ocr/ORNL-2278.txt

  
   

AEII RESEARCH AND DEVELOPMENT REPORT Re?jif Specia

Wi I iy

3 Y45k 035147 2

       
   

VISCOSITY MEASUREMENTS ON MOLTEN
FLUORIDE MIXTURES

  
      

S. |. Cohen
T. N. Jones

i BEBlﬂSSlHED

By AuTee H.n"r 0Oy: ? //t— -'(p /
il AP di. .. - é.z 1R 0D nm

       
    

(o) VA N (e AT
DOCUMENT COLLECTION

LIBRARY LOAN COPY
Do IIOT THMISFEFI TG LR PERSBH

3 lf you R someone else to see lhis
' document, send in name with dﬂwmm‘l
and the library will arrange a Iunn

  
 

.5A

ors

OAK RIDGE NATIONAL LABORATORY

OPERATED BY
UNION CARBIDE NUCLEAR COMPANY

A Division of Union Carbide and Carbon Corporation

POST OFFICE BOX X * OAK RIDGE, TENNESSEE

 
wtit ot 2270

This document consists of 50 pages.
Copy &~ 208 copies. Series A,

Contract No. W-T4O5-eng-26

Aircraft Reactor Engineering Division

VISCOSITY MEASUREMENTS ON MOLTEN

FLUORIDE MIXTURES

S. I. Cohen
T. N. Jones

DATE ISSUED
JUN = 8 1957

OAK RIDGE NATIONAL LABORATORY
Operated by
UNION CARBIDE NUCLEAR COMPANY
A Division of Union Carbide and Carbon Corporation
Post Office Box X
Oak Ridge, Tennessee

ARIETTA ENERGY SYSTEMS LIBRARIES y -

IR

3 445k 0351470 e

   

 

 

 
b | -ii- ORNL-2278 /
- \ ' Reactors-Special Features

' of Aircraft Reactors
INTERNAL DISTRIBUTION

1. C. E. Center 51. E. J. ??éeding

2. Biology Library 52. E. R, Dytko
3. Health Physics Library 53. R. Ji Gray
L-5. Central Research Library 54, D."P, Gregory
§. Reactor Experimental 55. . E. Mott
Engineering Library 564 G. L. Muller
T-11. fﬁﬁgratory'Records Department 57. R. D. Peak
12. Labowatory Records, ORNL R.C. gf58. W. B. Cottrell
13. A. M. inberg ~# 59. 8. J. Cromer
14, L. ;?%§§1§t (x-25) 4 60. J. H. DeVan

15. J. P. Murr
16, J. A. Swartout
17. E. H. Taylor %
18. E. D. Shipley
19, M. L. Nelson
20. 8. C. Lind
2l. F. L. Culler
22. A. H. Snell
23. A. Hollaender
2k, M. T. Kelley ;
25. J. H. Frye, Jr. &
26. K. Z. Morgan s
27. C. P. Keim

   
  
  
 

(Y-12) & 61. W. K. Ergen

K 62. A. P. Fraas
S 63. W. T. Furgerson
! 64. A. G. Grindell
65. W. H. Jordan
66. F. F. Blankenship
€7. W. R. Grimes
68. F. Kertesz
69. H. W. Savage
T0. W. D. Manly
71. E. P. Blizard
T2. R. E. MacPherson
73. E. R. Mann

28. R. 8. Livingston kN 7h. L. A. Mann

29. T. A, Lincoln 7 Ny, 75. W. B. McDonald

30. A. S. Householdey By 76. R. V. Meghreblian
31. C. S. Harrill " TT. A. M. Perry

32. C. E, Winters “. 78. D. H. Platus

33. D. W. Cardwell *79. D. L. Platus

34. E. M. King / - 80, R. D. Schultheiss
35. A. J. Miller/ 81.:D. B. Trauger

36. R. A. Charpie 82. M. M. Yarosh

37. J. A. Lane/ ' 83. W. F. Boudreau

38. M. J. Skirnner 8k. R. B. Lindauer

39. S. I. Colen 85. D. C. Hamilton

LO. W. R. Gawmbill 86. R. E. Moore

41, N. D. Greene 87. G. J. Nessle

42, H. W. Hoffman 88. R. F. Newton

43, T. N. Jones 89. L. G. Overholser
L, J. J. Keyes 90. R. E. Thoma

4. A. I, Krakoviak 91. G. M. Watson

46. F. E. Lynch 92, C. J. Barton

47. L. D. Palmer 93, C. M. Copenhaver
8. W. D. Powers oL, ORNL - Y-12 Technical Library,
b9, W. J. Stelzman Document Reference -Section
50. J. L. Wantland o . X

‘ . AP
¥
! 3

o - &
. . - g
(‘?

wtdid -

% EXTERNAL DISTRIBUTION

95. Air Technical Intelligence Center
0500, AP troject Cffice, TFort Worth
’ . Albuquerque Operations Office
. Argonne National Laboratory A
L. Armed Forces Special Weapons Project, Sandia
R, Armed Forces Special Weapons Project, Washing
\Assistant AF Plant Representative, Downey
gsistant Secretary of the Air Force, R&Dﬁ

  
  
    
 

   
 
   
 
  
   

105-110. gmic Energy Commission, Washington
111, Njcs International
112. Rle Memorial Institute

113-11k, Nplant (WAPD)
115. & Aeronautics

116. Bureau oMgAeronautics (Code 2&)

117. Bureau of ¥

118. Chicago Ope¥s

119. Chicago Patel§ Group ﬁf

120. Chief of Nava¥;Research

121. Convair-GeneralDynamics Cérporation

122. Curtiss-Wright s%rporatlon

123. Engineer ResearcAyand Development Laboratories
12k-127. General Electric Cjmpary (ANPD)

128. General Nuclear Engﬁwfering Corporation

129. Glenn L. Martin Comfigny

130. Hartford Area Offlceiﬁu
131-132. Headquarters, Air. Forﬁl

133, TIdaho Operations.Offic’b

134, Knolls Atomic Pﬁﬁer Labératory

135. Lockland Area Office W

136, Los Alamos Sgientlflc Labbgatory

137. Marquardt Ajrcraft Company A

138. National Aﬂvisory Committeegfor Aeronautics, Cleweland

139, National Adv1sory Committee 'f
140-142, Naval Ai¥ Development and Matey
143-1kk, Naval Research Laboratory (1 &dpy to L. Richards)

145, New Yqﬁk Operations Office S

146, i American Aviation, Inc. (Missile Development Division)
147. Nuclfar Development Corporation Rg

Special Weapons Center

  
  
 

 
 
  
  
  

148, ce of the Chief of Naval Operayions (OP~361)
149, ent Branch, Washington
150. BRtterson-Moos
151-155. @Pratt and Whitney Aircraft Division ox Project) (1 copy %o

#S. M. Kepelner)
f. San Francisco Operations Office
Sandia Corporation
School of Aviaition Medicine
Sylvania Electriec Products, Inc.
160. Technical Research Group
161. USAF Headquarters &
“462. USAF Project Rand RORY
163. U. S. Naval Radiological Defense Laboratory ‘.
16L4. University of California Radiation Laboratory;“leermore
165-182. Wright Air Development Center (WCOSI-3)
183-207. Technical Information Service Extension, Oak Rldge
208. Division of Research and Development, AEC, ORO ",

s,
*
2

TABLE OF CONTENTS

ABSTRACT .« v vv e es s enenananeennennnnnnesnenannasscenannns errene 1
INTRODUCTION . ........ e eraaean. et et e re e, 2
EXPERTMENTAL APPARATUS AND TECHNTQUES .o evvrvneeeencnnnneresnonnsons

