FFR_chap23.txt

CHAPTER 23

ENGINEERING DESIGN

23-1. RracTtor Drusiay*

The LMFR readily lends itself to a wide variety of designs and arrange-
ments. The concepts proposed to date may be classified according to type
as being internally or externally cooled and either compact or open ar-
rangement of cycle. Such clagssification has been brought about in an at-
tempt to present designs which minimize bismuth and uranium inventories.

If we assume the cost of U235 to be $20/g and that of bismuth to be
$2.25/1b, a U-Bi solution of 700 ppm uranium by weight would cost ap-
proximately $6000/ft3. This high volume cost makes it very important to
design the LMFR with the minimum possible holdup.

In addition to the variety of cycle arrangements, several different
coolants are possible, The U-Bi may be used directly to produce steam, or
a secondary fluid such as NaK or sodium may be used. The LMFR has
also been proposed as the heat source for a closed-eycele, gas-turbine power
plant [2].

23-1.1 Externally cooled LMFR. In an externally cooled LMFR the
fuel is circulated through the core to an external heat exchanger, where the
heat is removed by the secondary fluid. This type provides the simplest
core design, requiring simply an assembly of graphite pierced with holes
for ecirculation of liquid-metal fuel. The major problems of heat transter
are essentially removed from the core design.

23-1.2 Internally cooled LMFR. The internally cooled LMFR is de-
signed so that the liquid fuel remains in the reactor core. The core thus
acts as a heat exchanger in which the heat is transferred to a secondary
fluid flowing through it to an external heat exchanger or steam generator.

The internally cooled design offers a means of substantially reducing
the U—DBi inventory of the system, but considerably complicates the design
of the core. The core must be designed to accommodate two fluids and suf-
ficient surface for transferring heat from one to the other. The introduction -
of a secondary fluid in the core requires a greater uranium concentration
than in the externally cooled system, which has only U-Bi and graphite in
the core. The required concentration cannot be achieved with U-DB1 so-

*Based on material by T. V. Sheehan, Brookhaven National Laboratory, Upton,
L. I, New York.

832
23-1] REACTOR DESIGN 833

From
Recuperator
To Turbine

J .
\ Volatile

Fission Prod.

~——Pump

     
 
 

S

 

 

 

U-Bi

 

n Core

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Heat Exchanger Graphite Reflector

Dump

Fic. 23-1. Externally cooled compact arrangement LMFR for closed-cycle gas
turbine.

lutions, since these concentrations approach the solubility limits for the
temperatures presently being considered (400 to 550°C).

23-1.3 Compact arrangements. The compact arrangement may best be
described as an integral or “pot type” design and may be internally or ex-
ternally cooled. In such a design [1] the fluid fuel remains in the core ex-
cept for small amounts which are withdrawn for reprocessing. The breeding
fluid acts as a coolant by circulating through blanket and core and thence
through heat exchangers which are also contained within the primary
reactor vessel,

Figure 23-1 shows a concept of an externally cooled compact reactor
834 ENGINEERING DESIGN [cHAP, 23

arrangement for a closed-cycle, gas-turbine power plant [2]. In this ar-
rangement the fuel is circulated through the core and heat exchanger,
which are contained inside the same vessel. The compact arrangement
offers a means of reducing the U-Bi inventory over a particular reactor
designed with an open-cycle arrangement. It does, however, increase the
problems associated with design of the core, blanket, and reactor vessel.
The compacting of all the equipment into a single vessel reduces the flexi-
bility of mechanical design which the open arrangement allows, as well as
intensifying the problems of thermal expansion. The reactor vessel not
only becomes larger, but the number of openings is also increased, both of
which complicate the vessel design. Nevertheless, as operating experience
with materials and equipment becomes available, the compact arrange-
ment may provide a means of improving the economics of the LMFR
system.

23-1.4 Open arrangements. The open arrangement is the type receiving
the most consideration at present because of the flexibility and simplicity
of design it affords the system components. Figure 23-2 shows one con-
cept of an externally cooled LMFR using the open-cycle arrangement [3].
In this design both blanket and core fluids are circulated to heat ex-
changers located outside the reactor vessel. This type of arrangement also
allows greater freedom of design for maintenance of equipment. Means
must be provided for removal and/or maintenance of system components
under radioactive conditions. The open arrangement makes it easier to
provide such facilities. The major disadvantage of this arrangement is the
high U-Bi inventory.

The open-cycle arrangement may also be employed in an internally
cooled LMFR to reduce fuel inventory, but it introduces those problems
peculiar to internally cooled systems.

23-1.5 Containment and safety requirements. The high negative tem-
perature coefficient and low amount of excess reactivity available make the
LMFR inherently stable and safe. However, any rupture of the primary
system, whether by reactor excursion or otherwise, would release fission
products and polonium to the surrounding atmosphere. The primary
system must therefore be surrounded with a secondary vessel for contain-
ment of radioactivity in case of such a failure. Since all materials in the
reactor core have very low vapor pressures, the containment vessel need
not be designed to withstand any appreciable pressure. The containment
problem in the LMFR 1s one of containing the high-temperature liquid
metal together with fission products, and such containment can be ac-
complished by lining the reactor and primary circuit cells with a gastight
steel membrane. This containment vessel also acts as a catch basin for
recovery of U-Bi in case of leaks.
23-1]

REACTOR DESIGN

i
’? Saturated 600
| Steam To Super-
‘ heater

| Sodium Inlet
And Qutlet Box

 
 

835

- ——4"" "3+ Na To Superheater
And Reheater

Reheat And Superheat
Liquid Metal Pumps

-

 
   

Water Inlet ,[ -! { ‘
and ol )
Outlet Box E!! ! S jE

 

 

 

Gas Takeoff

 

 

o —
Processed Th-Bi

Slurry Return

Boiler Bundle 4

 

Slurry Suction

#° Header

Blanket
(Th-Bi, Graphite)

—

Liquid Metc|t

To Fuel e 4
4_=V/___ T
Processing

System —=
=
S
Vo

1

 

 

Superheat + Reheat
Bundle

oo e Slurry Drawoff
Grc:ph:te Core

   

From Slurry
Coolers

 

To Slurry
Coolers

 

Slurry
"/ Discharge
Header

Fig. 23-2. Externally cooled open-cycle arrangement LMFR.

 

 

 

 

 

{

I

-

To Fuel

Th-Bi Slurry Pump
Dump

Liquid Fuel Pumps

The arrangement of the containment vessel also depends on the heat-
removal design. If an intermediate heat-transfer fluid such as sodium or
Nal\ is used, the containment may be handled as above. If a direct U-Bi to
steam cycle is used, a double-wall heat exchanger must be used to maintain
double containment, unless the entire building is constructed to act as the
second containment barrier.

In the event of a leak in the system, the U~-Bi would be drained to a
dump tank. This tank would be provided with adequate cooling to remove
the decay heat from fission products.

23-1.6 Design methods. The vessels in an LMFR are designed in ac-
cordance with the Code for Unfired Pressure Vessels [4]. Vessels would be
of welded construction with all seams radiographed and stress-relieved.

The design temperature used can be as high as 1100°F. For 22% Cr-1%
Ao steel, this gives allowable stresses of 4200 psi for normal operating condi-
tions and 9200 psi for emergency, short-duration conditions. These figures
correspond to 19 creep strength for 100,000 hr and 10,000 hr, respectively.
836 ENGINEERING DESIGN fcHAP. 23

23-1.7 Maintenance and repair provisions. Provisions for maintenance
and repair of the LMFR raise several problems. It is anticipated that a
substantial level of activity will be induced in the system by the circulating
fuel. This means that the system should be designed so that it can be main-
tained despite the high radiation level. Several approaches, not mutually
exclusive, to this problem are being considered:

(1) If maintenance or repair to a component is required, the entire com-
ponent will be removed from the system and a new one inserted.

(2) All connections between components will be made in one area, fully
biologically shielded from the components themselves. When a com-
ponent is to be removed, its connections are shiclded from adjacent connee-
tions by portable shielding if the work is to be done directly rather than re-
motely. The connections are broken and the shielding is removed above the
pipes leading to the component in question. The component is removed
with the overhead crane and a new one set in place. The shielding is re-
placed, and the connections are remade. The connections are accessible
and pipes do not overlay each other so as to prevent removal of any
disconnected component. Unfortunately, placing all connections in one
channel increases the fuel inventory since the piping for this arrangement is
somewhat longer than that required for & more conventional arrangement.

(3) Both mechanical and welded connections are being studied, with a
view toward the ease with which connections can be made and broken both
directly and remotely.

(4) Remote methods of performing maintenance tasks (welding and cut-
ting pipe, making and breaking flanged joints and closures) are being
studied, since direct maintenance will not be possible in some areas.

(5) Fluidized powders, shot, and liquids are being studied as possible
portable shielding media.

23-2. HeaT TRANSFER®

In the open-cycle externally cooled, two-fluid LMFR, the bismuth-
uranium solution serves as the primary coolant as well as the fuel. In the
reactor itself, there is no actual heat transfer. Instead, the solution acts as
a transporter of heat to an external heat exchanger. In evaluating bismuth
as a primary coolant, it is helpful to make a comparison between it and
three other coolants: sodium, a typical alkali metal coolant; LiCI-KCl
cutectic, a typical alkali halide salt mixture; and water. (The salt eutectic
used here would not be a suitable primary coolant for a thermal reactor.
Its heat transfer properties, however, are typical of salt coolants.)

