Vo!' 58, No.

5, May 1957

Journal of Research of the National Bureau of Sta nda rds

Resea rch Paper 2757

Vessels for the Storage and Transport of Liquid Hydrogen

B. W . Birmingham, E. H. Brown, "_C . R. Class, 2 and A. F. Schmidt

The d esign a nd co nstru ction of a lar ge-scale hydr ogen liqu efi er by t he N a tion al Bureau
of Standards r eq uired a concurre n t developmen t of large, transpor table liq uid-hydrogen
con tainers. These co n tainers, known as D ewar vessels after t he in ven tor, J a mes D ewar,
ha d to be capa ble of road a nd air t ranspor t, whi ch implied ruggedness a nd ligh t weig ht .
This ar t icle reports o n lar ge-scale liquid-h ydrogen D ewar development condu cted by
t he N BS Cr yogenic E ngineerin g L a bor atory. A 440-li ter a nd a n 840-li ter D ewar a re
di scussed. E ver y effort has been mad e to design D ewars wi t h ver y low hydrogen-l oss rates
within t he limi tations of t he t ransp or t r equire men t .

The op timum D ewar configuration is spherical,

because the sphere has the smalles t area-to-volume
ratio of any shap e. I n practice, however, such
t heoretical configurations must b e temper ed by the
commercial availability of v ariously shaped m etal
heads. The sh ape of each of the N BS D ewar s [1] 3
was govern ed by different considerations. The 440liter D ewar was designed around r eadily available
commercial h eads. The 840-liter D ewar was design ed to fi t a specific space and r equired special
tooling for h ead fabrication. Figures 1 and 2 show
D ewar cross sections ; figures 3 and 4 show the compItted D ewars.

container and intercepts all ambien t temperat ure

radiation from th e outer shell. Thermal radiation
to t he liquid-hy drogen container is th en r educed by
a factor of at least 200 and b ecomes of the same order
of magnitude as the r esidual gas conduction and the
solid conduction .
Wh en considering hea t t ransfer to the ni trogencooled shield, r esidual gas condu ction for a vac uum
of th e order of 0.01-}.L H g is relatively unimportan t
compared with thermal r adiation . A low h eat leak
to th e nitrogen-cooled shield th er efore r equires
highJy r efl ectin g surfaces. Tl1Toughou t the N BS
D ewar developmen t program high insulating va,cuums and surfaces of high r efl ectivi ty were emphasized.

2. Insulation Methods

3 . Materials of Construction

Only two m ethods of D ewar insulation appeared

feasible a t the time of the design : the use of powder
insulation, su ch as silica aerogel, at a moder ate
vacuum, and th e use of an unfilled sp ace at a high
vacuum . Although the moder a te vacuum associa ted
with powder insulation brings m any advant agessuch as r educing th e impor tance of ver y minute
leaks and making th e achievem en t of a static vacuum
easier- the insula tion value when used in moder ate
thickn ess does not appear to b e as gr ea t as that of
an unfilled space a t a high vacuum (up to abou t
0.01-}.L H g pressure) with surfaces having a high
refl ectivity (low emissivity). As one of the primary
goals was the achievem ent of a low hy drogen loss
withou t making the D ewar too bulky, high va cuum
r ather than evacuated powder was chosen for t he
D ewar insula tion.
With high-vacuum insulation, h eat energy flows
to th e inner surfaces by three m ech anisms : th ermal
radiation, r esidual gas conduction, and solid conduction through piping and insulating suppor ts.
Withou t any shielding, thermal r adiation from
room temper ature directly to the liquid-hydrogen
con tainer would b e excessive and would m ak e it
difficul t to develop a low-loss hydrogen D ewar. I n
order to r edu ce this thermal r adiation, a liquidni trogen-cooled r adiation shield was planned. This
shi eld completely surrounds the liquid-hydrogen

Several factors must b e consider ed in varyin g degr ees in r egard to th e m aterials of construction of
D ewar vessels. :M ech anical proper ties, t hermal conductivity, surface reflectivity, ou t-gassin g r ates, and
diffusion proper ties ar e some of the more impor tan t
factors. For equipmen t designed for transport,
anoth er factor- weigh t- is also of consid erable impOltan ce.
Among th e various mechanical proper ties of metals,
brit tleness or no tch sensitivity is one of the most impor tant in cryogenic work, b ecause m any m etalsno tably the ferritic steels- b ecome ver y bri ttle just
below room temperature. Some of t ho common engineering m etals whose impact strengths (as measured
by Charpy or Izod no tch tests) are satisfactory down
to liquid-hy drogen tempera tures, ar e austenitic steels
(such as the stainless t yp es, 304, 316, 321 , and 347),
copper , many copper alloys, and aluminum.
For piping, good strength and low t hermal condu ctivity are necessary. Of the more common materials,
stainless steel and :M onel h av e th ese characteristics.
Of these two, Monel no t only has a thermal conductivi ty approximately twice that of stainless steel, but
it is also more expensive. Thus, in t he absence of any
other compelling factors, stainless steel is ordinarily
used for all internal piping.
The choice of shell materials is more difficul t b ecause a gr ea ter number of factors are involved. F or
low weigh t, aluminum is indica ted. For high r eflectivi ty, both copper and aluminum are good, bu t
aluminum probably main tains its r eflectivity b etter

