control. The pump units will probably be held in place
with a bolted flange. It is recognized that much detailed design and develop-
ment work will be required to insure adequate operation of remote maintenance
-equipnment.

Surrounding the steel shell and behind concrete shielding are eight
compartments, each one in turn shielded from the others, These compartments
contain the secondary heat exchangers (Flinak-to-Na) and Flinak pumps. Remov-
able shield plugs above each compartment will allow access to, and replacement
of, the components containeq therein,

The sodium is conducted from the above-mentioned compartments to
four separate layers of boilers, superheaters, pumps, and blenders; The top

layer is to serve the blanket cooling system.

 
UNCLASSIFIED
ORNL-LR-DWG 19337

 
    
   
  

__—SECONDARY
HEAT EXCHANGER

CONDENSER SERVICE WELL

    
   
   
 
 
  
    
  
 

-4

HEAT EXCHANGER
TO REHEAT-- —

 

 

 

GENERATOR U !
) |

 

 

 

 

~— TN
"= | Ow PRESSURE TURBINES

AT —_ {0 O ACCESS REGION

   
   
 
   
 

FRIMARY
HEAT EXCHANGER

 

 

 

‘o T v
CY - ) \

‘ ‘

‘ |

i R
— . !

.
i\

 

mREHEATER

   
 

 

CCONTROL ROCM {

 

 

 

 

 

 

 

 

b

 

 

Fig. 10. Plan View of Power Plant.

0¢l
 

UNCLASSIFIED
ORNL-LR-DWG {9373

   
    
 

 
  
  
 
   

_-Na BOILER
CIRCUIT PUMP

 
  
  
  
 
 

SECONDARY
HEAT EXCHANGER
SPACE

 

LOW PRESSURE TURBINE

 

LOW PRESSURE TURBINE

 

    
  

HIGH PRESSURE
TURBINE

 
 
  
 

GENERATOR

GENERATOR
SHIELD

 
   
 

REMOVABLE
PLUG —

 
   

 
     

NeTONa .~ * ~ o
- HEAT EXCHANGER
FOR REHEAT

 
 

100 10 20 30
SCALE IN FEET

Fig. 11. Section Through Reactor and Power Plant.

 

12t
- 122 -

From the boilers and superheaters on, the system and building arrange-
ment closely resemble a conventional steam-electric power plant. However, in
place éf coal and ash handling equipment; space not shown im the layouts must
be provided for fuel, Flinak, sodium, and blanket salt sforage vessels,'and
chemical processing eQuipment, It is visualized that these facilities would
be located below and at the side of the reactor complex opposite the steam
plant.

Space for the auxiliary features of a conventional power plant, such
as water processing, machine shop, instrument shop, offices, etc., mflst, of

course, be provided.

E, Chemical Processing and Fuel Cycle Economics

 

 

1., Core Processing
The core salt chemical processing system is a combination of the
ORNL fluoride volatility and the K-25 UF6 reduction processes discussed in
Secfion, Part F. The core salt is transferred by gravity or inert gas pressure .
from the core loop to a holdup vessel where short-half-lived fission products
are allowed to decay before the salt is transferred to the fluorination vessel.
The reactor fill-and-drain tank is felt to be the ideal holdup vessel since
it is already designed for cooling hot sal£ and since its volume is large com-
pared to the heldup required by the chemical'plant, If thirty days proves to
be an adequate cooling period, about sixty cublc feet of core salt will be
stored at one time,
The core salt will be fluorinated in 1.l ft” batches (ORNL pilot
plant size) at one=-two batches per day (600 ft3/yr), The barren salt, stripped

of uranium but containing most of the Pu, fission products, and corrosion

products, is transferred to tank storage where it is held for future salt

 
- 12% -

recovery., The 10-kgU-capacity NaF pellet beds (ORNL pilot plant size) will
be’operated on a once-or-twice-per week cycle (i,e,, severalifluorination
cycles per sorption cycle). The volatility part of the chemical plant produces
liquid UF6 in cylinders, The UF6 reduction tower will operate simi-continuocusly,
using UF6 from the volatility plant as feed. It will discharge its UFh product
directly to a fuel salt mixing pot, to which is also fed fresh salt and make-
up UFb{°

Uranium losseg in chemical processing are quite low, apout 10 ppm
in waste salt and 2 ppm in waste gases. This is approximately 1 kg/yr uranium

ioss,

2. Blanket Processing

 

