Вы находитесь на странице: 1из 545

ECSS-E-HB-31-01 Part 14A

5 December 2011

Space engineering
Thermal design handbook - Part 14:
Cryogenic Cooling

ECSS Secretariat
ESA-ESTEC
Requirements & Standards Division
Noordwijk, The Netherlands
ECSSEHB3101Part14A
5December2011

Foreword
ThisHandbookisonedocumentoftheseriesofECSSDocumentsintendedtobeusedassupporting
material for ECSS Standards in space projects and applications. ECSS is a cooperative effort of the
EuropeanSpaceAgency,nationalspaceagenciesandEuropeanindustryassociationsforthepurpose
ofdevelopingandmaintainingcommonstandards.
The material in this Handbook is a collection of data gathered from many projects and technical
journals which provides the reader with description and recommendation on subjects to be
consideredwhenperformingtheworkofThermaldesign.
Thematerialforthesubjectshasbeencollatedfromresearchspanningmanyyears,thereforeasubject
mayhavebeenrevisitedorupdatedbyscienceandindustry.
The material is provided as good background on the subjects of thermal design, the reader is
recommended to research whether a subject has been updated further, since the publication of the
materialcontainedherein.

This handbook has been prepared by ESA TECMT/QR division, reviewed by the ECSS Executive
SecretariatandapprovedbytheECSSTechnicalAuthority.

Disclaimer
ECSSdoesnotprovideanywarrantywhatsoever,whetherexpressed,implied,orstatutory,including,
butnotlimitedto,anywarrantyofmerchantabilityorfitnessforaparticularpurposeoranywarranty
that the contents of the item are errorfree. In no respect shall ECSS incur any liability for any
damages,including,butnotlimitedto,direct,indirect,special,orconsequentialdamagesarisingout
of,resultingfrom,orinanywayconnectedtotheuseofthisdocument,whetherornotbasedupon
warranty,businessagreement,tort,orotherwise;whetherornotinjurywassustainedbypersonsor
propertyorotherwise;andwhetherornotlosswassustainedfrom,oraroseoutof,theresultsof,the
item,oranyservicesthatmaybeprovidedbyECSS.

Publishedby: ESARequirementsandStandardsDivision
ESTEC,P.O.Box299,
2200AGNoordwijk
TheNetherlands
Copyright: 2011bytheEuropeanSpaceAgencyforthemembersofECSS

2
ECSSEHB3101Part14A
5December2011

Table of contents

1 Scope.....................................................................................................................26

2 References ............................................................................................................27

3 Terms, definitions and symbols..........................................................................28


3.1 Terms and definitions ............................................................................................... 28
3.2 Abbreviated terms .................................................................................................... 28
3.3 Symbols....................................................................................................................30

4 General introduction ............................................................................................42


4.1 Radiant coolers ........................................................................................................43
4.2 Stored solid-cryogen coolers ....................................................................................44
4.3 Stored liquid Helium (He4) coolers........................................................................... 44
4.4 Trends toward lower temperatures........................................................................... 45
4.5 Mechanical refrigerators ........................................................................................... 46
4.6 Low temperature requirements to IR sensors .......................................................... 47
4.6.2 Radiation from the optical system ..............................................................48
4.6.3 Noise from the detector .............................................................................. 49

5 Refrigerating systems ..........................................................................................51


5.1 General.....................................................................................................................51
5.2 Closed cycle .............................................................................................................51
5.2.1 Reverse-Brayton cycle ...............................................................................52
5.2.2 Reverse-Brayton and Claude cycle refrigerators........................................ 54
5.2.3 Gifford-McMahon/Solvay cycle refrigerators .............................................. 55
5.2.4 Joule-Thomson Closed Cycle Refrigerator................................................. 57
5.2.5 Stirling cycle refrigerators ...........................................................................58
5.2.6 Vuilleumier cycle refrigerator ...................................................................... 66
5.2.7 Existing systems.........................................................................................69
5.3 Open cycle .............................................................................................................105
5.3.1 Joule-Thomson open cycle refrigerators .................................................. 105
5.3.2 Existing systems.......................................................................................109
5.3.3 Stored liquid or solid cryogen open refrigerators...................................... 114

3
ECSSEHB3101Part14A
5December2011
6 VCS Dewars ........................................................................................................115
6.1 General...................................................................................................................115
6.2 Theoretical analysis................................................................................................117
6.2.1 Introduction............................................................................................... 117
6.2.2 The idealized model ................................................................................. 118
6.2.3 Evaluation of the restrictions involved in the idealized model .................. 123
6.3 Supports .................................................................................................................162
6.3.1 Introduction............................................................................................... 162
6.3.2 Support materials .....................................................................................163
6.3.3 Low thermal conductance tubing.............................................................. 165
6.3.4 Tensile and flexural supports.................................................................... 170
6.3.5 Compressive supports.............................................................................. 175
6.4 Phase separators ...................................................................................................176
6.4.1 Introduction............................................................................................... 176
6.4.2 Thermodynamic vent system.................................................................... 182
6.4.3 Capillary barriers ......................................................................................183
6.4.4 Porous media ...........................................................................................190
6.4.5 Baffled tanks.............................................................................................193
6.4.6 Empirical data for design ..........................................................................205
6.4.7 Testing......................................................................................................216
6.5 Existing systems.....................................................................................................218
6.5.1 Introduction............................................................................................... 218
6.5.2 Data on existing systems..........................................................................220

7 Superfluid Helium...............................................................................................234
7.1 Dynamics of superfluids .........................................................................................234
7.1.1 Relevant equations of superfluid dynamics .............................................. 235
7.1.2 Frictional effects .......................................................................................240
7.1.3 Counterflow heat transfer ......................................................................... 246
7.1.4 Heat transfer at arbitrary combinations of vn and vs ................................ 257
7.1.5 Vapor formation ........................................................................................258
7.1.6 Superfluid Helium film............................................................................... 259
7.2 Kapitza conductance ..............................................................................................267
7.2.1 Measuring methods ..................................................................................269
7.2.2 Experimental data.....................................................................................272
7.3 Thermo-acoustic oscillations .................................................................................. 303
7.4 The superfluid plug.................................................................................................305
7.4.1 Phase separation in superfluid helium...................................................... 305

4
ECSSEHB3101Part14A
5December2011
7.4.2 Simplified theory of the superfluid plug..................................................... 306
7.4.3 Characteristics of porous media ............................................................... 330
7.5 Filling a superfluid helium container ....................................................................... 338
7.5.1 Liquid loss because of pump down .......................................................... 338
7.5.2 Pumping down requirements.................................................................... 340
7.5.3 A typical filling sequence ..........................................................................340

8 Materials at cryogenic temperatures ................................................................343


8.1 Normal cryogens .................................................................................................... 343
8.1.1 General properties.................................................................................... 343
8.1.2 Entropy diagrams ..................................................................................... 391
8.2 Superfluid Helium-4................................................................................................443
8.3 Normal Helium-3 ....................................................................................................449
8.4 Metallic materials....................................................................................................452
8.5 Composite materials...............................................................................................466
8.5.1 Structural tubes ........................................................................................493
8.6 Miscellaneous materials .........................................................................................495

9 Safety with cryogenic systems .........................................................................496


9.1 General...................................................................................................................496
9.1.1 Physiological hazards............................................................................... 496
9.1.2 Fire and explosion hazards ...................................................................... 496
9.1.3 Pressure hazards ..................................................................................... 497
9.1.4 Materials hazards .....................................................................................497
9.1.5 Safety provisions ...................................................................................... 498
9.2 Hazards related to properties of cryogens.............................................................. 499
9.2.1 Combustion in an oxygen environment .................................................... 501
9.2.2 Combustible cryogens .............................................................................. 502
9.2.3 Fluorine.....................................................................................................510
9.2.4 O2 deficiency ............................................................................................511
9.3 Change of properties of structural materials .......................................................... 511
9.3.1 Temperature embrittlement ......................................................................511
9.3.2 Hydrogen embrittlement ...........................................................................521
9.3.3 Design codes and acceptance tests......................................................... 528

Bibliography...........................................................................................................529

Figures
Figure 4-1: He3 cooler being developed by NASA. From Sherman (1978) [216]. .................. 46

5
ECSSEHB3101Part14A
5December2011
Figure 4-2: Procedure to reduce the background flux from the optics. From Caren &
Sklensky (1970) [37]. ........................................................................................... 48
Figure 4-3: Detectivity, D*, of a photon noise-limited detector as a function of cutoff
wavelength, c, for several values of the optics temperature, T. From Caren
& Sklensky (1970) [37].........................................................................................49
Figure 4-4: Typical detector operating temperature, T, vs. detectivity, D*. The detector
is germanium doped either with mercury, with cadmium or with copper.
From Caren & Sklensky (1970) [37]. ...................................................................50
Figure 5-1: Reverse-Brayton Cycle Refrigerator. From Sherman (1978) [216]...................... 52
Figure 5-2: Compressor cross section of ADL rotary-reciprocating refrigerator. From
Donabedian (1972) [59]. ...................................................................................... 53
Figure 5-3: Claude Cycle Refrigerator. From Donabedian (1972) [59]. ................................. 54
Figure 5-4: Solvay Cycle Refrigerator. From Donabedian (1972) [59]. .................................. 56
Figure 5-5: Joule-Thomson Closed Cycle Refrigerator. From Donabedian (1972) [59]. ........ 57
Figure 5-6: Stirling Cycle Refrigerator Operation. From Sherman (1978) [216]..................... 58
Figure 5-7: Stirling Cycle Refrigerator Ideal Pressure-Volume and Temperature-
Entropy Diagrams. From Sherman (1978) [216].................................................. 59
Figure 5-8: Schematic representation of North American Philips refrigerator, showing
rhombic drive mechanism. The drive has two counter-rotating crankshafts,
each powered by a drive motor. By adjusting the mass of the reciprocating
members of the drive and by adding appropriate counterweights to the
crankshafts, the center of the gravity of all the moving parts can be kept
stationary. From Balas, Leffel & Wingate (1978) [16]. ......................................... 60
Figure 5-9: Schematic representation of North American Philips Magnetic Bearing
refrigerator, showing the linear motors for piston and displacer and the
magnetic bearing. The displacer rod passes through the piston. From
Sherman, Gasser, Benson & McCormick (1980) [221]........................................ 61
Figure 5-10: Coupling of two refrigerator units to provide cooling of a single detector.
The complete refrigerator can be seen in Figure 5-8. Here, on the contrary,
only the first and second stages of both refrigerators are shown. From Naes
& Nast (1980) [160]..............................................................................................62
Figure 5-11: Ground Test temperatures, of the first and second stage vs. Second stage
heat transfer rate, Q2, for different values of the first stage heat transfer
rate, Q1, and motor rpm. The data correspond to refrigerator 2 but are
typical of the four units. From Naes & Nast (1980) [160]. first stage, Q 1 =
1,5 W, 1000 rpm; first stage, Q 1 = 1,5 W, 1150 rpm; second stage Q 1
= 1,5 W, 1000 rpm; second stage Q 1 = 1,5 W, 1150 rpm; first stage, Q
1 = 2 W, 1000 rpm; Q1 = 2 W, 1000 rpm. ......................................................... 63
Figure 5-12: In orbit temperature, T, of several components of Gamma 004 systems vs.
Orbital time, t . From Naes & Nast (1980) [160]. cold tip of refrigerator 3;
cold tip of refrigerator 4; shroud; ground test value of cold tip of
refrigerator 3; ground test value of shroud....................................................... 63
Figure 5-13: In orbit temperature, T, of several components of Gamma 003 systems vs.
Orbital time, t . From Naes & Nast (1980) [160]. cold tip of refrigerator 2;
cold tip of refrigerator 1; shroud; ground test value of cold tip of
refrigerator 2; ground test value of shroud....................................................... 64

6
ECSSEHB3101Part14A
5December2011
Figure 5-14: In orbit heat transfer rates, Q, from Gamma 003 detector to refrigerators 1
and 2, vs. orbital time, t. From Naes & Nast (1980) [160]. detector heat
load. Refrigerator 2 on; heat load through meter 1, Q 1 . Refrigerator 1
off; heat load through meter 2, Q 2 . Refrigerator 2 on; refrigerators 1
and 2 on; refrigerators 1 and 2 on; refrigerators 1 and 2 on. ....................... 64
Figure 5-15: Schematic of the Vuilleumier-Cycle Refrigerator. From Sherman (1978)
[216].....................................................................................................................67
Figure 5-16: Vuilleumier-Cycle Refrigerator. From Sherman (1971) [218]............................. 67
Figure 5-17: Pressure-Volume Diagrams, for the Cold Cylinder, Hot Cylinder and Total
Gas, of the Vuilleumier-Cycle Refrigerator. From Sherman (1971) [218]. ........... 67
Figure 5-18: Inverse efficiency (required power per unit of refrigeration power) -1, vs.
operating temperature, T, for several closed cycle refrigerators. a - Brayton
refrigerators (Turbo machinery Systems). b - Stirling refrigerators. c -
Vuilleumier refrigerators. d - Gifford-McMahon/Solvay refrigerators. From
Donabedian (1972) [59]. Also shown are curves for closed cycle
refrigerators operating with the quoted efficiencies (in percentages of
Carnot) through the whole temperature range. From Haskin & Dexter
(1979) [83]. The Carnot efficiency for a machine working between TC and
TH temperatures is given by c = 1 TC/TH. Very low operating
temperatures result in a reduced efficiency for a given cooling load and a
given cycle. ..........................................................................................................70
Figure 5-19: System mass per unit of refrigeration power, Mp, vs. operating
temperature, T for several closed cycle refrigerators. a - Gifford-
McMahon/Solvay refrigerators. b - Stirling refrigerators. From Donabedian
(1972) [59]. ..........................................................................................................71
Figure 5-20: System mass per unit of refrigeration power (or cooling load), Mp, for
representative closed cycle refrigerating systems and for passive radiant
coolers. Closed cycle refrigerators, Q = 0,1 W. Closed cycle
refrigerators, Q = 1 W. Closed cycle refrigerators, Q = 10 W. Passive
radiant coolers; Q = 0,1 W. Passive radiant coolers; Q = 1 W. Passive
radiant coolers; Q = 10 W. From Haskin & Dexter (1979) [83]. Smallest
temperature attained by closed cycle refrigerators in orbit. Smallest
temperature attained by passive radiant coolers in orbit. From Sherman
(1982) [217]. ........................................................................................................72
Figure 5-21: System area per unit of refrigeration power (or cooling load), Ap/Mp, for
closed cycle refrigerating systems and for passive radiant coolers.
Closed cycle refrigerators, Q = 1 W. Passive radiant coolers; Q = 1 W.
From Haskin & Dexter (1979) [83]. Although the areas, Ap, have been
calculated for 1 W cooling, they could be scaled in approximately direct
proportion to the cooling load. Ap/Q = 7,13x107 T4 is the best fitting, by the
least squares method, to the data for passive radiant coolers. Smallest
temperature attained by closed cycle refrigerators in orbit. Smallest
temperature attained by passive radiant coolers in orbit. From Sherman
(1982) [217]. ........................................................................................................75
Figure 5-22: 80 K cooler schematic. From Jewell (1991) [103]............................................ 101
Figure 5-23: Cooler heat lift performance vs. gross compressor input power. From Scull
& Jewell (1991) [211]. ........................................................................................102
Figure 5-24: 20 K cooler schematic. From Jones et al. (1991) [110]. .................................. 103