1. Dewar Configuration

1 Work SUI)ported by the U. S. A tomie E nergy Commission an d t he U. S. Air

F orce.
, Now with Beech Aircraft COrl).
a J<' igures in brackets ind icate the li terature references at t he end of t his paper.

243

7 LlQ. N VENT
MULTIPLE - CONTACT
8 INSULATING SUPPORTS
9 RADIATION SHIELDS (2
TO ST. STEEL
x ALUMINUM
SOLDERED CONNECTIONS

DESCRIPTION

No.

LIQUID Hz VALVE
Hz RELIEF VALVE
Nz RELIEF VALVE
LlQ. Hz FILL a EMPTY
Hz VENT LINE
LlQ. Nz FILL a EMPTY

10

I
I

- --- --4
VACUUM

440 LITERS
LIQUID HYDROGEN

60

_ ..r

Lc-------37--~J

--------------------- 42 ----------------~

-------------------------54--------------------------F IGUR E

1.

440-1i ter liquid hydTogen Dewar.

244

4
5

PART
DESCR I PTI ON
NO.
I
L10U I D H2 VALV E
2
Hz RELI EF VALVE
3

6
7

I..-~

H2 J -T VALVE

9
10
II
12

LlOUID H2 FILL a EMPTY

H2 VENT a L . P. RETURN
H2 H. P. INLET
L .P. H. REFRIG . CONNECTION
H. P.

"

"

13

"

LIQ UI D N2 VALV E
N. RELI EF VAL V E
LlOUID N2 FI LL a EMPTY
N. VENT
MULTIPLE -CONTACT INSULATING SUPPORTS

14
15
16
17
18

J-T HEAT EXCHANGERJ( FINNED CONCENTRIC_

H. T .
"
"
I
TUBING)
N2
"
J-T VALVE CONTROL
SUSPENSION SYSTEM FOR AIRCRAFT LOADING

~II

BID
LIOUI D

...

LITERS

B40

N I TROGEN

LIQUID

'I

LITERS
HYDROGEN

._ ,

CJ1

- --

----52"

C_

Cif/l'L "'- -'\."\ \

'\--.
(I s)

_~I
_~

VACUUM

P7lll111

I, I, -\

\... - ~ (1 6)
~

~ IIiIIIJI77l

\.:.

\.... \

(1 4)

(13)

VACUUM

-/'l
:::n
95"

FIC U RE 2.

8 SUCH INSULATING SUPPORTS

EO UALL Y SPACED ON CI RCUMFERENCE

- - - - - - -- - - -- - - - - - - - -- -1

840-liter liquid hyd rogen Dewar.

FIGUR E

3.

440-liter liquid hydrogen Dewar.

A, Liquid-level gages (N, and H ,); B , pressure gages (N,

and IT,) ; c , weight-loaded relief valve; D, insulating vacuum
gage; E , aux.i1iary vacuum pump coru1cction ; F , blowers;
OJ shielded air-cooled diffusion pump ; H , mechanical vacuum pump ; I , liquid Nz fill-aDd-discharge val ve; J, electrica l
system switch panel and vacuum gage control unit; K ,
liquid H , d ischarge v al ve; L , liquid H, fill valve; :111:, relief
valves (H , and N,).

F I GURE

4.

840-1iter liquid hydrogen Dewar.

A , J-T v alYc control ; D , H . P . H 2 refri gerator connection ; C , H . P . H 2 line filter; D , L . P . H 2 refrigerator cOllnection ; E , N 2 relief valve; F, H 2 relief vah'e; G ~
liqu id H , fill a nd discharge ml ve; H, liquid N, fill and discharge valve; I , liquid level a nd pressure gages (IT, and N ,); J, pressure gages for monitoring refrigera tion
process; K , refrigerator heat exch anger warmup connections; L , suspension system for aircraft loading; 1\1 , vacuum pumping system enclosure.

