FFR_chap18.txt

Part 111

LIQUID-METAL FUEL REACTORS

I'rank Maspan, Editor
Brookhaven National Laboratory

18. Liquid-Metal Fuel Reactors

19. Reactor Physics for Liquid-Metal Reactor Design
20. Composition and Properties of Liquid-Metal Fuels
21. Materials of Construction—Metallurgy

22, Chemical Processing
23. Engincering Design

24. Liquid-Metal Fuel Reactor Design Study
25. Additional Liquid-Metal Reactors

R. Bourpnrav
M. B. Bropsky
J. 5. BRYNER

J. CHERNICK

J. G. Y. Coow
0. E. Dwyer
W. P, EATHERLY
J. J. Egax

A, M. Egaya
W. 8. GINELL

L. GrREEN

R. J. IsLER

D. H. GURINKsY
D. HaLn

. B. HiLL

CONTRIBUTORS

M. JANES

0. F. KAMMERER
C. J. KLamur

R. M. Kieun

R. L. MANSFIELD
R. A. MEYER

¥F. T. MILEs

C. RASEMAN

W. RosBa

D. G. SCHWEITZER
T, V. SHEEHAN
H. SusskinND

C. Wane

J. R. WEEKs

R, H. WiswaiL
PREIFACE

This is the most extensive discussion of liquid-metal fuel reactor devel-
opment yet published in the United States. Emphasis has been placed on
the Liquid Metal I'uel Reactor being developed by Brookhaven National
Laboratory and Babcock & Wilcox Co. because it is the most advanced
project. Work on various phases of liquid-metal fuel reactors is being
carried out by Los Alamos Scientific Laboratory, Raytheon Manufacturing
Co., Argonne National Laboratory, Ames Laboratory, and Atomics
International. The editor would like to have given more coverage to work
at the last three locations but was unable to because time was lacking.

The liquid-metal fuel reactor development at Brookhaven started as
an organized program in 1951. Before that, work had been conducted on
bismuth-uranium fuel and other components. In 1954, Babcock & Wilcox
Co., in collaboration with representatives of sixteen other companies,
prepared a reference design and report. In 1956, Babcock & Wilcox con-
tracted with the Atomic Energy Commission to design, build, and operate
a low-power experimental reactor (LMI'R Experiment No. 1). Research,
development, and design studies are being carried on concurrently by
B & W and Brookhaven. LMFR Experiment No. 1, on which construc-
tion is scheduled to start in 1960, is intended to demonstrate feasibility
and provide information on the physics, metallurgy, chemistry, and
mechanical aspects of this type of reactor.

The editor expresses appreciation to many of his colleagues at Brook-
haven and Babcock & Wilcox for working with him on these chapters. He
wishes particularly to thank those whose material he drew upon, also
C. Williams, O. E. Dwyer, D. Gurinsky, H. Kouts, I'. T. Miles, and T. V.
Sheehan, of Brookhaven National Laboratory; R. T. Schoemer, H. H.
Poor, and J. Happell, of Babcock & Wilcox Co.; R. Rebholz and G. Goring,
of Union Carbide Corp.; D. Hall, of Los Alamos Scientific Laboratory;
and W. Robba, of Raytheon Manufacturing Co. Special appreciation is
due Miss Gloria Ministeri for her laborious and prolonged secretarial
work and Miss Dolores Del Castillo for coming to our aid in emergencies.

Upton, New York Frank Maslan, Editor
June 1958
CHAPTER 18

LIQUID METAL FUEL REACTORS

18-1. BAcKGROUND

Liquid metal fuel reactors have received attention since the early days
of reactor technology. The concept of a high-temperature fluid fuel which
could be ecirculated for both heat exchange and chemical processing has
been an intriguing one [1-1].

This type of reactor was first suggested in 1941 but received little research
and development attention until approximately 1947. At this time the
Nuclear Engineering Department at Brookhaven National Laboratory
began its Liquid Metal Fuel Reactor (LMFR) development. A solution
of uranium 1 bismuth was suggested because of the low melting point and
low neutron-capture cross section of bismuth. Coupled with these factors
1s the very high boiling pommt of bismuth, which makes possible the high-
temperature operation of a bismuth-cooled reactor at relatively low
pressures,

Modern steam power plants have a thermodynamic efficiency of approxi-
nuitely 409¢. For a nuclear system to achieve comparable efficiencies, the
working fluid will have to have a reactor outlet temperature in the neigh-
borhood of 500°C. The LMIR is one of the new types of nuclear reactors
having this desirable characteristic. Thus, it is one of the few with poten-
tialities for producing power competitive with the best of the present steam
systems.