7
ECSSEHB3101Part14A
5December2011
Figure 5-25: Heat lift performance of: a) development model; b) engineering model.
From Jones et al. (1991) [110]........................................................................... 104
Figure 5-26: 4 K cooler layout. From Bradshaw & Orlowska (1988) [27]. ............................ 104
Figure 5-27: Cooling power/mass flow vs. precooler temperature. From Bradshaw &
Orlowska (1991) [28]. ........................................................................................105
Figure 5-28: Isenthalps and inversion curve for different gasses. a Hydrogen. b Helium.
c Nitrogen. From Zemansky (1968) [272]. Data in b, after Hill &
Loumasmaa (1960) [89], are no longer valid for above 20 K. Upper
isenthalps are instead from Angus & de Reuck (1977) [6], pp. 64-127 . The
locus of the maxima has been drawn by the compiler as a dotted line. ............ 106
Figure 5-29: Schematic of a typical JT cryostat-dewar system. From Hellwig (1980)
[86].....................................................................................................................107
Figure 5-30: Schematic of a self-demand flow JT cryostat-dewar system. From Oren &
Gutfinger (1979) [175]. The sketch of the variable-orifice controlling device
is from Buller (1970) [35]. ..................................................................................108
Figure 6-1: Schematic representation of a solid gryogen cooler. From Breckenridge
(1972) [29]. ........................................................................................................116
Figure 6-2: Sketch of a typical VCS Dewar. From Niendorf & Choksi (1967) [169]. ............ 117
Figure 6-3: Heat transfer mechanism through a normal attachment VCS Dewar. From
Niendorf & Choksi (1967) [169]. ........................................................................ 117
Figure 6-4: Insulation model geometry................................................................................. 119
Figure 6-5: Ratio m/m0 against the cryogen sensibility, S, for different values of the
heat additions to the cryogen other than those across the insulation. No
cooled supports (msj = 0). Calculated by the compiler. ...................................... 121
Figure 6-6: Corrective factor, k, for the dependence of insulation thermal conductivity,
k, on temperature, T, against the sensibility, S, of the cryogen, for several
values of the temperature ratio, TC/TH. A linear thermal conductivity vs.
temperature dependence has been assumed. Calculated by the compiler. ...... 128
Figure 6-7: Insulation model with finite number of shields. .................................................. 129
Figure 6-8: Corrective factor, n, accounting for the influence of the finite number, n, of
shields, vs. the sensibility S of the cryogen, for several values of n.
Calculated by the compiler. ............................................................................... 134
Figure 6-9: Contours of constant values of the ratio of the heat flux through the VCS
system to the uncooled shield heat flux, mapped as functions of the
dimensionless distances, 1 and 2, of the two vapor cooled shields to the
cold face of the insulation, for several values of the sensibility, S, of the
cryogen. Uniform insulation thermal conductivity. The numerical values
labelling the contours corresponds to n/nopt 1. Calculated by the
compiler. ............................................................................................................140
Figure 6-10: Contours of dimensionless displacements of a single shield from its
optimum position (1 = 0,25) which produce a 10% increase in the heat flux
through a three shield system. The contours are mapped as functions of
the remaining two shields dimensionless positions. Numerical values are
for helium between 4 K and 300 K. From Atherton & Prentiss (1973) [12]. ....... 141
Figure 6-11: Contours of constant values of the ratio of the heat flux through the VCS
system to the uncooled shield heat flux, mapped as functions of the
dimensionless distances, 1 and 2, of the two vapor cooled shields to the

8
ECSSEHB3101Part14A
5December2011
cold face of the insulation, for several cryogens in typical cases.
Temperature dependent insulation thermal conductivity (k = k1T). The
numerical values labelling the contours corresponds to n/nopt 1.
Calculated by the compiler. ............................................................................... 142
Figure 6-12: Factor Nu, accounting for finite convective heat transfer in the venting
duct, vs. coefficient r, for several cryogens. TH = 300 K. Calculated by the
compiler. ............................................................................................................145
Figure 6-13: Factor Nu, accounting for finite convective heat transfer in the venting
duct, vs. coefficient r, for several cryogens. TH = 200 K. Calculated by the
compiler. ............................................................................................................146
Figure 6-14: Factor Nu, accounting for finite convective heat transfer in the venting
duct, vs. coefficient r, for several cryogens. TH = 150 K. Calculated by the
compiler. ............................................................................................................147
Figure 6-15: Helium vapor bulk temperature, Tb, vs. insulation temperature, T, for
different values of the dimensionless heat transfer coefficient, r. TH = 300 K.
Calculated by the compiler. ............................................................................... 148
Figure 6-16: Temperature, T, across the insulation for different values of the
dimensionless heat transfer coefficient r. Helium vapor cooling. TH = 300 K.
Calculated by the compiler. ............................................................................... 149
Figure 6-17: Sketch of a VCS insulation in the nearness of the venting duct. Normal
attachment. After Paivanas et al. (1965) [177]. ................................................. 149
Figure 6-18: Sketch of the insulation and of the simplified configurations used to
analyze the influence of the finite thermal conductivity of the shields. (a)
Insulation. (b) Simplified configuration in the physical coordinates x, y. (c)
Simplified configuration in the stretched coordinates, , ................................. 151
Figure 6-19: Sketch of a typical spaceborne Dewar. All the dimensions are in mm. ........... 152
Figure 6-20: Coefficient, (y 1)/, of the first order correction accounting for the
influence of the finite thermal conductivity of the VCSs on the cryogen boil-
off rate, as a function of the cryogen sensibility, S, for two values of the
dimensionless outer radius of the venting duct, . The results have been
obtained by means of a perturbation scheme in the small parameter, ,
which measured the ratio of normal to lateral heat flux, and are valid
provided that terms of order 3/2 can be neglected. Calculated by the
compiler. ............................................................................................................160
Figure 6-21: Cryogenic supports tubes. a) Composite. b) All-metal. All dimensions are
in mm. From Hall & Spond (1977) [81]. ............................................................. 166
Figure 6-22: Heat transfer rate, Qs, through fiber-glass overwrapped and through all-
stainless-steel supports vs. support length, L, for several values of liner wall
thickness, tl, and overwrap thickness, to. (a) Inner diameter of the tube, d =
12,7x103 m. (b) d = 50,8x103 m. From Hall et al. (1971) [80]. ......................... 167
Figure 6-23: Heat transfer rate, Qs, through fiber-glass overwrapped supports vs. liner
wall thickness, tl, for several support lengths, L and overwrap thickness, to =
0,762x103 m. Hoop wrapping. (a) Inner diameter of the tube, d = 12,7x103
m. (b) d = 50,8x103 m. From Hall et al. (1971) [80]. ......................................... 168
Figure 6-24: Heat transfer rate, Qs, through fiber-glass overwrapped stainless-steel
supports vs. overwrap thickness, to, for several supports lengths, L, and
liner wall thickness tl = 0,51x103 m. Hoop wrapping. (a) Inner diameter of
the tube, d = 12,7x103 m. (b) d = 50,8x103 m. From Hall et al. (1971) [80]. .... 169

9
ECSSEHB3101Part14A
5December2011
Figure 6-25: Heat transfer rate, Qs, through fiber-glass overwrapped stainless-steel
supports vs. warm boundary temperature, TH, for several values of the cold
boundary temperature, TC. Tube length, L, liner wall thickness, tl, and
overwrap thickness, to, as indicated in the insert. Hoop wrapping. (a) Inner
diameter of the tube, d = 12,7x103 m. (b) d = 50,8x103 m. From Hall et al.
(1971) [80]. ........................................................................................................170
Figure 6-26: Typical supporting methods. Notice how the rods shown in (a) are crossed
to minimize the effect of thermal contraction and to increase the length of
the heat flow path. In (b), long suspension rods are accommodated in
standoffs. From Barron (1966) [18]....................................................................171
Figure 6-27: Tensile support of a liquid helium tank. From Lemke, Klipping & Rmisch
(1978) [131]. ......................................................................................................171
Figure 6-28: Spacing discs. From Bennett et al. (1974) [23]................................................ 172
Figure 6-29: Support tube for a liquid helium Dewar. From Bennett et al. (1974) [23]......... 173
Figure 6-30: Two ways of supporting cryogenic containers by means of tensile ties.
After Glaser et al. (1967) [75]. ...........................................................................174
Figure 6-31: Sketch of the Superfluid Helium Cryostat for Space Use (CRHESUS)
showing the tensile ties used for supporting the helium tank. From Lizon-
Tati & Girard (1978) [134]. .................................................................................174
Figure 6-32: CRHESUS heat flow diagram. From Lizon-Tati & Girard (1978) [134]. ........... 175
Figure 6-33: Composite column compressive support. From Heim & Fast (1973) [85]........ 176
Figure 6-34: Schematic of thermodynamic vent system. a) Forced convection. From
Mitchell et al. (1967). b) Pulsed constant pressure. From Mller et al.
(1983) [157]. ......................................................................................................182
Figure 6-35: Thermodynamic phase separator. From Fradkov & Troitskii (1975) [71]......... 183
Figure 6-36: A capillary barrier in static equilibrium. From McCarthy (1968) [144]. ............. 184
Figure 6-37: Container with a capillary-barrier partition. From McCarthy (1968) [144].
(a) An angular acceleration appears when the interface is formed at the
barrier. (b) The configuration reaches a steady angular velocity before
interaction of the interface with the barrier. See Table 6-12 for the definition
of the experimental conditions. .......................................................................... 185
Figure 6-38: Results of barrier dynamic stability tests. Bond number-controlled mode.
Tests were insufficient for determining the effect on barrier stability of the
various dimensionless parameters. From McCarthy (1968) [144]. .................... 186
Figure 6-39: Results of dynamic stability tests with different barriers. Bond number-
controlled mode. The acceleration, g, is parallel to the barrier. From Fester
(1973) [67]. A Reynolds number through the hole has been plotted vs. the
critical Bond number. ......................................................................................... 187
Figure 6-40: Results of barrier dynamic stability tests. Weber number-controlled mode.
From McCarthy (1968) [144]. The Weber number in abscissae is
normalized with an analytical critical Weber number Wec, which is given in
Figure 6-41 below. ............................................................................................. 188
Figure 6-41: Critical Weber number, Wec, as a function of geometry, l/D, and position of
the axis of rotation, L/D. These results have been obtained by use of a
potential (incompressible, inviscid, irrotational flow) theory with Op 1,
although assuming that the barrier induces a capillary pressure difference.
From Gluck (1970) [76]......................................................................................188

10
ECSSEHB3101Part14A
5December2011
Figure 6-42: Damping performance of selected barriers. From Fester (1973) [67]. The
damping categories A to G are associated to the flow patterns resulting
after impingement of the liquid with the barrier, from orderly (A) to irregular
(G)......................................................................................................................189
Figure 6-43: Compartmented tank device. From Fester, Eberhardt & Tegart (1975) [68]. .. 190
Figure 6-44: Sustained liquid height in a capillary tube........................................................ 190
Figure 6-45: Sustained ethanol height, l, vs. diameter of fiber, do. Gravity level 40go.
From Enya, Kisaragi, Ochiai, Sasao & Kuriki (1981) [64]. ................................. 191
Figure 6-46: Sustained liquid height, l, vs. gravity level, g/go. Liquids are: Ammonia
(circle), underfilled Ammonia (square), and ethanol (triangle). Matrix is
glass wool, do = 10-6 m. Solid lines have been deduced from Eq. [6-91] with
= 0 and the quoted values of d. From Enya, Kisaragi, Ochiai, Sasao and
Kuriki (1981) [64]. ..............................................................................................192
Figure 6-47: Criterion for the onset of nucleation in subcooled boiling. After Collier
(1981) [46]. ........................................................................................................193
Figure 6-48: Post height, l, required to position a given ullage, U, under reduced
gravity. See Clause 6.4.5.2 for explanation of curves d/R =/ 0, B and C........... 194
Figure 6-49: Experimental glass tank with a standpipe. From Petrash, Nussle & Otto
(1963) [184]. All the dimensions are in mm. ...................................................... 195
Figure 6-50: Minimum ullage centering capability of the standpipe. .................................... 196
Figure 6-51: Liquid acquisition by the standpipe for large ullages. From Petrash, Nussle
& Otto (1983) [184]. ...........................................................................................196
Figure 6-52: Central post with thin, off axis, posts (fingers). From Tegart et al. (1972)
[233]...................................................................................................................197
Figure 6-53: Criteria to deduce vane profile limits. From Tegart et al. (1972) [233]. ............ 198
Figure 6-54: Limiting vane profiles, Rmin/R and Rmax/R for n = 6, 8 and 12 vanes. Rmin/R
has been calculated for an ullage U = 0,05. Rmax/R is ullage-independent.
After Tegart et al. (1972) [233]........................................................................... 199
Figure 6-55: Simplified bubble geometry. The bubble is held by two contiguous vanes
and shapes up as if it were held by the "effective" vane. From Tegart et al.
(1972) [233]. ......................................................................................................200
Figure 6-56: The ideal distorted axisymmetrical bubble....................................................... 201
Figure 6-57: Angle a which measures the distortion of the bubble vs. ratio, Ro/R, of
inner body radius to tank radius. Calculated by the compiler. ........................... 202
Figure 6-58: Typical effective vane profiles, Ro/R, and dimensionless restoring force,
RK, vs. displacement angle, . The Figure has been replotted by the
compiler after a representation in polar coordinates by Tegart et al. (1972)
[233]...................................................................................................................203
Figure 6-59: Typical effective vane profiles, Ro/R, and dimensionless restoring force,
RK, vs. displacement angle, . The vane profiles have been calculated by
Eq. [6-99] with the shown values of k and m. Forces have been deduced
from Eqs. [6-96] to [6-98]. ..................................................................................204
Figure 6-60: Bond length, Lb, as a function of T, for saturated Argon, Methane, Nitrogen
and Oxygen. ......................................................................................................206