246

-than copper. It is po sible to use stainless steei

covered with aluminum foil to obtain a high reflect ivity. For high th ermal conductivity, which is desirabl e in some places, either copper or aluminum can
bc used. Finally, because the solubility of hydrogen
in aluminum is much less t han that in stainless
steel [2], aluminum should be a very good material
with respect to hydrogen diffu sion.
For th e ou tel' shell, where low weight and high reflectivity are important, aluminum would appear to
be the ideal material. For the liquid-nitrogen-cooled
radiation shield, low weigbt, high reflectivity, and
high thermal conductivity are desirable. 'rh e latter
factor- high t hermal conductivity- i s importan t b ecause the liquid nitrogen is in contact with only
a restricted area of the shield, which means that thermal energy reaching the remainder of the shield must
flow to the liquid nitrogen b.v solid conduction. This
co.mbination of requirements again indicates alulTIlnum.
For the liquid-h ydrogen container, high reflectivity, low hydroge n d iffu sion, a nd, agai n, high thermal conductivity are the impor ta nt factors. Liquid
hydrogen when kep t in a shell of r ela tively Jow co nductivity, i. e., sLainless steel, suppor ts a temperature
gradient that is undesirable when t he vessel is scaled
to allow pressme buildup. The top la:ver of the liquid
ma:\T be as much as 1 to 1 7~ deg C warmer t ha n t he
bottom layer, a nd the vapor pressure in t he vapor
space above tbe liquid is correspondingly bigh. This
pressure buildup is undesirable in a scaled-off container because it causes t he safeLy valves to relieve
t he vessel prematurely, and in a refrigerated vessel
causes the refrigerator to cycle more frequently. In
stainless 01' other poorly co nducting sh ells, t his str atification ca n be overcome by thermal equilibratorsyertical copper or aluminum sheets 01' rods which
conduct heat from the top to t he bottom of th e liquid- but it is simpler, if possible, to accomplish this
by usc of a good t hermal-co ndu ctivity material in
the shell itself, e. g., aluminum or copper.
Because of the above combination of reasons it was
decided to fabricate most of th e D ewar sh ells from
aluminum. The particu lar type of aluminum chosen
was the weldable alloy 52S, which in liquid oxygen
equipment had been found to be even less suscep tible
than 61S to cracking in the welds at low temperatures.
Although aluminum had been used exte nsively in
liquid-oxygen plants and D ewars, as mentioned by
Gillette [3], this appears to be the first time its use was
considered both for liquid-hydrogen temperatures
and for high-vacuum vessels. Tb e solution of th e one
foreseeable pro blem that arose b ecause of this choice,
transition joints between aluminum shells and stainless steel tubin g, is discussed below.

4. Insulating Supports
Acceleration, shock , and vibration conditions can
b e quite severe in equipment subj ect to road and air
t ransport. For instance, the following m echanical
se rvice condit ions were imposed as 840-liter D ewar
design objecLives :

(1) Acceleration:

(a) Vertical acceler ations from -3 g to + 5 g, ineluding acceleration of gravity.

(b) Horizontal accelerations from -4 g to + 4 g.
Accelerations arc continuously applied for periods
not exceeding 1 sec.
(2) SIwek:
Shocks of 5 g applied in either direction along three
mutually p erp endicular axes, wi th each shock being
{mm 10 to 15 msec dlll'ation .
(3) Vibration:
(a) 0.05-in. double amplitude (total exc ursion)
from 5 to 10 cps.
(b) 0.036-in . double amplitude from 10 to 75 cp s.
Vibration continuous in anyone or all of three
mu tually p erpendicular planes.
Thus it is r eadily seen th at a D ewar capable of
road and air tr ansport requires a sturdy suspension
system (i. e., a system for supportin g one sh ell from
a nother) . This sllspension s:\'stem mu st also be a
good t hermal insulator in order to minimize cond uction h eat leak to th e cold shells. From a mechanical
v iewpoint it is des irable to h ave the suspe nsion
system rigid to eliminate the possibility of internal
r esonances th at co uld be ca used b~7 vibration dUl'ing
t ransport, a nd from an assembly viewpoin t the
sLlspension system should be simple.
Fo r th e suspension system used on th e NBS
Dewars, there appeared to be a choi ce between
either stainless-steel cables or rods , 0 1' multi pleco ntact thermal insul ators which h ave been developed
by thc NB S C ryogenic Engineering Laboratory.
Stai nIess-steel cables have been emplo.\rrcl ve ry SLlecessf ully, from a thermal point of view, in liquidhydrogen tra nspor t Dewars. However, th eir use h as
r es ulted in two major difficulties . First, getting the
proper distribution of tension on all support cables
causes a very diffi cult assembl.\' problem . Second,
experience h as shown th a,t s uch cables may bc
str ctch ed considerably under repeated loading and
thermal cycling. This stretching lowers t he natUl'al
frequency of the sLlspension s.,' stem and makes it
difficult for t he D ewar to m eet acceleration and shock
specifications.
Some r em arks on th e development of multiplecontact thermal insulators b~' the Bureau might be
of interest. During various experiments it h as been
observed that difficulty is experienced in transferring
heat across m etal surfaces clamped together in
supposedly good contact in a vacuum. The surfaces
may appear quite smooth and yet, due to microscopi c
irreg ularities, may present only a small area of m etalto-metal contact; thus, the net area for solid conduction is quite small . This area depends on load and
metal hardn ess. vVhen one multiplies this effect by
lamination, rigid memb ers that arc also good thermal
insulators can be r eadilv devised . Preliminary
experiments by the NBS Cryogenic Engineering
Laboratory h ave indicated that the apparent thermal
conductivity of a laminated stainless steel or Mon~l
insulating support under high unit loads (1,000 pSI)
would be about one-fiftieth that of the metal itself [4] .
Fmther experiments [5] indicate that, by proper