*Based on material by O. E. Dwyer, Brookhaven National Laboratory.
23-2] HEAT TRANSFER 837

The ideal primary coolant for a nuclear power reactor should have the
following characteristics:

(1) High heat-transfer rates.
(2) Good gamma, absorptivity.
(3) Low pumping power requirements.
(4) Low melting point.
(5) Low vapor pressure.
(6) Low corrosion rate.
(7) Low chemical reactivity.
(8) Low neutron absorption.
(9) Low induced radioactivity.
(10) Low cost.

In order to have the above characteristics, the coolant should have the
following physical properties in either a high or low amount:

(1) Density (high): affects pumping power requirements, heat-transfer
characteristics, and gamma shielding requirements.

(2) Thermal conductivity (high): affects heat-transfer characteristics.

(31 Specific heat (high): affects heat-transfer characteristics and coolant
flow rate.

(4) Viscosity (low): affects pumping power requirements and heat-
transfer characteristics.

(5) Melting point (low): affects auxiliary heating requirements.

() Vapor pressure (low): affects mechanical design of reactor and system
components.

(71 Volume change on fusion (low): affects startup and shutdown pro-
cedures.

(81 Coefficient of volumetric expansion (high): affects thermal pumping
capacity and, where primary coolant is also the fuel, reactor reactivity.

(91 Electrical resistivity (low): affects applicability of electromagnetic
pumps.

Table 23-1 summarizes the physical properties of bismuth which are
relevant to nuclear reactor design and in the temperature range of practical
interest from the standpoint of electrical power generation [5,6].

23-2.1 Nuclear aspects of coolants. From the nuclear standpoint, five
important characteristics of primary reactor coolants are their capacities
for (1) absorbing thermal neutrons, (2) slowing down neutrons to the
thermal energy level, (3) absorbing gamma radiation, (4) developing in-
duced radioactivity, and (5) resisting radiation damage.

In Table 23-2 the thermal neutron absorption cross section and neutron-
slowing-down power of Bi are compared with those of Na and H20. Bis-
338 ENGINEERING DESIGN [cHAP. 23

TasLE 23-1

PaYsicalL ProprerTIES oF BIsMUTH

 

Atomic weight 209

Melting point 271.0°C (520°F)
Boiling point 1477°C (2691°F)
Volume change on

fusion —3.329,

Temperature, °C 300 400 500 600
Temperature, °F 572 752 932 1112
Vapor pressure,

mm Hg 1079® 3.5X107%*12.3x1074)6.3 x 1074
Density, g/cm? 10.03 9.91 9.79 9.66
Specific heat,

cal/(gm)(°C) 0.0343 0.0354 0.0365 0.0376
Viscosity, centipoises 1.66 1.37 1.16 1.00
Thermal conduetivity,

Btu/(hr)(ft)(°F) 9.9 9.0 9.0 9.0
Electrical resistivity,

ohms 128.9 134 .2 139.8 145.2
U solubility, ppm 480 1850 5100 13000

 

 

 

 

 

 

 

*Extrapolated results.

muth with a macroscopic cross section of 9.0 X 107% em~! at 450°C has
the lowest neutron absorption characteristic of any practical coolant, with
the exception of D20 and certain gases. Its “reactor poisoning’ effect is
at least an order of magnitude below those of sodium and water. The
slowing-down power of Bi i1s very low, however, which means that when it is
used as the primary coolant in a thermal reactor it contributes very little
moderating capacity. The term £No, in Table 23-2 represents the decrease
in the natural logarithm of the neutron energy per centimeter of travel
through coolant.
The gamma absorption coefficient, u, is defined by the equation

dl = —uldx (23-1)

and has the units of em™! Values of u for 450°C Bi at several gamma
energies are shown in Table 23-3, along with those for Na and H:O.
Bismuth, because of its high density, 1s an excellent absorber of gamma
radiation, which means that it provides considerable internal shielding.
The values presented in Table 23-3 are estimates based on the theoretical
calculations of Davission and Evans [8].
23-2]

 

 

 

HEAT TRANSFER 839
TaBLE 23-2
SomME NUCLEAR PROPERTIES OF VARIOUS REAcTOR COOLANTS
Thermal Macero- Thermal q
. . Slowing-
neutron scople scattering "
£ down
Coolant Temp.,| cross Cross Cross | o ot one power
°C  |section [7],} section [7],; Density| section, 1 o
: ess (No,
Ua) ?\ aO', :O) Us? _.1!
barns em ! g/cm3 | barns em
Bi 450 0.032 0.00090 | 9.82 9 0.0095 | 0.0024
Na 450 0.505 0.011 0.841 4.0 0.084 0.0074
H.0 250 — 0.018 0.802 — 1.23

 

 

 

 

 

 

 

 

 

 

*Average decrease in the natural logarithm of the neutron energy per collision.

TABLE 23-3

VALUES oF u, THE GaMMA ABSORPTION COEFFICIENT,
FOoR VARIOUS REacTOR Co0OLANTS A8 A FuncTioN oF ENERGY

 

 

 

 

 

Energy, Mev
. Temp.,
Coolant °C
0.5 1.0 1.5 2.0 3.0
| Bi 450 1.57 0.70 0.52 0.44 0.41
; Na 450 0.070 0.051 0.042 0.036 0.029
-0 250 0.078 0.057 (.046 0.039 (.032

 

 

 

 

Regarding the tendency for developing induced radioactivity, Bi has a
serioux disadvantage, owing to the formation of Po?!Y a very energetic
alpha emitter with a 138.3-day half-life. Its formation and decay can be
represented as follows:

Bir(19mb) —_y. 3j210 ;6, Po2l0 — % Pp206
5d 138.3d

0?7 is one of the most poisonous materials known, the maximum allow-
able concentration in air being 7 X 107 ue/ml or 3.75 X 1078 ppm.
Another troublesome feature of Po2!% is its tendency to scatter throughout
840 ENGINEERING DESIGN [cHAP. 23
any accessible volume, due to recoil from its high-energy alpha emission.
Thus, spillage of solutions containing Po?!? constitutes a most serious phys-
1ological hazard. Inthe LMFR, however, it is not believed that the presence
of Po?!0 in the fuel stream creates a more serious radioactivity problem
than already exists as a result of the fission products.

Sodium is not free of the radioactivity problem either, but as a primary
coolant it is not as bad in this respect as Bi. Water is comparatively free
of induced radioactivity after short holdup times. For the same flux
conditions, Na will give over 20,000 times as much radioactivity, on a
roentgen basis, as HoO.

Liquid metals, because of their simple atomic structure, suffer no radia-
tion damage.

23-2.2 Pumping-power requirements. An important criterion for as-
sessing the relative merits of different coolants is the amount of pumping
power required for a fixed rate of heat removal in a given application. The
three main pressure drops in the primary coolant circuit are those in the
reactor, the external heat exchanger, and the interconnecting piping. A
comparison of the four different types of coolants will now be made on the
basis of their relative pumping-power requirements, with respect to the
interconnecting piping and the heat exchangers. The physical properties
of the coolants are listed in Table 23-4. The properties of the first three
are evaluated at 450°C, as a typical average primary coolant temperature
for such coolants, and those for water at 250°C.

TAaBLE 234

PuaysicaL ProrERTIES oF SoME Tyrical Rracror CooLANTS

 

 

 

 

 

 

 

 

Property Bi Na KCHLiCl H.0
450°C 450°C 450°C 250°C
Melting point, °F 520 208 664 32
Boiling point, °F 2691 1621 — 212
Density, 1b/ft3 615 52.5 103 o0.0
Specific heat, Btu/(1b)(°F) 0.036 0.304 0.31 1.16
Heat capacity, Btu/(ft3)(°F) 22.1 15.95 31.9 57.8
Thermal conductivity,
Btu/(hr) (ft)(°F) 8.95 39.5 1.47 0.357
Viscosity, ep 1.28 0.245 3.4 0.110
Prandtl number, C'p, u/k 0.0125 0.00454 1.7 0.863

 

 
23-2] HEAT TRANSFER 841

The pumping power required to circulate the coolant through the piping
system per unit rate of heat transport for a fixed temperature rise in the
coolant has been shown [9] to be

: 02
bp = % = (a constant) ;#GF. (23-2)
The quantity u%2/p2C,28, represented here by the symbol X, is an index
of the pumping power required to circulate a coolant through a fixed
piping system, for a given heat load. Table 23-5 gives relative values of X
for the four typical coolants mentioned above. The units and values of the
physical properties used in evaluating X are the same as those given in
Table 23—4.

TaAaBLE 23-5

RELATIVE VALUES oF X vorR VARriorus COOLANTS
Frowing THrROUGH A FIXED PIPING SYSTEM

 

 

 

 

Coolant Temp., °C X x 104
Bi 450 308
Na 450 77
LiCHKCI 450 32

eutectic
H.0 250 1.7

 

 

 

 

The very large spread in pumping-power requirements is striking. Bis-
muth has about four times the pumping-power requirements of sodium and
both have manifold greater requirements than that of water, which has
the least of any known liquid. The tremendous superiority of water as a
heat-transport medium is due to its low viscosity and very high volumetric
heat capacity.