11
ECSSEHB3101Part14A
5December2011
Figure 6-61: Bond length, Lb, as a function of T, for saturated Ethane, Carbon Dioxide
and Ammonia.....................................................................................................207
Figure 6-62: Bond length, Lb, as a function of T, for saturated Hydrogen, Helium and
Neon. .................................................................................................................208
Figure 6-63: Relation between contact angle, , and surface tension, , for several
liquids on the quoted surfaces. ..........................................................................210
Figure 6-64: Sketch of a dual stage solid cooler. From Nast et al. (1976) [161]. ................. 218
Figure 6-65: Liquid helium (He4) coolers. a) Single stage. b) Dual stage. From Sherman
(1978) [216]. ......................................................................................................219
Figure 6-66: Normal attachment of the VCSs to the cooling duct through heat stationis.
From Glaser et al. (1967) [75]............................................................................ 219
Figure 6-67: Tangential attachment of the cooling duct to the shields. Sketched by the
compiler after Hopkins & Chronic (1973) [94].................................................... 220
Figure 6-68: Detector, T1, and optics, T2, temperature vs. orbital time. ............................... 225
Figure 6-69: JPL-Caltech IR detector cooler arrangement................................................... 227
Figure 6-70: Heat Flow diagram of the Ball Brothers Liquid helium Dewar.......................... 232
Figure 7-1: Phase diagram for He4 (not to scale). From Arp (1970) [10].............................. 234
Figure 7-2: Schematic of the apparatus used by the Leiden group to produce helium
flow through capillaries with independent variation of superfluid and normal
velocities. a) From Van der Heijden, Van der Boog & Kramers (1974) [247].
b) From De Haas & Van Beelen (1976) [55]...................................................... 242
Figure 7-3: The superfluid friction, LFs, vs. relative velocity, vn-vs, for various runs with
svs+nvn = Const. From van der Hejden, van der Boog & Kramers (1974)
[247]...................................................................................................................244
Figure 7-4: The mutual friction, LFsn, vs. relative velocity, vn-vs, from various constant
mass flux runs. From van der Hejden, van der Boog & Kramers (1974)
[247]...................................................................................................................245
Figure 7-5: Mutual friction to superfluid friction ratio, Fsn/Fs, vs. relative velocity, vn-vs,
from various runs with vs 0 and vn 0. From van der Heijden, van der
Boog & Kramers (1974) [247]. ...........................................................................245
Figure 7-6: Isothermal and iso chemical-potential flows in the vn,vs plane. The shaded
region corresponds to subcritical flow (=0). From van der Heijden, van
der Boog & Kramers (1974) [247]......................................................................246
Figure 7-7: Correlations between the critical superfluid velocity, vsc1, and the tube
diameter, DE. The experimental data have been re-plotted by the compiler
after van Alphen et al. (1969) [246]. They correspond to widely different
flow conditions. * Clow and Reppy, T-T 50 x 10 -3 K. Fokkens, film flow.
Pellman, "superfluid wind tunnel". Chase, heat conduction TT ; v n
0. Van Alphen, adiabatic flow rate. Van Alphen, energy dissipation
technique. Kramers, second sound attenuation in puresuperfluid flow.
Van Alphen, critical flow through jeweller's rouge. Keller and Hammel,
isothermal flow. Data from reviews of Atkins, and Hammel and Keller. .......... 248
Figure 7-8: Schematic of pressure and temperature drop data as a function of heat flux.... 249
Figure 7-9: Schematic of Lv1/2DE vs. vsDE under steady-state conditions. From Childers
& Tough (1976) [44]...........................................................................................251

12
ECSSEHB3101Part14A
5December2011
Figure 7-10: Critical Reynolds number for counterflow heat exchange, Rec, as a
function of temperature, T. From Arp (1970) [10]. ............................................. 253
Figure 7-11: Diagrams which relate the thermal gradient, dT/dx, to the heat flux, q, in
counterflow heat exchange. T=1,5 K to 2 K. Calculated by the compiler
after Arp (1970) [10]...........................................................................................254
Figure 7-12: Temperature profile along a channel filled with He II at atmospheric
pressure in conterflow heat exchange. From Bon Mardion, Claudet &
Seyfert (1979) [26]. ............................................................................................256
Figure 7-13: Tube and He II bath arrangement.................................................................... 258
Figure 7-14: Film and bulk liquid configuration..................................................................... 259
Figure 7-15: Bernoulli thinning. The full line corresponds to Eq. [7-46]. The dotted line
is the Kontorovich (1956) [125] solution. Neither solution gives the correct
transition of the film interface to the horizontal free surface in the reservoir,
because capillary pressure has been neglected. Curves labelled with the
values of Bo correspond to Eq. [7-49]. .............................................................. 262
Figure 7-16: Cell used to perform reduced-gravity test. The film thickness experiments
were performed in the left hand side compartment. From Yang & Mason
(1980) [268]. ......................................................................................................267
Figure 7-17: Kapitza conductance, hk, of low Debye temperature metals, Mercury,
Lead, Gold and Silver in contact with Liquid Helium, vs. temperature, T.
See Table 7-2 below. .........................................................................................273
Figure 7-18: Kapitza conductance, hk, of Copper in contact with various low acoustic
impedance materials vs. temperature, T. See Table 7-2 and Table 7-41
below. Theoretical results are also shown in this figure. ................................... 274
Figure 7-19: Kapitza conductance, hk, of Tungsten, Aluminium, Molybdenum and
Beryllium, in contact with Liquid Helium, vs. temperature, T. See Table 7-2
below. ................................................................................................................275
Figure 7-20: Kapitza conductance, hk, of Nonmetals in contact with Liquid helium vs.
temperature T. See Table 7-32 below. ..............................................................275
Figure 7-21: The neutral stability curve for Taconis oscillations when = 1. D E = 2,4
x 10 -3 m, T H = 288 K; D E = 2,4 x 10 -3 m, T H = 77,3 K; D E = 4,4 x 10 -3
m, T H = 288 K; D E = 4,4 x 10-3 m, TH = 77,3 K From Yazaki, Tominaga &
Narahara (1979) [269]. ......................................................................................304
Figure 7-22: Device for preventing Taconis oscillations. All the dimensions are in mm.
From Hilal & McIntosh (1976) [88]. .................................................................... 305
Figure 7-23: Superfluid plug arrangement. The intake face of the plug is located at x =
0.........................................................................................................................306
Figure 7-24: Backward pressure, p2, as a function of mass flow rate, m, through the
plug. Experimental points are from smooth curves by Karr & Urban (1978,
1980) [113] & [114]. The curve shown in the figure and the Reynolds
number in the abscissae axis correspond to turbulent flow (neglecting
entrance effects, see ECSS-E-HB-31-01 Part 13 clause 7.2.5) in a straight
tube of circular cross-section, under the validity of Blasius formula, for the
data shown in the insert. Calculated by the compiler. ....................................... 312
Figure 7-25: Quadrangle of data required in porous plug performance evaluation.............. 313
Figure 7-26: Mass flow rate, m, vs. pressure drop, pl-p2, for slits of various lengths, t,
and two different bath temperatures, Tl. From Denner et al. (1980) [56]. .......... 324

13
ECSSEHB3101Part14A
5December2011
Figure 7-27: Active Phase Separator (APS). From Denner et al. (1982) [57]. ..................... 325
Figure 7-28: Three typical positions of the liquid-vapor interface. a) Ideal flow
separation. b) Choking. c) Gorter-Mellink flow. From Schotto (1984) [209]....... 327
Figure 7-29: Temperature distribution within a 4 x 10-2 m thick. Ceramic plug for several
pressure differences. p2 = 2,55 x 10-3 Pa in any case. From Elsner (1973)
[63].....................................................................................................................329
Figure 7-30: Time constant, b, as a function of heating power, Q, for the plug described
by Karr & Urban (1978,1980) [113] & [114] in clause 7.4.2.6. Position of the
heaters, H, is also shown in the figure. White circle: upstream heater
power-on; blackcircle: upstream heater power-off; white square: heater at
the plug exit power-on; black square: heater at the plug exit, power-off.
From Karr & Urban (1978,1980) [113] & [114]. There is no consistent
difference between power-on and power-off. .................................................... 329
Figure 7-31: Fraction, f, of liquid mass lost because of pump down vs. final
temperature, Tf. Curves labelled REVERSIBLE correspond to Eqs. [7-80]
and [7-81] respectively. Experimental results are also shown. From Nicol &
Bohm (1960) [168]. ............................................................................................339
Figure 7-32: Mass flow rate, m/, required for a refrigerating load of 102 W as a
function of final temperature, Tf, under three different situations. (a) Liquid
He4 is continuously supplied at 4,2 K for evaporation. (b) No supply of He4.
(c) Liquid He3 is continuously supplied at 3,2 K for evaporation. From Nicol
& Bohm (1960) [168].......................................................................................... 340
Figure 7-33: Superfluid helium filling assembly. Explanation: NV1 to NV4, ruby needle
valve; NV5, standard needle valve; V1, remote controlled QSB for flap
valve; V2 to V8, standard valves; F1 and F2, external fittings to maintain
cleanliness; T1, 120 flexible transfer tube continuous with filling cryostat
and having a 4,2 K radiation shield; T2, long flexible transfer tube for filling
4,2 K tank; R1, 4,2 K reservoir and header tank; R2, 1,5 K reservoir. (NV2
is the porous plug seal. NV3 is the gas vent hole seal). From OXFORD
INSTRUMENTS (1976) [176]. ...........................................................................341
Figure 8-1: Density, , of Saturated Liquid Argon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................348
Figure 8-2: Density, , of Saturated Solid Argon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................349
Figure 8-3: Density, , of Saturated Liquid Methane vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................349
Figure 8-4: Density, , of Saturated Solid Methane vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................350
Figure 8-5: Density, , of Saturated Liquid Ethane vs. temperature, T. From Johnson
(1975) [107]. ......................................................................................................350
Figure 8-6: Density, , of Saturated Liquid Carbon Dioxide vs. temperature, T. From
LEFAX [130]. .....................................................................................................351
Figure 8-7: Density, , of Saturated Solid Carbon Dioxide vs. temperature, T. From
LEFAX [130]. .....................................................................................................351
Figure 8-8: Density, , of Saturated Liquid Hydrogen vs. temperature, T. From Vargaftik
(1975) [253]. ......................................................................................................352

14
ECSSEHB3101Part14A
5December2011
Figure 8-9: Density, , of Saturated Liquid Helium-4 vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................352
Figure 8-10: Density, , of Saturated Solid Helium-4 vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................353
Figure 8-11: Density, , of Saturated Liquid Helium-3 vs. temperature, T. From Keller
(1969) [119]. ......................................................................................................353
Figure 8-12: Density, , of Saturated Liquid Helium-3 -along the freezing curve- vs.
temperature, T. From Keller (1969) [119]. ......................................................... 354
Figure 8-13: Density, , of Saturated Solid Helium-3 vs. temperature, T. From Keller
(1969) [119]. ......................................................................................................354
Figure 8-14: Density, , of Solid Helium-3 vs. temperature, T. Values of are shown
along the melting curve as well as along curves of constant isobaric
compressibility. hcp and bcc stand for hexagonal-close-packed and body-
centered-cubic phases of Solid Helium-3, respectively. From Straty (1966)
[228]. Additional data for a wider temperature range are given in Figure
8-15....................................................................................................................355
Figure 8-15: Density, , of Solid Helium-3 vs. temperature, T, along the melting-
freezing curve. Since Helium-3 samples were contaminated with around 0,2
% of Helium-4, two freezing curves appear. The figure also reveals the
existence of another solid phase, cubic-close-packed (ccp), at high
pressures. From Sample (1966) [205]. .............................................................. 356
Figure 8-16: Density, , of Saturated Liquid Nitrogen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................356
Figure 8-17: Density, , of Saturated Solid Nitrogen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................357
Figure 8-18: Density, , of Saturated Liquid Ammonia vs. temperature, T. From
Vargaftik (1975) [253]. .......................................................................................357
Figure 8-19: Density, , of Saturated Liquid Neon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................358
Figure 8-20: Density, , of Saturated Solid Neon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................358
Figure 8-21: Density, , of Saturated Liquid Oxygen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................359
Figure 8-22: Specific heat, cp, of several gases vs. temperature, T. Sources of data,
and pressures are: Ar, CO2, H2, N2 and O2, from Hilsenrath et al. (1960)
[90], 1 atm. CH4 and Ne, from Johnson (1961) [109], 1 atm. C2H6 from
Vargaftik (1975) [253], 105 Pa. He4 from Angus & de Reuck (1977) [6], 105
Pa. NH3 from Norris et al. (1967) [171], 1 atm. 1 atm = 1,0135x105 Pa. ......... 359
Figure 8-23: Specific heat, cp, of Gaseous Helium-3 vs. the deviation TTcfrom
critical temperature, Tc, along nearly critical isochores. T < T c .
Experimental. cp = R[2,7-3,7ln((Tc-T)/T)] T > T c . Experimental.
cp = R[0,5-3,7ln((T-Tc)/Tc)] From Keller (1969) [119]. Tc = 3-324 K,
R = R/M R = 8,31432 J.mol1.K1 M = 3,01603x103 kg.mol1. ........................ 360
Figure 8-24: Specific heat, cp, of Liquid Helium-3, vs. temperature, T, at several
pressures. a) is for p = psat, and the shazed zone is enlarged in b). Data in
b) are for the following pressures p = p sat ; p = psat p = 14,9x10 5
5 5
Pa. p = 11,7x10 Pa. p = 28,3x10 Pa. p = 27x105 Pa.

15
ECSSEHB3101Part14A
5December2011
Data points are from Strongin et al. (1963) [230] and curves from Keller
(1969) [119]. ......................................................................................................361
Figure 8-25: Specific heat, c, of Solid Helium-4 (dashed line) and Solid Helium-3 (full
lines), vs. temperature, T. Numbers on the curves are densities in kg.m3.
From Sample (1966) [205]................................................................................. 362
Figure 8-26: Heat of conversion, h, from Normal to Para Hydrogen vs. temperature, T.
From Johnson (1961) [109]. .............................................................................. 363
Figure 8-27: Heat of vaporization, hfg, of Saturated Liquid Argon vs. temperature, T.
From Vargaftik (1975) [253]...............................................................................363
Figure 8-28: Heat of vaporization, hfg, of Saturated Liquid Methane vs. temperature, T.
From Vargaftik (1975) [253]...............................................................................364
Figure 8-29: Heat of vaporization, hfg, of Saturated Liquid Ethane vs. temperature, T.
From Vargaftik (1975) [253]...............................................................................364
Figure 8-30: Heat of vaporization, hfg, of Saturated Liquid Carbon Dioxide vs.
temperature, T. From Angus, Armstrong & de Reuck (1976) [5]. ...................... 364
Figure 8-31: Heat of vaporization, hfg, of Saturated Liquid Normal Hydrogen vs.
temperature, T. From Johnson (1961) [109]. ..................................................... 365
Figure 8-32: Heat of vaporization, hfg, of Saturated Liquid Helium-4 vs. temperature, T.
From Angus & de Reuck (1977) [6]. .................................................................. 365
Figure 8-33: Heat of vaporization, hfg, of Saturated Liquid Helium-3 vs. temperature, T.
From Keller (1969)[119].....................................................................................366
Figure 8-34: Heat of vaporization, hfg, of Saturated Liquid Nitrogen vs. temperature, T.
From Johnson (1961) [109]. .............................................................................. 366
Figure 8-35: Heat of vaporization, hfg, of Saturated Liquid Ammonia vs. temperature, T.
From Vargaftik (1975) [253]...............................................................................366
Figure 8-36: Heat of vaporization, hfg, of Saturated Liquid Neon vs. temperature, T.
From Johnson (1961) [109]. .............................................................................. 367
Figure 8-37: Heat of vaporization, hfg, of Saturated Liquid Oxygen vs. temperature, T.
From Johnson (1961) [109]. .............................................................................. 367
Figure 8-38: Heat of sublimation, hfg, of several solid cryogens vs. temperature, T. T is
bounded from above by the triple point and from below by a vapor pressure
of 1,33 Pa. From Nast, Barnes & Wedel (1976) [161]. ...................................... 368
Figure 8-39: Vapor pressure, psat, of Liquid Argon vs. temperature, T. From Hilsenrath
et al. (1960) [90].................................................................................................369
Figure 8-40: Vapor pressure, psat, of Solid Argon vs. temperature, T. From Hilsenrath et
al. (1960) [90].....................................................................................................369
Figure 8-41: Vapor pressure, psat, of Liquid Methane vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................370
Figure 8-42: Vapor pressure, psat, of Solid Methane vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................370
Figure 8-43: Vapor pressure, psat, of Liquid Ethane vs. temperature, T. From Vargaftik
(1975) [253]. ......................................................................................................371
Figure 8-44: Vapor pressure, psat, of Liquid Carbon Dioxide vs. temperature, T. From
Hilsenrath et al. (1960) [90]. .............................................................................. 371