The DrybDOX +.ceveescoresacocrcosscsoonnsonsoorssoenssssovosasonsscss

The Fm‘mce 0 A S PP RS e R NP b YD st R T RE DO R0 LT OE NS E RS rODY S SDO G

-~ W\

The Capillary Viscomeler .c.ceescvooconnes P -~
The Brookfield Viscometer ....... ceoscsccsnoenss sececeseiesnescosease 1O
INSTRUMENT CALIBRATION ...-6ecceovseococooecsecnsnssaancscansnsens cassess 19
EXPERTMENTAL RESULTS ....c00v000. tsccsssessacesotnansens cresecsenscsss 29
CORRELATIONS ..... crsenone cecavan tsecsceseranona G sconasuntasast e as . 34

RHERMCES * & & % @ & v 9T *® % % 0 3 B B KB B8 F A OO0 0 & P R GO T DL R R OO H SN EF O 9 o 9 & 3 & 0 O 45
 

ABSTRACT

This report is a summary of the experimental viscosity
program on fused fluoride mixtures which has been carried
out in support of the ANP effort at the Oak Ridge National
Laboratory. The experimental techniques which have been de-
veloped are described, data on the visceosity of 36 mixtures

are tabulated, and several correlations involving these data

are discussed.
IRTRODUCTION

The value of fused fluorides as coolants and fuels for high tempera-
ture, circulating-fuel reactors was recognized early in the CORNL-ANP program.
However, the scarcity of known and reliable data on the thermal properties
of these flulids seriously hampered the initial design efforts. To remedy
this situation, a program was outlined to obtein experimental data on the
viscosity, enthalpy and heat capacity, density, and thermal conductivity of
the fused fluorides and their mixtures. This report is a summary of the
portion of this prdgr&m concerned with the measurement of the viscosity of
the fused fluoride salts.

The experimenter is confronted with a number of difficulties in the
measurement of the viscosity of these fluids:

(1) The high melting points and wide extent of the liquid state

of the fluoride salts establishes a temperature region of
experimental interest between SOOOC and l,OOOOC.

(2) All of the common materisls available for the construction

of the experimental apparatus corrode in the molten fluoride
salts. In the presence of even small amounts of oxygen or
water vapor, the corrosion rates are greatly accelerated.

(3) Since these salts in the molten state combine strongly with

both oxygen and water vapor, the purification of the salt is
difficult. Further, to insure & continuing purity, and hence

avoid large amounts of contaminating corrosion products, it is
necessary to maintain the molten salts in an inert atmosphere.
() Some of the salts, such as beryllium fluoride, present prob-
lems in handling due to their toxicity.
(5) The fluid viscosity in the temperature range of intérest is
low, varying from approximately 2 to 15 centipoises.
These difficulties impose severe requirements on the measuring equipment by
limiting the choice of techniques, construction materials, and instrumentation
end by complicating the experimental manipuletion of the apparatus.

A number of techniques hasve been investigated by the experimental groups
at the Oak‘Ridge National Leboratory in developing suitable instruments for
measuring the viscosities of fused salts. Knox and KErteszl and later Redmond2
3 used a modified form of the Brookfield rotational viscometer. El‘cvbiasslF

developed a gravity flow capillary efflux viscometer. Knox and KErtesz5 oper-

and Cisar

ated a capillary device in which fluid flow was accomplished by gas pressure.
A "falling-ball" viscometer using & radioactive cobslt-plated Pyrex bead was
developed by Redmond and Kaplans.

After review of this early work on the measurement of the viscosity of
the fluoride salts, it was decided that the most expedient route to results
of the desired sccuracy was through further refinements of the apparatus and
techniques for the modified Brookfield and capillary efflux viscometers. Dur-
ing this same period of apparatus development, salt preparation'techniques
were greatly advanced by the Materisls Chemistry Division. Thus, the resultis
reported are in genersl for high purity salt mixtures.

The general techniques for fused salt viscosity messurements have been

developed sufficiently to place these measurements in the category of routine.
g

_— "

The detalled discussion of these techniques and the apparastus modifications
is contained in the section of this report on experimental equipment and
procedures.

Some 36 mixtures have been studied at ORNL*. These have included
binary, ternary, and quaternary mixtures of the following fluorides:

NeF, LiF, KF, RWF, BeF,, ZrF, UF,, UF,, and TuF,.

3)
A modest program of viscosity measurement on fused fluorides will be
continued. As new mixtures are formulated and found to be of sufficient
interest, viscosity data will be supplied to the various design groups
studying fused fluoride reactors. In addition, a program to investigate
fused salt systems other than fluorides is planned. At present this plan
includes studies on chlorides, and mixtures of chlorides and fluorides.

Sonme data7 have already been obtained in the chloride system. Other sys-

tems will be studied as they become of interest.

 

In addition to the ORNL program, the Mound Laboratory is involved in an
extensive investigation of the viscosity of fluorides comprised mostly of
mixtures of the alkali fluorides with BeF2 and UF#' This work is being
carried out by B. C. Blanke using a rotational viscometer developed at

the Mound Laboratory.

o R
— -5~
EXPERIMENTAL APPARATUS AND TECHNIQUES

The experimental apparatus and techniques used in the measurement
of the viscosity of the fused fluoride mixtures are described in the
paragraphs which follow. This equipment includes the two viscometers
(capillary efflux and rotationsl), the viscometer furnace, and the dry-

box.

The Drybox. Because of the necessity for handling these salts in an
inert atmosphere, all measurements were made In a drybox under an argon
blanket. This box was constructed from 1/8-inch mild steel with the top
and front consisting completely of Plexiglas windows. The box was 50
inches high, 30 inches wide, and 24 inches deep. Long rubber gloves were
attached to Plexiglas rings located on the front at shoulder height to
ensble convenlient and comfbrtable manipulation of the equipment located
within the box. A photograph of the box is shown in Figure 1.

One side of the drybox was constructed as a door to provide access
to the inside. This door was constructed of L/h-indb aluminum to reduce
its weight and was sealed against a live rubber tubular gasket by KHUfVISE
H-200 clamps located around its periphery. Shelves and a bracket to support
the Brookfield viscometer were welded to the inside wells and a cooling coil
made from L/h~inch copper tubing was located on the wall opposite the door.
This coil was provided to reduce the temperature rise in the drybox due to
prolonged operation of the furnace, The efficiency of this coil could be
increased for high heat loads by the use of a 15 cfm blower to establish a

forced draft across the tubes.
\ _
o L UNCLASSIFIED:
T PHOTO 27741

 
  
  
 
 

     
   

  
  
 

   
      

 

    
  

P N .
et . . .
o .

T

 

 

 

=

1

Fig. 4. Photograph of Drybox Showing Furnace and Brookfield Viscometer

in Operating Position.
| -7-

A number of precautionary measures were taken during construction of
the box to insure gas tightness. Thermocouple leads were brought through
the drybox wall in Conax fittings and attached to terminal blocks on both
the inside and ocutside surfaces of the wall to eliminate strain on the
seal. Electrical leads were introduced through similar special fittings.
These fittings, as well as the fittings used for the cooling coil, gas in-
let, and gas outlet, were welded in place. The welds were heavily pasinted
with Glyptal before the finasl paint was applied to the drybox. The box
wes periodically checked for tightness with a Freon lesk detector.

A series of {ests was conducted to ascertain ﬁhe optimuﬁ procedure
for obtaining argon drybox atmosPheresg. It was found that best results
were produced by introducing argon at the bottom of the box and allﬁwing
it to exit near the top through an oil bubbler. The box was usually purged
in this manner overnight with a flow of from 4 to 8 cubic feet per hour of

argon before the test mixture was exposed to the drybox stmosphere.