23-2.3 Heat transfer for LMFR. So far as is known, no heat-transfer
data have been obtained for liquid bismuth. However, several investigators
[10-14] have published experimental heat-transfer results on the bismuth
lead eutectic and on mercury. For these results the Lubarski and Koffman
equation [15] expresses the results most closely:

"D 0.625(DV,Ca/R0 (23-3)
842 ENGINEERING DESIGN [cHAP. 23

This equation may be used for turbulent flow in round tubes or for turbulent
flow outside round tubes.

In obtaining the heat-transfer coefficients for comparison with bismuth,
the sodium coeflicients were calculated from the Martinelli-Lyon relation-
ship. The coeflicients for molten salt and water were calculated from the
conventional Dittus-Boelter equation.

Using the above relationships and assuming (1) total fixed heat load,
(2) fixed diameter of tubes, (3) fixed inlet and outlet temperatures, (4) av-
erage bulk temperature of coolants same as in Table 23-4, and (5) combined
heat-transfer resistance of tube wall and second fluid equals 0.001, a typical
value for 1-in. ID alloy steel tubes with 0.1 in. wall the values in Tables
23-6 and 23-7 were calculated. Although the heat-transfer characteristics
of bismuth are slightly inferior to those for sodium, it is clear from these
two sets of calculations that all four coolants hehave similarly.

The heat-transport capability of bismuth are simply related to its volu-
metric heat capacity. The values of this property are given in Table 23-4.
Bismuth is definitely superior to sodium but inferior to the fused salt and
water.

To achieve good thermal contact between bismuth and a solid metal
surface, the surface must be cleaned to a high polish, the bismuth must
be free of oxide and dissolved gases, and the system must be filled under a
high vacuum. Guses or oxides on the heat-transfer surface can greatly
reduce the heat-transfer coefficient for bismuth. Bismuth hags a less stable
oxide than the oxides of iron, chromium, and nickel which may be present
on the tube surfaces. Hence the bismuth would have a tendency to non-
wet the walls.

Good wetting of alloyed steels by bismuth may be achieved by adding
small amounts of alkali or alkaline earth metals, by heating to high tem-

TaBLE 236

CoMPARISON OF CooLANTS IN HEAT-EXCcHANGER DESIGN
WHEN NuMBER oF TUBES IN PARALLEL 18 FIXED

 

 

 

Coolant \o?lgf)? noo U, Relative size of
ft/sec’ Btu/(hr)(ft)2(°F) | Btu/(hr)(ft)2(°F) | heat exchangers
Bi1 15 2700 730 1.00
Na 20.8 10230 910 :
LiCI-KCI 10.4 2400 706 1.12
eutectic
H,0 5.73 2360 703 1.12

 

 

 

 

 

 

 
23-3] COMPONENT DESIGN 843

peratures (above 1200°F), or by both. FFor good heat transfer with bismuth
extreme care must be taken to ensure oxide- and gas-free systems.

23-2.4 Heat-exchanger design. In a commerecial liquid-metal fuel sys-
tem, the primary bismuth coolant would probably exchange heat with a
secondary metal coolant before generating steam. Typical conditions for
a 5-Mw countercurrent bismuth-sodium heat exchanger are given in
Table 23-8.

23-3. CompoNENT DEsign*®

This section discusses the design and development experience obtained
on components required in LMFR systems. Besides the requirements for
these systems, considerable component development is needed in the re-
search und development program for experimental apparatus. Both kinds
of components are treated here in detail and by case histories,

23-3.1 Pumps. In the case of liquid-metal pumps, which can be classified
as mechonieal or electromagnetie, a good deal of preliminary development
work hax been done by the Iairchild Iingine and Airplane Corporation
Nuclear Fuoerev for Propulsion of Aireraft Division (NEPA), the Allis-
Cliliners Co., the Babeock & Wilcox Co., and the Government Labora-
tories, WAPL, ORNL and ANL [19].

TasLe 23-7

CovMrarisoN oF CooranTts 1N Hear-Excraxcer DEsIGN
AT FIXEDp LiNEAR VELOCITY OF 13 FT, SEC

 

 

 

 

o
Conlait of tube% Temp., h, U, size of
e | °C Btu/ (W) (i0)2C°F) | Btw/ () (f)2(°F) | heat
parallel exchangers
I3 n 450 2770 730 1.00
N ©1.38n 450 8810 897 0.88
LiC-IKCT 0.6Y9n 450 3200 762 1.03
ertertic
H.0 (¢ 42n 250 5150 837 (.94

 

 

 

 

 

*Busceil on a contribution by C. Raseman, H. Susskind, and C. Waide, Brook-
hiaven Nutional Laboratory.
844 ENGINEERING DESIGN [cHAP. 23

TABLE 23-8

TypicaL ConNpITIONS IN A COUNTERCURRENT,
B1-NaA HeaT EXCHANGER

 

Tube material Low Cr-Steel
Thermal conductivity of tube, Btu/(hr}(ft) (°F) 15.8
Tube inside diameter, in. (.70
Tubhe thickness, in. 0.100
Tube spacing (triangular), in, 0.250
Bi temperature (bulk), °If 850

Bi velocity (outside tubes), ft/sce 15.0
Bi heat transfer coeflicient, Btu/(hr) ({t)2(°I") 3,390

Na temperature (bulk), °F 750

Na veloeity (inside tubes), ft/sec 25.5
Na heat transfer coefficient, Btu/(hr)(ft)2(°F) 12,300
Over-all heat transfer coefficient, Btu/(hr)({t)2(°T) 1,015
Fraction of resistance offered by tube wall 0.60
Heat flux (outside tube surface), Btu/(hr}(ft)?2 101,500
Power density, Btu/(hr)(ft)3 510,000

By, ft3/mw heat 0.56
Na inventory, {t3/mw heat 0.45

 

 

 

 

Lleetromagnetic pumps. In the early days of the LMI'R project, a mag-
netic pump for Bi was deseribed by B. 'eld and L. Szilard [20,21]. The
Fuel Processing Group of Brookhaven National Laboratory required pilot-
plant pumps that would circulate uranium-hismuth fuel with absolutely
no leakage. The U-DBi fuel was eventually to be circulated through an
experimental hole in the Brookhaven reactor where fission products and
polonium would be generated. Since a small flow rate of approximately
1 gpm was desired and efliciency was of little concern, it was decided to
use an electromagnetic pump.

An experimental loop [22] was set up to circulate nonradioactive U-Bi
by means of a General Eleetric Model G-3 ac (IParaday) electromagnetic
pump. This loop ran continuously for 2400 hr. During the first 160 hr the
rig was operated isothermally at a temperature of 645°F; during the
remainder of the time, the loop was run isothermally at 840°F. The U-Bi
solution was circulated for most of this period at a rate of 1 gpm. There
was no sign of plugging or flow restriction.

The General Electric G=3 ac pump was calibrated (Figs. 23-3 and 23-4)
in another AISI type-347 stainless steel liquid bismuth loop at 930°I°
[22]. It was operated continuously for over 13,000 hr.
23-3] COMPONENT DESIGN 845

 

0.08 — —

Efficiency, %

0.06 }— —

0.04 — —]

 

 

| l | 1 | | i |
0 0.2 04 06 08 1.0 1.2 1.4 16 18
Flow, GPM

Fig.23-3. AC electromagnetic pump efficiency. Molten bismuth in AISI type-347
stainless steel cell. (Manufactured by General Electric Co.)

 

 

 

 

 

 

 

 

 

119}
I | I I [ l
200 Vaolts
8 — —
I —
vy
= 150 Volts
-
o
@
I 4 | —
2 100 Volts -
50 Vaolts
{ | . | | L l 1
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

Flow, GPM

Frg, 23-4. AC electromagnetic pump characteristics. Molten bismuth in AISI
tvne-347 stainless steel cell. (Manufactured by General Electric Co.)

The =ame pump was used to circulate bismuth at 930°17 in a 239, Cr-19,
Mo steel loop. - The efheiency and characteristic curves were somewhat
lower than those obtalned in a stainless steel loop. This is probably due
1o =hort-circuiting of magnetie flux in the ferritic steel walls.

A theoretical study [23] was prepared by the Atomic Energy Research
Israblishment at Harwell, Iingland, for linear-induction pumping of
bismunth, The report indicates the feasibility of using this type of pump.
Linear-imduction pumps have been built and successfully used at Ames
846 ENGINEERING DESIGN [cHAP. 23

Laboratory to circulate Mg-Th eutectic (37 w/0o Th) and Bi-U alloy
(dbw/oU) in an Inconel-enclosed tantalum loop [24,25]. The pump
operated successtully in the Mg-Th system for 2000 hr at 1470°F with a
temperature differential of 250°I°, and in Bi-U for 5250 hours at 1740°T
with a temperature differential of 210°F. For ecalibration, about 1 gpm
of Bi-U was pumped at 750°F against a head of 0.5 in., with an efliciency
of 0.169.