16
ECSSEHB3101Part14A
5December2011
Figure 8-45: Vapor pressure, psat, of Solid Carbon Dioxide vs. temperature, T. From
Caren & Coston (1968) [36]...............................................................................372
Figure 8-46: Vapor pressure, psat, of Liquid Hydrogen vs. temperature, T. From
Vargaftik (1975) [253]. .......................................................................................372
Figure 8-47: Vapor pressure, psat, of Solid Hydrogen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................373
Figure 8-48: Vapor pressure, psat, of Liquid Helium-4 vs. temperature, T. From Angus &
de Reuck (1977) [6]. ..........................................................................................373
Figure 8-49: Vapor pressure, psat, of Liquid Helium-3 vs. temperature, T. From
Mendelssohn (1960) [148]. ................................................................................ 374
Figure 8-50: Vapor pressure, psat, of Liquid Nitrogen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................374
Figure 8-51: Vapor pressure, psat, of Solid Nitrogen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................375
Figure 8-52: Vapor pressure, psat, of Liquid Ammonia vs. temperature, T. From
Vargaftik (1975) [253]. .......................................................................................375
Figure 8-53: Vapor pressure, psat, of Solid Ammonia vs. temperature, T. From
Kutateladze & Borishankii (1966) [127]. ............................................................ 376
Figure 8-54: Vapor pressure, psat, of Liquid Neon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................376
Figure 8-55: Vapor pressure, psat, of Solid Neon vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................377
Figure 8-56: Vapor pressure, psat, of Liquid Oxygen vs. temperature, T. From Johnson
(1961) [109]. ......................................................................................................377
Figure 8-57: Thermal conductivity, k, of several gases -at a pressure of 105 Pa- vs.
temperature, T. From Vargaftik (1975) [253]. .................................................... 378
Figure 8-58: Thermal conductivity, k, of gaseous Carbon Dioxide -at a pressure of 105
Pa- vs. temperature, T. From Vargaftik (1975) [253]. ........................................ 378
Figure 8-59: Thermal conductivity, k, of Gaseous Hydrogen and Helium-4 -at a
pressure of one atmosphere (1,013x105 Pa) vs. temperature, T. From
Johnson (1961) [109].........................................................................................379
Figure 8-60: Thermal conductivity, k, of Gaseous Helium-4 and Helium-3 vs.
temperature, T. Calculated curves and experimental points are from
different sources. From Keller (1969) [119]. The thermal conductivity of
Gaseous Helium-4 in a much larger temperature range is given in Figure
8-59....................................................................................................................380
Figure 8-61: Thermal conductivity, k, of Liquid Helium-3 -at several pressures- vs.
temperature, T. p = 10 4 Pa; p = 105 Pa p = 6,7x10 5 Pa;
6 5
p = 10 Pa p = 26,9x10 Pa; p = 34,4x105 Pa From
Keller (1969) [119]. ............................................................................................380
Figure 8-62: Thermal conductivity, k, of gaseous Ammonia -at a pressure of 105 Pa- vs.
temperature, T. Calculated curves and experimental points are from
different sources. From Vargaftik (1975) [253]. ................................................. 381
Figure 8-63: Thermal conductivity, k, of gaseous Neon -at a pressure of one
atmosphere (1,013x105 Pa)- vs. temperature, T. Calculated curves and
experimental points are from different sources. From Johnson (1961) [109]. ... 381

17
ECSSEHB3101Part14A
5December2011
Figure 8-64: Dynamic viscosity, , of several gases -at a pressure of one atmosphere
(1,013x105 Pa)- vs. temperature, T. All the data are from Johnson (1961)
[109] except those corresponding to Argon and Carbon Dioxide which are
from Hilsenrath et al. (1960), and those from Ammonia which are from
Raznjevic (1970) [190]....................................................................................... 382
Figure 8-65: Dynamic viscosity, , of Gaseous Helium-4 and helium-3 vs. temperature,
T. Calculated accounting for quantum-mechanical effects.
Classical Mechanics calculations. From Keller (1969) [119]................... 383
Figure 8-66: Dynamic viscosity, , of Liquid Helium-3, at several pressures, vs.
temperature, T. p = p sat , p = 0,23x105 Pa. From Conte (1970) [48]
Shaded region enclosed experimental points for p = psat. From Keller
(1969) [119]. The line has been calculated from Hone (1962), p = 0. .............. 384
Figure 8-67: Prandtl number, Pr = cp/k, of several gases -at a pressure of one
atmosphere (1,0135x105 Pa)- vs. temperature, T. All the data are from
Hilsenrath et al. (1960) [90] except those corresponding to Methane,
Helium-4, Ammonia and Neon which have been calculated by the compiler. ... 385
Figure 8-68: Coefficient of linear thermal expansion, , of Solid Argon vs. temperature,
T. From Johnson (1961) [109]. .......................................................................... 386
Figure 8-69: Coefficient of volumetric thermal expansion (isobaric compressibility), , of
Liquid Nitrogen vs. temperature, T. From Johnson (1961) [109]. ...................... 386
Figure 8-70: Coefficient of volumetric thermal expansion (isobaric compressibility), , of
Liquid Helium-3 -at pressures near psat- vs. temperature, T. Curves come
from a wide variety of sources, see Keller (1969) [119]. Notice that is
negative in the pressure and temperature ranges under consideration. See
also Figure 8-71 for values of vs. T at the melting curve. ............................... 387
Figure 8-71: Coefficient of volumetric thermal expansion (isobaric compressibility), , of
Liquid Helium-3 -at the melting curve- vs. temperature, T. From Straty
(1966) [228]. ......................................................................................................388
Figure 8-72: Coefficient of volumetric thermal expansion (isobaric compressibility), , of
Solid-Helium-3 vs. temperature, T. Numbers on the curves are densities in
kg.m3. From Straty (1966) [228]. See also Figure 8-14. ................................... 388
Figure 8-73: Surface tension, , of Saturated Liquid Argon vs. temperature, T. From
Johnson (1961) [109].........................................................................................389
Figure 8-74: Surface tension, , of Saturated Liquid Methane vs. temperature, T. From
Johnson (1961) [109].........................................................................................389
Figure 8-75: Surface tension, , of Saturated Liquid Ethane vs. temperature, T. From
Vargaftik (1975) [253]. .......................................................................................389
Figure 8-76: Surface tension, , of Saturated Liquid Carbon Dioxide vs. temperature, T.
From Kutateladze et al. (1966) [127]. ................................................................ 390
Figure 8-77: Surface tension, , of Saturated Liquid Normal Hydrogen vs. temperature,
T. From Johnson (1961) [109]. .......................................................................... 390
Figure 8-78: Surface tension, , of Saturated Liquid Helium-4 vs. temperature, T. From
Johnson (1961) [109].........................................................................................390
Figure 8-79: Surface tension, , of Saturated Liquid Nitrogen vs. temperature, T. From
Johnson (1961) [109].........................................................................................391

18
ECSSEHB3101Part14A
5December2011
Figure 8-80: Surface tension, , of Saturated Liquid Ammonia vs. temperature, T. From
Dunn & Reay (1976) [61]. ..................................................................................391
Figure 8-81: Surface tension, , of Saturated Liquid Neon vs. temperature, T. From
Johnson (1961) [109].........................................................................................391
Figure 8-82: Temperature-entropy diagrams for normal cryogens. More details are
given in Figure 8-83 to Figure 8-94.................................................................... 394
Figure 8-83: Temperature, T, entropy, s, diagram for Argon. From Vargaftik (1975)
[253]. Tabulated values up to T = 1300 K and p = 108 Pa are given in pp.
543 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 395
Figure 8-84: Temperature, T, entropy, s, diagram for Methane. From Vargaftik (1975)
[253]. Tabulated values up to T = 1000 K and p = 108 Pa are given in pp.
211 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 396
Figure 8-85: Temperature, T, entropy, s, diagram for Ethane. From Vargaftik (1975)
[253]. Tabulated values up to T = 500 K and p = 5x107 Pa are given in pp.
225 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 397
Figure 8-86: Temperature, T, entropy, s, diagram for Carbon Dioxide. From Angus,
Armstrong & de Reuck (1976) [5]. Tabulated values up to T = 1100 K and p
= 108 Pa are given in pp. 84 and ff. of the mentioned source. Saturation
curve and typical isobars and isochores are shown in the figure. ..................... 398
Figure 8-87: Temperature, T, entropy, s, diagram for Normal Hydrogen. From Vargaftik
(1975) [253]. Tabulated values up to T = 500 K and p = 108 Pa are given in
pp. 8 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 399
Figure 8-88: Temperature, T, entropy, s, diagram for Parahydrogen. From Vargaftik
(1975) [253]. Tabulated values up to T = 500 K and p = 108 Pa are given in
pp. 9, and 16 and ff. of the mentioned source. Saturation curve and typical
isobars and isochores are shown in the figure. ................................................. 400
Figure 8-89: Temperature, T, entropy, s, diagram for Helium-4. From Angus & de
Reuck (1977) [253]. Tabulated values up to T = 1400 K and p = 7x107 Pa
are given in pp. 64 and ff. of the mentioned source. Saturation curve and
typical isobars and isochores are shown in the figure. ...................................... 401
Figure 8-90: Temperature, T, entropy, s, diagram for Helium-3. From Conte (1970) [48].
Only the saturation curve is shown in this figure. .............................................. 402
Figure 8-91: Temperature, T, entropy, s, diagram for Nitrogen. From Vargaftik (1975)
[253]. Tabulated values up to T = 1300 K and p = 108 Pa are given in pp.
433 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 403
Figure 8-92: Temperature, T, entropy, s, diagram for Ammonia. From Vargaftik (1975)
[253]. Tabulated values up to T = 560 K and p = 1,1x108 Pa are given in pp.
464 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 404
Figure 8-93: Temperature, T, entropy, s, diagram for Neon. From Vargaftik (1975)
[253]. Tabulated values up to T = 300 K and p = 2x107 Pa are given in pp.
536 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 405

19
ECSSEHB3101Part14A
5December2011
Figure 8-94: Temperature, T, entropy, s, diagram for Oxygen. From Vargaftik (1975)
[253]. Tabulated values up to T = 1300 K and p = 108 Pa are given in pp.
477 and ff. of the mentioned source. Saturation curve and typical isobars
and isochores are shown in the figure. .............................................................. 406
Figure 8-95: p-v diagram of a single-component system, illustrating the Maxwell equal
area rule.............................................................................................................410
Figure 8-96: The five regions I to V. a) in p, T coordinated, b) in , T coordinates. ............. 428
Figure 8-97: Concentration of normal fluid, n/, as a function of temperature, T, for
bulk He II. From Andronikashvili (Mendelssohn (1960) [148]). From
values of s/ calculated under the assumption the superfluid critical
velocity is temperature independent (van Alphen et al. (1969) [246])................ 447
Figure 8-98: Entropy per unit mass, s, of liquid Helium II as a function of temperature,
T. From> Dimotakis & Broadwell (1973) [58]. From Broz &
Khorana (1976) [31]. From fountain-effect measurements through 5x107 m
pore-size filters. .................................................................................................448
Figure 8-99: Dynamic viscosity, n, of the normal fluid as a function of temperature, T.
From Heikkila & Hollis Hallet (1955) [84]. From Staas, Taconis & van
Alphen (1961) [227]. From Woods & Hollis Hallet (1963) [267]. .................... 449
Figure 8-100: He3 to He4 vapor pressure ratio, psatHe3/psatHe4, vs. temperature, T.
Calculated by the compiler after data tabulated in Mendelssohn (1960) . ......... 451
Figure 8-101: Specific heat, c, of Silver, Beryllium, Nickel, Stainless Steel and Titanium
vs. temperature, T. Details concerning these materials are given below........... 453
Figure 8-102: Specific heat, c, of Aluminium and Copper vs. temperature, T. Details
concerning these materials are given below...................................................... 454
Figure 8-103: Thermal conductivity, k, of Silver, Beryllium, Nickel, Stainless Steel and
Titanium vs. temperature, T. Details concerning these materials are given
below. ................................................................................................................455
Figure 8-104: Thermal conductivity, k, of Aluminium and Copper vs. temperature, T.
Details concerning these materials are given below.......................................... 456
Figure 8-105: Thermal conductivity, k, of Aluminium alloys vs. temperature, T. Details
concerning these materials are given below...................................................... 458
Figure 8-106: Thermal conductivity, k, of several Copper alloys vs. temperature, T.
Details concerning these materials are given below.......................................... 459
Figure 8-107: Total fractional expansion, L/L293 = (L293L)/L293, and coefficient of
expansion, (1/L)(dL/dT), of several metallic materials vs. temperature, T.
Details concerning these materials are given below.......................................... 460
Figure 8-108: Ultimate tensile strength, ult, of Aluminium and Copper vs. temperature,
T. Details concerning these materials are given below. .................................... 462
Figure 8-109: Ultimate tensile strength, ult, of Aluminium alloys vs. temperature, T.
Details concerning these materials are given below.......................................... 463
Figure 8-110: Ultimate tensile strength, ult, of a representative copper alloy vs.
temperature, T. Details concerning these materials are given below. ............... 464
Figure 8-111: Ultimate tensile strength, ult, of Titanium, Titanium alloys and Stainless
Steel vs. temperature, T. Details concerning these materials are given
below. ................................................................................................................465