Chemical processing of the blanket salt 1s physically much the same
as that of the core salt, except that after fluorination,the blanket salt with
the Pa and fission products that it contains is returned to the blanket system.
Because of the much lower power density in the blanket, holdup for the fission
product cooling is not a problem,

Separate fluorinators for core and blanket salts will prevemnt pqssible
cross-contamination. A continuous fluérinator for the blanket is assumed since
complete uranium recovery is not necessary, and although continuous fluorina-
tion has not been demonstrated, no new basic development is required., (Con-
tinuous fluorination of the core salt would require further development to
demonstrate complete uranium recovery.) A conservative blanket processing
rate of one blanket loop volume per year is assumed (achievable even with
batch fluorination) though a much faster rate is probably possible. Separate
NaF beds and cold traps are provided to enable withdrawal of pure U235F6 from

the system if desired, It is assumed that the same UF, reduction tower will

serve both core and blanket,

 

 

 

 
- 124 -

 

 

 

3. Chemical Processing Costs s
a, Investment Costs
- »
The ORNL volatility pilot plant'capital costs can be broken down
as follows:
Replacements and
Design Construction Modifications
Through FY-56 $ 130,000 $350,000 $ ---
Budget FY-57-58-59 210,000 296,000 328,000
Subtotals $ 340,000 $646,000 $328,000
Grand Total $1,314,000
These figures include replacements and modifications which should
not be required in a second plant. They also include solid fuel element han-
dling and dissolution facilities which would not be fequired in the reference
design reactor plant., These costs do not include building and service facil- »
ities for reducing UF6 to UFh and reconstituting fuel. The over«ali plant cost
estimate in Appendix I includes the following capital costs assignablg‘to the "
chemical plant:
v Building and site $ 250,000
VI Equipment and installation | 1,500,000
VII Design 262, 500
VIII Prime Contractor 402,500
IX Contingency __ho2,500
$2,817,500
|
While the breakdo#n may not be exactly accurate, the total is felt
to be an adequate estiméte of the cost. These investment costs are charged v

at sixteen percent per year.
- 125 -

b, Operating costs

 

The ORNL pilot plant operating budget for three fiscal years--1957,
1958, and 1959--totals $1,368,000, The reference design reactor chemical plant
would have lower "unusual costs" but higher "production-proportional costs" and
is estimated to cost $500,000/yr to operate, not including core salt replace-
ment cost which adds $810,000/yr (600 ft5/yr at $1350/ft5). These costs are
included with other fuel cycle costs in the discussion cf the over-all power
costs that follows, Other fuel cycle costs are explained in Section II, Part

G, of this report.

4, Fuel Cycle Economics

 

The nuclear characterlstics of two region molten salt reactors have
been discussed in Secticn II, Part D, General aspects of fuel cycles for
molten salt power reactors have been discussed in Section II, Parts F and G,
together with the interdependence of dollar and néutron economics, For the
purpose of estimating fuel cycle costs for the reference design reactor, the
reactor of Column D, Table XIV, was chosen. (Reactor "E" had a much lower
fuel cost but assumed U235 make-up at $17/g.. If U255 were available at $34/g
such a reactor would have fuel costs about like "D" but would "look better"
because of the higher breeding ratio. Reactor "G" had a fuei cost ~~ ten per-
cent lower than "D" but had only a five foot core., Reactor "C" had very nearly
the same fuel cost as "D", with better breeding ratio but higher salt make-up
charge ) In Section III, Part F, the chemical plant construction cost has
been caMbined with the other reactor complex capital costs The fuel cycle

"operating" costs are shown in Section III, Part F, as 2, 3 mills/kwh, which

results from Reactor "D" as follows:

 

 

 
 

- 126 -

$/yr Mills/kwh »

23 + 25 rental’ 441,000

0.3

25 make-up 2,030,000 1,2 &
Salt make~up 810,000 0.5
- Chemical plant operation 5002000 0.3
) 3,781,000 2.3

The first three $/yr figures come from Table XIV., The chemical
plant operation estimate is given above in this section. The conversion to

mills/kvh 1s based on 240 Mv electricity for 7000 hrs/yr.

 

The cost difference between having the reactor on stand-by or on
the line at full power is 1.7 mills/kwh, the sum of "25 mgke-up" and "salt

make-up,"

F. Cost,Analzsis
1. Introduction '

Any cost analysis of a system such as this natura;;y breaks into
three categories: (l) materials and component development costs. necessary be-
fore construction; (2) design and constructiéncost; and (3) operating costs,
if is on this basis £hat_costs are estimated for the reference design reactor.