Mechanical pumps. Most pump development work has been aimed at
pumping sodium or sodium-potassium alloys. The most serious problem
relative to the design of a mechanical Liquid-metal pump appears to be
that of suitable bearings and seals.

Bismuth was pumped by NIEPA in 1950 [26]. The system was operated
for 37 hr, the maximum flow rate measured was 2 gpm, the maximum head
developed was 66 psi, and the maximum bismuth temperature reached was
1765°F. The pump was a modified Browne and Sharpe No. 206, machine-
tool-coolant pump.

In another experiment [27] NIEPA circulated bismuth with a 50-gpm
centrifugal pump for 100 hr at a mean temperature of 1500°L" with a
temperature differential of 500°F.  An accumulation 1n the sump of a
residue high in oxide content and dissolved elements reduced the flow and
forced suspension of operation. This residue probably resulted from an
impure inert atmosphere above the liquid metal. The container material
selected was AISI type-:3147 stainless steel which had shown some promise
in bismuth solubility tests at temperatures up to 1800°I.

The California Research and Development Corporation made a survey
of the various types of pumps that might be used for liquid bismuth and
came to the conclusion that a centrifugal pump would best fit the need.
A test unit was built that operated for 1037 hr, and a report [28] stated
that the centrifugal pump proved to be a very satistactory means for cir-
culating bismuth in an isothermal system at 700 to 750°L". This pump and
its driver are on a common shaft, the shaft being top-suspended with all
bearings in the motor chamber. Space was provided for a labyrinth to
separate the pump chamber from the motor chamber, although no seal was
used during operation. This pump has also been used to circulate mercury
in a test loop at BNL. It has been run successfully for an accumulated time
of over 4000 hr.

Brookhaven has developed a totally canned overhung-impeller centrit-
ugal pump. Iigure 23-5 shows the major design features of this pump.
These units pump 5 to 25 gpm Bi against heads up to 30 ft while operating
at 525°C. These sump-type pumps run with no bearings in the liquid metal
and have proved reliable so long as sufficient internal baffling is included
to stop surface splashing.

There are several centrifugal pumps that have been used to circulate
23-3] COMPONENT DESIGN 847

 

 

4 i _~Induction Motor

 

 

 

 

 

 

 

 

 

 

 

 

 

   
   

Gas & Vac.
Liquid
Level Cooling
Probes __ Water
2
[ g LJ ™~ Flange Cooling
gy Tubes
Shaft Cooling
Chamber
] Pump Baffles &
P
P j §7 Splash Can
Overhung

 

ol Impeller Shaft

 

 

 

 

 

 

Liquid Level
Side Arm Discharge Inlet

Fig. 23-5. Canned-motor centrifugal pump developed at Brookhaven.

lead-bismuth eutectic [29,30]. They are all vertically mounted sump
pumps with overhung shafts and impellers. All would require a can around
the motor and shaft for a hermetie seal.

The University of California has used a double-volute pump which is
rated at 30 gpm and a 40-ft head at 1000°F.* The lower bearing is 2 ft
wbove the liquid metal. The pump utilizes a packing gland (Johns-Manville
=uper ~eal No. 6) adjusted to allow helium at 2 ps:g to leak out of the
~v=tent at a rate of about 10 ft3/hr.

The pump used at North American Aviation, Inc. [30] is made of cast
~teel. The lower bearing is cooled with a water jacket and a graphite seal

“The vendor is Berkeley Pump Co., Berkeley, Cal. The 71-in.-diameter impeller
and the pump casing are made of AISI type-410 steel. The pump is V-belt driven
by a 30-hp motor.
818 ENGINEERING DESIGN [caAP. 23

minimizes gas leakage from the casing. A flow of 0.82 gpm at 400-rpm
shaft speed and a temperature of 700°F was maintained until oxide dross
forced shutdown of the pump after 496 hr.

A completely canned, modified Series T-34 MD Duval stainless steel
pump was used at the University of California to circulate mercury [31].
The packing gland was replaced with a bushing and any metal leakage
was drained to a reservoir. The pump was driven by a 5-hp, 3-phase in-
duction motor at a shaft speed of 1200 rpm.

North American Aviation, Inc. has circulated tin with a graphite pump
at 2 gpm against a head of 22.5 psi at 1830°F" [32]. The pump has a 4-in.-
diameter impeller and is driven by a variable speed (20 to 2000 rpm) pc
compound-wound motor mounted outside the gastight enclosure to avoid
the high temperatures. A rotating Graphitar bushing on hardened steel
provides the gas seal. The spindle bearings are in a cooled housing. The
pump was operated for 500 hr in one run; this was followed by additional
runs. To overcome differential thermal expansion, a molybdenum adapter
joins the graphite shaft to the stainless steel spindle.

A miniature canned centrifugal pump to circulate bismuth, ideally
suited for in-pile work, has been developed by the Atomic Energy Research
Establishment at Harwell England. The over-all pump dimensions are
33-in.-diameter by 24% in. long, with a 2-in.-diameter impeller. The bis-
muth flow is 1.5 gpm with a head of 9 ft. The motor rating is 0.75 hp and
2800 rpm. Two gas-lubricated bearings are utilized. The material of con-
struction is 23 Cr-1 Mo steel.

The Allis-Chalmers Manufacturing Co. has built a canned rotor centrif-
ugal pump with fluid piston-type bearings to pump bismuth at 1050°F.
The pump is rated at 10 gpm and a head of 25 ft, with an efficiency of 10%,.
Those parts of the pump in contact with the bismuth are made from AISI
type—410 steel. The pump was used in loop GG at BNL to pump bismuth
at 1020°F with a temperature differential of 300°F. After 15.5 hr the pump
failed, due to scoring of the bearings and seizure of the can by the rotor.

23-3.2 Valves. The standard-stem packed gate valves used in early
NEPA bismuth tests [26] proved that special valves would be required
for successful liquid-bismuth operation. High leakage rates through the
packing caused maintenance difficulties throughout the tests.

A 14-in. Fulton-Sylphon bellows-type stainless steel valve was cycled
1000 times at the rate of 77 times/min against bismuth at a temperature
of 1000°F and a pressure of 25 psig. No failure of the bellows or other
valve parts occurred. NEPA also checked valves for metal-to-metal
self-welding effects [33]. Tests of valve operation reached 1500°F with
liquid bismuth on Standard Stellite-faced poppets and seats without indi-
cation of self-welding effects.
23-3] COMPONENT DESIGN 849

The two types of valves which have seen extensive service up to 1050°F
in liquid-metal fuel systems are standard Y pattern globe valves and
needle valves. Due to the stringent requirements of zero gas leakage
{(into or out of the metal systems), the only acceptable stem seal has been
a steel bellows. Packings are unacceptable.

Brookhaven National Laboratory has used both types of valves exten-
sively [22,34]. The 1/2-in. IPS 150-1b Y pattern globe valves constructed
from AISI type-347 stainless steel for all parts in contact with bismuth
(including bellows, stem, and disk) have heen used continuously for over
8000 hr at 930°" without mishap. Similar valves with mild carbon steel
disks (instead of type—347 stainless steel) have been used at 930°F for over
13,000 hr without failure or extensive corrosion.

A high-velocity loop operating with bismuth at 1020°I" at BNL uses
I-in. IPS 150-1b Y pattern globe valves made from 239, Cr-19, Mo steel,
AIST type—130 =teel bellows and disk, and AISI type-416 steel stem.

Needle valves (1 8in. IPS AISI type-347 stainless steel construction,
mcluding the bellows) have been in use for intermittent service (i.e., drain
valves

A~ un additional safety measure, 1/2-in. IPS globe valves used in an
m-pile Toop at Brookhaven National Laboratory have utilized two sets of
bellows [54]. The space between the two bellows was pressurized with inert
gus which wus continually monitored to detect pressure changes (thus
mdieating a valve leak). None was detected.

The valve drives have been modified to facilitate remote operation.
The wlohe valve handwheels are replaced by gears and these are, in turn,
connected to extension rods projecting through the enclosures, Extension
rod~ wre welded directly to the needle valve bellows, Universal joints and
richt-umgle gear drives are used for changes in direction between valve and
operator. When relatively gastight enclosures are desired, as in in-pile
loup~. the extension rods project through rubber-gasketed compression
<ig s,

Ok Ridee National Laboratory has reported on the use of special high-
terperature packing [35] for valve stems. This packing consists of suc-
ces~1ve luvers of Inconel braid, graphite, nickel powder, and another layer
of Treonel braid.

[t hax heen shown at practically all AEC installations that two sections
of x cireukition system can be isolated from each other by freezing a short
seetion of counecting pipe. This plug can be remelted and flow resumed
after a =hort wait. This type of seal is undesirable for uranium-bismuth
solution=, however, since the uranium will deposit at the cold surface.

23-3.3 Piping. Layout features. The most important considerations in
designing piping for a liquid-metal fuel system are the considerable thermal
850 ENGINEERING DESIGN [caaPp. 23

expansion of the pipe when heated from room temperature to operating
temperature, and the expansion of bismuth upon freezing (3%). The former
condition prescribes the type of supports required, while the latter de-
termines the methods and techniques for freezing the metal.