18~1.1 Work at Brookhaven National Laboratory. In 1948, an appraisal
of various low-melting alloys was made at Brookhaven. Attention was also
given to metallie slurries consisting of uranium in the form of intermetallie
compounds suspended in liquid metal carriers,  The uranium-bismuth
system appeared to show considerable promiise. Preliminary solubility
studies were completed by 1950 and a start was made on fuel processing
Investigations.

Nince that time the project has steadily accelerated. Chemical aspects
of the fuel and fuel-processing systems have been and are being investigated
in considerable detail. Metallurgical studies of corrosion, mass transfer,
and stability of fuel systems have advanced from short-time crucible tests
to circulating loops of alloy steel operated for many thousands of hours.
Consideration has also been given to the design of such various reactor
components as pumps, piping, valves, heat exchangers, and instruments.

703
704 LIQUID METAL FUEL REACTORS [cHAP. 18

18-1.2. Work of study groups. In common with other reactor concepts,
the LMI'R has been evaluated from time to time as part of the general
Atomic Energy Commission Reactor Development Program. During the
summer of 1953, the LM FR was evaluated under Project Dynamo, and it
was concluded that 1t was an extremely attractive concept if proven tech-
nically feasible. In 1955 an industrial study group, under the direction of
Babeock & Wilcox, made a detailled appraisal and design of the LMFR
concept [19], and reported that it could be proved technically feasible in
the near future and that it appears attractive from an economic point of
view. In 1957, the Babcock & Wilcox Company re-evaluated the LMI'R
and found the outlook as good as indicated previously [21]. Of course, the
development of a new reactor concept of this kind 1s a long-range program.

Present plans call for a buildup of knowledge through the construction
and operation of several LMI'R experiments. The first of these 1s currently
being designed by Babecock & Wilcox.

18-2. GENERAL CHARACTERISTICS OoF Liuip METAL
Fuern REacTORS*

18-2.1 Comparison of fluid- and solid-fuel reactors. In order to better
understand the development and characteristics of the Liquid Metal
Fuel Reactor, fluid- and solid-fuel reactors should be compared, and a
distinetion should be made between the features of fluid fuels in general
and those of liquid metal fuels in particular.

A reactor using a fluid fuel may have the following advantages over one
with solid-fuel elements:

(1) Simple structure. A fluid fuel can be cooled in an external heat
exchanger separate from the reactor core. Thus the nuclear requirements
(of the core) and the heat flow requirements (of the exchanger) need not
both be satisfied at the same place. This may allow design for very high
specific power. For example, material of high cross section, such as tung-
sten or tantalum, which could not be used in the core, could be used in the
heat exchanger.

(2) FKasy fuel handling.

(3) Simplified reprocessing. The reduction to metal, fabrication, canning,
and dissolving steps are eliminated. Because manual steps in refabrication
are unnecessary, decontamination need not be complete. The cooling time
could be made much shorter, resulting in a smaller holdup of fissionable
material.

(4) Simplified waste disposal.

(5) Continuous removal of fission products. The removal of poisons
would improve neutron economy and permit higher burnup. With a lower

*(Contributed by F. T. Miles, Brookhaven National Laboratory.
18-2] CHARACTERISTICS OF LIQUID METAL FUEL REACTORS 705

inventory of radioactive material, the potential hazard would be decreased;
this might reduce the size of the exclusion area required for safety.

(6) Inherent safety and ease of control. Any liquid fuel which expands
on heating gives an immediate negative temperature coefficient of re-
activity. This effect is not delayed by any heat-transfer process. The
rate of expansion 1s limited only by the speed of sound (shockwave) in the
liquid. This instantaneous effect tends to make the reactor self-regulating.
Adjustment of fuel concentration can be used as an operating control.