For a detaile& analysis, certain assumptions and decisions had to
be made, It is assumed for this cost analysis that the reference design
reactor will be the next molten salt power reactor constructed, This means
qpecifically that it is assumed its construction wouid not be preceded by
the construction of a smaller reactor and that most of the development work
undertaken would be pointed specifically at this reactor. Implicit in the
cost analysis are all of the decisions outlined in the above description of .

the reference design reactor. It is further assumed that for both the

o

 
-127-

development and cogstruction the timing will be such that this work can proceed

in an orderly and businesslike manner without either undue delay or extreme

urgency.

2. Materials and Components Development Costs

 

The present state of knowledge of this system is such that it is
reasonable to expect that a large part 6f the costs to be incurred prior to
consfiructing a power station will be for extrapolating and improving present
designs and testing particular components. |

Section IT of this report gives a review of the present-state of
technology as applied to this reactor, and any development program is necessar-
ily based on the background presented. Specific points of importance are:

1. An alloy (INOR-8) exists which has satisfaptory mechanical and
fabrica@ion properties and fqr which there is reason to expect adequate corro-
seion resistance to the sgits used., However, long-term corrosion tests under
the specific condition imposed by the referénce design reactor are yet to be
conducted. |

2. DSatisfactory %iquid metal and moltén salt handling techniques
are known.

3« Batisfactory equigmept for pump%ng liquid metals and molten
salts has been developed and adquateky proven, but will have to be redesigned
and reproyen in sizes and for operating periods appropriate to the needs of
thls system.

L, A satisfactory chemical process has been devised and is in pilot
plant operation for recovery of uranium from molten salt..

| 5. The temperature coefficient of reactivity of the syétem is such

that no mechanical reactivity control devices are needed other than equipment

for fuel addition,

 

 
- 128 -

O this basis the items listed below would constitute the develop- ¢
ment effort required. Items one and two are required to prove that the y
materials proposed are compatible and suitafile for long-term reactor use.
(1) Out-of-pille pumped loops and
natural convection loops $1,000,000
(2) 1In-pile loops, at least one each
for the core an& blanket fluilds 125002000
| Subtotal $2,500,000
These two items constitute the initial investment in déveiopment
work, and further expenditures for development and construction Qf the reference
design reactor would be spent only 1f the expected favorable results were real-
ized. The $2,500,000 listed for items one and two represent this optimistic
expectation., If disappointing results were obtained, or if a great number
of experiments were performed, the amount could be considerably greater, -
Before construction of a reference design reactor, the folldwing .
items should have development attention:
(3) Pump for.circulation of reactor core salt, blanket salt, and
intermediate coolant salt
() TFuel-to-salt heat exchangers
(5) Coolant salt-to-sodium heat exchangers
(6) Piping and vessels
(7) Instrumentation
(8) Chemical processing
(9) Critical experiments
(10) Remote maintenance equipment »
(ll) Sodium pumps
»

(12) Sodium heat exchangers, boilers, superheaters, blenders, and valves

 

 
- 129 -

For the last two items, only acceptance tests are.included in the estimate.
It is thus assumed that any basic problems of sodium-to-water boilers and
superheqters which are not already solved will be solved elsewhere,

Although rough estimates for individual items 3-12 have been
attempted, it is not believed that their accuracy warrants listing them
individually. Estimates of the costs of development for all the items 3-12
vary from $12,000,000 to $19,000,000, A very large uncertainty is associated
with item ten--remote maintenance equipment--this item canndt be assessed
completely without further study. |

If a molten salt reactor program is to be taken seriously, there
must also be supporting research carried out in metallurgy9 chemistry, and
solid state physics. A portion of this work would be aimed at'future better
modifications of the molten salt system. It is important that this research
bé started early because cf fhe natural time lag between early research re-
sults and practical operating systems, Hence, the following item has to be
included in the total research and development costs:

(13) Supporting research and development --$600,000 per year

Estimates of the total research and development costs (items 1-13)

range from $18,000,000 to $27,000,000.

3. Deslgn and Construction Costs

The best available system flow diagrams for the reference design
resctor have been broken down in@o individual components insofar as possible,
and costs of purchasing, inspecting, and installiné“these components have
been estimated on the basis of standard engineering cost estimating procedure
as modified by experience in the nuclear power field., These modifications are

extensive, as & result of the higher standards required and the necessity for

multiple inspection of every piece that goes into a reactor.