20
ECSSEHB3101Part14A
5December2011
Figure 8-112: Specific heat, c, vs. temperature, T, of several Glass-Reinforced
Composites. From Kasen (1975)a [115]............................................................ 468
Figure 8-113: Thermal conductivity, k, vs. temperature, T, of several Glass-Cloth
Reinforced Composites. From Kasen (1975)a [115]. ........................................ 469
Figure 8-114: Thermal conductivity, k, in the plane vs. temperature, T, of several Glass-
Fiber Reinforced Composites. From Kasen (1975)a [115]. ............................... 470
Figure 8-115: Thermal expansion, L/L, vs. temperature, T, of several Glass-Fiber
Reinforced Composites. From Kasen (1975)a [115]. ........................................ 471
Figure 8-116: Thermal expansion, L/L, normal to the plane, vs. temperature, T, of
several Glass-Cloth Reinforced Composites. From Kasen (1975)a [115]. ........ 472
Figure 8-117: Thermal expansion, L/L, in the plane, vs. temperature, T, of several
Glass-Cloth Reinforced Composites. From Kasen (1975)a [115]...................... 473
Figure 8-118: Ultimate tensile strength, ult, vs. temperature, T, of several Glass-
Reinforced Composites. From Kasen (1975)a [115]. ........................................ 474
Figure 8-119: Ultimate compressive strength, ult, vs. temperature, T, of several Glass-
Reinforced Composites. From Kasen (1975)a [115]. ........................................ 475
Figure 8-120: Ultimate flexural strength, ult, vs. temperature, T, of several Glass-
Reinforced Composites. From Kasen (1975)a [115]. ........................................ 476
Figure 8-121: Specific heat, c, vs. temperature, T, of several Advanced Composites.
From Kasen (1975)b [116]. Details concerning these composites are given
below. ................................................................................................................477
Figure 8-122: Thermal conductivity, k, in the plane, vs. temperature, T, of several
Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 478
Figure 8-123: In plane, longitudinal thermal expansion, L/L, vs. temperature, T, of
several Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below .................................................. 480
Figure 8-124: In plane, transverse thermal expansion, L/L, vs. temperature, T, of
several Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 482
Figure 8-125: Ultimate tensile strength, ult, in the plane, vs. temperature, T, of several
Uniaxial Graphite-Epoxy Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 484
Figure 8-126: Ultimate tensile strength, ult, in the plane, vs. temperature, T, of several
Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 486
Figure 8-127: Ultimate compressive strength, ult, in the plane, vs. temperature, T, of
several Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below .................................................. 488
Figure 8-128: Ultimate flexural strength, ult, in the plane, vs. temperature, T, of several
Uniaxial Graphite-Epoxy Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 489
Figure 8-129: Ultimate flexural strength, ult, in the plane, vs. temperature, T, of several
Uniaxial Advanced Composites. From Kasen (1975)b [116]. Details
concerning these composites are given below. ................................................. 491

21
ECSSEHB3101Part14A
5December2011
Figure 8-130: Cost of several fibers (for orders above 50 kg). From Greszczuk et al
(1975) [78]. ........................................................................................................492
Figure 8-131: Thermal conductivity, k, vs. temperature, T, of several structural tubes.
From Foster, Naes & Barnes (1975) [70]. Details concerning these tubes
are given below..................................................................................................493
Figure 8-132: Mechanical and thermal properties of Bone. From Wipf & Gibney (1975)
[266]. a) Ultimate compressive strength, ult, vs. temperature, T. Values
from Armstrong et al. (1971) [9]. Values from Wipf & Gibney (1975)
[266]. b) Thermal conductivity, k, vs. temperature, T........................................ 495
Figure 9-1: Ignition temperature, Ti, of several metals as a function of Oxygen
pressure, p. From Clarck & Hust (1974) [45]. .................................................... 506
Figure 9-2: Flash point temperature, Tfl, of several nonmetals as a function of Oxygen
pressure, p. Also shown the ignition temperature, Ti, of the same materials
at high pressure. From Clarck & Hust (1974) [45]. ............................................ 506
Figure 9-3: Effect of pressure in the limits of flammability of natural gas-air mixtures
ignited by means of an induction coil. From Lewis & von Elbe (1961) [132]...... 509
Figure 9-4: Minimum ignition energy, E, and quenching distance, D, for hydrogen-
oxygen-inert gas mixtures at atmospheric pressures. From Lewis & von
Elbe (1961) [132]. ..............................................................................................510
Figure 9-5: Dependence of impact energy, J, on temperature, T, for several materials.
From Edeskuty, Reider, Williamson (1971) [62]. ............................................... 512
Figure 9-6: Stress intensity at crack arrest, KTH, for AISI 4340 in Hydrogen at ambient
temperature, as a function of Hydrogen pressure. From Chandler & Walter
(1975) [41]. ........................................................................................................527
Figure 9-7: Fracture ductility, measured by the tensile reduction in area at fracture, RA,
for several steels, as a function of Hydrogen content, CH2. From Johnson &
Kumnick (1975) [107].........................................................................................528

Tables
Table 4-1: NASA Mission Categories Requiring Cryogenic Cooling in Space ....................... 42
Table 4-2: Spacecraft Cryogenic Cooling Techniques ........................................................... 43
Table 5-1: Development Problem Areas of Brayton Cycle Rotary-Reciprocating
Refrigerators ........................................................................................................53
Table 5-2: Development Problem Areas of Brayton and Claude Cycle Refrigerators............ 55
Table 5-3: Potential Problem Areas Associated with the Stirling Refrigerators...................... 65
Table 5-4: Development Problem Areas of Vuillemier Cycle Refrigerators............................ 68
Table 5-5: Maximum Inversion Temperature and Pressure of Selected Cryogens.............. 107
Table 6-1: Benefit Obtained from a VCS System in a Storage Container............................ 122
Table 6-2: Relaxation of the Restrictions Involved in the Idealized Model ........................... 123
Table 6-3: (Effective) Thermal Conductivity of Several MLIs vs. Temperature. ................... 126
Table 6-4: Corrective Factor, n, Giving the Influence of the Number, n, of Conductive
Shields on the Boil-off Rate of Several Cryogens. (k = k )............................... 133

22
ECSSEHB3101Part14A
5December2011
Table 6-5: Cooled Shield Temperatures, Ti, and Dimensionless Shield Positions, i,
Which Minimize the Boil-off Rate for Several Cryogens. (k = k1T)..................... 135
Table 6-6: Cooled Shield Temperatures, Ti, and Dimensionless Shield Positions, i,
Which Minimize the Boil-off Rate for Several Cryogens. (k = k1T)..................... 135
Table 6-7: Cooled Shield Temperatures, Ti, and Dimensionless Shield Positions, i,
Which Minimize the Boil-off Rate for Several Cryogens. (k = k1T)..................... 136
Table 6-8: Corrective Factor, nk, Giving the Influence of the Number, n, of Conductive
Shields on the Boil-off Rate of Several Cryogens. (k = k1T). ............................. 138
Table 6-9: Figure of Merit, /k, of Several Tensile Support Materials at Cryogenic
Temperatures a ..................................................................................................164
Table 6-10: Main Features of Separating Systems for VCS Dewars ................................... 177
Table 6-11: Phase Separating Systems...............................................................................178
Table 6-12: Experimental Conditions for Capillary Barrier Stability Studies (Figure
6-37). .................................................................................................................185
Table 6-13: Sources of Data for Calculating Bond Lengths ................................................. 209
Table 6-14: Compatibility of Materials with Ammonia. Non-metals. ..................................... 212
Table 6-15: Characteristics of the Lockheed VCS Dewar .................................................... 222
Table 6-16: Predicted Heat Loads........................................................................................ 225
Table 7-1: Several Attempts to Experimentally Verify the Bernoulli Thinning (BT) .............. 265
Table 7-2: Kapitza Conductance, hk, of Metals in Contact with Liquid Helium (He II)............. 276
Table 7-3: Kapitza Conductance, hk, of Mercury 72 in Contact with Liquid Helium (He II)... 277
Table 7-4: Sample Description of Mercury 72 in Table 7-3 .................................................. 277
Table 7-5: Kapitza Conductance, hk, of Lead 105 in Contact with Liquid Helium (He II)...... 278
Table 7-6: Sample Description of Lead 105 in Table 7-5 ..................................................... 279
Table 7-7: Kapitza Conductance, hk, of Indium 108 in Contact with Liquid Helium (He II)... 280
Table 7-8: Sample Description of Indium 108 in Table 7-7 .................................................. 281
Table 7-9: Kapitza Conductance, hk, of Gold 164 in Contact with Liquid Helium (He II) ...... 281
Table 7-10: Sample Description of Gold 164 in Table 7-9 ................................................... 281
Table 7-11: Kapitza Conductance, hk, of Tin 199 in Contact with Liquid Helium (He II)....... 282
Table 7-12: Sample Description of Tin 199 in Table 7-11 .................................................... 282
Table 7-13: Kapitza Conductance, hk, of Platinum 240 in Contact with Liquid Helium
(He II).................................................................................................................282
Table 7-14: Sample Description of Platinum 240 in Table 7-13 ........................................... 283
Table 7-15: Kapitza Conductance, hk, of Silver 225 in Contact with Liquid Helium (He II)... 283
Table 7-16: Sample Description of Silver 225 in Table 7-15 ................................................ 284
Table 7-17: Kapitza Conductance, hk, of Palladium 274 in Contact with Liquid Helium
(He II).................................................................................................................284
Table 7-18: Kapitza Conductance, hk, of Niobium 275 in Contact with Liquid Helium (He
II)........................................................................................................................285
Table 7-19: Sample Description of Niobium 275 in Table 7-18............................................ 285

23
ECSSEHB3101Part14A
5December2011
Table 7-20: Kapitza Conductance, hk, of Copper 343 in Contact with Liquid Helium (He
II)........................................................................................................................286
Table 7-21: Sample Description of Copper 343 in Table 7-20 ............................................. 288
Table 7-22: Kapitza Conductance, hk, of Tungsten 400 in Contact with Liquid Helium
(He II).................................................................................................................291
Table 7-23: Sample Description of Tungsten 400 in Table 7-22 .......................................... 292
Table 7-24: Kapitza Conductance, hk, of Aluminium 428 in Contact with Liquid Helium
(He II).................................................................................................................292
Table 7-25: Sample Description of Aluminium 428 in Table 7-24 ........................................ 293
Table 7-26: Kapitza Conductance, hk, of Nickel 450 in Contact with Liquid Helium (He
II)........................................................................................................................294
Table 7-27: Sample Description of Nickel 450 in Table 7-26 ............................................... 295
Table 7-28: Kapitza Conductance, hk, of Molybdenum 450 in Contact with Liquid
Helium (He II)..................................................................................................... 295
Table 7-29: Sample Description of Molybdenum 450 in Table 7-28 .................................... 296
Table 7-30: Kapitza Conductance, hk, of Beryllium 1440 in Contact with Liquid Helium
(He II).................................................................................................................296
Table 7-31: Sample Description of Beryllium 1440 in Table 7-30 ........................................ 297
Table 7-32: Kapitza Conductance, hk, of Nonmetals in Contact with Liquid Helium (He
II)........................................................................................................................297
Table 7-33: Kapitza Conductance, hk, of Graphite 420 in Contact with Liquid Helium
(He II).................................................................................................................297
Table 7-34: Sample Description of Graphite 420 in Table 7-33 ........................................... 298
Table 7-35: Kapitza Conductance, hk, of Quartz (SiO2) 470 in Contact with Liquid
Helium (He II)..................................................................................................... 298
Table 7-36: Sample Description of Quartz (SiO2) 470 in Table 7-35................................... 298
Table 7-37: Kapitza Conductance, hk, of Silicon 636 in Contact with Liquid Helium (He
II)........................................................................................................................299
Table 7-38: Sample Description of Silicon 636 in Table 7-37 .............................................. 299
Table 7-39: Kapitza Conductance, hk, of Lithium Fluoride 730 in Contact with Liquid
Helium (He II)..................................................................................................... 300
Table 7-40: Sample Description of Lithium Fluoride 730 in Table 7-39 ............................... 300
Table 7-41: Kapitza Conductance, hk, of Metals in Contact with Low Acoustiv
Impedance Media (LAIM) .................................................................................. 300
Table 7-42: Kapitza Conductance, hk, of Indium 108 in Contact with Low Acoustic
Impedance Media (LAIM) .................................................................................. 301
Table 7-43: Sample Description of Indium 108 in Table 7-42 .............................................. 301
Table 7-44: Kapitza Conductance, hk, of Copper 343 in Contact with Low Acoustic
Impedance Media (LAIM) .................................................................................. 302
Table 7-45: Sample Description of Copper 343 in Table 7-44 ............................................. 302
Table 7-46: Variables Characterizing the Porous Media...................................................... 331

24
ECSSEHB3101Part14A
5December2011
Table 7-47: Expressions Relating the Permeability, K, to Geometrical Properties of the
Porous Medium. All expressions in this Table are from Bear (1972) [19].......... 333
Table 7-48: Data on Porous Media ...................................................................................... 335
Table 8-1: Thermodynamic and Transport Properties of Normal Cryogens......................... 344
Table 8-2: Thermodynamic and Transport Properties of Normal Cryogens......................... 345
Table 8-3: Thermodynamic and Transport Properties of Normal Cryogens......................... 347
Table 8-4: Entropy Departure of Saturated Liquid and Vapor .............................................. 407
Table 8-5: Values of the Coefficient ai for Argon .................................................................. 413
Table 8-6: Values of the Coefficients ai, n and n for Methane ........................................... 417
Table 8-7: Values of the Coefficients ai, n and n for Ethane ............................................. 421
Table 8-8: Values of the Dimensionless Coefficients nij, an and bn for Carbon Dioxide ....... 424
Table 8-9: Values of the Dimensionless Coefficients nij and for Helium-4 ......................... 428
Table 8-10: Values of the Coefficient ai for Nitrogen ............................................................ 432
Table 8-11: Values of the Coefficient ai for Neon ................................................................. 438
Table 8-12: Values of the Coefficient ai for Oxygen ............................................................. 442
Table 8-13: Relevant Properties of He II as a Function of Temperature.............................. 444
Table 9-1: Relevant properties of Cryogen ..........................................................................499
Table 9-2: Several Useful Definitions ................................................................................... 501
Table 9-3: Thermal Data of Relevant Materials in an Oxidizing Atmosphere....................... 503
Table 9-4: Ranking of Materials for Oxygen Compatibility ................................................... 507
Table 9-5: Properties of H2 and CH4 Related to their Combustion Hazards ........................ 508
Table 9-6: Symptoms of O2 Deficiency.................................................................................511
Table 9-7: Sources of Mechanical Properties of Structural Alloys at Cryogenic
Temperatures. ................................................................................................... 512
Table 9-8: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(ALUMINIUM ALLOYS) AT CRYOGENIC TEMPERATURES .......................... 513
Table 9-9: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(IRON BASE ALLOYS) AT CRYOGENIC TEMPERATURES ........................... 515
Table 9-10: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(NICKEL BASE ALLOYS) AT CRYOGENIC TEMPERATURES ....................... 516
Table 9-11: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(STAINLESS STEELS, AUSTENITIC) AT CRYOGENIC TEMPERATURES.... 517
Table 9-12: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(STAINLESS STEELS, MARTENSITIC) AT CRYOGENIC
TEMPERATURES .............................................................................................519
Table 9-13: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(STEELS) AT CRYOGENIC TEMPERATURES................................................ 519
Table 9-14: SOURCES OF MECHANICAL PROPERTIES OF STRUCTURAL ALLOYS
(TITANIUM ALLOYS) AT CRYOGENIC TEMPERATURES.............................. 521
Table 9-15: Susceptibiliy of Metals to Hydrogen Embrittlement as Measured by Tensile
Tests a ................................................................................................................523

25
ECSSEHB3101Part14A
5December2011

1
Scope

In this part Part 14 cooling methods below 100 K are described. These low temperature levels are
mainlyrequiredbyspaceborneelectronicsystemsoperatingunderverylownoiseconditions.Details
onthematerialsusedandsafetyfactorsaregiven.