247

STACK ED DISCS

STACKED WASHERS

NOTE:
ARROWS INDICATE
DIRECTION OF
APPLIED LOAD.

COILED STRIP

FIGURE

5.

Insulating su pports.

design (dusted surfaces and moderate unit lo ads) ,

ratios as low as one two-hundredth of the base metal
can be achieved.
The insulatin g supports developed for the NBS
Dewars were rigid laminae formed from thin stainless
steel (0.002 to 0.005 in. thick). Washers and disks
were used to support axial lo ads, and tightly wound
coils were used to support lo ads that were radial (with
respect to the coil) (see fig. 5).

5. Joining Aluminum to Stainless Steel

To use metallic tubing of low thermal conductivity
such as stainless steel in Dewars of aluminum-allo
shell construction, it is necessary to make aluminumto-stainless-steel tubing joints t hat will remain
vacuum-tight over a possible range of + 130 to
- 420 F. Because the coefficient of thermal
expansion of aluminum and most of the common
aluminum alloys is approximately 50 percent greater
than that of stainless steel, the joint must be sufficiently ductile to withstand all stresses imposed
upon it by thermal expansion and contraction. In
addition, the joint must be designed to accommodate
an amount of vibration and shock loading commensurate with specified service requirements. Finally,
the simplicity and ease of actually making the joint is
a desirable and almost essential feature , as some form
o.f field repair work may prove necessary upon occaSlOn.

Four basic methods of joining aluminum to stainless steel have b een su ccessfully used, each of which
employs soft solder as the filler material. (For a
survey of methods of joining aluminum to stainless

steel, see [6]. ) In all cases, the aluminum and stainless-steel tubing were initially of the same nominal
size; however, prior to the joining operation , the
aluminum tube was swaged out at one end for a
length of approximatel.'- 1 in. so that it would fit
easily over the outside diameter of the stainless-steel
tube, thus forming a tubl1 1ar lap joint (see fig. 6).
In an aluminum-to-stainlcss-steel con nection such
as is described here, it is of d efinite value to utilize
the ahuninum as the outer member, b ecause in cooling from the soldering temperature to room temperature, and event ually fronl. room temperature to
- 420 F, the joint tends to close up and become
even tighter than when fabl'icated , due to th e greater
contraction taking place in the aluminum as compared with the stainless steel.
In brief, the four joining methods are as follows:
(1) To eliminate the usual amount of difficulty
involved in effectively tinning aluminum, the portion
of aluminum to be used is first copperplated in an
approved manner, thus providing an excellent soft
solder base. Subsequent tinning of the copperplated
area and the stainless-steel member provides the
basis for a satisfactory joint. (2) An alternative
method to that described above concerns the use of
ultrasonic soldering equipment, which removes the
troublesome aluminum oxide layer by cavitation,
thereby permitting the oxide-free aluminum alloy to
b e tinned directly. (3) When it is found necessary
to make, or repair, an aluminum-to-stainless-steel
connection without benefit of copperplating or
ultrasonic equipment, such as in the field, friction
tinning can b e of value. In this method, molten
solder is applied to the aluminum working surface

248
I

l_

STAINLESS STEEL TUBE

SOLDER
ALUMINUM TUBE

\
I

FILLET
F I GUR E

6.