 
- 130 -

Wherever individual components were too small or numerous to be properly J

 

isolated, an attempt has been made to assign a cost figure to an entire subsystem,
such as in the cases of the inert gas systems of helium and nitrogen. In this
manner, an estimated construction cost that is felt to be on the conservative
gide was reached. This includes 15 percent for engineering design; a sum of
$1,000,000 for a period of start-up operations before the plant is "on line";
a contingeqcy fartor of 20 percent of all reactor costs; and a factor of 23
percent which has been found normal for prime contractor fees in reactor con-
struction: The conventionai portibn of the plant, turbine; generator, etc.,
has not been treated in this manner as costs here on an installed basis can be
obtained with a high degree of accuracy §Z/o

This preliminary, but detailed, cost analysis shows an anticipated
cost of $238 per installed kilowatt of generating capacity. A detailed break-

down of the cost estimate is shown in Appendix T.

4y, Cost of Power from the Reference Design Reactor ,

 

Power costs fall naturally into three categories: (1) fixed costs,
(?) operation and maintenance, and (3) fuel cycle costs,

Fixed costs are those charges resulting from capital investment in
the plant. In this study, this investment 1s calculated as shown in Appendix
I to be roughly $57,000,000 or $238/kw of installed capacity. Of this invest-
ment, $99/kw is for the conventional portion of the plant and $139/kw for the
reactor complex portion, incluaing_chemical processing equipment, As pointed

out by Mr, W, K. Davis §§/, a charge of roughly 12 percent per year is applicable

 

87/ See, for example, "Ninth Steam Station Cost Survey", Electrical World,
October 1955

 

88/ Progress in Nuclear Energy, Vol., VIII, McGraw Hill (1957), p 215

O

 
- 131 -

to both these investment figures due to financing and taxes., If, for pur-
poses of computing amortization costs, one assumes & 4O-year life for the
conventional portion of the plant and a 20-year life for the remainder, these
add two and four percent, respectively, to these charges, giving a fixed

cost of 1k percent per year on the $99/kw of conventional plant and 16 per-
cent per year on the $l§9/kw of reactor portion. Using the accepted load
factors of 80 percent for a base load plant such as this results in a fixed
charge of 5.1 mills per kwh,

Operation and malntenance expérience in conventional coal-fired
plants since the last war has been that these charges in plants of this size
have amounted to O.h mills/kwh, and that about 0.3 mills of this charge was
for thg coal burning and heat transfer equipment., Assuming reactor experi-
ence three times as bad leads to a cost (0.3 x 3 + 0.1) for operation and
maintenance of 1.0 mills/kwh,

Fuel and fuel processing costs, as discussed in Section III, Part E,
of this report, amount to 2.3 mills/kwh.

These costs thus add up as follows:

Fixed costs 5.1 mills/kwh
Operation and maintenance 1.0
Fuel and fuel processing 2.3

Total power cost 8.4 mills/kwh

It is probable that a reactor of this type could be built and would
operate safely for some time. The principal uncertainties concern the life of
the components and the ability to replace them in case of failure. These un-
certain points should be investigated as recommended in Section I before the

construction of a large reactor is undertaken.

 
- 132 -

 

The costs outlined above for the construction and operation of the
reference design reactor are predicated on obtaining favorable results from
a development program aimed at removing these uncertainties; with this pro-

vision they are believed to be as realistic as is feasible at this time,

 
I,

- 133 -

APPENDIX

COST ESTIMATE OF REFERENCE DESIGN REACTOR

Fuel Circult

A,

B.

D.

Reactor core

Heat transfer equipment
1, six 6000 gpm pumps )
2, s8ix 2-speed motors, 300 HP )
3, six 100 mw fuel-to-coolant salt exchangers,
16 £t L x 29 inches D, 8000 1b,
3040 sq ft at $55/ft2
4. 100 ft of 12-inch INOR pipes, 5700 1lb at $4/1b
5. 2b welds of 12-inch INOR pipes at $600
6, insulation
T. heating equipment

Fuel handling equipment

1. 2 full volume fill-and-drain tanks (INOR pipes),
450 ft3 each for radioactive core salt

o, five 2-inch shutoff valves (INOR) (remote control)

3., 50 ft 2-inch INOR pipe

Ik, heating equipment

5. fifteen 2-inch welds, etc.