Disadvantages of fluid fuels are listed below:

(1) Possible fluctuations of reactivity caused by density or concentra-
tion changes in the fuel, e.g., bubbling.

(2) Loss of delayed neutrons in the fuel leaving the core.

(3) External holdup of fissionable material.

(4) Induced activity in pumps and heat exchangers and possible de-
position of fuel and fission products.

(5) Corrosion and erosion problems. Each fuel system has its particular
corrosion problems. These differ greatly from one system to another, but in
every case corrosion is a critical problem which must be solved.

(6) Iligh radiation levels in the reactor and in the component piping
require devclopment of remote maintenance techniques.

18-2.2 Advantages and disadvantages of LMFR. Comparing one liquid
fuel system with another involves relative advantages and disadvantages.
Liquid metal solution systems (in particular, solutions of uranium in
bismuth) [5-12] have the following advantages over aqueous systems:

(1) Metals can be operated at high temperatures without high pressures.

(2) Metal solutions are free from radiation damage and do not give off
bubbles. By using liquid metals, therefore, two factors that may limit the
specific power of aqueous systems are avoided.

(3) Liquid metals have better heat-transfer properties than water.

(4) Metal systems do not have inherent moderating properties and can
be used for fast and intermediate reactors as well as for thermal reactors,
provided the critical mass requirements are not excessive.

(5) Liquid metals can be circulated by electromagnetic pumps if desired,
although the efficiency may be poor, as with bismuth.

(G) Some suitable metals, e.g., bismuth, are cheaper than D-0O.

(7) Polonium, formed from bismuth by neutron capture, may be a
valuable by-product.

Liquid-metal systems have the following disadvantages in comparison
with aqueous systems:

(1) The heat capacity is less than with water.

(2) The higher density may be a disadvantage.

(3) Liquid metals are more difficult to pump.
706 LIQUID METAL FUEL REACTORS [cHAP. 18

(4) The absorption cross sections of the best metals (e.g., bismuth
g, = 0.032 barn) are inferior to D20, although better than H2O. The cross
section of bismuth may be low enough, however, to allow breeding of U233
from thorium by means of thermal neutrons.

(5) For a thermal reactor, moderator must be supplied.

(6) The limited solubility of uranium in bismuth necessitates the use of
enriched U230 or U2?3 as fuel. Uranium-238 or thorium cannot be held in
solution in sufficient concentration to give internal breeding.

(7) Because of items (4) and (5} above, liquid metal fuel reactors are at
least 2 ft in diameter [13] and cannot be scaled down as far as aqueous
reactors can.

(8) The high melting point of most metals makes the startup of a reactor
difficult.

(9) Polonium may represent an additional hazard. However, if the
polonium remains with the fission products, it should not add te the prob-
lems already present.

18-3. Ligvip MEeTaL Frer REacTor TYpES

As a solvent for liquid-metal fuels, bismuth is a natural choice because
it dissolves uranium and has a low cross section for thermal neutrons. Asa
result, research work at Brookhaven National Laboratory has centered
on bismuth-uranium fuels. Other possible liquid-metal fuels are the Los
Alamos Molten Plutonium System (LAMPRE) [14] and dispersions of
uranium oxide in liquid metals, NalX [153] or bismuth [16]. The limited
solubility. of uranium in bismuth is troublesome in some designs. More
concentrated fuels can be obtained by using slurries or dispersions of solid
uranium compounds in bismuth. Among the solids which have been sug-
gested are intermetallic compounds [10] uranium oxide [16], uranium
carbide, and uranium fluoride. Use of a dispersion avoids the limited con-
centration but introduces other problems of concentration control, sta-
bility, and erosion.

Liquid metal fuel reactors would appear to be most useful for large
central station power plants [6,11,17-20] where the integrated chemical
processing, one of the attractive features of an LMFR system, would be
important.

The uranium-bismuth fuel system is flexible and can be used in many
designs. Although other types of liquid-metal systems are certainly possible,
the LMFR at Brookhaven is being designed as a thermal reactor in which
the fuel is dissolved or suspended in a liquid heavy-metal carrier. Ordi-
narily, the liquid metal is bismuth for highest neutron economy, but other
systems such as lead or lead-bismuth eutectic may be used. The moderator
is graphite, although beryllium oxide has also been considered.
18-3] LIQUID METAL FUEL REACTOR TYPES 707

| LMFR
r———_—_—----——-——— S I SIS SR
infegrul‘ | Externally Cooled | Internaily Cooled
{Pot-Type)

| One Fluid !