TheThermaldesignhandbookispublishedin16Parts
ECSSEHB3101Part1 ThermaldesignhandbookPart1:Viewfactors
ECSSEHB3101Part2 ThermaldesignhandbookPart2:Holes,GroovesandCavities
ECSSEHB3101Part3 ThermaldesignhandbookPart3:SpacecraftSurfaceTemperature
ECSSEHB3101Part4 ThermaldesignhandbookPart4:ConductiveHeatTransfer
ECSSEHB3101Part5 ThermaldesignhandbookPart5:StructuralMaterials:Metallicand
Composite
ECSSEHB3101Part6 ThermaldesignhandbookPart6:ThermalControlSurfaces
ECSSEHB3101Part7 ThermaldesignhandbookPart7:Insulations
ECSSEHB3101Part8 ThermaldesignhandbookPart8:HeatPipes
ECSSEHB3101Part9 ThermaldesignhandbookPart9:Radiators
ECSSEHB3101Part10 ThermaldesignhandbookPart10:PhaseChangeCapacitors
ECSSEHB3101Part11 ThermaldesignhandbookPart11:ElectricalHeating
ECSSEHB3101Part12 ThermaldesignhandbookPart12:Louvers
ECSSEHB3101Part13 ThermaldesignhandbookPart13:FluidLoops
ECSSEHB3101Part14 ThermaldesignhandbookPart14:CryogenicCooling
ECSSEHB3101Part15 ThermaldesignhandbookPart15:ExistingSatellites
ECSSEHB3101Part16 ThermaldesignhandbookPart16:ThermalProtectionSystem

26
ECSSEHB3101Part14A
5December2011

2
References

ECSSSST0001 ECSSSystemGlossaryofterms
ECSSEHB3101Part5 ThermaldesignhandbookPart5:StructuralMaterials:Metallicand
Composite
ECSSEHB3101Part7 ThermaldesignhandbookPart7:Insulations
ECSSEHB3101Part8 ThermaldesignhandbookPart8:HeatPipes
ECSSEHB3101Part9 ThermaldesignhandbookPart9:Radiators
ECSSEHB3101Part13 ThermaldesignhandbookPart13:FluidLoops

AllotherreferencesmadetopublicationsinthisPartarelisted,alphabetically,intheBibliography.

27
ECSSEHB3101Part14A
5December2011

3
Terms, definitions and symbols

3.1 Terms and definitions


ForthepurposeofthisStandard,thetermsanddefinitionsgiveninECSSSST0001apply.

3.2 Abbreviated terms


ThefollowingabbreviatedtermsaredefinedandusedwithinthisStandard

A aged

AAD averageabsolutedeviation

AC aircooled

alternatingcurrent

Ann annealed

AnnA annealedandaged

APS activephaseseparator

BLIP backgroundlimitedinfraredphotoconductor

CHRESUS superfluidheliumcryostatforspaceuse

CTFE polychlorotrifluorethylene

CW coldworked

DA doubleaged

DTA detectcapsuleassembly

DC directcurrent

EMI electromagneticinterference

GIRL Germaninfraredlaboratory

GR generationrecombination

28
ECSSEHB3101Part14A
5December2011

GRS glassfabricandsiliconerubber

HEOA highenergyastronomicalobservatory

HIP hotisostaticpressed

IR infrared

JT JouleThomson

LAIM lowacousticimpedancemedium

LNG liquifiednaturalgas

MLI multilayerinsulation

NDT nilductilitytransition

NEP noiseequivalentpower

RMS rootmeansquare

RRR residualresistanceratio

SHFE superfluidheliuminzerogsetofexperiments

OFHC oxygenfreehighconductivity

QCM quartzcrystalmicrobalance

Q quenched

QA quenchedandaged

RevAnn reverseannealed

SA solutionannealed

SHFE superfluidheliuminzerogsetofexperiments

SPT standardpressureandtemperature

SS stainlesssteel

ST solutiontreated

STA solutiontreatedandaged

STDA solutiontreatedanddoubleaged

STQA solutiontreated,quenchedandaged

STQW solutiontreated,waterquenched

TEMPER tempered

29
ECSSEHB3101Part14A
5December2011

VCS vaporcooledshield

WQ waterquenched

bcc cubicbodycentered

ccp cubicclosepacked

ed electropolishingdepth

hcp hexagonalclosepacked

id innerdiameter

ppm partspermillion

3.3 Symbols
A Clause6:insulationarea,[m2]itismeasurednormally
tothemaintemperaturegradient

Clause7:crosssectionalarea.[m2]

Clauses7and8:Gorter&Mellinkconstant,[m.s.kg1]

A Helmholtzpotential,[m2.s2]itisdefinedinclause
12.1.2.2

AFL internalcrosssectionalareaofaDdctoraporous
plug.[m2].Alsocalledfreeflowarea.Inthecaseofa
porousplug,AFL=AFR.

AFR frontalareaofaporousplug,[m2]

Ad detectorarea,[m2]

Ap systemarea,[m2]

As supportcrosssectionalarea,[m2]

B dimensionlessforceparameterinVinenstheory

Bo bondnumber,Bo=glR/

Ca capillarynumber,Ca=V/

Co Besselcylinderfunctionoforderzero

D Clauses6and8:diameter,[m]

Clause9:quenchingdistance,[m]

DE equivalentorhydraulicdiameterofaductofnon

30
ECSSEHB3101Part14A
5December2011

circularcrosssection,[m]

Dj dimensionlessheatconductionfactorforthe
insulationlayer,itisdefinedinEq.(10.11)

D* detectivityofadetector,[m.W1.s1/2]

E minimumignitionenergy,[J]

Ec Eckertnumber,Ec=p/(V2)

F fluiddynamicimpedanceofaduct,[m.s]

Fr froudenumber,Fr=V2/(gl)

Fl auxiliaryfunction,itisdefinedinEq.[770]

Fn frictionforceperunitvolumebetweenthenormal
fluidandthewall,[N.m3]

Fp fluiddynamicimpedanceofaporousplug,[m.s]

Fs frictionforceperunitvolumebetweenthesuperfluid
andthewall,[N.m3]

Fs fsaboveplusthelossofmomentumatthetubeexit,
[N.m3]

Fsn mutualfrictionforceperunitvolumebetween
superfluidandnormalcomponent,[N.m3]

Gr Grashofnumber,Gr=Re/Fr

J impactenergyabsorbed,[J]

J(CU) charpyUnotchimpactenergyabsorbed,[J]

J(CV) charpyVnotchimpactenergyabsorbed,[J]

JC Jintegraltoughnessbasedonmaximumload,[J.m2]

JIC criticalvalueofJintegralforcrackinitiation,[J.m2]

J Besselfunctionoffirstkindandorder

Jintegral integraltoughness,[J.m2]

Ji uppervalueofsubscriptj

K permeabilityorgeometricfactororaporousmedium,
[m2]

KD dynamicfracturetoughness,[Pa.m1/2]

KIC planestrainfracturetoughness,[Pa.m1/2]

31
ECSSEHB3101Part14A
5December2011

KIC(J) KICcomputedfromJIC,[Pa.m1/2]

KTH stressintensityfactoratcrackarrest,[Pa.m1/2]

K stressintensityfactor,[Pa.m1/2]isisbasedontheload
correspondingto5%secantline

Kc fracturetoughness,[Pa.m1/2]

Kmax Stressintensityfactor,[Pa.m1/2]itisbasedonthe
maximumloadatfailure

Kt stressconcentrationfactor

L length,[m]

Lb bondlength,Lb=[/(g)]1/2

Lv lengthperunitfluidvolumeofvortexlineinVinens
theory,[m2]

L293 lengthatT=293K,[m]

M Clause7:specificsurfaceoftheporousmedium,[m1]
itisdefinedinTable746.

Clauses7and8:molarmass,[kg.mol1]

Mp refrigeratingsystemmassperunitofrefrigeration
power,[Kg.W1]

Ms specificareaoftheporousmedium,basedonthe
apparentvolumeofthemedium,[m1]itisdefinedin
Table746.

N Clause6:numberofshieldsinanMLI

Clause8:numberoflayersinacompositematerial

Clause9:numberofloadcyclesinfatigue

Nu Nusseltnumber,Nu=hDE/k

Op Ratioofopentototalareaofacapillarybarrier

P Clause6:mechanicalload,[N]

Clause7:percentageofmaterialbetweenadjacent
sieves,itisintroducedinTable747.

Pr Prandtlnumber,Pr=cp/k

Q heattransferrate,[W]

Ql heattransferratefromtheheatsource,[W]alsocalled

32
ECSSEHB3101Part14A
5December2011

heatload

Qm heattransferratethroughtheinsulation,[W]

Qs heattransferratethroughthesupports,[W]

R Clause6:characteristiclengthalongtheinsulation,
[m]tankradius,[m]

Clauses7and8:gasconstantofaparticulargas,[m2.s
2.K1]

Ro localradiusoftheeffectivevaneprofile,[m]

R1,R2 principalradiiofcurvatureofaninterface,[m]

R universalgasconstant,R=8,31432J.K1.mol1

RA tensilereductioninarea

Re Reynoldsnumber,Re=VD/

S Cryogensensibility,S=[cp(THTC)]/hfg

T temperature,[K]

TB normalboilingtemperature,[K]

TM meltingtemperature,[K]

Tb fluidbulktemperature,[K]itisdefinedinthelistof
symbolsofECSSEHB3101

Tfl flashpoint,[K]

Ti Clause6:temperatureoftheithvaporcooledshield,
[K]

Clause9:ignitiontemperature,[K]

Tr reducedtemperature,Tr=T/TC

Tw walltemperature,[K]

U Clause6:ullage

Clause7:potentialofanyconservativebodyforceper
unitvolume,[m2.s2]

UPE unitpropagationenergy,[J.m2]

V Clause5:volume,[m3]

Clause6:meanfluidvelocity,[m.s1]

33
ECSSEHB3101Part14A
5December2011

We Webernumber,We=V2L/

X,Y strainedcoordinates

X(T) bonMardionetal.function,X(T)=L(T)q3,4

Y inversedimensionlessStokeslayerthickness

Y Besselfunctionofsecondkindandorder

Z compressibilityfactor,itisdefinedinclause8.1.2.1

a cracklength,[m]

ao Vortexcoreradius,[m]

b timeconstant,[s1]

c Clause4:speedoflightinvacuum,c=2,9978x108
m.s1

Clause8:specificheatofasolid,[J.kg1.K1]

Clause9:coefficientinfatiguecrackgrowthrate
equation(Parisequation),[m1n/2.Pan.cycle1]

cH2 Hydrogencontent,[m3.kg2]

co numericalfactorinKozenysequationTable747.

cp constantpressurefluidspecificheat,[J.kg1.K1]for
gasescpisusedindistinctiontocv,theconstant
volumespecificheat,forliquidsnosuchdistinctionis
needed,neverthelesscpwillbeused

d Clauses6and7:diameter,[m]

Clause7:particlediameter,[m]itisdefinedinTable
746.

dm meanparticlesize,[m]Table747.

dv porediameter,[m]

dl vectorelementofstreamline,[m]

f Clause4:diametertofocallengthratioofalens,f1is
usuallyknownaslensaperture

Clause7:forceperunitlengthonavortexlinein
Vinenstheory,[N.m1]fanningfrictionfactor,itis
definedinECSSEHB3101,clause7.2.2

g accelerationduetogravity,[m.s2]

34
ECSSEHB3101Part14A
5December2011

h Clause15Plancksconstant,h=1,0545x1034J.s

Clauses5and8:specificenthalpy,[J.kg1]

Clause6:convectiveheattransfercoefficient,[W.m
2.K1]

Clause7:thermalconductanceoftheporousmedium
plusthefluidfillingitscavities,[W.K1]

Clause8:heatofconversionfromnormaltopara
Hydrogen,[J.kg1]

hk Kapitzaconductance,[W.m2.K1]

hfg heatofvaporizationorsublimation,[J.kg1]

hl thermalconductanceoftheliquidwithinaporous
medium,[W.K1]

hq heatofcombustion,[J.kg1]

hs thermalconductanceoftheporousmedium,[W.m1]

massfluxdensity(Vector),[kg.m2.s2]

k Clauses5and8:Boltzmannsconstant,k=1,38054x
1023J.K1

Clauses7and8:thermalconductivity,[W.m1.K1]

kj coefficientintheapproximateexpression,k=kjT.
[W.m1.K(j+1)]

kx thermalconductivityacrosstheinsulation,[W.m1.K1]

ky thermalconductivityalongtheinsulation,[W.m1.K1]

k temperatureaveragedeffectivethermalconductivity
oftheinsulation,[W.m1.K1]
1
k k d
0

l length,[m]heightinthegravitationalfieldabovea
referencelevel,[m]

m Clause6:Cryogenboloffrate,[kg.s1]

Clause7:fluidmassflowrate,[kg.s1]

mHe atomicmassofhelium,mHe=6,6435x1027kg

mo Cryogenboiloffratebecauseoftheheatleaks

35
ECSSEHB3101Part14A
5December2011

throughaconventionalMLIsystem.[kg.s1]
k ATH TC
mo .
h fg t

n Clause6:numberofvaporcooledshields

Clause9:exponentinfatiguecrackgrowthrate
equation(Parisequation)

nB backgroundphotonnumberflux,[s1.m2]

p Clause6:numberofsupports

Clauses5,7,8and9:pressure,[Pa]

pb boilingpressureat300K,[Pa]

pr reducedpressure,pr=p/pc

q tortuosityofaporousmedium,itisdefinedinTable
746.

q heatflux(vector),[W.m2]

qm heatfluxthroughaguardshroudusedinthethermal
testofasupport,[W.m2]

r Clause6:dimensionlessconvectiveheattransfer
coefficient,itisdefinedinEq.[637]

Clause7:roughnessfactor,definedastheratioof
roughtomacroscopicsurfacearea

s specificentropy,[J.kg1.K1]

t thickness,[m]

t time,[s],[d]or[yr]

tl linerwallthickness,[m]

to overwrapthickness,[m]

u fluidvelocity(vector),[m.s1]

Clause7:crosssectionalaverageofthelocalvelocity
u.[m.s1]
Clause8:molarvolume,[m3.kg1]

vd diffusionvelocity(Vector),[m.s1]

vnc2 secondnormalcriticalvelocity,[m.s1]

36
ECSSEHB3101Part14A
5December2011

vsc1 firstsuperfluidcriticalvelocity,[m.s1]

vt relativevelocitybetweennormalfluidandvortexline,
[m.s1]

x Clause6:distancetothecoldfaceoftheinsulation,
[m]itismeasuredacrosstheinsulation

Clause7:axialdistancetotheentryinatubeor
porousplug,[m]

y distancebetweenapointintheinsulationandthe
ventingductaxis,[m]itismeasuredalongthe
insulation

z verticaldistance,[m]

difference

K Clause6:curvaturedifference,[m1]

Clause9:stressintensityfactorrangeduringfatigue
cycling,[Pa.m1/2]

p pressuredrop[Pa]

j jthterminthepowerseriesdevelopmentofthe
innerexpressionofneartheinsulationcoldface.

aspectratio

stretcheddimensionlessdistancefromtheinsulation
coldface,=(ky/kx)1/2(t/R)

Clauses6and7:volumeporosity,itisdefinedinTable
746.