6., insulation

Auxiliary equipment
l, one 4000 £t water tank plus pipes, pumps, etc,
2, one 1000 £t° S8 30k pipe to be put inside H,0
tank to hold gases temporarily
3, one 1300 £t~ SS 304 pipe full of charcoal
L, one 150-ft exhaust stack, 4-ft D,, metal
5. fuel enrichment and sampling equipment
(1n addition to chemical plant)
Subtotal

$ 650,000

660,000

640,000

22,800
14,400
9,600
9,900

240,000

8,000

800
2l 000
3,000
1,500

12,000
9,000

13,000
5,000
50,000

$2,373,000

 

 
- 134 -

II. Blanket Circuit

 

1.
2.
5.
L,
De

A, Container

vessel, pump inlet header, expansion tank
heating equipment

insulation

supporting structure and foundation
biological shield, reactor room

B. ‘Heat transfer equipment

1,
20
3.

L,
Do
6.
To

two 4000 gpm pumps )

two 2-speed motors, 150 HP )

two 35 mw salt-to-salt exchangers, 1500 fte,
4000 1b, $h5/ft2

30 £t of 10-inch pipe (INOR-8)

8 welds 10-inch pipe

insulation

heating equipment

C. Blanket salt handling equipment
2 full volume fill-and-drain tanks for non-radicactive 40,000

1.

2.
5.
b,
50
60

blanket salt, 500 ft° each

five 2-inch shutoff valves (remote control)
50 f£ 2-inch INOR-8 pipe

heating equipment

insulation

fifteen 2-inch welds

D, Auxiliary equipment

1,
2.

30k SS connections to off-gas system
sampling system (in addition to chemical plant)

Subtotal

III. Coolant Salt System (6-core, 2-blanket)

A. Core system

1.
2.

 

900 ft li-inch INOR pipe
six 100 mw salt-to~-Na exchangers, 31 ft L x
30 inches D, 1100 1b, 3600 t2 at $5o/ft2

$ 250,000
4,000
4,000

100,000
250,000

140,000
135,000

5,500
- 4,000
3,000
2,000

8,000

800
7,000
6,000
3,000

10,000
30,000
$1,043,000

25,000
650,000

 
‘-

- 135 -

IIT-A (continued)

 

3, six 6000 gpm pumps ) $ 540,000
L, 6 motors, .300 HP )
5. 2L shutoff valves (remote control) 38,000
6., 2 INOR-8 fill-and-drain tanks (for all 8 systems), 64,000
6450 Tt
7. 2 INOR-8 fill-and-drain tanks, capacity 1 system each, 15,000
200 £t°
8. heating equipment 35,000
9, insulation 40,000
10, welding 100,000
B. Blanket system
1., 210 ft 10-inch pipe 36,000
2, two 25 mw salt-to-Na exchangers, 31 ft L x 20 inches 96,000
D, S 304 shell, 4500 1b, 1200 ft° at $40/rt°
3, two 3000 gpm pumps ) 100,000
4, 2 motors, 125 HP )
5. 6 shutoff valves (remote control) 12,800
6. heating equipment 11,400
7. insulation 9,600
8. welding | 18,000
Subtotal $2,011,000
IV. Sodium Clrcuits
A, From core loops
1, 23%0 ft 18-inch pipe SS 38,800
2. 1450 ft 1l6~inch pipe SS 191,500
3. 3%0 ft 12-inch pipe SS 28,200
L, 360 ft 10-inch pipe SS 23,200
5. 690 ft 8-inch pipe S8 31,600
6. 540 ft 6-inch pipe SS 16,500
7. six 14,000 gpm pumps ) 660,000
8. six 2-speed motors, 375 HP )
9. 6 Na-toE0 boilers, 45 ft L x 32 inches D, 381,000

Croloy, 24,000 1b, 2540 £t at $25/ft

 
IV-A

10,

11,

12,

13,
14,

15,
16,
17.

18.

19,
20,
21,
22,

- 136 -

(continued)

6 Na-to-steam superheaters, 46 ft L x 32 inches D,
SS, 2000 £t2 at $30/ft2

6 Na-to-Na exchangers, 1k ft L x 16 inches D,
2000 1b, 450 2 at $30/ft2

six 18,000 gpm Na-to-Na blenders,
3 ft Dx 3 ft L, 3/8-inch t

six 1L,000 gpm Na-to-Na blenders

6 cold traps and pro rata share of accessory
equipment

30 control valves (incl. control mechanism)

36 shutoff valves (Na)

2 full Na system fill-and-drain tanks, 7000 f£t° at
$11/£¢°

2 Na fill-and-drain tanks (largest single circuit)
1000 £t°

six 18,000 gpm pumps (Na) )

six 2-speed motors, 115 HP )

welding

heating equipment and insulation

B, From blanket locps

l.