In general, it is desirable to hang pipe from overhead supports, prefer-
ably spring-loaded hangers with straps around the pipe insulation. Heavy
vessels may be anchored to hangers by brackets welded to the wall. Care
should be taken to see that these brackets do not act as a large heat sink.
If the system is supported from below, heavy vessels should “float” by
locating them on freely moving bearing raceways.

Freezing the liquid metal in the system, especially in components with
bellows, should be avoided. However, in case of emergency, the metal
should be frozen towards the free surface. For this reason, a system should
always contain a surge (or expansion) tank, located at the highest ele-
vation.

The use of an integral fill tank, located at the lowest point in the system
to permit charging the loop with metal through a pipe “dip leg” com-
pletely immersed in the metal, is recommended. The application of gas
pressure on the fill tank will transfer the metal slowly into the loop. By
charging the metal from the bottom, into a previously evacuated system,
gas entrainment will be minimized. A sintered metallic filter should be
used to remove oxide and other scum from the metal while filling the loop.
This filter should always be located outside the fill tank, since this will fa-
cilitate removal of the filter when it becomes clogged and will prevent crack-
ing of the pores if the contents of the fill tank freeze.

The loop may be drained into a vessel which can be either the fill tank
or a separate drain tank. Piping lines should be sloped to facilitate drain-
age: undrainable pockets should be provided with separate drain lines or,
if possible, eliminated. A typical liquid-bismuth loop layout is shown in
Fig. 23-6.

Bellows. Several types of metal bellows have been used at Brookhaven
National Laboratory in bismuth systems at 930°F.  AISI type-347 stainless
steel welded bellows have been used continuously in 1/2-in. IPS globe valves
for periods as long as 13,000 hr. The bellows have not, however, been ex-
tensively cycled in bismuth. Their dimensions are 23 in. OD by 1-in. ID
by 0.018-in. thick and contain 32 convolutions. Two AISI type-410 steel
welded bellows have been bench eycled 32,000 and 120,000 times, respec-
tively, in bismuth at 1020°F and should, therefore, be satisfactory. They are
used in pressure transmitters, and are 1} in. OD by 3/8-in. ID by 0.009-1n.
thick and contain 22 convolutions. At this time, one ATSI type—430 steel hy-
draulically formed bellows, used in 1/2-in. IPS globe valves, has been
bench-cyecled with helium over 200,000 times at 1020°F without failure. Its
dimensions are 13-in. OD by 7/8-in. ID by 0.008-in. thick (two-ply).
23-3] COMPONENT DESIGN 851

Bellows tests at Argonne National Laboratory [36] have yielded the
foliowing data:

(1) Failures have generally occurred at a weld; therefore bellows with
the least number of welds are favored. However, mechanically formed
bellows should be examined for cracks and other flaws that may be intro-
duced in the forming.

(2) There was no evidence that corrosion played a part in the failure of
any bellows.

(3) One predominant factor determining bellows life is the relative
amount of travel.

(4) Other factors affecting bellows life are temperature and the relative
distribution between compression and extension. It was found that the
outer bellows failed before the inner bellows which operated at a higher
temperature.

(51 Some bellows designs had not failed up to 106 ¢ycles, at which point
the test was <topped.

Joints. Metal systems. In general, in these metal systems, all joints
should be welded for tightness and structural soundness. All weld joints
are made by standard inert-arc procedures. Complete procedure specifica-
tion= have been prepared by BNL for inert-arc welding of AISI type-347
stainless steel pipe, fittings, and vessels for use with liquid metals. This
procedure was developed through the cooperation of the Metallurgy
Division of Oak Ridge National Laboratory. Specifications have, likewise,
been prepared at BNL for welding 239% Cr-19% Mo steel. AIST type-502
~teel welding rods are used in welding 23% Cr-19 Mo steel pipe. A pro-
cedure tor welding 0.030-in.-thick tantalum tubing, as well as AISI type-316
~tiinless steel to tantalum, has been prepared at Ames Laboratory [24,25].

Fxpermmental and operating procedures, however, often make it ad-
vantauenils to have removable joints. These have been successfully used
at o number of installations.  An oval cross-sectional ring for a flanged
joint wus used by NEPA [37] in a bismuth system between 520 and 660°F
at 500 p~ix. and by the California Research and Development Corporation
[25] o Th-in. piping containing bismuth at 700 to 750°F.

Stwerdard metallie ring-joint flanged connections have also been satis-
fuctorily used at the University of California and Brookhaven National
Liborarory [22.29]. The rings were of soft iron (in lead-bismuth systems)
and NI=T tvpe—347 stainless steel (in bismuth systems). At a temperature
of 45071 the AIST type-347 stainless steel joint has been found to be
heliin leaktight to a mass spectrometer.

The wbility of liquid metals and liquid salts to leak through extremely
=m:ll openings has made the use of helium mass-spectrometer leak testers
a =pecified test step. Halogen leak testers should never be used because of
the ub=orbed halogen which remains in the surfaces after the tests.
852 ENGINEERING DESIGN [cHAP. 23

Graphite system. Several graphite loops have been operated with bis-
muth at a maximum temperature of 2550°F [38] and with tin at tempera-
tures up to 2730°F [32]. Spherical joints held together with steel flanges
and bolts, or tapered joints threaded for assembly under tension, have been
the best. In addition, the joints may be fused to reduce leakage by coating
furfural on one face and hydrochloric acid on the other. However, even
with all these precautions, the systems were not absolutely tight to bis-
muth or tin,

Sight ports. Sight ports have been used to facilitate viewing the liquid
metal inside a closed system at the University of California [29] and at
Brookhaven National Laboratory. A satisfactory port consists of a glass
plate at the end of a steel bellows welded to the pipe. A normally closed
butterfly valve isolates the glass from lead or bismuth vapors. The valve
is moved by an externally mounted magnet or a handle projecting through
a Teflon-packed gland.

23-3.4 Heating equipment. Flexible Nichrome heater wire consisting of
a Nichrome inner wire, asbestos and glass insulation, and a flexible stainless
steel protective braid, is extremely useful for maintaining systems at tem-
peratures up to 1100°F for periods of time in excess of 10,000 hr [22].
Figure 23-6 shows the application of this type of heater in loop work.
Strip and tubular heaters have heen in in-pile service for over 8000 hr [34].
A resistance heater has also been used as an internal heater submerged in a
lead-bismuth eutectic system [29].

Induction heating has been used on bismuth with good results [27].
A heating transformer in which the metal stream is the secondary circuit
has been used at Ames Laboratory in magnesium-thorium and uranium-
bismuth systems at temperatures up to 1740°F for periods of up to
5000 hr [24].

The use of graphite as a resistance heater in graphite loops has been suc-
cessful at temperatures up to 2700°I" for short times (about 500 hr) [32,38].

23-3.5 Insulation. Samples of 26 insulating materials were tested for
possible reaction with molten bismuth [51]. In general, results indicated
that little or no reaction occurred when molten bismuth at 1832°F came
into contact with the unheated materials, but that none of the materials
would withstand contact with the bismuth for more than a few hours when
both were at 1832°1F.

At BNL, Johns-Manville Co. Superex preformed pipe insulation and
Carborundum Co. Fiberfrax bulk insulation have been used extensively.

23-3.6 System preparation. Cleaning of equipment. Owing to the cor-
rosive nature of most bismuth compounds and the necessity for maintain-
23-3] COMPONENT DESIGN 853

Sprlng-Lcodeleongers i _ _ E ;

 

F1g, 23-6. Typical liquid bismuth loop.
854 ENGINEERING DESIGN [cHAP. 23

ing definite concentrations of additives in the fuel systems, the type of
container employed and the condition of the container-liquid interface is
of great importance. The presence of oxygen and other impurities in
soluble or insoluble form can accelerate the attack upon the container
material. As a result, it is desirable to remove all foreign material from
liquid-metal fuel systems before charging. Cleaning techniques for the
more important liquid-metal container materials are summarized as
follows.

Stainless steels. The committee of stainless steel producers of the Ameri-
can Iron and Steel Institute [40] recommend several techniques, depending
upon the type of impurity to be removed. In addition to these methods,
BNL has found electropolishing to be useful in removing surface oxides
[34]. In all cases, after the use of cleaning solutions the material is rinsed
thoroughly with water and dried by allowing a final aleohol or acetone
rinse to evaporate.

Low-chrome steels. Several methods have been used for cleaning metals
of this type. One method is described [30] for PbBi systems m which
boiling detergent solution is used to remove dirt and scale, followed by a
distilled water rinse and drying under conditions of heat and vacuum.
The same reference describes the following cleaning procedure:

(1) Inhibitor: 109, HCI for 12 hr.

(2) Neutralization of HCl with Na2COs.
(3) Water rinse.

(4) 109, phosphoric acid wash.

(5) Drying with heat and vacuum.

The following technique has been developed at BNL for use with large
vessels:

(1) Degrease with trichlorethylene.

(2) 39, HNO3-179, HCI solution for 30 min at room temperature.
(3) Flush with water.

(4) Repeat steps (2) and (3).

(5) Add 209, HCI solution for 5 min at room temperature.

(6) Rinse with water.

(7) Rinse with aleohol.

(8) Dry with inert gas blast.