Clause7:Rayleighviscousdissipationfunction,[W.m
3]

Clause6:outerradiusoftheventingductmade
dimensionlesswiththeradius,R,oftheinsulation
surface

Clause7:wallinteractiondimensionlessconstantin
Vinenstheory

Clause7:wallinteractionVanderWaalsconstant,
[kg.m2.s2]thermalaccommodationcoefficient,
dimensionlesstemperature,=TH/TC,particleshape
factorTable747.

Clause9:thermaldiffusivity,[m2.s1]=k/c,wherec
isthespecificheat

37
ECSSEHB3101Part14A
5December2011

fluidisobariccompressibility,[K1]alsoknownas
coefficientofvolumetricthermalexpansion

Clause7:superfluidfilmthickness,[m]

Clause9:tensileelongation

c crackopeningdisplacementatthecracttip
correspondingtotheonsetfracture,[m]

Clauses4:emittance

Clause5:parameterinintermolecularpotential
function,[J]

Clause6:normaltolateralheatconductionparameter,
=(kx/ky)(R/t)2

dimensionlessdistanceacrosstheinsulation
k m Nu

k TH mo

Clause5:efficiency

Clause6:dimensionlessdistancealongtheinsulation,
=y/R

c Carnotefficiency,c=1TC/TH

Clause6:dimensionlesstemperature,=(TTC)/(TTH)
liquidsolidcontactangle[angulardegrees]polar
angle,[angulardegrees]

angularvelocity,[rad.s1]

angularacceleration,[rad.s2]

a spreadingangleofabubblecontactingtwoconcentric
spheres,[angulardegrees]

D Debyetemperature,[K]

strengthofaquantizedvortexring,[m2.s1]

Clause4:wavelength,[m]

Clause7:frictionfactor,itisdefinedinmajorclause
11.2.2

c detectorcutoffwavelength,[m]

Clause5:JouleThomsonparameter,[K.m2.N1]
T / p h .

38
ECSSEHB3101Part14A
5December2011

Clauses6and8:fluiddynamicviscosity,[Pa.s]itis
alsoknownasviscositycoefficient

Clause7:chemicalpotential,[m2.s2]

n Clause6:nthrootofY1(n)J1(n)J1(n)Y1(n)=0,J1
andY1beingtheBesselfunctionsoforderone(first
andsecondkindrespectively)

Clause7:dynamicviscosityofthenormalfluid
component,[Pa.s]

Kinematicviscosity,[m2.s1]

Clause6:dimensionlessdistanceacrosstheinsulation,
=x/t

Clause7:dimensionlessdistancealongatube,=(L
1)/l

density,[kg.m3]

l liquiddensity,[kg.m3]

v gasdensity,[kg.m3]

Clause6:strength,[Pa]

Clauses6,7and8surfacetension,[N.m1]

f fatiguestrength,[Pa]

ult ultimatetensilestrenght,[Pa]

2 tensileyieldstrength,[Pa]2%offset

dimensionlesstemperature,=T/TH

1 dimensionlessvortexgenerationparameterinVinens
theory

2 dimensionlessvortexdecayparameterinVinens
theory

Clause6:dimensionlessheattransferrate,=
Q/(mohfg)

Clause7:superfluidvelocitypotential,[m2.s2]

1 dimensionlessheatload,1=Ql/(mohfg)

frequency,[rad.s1]

Nu factoraccountingfortheinfluenceofthefinite
convectiveheattransferonboiloffrate

39
ECSSEHB3101Part14A
5December2011

i factoraccountingfortheinfluenceoftheithdeviation
fromtheidealmodelontheboiloffrate

k factoraccountingfortheinfluenceofthetemperature
dependent.thermalconductivityontheboiloffrate

n factoraccountingfortheinfluenceofthefinite
numberofvaporcooledshieldsontheboiloffrate

nk factoraccountingforthesimultaneousinfluenceofthe
finitenumberofvaporcooledshieldsandofthe
temperaturedependentthermalconductivityonthe
boiloffrate

y factoraccountingfortheinfluenceofthefinitethermal
conductivityofthevaporcooledshieldsontheboil
offrate

Subscripts
A ambient

C cold

H hot

Nu referstoavaporcooledshieldsystemwhenfinite
convectiveheattransferintheventingductistaken
intoaccount

b fluidbulkproperties

c criticalconditions

eff effective

i Clauses5and7:inletconditions

Clause6:ithelementinasetofseveralelements

j jthelementinasetofseveralelements

k referstoavaporcooledshieldsystemwith
temperaturedependentsthermalconductivity

l liquidconditions

max maximum

min minimum

n Clause6:referstoavaporcooledshieldsystem
havingncooledshields

40
ECSSEHB3101Part14A
5December2011

Clauses7and8:normalcomponentinsuperfluid
heliumflow

nk referstoavaporcooledshieldsystemhavingncooled
shieldsandtemperaturedependentthermal
conductivity

o Clause7:outletconditions

Clauses7and8:referencestate

phon phonon

r reducedvalue

s Clause6:referstosupports

Clauses7and8:superfluidcomponentinsuperfluid
heliumflow

sat conditionsalongthesaturationcurve

v vaporconditions

y Clause6:referstoavaporcooledshieldsystemwhen
heattransferalongthecooledshieldsistakeninto
account

Clause7:ycomponent

z zcomponent

, helium4conditionsatthelambdapoint

Superscripts
o perfectgasconditions

averagedvalue.

* Localboilingconditions.

41
ECSSEHB3101Part14A
5December2011

4
General introduction

ThisPartconcernscoolingtotemperaturesbelow100K.
These low temperature levels are dictated by the use of improved Earthapplications instruments
(promisinghigherresolutionintheinfrared(IR)region),byagrowinginterestininfrared,gammaray
and highenergy astronomy, and by the advent of a large number of opportunities for instrument
launchesbothbyuseoftheShuttle(7to30days)andofunmannedspacecraft(1to5years).
Table41,fromSherman(1978)[216],listsdisciplineswhichrequirecryogeniccoolinginspace,with
estimates of temperatures and cooling loads. The table concerns only NASA involvement, but ESA,
DOD,DFVLRandCNES,amongothers,alsoneedcryogenicsystemsinspace.Thedevelopment,by
ESAofaHeliumIIcryostatforSpacelabpayloadsisdescribedbyLizonTati&Girard(1978)[134].A
similareffortisbeingundertakeninGermanybyDFVLR(Lemke,Klipping&Rmisch(1978)[131],
Seidel (1978) [212]. On the other hand, CNES in France is developing a solid cryogen cooler for
spaceborneIRdetectors(Rolfo&Prost(1978)[198].
The cryogenic cooling techniques that are at present either in use or in development state, or that
promising for the near future, are listed in Table 42, which has been also borrowed from Sherman
(1978)[216].

Table41:NASAMissionCategoriesRequiringCryogenicCoolinginSpace
Discipline Temp.Range[K] CoatingLoad

ApplicationsMissions(Weather,Earth
10to100 Milliwattsto10W
Resources,PollutionMonitoring,etc.)

HighEnergyandGammaRay
4to100 Milliwattsto10W
Astronomy

IRAstronomy 0.3to10 Microwattsto0,1W

RelativityMissions 0,001to1,5 MicrowattstoMilliwatts

SuperconductingDevices 1to15 WideRange

BasicResearchExperiments 1to10 Below0,1W


NOTE FromSherman(1978)[216]

42
ECSSEHB3101Part14A
5December2011

Table42:SpacecraftCryogenicCoolingTechniques
CoolingLoadforYear
CoolingTechnique Temp.Range[K]a
Missionb[W]

0to0,01todate;higher
RadiantCoolers 70to100
capacityforeseeable

StoredSolidCryogenCoolers 10to90 0to0,8

StoredLiquidHeliumCoolers 1,5to5,2 0to0,1

He3Coolers 0,3 0to104

DilutionRefrigerators&Adiabatic
0,001to0,3 0to104Adiabatic
DemagnetizationRefrigerators

MechanicalCoolers 4to100 0to300

a Thesevaluesarenottheoreticallimitsbutareestimatesoftemperaturesandloadsbasedonthedesignsas
theyappeartobeevolving.
b Formissionsofshorterduration(7to30dayShuttlesortie)significantlyhighercoolingloadscouldbe
accommodated.
NOTE FromSherman(1978)[216].

4.1 Radiant coolers


Radiant coolers are at present the most widely used system for cooling at cryogenic temperatures
aboardspacecraft.ThesedevicesaredescribedinECSSEHB3101Part9,clause6.
Radiant coolers offer a relatively simple, passive, low weight technique for cooling spaceborne
sensors.Theypresent,however,severelimitationswithrespecttotemperature(>70K),coolingload
(milliwatts),placementonspacecraft,andspacecraftorbit.
Studiestoadvancethedesignofradiantcoolershavebeenmade.Forexample,Sherman&Brennan
(1976) [219] suggested the addition of a cryogenic heat pipe between the detector and the radiant
coolerinordertoallowformoreflexibilityinlocatingtheinstrumentwithrespecttothecooler.They
found that in the 80 K to 100 K temperature range, the parasitic heat loads to the heat pipe require
radiant coolers larger than those being developed at present (by NASA). As the cooler temperature
levelincreases,theeffectoftheheatpipeinsystemperformancebecomeslesssignificantbecauseits
parasiticlossesdecreasewhenthecoolercapacityincreases.
Theuseofcryogenicdiodeheatpipes(seeECSSEHB3101Part8,clause9.2)hasbeenconsideredfor
applicationsinwhichthecoolerisexposedtoahotenvironmentsuchascyclicsolarorearthinputs.
Whenthecoolerbecomeswarmerthantheheatsource,thediode)shutsdowntoisolatethesystem
fromtheenvironment.
In spite of these or similar studies for alleviating some of the limitations of radiant coolers, other
spacecraft cryogenic systems are being developed to meet the increasing demands of lower
temperatures,highercoolingloadsandfurtherflexibilityofintegrationandoperation.

43
ECSSEHB3101Part14A
5December2011

4.2 Stored solid-cryogen coolers


Storedsolidcryogensystemsforlongtermcoolingdetectorsinspacehavebeenunderdevelopment
for the last ten years. These cryogenic systems are passive and not limited by orbit, spacecraft
orientation,orlocationswithinthespacecraft.
Coolingisperformedinthesesystemsthroughsublimationofasolidorboilingofaliquid.For10Kto
90Kdetectorinstrumentcoolinginspace,storedsolidcryogensystemsoffermanyadvantagesover
storedliquidsystems.
1. Cryogensforthistemperaturerangehaveheatsofsublimationabout10%to15%higher
thantheirheatsofvaporization.
2. With a storedliquid system, liquid phasegas phase separation under microgravity is
difficult,becauseoftheuncertainlocationoftheliquidfreesurface.
3. Foragivenvolume,thecoolingcapacityofthestoredsolidcryogenislargerthanthatof
thestoredliquidcryogensincethesolidcryogenwillhaveadensity10%to15%higher
thantheliquid.
Theoperatingtemperaturerangeofasolidcryogenislimitedfromabovebythetriplepoint,andfrom
below by the requirement that pressure around the sublimating solid be large enough to overcome
pressurelossesintheventingtube,allowingtheventingofvaporsresultingfromsolidsublimation.
Although the mentioned pressure losses obviously depend on the geometrical configuration of the
ventingplumbing,theouterpressureandthemassflowrate,avalueoftheorderof0,1torr(13,33Pa)
isappropriateinmostcases.
SeveralstoredsolidcryogencoolersaredescribedinClause6.5ofthisPart.

4.3 Stored liquid Helium (He4) coolers


ManyapplicationsinthefieldsofIRastronomy,cosmicraysatthetopoftheatmosphere,andbasic
physicswilldemandcoolinginthe1,8Kto4,2Ktemperaturerange.Superfluidliquidhelium(He4)is
wellsuitedtothispurpose.
Superfluidliquidheliumpresentsanumberofunusualpropertieswhichshouldbetakenintoaccount
bythedesignerofthecooler.ThesepropertiesarediscussedinClause3ofthisPart.
Thecrucialareasinthedesignofstoredliquidheliumcoolersare:
1. Insulation.Heliumhasaheatofvaporizationinthe1,8Kto5,2Ktemperaturerangeof
about22x103J.kg1,whichisanorderofmagnitudebelowtheheatofsublimationofsolid
cryogens.Thisunderscorestheimportanceofminimizingtheheatleakstoliquidhelium.
Ontheotherhand,sincegaseousheliumhasahighspecificheat,heliumvaporscanbe
usedveryefficientlytocooltheinsulationbeforeventingtospace.Thisisachievedinthe
socalledVaporCooledShield(VCS)Dewars.VCSDewarsareextensivelydiscussedin
Clause6ofthisPart.
2. Containment and liquid/vapor separation. liquid cryogen into the venting tube of the
Dewar should be avoided by all means. Otherwise a waste of cryogen will result.
Fortunately,thereisapeculiareffectinsuperfluidhelium,thethermomechanicaleffect
(seeclause7.1.1.1),whichallowstheventingofthevaporsthroughaporousplugwhile
still retaining the superfluid liquid. The superfluid porous plug is considered in clause
7.4ofthisPart.
3. Fillingofasuperfluidheliumcontainer(clause1.1)isnotaneasytask,sincemuchcare
should be exercised to minimize the liquid loss in the container while pumping down.

44
ECSSEHB3101Part14A
5December2011
Twodifferentfillingapproachesarepresentlyenvisaged.1)Storageoftheliquidhelium
in the normal state at ground, and pumping down in space upon opening the effluent
port.2)Fillingofsuperfluidliquidheliumatground.Thefirstapproachlookssimpler,in
termsofgroundsupportequipment,butwouldresultinabout40%heliumloss.
4. Sloshingofthesuperfluidliquidheliumisanotyetsatisfactorilysolvedproblem.Notice
that the mass of the liquid is significant and that the cryostat may be required to be
highly stable with reference to its own experiments (for instance, when an instrument
scans)andwithreferencetootherexperimentsinthesamespacecraft.
ThestudyoftheHe4freesurfaceposesasubstantialchallengetoboththeoristsandexperimentalists,
nottomentionthenewproblemsassociatedtothemicrogravityenvironment.AccordingtoSherman
(1978)[216]zerogtestingaboardaircraftindicatesthattheincorporationofaporousmaterialwithin
theDewartocapturetheliquidcansolvethesloshingproblem,butthisisbynomeansclearbecause
ofthesuperfluidvelocitycomponentwhichmaynotbeimpededbytheporousmaterialunlessvery
narrowchannelsareused.