T ubu lar lap j oint.

and a sharp implemen t is used to rub the surface fr ee

of oxide ben eath th e solder, again permitting t he
oxide-fr ee aluminum alloy to bc tinned directly. (4)
In order to avoid the annoying and ever-present
possibility of oxidizing th e stainless steel when m aking the joint, i t is often beneficial t o use a copper
transition, or in termediate memb er in the connection . The copper sleeve may b e joined either to th e
inner or outer surface of the stainless steel by suitable
silver brazing technique, and the fr ee end is tinned
with soft solder. The joint is made after trea ting
the aluminum por tion as in (I ), (2), or (3) above.
A numb er of other m ethods h ave b een tried for
this type of work, some of which have b een erratic
in r eproducibility and other s of which h ave m et wi th
lit tle or no success.
Possibly on e of the more desirable m ethods of
making a bimetallic joint involves th e molecular
bonding of aluminum and its alloys to s teel. In this
process, the aluminum is cast by any of the standard
casting m ethods, against a specially prepared, clean,
ferrous surface. In this manner, any practicable
aluminum form can b e incorporated in to th e joint
design. T ests on a preliminary batch of 5 such
bimetallic joints indica ted that 4 r emain ed vacuumtight b etween room temp era ture and - 320 F , and
1 leaked at any t emp erature. A very poor intermetallic bond was observed on the one failur e that
occurred, but the r emaining four app eared good.
It appears that the b est method of joining aluminum and stainless steel depends greatly upon the
application involved as well as the availability and
suitability of equipment to do the particular job.
Although it is not n ecessary to employ an exper t to
fabri ca te connections of this type, a cer tain amount
of skill and car e is of considerable value in doing good
work.

6 . Auxiliary Design Features

The transfer of liquid hydrogen from these large
vessels is usually accomplished by increasing the gas
pressure in the free space above the liquid until it is
sufficien t t o force the liquid through an insula ted
transfer lin e t o the r eceiving vessel at a lower

pressure. A simple, convenien t method for pressurizing has been developed at the Bureau . It consists
of a circula ting pump , whieh wi thdraws cold hy drogen gas from the D ewar through a copper warmup
t ub e and then returns the gas to the space above the
liquid in the D ewar . In su ch a sys tem , for every
standard cubic foot of hy drogen gas circulated, abou t
2.5 whr is available to cause evaporation. This
raises the pressure over th e liquid. If pressurized
quickly enough, only th e top layer of liquid hy drogen
will come t o equilibrium with t he vapor . This
technique makes possible th e t ran sfer of sub cooled
liquid hy drogen and r esults in smaller losses th an
when transferring liquid hy drogen at satura tion
temp erature.
The pump used with t he 440-liter D ewar is a
~~ -hp oilless diaphragm paint-sprayer compre SOl'
with a 5.8-cfm free gas displacem ent. The pump is
small and compact enou gh t o be permanently
mounted on th e Dewar. The time required to
pressurize th e full D ewar from 1 t o 15 psig is about
3 min. The time required t o pressurize the D ewar
when half-full or nearly empty is some"wha t longer.
Transfer rates of over 2,000 liters/hr have been
accomplished with this pressurizing techniqu e.
On e of the predominantly experimen tal features
of the 440-liter D ewar is cen ter ed around th e use of
two liquid-hy drogen lines, one at th e top , the other
at the bottom of the hydrogen container. Prior to
fabrication, various theories existed as to the desirability of filling the liquid-hydrogen container from
the top rather than from the bottom, and vice versa.
If the filling takes place from the top , a great deal
of splashing occurs, and an additional line a t the
bottom is required to empty the D ewar. If, on t he
other hand, the filling takes place from the bottom,
flash vapor from the t ransfer operat ion bubbles
through the liquid as it collects in the container ,
possibly entraining liquid and carrying it ou t the
vent line. However , the v essel may be emptied
through the same line. Cont rary to speculation, no
difference was noted between top and bottom ports
in the filling operation, so it appears advisable to
use a single valve and line leading to th e bo ttom of
the hy drogen container for both filling and emptying.

249

-~---~--

The insulating vacuum of each Dewar was provided by an air-cooled oil-diffusion pump having a
rated speed of 260 liters/sec, together with associated
forepump, valves, and safety features . The 840liter Dewar was required to operate at ambient
temperatures up to 130 F , at which temperature
considerable backstreaming of oil vapor from the
diffusion pump was found in laboratory tests. To
prevent the oil from condensing on the cold goodre[-[ective surfaces in the Dewar and thereby raising
the emissivities, a baffle cooled by metallic contact
with the liquid-nitrogen-cooled shield was installed
in the pumping line to the Dewar vacuum space.