The Furnace. It was necessary in carrying out this experimental program
to design a tube furnace which would bring a capsule consisting of a
1k-inch section of one-inch I.P.S. pipe filled with a fused fluoride to
red heat quickly, hold the temperature constant along the entire length
of the capsule, and attain equilibrium rapidly after adjustments in the
temperature setting. A disgram of the furnace along with the temperature
measurement and control systems ig shown in Figure 2, and a photograph of
the furnace is shéwn in Figure 1.

The furnace shell was a cylinder 16 inches high and 10 incheg in dia-

meter, mounted on casters and equipped with leveling screws which can be
THERMOCOUPLE
LOCATION

 

UNCLASSIFIED
ORNL-LR-DWG 19834

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

   

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

SALT CAPSULE BROWN
4 Cr-Al THERMOCOUPLES MULTIPOINT
HEATER ELEMENT—\ RECORDER
N 7 N\
z Y - A8
c|x % PN Cr-Al
:;: %J Z %%% 2 SIMPLITROL
@ | w Z 8/ .
Sy . 7'2\2 | RELAY
°] AN /\é
SN /\_
§ 7 . VARIAC "0 v
1 T
|y
A
7
\24 N
7
7
E \i /
N g %8 SIMPLITROL
T g ,
0 4\g
=2 é %
e
— 7 \
A
/ “ \-msuumom
[ L : 1

 

 

Y / Y -
CERAMIC BLOCK COPPER LINER

s

Fig. 2. Diagram of Furnace with Temperature Control and Measurement Systems.

_8-
pamnitl a Vo

operated from several inches above the furnace. It was insulated with semi-
circular blocks cut from slabs of JM-3000, an aluminum silicate insulating
brick manufactured by Johns-Manville.

Two pairs of 3-inch I.D. "clam-shell" resistance heaters made up the
furnace element. These two units, having léngths of lO-L/E inches and
3-1/2 inches, were located in the furnace with the short section on top
and were controlléd separately (see furnace diagrsm, Figure 2). The lower
unit heated the liquid and the upper unit served as a "guard zone" heater.
This guard heater was necessary for two reasons., First, the zone just
above the liguid surface must be kept at the temperature of the ligquid or
the resulting tempersture gradients will cause convection currents giving
slightly erroneous readings on the Br@okfiéld viscometer. Second, 1t is
essentiel that the efflux cup viscometer be held in a zone at the tempera-
ture of the liquid while it is draining or an error is obviously introduced.
It was found that two separately controlled elements provided the simplest
means of maintaeining the temperature of this zone at the liquid temperature.

Temperatures of the two elements were controlled by circuits containing
chromel-alumel thermocouples, relays, and "Simplitrols.” Variacs were also
inclﬁded in these circuits to afford more sensitive adjustment. The hot
Junctions of the thermocouples were imbedded in the outer face of a heavy
copper pipe occupying the annuius between the furnace element and the capsule
containing the sample. This copper pipe, having a wall thickness of almost
one inch, served as a thermal diffuser to provide very even temperature dis-

tribution in the sample.
L -10- y

Liquid temperatures were measured using four chromel-alumel thermo-
couples inserted in wells of 0.08-inch 0.D. thin-walled tubing which were
snapped into 0.08-inch square grooves machined longitudinally in the outer
wall of the capsule (see Figure 2). The wells were cut to such length that
the junctions of two thermocouples were located in the liquid zone and two
were in the zone above the liquid. A diagram showing the relative positions
of the four thermocouples to the two viscometers in operation may be found
in Figure 3. Temperatures measured by these four couples were read on a
Brown multipoint recorder.

Messurements were made with both the capillary and rotational instru-
ments using the same sample of salt during = single heat-up period. About
100 cc of molten salt were required to fill the capsule to sufficient depth
to operate the two instruments. A complete viscosity-temperature curve was
determined using one of the instruments before measurements were begun with
the other.

After a satisfactory inert atmosphere had been obtained in the drybox,
the furnace controls were set at a temperature 25 to 50 degrees above the
liquidus temperature of the salt and the system was allowed to come to
equilibrium. The small guantity of graphite present in the salt from the
preparation equipment rose to the surface as the salt melted. This graphite
was removed by touching the surface of the salt with a cold tamper, asllowing
the superficial layer conteining the graphite to freeze on the tamper.

The capillary cup was usually used in making the first measurements
on a new salt. To establish the entire viscosity-temperature eurve, the

température of the melt was incrementally raised and measurements were made

v —
-11-

ORNL-LR-DWG 19835

<t
—

@w@w%ﬁ

B N N N N N

 

 

 

 

 

 

 

 

 

 

 

 

B N W N N SN SN NN NN NN

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o M <
_.I + — _I

. —

_OOOOEDOOOOOOCCC/OJAH

| AN N NN N S N N RN N N m —

 

 

 

 

 

 

L= \T

 

 

 

 

 

B S ™ S NN N NN % !

4
]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OO0 000000I0 0000000000000 0S——2

 

 

i i

LR N NS N NN N NN DN Nl N

- s
- =
o
I

L N L U N N N =

 

 

 

 

 

 

   
 

 

 

 

 

 

 

 

N NN

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

COOCOO0O000l

 

 

 

EFFLUX BROOKFIELD SPINDLE

FILL

Fig. 3. Diagram Showing Relative Positions of Thermocouples and Viscometers in Furnace.
— z-

after attainment of equilibrium at each temperature step. Then, the salt
was allowed to cool to the original starting temperature, and a similar

experimentsl procedure was followed with the Brookfield viscometer,

The Capillary Viscometer. The efflux cup or capillary viscometer, origin-

 

ally designed by Tobiass, is still being used with slight modifications.
The instrument consists of a reservoir, containing the test fluid, located
above and integrally connected with a capillary tube. A photograph of the
device is shown in Figure 4, and a detailed diagrem is given in Figure 5.
The fluid drains through this capillary and the time of efflux is obtained.
By comparing the efflux time for the test fluid with the times for fluids
of known viscosity flowing through the same cup, the unknown viscosity is
established. The determination of the calibration curve will be discussed
in detail in the section on instrument calibration.

This viscometer, while it does not provide an absolute measurement,
does possess the advantages of simplicity of operation, ease of cleaning,
and low cost of fabrication. Various construction materials can be used
to provide compatibility with many classes of liquids, and the capillary
bore can be varied to cover a wide viscosity range. The principal dis-
adventage of the instrument, plugging of the capillary, has been virtually
eliminated by improvements in fuel preparation and drybox atmogphere tech-
nigques. The occasional partial plugging which does occur is easily remedied
by reaming with a fine, stiff wire,

The cup used in these studies had a reservoir of approximstely 6 cc

capacity, connected to a capillary one inch in length. Two capillary bore
" UNCL ASSIFIED
PHOTO 27740

 

 

 

 

 

 

 

 

 

Fig.4. Photograph of Efflux Cup or Capillary Viscometer and Spindle of

Brookfield Viscomefter.
14~

UNCLASSIFIED
ORNL-LR-DWG 19836

 

 

- fin. -

 

 

 

~— (0.881 in. —

 

 

OOONNNNNNNN

 

SO

 

CAPILLARY DIAMETER:
0.033 in. OR 0.039 in.

OOV SIS
OONVNNNNINNNNNNNNY

=

 

 

 

 

 

Fig. 5. Diagram of Capillary or Efflux
Cup Viscometer.
— 25-

dismeters were used, 0.0330 inch and 0.0394 inch, ILiquids with a kinematic
viscosity of one centistoke, the lowest value usually encountered in measure-
ments of fluorides, will drain from the smaller bore cup in about 45 seconds.

Measurements were made with the efflux cup using a visual procedure.