 

 

 

 

 

 

   

 

Two Fluid

 
   

 

 

 

 

 

1 1
Slightly . U-Bi or Th-Bi or| | Solid Th
;'T“ Enriched Ul"_3' UQ,.-Bi UTh | 1 1ho.,-8i | | Blanket
vy U-Slurry Solution Slurry Shurry Slurry Elements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fic. 18-1. Classification of Liquid Metal Fuel Reactors.

Liquid metal fuel reactors are classified on the basis of their heat-
transfer characteristics (Fig. 18-1) [21]. If heat is transferred within the
core the reactor is said to be internally cooled. If heat is transported by the
fuel to the primary heat exchanger external to the core, the reactor is
externally cooled. The term “‘integral reactor’” implies an externally cooled
system, but one so compact that the reactor and primary heat exchangers
can be placed in the same container.

Iixternally cooled LMFR’s can be divided into two classes, single-fluid
and two-fluid. In the single-fluid reactor the fissionable and fertile ma-
tertals are combined in a single fluid carrier, bismuth. This type of reactor
has no separate blanket, and conversion or breeding takes place within the
core fluid itself. The conversion ratio can be made to approach unity with
the proper choice of such parameters as core size, graphite-to-fuel ratio, and
thorium concentration. However, the most economic design is not neces-
sarily the one having the highest conversion ratio (see Chapter 24). If
no fertile material is mixed with the fuel, the concept reduces to the simple
burner.

The two-fluid externally cooled LMFR (Fig. 18-2) is somewhat more
complex because it has a physically separate core and blanket, but higher
conversion ratios are possible. The blanket can be made in a variety of
ways, making use of either solid or liquid blanket materials. In exploiting
the LMEFR concept to the full, a fluid blanket consisting of a slurry of
ThBis or ThOg in bismuth is used.

A variety of fuels is also possible. In the two-region reactor, critical
concentrations of uranium in bismuth could be below solubility limits;
708 LIQUID METAL FUEL REACTORS [cuaPp. 18

    
    
 
  

 

J’“—MM ‘ ) .~ Radioactive -~ Vac
Salt Fuel Bi-U233. F ps 3.\ Gas Storqge ‘ Pumps
fi Fpsl Process f;T i Po
P
] ump Xenon
Salt Trup

      
 
   
  

U Concn |Bi-U233 Sh[l

EXCESS U233 Control -
‘?‘:’ \ Bi-Po Steam Plant
S e ; Trap
L U233 parg :
~ Storage 4~Pump(s) # (L:
Y

o Bi-Th3Bi3-Po233.U233.Fp's (
| Storage - | 4233 Blanke! | .n.'... 1\ )
r Bi-Th3Bis

U233 Th3 Bis-Bi 1: ‘ '
Blanket ~ % H Il _:“,":'/.c
Bianket
+ FP's Process Graphite ’ L
- - Moderator
Salt Thflf Bi-UZ233 Fyel

Fic. 18-2. Schematic diagram of LMFR, showing reactor, steam plant, and
chemical processing.

 

 
   

therefore solution fuels are possible. Such a fuel for the single-region re-
actors is possible only for small thorium loadings or for burners. Higher
fuel concentrations can be utilized only through the use of slurries. On the
basis of experiments, a maximum slurry content of 10 w/o (weight percent)
of either uranium or thorium as bismuthide compounds in bismuth can be
assumed. If an oxide slurry is used, approximately 20 w/o can be carried
by the bismuth. So far only fuels of U233 and U235 have been investigated
in the LMFR program.