where the subscripts 1 and 2 refer to the inner and

outer surfaces, respectively. A value of 0.04 for EI
and E2 was used in the radiation calculations, al
though at the time there were no experimental data
for the particular alloy used to substantiate this
value. Subsequent experiments resulted in a value
of 0.03 for these surfaces. However, irregularities
due to piping and supports would increase the
effective emissivity so that 0.04 appears to be a
reasonable value.
7 .2 . Residual Gas Conduction

The residual gas conduction was calcula ted from

an equation derived by S. S. Kistler [8]. In slightly
modified form this equation can be written,

7 . Thermal Calculations
Preliminary heat-transfer studies were carried out
on various support and insula tion methods in order
to obtain predicted values for the heat [-[ow into the
liquid-nitrogen-cooled shield and the liquid-hydrogen
container of each Dewar. As mentioned above, the
heat [-[ow to each of these sh ells can be broken down
into three mechanisms: (1) Thermal radiation, (2)
residual gas conduction, and (3) heat flow through
piping and insulating suppor ts.
For heat-transfer calculations the Dewars may be
considered as two separate units. The first unit
consists of a liquid-nitrogen container at 77 K surrounded by the outer shell at 300 K; the second,
a hydrogen container at 20 K surrounded by the
liquid-nitrogen-cooled shield at 77 K. Each of
these units is considered separately in calculations
because the two important figures in the results are
the evaporative loss rates of liquid nitrogen and
liquid hydrogen.

.
q= 0.374

1M

T z- T I

-I I +(2-a -1 )P-l'

v-V TmL

where
q= watts per square inch of inner surface,
Ov= specific heat of the residual gas at constant
volume in joules per gram per degree K ,
Tm= (TI + T 2)/2= m.ean tempera ture of the gas
in degrees K ,
TI = temperature of the inner surfa ce in degrees

K,
T 2 = temperature of the ou tel' surface in degrees
K,
L = p erpendicular distance between inner and
outer surface in inches,
ll= mean free path of the gas in inches at T m
and a pressure of 1 mm Hg,
P = pressure of the gas in mm Hg,
a = accommodation coefficient factor of the
surfaces,
M = molecular weight of gas.

7.1. Radia hon

Radiant energy transfer may be calculated from

the Stefan-Boltzmann equation,

It should be noted that q, the rate of energy [-[ow

per unit of area, is used diI'ectly, because at low
pressures the concept of specific conductivity for
the gas is meaningless. If the residual gas is air
the value of II will be II = 6.56 X 10- 6 T min. , whereas
if t h e residual gas should be hydrogen, the value of
II would be II = 12.2 X 10- 6 T m in. As in the radiation calculations, a correction for the Dewar configuration must be applied. It seems reasonable to
use a Christiansen-type correction factor to define
an effective accommodation coeffi cient, a.

where
Q= watts,
o- = 0.0037 X 10- 8 w in. - 2 deg K -I,
E= emissivity factor ,
TI = tempera ture of inner surface in degrees K ,
T2 = temperature of outer surface in degrees K ,
Al = area of inner surface in square inches.
The emissivity factor used in the above equation is
obtained from a correction formula developed by
Christiansen [7] for concentric spheres and cylind ers:

Al and A2 are the areas of the inner and outer surfaces, respectively, and a l and a2 are the accommodation coefficients of the surfaces for the residual
gas molecules.

250

:e

7 .3 . Piping and Insulating Support Conduction

9 . Refrigeration

Heat flow through the piping was computed by

use of the standard heat-conduction equations and
values of thermal conductivity available in the literature [9]. Calculations on insulating support conduction were made at a time when limited data
were available on thermal contact r esistance (see
section 4). An informal communication contained
the only data available at that time. A figure of
approximately 0.75 mw/cm C was given for the
equivalent thermal conductivity of a Monel coil.
Because stainless steel provides better insulation
than Monel, and b ecause the thermal conductivity
decreases with decreasing temperature a value of
k = O.S mw/cm C was arbitrarily chosen for t h e
calculations.

It is frequ en tly desirable to store liquid hydrogen

without loss. For short p eriod s this may be accomplished by sealing the hy drogen container and allowing a moderate pressure ri se. In D ewars of the type
discussed h ere, storage durations of 1 week or more
result in a pressure increase of about 1 atm within
the hydrogen sys tem when the hydrogen container
is initially about 90 p ercent full of liquid.
2.

TAB LE

P eljonnance data f or 840-liter Dewal'

Liquid hyd rogen

Resid ual
pressure

Outer shell
te mperature

7n7n U g

op

10

52

2. 5

8. Thermal Performance
10-'

The average total h eat leak by piping, r esidual

gas and support conduction, and t hermal radiation
for the 440-liter D ewar, during a p eriod of time
exceeding 1~ years of actual D ewar operation, is
shown in table 1.