The cup was submerged in the molten salt until complete thermal equilibrium
was attained. It was then raised above the surface of the salt, held in the
heated "guard zone," and sllowed to drain (see Figure 3). The close clesr-
ance between the outside walls of the cup and the inner walls of the tube
containing the salt afforded assurance that the cup was uniformly vertical
from reading to reading.

A timer was starfed as the top edge of the reservoir cleared the sur-
face on being withdrawn from the liquid. The cup drained while being held
in the heated zone above the liquid, and the timer was stopped when the
liquid level in the cup reached the bottom of the conical section of the
reservaoir. The time of efflux was recorded, and the procedure was repeated
a number of times at the same temperature. The precision at a given tempera-
ture was within + 2 - 3 per cent. Results with greater spread were ususally
due to partial plugging of the capillary. In the event of plugging, the
capillary was reamed and the procedure was repeasted until a series of meas-~
urements of satisfactory precision was obtained.

Because of the high purity of the salts and the improved inert atmos-
pheres, ceorrosion has been slight enough té enable cleaning, recalibration

and continued use of the cups on an almost indefinite basis.
. -16- . |
' e

The Brookfield Viscometer. This instrument is a rotational viscometer
consisting of a spindle which is immersed in the liquid being studied and
a synchronous motor which rotates the spindle. The torque spplied to the
spindle by the viscous forces of the liquid produces a deflection on a
calibrated spring made of a copper-beryllium alloy. A dial graduated in
centipoises rotates with the motor and s pointer attached below the spring
indicates the deflection of the spring on this disl.

As with the capillery cup viscometer, this instrument does not pro-
vide an absolute measurement and requires frequent caelibration. However,
even though more complex than the capillary device, it possesses similar
advantages for measurements on fused salis. The spindle, thé only part of
the system which comes into contsct with the salts, is easily fabricated
from a cholice of metals. It is simple to construct, easily cleaned, and
may be reused indefinitely. The principal disadvantage with this instru-
ment results from its application to a lower viscosity range (2 to 15 cen-
tipoise) than that for which it was designed. Thus, a number of modifi-
cations of the manufacturer's design were necessary. These modifications
will be discussed in the section on instrument calibration.

For messurements with the fluorides, the spindle was connected to the
motor by a long, linked drive shaft. This isolated the delicate mechanism
of the instrument from the extreme temperatures at the top of the furnace.
A shiny metallic reflector, placed just beneath the motor housing, served
as additional thermasl protection for the motor. A photograph of the Brook-
field in operating position in the drybox is shown in Figure 1, and a photo-

graph of the spindle is presented in Figure b4,

. -
té%ﬁﬁ@?
— -

Centering of the spindle in the salt capsule, leveling of the motor,
and even temperatures throughout the entire sample of salt are all essen-
tial if asccuracy is to be obtained with the Brockfield., Poor leveling of
the motor causes bumpy, uﬁeven rotation and & resulting error. A leveling
bulb on the motor housing provided a means of avoiding this difficulty.

A uniform annulus was produced between the surface of the spindle and the
wall of the cépsule by carefully adjusting the capsule to an exactly verti-
cal position. This was done by placing e level on the machined top surface
of the capsule which extends out of the furnace and making necessary adjust-
ments with the leveling screws on the furnsce (see Figure 1). Thermal con-
vection was minimized by adjustment of the "guard zone" heater..

Errors produced by improper centering of the spindle result in high
readings. It was therefore possible to properly center the instrument by
moving the spindle within the capsule to give the lowest consistent reading.
Once this was done, only vertical adjustments were made during the course
of the experimental measurements over the complete femperature range so as
to maintain the liquid level at the notch located on the spindle shaft,

This was necessary to compensate for the expansion of the liquid with in-~
creasing temperature.

The results obtained with the Brookfield viscometer at a single temper-
ature are highly reproducible. Variation seldom exceeds + 0.1 centipoises
even though the clutch is disengaged and the spring allowed to completely
relax between readings.

The Brookfield instrument used is designated Model LVF by the manu-

facburéf@‘ It has four speeds: 60, 30, 12, and 6 rpm. Only the 30 rpm
— -18- t’:

speed was used in this program. Operation at 60 rpm resulted in turbu-

lence in the lower viscosity liquids. The spindles used were modified

to give a larger sheasr surface than those provided by the manufacturer. .

This was done by doubling the spindle length while maintaining the same

cylinder diameter.

 
. /a9 vt

INSTRUMENT CALIBRATION

The viscosities of nearly all fluoride mixtures of practical interest
fall between 2 and 15 centipoises over the pertinent temperature range.
Since the minimum full-scale reading of the Brookfield viscometer as manu-
factured was 100 centipoises, it was clear that its use in an absolute
menner would result in readings of only a few per cent of full scale and
would introduce large errors. To avold this pitfall, it was necessary to
modify the instrument to produce a higher reading at a given viscoslty and
then carefully calibrate the modified instrument in liquids of known pro-
perties. It was also necessary to calibrate the individusl capillary vis-
cometers as it was found that the capillaries were not bored with sufficient
precigion to allow theoretical determination of their characterlstics;
callbration curves were found to vary substantially between cups. The
following paragrsphs will describe the development of the calibration tech-
niques presently used.

Early Brookfield measurements were msde at 60 rpm using = spindle
identical in geometry with the lowest range spindle provided by the manu-
facturer. This spindle was cylindrical, having an 0.D. of about 3/h inch
and e length of about 2-3/& inches. The salts were contained in & capsule
made from 1-1/2 inch, Schedule 40 I.P.S, pipe. Calibration was carried out
in a series of glycerol-water solutions having the proper viscosities, and
a calibration curve of Brookfield reading versus actual viscosity was plotted
from the data. Capillary viscometers were also calibrated in these solutions

and the results were plotted in the form: kinematic viscosity in centistokes
‘ -20- y il

versus time of efflux in seconds. These solutions were easily prepared and
their viscosities were determined in Ostwald viscometers in a constant temper-
ature bath., The solutions were kept in wide-mouth jars containing s length
of 1-;/2 inch pipe to reproduce the annuwlus in the salt capsule. However,
these solutions were somewhat disadventageous in that they were subject to
property changes due to evaporation and those of lower glycerol concentration
had a tendency to grow molds. In an effort to avoid these disadvantages, &
series of puré alchols were tried in place of these solutions and used with
some success,

The results obtained on fused fluorides with the two separate instru-
ments calibrated in this way differed substantially. Deviations up to
30 per cent were found, and it was noted that the Brookfield readings in-
variably were high, A visual study was made on the flow patterns of the
liguid around the Brookfield spindle by operating it in a dilute solution
of Milling Yellow NGS, a commercially available dyestuff having double re-
fracting properties. Lines of strain and consequently, flow patterns, are
visible in this liquid when viewed with polarized light. Fortunately, solu-
tions in the concentration range exhibiting this effect also had viscosities
in the desired range. The sclution was held in a glass container having
about the same geometry as the 1~1/2 inch pipe used for the salt capsules
and in the calibrating liquids.

Turbulent cells were observed between the spindle and cépsule surfaces
when the instrument was operated at 60 rpm. Since the fluoride melts were

2 to 4 times more dense than the dye or calibrating solutions, the Reynolds
— -21- F.n

modulus in the annular region between the capsule and the spindle was 2 to 4
times higher in the salts than in the calibrating liquids. As a result,
this turbulence would be even more pronounced in the salts.