Leak testing. Liquid-metal fuel systems which involve solutions con-
taining uranium and oxygen-sensitive additives (such as the magnesium
used in LMFR systems) require that precautions be taken to prevent air
leakage into equipment. In general, a sequence of leak detection is followed
in which gross leakage and structural faults are first eliminated by pressure
testing. Suspected leaks can be verified by application of soap solution.
23-3] COMPONENT DESIGN 855

The hellum mass-spectrometer leak detector has been found to be the most
useful as a final test. Systems found leaktight to helium are acceptable for
use in uranium-bismuth systems.

Prehealing. A procedure for preheating equipment has been used at
BNL and elsewhere [3441] in which the equipment is first evacuated to
less than 100 microns pressure and then heated, at a rate slow enough to
prevent pressure surges above 100 microns, to operating temperature.
This procedure has the advantage of removing condensables from the con-
tainer walls before they can react with the wall at eclevated temperatures.

With the equipment at or above operating temperature, purified hydro-
gen may be introduced to reduce any surface oxide that might be present.
This step is frequently done with the liquid metal present in the charging
vessel in order to reduce oxides present in the charge.

Charging proccdires. The procedures deseribed here are specific for the
preparation of LMI'R fuel solutions, but they are at the same time some-
what typical of the handling techniques necessary for other liquid-metal
fuel svztems that have been suggested. Basically, the procedures result from
the need for maintaining system cleanliness, stability of additives, minimum
oxveen contamination, and uniformity of solutions.

Bismodth prcpuration. Bismuth ingots ave cut to a size suitable for loading
and surface oxide deposits are mechanieally removed. The metal is then
charged to a melt tank and heated to the charging temperature under
vacuun. - Zirconium and magnesium, in the appropriate amounts, are
suspended in the melt to establish the proper concentrations of additives.
samples are taken to verify this. When the concentrations of the additives
are =tuble. the bismuth is considered satisfactory for charging to the test
equipment.

Eguipmant charging. The bismuth from the charging vessel is foreed, by
mmert-uas pressure, through a porous metal filter to remove oxides, and into
a =urp tank in the test equipment. From this tank the metal can be raised
by gia= pressure into the operating sections of the equipment.

Addition to flowing bismuth. The addition of uranium, magnesium, and
zivconinm to flowing streams is accomplished by inserting a steel basket,
containing the additive into the bismuth stream through a sampling port.
Inmtial wranium additions to a system are not made until sampling has
shown that the concentrations of magnesium and zirconium are stable.

23-3.7 Operation and handling. Blanket gas. The blanketing of bismuth
with inert gases 1s necessary to provide protection against oxidation. In
many cases it has been found necessary to purify commercial grades of gas
to meet svstenu requirements. A survey of active metals for use in the puri-
fication of rare gases has been made at Ames [421.

Several methods are in use for the determination of oxygen in gases in the
356 ENGINEERING DESIGN [crAP, 23

 

 

 

 

 

 

 

Helium
l

Vacuum

Compression Hollow

Sec1|\ ——Sampling Rod
flfl Vacuum
g M
Atm
)_,__ Ball @
Valve
Bubbler
1/16 In. (

 

 

Opening

 

Fig. 23-7. Thief-type sampler.

ppm range of concentration. KAPL [43]and Oak Ridge [44] have developed
techniques for this analysis and commercial units have also been developed
for use in this range. At BNL, the purity of gas is checked by passing it
over a polished uranium chip at 550 to 600°C. If the chip is not tarnished,
the gas is considered suitable for use.

Conditioning operation. In addition to the system preparation steps de-
scribed in previous sections, it has been found desirable to provide a period
of system operation in which a corrosion-inhibiting layer of zirconium ni-
tride can be formed on the container walls. In general, this is done by
charging the system with bismuth to which zirconium and magnesium
have already been added and then operating the system isothermally until
analyses have shown the additive concentrations to be stable.

Sampling. 'Thief-type samplers have been used almost exclusively for
liquid-metal fuel systems. Sampling in this manner is accomplished by
inserting a sample tube into the metal through an airlock mounted above
the vessel. The airlock is separated from the vessel chamber by a full-
opening ball valve. By bubbling helium through a hole near the bottom of
the sample tube, it is possible to control the depth at which the sample is
taken. At the time of sampling the pressure inside and outside the tube is
eaualized and the liquid enters the tube, which is then withdrawn [22].
23-3] COMPONENT DESIGN 857

    
 
 
 
   

Splash Shield
Semple Cup Holder
Sample Cup Gas &
Vacyum

Connection
Gas & Vacuum Connection

 

 

 

f Sample Cup
A Qutlet Line

  
 

Guide Support

Ball Valve Ball Valve Water Cooling Passeges

Air Lock End This End Heated With Resistance Wire

F1c. 23-8. In-line bismuth sampler.

This method is shown in Fig. 23-7. A variation of this technique has been
adapted for taking filtered samples; an inverted sample cup, which has
been closed at one end by a filter, is lowered into the metal stream and filled
by increasing the system pressure. Another variation involves the use of a
<liding valve on the sample tube. This valve is opened and closed by a
rotary bellows-sealed drive that controls the time at which the sample is
taken. Radioactive samples have been taken using thief-sampler tech-
niques. The activity levels encountered were not high enough to require
remote manipulation, but drybox techniques were necessary to protect
against alpha contamination.

Corrosion study samples are used extensively in developmental systems
and consist of carefully prepared and examined metal or graphite pieces
which are included in the system piping during fabrication and removed
after each experimental run. Samples have also been inserted into flowing
streams through thief-sampler airlocks to study corrosion effects and inter-
actions between the sample and fuel stream components.

A line-type sampler, in which the liquid-metal stream is drawn through
a sample line to a sample container, is shown in Fig. 23-8. In this device,
small cups may be filled in succession and then withdrawn through the
airlock. The sampler is manipulated externally by the pinion gear.

High-temperature radiography. Techniques for radiographing operating
bismuth syvstems at elevated temperatures have been developed to study
plug formation, gross corrosion effects, and operating characteristics such
as liquid levels and gas entrainment. Gamma-ray sources are used in this
work [45].
858 ENGINEERING DESIGN [cHAP. 23

Repair techniques. In making repairs on systems which have contained
liquid-metal fuels it is essential to observe certain precautions:

(1) Whenever possible, the system should be thermally cold.

(2) Blanket gas should always be maintained on the inside of the system.
When the system is opened, a flow of gas from the system should be
maintained.

(3) In making welds, any surface deposit of bismuth must be removed
before a successful weld ean be assured. Removal of a part of the inner
pipe wall by reaming has been found necessary. Cooling coils placed on
the pipe at the end of the reamed section will keep bismuth from melting
and flowing into the weld.

(4) In cases where bismuth fuels have undergone neutron irradiation,
proper protection against polonium contamination must be provided. It
has been found that polonium and nonvolatile fission products contained
in solid bismuth can be handled without little difficulty, since they are
largely immobilized by the bismuth. Repairs of contaminated equipment,
including welding operations, have been made without hazard [34].

23-3.8 Instrumentation. Liquid level measurement. Determination of
liquid levels in a closed metallic system, such as that generally en-
countered in liquid-metal work, can be approached either as a single-
level problem or as a continuously indicating level problem. The require-
ments for the former are:

(1) A metallic probe, preferably of the same material as the metallic
container.

(2) High-temperature insulation between the probe and the vessel in
which the liquid level is to be determined.

(3) A gastight seal between insulation and both adjacent metallie parts.

(4) An appropriate external circuit to note the attainment of the par-
ticular level.

Experience at Brookhaven National Laboratory [22] has shown that the
most successful method for providing both good insulation and a satisfactory
high-temperature seal in a single-level probe is by the use of automotive
spark plugs. It is suggested that the seal be removed from direct contact
with the heat source by means of an appropriate pipe extension. A probe
can be welded to the spark plug after removal of the bent side electrode.
The probes may be made from AISI types—347 and 502 steel for bismuth
systems or of tungsten in a tin system [32]. The external circuit consists
of a transformer, relay, and indicating lights. By the use of two probes and
interlocked relays, it is possible to indicate a level beneath the lower probe,
between probes, or above the upper probe.
23-3) COMPONENT DESIGN 859

There are two general types of continuous level indicators: a manually
adjustable resistance probe, and a variable inductance probe.

The movable probe, consisting of the proper metal rod or tube, is adjusted
through a suitable compression fitting. Modified Parker fittings [29] and
Wilson fittings with Teflon packing glands are recommended. The liquid
level is determined by comparing the probe height with a previously cali-
brated scale.

The variable inductance probe consists of a doubly wound coil in a ce-
ramic form [22]. The coil is inserted into a pipe well inside the tank and,
as the liquid-metal level rises, the inductance of the coil changes. The
change of inductance is detected in a bridge circuit, with the degree of
unbalance being a measure of the level. This method has the advantage,
especially important in handling radioactive fluids, that the system is
hermetically sealed at all times.