A represen tative set of performan ce data for the

840-liter D ewar, over an operational period of 6
months, is 'given in-table 2.
Thus it is seen that under comparable conditions10- 5 mm Hg insulating vacuum and SOo to 60 F
ambient temperature- the hydrogen capacity, hydrogen heat leak, and nitrogen heat leak of the 840liter D ewar is approximately twice that of the 440liter D ewar. The daily evaporation of liquid
hy drogen expressed as a percentage of the total
capacity is n early the same for both vessels. With
similar design obj ectives some improvement percentagewise should be expected in the performance
of the larger vessel. In this case, however, more
stringent m echanical service conditions r esul ted in
more h eat flow through insulating supports a.nd
pullified such improvement
T ABLE

7nmHg
10-'
5X lO-'
10-'
10-'

OF
60
60
60
60

U'

3.7
2.0
b 1. 3 (1. 4)
b O. 9 (0.8)

0. 80

10

118.0

Liters
63.2

1. Average total heat leak

Liquid hydrogen

Heat leak

To keep hydrogen without loss indefinitely, it is

n ecessary to periodically r efrigerate the hydrogen to
prevent excessive pressure rise in the hydrogen shell.
Several refrigeration schem es have b een proposed
and developed, whi ch operate on the principle of
r econdensing the vapor formed from the liquid hydrogen evaporation by circulating a colder gas, used as
a r efrigerant, through a closed condensing co il. The
Bureau has developed a r efrigerator that utilizes
th e principle of rcliq uefying the v apor formed from
evaporation of liquid hydrogen by circulating the
hydrogen vapor in a simple Joule-Thomson-type
hydrogen liquefi er. A high r efrigeration efficiency
is achieved. The compon ents of this r efrigera tion
system include the usual compression, purification,
purging, and control equipment, along with th e
n ecessary h ea.t exchanger s and expansion valve.
The 840-litel' D ewar is capable of b eing refrigerated
by equ ipment using this n ew application of the
Joule-Thomson process. The only components operating at low temperatures, and thus requiring insulation , are the heat exchangers and expansion
valve. These were mounted in the Dewar insulating
vacuum space. The r emainder of the r efrigeration

8 .2. The 8 40-Liter Dewar

Residual
pressure a

Dai lyeyapD ail yevaporati on (peroration

H eat leak
cen tage or H eat leak
total
capacity)

8 .1. The 440-Liter Dewar

Onter
shell
tem perature

Liquid ni trogen

Dailyevaporation (percen tage of total

capaCity)

2. 29
1. 24
b O. 81 (0.89)
b. 56 (0.46)

Liquid nitrogen
Wit hout add itional shields
H eat leak

----- - .-._-.- ---- ---- -b (63. 8)

b 62. 5 (61.1)

Daily
evaporation
Liters

----- --- ----

---. ------ --

b (34. 2)
b 33. 5 (32. 7)

With additional shields

Heat
leak c
U'

-- .-._---- - ---.- - - ---52.0

Daily
evaporation
Liters

------- -----

------- --------.-- ----27. 8

For the purpose of stuciying t he effect of a variable Insulating vacuum, t he tabulated d~ta fo r pressures a bove 10-' mm H g were obtained by introducing a small
leak to the system, because a vacuulll between 10.... and 10- 7 mill H g was usually maintained during normal operation.
b N umbers ill parentheses represent deSign calculations.
, It will be Doted that t he add itio n of two rigidl y supported aluminum radiation shields [9, fig . 1J bet ween tbe outer shell and the li quid-nitrogen-cooled radiation
shield red uced t he actual beat leak to tbe liquid nitrogen shield approximately 16 percent.

251

-equipment is mounted on a dolly, which can be

brought easily to the D ewar and connected. Construction of the refrigerator in this manner has resul ted in maximum efficiency and simplicity because
all connections are warm and the D ewar insulating
vacuum is utilized to insulate all low-temperature
components of the refrigeration cycle.
A detailed description of this reliquefying hydrogen
refrigerator is planned for a subsequent report.