In view of these observations, a number of modifications were made in
calibration and operation procedures. The speed of rotation was reduced
to 30 rpm, and the length of the spindle was doubled to compensate for the
loss in shear resulting from this reduction in speed. Further, the salt
capsules were constructed of one-inch Schedule 40, I.P.S. pipe to reduce
the thickness of the salt annulus. The net result of these changes was an
increase in the shear, and hence iarger ingtrument readings. At the same
time the turbulent flow cells were eliminated. The effect of & reduction
in anmmulus thickness in & rotational viscometer on the critical speed, or

speed st which vortices begin to form, can be determined from the following

relation for rotational viscometers with the outer cylinder stationary9:
2 El.oglo Ut/\f' o1t * 1081y t/R) = 3.232 (1)

where

U = criticsl speed at which vortices begin to be produced

Rl = outer radius of annulus

R2 = inner radius of annulus

t = Rl - RQ = thickness of annulus

v = kinematic viscosity of liquids.

Solving equation (1) using the dimensions of the one-inch capsule and &
kinematic viscosity of one centistoke (the lowest value usually encountered

in measurements on fluorides) yielded a critical speed of 58 rpm. Thus,

W ‘
operation at 30 rpm assured leminar flow in this capsule. When equation (l)
was solved using the same kinematic viscosity and the dimensions of the
l-l/E inch capsules, a critical speed of 13 rpm was obtained. It was there-
fore likely that considersble turbulence existed in the salt in the l—L/2
inch capsules, particularly st the operating speed of 60 rpm. The equation
was solved for v using the dimensions of the large capsule and a value of
4.64 centistokes was indicated as the minimum viscosity which could be
measured with laminar flow at a spindle speed of 60 rpm. Most of the salts
studied in this program have viscosities less than 4,64 centistokes.

Since the nature of flow is a function of density as well as viscos-
ity, & search was made to find calibrating liquids having both densities
and viscosities comparable to fluorides. Two liquids, HTSlO and
s-tetrabromoethane, were found to have suitable properties., HTS, the
liguid principally used, is a fused salt mixture having the composition:
NaNOE-NaNO3-KNO3;

air, and melts at 142°C. The density of this mixture varies from about

40-7-53 weight %. It is noncorrosive, easily handled in

1.965 gm/cc at 160°C to about 1.790 gm/cc at 400°C. The viscosity varies
from about 14.5 centipoises at 160°C to about 2 centipoises at 400°¢C.
Density-temperature and viscosity-temperature curves for HTS are given in
Figure 6. Regular checks were easily and precisely made on the viscosity
of the samples of HTIS used for calibration in an Ostwald viscometer sus-
pended in a bath of HTS.

Both instruments were calibrated in this ligquid using a capsule and
furnace identical with those used with the fluorides. Measurements were

made at several temperatures, and calibration curves were plotted in a

A
)
il
ABSOLUTE AND KINEMATIC VISCOSITY

=23~

UNCLASSIFIE[}

 

ORNL-LR-DWG 19837

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.05
| DENSITY 1955
\ >
=
\ g
\
18 \ \\ 1.85 E‘JJ
\ =
16 \ 1.75
4 \\
2 \\
o \\
8 AN
ABSOLUTE VISCOSITY
\ \ (centipoises)
6 \ \
\ \\
4 \ \
— KINEMATIC VISCOSITY \ '\\
tistok
2 centistokes] \\ ~—
\
0
100 150 200 250 300 350 400
i

Fig. 6. The Density and Viscosity of "HTS".

TEMPERATURE (°C)
g _2k-

manner similar to that deseribed in conjunction with the glycerol solution
calibrations. Use of this salt mixture had the additional advantage that
it reduced the already small error resulting from the slightly different
dimensions of the instruments in the glycerol solutions st room tempera-
ture and the fluorides at high temperatures,

The other calibrating liquid, s-tetrabromoethane, is a heavy, highly
toxic, organic halogen compound having a room tempersture density of
2.964 gm/cc. The viscosity of this liquid varies from about 11.5 centi-
poises at 20°C to about 3.8 centipoises at 6500. The viscosity-temperature
curves for this liquid may be found in Figure 7. Calibration was carried
out in this liquid by containing it in one of the capsules used for the
fluorides and supporting this capsule in a constant temperature bath. This
liquid was used less frequently than HTS because of its toxicityll.

A complete calibration curve was determined for each new cup or spindle
before it was put into use. After each set of messurements on a fluoride
mixture, two or three recalibration points were made to insure that the in-
strument had not been altered during use. Representative calibration curves
for the Brookfield and capillary viscometers are shown in Figures 8 and 9,
respectively. Above a resding of about 10 centipoises, the ratio of Brook-
field reading to viscosity becomes constant at 1-1/2.

The changes made in the use of the Brookfield, coupled with the furnace
design employing two elements and the use of calibrating liquids of higher
density, have virtually eliminated the earlier poor agreement between the

two different viscometers. The results with the two instruments now agree

o
-25-

UNCLASSIFIED
ORNL-LR-DWG 19838

 

 

 

N\ KINEMATIC VISCOSITY

 

 

N

 

N

 

N

ABSOLUTE VISCOSITY (centipoises)

14

12

10

 

 

 

 

 

 

 

 

\éBSOLUTE\HSCOSHY

 

N

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\\
\\
20
DENSITY: p4 =2.964 g/cc
10 20 30 40 50 60

TEMPERATURE (°C)

Fig. 7. The Viscosity of s-Tetrabromoethane.

70

KINEMATIC VISCOSITY (centistokes)
VISCOSITY (centipoises)

14

10

UNCLASSIFIED

ORNL-LR-DWG 19839

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O ORIGINAL CALIBRATION -~
A REGALIBRATIONS
]
X ,/
a
B
° d}/
A /
P
//
/O
o
@/.
./
s~
></
-~
/U
AN
/
o
;/O ]
g |
//
4 6 8 10 12 14 16 18

BROOKFIEL.D READING {centipoises)

Fig. 8. Brookfield Viscometer Calibration Curve.

20

_93_
KINEMATIC VISCOSITY (centistokes)

7.0

6.0

5.0

4.0

3.0

2.0

1.0

UNCLASSIFIED

ORNL-LR-DWG 19840

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CUP NO. 22 (INTERMEDIATE RANGE)
CAPILLARY DIAMETER =0.039 in. /
o CUP NO. 41 (LOW RANGE)
/ CAPILLARY DIAMETER = 0.033 in.
/] _~
0O
/ _~
// G
/O /,O
/ o
/O u
1‘ //
/ 0
/ S
. (
g /C"f O ORIGINAL CALIBRATION
7 A RECALIBRATIONS
/
/o /G/ A
a
a O
/ o‘QP .
O
/T °
20 40 60 80 100 120 140 160 180

TIME OF EFFLUX (sec)

Fig. 9. Capillary Viscometer Calibration Curves.

200

—LZ_

“g
4

o

*’.)ﬁ

Ny

LA

e g

e 2
{,51““;4 a
-— -20-

to within + 10 per cent. All of the data presented in this report were
taken after this technique development program had been completed and the

changes instituted.

SECRET
e/ 20 i

EXPERTMENTAL RESULTS
{ Experimental viscosity measurements have been made on 36 fluoride
mixtures. The constituents of these mixtures were NaF, LiF, KF, RUF,
BeFé, Zth, UFH’ and ThFu.‘ Since the system of principal interest to
the ANP project at the present time is NaF-Zth~UFu, seven binaries and
ternaries from this system have been studied. Mixture 30 (NaF-Zth—UFh;
50-46-4‘moie %) is presently being plenned for use in the ART. This mi#-
ture will be formed in the reactor and the reactor made critical by the
addition of Na,UF, (mixture 43) to NaZrF5 (mixture 31).

""" Conslderable interest has been shown in mixtures in the NaF-LiF-BeFe-

UFh system., Although the proportion of BeF2 must be somewhat limited as
excessive amounts result in very high viscosities, the other properties
of many of these mixtures make them highly desirable for reactor appli-
cations. Measurements have been made on severasl mixtures at this labora-
tory, and the viscosity program of Blanke, et al, at Mound Laboratoryle
has been devoted almost entirely to an extensive and complete examination
of this systen.