If it is not possible to utilize the fluid itself for level indication, the liquid
level may be obtained in a roundabout manner by means of a stainless-steel
float. A stainless-steel tube long enough to protrude from the tank is
attached to the float. A short length of cold-rolled steel rod is contained
in the uppermost section, which is completely enclosed so that no liquid can
come in contact with it. The liquid level is obtained by locating the position
of the cold-rolled steel rod with a search coil wound about a tube concentric
with the one protruding from the tank.

Pressure measurement. Several methods are available for measuring
the pressure exerted by liquid-metal fuels. These include seal pots, gas- or
spring-balanced nullmatic transmitters, and bourdon-type gauges.

The seal pot measuring devices are simple to construct and have been
used most extensively [22,29,30] in this work. The pressure is transmitted
from the metal to a trapped inert gas that is monitoned by a conventional
gas-pressure gauge. This inert gas maintains a constant metal level in the
seal pots, as determined by means of a float [29] or spark plug probes
[22,30]. The float (with an extension rod) or the High-Low spark plug
probes actuate solenoid valves connected to gas supply and vent lines.
The probe separation is 1/4-in., thereby regulating the liquid level to +1/8
in. Since there is no barrier between metal and gas, metal may splash into
the gas space and freeze the gas lines. This may be partly alleviated by
providing long vertical gas lines, a means of heating these lines, and
baffles.

A variation consists of measuring the relative height of a column of
bismuth, backed up by gas pressure in a steel pipe [25]. The level 1s deter-
mined by radiography with an Ir'®? source. This method finds special
application in measuring differential pressure heads (1.e., orifice).

The gas-balanced [46,47] or spring-balanced [48] nullmatic pressure
transmitters provide a metal bellows or diaphragm seal between the liquid
360 ENGINEERING DESIGN [crAP. 23

Inlet Gas Pressure
Zero Adjustment Screw

   

Transformer Care

   
   
   

   

 

   

 

 

 

Return To PR
Pilot Relay {
Transformer . NaK Filled
Balancing .
| Capillary ||,
Nezzle Tube I\ } \l
Calibrated Sensing R
Balancing Bellows NaK Fill
Spring =
Bismuth U_ Tube
Secondary Process w0
Sealing Stream [rrJ
Bellows Nullmatic Gas
Ballasted Type
Cooling Fins Sensing & 1 TER
Diaphragm 5
<+ Holes
Bismuth ; N
Bismuth f = Process Stream 2
Process S'rreom’ 3
Differential Transformer Tyne NaK Filled Bourdon Tube Type

Frc. 23-9. Pressure transmitters.

metal and a gas or mechanical pressure balance; this balancing pressure is
then measured. Figure 23-9 illustrates the basic design of three types of
these transmitters.

The nullmatic pilot-operated pressure transmitter can be made to be
very sensitive, with rapid response. A thin metallic bellows seals the unit
and is the sensing element. The full-range bellows movement is only a few
thousands of an inch. The backing gas is nitrogen and the sensing system
1s adjusted to maintain a maximum differential of 10 psig across the bellows.
One of the difficulties with this type of element is the incomplete drainage
of Bi from the convolutions of the bellows. This trapped Bi may rupture
the bellows when it freezes. Another disadvantage is its large consumption
of instrument gas.

Another type of pressure transmitter utilizes a bellows-sealed differen-
tial transformer. The sensing element of this transmitter is similar to that
of the previous unit and consists of a metallic bellows. The very slight
movement of the bellows during a pressure change is transmitted to a
differential transformer by a rod with a secondary bellows seal. A matching
transformer installed in a bridge circuit allows a calibrated instrument to
indicate or record the actual pressure in the system.

The diaphragm-sealed, NaK-filled bourdon-tube type of pressure trans-
mitter has been used with two different styles of diaphragms. A thin,
0.010- t0 0.015-in.-thick metallic diaphragm is used to separate the Bi system
23-3] COMPONENT DESIGN 861

from a Nal capillary system that extends from the diaphragm chamber
to a bourdon tube in a conventional pressure transmitter. Capillary lengths
up to 20 ft allow the transmitter to be placed remotely with respect to the
system. The other diaphragm style consists of two thin sheets of metal
welded together to form an envelope. The inside of the envelope contains
NaK and is connected to a bourdon element by a length of eapillary tubing.
The envelope diaphragm is suspended in the Bi in an all-welded container.
This type of transmitter has proved to be reliable in the pressure range
between 10 and 175 psig.

Flow measurement. Orifice. Tlow of liquid-metal fuels, much like flows
of water or other liquids, is most commonly measured with standard
orifices [22,25.30]. Work done at the Engineering Research Center,
University of California [39,50] has demonstrated that an orifice may be
calibrated with water, and the calibration may then be used directly for
heavy metal (Bi or Ph-Bi) flow metering. The error introduced in this
manner i< only between 3 and 5%.

Orifice a==emblies have generally been installed in the piping systems
with ring-joint and flange connections; one-picce orifice plate and metallic
O-rines ure used. Fither flange or vena contracta taps are used and the
pres<ure i~ measured as indicated in the previous section. Mild steel orifice
plutes with sharp-edged holes are satisfactory for use in lead-bismuth
svatems [20.30]. After 500 hr at 350°F, and a throat velocity of 1.5 fps,
there was no detectable erosion in one such onfice.

A ronnded-edge orifice (with flange taps) made from AISI type-347
stainfess steel gave very satisfactory service at Brookhaven National
Laborgtory in a 1/2-in. IPS bismuth loop for 13,500 hr at 930°F [22]. The
flow was 5.5 fps through the throat. Upon examination, the hole diameter
hid ncrea=ed by 397 (from 0.2662 in.) during loop operation.

A submerged orifice made from 239, Cr—1 Mo steel has been successfully
used ut Brookhaven National Laboratory in over 4000 hr of operation with
bi=muth at 1020°F. Its special appeal lies in the fact that liquid levels
theads. instead of pressures are measured. Ordinary liquid level probes are
used.,

Elictromagnetic flowmeter. An electromagnetic flowmeter has been de-
signed and analyzed theoretically by General Electric Company and by
Babcock & Wilcox. A permanent magnet is mounted around the pipe
through which molten metal is flowing, with the faces of the magnet creating
a field perpendicular to the pipe. Two leads are welded to the pipe wall,
mutually perpendicular to both the pipe and magnetic flux. The emf gen-
erated by the molten metal when cutting the lines of flux is picked up by
these leads and can be transmitted to any potential-sensitive instrument.
The theoretical analysis of this type of flowmeter agrees within 69, with
experimental results.
862 ENGINEERING DESIGN [cHAP. 23

The electromagnetic flowmeter has been successfully used to meter
bismuth flows in AISI type-347 stainless steel at Brookhaven National
Laboratory [22]. The measured flow agreed within 1097, with the theoreti-
cally determined value.

Preliminary results have shown that these flowmeters may also be used
in a 23%, Cr-19%, Mo steel system. However, corrections must be made for
the short-circuiting of magnetic flux in the ferritic steel pipe walls. One way
of minimizing this correction might be to use a bimetallic cell, that is, a
thin (0.010 in.) liner of 239 Cr-19% Mo steel surrounded by an AISI
type-347 stainless-steel pipe to provide structural strength.

Temperalure measurement. The temperature of liquid metal fuels is
usually measured with thermocouples of duplex Chromel-Alumel, No. 20
BWG gauge. Each wire is individually insulated with fiberglass and
asbestos and each pair is covered again with insulation.

The best and most accurate service in low-chrome or stainless-steel
systems is obtained by welding the thermocouple junction directly to the
outside of the pipe wall. The difference between the temperature on the
pipe wall and the bulk bismuth at 930°F is no greater than 10°F. If re-
quired, thermocouples located in wells have also been used in bismuth
systems.

In graphite systems the thermocouples are inserted in drilled holes, and
then cemented in place with alumina cement [32].

Temperature control for isothermal loops is obtained as follows [22].
The various parts of the loop are heated by means of individual heater
circuits. Since the current demand varies, depending on the position of the
heater in the loop, the current to the heaters is adjusted by means of in-
dividual autotransformers on each circuit. The entire heater group is
supplied from a single line whose voltage varies according to the signal
supplied to a controller by a single, centrally located thermocouple. The
voltage 1s varied by means of a transformer whose primary is in the feed
line. While the loop temperature remains within the neutral band around
the set point of the controller, the secondary coil circuit is closed. If the
temperature drops below the neutral band, the relay opens the secondary
coil circuit, thus decreasing the inductance of the primary, and increases
the voltage to the heaters. If the temperature rises above the neutral band,
the controller relay opens the main circuit breaker and cuts off current
to the heaters.