O i F W
o

o

10,
11,
12,

 

490 ft 10-inch diameter pipe, SS

200 ft 6-inch diameter pipe, SS

2%0 ft h-inch diameter pipe, SS

two 5000 gpm pumps )

two 2-speed motors, 110 HP )

2 Na=t0=H20 boilers, 45 ft L x 19 inches D, 9000 1b
850 £t° at $35/ft°

2 Na-to-steam superheaters, U6 ft L x 19 inches D,
700 £t° at $ho/ft2

2 Na-to-Na 6000 gpm blenders
2 Na<to=Na 5000 gpm blenders
2 cold traps

6 control valves

6 shutoff valves

$ 360,000

81,000
18,000

15,000
30,000

172,500
7,200
140,000

22,000
300,000

108,000
150,000

31,600
6,100
3,800

120,000

55,000
56,000

5,000
k000
8,000

22,500
1,200

 
- 137 -

 

C
IV-B (continued)
v 13, two 6000 gpm pumps ) $ 70,000
1L, two 2-speed motors, LO HP ) |
15, heating equipment ) 55,000
16, insulation )
C. Reheat loops
1., two 11,000 gpm Nae Pumps ) 180,000
2, two 2-speed motors, 300 HP ) |
3., 6 control valves 21,000
4, 3 shutoff valves (Na) , 600
5. 1 cold trap | 4,000
6. heeting equirment ) 11,500
7. 1insulation )
8, 1 reheat Na-steam superheater, 30 ft L x 68 inches D, 226,500
60,000 1b, 7550 £t° at $30/t°
Subtotal $3,656,000
Ve Miscellaneous Components |
: A, Reactor building 2,000,000
B, Site and site improvement | 500,000
C. Reactor isolation conteiner | 450,000
D, Instrumentation 750,000
E. Auxiliary systems
l. helium system 75,000
2, nltrogen system 150,000
3, auxiliary cooling 250,000
4k, stand-by power system 150,000
5, reactor crane (25-ton) 50,000
6., remote handling equipment 800,000
. 7. 8alt-to-Na exchanger shielding 80,000
8. electric gear for heating system 315,000
. 9, electric gear for pump operation, etc. 400,000

 

- Subtotal $5,970,000

 

 
Vi1,

VIII.

IX.

)

XI.,

XII.

XIII.

- 138 -

Chemical Processing Equipment

 

Engineering Design
15% of Items I, II, III, IV, V, VI

 

Prime Contractor
23% of Items I, II, III, IV, V, VI

 

Spare Parts
A, Pumps
B, Heat exchangers
C. Miscellaneous
Subtotal

Start-up Operations

 

Contingency Reserve
20% of Items I, II, III, IV, V, VI, VII, IX, X

Original Inventories

 

l., sodium 1.3 charges
2. coolant salt 1.2 charges
3, fuel salt 1.2 charges
Lk, blanket salt 1,1 charges

Subtotal

Conventional Equipment

 

l. turbine

2. condenser

3. water works

k., bullding

5. feed water

6. electric gear

T. transmission equipment

8. miscellaneous, office, crane, etec,

Subtotal

$1, 500,000

2,483,000

3,805,000

400,000

250,000

2002000
$ 850,000

1,000,000

4,160,000

75,000
1,160,000
720,000

2,500,000

$l,455,000

7,500,000
- 1,000,000
1,750,000
b, 750,000
2,750,000
1,750,000
2,000,000

1,250,000

$23,750,000
II

IIT

IV

VII
VIII

IX

X1I

XIIT

Item

 

Fuel Circuit
Blanket Circuit
Coolant Salt System
Sodium Circuits

Miscellaneous Components

- 139 -~

RECAPITULATION

Cost

 

$ 2,373,000
1,043,000
2,011,000

3,656,000

2,970,000

Tota; for Reactor

Chemical Processing Equipment

Engineering Design
Prime antractor
Spare Parts

Start-up Operations
Contingency Reserve
Original Inventories

Conventional Equipment

GRAND TOTAL

$15,053,000
1,500,000
2,483,000
3,805,000
850,000
1,000,000
4,160,000

k,455,000

23,750,000

$57,056,000