Another mixture which has attracted considerable attention is the
eutectic of KF, LiF, and NaF known as Flinak. This salt has the most de-
sireble combination of physical properties, including viscosity, of any salt
studied to dete. However, its use has been limited because of asscocisted
corrosion problems. Measurements are reported on Flinsk (mixture 12) and

on two fuels formed by the addition of varying smounts of UFh to Flinek

(mixtures 1b and 107).

3w q
- -30~- § 8

In addition to the mixtures from these three systems, a number of
others from various systems have been studied as part of a continuous pro-
gram devoted to finding the coptimum fused fluoride mixture for use in re-
actor work*. Mixtures containing RbF show considerable promise from a
viscosity standpoint.

Table I presents the viscosity data on the mixtures which have been
studied to this time. Mixture numbers, compositions in mole per cent,
liguidus temperatures, and viscosities in centipoises at lOOOC intervals

are given. Equations for the viscosity in the form
o
w = 2700 (2)

are given for liquids where the experimental data when plotted as log u
versus L/T(OK) yields a straight line. The initial reference is also given.
Nearly all of these measurements have been previously reported in memos and
quarterlies; all except Nos. 117, 118, and 119 may be found in the recently
published physical properties summaryl3. This summary report supplies the
data in engineering units as well as centipoises and also provides kinematic
viscosiﬁies in CGS and engineering units.

The accuracy of these measurements is indicated by the agreement between

values obtained with the two completely different viscometers described in

the experimental section; one of these instruments involves flow through a

 

A tabulation of all the mixtures that have been formulated to date within
the ANP project may be found in reference 13. This tabulation is complete
except for mixtures 117, 118, and 119, which are given here and any compo-
sitions which may have been named since this report was written.,
Q;—

IS
o CMéggurements were made by B. C. Blanke at Mound Laboratory as well as by the authors. Results were in satisfactory asgreement.

TABLE I

THE VISCOSITIES OF FLUORIDE MIXTURES STUDIED TO DATE

 

 

 

 

Composition (Mole %) Liquidus Viscosity (Centipoises) B/T(OK)
Mixt NaF 1iF KF RoF BeF, ZrF, UF, ThF Tgup. p=Ae? * * Initial
ure  fa + 2 4 4 4 C H500 Fsoo  H700  M800 H900 A B Ref.
2 %6.5 26.0 27.5 530 17.3 9.8 6.3 4.35 0.0767 4731 15
12 11.5 #6.5 42,0 454 9,2 4,75 2.9 1.95 0.0400 4170 16
14 10.9 44.5  43.5 1.1 452 8.8 4,6 2.75 1.85 0.0348 L4265 16
20 5.0 52.0 43.0 425 10.5 6.1 .1 3.1 0.161 3171 17
30 50.0 L6.0 4.0 520 8.5 5.4 3.7 0.0981 3895 18
31 50.0 50.0 510 8.4 5.2 3.45 0.0709 4168 19
33 50.0 25.0 25.0 610 8.5 5.0 3.5 14
35 . 57.0 43.0 360 12.8 7.0 4.25 0.0346 5164 20
43 66.7 33.3 665 10.25 7.0 5.15 0.181 3927 21
Lh 53.5 4.0 6.5 540 8.5 5.7 4.2 0.194 3302 13
45 53.0 47.0 520 7.5 4.6 3.2 13
70 56.0 39.0 5.0 530 8.1 5.2 3.6 0.104 3798 22
T2 20.9 38.4 35.7 5.0 450 20.0 9.9 6.0 4.25 23
Th 69.0 31.0 505 7.5 4.9 3.45 0.118 3624 12%,20
78 56.0 16.0 28.0 478 6.0 4,0 2.85 0.111 3486 12%,2h
81 22.0 55.0 23,0 570 12.0 7.0 L.45 0.585 LehT 25
82 20,0 55.0 21.0 4.0 545 12.0 7.0 L.h45 0.0585  L64kT 25
86 32.0 35.0 29.0 4.0 Lus 20.5 10.5 6.45 L4.55 26
87 48,0 48.0 4,0 ko5 7.1 L.65 3.3 0.116 3590 27
88 64.0 5.0 31.0 555 7.1 4,75 3.4 0.138 3435 12%,28
R0 49,0 15.0 36.0 555 8.0 5.0 3.4 0.0811 4008 29
g5 50.0 46.0 4.0 500 7.05 .35 2.95 0.0657 hLoB1 27
96 53.0 2.4.0 23.0 535 5.9 4.1 3.0 0.157 3168 29
97 49.0 36.0 15.0 597 5.65 3.95 2.95 0.173 3043 30
o8 56.0 21,0 20.0 3.0 505 7.3 4.6 3.1 0.0737 Lol2 30

 

Average values are listed.

_IE-
TABLE I (Continued)
THE VISCOSITIES OF FLUORIDE MIXTURES STUDIED TO DATE

Composition (Mole %) Liguidus Viscosity (Centipoises)

 

 

 

B/T(°K) .

. Temp., = Ae Initial

Mixture NaF LiF KF RbF BeF, IrF), UFlL ThF4 Sq f500 Moo Mooo  H8oo Ho00 E;"ET'"“'“ii"‘ ot
'ioo 40.0 60.0 652 3.2 2.35 0.116 3225 1k
104 43,0 57.0 475 9.0 4,5 0.0212 4678 27
107 11.2 45,3  L41.0 2.5 490 5.1 3.0 0.0292 4507 16
111 71.0 16.0 1.0  12.0 13.0 7.1 4.8 0.0620 4666 13
113 50.0 50,0 380 15.3 8.4 5.25 0.0493 5009 31
114 50.0 50.0 Lhs 15.3 6.7 3.45 0.00517 6976 31
115 50.0 50.0C 400 11.5 5,2 2.75 0.00534% 6701 13
116 79.0 21.0 730 2.2 0.077C0 3600 29
117 5.0 54,5 34.5 6.0 545 8.0 4,8 3.3 32
118 Wy,5 15.0 34.5 6.0 540 7.5 .75 3.5 32
119 37.5 9,5 Y7.0 6.0 540 9.5 5.2 3.5 32

—BE-

 
 

capillary and the other is concerned with shear in an annulus with the inner
cylinder rotating. An error analysis indicates that the results are within
+ 10 per cent of the true value.

Measurements were made on all of the salts reported here except those
containing BeF2 with both of the instruments previously described. Salts
containing BéF2 were studied with only capillary viscometers in a separate
drybox facility built for this purpose from & 55-gallon drum. Two or three
capillary viscometers were used in studying each beryllium salt mixture to

furnish a check. The agreement between cups was within + 10 per cent.

-
Ex
— "3 8 . if

CORRELATIONS

One of the purposes of any experimental program is the development of
correlations based on the experimentally determined data. These correlations
describe the behavior of the materials studied and, if possible, serve as s
basis for prediction of the properties of similar materials. Several such
attempts have been made to correlate the viscosity data obtained in this
program, and these attempts have not been without some degree of success.
However, because of the complex nature of the phenomenon of fluid viscosity,
these correlations cannot be depended on for predicting the viscosities of
mixtures far from the measured compositions. They do serve to illustrate
the various trends observed in the data.

Before describing the correlations which have been developed, the ob-
served trends in the data will be discussed. In general, the viscosities
of the fluorides are found to be a function of the fluid density and molec-
uwlar weight. This reletionship is as predicted since most classes of liquids
behave in this manner. However, for some mixtures of fluorides this general
correlation does not hold. For example, mixtures containing appreciable

amounts of BeF_,, the least dense of the fluorides studied, exhibit very high

07
viscosities. Mixtures containing BeF'2 have much higher viscosities than cor-
responding BeFe—free mixtures, and the introduction of smounts of BeF2 in
excess of 30 - 40 mole per cent results in very sharp increases in viscosity.
Blanke33 obtained a reading of about 900 centipoises on HaF-BeF2 (21.9-78.1
mole %)_at 600°C. The fact that pure BeF2 is a glass provides a partial ex-

planation of the high viscosities of BeFE-containimg mixtures.