By proper adjustment of the individual Variacs it i1s possible to maintain
the temperatures around the loop within 20° of the desired value and to
operate so that the main circuit breakers are rarely opened.
863

REFERENCES

1. R. J. Triter, An Internally Cooled Liquid Metal Fuel Reactor Design, in
Proceedings of the First Nuclear Engineering and Science Congress, Vol. 1, Problems
in Nuclear Engineering. New York: Pergamon Press, 1957. (pp. 292-301)

2. T. V. Sueena~x and L. D. Stovenron, The Liquid Metal Fuel Reactor
Closed-Cycle Gas Turbine Power Plant, Mech. Eng. 78, 609-702 (1956).

c 3. C. Wrnnuiams and ¥. T. Mines, Liquid-Metal-Fuel Reactor Systems for
Power, in Chemical Ingineering Progress Symposium Series, Vol. 50, No. 11.
New York: American Institute of Chemical Engincers, 1954, (p. 245)

4. Frank W. Davis, Feasthility Study of Pressure Vessels for Nuclear Power
Generating Reactors, USALC Report AKCU-3062, Division of Reactor Develop-
ment, AILC, December 1955, (pp. 5-6)

5. C. WiLniams and F. T, Miues, Liquid-Metal-Fuel Reactor Systems for
Power, 1 Chemical Engineering Progress Symposium Series, Vol. 50, No. 11.
New York: American Institute of Chemical Engineers, 1954, (pp. 245-252)

6. R. N. Lyox et al., Liquid Metals Handbook, U. S. Atomic Energy Commis-
sion and U, 8. Navy. 2nd ed. Washington, D. C.: U. 8. Government Printing
Ofhee, 1952,

. D.J. Huenes and J. A. Harvey, Neutron Cross Sections, USAEC Report
BXN1.-325. Brookhaven National Laboratory, May 1955.

s. AL Davisson and R. D. Evans, Gamma-Ray Absorption Coefficients,
Rev. Modern Phys. 24(2), 79-107 (1952).

9. 0. E. DwyER et al., Liquid Bismuth As a Fuel Solvent and Heat Transport
Medium for Nuclear Reactors, USALLC Report BNL-2432, Brookhaven National
Laboratory, 1935.

10. L. M. TrererHAN, Heat Transfer Properties of Liquid Metals, Cambridge
University, England, Christ’s College, July 1, 1950.

11. =. E. Isaxkorr and T. B. DrEw, Heat and Momentum Transfer in Tur-
bulent Flow of Mercury, in Proceedings of the General Discussion on Heat Trans-
fer. Institution of Mechanical Engineers (London) and American Society of
Mechanical Engincers, 1951, (pp. 405-409)

12, W. K. StroMQUuisT, Effect of Wetting on Heat Transfer Characteristics of
Liguid Metals (thesis), USAEC Report ORO0-93, University of Tennessee,
March 1953,

13. H. A. Jou~soxn et al., Heat Transfer to Mercury in Turbulent Pipe Flow,
USALC Report AECU-2627, University of California, Berkeley, Institute of
Engineering Research, July 1953.

14. . A. Jouxson et al., Heat Transfer to Molten Lead-Bismuth Futectic in
Turbulent Pipe Flow, Trans. Am. Soc. Mech. Engrs. 75(6), 1191-1198 (1953).

15. B. Lusarsky and S, J. KaurMAN, Review of Experimental Investigations
of Ligad-Metal Heat Transfer, Report NACA-TN-336, Lewis Flight Propulsion
Laboratory, March 1955,

16. R. N. Lyox, Liquid-Metal Heat Transfer Coefficients, Chem. Eng. Progr.
47(2), 75-79 (1951).
864 ENGINEERING DESIGN {[cHAP. 23

17. R. C. MartiNELLl, Heat Transfer to Molten Metals, Trans. Am. Soc.
Mech. Engrs. 69(8), 947-959 (1947).

18. O. E. DwyEer, Heat Exchanger in LMF Power Reactor Systems, Nucle-
onies 12(7), 30-39 (1954).

19. R. L. MorGaN, Technical Information Service, AEC, 1952. Unpublished.

20. B. Fewp and L. SziLarn, A Magnetic Pump for Liquid Bismuth, USAEC
Report CE-279, Argonne National Laboratory, 1942,

21. B. Fewp, More Calculations in the Bismuth Pump, USAEC Report CP-326,
Argonne National Laboratory, Oct. 17, 1942,

22. C. J. RaseEman and J. Wersman, Liquid Metal Fuel Reactor (LMFR)
Processing Loops. Part 1. Design, Construction, and Corrosion Data, USAEC
Report BNL-322, Brookhaven National Laboratory, June 1954.

23. D. A, Warr, A Study tn Design of Traveling Field Electromagnetic Pumps
for Liguad Metals, Report AERE-ED/R-1696, Great Britain Atomic Encrgy
Research Establishment, June 12, 1955,

24. G. R. Winpers and R. W. Fisuer, An Electro-magnetic Pump and Healing
Transformer for High Temperature Liquid Metals, USAEC Report ISC-547,
Towa State College, Dec. 6, 1954,

25. R. W. I'isner and G. R. Winpers, High Temperature Loop for Circulat-
ing Liquid Metals, in Chemical Engineering Progress Symposium Series, Vol. 53,
No. 20. New York: American Institute of Chemical Engineers, 1957. (pp. 1-6)

26. R. 8. WinGarp, Jr., Fairchild Engine & Airplane Corp., NEPA Division,
1950. Unpublished.

27. J. F. Corrins, Fairchild Engine & Airplane Corp., NEPA Division, 1950.
Unpublished.

28. J. E. WaLkey, California Research Corporation, 1951, Unpublished.

29. H. A. Jounsox et al., The Design and Operation of a 30 Gpm 40 Kw Pb-Bt
Eutectic Heat Transfer System, USAEC Report AIKCU-2848, University of Cali-
fornia, Berkeley, Institute of Engineering Research, February 1954,

30. R. Cyaan, Cerculation of Lead-Bismuth Futectic at Intermediate Tempera-
tures, USALLC Report NAA-SR-253, North American Aviation, Inc., Oct. 1, 1953.

31. H. A. Jouxsox et al., Heaf Transfer to Mercury in Turbulent Pipe Flow,
USALEC Report AECU-2627, University of California, Berkeley, Institute of
Engineering Research, July 1953.

32. R. D. Kren, High Temperature Ligpuad Metal Circulating System, USAEC
Report NAA-SR-985, North American Aviation, Inec., Aug. 1, 1954,

33. T. A. Simms, Fairchild Engine & Airplane Corp., NEPA Division, 1950,
Unpublished.

34. C. J. RasEMan et al., Liquid Metal Fuel Reactor In-pile Fuel Processing
Loop (Loop B); Construction, Operation, Erperimental Results, USAEC Report
BNL-403, Brookhaven National Laboratory, January 1957.

35. W. B. CorrrELL, Oak Ridge National Laboratory, 1952. Unpublished.

36. W. P. BiGLER, FReactor Engineering Quarterly Report for March 1, 1950,
Through May 31, 1950, USAEC Report ANL-4481, Argonne National Labora-
tory, July 1, 1950.

37. R. Porrer et al., Fairchild Engine & Airplane Corp., NEPA Division,
1950. Unpublished.
REFERENCES 805

38. W. J. HaLLerr et al., Dynamic Corrosion of Graphite by Liquid Bismuth,
USAEC Report NAA-SR-188, North American Aviation, Inc., Sept. 22, 1952.

30. R. A. Sepax et al., Flow Metering of Mollen Lead-Bismuth Eutectic, al
University of California, Berkeley, California. University of California, Berkeley,
Institute of Engineering Research, April 25, 1949.

40, Am. Machinist, Nov. 12, 1951.

41, O.J. Evaerr et al., Dynamic Corrosion of Steel by Liquid Bismuth, USAEC
Report LWS-24891, California Research and Development Co., Aug. 29, 1952.

42, . S. Giess ot al., Purification of Rare Gases. 1. A Comparison of Aclive
Metals in the Purification of Rare Gases, USAEC Report ISC-560, Towa State
College, Dece. 30, 1954.

43. L. P. Prrkowrrz and E. L. SHirLey, Quantitative Determination of
Oxygen in Gases, Anal. Chem. 25, 1718-1720 (November 1953).

44, Lrnanp A, Manw, Oak Ridge National Laboratory, personal communica-
tion.

45. ], C. AvstiNy and P. Ricuagrps, Radiography As a Hot Lab Service,
Nucleonies 12(11), 78 (1954).

16, P.W. Tavyvror, Moore Pressure Transmitler Test Summary, USAEC Report
CF-533-1-260, Oak Ridge National Laboratory, Jan, 22, 1933.

47. 2. T. Moraax, Hermetically Sealed Iligh-Temperature Pressure Trans-
witter and Hermelically Sealed Iigh-Temperature Liquid Level Probe, USAEC
Report ORNL-1939, Oak Ridge National Laboratory, Sept. 15, 1955.

i~ 15O Kixe and V. K. Heckey, High Temperature Pressure Gauge, Tech-
nicul Report No. 45, Mine Safety Appliances Co., Jan. 5, 1956.

10, E. A\, Lueskn, Knolls Atomic Power Laboratory, 1952. Unpublished.

500 1. A, Jounson et al., Orifice Metering Coefficients for Lead-Bismuth
Footorde, USALC Report AECU-2798, University of California. Berkeley, Insti-
tut: o Enainecering Research, December 1953,

51, W, 2. Fresaman and C. G. Covuins, The Effect of Molten Bismuth on In-
suli'ing Matersals, Report NEPA-1306, Fairchild Engine & Airplane Corp.,
NEPA Division, Feb. 9, 1950.

52. 1. Cyeax, Lead-Bismuth Eutectic Thermal Convection Loop. USAEC
Leport NAA-SR-1060, North American Aviation, Inec., Oct. 15, 1954.