10. Conclusions
The National Bureau of Standards has develop ed
r ugged, lightweight vessels for storing and transporting liquid hydrogen. These vessels have a very low
loss rate. T echniques employed in this development
include vacuum-tight aluminum welding, joining
aluminum to stainless steel, the use of novel insulating supports, and methods for improving the reflectivity of metallic surfaces. It appears entirely
feasible , by fur ther refinements in producing surfaces
of high r efl ectivity and employing mechanical designs
with less support conduction, to effect a still further
improvement, making possible the low-loss transportation of liquid helium.

improvements. A comparison of figures I and 7 will

service to illustrate these changes. In brief, the
aluminum 440-liter spherical container was replaced
with a 500-liter stainless-steel dish-ended cylindrical tank, externally silver-sprayed in order to increase the surface reflectivity. The vertical stainlesssteel center support tube was decreased in wall
thickness to 0.016 in. (one-fifth of its former value) ,
and lateral motion of the inner container with respect to its radiation shield was nullified by 2 sets
of 3 stainless-steel aircraft cables, %2 in. in diameter ,
which r eplaced the stainless-steel co il strips previously used; }~-in . aircraft cables also replaced the
entire bottom coil assembly. The rigidly supported
double set of "floating radiation shields" was exchanged for a single "floa ting shield" attached to the
outer shell by 3 long stainless-steel tubes, and Teflon
bumpers spaced the shield from the nitrogen shell
at the top. In order to reduce the over-all height of
the assembly, vent lines and valves were brought
through the side of the Dewar.
The thermal performance of this 500-liter Dewar
over a I-month period is given in table 3.
TA nLE

3.- Thennal performance of 500-liter Dewm'

11 . References
[1] Government er ects $3,500,000 low temperature la boratory,
Refrig. Eng. p. 1082 (1953).
[2] Donald P. Smit h, Hydrogen in metals (U niv . Chicago
Press 1948) .
[3] H. W. Gillette, The behavior of engineering metals, p . 353
(John Wi ley & Sons, Inc. , New York , N . Y ., 195 1) .
[4] Russell B. Scott (pri vate commu ni cation).
[5] R. P. Mikesell and R. B. Scott, Heat conduction t.hrough
insulating supports in very low temperat ure equipment,
J . Resear ch NBS 57, 371 (1956) RP2726.
[6] Morton C. Smith and David D . Rabb (private communication).
[7] Max M. Jakob and G. A. Hawkins, E lements of heat
transfer, 1st ed. , p. 125 (J ohn "Wi ley & Sons, Inc.,
N ew York , N. Y. , 1942).
[8] S. S. Kistler, The relation be tween heat cond uctivity a nd
struct ure in Silica Aerogel, J. Phys. Chem. 39, 79
(1935) .
[9] Robert L. Powell and \Yilli am L. Blanpied, Thermal conductiv ity of metals a nd alloys at low temperatures,
NBS Circ. 556 (1954).

12. An Improved Vessel

A. F. Schmidt and B. W . Birmingham
Subsequent to the work reported in the preceding
sections, the 440-1iter liquid-hydrogen D ewar was
converted in capacity and design to a 500-liter
Dewar with several predominant modifications and

Liq ui d hydrogen
R esid ual

pressu re

Liquid nitroge n

Outer shell

termperature

H eat
lea k

Dail y

Dail y eva poration (percen t

age of total
ca paC ity)

H eat
leak

eva p o r a.~

tion

1--- ------------------o F'

rnm H g
Liters
'W
%
10-<

96

0.47

0. 26

2;V 5

15.8

A note of interest may be inserted here relative to

one aspect of safety involved in the use of this vessel.
Users have been concerned about the consequences
of sudden loss of insulating vacuum for containers
of this type. A dangerous rise of pressure was feared.
In this container a I-in.-diameter hole was suddenly
opened in the outer wall of the Dewar vacu um space,
resulting in the entire contents- both hydrogen and
nitrogen- evaporating and being vented through the
respective %-i n. vent lines. The pressure rise resulting from the rapid evaporation did no damage.
In conclusion, these changes confirm that appli cation of design modifications recommended in section
10 were justified. A liquefied-gas-storage vessel has
been developed with a daily hydrogen evaporation
rate of approximately ~ percent of the total capacity
in the 500-liter size range. It is planned to evaluate
this vessel as a li q uid-helium-storage Dewar.

252

P~~T
I

2
3
4
5
6

DESCR I PTI ON

7 N VENT LINE
8 MULTIPLE -CONTACT
INSULATING SUPPORTS
9 RADIATION SHIELD
10 TEFLON BUMPER
X ALUMINUM TO STAINLESS
SOLDERED CONNECTIONS

LIQUID Hz VALVE
H RELIEF VALVE
N2 RELIEF VALVE
LIQUID Hz FILL 8 EMPTY
H2 VENT LINE
LIQUID Nz FILL 8 EMPTY

500

FIG U RE

7.

L I TERS
LIQU I D HYDROGEN

500-1iler liqui d hydrogen Dewar.

BO UL DER, COLO.,

253

October 27, 1955.