. . . 1 g 2
o
oA e R
e e
- 3

Another pronounced trend which has been noted involves the relative

 

effect of each alkali fluoride on the viscosity of a mixture. It was found
that the viscosity of a fluoride mixture will decrease as the lower molecular
weight alkall-metal fluorides are replaced by those of higher molecular weight.
Thus, the viscosity of a RbF-containing mixture will be considerably lower
than that of a prototype mixture containing LiF in place of the RbLF.

These treﬁds can be seen in Figure 10 in which the viscosity at TOOOC
of each salt is plotted as a function of its room temperature density. The
room temperature density values were calculated from the mixture composition

using the equation:

(3)

where:

density at room temperature of the mixture (gm/cc)

Mi = molecular weight of a component of the mixture
fi = mole fraction of that component
p; = density at room temperature of that component (gm/cc).

Values obtained from this relation represent well the room temperature experi-
mental values. This equation was used in the correlation for predicting the

34

liguid densities of the fluoride~ mixtures.
An sverage line drawn through the non-BeF2 points on Figure 10 indicates

the trend of the relationship between density and viscosity. The two extremes,

——
VISCOSITY (centipoises)

-36-

  

AN e L
A
!ﬁ R “3"!..

ORNL-LR-DWG 19841

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50
® NON-BeF, MIXTURES
A Bef,—BEARING MIXTURES
20
+30%
)
.
113 72__%
t33 g1, M = 23
aa * 52" 5 - 7307
"s ,/ 44 20 //' °
5|4 P A 7 SeePe 7 4
96097~ _& 45 /“°>?<\8770
800 " 20 / [ Tes
— - ~— /[ 7
1o TIOT 0 T4
14 —
/
2
1
2.0 2.5 3.0 35 4.0 45 5.0 5.5

CALCULATED ROOM TEMPERATURE DENSITY (g/cc)

s Fig#10. Viscosity at 700°C vs. Calculated Room Temperature
Density.
_ 37 b

mixtures 12 and 43, are the lightest and densest salts, respectively, that
are likely to be encountered in non—BeF2 fluoride mixtures. It is seen that,
with the exception of two mixtures, the data fall within + 30 per cent of
this average line. Figure 10 also.shows that all of the mixtures containing
BeF2 are above the + 30 per cent line and that the viscosities of the BeF2
salts vary as their BeF2 content over the concentration range (15 to 50 mole
per cent) studied.

The relstive effect of the various alkali fluorides on the viscosity is
indicated in Figure 10. Mixtures 81, 82, 86, and T2, vwhich are considerably
above the average line, are zirconium-base fuels containing LiF as the prin-
cipal alkali fluoride. Salts 87, 95, and 117, which fall somewhat below the
average line, are zirconium-base fuels containing rubidium as the alkali flu-
oride. The mixtures of intermediate viscosity, suéh as 30, 44, and 70, are
similar mixtures conteining NaF.

All of téeae trends are substantiated by a correlsation resulting from
g8 statistical treatment of the data. This approach was based on & number of

empirical equations cited by Hatgchek35

relating the viscosities of mixtures
to their compositions. The simplest assumption is that the viscosity of an

ideal mixture should be additive. Thus,
bo=pX+ o, (1-x) (4)

where pu is the viscosity of the mixture, and u, are the viscosities of the

Hy

components, and x is the mole fraction of one of the two components. Bingham

stated that the reciprocal of viscosity, the fluidity (@), was additive. Thus,

¢ =0.x+ 0, (1-x) (5)

. a5 U
ot -38- b

Arrhenius proposed the purely empirical function,

log = x log p, + (1 - x) log Mo (6)
Kendall and Monroe found for some mixtures of nonassociated liquids that

1

—

+ (1 - x) uz | (7)

[~

WYo= X p

L
Wik

From these empirical equations and the observed data trends, a corre-
lation based on equation (%), assuming the viscosities of the fluorides to
be approximately additive, was indicated. The viscosity of the mixture was
represented by the equation

b= up Ty +oppfp + e+ f (8)
where K represents the viscosity of pure component A, fA is the mole frac-
tion of that component, and so forth, for the eight components. A matrix
containing as its elements the mole fractions of the components in all the
mixtures studied along with the viscosities of the mixtures at 600°C, 700°C,
and 800°C was programmed for the ORACLE, Solutions were obtained at»the
three temperstures for the viscosities of the pure components. It must be
understood that these "pure component viscosities” are merely empirical rep-
resentations for use in this relationship since the pure components are not
liquids at these temperatures. Solutions wére also obtainéd at the three
temperatures using the same matrix along with the viscosities expressed as
log p and as fluidity, or 1/u, as in equations (5) and (6). Compsrison of
experimental and calculated viscosities using equations (4), (5), and (6)

indicated that the three correlations were of about equal merit.
‘ -39- N

However, the degree of correlation was only feir. Viscosities calcu-
lated by this correlation for four salts at TOOOC differed from the experi-
mental values by more than 20 per cent. The average deviation between
experimental and calculated viscosities at 70000 for all of the mixtures
studied was approximstely 10 per cent. It must be concluded, therefore,
that this correlation would not be a safe basis for predicting the viscos-
ities of new mixtures. The "pure component viscosities" obtained from the
ORACLE for the equation involving linear additivity, the simplest of the
three functions tested, are found in Table II. A comparison of the experi-
nmental and calculated viscosities at TOOOC based on this function is given
in Table III.

Another correlation which may be applied to the experimental results
involves the relastionship between fluidity and the number of "holes” in a
liguid. If holes must be present in a liquid before flow can take place,

36

as has been postulated by‘Eyring , it is reasonable to assume that the
fluidity of a liquid will be.proportional to the number of holes. In addi-
tion, the essential difference between a solid and a liquid may be regarded
as the introduction of holes. If V is the molér volume cf the liquid and
V, is the molar volume of the unexpended solid, the difference, (V - V_),
is proportional tc the number of holes and, hence, to the fluidity. Since
the fluidity is the reciprocal of the coefficient of viscosity, it follows

that
(9)

 

HEV Y
8

™
where ¢ 1s a constant. This equation was originally proposed by Batschinaki3‘

and w&s fqud to hold for a large number of nonassocisted liquids.
 

 

[ -h0- ¢ i
TABLE II
"PURE COMPONENT VISCOSITIES" BASED ON EQUATION (8)
"Pure Component Viscosity" (Centipoises)

Component 600°¢ 700°¢ 800°¢
NeF (A) - 1.33319 0.414287 0.495901
LiF (B) 8.58465 5.27885 3.38107
KF (c) 1.03168 0.367723 0.332103
RbF (D) - 2.90961 - 1.43543 - 1.93418
BeF, (E) 27.3054 13.4185 8.10863
ZrF), (F) 15.1988 9.09101 6.54347
UF), (@) 60.9819 31.2966 20.5855
ThF), (B) 16.0516 7.43398 7.46834

 
 

TABLE III

COMPARTSON OF EXPERIMENTAL VISCOSITIES AT TOOOC

AND PREDICTED VALUES BASED ON EQUATION (8)

 

 

Viscogity at TOOOC, Centipoises Per Cent
Experimental Predicted Deviation
9.8 8.89 9%
2.9 2.66 8%
2.75 2.90 5%
.1 4,12 0
5.4 5.6k 44,
5.2 k.75 %
8.5 10.30 21%
7.0 6.01 14%
10.25 10.70 4%
5.7 5,89 3%
4.6 b 49 2%
5,2 5.3h 3%
6.0 6.92 15%
4,9 7.80 59%
4,0 4,83 21%
7.0 5.09 27%
7.0 6.15 12%
6.45 5.87 9%
4.65 4,93 6%
4,75 4,69 1%
5.0 5.09 2%
k.35 b, 71 8%
b1 4.57 11%
3.95 4,12 49,
4.6 k. o6 8%

 

Per cent deviation is defined as (difference/experimental) x 100.