184. LMFR Procram

In the following chapters detailed discussions of the liquid metal fuels
research, development, and engineering work are given. Practically all
the LMFR work is in the research and development stage. In the first
group of chapters, the physics, chemistry, and engineering design of the
LMFER are discussed. In the last chapters, several liquid metal fuel re-
actor designs, based on current research and development, are presented.
It should be understood that these are design studies and it is expected that
more than one liquid metal fuel experimental reactor will have to be built
and operated before a final commercial design is evolved.
REFERENCES 709

REFERENCLS

1. H. Harpax and L. Kowarski, Cambridge University, England, Cavendish
Laboratory, 1941. Unpublished.

2. M. B Le, Fairchild Engine & Airplane Corp.,, NIEPA Division, 1950.
['npublished.

3. E. P. WienuR et al., Argonne National Laboratory, 1944, Unpublished.

4. G. Youna, Outline of a Liquid Melal Pile, USALC Report MonP-264, Oak
Ridge National Laboratory, Mar. 5, 1947,

5. 0. E. Dwyer, Heat Transfer in a Liquid-Metal-Fuel Reactor for Power, in
Chemical Engineering Progress Symposium Series, Vol. 50, No. 11. New York:
American Institute of Chemical Iingineers, 1951, (pp. 75-91)

6. C. Wirnrams and F.o T Mines, Liquid Metal Tuel Reactor Systems for
Power, ibid., No. 11, (pp. 244-252)

7.5 D Arnesrroxn et al,, Studies in the Uranium-Bismuth Fuel System, ibid.,
No. 120 (p. 23)

8. (. J. Rasemax and J. Wemsaan, Liquid-Metal-I'uel Reactor Processing
Loops, ibid., No. 12, (p. 153)

9. D. W. Bareis et al., Processing of Liquid Bismuth Alloys by Fused Salts,
ibid., No. 12, (p. 228)

16. R. J. Trrrern et al., Liquid-Metal Fuels and Liquid-Metal Breeder Blan-
kets, 1bid., No. 13, (p. 11)

1. Nvcrear ENGINEERING DEPARTMENT, BROOKHAVEN NATIONAL LABORA-
rory, Liquid Metal Fuel Reactor Systems, a collection of seven papers, Nucleonics
12(7), 11-12 (1954).

12, 0. E. Dwyer et al., Liguid Bismuth As a Fuel Solvent and Heal Transport
Medium for Nuclear Reactors, paper presented at the Nuclear Engincering and
Scienee Congress at Cleveland, Ohio, Dec. 12-16, 1955, (Preprint 50)

13. J. Cueryick, Small Liquid Metal Fueled Reactor Systems, Nuclear Ser.
and I'ng. 1, 135-155 {1956).

I4. R. M. Kienn, .4 Molten Plutontum Reactor Concept— LAMPRE, USAEC
Report LA-2112) Los Alamos Scientific Laboratory, January 1957: Los Alamos
Molten Plutonium  Reactor Equipment (LAMPRID), Nucleonies 14(2), 14
(February 1956); Molten Plutonium Reactors, in Radialion Safety and Major
Activiites tn the Alomic Inergy Programs, July—December 1956, U, S. Atomic
Iinergy Commigsion. Washington, D. C.: Government Printing Office, January
1957. (p. 43)

15. B. M. Asranam et al,, UOo-NalX Slurry Studies in Loops to 600°C,
Nuclear Sci. and Eng. 2, 501-512 (1951).

16. J. K. Davipson et al., A UOg-Liquid Metal Slurry for Economic Power,
paper presented before the American Nuclear Society at Washington, D. C.,
Dec. 10-12, 1956.

17. F. T. MiLes and C. WinLiams, Liquid Metal Fuel Reactor, in Proceedings
of the International Conference on the Peaceful Uses of Atomic Energy, Vol. 3.
New York: United Nations, 1956. (P/494, p. 125)
710 LIQUID METAL FUEL REACTORS [cHAP. 18

18. D. J. SunaesTaken and . Duruam, Liquid 3Metal Fuel Reactor for Central
Station Power, paper presented at the Nucelear Engineering and Scienee Congress
at Cleveland, Ohilo, Dee. 12-16, 1955, (Preprint 39)

19. Barcock & Wircox Co., Liquid Metal Fuel Reactor; Technical Feasibility
Report, USALC Report BAW-2(Del)), June 30, 1955,

20, . Mars et al., Preliminary Design of an LMFR Power Plant, Nuclear
Set. and Eng., in preparation.

21. Bascock & Wincox Co., 1958. Unpublished.