 
 

 

“hoo ‘;m:f““
TABLE III (Continued)
COMPARISON OF EXPERIMENTAL VISCOSITIES AT 700°C
AND PREDICTED VALUES BASED ON EQUATION (8)
Viscosity at TOOOC, Centipoises Per Cent
Mixture Experimental Predicted Deviation#

100 3.2 3.33 4%
107 3.0 3.37 12%
111 7.1 7.10 0

113 8.4 6.92 18%
114 6.7 6.89 3%
115 5.2 5.99 15%
117 4.8 h.25 13%
118 b.75 4,98 5%
119 5.2 6.17 19%

 

*
Per cent deviation is defined as (difference/éxperimental) x 100.

e
Wt -43-

This relationship was applied to the experimental data on fluorides
reported here, and the resulis are shown graphically in Figure 11. The

term p in the equation is the viscosity at 700°C, V is the molar volume

 

or reciprocal of \§iliUMsity at 700°C, and V, is the molar volume or re-
ciprocal of the density at room temperature calculated from equation (3).
Experimental values are used for the liquid density where available; cal-
culated values are used in the other instances.

Although the points shown on the figure exhibit considerable scatter,
the hyperbols described by the equation éan be discerned. In contrast to
the density-viscosity trend shown in Figure 10, this relationship appears
to correlate the data for liquids containing BeF2 as well as for the
non—BeF2 mixtures. However, the BeFa—bearing ligquids studied have contain-
ed only moderate proportions of this constituent. Calculations using some
of the data taken at Mound Laborstory on liquids containing enough BeF2 to
sharply increase the viscesityl2 result in greatly increased scatter.

Some of the scatter shown in Figure 11 is probably due to the error
limits to which the experimental and predicted date are subject. However,
it is safe to assume that most of the divergence encountered with this
correlation, as well as with the others described, is due to factors such

as the general complexity of the viscosity phenomenon and the wide varia-

tion in the physical nature of the fluoride mixtures.
ORNL - LR—DW! 19842

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

042
0.10 0 LO
o
$
0.08 H—@
N 'F % 0 E o
| ° .8 o ®
o O%Oo O
o 0.06 5
o
i\-
o
0.04 f ©
o NON-BeF, SALTS o
0.02 ® BeF, —BEARING SALTS
O
0 2 4 o 8 10 12

VISCOSITY AT 700°C (centipoises)

Fig. 11. Viscosity at 700°C vs. (VTOOOC - Vs).
] ~45- A

10O.
11.

3.2.

130
1k,

15,
16.

tggﬁﬁgl

REFERENCES

Knox, F. A., and Kbrtesz, F., ANP Quarterly Progress Report for Period
Ending September 10, 1951, ORNL-1154, p. 136.

 

Redmond, R. F., ANP Quarterly Progress Report for Period Ending June 10,
1952, ORNL-129}+, Po 12}4‘-

 

Cisar, J. M., ibid.

Tobias, M., ANP Quarterly Progress Report for Period Ending December 10,
1951, ORNL -1170, p. 12k.

 

Knox, ¥. A., and Kertesz, F., ANP Quarterly Progress Report for Period
Ending September 10, 1952, ORNL-1375, p. 145.

 

Redmond, R. F., and Kaplan, 8. I., "Remarks on the Falling Ball
Viscometer,"” ORNL CF 53-1-248, January 1%, 1953.

Cohen, 8. I., ANP Quarterly Progress Report for Period Ending
December 31, 1956, ORNL-2221, p. 228.

 

 

Cohen, S. I., and Peele, J. M., "Determination of the Optimum Procedure
for Obtaining Inert Atmospheres in Non-Vacuum Dryboxes,"
ORNL CF 55-10-132, October 26, 1955.

"Modern Developments in Fluid Dynsmics," Vol. II, pp. 388-390, Oxford
Press, Oxford, (1938).

Kirst, W. E., Nagle, W. M., Castner, J. B., "A New Heat Transfer Medium
for High Temperatures," Trans. AIChE, 36, p. 371, (1940).

Sax, N. Irving, "Handbook of Dangerous Materials," p. 5, Reinhold,
New York, (1951).

Blanke, B. C., MIM CF 55-11-14%, November 7, 1955.

Cohen, S. I., Powers, W. D., and Greene, N. D., "A Physical Property
Sumuary for ANP Fluoride Mixtures,"” ORNL-2150, August 23, 1956.

Cohen, S. I., ANP Quarterly Progress Report for Period Ending
December 10, 1955, ORNL-2012, p. 100,

 

 

Cohen, S. I., and Jones, T. N., ORNL CF 55-%-32, April 1, 1955.

Cohen, S. I., and Jones, T. N., ORNL CF 56-5-33, May 9, 1956.
‘ - 1 | -46- & t‘.

REFERENCES (Continued)

17. Cohen, S. I., and Jones, T. N., ORNL CF 55-2-20, February 2, 1955.

18, Cohen, S. I., and Jones, T. N., ORNL CF 56-4-148, April 17, 1956.

19, Cohen, S. I., and Jones, T. N., ORNL CF 55-12-128, December 23, 1955.

20. Cohen, S. I., and Jones, T. N., ORNL CF 55-2-89, February 15, 1955.

21. Cohen, 8. I., and Jones, T. N., ORNL CF 55-3-137, March 16, 1955.

22. Cohen, S. I., and Jones, T. N., ORNL CF 55-9-31, September 6, 1955.

23. Cohen, S. I., and Jones, T. N., ORNL CF 55-3-61, March 8, 1955.

24, Cohen, S. I., and Jones, T. N., ORNL CF 55-5-59, May 16, 1955.

25. Cohen, S. I., and Jones, T. N., ORNL CF 55-5-58, May 16, 1955.

26. Cohen, S. I., and Jones, T. N., ORNL CF 55-7-33, July 7, 1955.

27. Cohen, S. I., and Jones, T. N., ORNL CF 55-11-27, November 4, 1955.

28. Cohen, S. I., and Jones, T. N., ORNL CF 55-8-21, August 1, 1955.

29. Cohen, S. I., and Jones, T. N., ORNL CF 55-11-28, November 8, 1955.

30. Cohen, S. I., and Jones, T. N., ORNL CF 55-12-127, December 23, 1955.

31. Cohen, S. I., and Jones, T. N., ORNL CF 55-8-22, August 1, 1955.

32. Cohen, S. I., and Jones, T. N., Unpublished data.

33. Blanke, B. C., MIM CF 55-10-34, October 24, 1955.

34. Cohen, S. I., and Jones, T. N., "A Summary of Density Measurements on
Molten Fluoride Mixtures and a Correlation for Predicting Densities
of Fluoride Mixtures," ORNL-1702, July 19, 195k,

35. Hatschek, Emil, "The Viscosity of Liquids,” pp. 135-139, D. Van Nostrand,
London, (1928).

36. Glasstone, 8., Laidler, K. J., and Eyring, H., "The Theory of Rate
Processes, " p. 481, McGraw-Hill, New York, (194l1).

37. Batschinski, A. J., Z. physik Chem., 84, p. 643, (1913).

g *
 