4.4 Trends toward lower temperatures


TheperformanceofIRbolometersunderlowbackgroundradiationisgreatlyenhancedbyoperation
at the lowest possible temperatures. For instance, Sherman (1978) [216] indicates that operating a
bolometer at0,3 Kas opposed to the1,8 K, which can be reached by He4 Dewars, resultsina noise
equivalentpowerreductionofanorderofmagnitudeormore.
The simplest way for reaching temperatures significantly below 1 K is by using a He3 cooler. He3,
whichisanisotopeofhelium,presentstwofavorablecharacteristicswhicharewellsuitedtothisaim.
First,thevaporpressureofHe3isatalltemperatureshigherthanthevaporpressureofHe4.Theratio
of He3 to He4 vapor pressures is 74 at 1 K, 610at 0,7 K and 9800 at 0,5 K (Lounasmaa (1974) [138]).
Second,thereisnoHe3film.AHe3bathmaythusbepumpedviaawidetubeatthelowtemperature
and,withoutneedforanarrowconstrictiontoimpedefilmflow.
The normal boiling point of the He3 is 3,19 K, its critical temperature and pressure are 3,32 K and
1,16x105Parespectively,anditsheatofvaporizationislessthanhalfthatofHe4.Inadditiontoitslow
heatofvaporization,He3presentsthedrawbackofitsextremeexpensiveness,approximately250$/l
under normal pressure and temperature conditions (106 m3 of liquid correspond to 0,6 l of gas at
normalconditions).ThishighcostofHe3urgesusingclosedcyclecoolers.
Lounasmaa (1974) [138] presents the basis of He3 cryostat design and describes several laboratory
apparatus. Figure 41 shows a groundbased closedcycle He3 cryostat, intended to cool three
bolometers to 0,3 K, and which is designed, fabricated and being tested by NASA Goddard Space
Flight Center. The heat load (60x106 W) is absorbed by vaporization of the He3 asit is pumped via
charcoaladsorption.Undernormaloperatingconditionsthecharcoalcanisteriskeptatabout1,8Kby
thermal switching to the He4 Dewar. When all the liquid He3 (5x106 m3) is adsorbed, the cycle is
reversedbyopeningthethermalswitchandheatingthecanister.Asthecharcoaliswarmedtoabout
40KtheadsorbedHe3isdrivenoffandissubsequentlycondensedinaninternallyfinnedcondenser.
TheHe3thenflowsbackintothecontainer.Theheatswitchinthenclosed,coolingdownthecharcoal
to1,8K,andstartingupanewcycle.

45
ECSSEHB3101Part14A
5December2011

Figure41:He3coolerbeingdevelopedbyNASA.FromSherman(1978)[216].

CriticalproblemsintheadaptionofthebasicHe3cycletospacecraftare,accordingtoSherman(1978)
[216],thecapillarysystemsustainingtheHe3inthecup,andthestructurecapableofsurvivinglaunch
whilelimitingheatleakstoextremelylowvalues.
A similar He3 cryostat, developed for the cooling of bolometers in ballonborne astronomy and
aeronomyexperimentshasbeenflighttested(Torre&Chanin(1978)[240].
Therearethreemethodspresentlyavailableforcoolingto102Kandbelow.Theseare:theHe3/He4
dilutionrefrigeration,Pomeranchukcooling,andadiabaticnucleardemagnetization.Theprinciplesof
allthreearepresentedinLounasmaa(1974)[138].Thefirstandlastmethodsarebeingconsideredfor
103 K spacecraft coolers. The dilution refrigerator operates continuously, it is quite simple in
construction and easy to operate (systems for terrestrial laboratory use are commercially available),
but presents many zerog fluid management problems that do not exits with adiabatic
demagnetization. This last technique, however, is basically a singlecycle method of cooling, after
demagnetizationthesystembeginstowarmupand,thence,canbeonlyusedinsingleshottypeof
experiments.

4.5 Mechanical refrigerators


Mechanicalrefrigeratorswillberequiredforprovidingcoolingoverlifetimesoftheorderofyears.A
greatdealofefforthasbeenexpendedforthelasttenyearsinthedevelopmentofthesedevices.An
accountoftheseveraltypes,theirdevelopmentproblems,andthestateoftheart(mainly)upto1982
is given in Clause 5 of this Part. A longlifetime (2 yr to 5 yr) mechanical cooler with stable and
consistentperformanceisstilltobedeveloped.
The spacecraft mechanical coolers that have received the greatest attention, among the many
refrigerating cycles available, include reverse Braytoncycle, Stirlingcycle and Vuilleumiercycle
machines.
TheidealreverseBraytonthermodynamiccycleisdescribedinClause5.2ofthisPart.Thisidealcycle
is not thermodynamically reversible. Further, with this cycle it is difficult for machinery to achieve
highefficiency,especiallyforsmallcoolers.
ThereverseBraytonthermodynamiccyclecanbeapproximatedwithturbomachinerysystems.These
cooling systems have a longlifetime potential because its high rotational speed (above 10 rpm)

46
ECSSEHB3101Part14A
5December2011
allows the efficient use of gas bearings. Coyler et al. (quoted by Sherman (1978) [216] describe a
detaileddesignstudyofaturborefrigeratorforspaceapplication,withnominalcoolingloadsof1,5
Wat12Kand40Wat60K,3x10hlifetime,andamaximumpowerconsumptionof4x10W.
Miniature refrigerators based on the reverseBrayton thermodynamic cycle use the rotary
reciprocating mode of action. In the rotaryreciprocating system (clause 5.2) the pistons used to
compress(ortoexpand)theworkingfluidarerotatedaswellasreciprocated.ArthurD.Littlesrotary
reciprocatingrefrigerator(clause5.2andclause5.2.1)employsreciprocatingmotionforrefrigeration
action and rotary motion to allow the use of gas bearingsfor longlifetime potential. In accord with
Sherman (1978) [216], no complete test program for these coolers has been performed up to this
moment.
Stirlingcyclerefrigerators(clause5.2)havebeenemployedinaircraftandshorttomediumduration
space missions. Several Stirlingcycle refrigerators under development are listed in clause 5.2.1. In
orbitperformanceofspacequalifiedStirlingcyclerefrigerator(rhombicdrive)isdiscussedinclause
5.2.1.
PreliminarytestresultsofamagneticallysuspendedStirlingcyclerefrigerator,underdevelopmentby
NorthAmericanPhilips(seeclause5.2)havebeenreportedbyDanielsetal.(1984)[50].
The Vuilleumier refrigerator presents many features which offer the possibility of longlifetime
operation. These include relatively low bearing loads and a less stringent seal requirement as
comparedtotheStirlingcyclerefrigerators.
The main features of the Vuilleumier refrigerators are described in Clause 5.2 of this Part. These
refrigerators incorporate regenerators designed for low pressure drop, thus the cooler requires little
compressionorexpansionoftheworkingfluidandthedisplacersseparatechamberswhosepressures
arenearlyequalatanymoment.Consequently,theelectricmotorpower,bearingloadsandvibrations
aresmall.
SeveralVuilleumierrefrigeratorsunderdevelopmentarelistedinclause5.2.1.
NASA Goddards Vuilleumier refrigerator program started in 1969. According to Sherman (1978)
[216],amachinebuiltbyGarretAiResearchmetthermalperformancegoalswith299Wrequiredto
achieve7Wofcoolingat75K.Theunitran6x10hwithoutamechanicalfix.
The US Air Force has had an extensive Vuilleumier refrigerator program during the past decade.
Muchworkhasbeenconcentratedonthedevelopmentofthreestagemachinewith12W,10Wand
0,3Wcoolingloadsat75K,33Kand11,5Krespectively.Thismachinehasalifetimegoalof2x10hof
unattended operation with a maximum of 2,7x10 W of input power. To accomplish these goals,
paralleldevelopmenteffortshavebeencarriedoutbyPhilipsLaboratoriesandHughesAircraft(see
clause5.2.1).

4.6 Low temperature requirements to IR sensors


Available IR sensors are capable of detecting extremely small amounts of incoming radiant energy,
providedthatextraneoussourcesofradiantandthermalenergyarereducedtolevelsfarbelowthose
emanatingfromthetarget.
Thesourcesofenergyinterferingwiththedetectionprocessare:
1. Sourcesotherthanthetargetexistinginthescene.
2. Radiationfromthesensoropticalsystem.
3. Excitationsofthedetectorduetoitsowntemperature.

47
ECSSEHB3101Part14A
5December2011
The aim of cryogenic cooling is to reduce the effect of the last two sources of interference. The
followingdiscussionisasummaryofapaperbyCaren&Sklensky(1970)[37].

4.6.2 Radiation from the optical system


ThesensordetectioncapabilitydependsontheratioofthenumberoftargetphotonstotheRMSofthe
fluctuationin the number of background photons arrivingat the detectors during the dwell time of
theopticalsystem(signaltonoiseratio).
Thesourcesofbackgroundradiationfromtheopticalsystemaretheinterioroftheopticalcavityand
theopticsitself.Althoughtheopticshavegenerallyalowemittance,mostofthebackgroundenergy
arrivingatthedetectorcomesfromtheoptics,sincetheradiationfromtheopticalcavityisshieldedas
sketchedinFigure42.

Figure42:Proceduretoreducethebackgroundfluxfromtheoptics.FromCaren&
Sklensky(1970)[37].

The monochromatic background power reaching the detector is equal to the power radiated by the
opticsperunitarea,perunitspectralbandwith,multipliedbytheopticssurfaceareaandbytheview
factorbetweenopticsanddetector.
The monochromatic power depends on the wavelength, , and its maximum value moves with
temperaturetoshorterwavelengths,asexpressedbyWienslaw(maxT=2,8978x103m.K).
Thephotonnumber,correspondingtoagiven ,andleavingtheunitareaoftheemittingsurfaceis
equaltothemonochromaticpowerdividedbytheenergycarriedbyonephoton,whichis2hc/(2h
=6,6256x1034J.sisthePlanckconstant,andc=2,9979x108m.s1isthespeedoflightinvacuum).
Tocalculatethebackgroundphotonnumberfluxonthedetector,themonochromaticcontributionis
integrateduptothedetectorcutoffwavelength,c.
In order to reduce the background power reaching the detector, c (which depends on the
characteristics of the target) ought to be smaller than max. Since maxT is constant, the detector
temperature,T,isdecreasedwhenthecutoffwavelength, c,isincreased.ThisisillustratedinFigure
43, which gives the detectivity, D*, of the detector as a function of the cutoff wavelength, c, for
differentvaluesoftheopticstemperature,T.

48
ECSSEHB3101Part14A
5December2011

Figure43:Detectivity,D*,ofaphotonnoiselimiteddetectorasafunctionof
cutoffwavelength,c,forseveralvaluesoftheopticstemperature,T.FromCaren
&Sklensky(1970)[37].

Thedetectivity,D*,whichisafigureofmeritofthedetector,isdefinedas:

Ad / t
D* [41]
NEP

where Ad is the detector area in m2, t is the dwell time of the system in s, and NEP the Noise
EquivalentPowerinW.NEPisthepowerthathavetoreachthedetectortoprovideanoisesignalof
themagnitudeofthenoiseobserved.

4.6.3 Noise from the detector


The detector which are currently used for infrared sensing are semiconductor materials exhibiting
photoconductiveproperties.
When a photon of energy equal to or greater than some given threshold energy (the energy gap) is
absorbedbyonesuchmaterialariseintheconductivityofthematerialresults.Thiseffect,whichis
duetothegenerationofaholeelectronpair(freecarrier),disappears,andthematerialexperiencesa
transition to its previous undisturbed state, once a given time is elapsed. This transition is called
recombination.
Fluctuationsinthearrivalofbackgroundphotonsfromtheopticsresultinfluctuationsintheelectrical
resistance of the material. This, in turn, is converted into a voltage or current fluctuation which
constitutesthenoisesignal.
There are several alternative sources of noise in the detector itself and in the electronics associated
with it. Usually, generationrecombination of free carriers (GR) dominates the noise producing
mechanism. The free carriers are constantly generated optically and thermally, and recombined
throughmanypossibleprocesses.
When, for a given flux level, the background photon generation rate of free carriers exceeds the
thermalgenerationrate,thedetectorissaidtobeBLIP(BackgroundLimitedInfraredPhotoconductor)
atthisfluxlevel.

49
ECSSEHB3101Part14A
5December2011
AnupperboundofthedetectoroperatingtemperatureforBLIPoperation,canbeestimatedequating
thebackgroundphotonandthermalgenerationrates.

Figure44:Typicaldetectoroperatingtemperature,T,vs.detectivity,D*.The
detectorisgermaniumdopedeitherwithmercury,withcadmiumorwithcopper.
FromCaren&Sklensky(1970)[37].

Figure 44 gives the upper bounds of the detector operating temperatures, T, vs. detectivity, D*, for
germanium doped with three types of dopants. To relate the background photon number flux, nB,
reachingthedetector,tothedetectivity,D*,thefollowingexpressionhasbeenused:


D* [42]
2h n B

Where isthewavelengthofthedetectorpeakresponse.
It can be deduced from Figure 44that low detector operating temperatures are required to achieve
highsensitivity.

50
ECSSEHB3101Part14A
5December2011

5
Refrigerating systems

5.1 General
Thevarioustypesofrefrigeratingsystemsapplicabletospacecraftaredescribedwithsomedetailin
thisClause.
Clause5.2concernsmechanicalrefrigeratorsorclosedcyclesystems.Thesesystemswillbeusedinthe
near future for thermal conditioning payloads with high cooling loads (above 5 W), low operating
temperatures (below 100 K), and log lifetimes of unattended service (2 to 5 yr). Unfortunately, the
state of development of these systems is far from satisfactory and, thus, only a description of the
thermodynamiccyclesonwhichtheyarebased,andofthedevelopmentproblemstheyface,aregiven
here.
Opencycleorexpendablerefrigeratingsystems,whicharesubjectofclause5.3,aremuchsimplerthan
mechanical refrigerators although their lifetime is obviously smaller (see Table 42, in General
IntroductionofClause4).Opencyclesystemsarebasicallyoftwotypes:thoseusingahighpressure
gas whose cooling effect is produced by a JouleThomson adiabatic expansion, and those using a
storedsolidorliquidcryogeninthermalcontactwiththeheatsource.Sincestoredcryogencoolersare
extensivelydealtwithin

5.2 Closed cycle


AnumberofmechanicalrefrigeratorsadaptabletospaceborneoperationispresentedinthisSection.
Thebasicthermodynamiccyclesandthedevelopmentproblemsassociatedtoeachtypeofrefrigerator
arediscussed.Thedataarearrangedinthefollowingorder:Brayton(RotaryReciprocatingSystems),
Brayton & Claude (Turbo machinery Systems), GiffordMcMahon/Solvay, JouleThomson, Stirling,
Vuilleumier.
Spaceborne mechanical refrigerators are characterized by: low mass (which includes masses of
refrigerator,powersupply,andheatrejectionsystem),lowpower,lowacousticnoise,andextended
reliabilitywithoutopportunityformaintenanceand/orrepair.Tomeettheserequirementsemphasis
mustbeplacedonbothmechanicalandthethermaldesign.
Muchefforthasbeenexpendedforthelasttenyearsinthedevelopmentofmechanicalrefrigerators.
Severalofthedevelopmentprogramshavebeengeneralinnaturewithemphasisinraisingtherelated
technologylevel,whileothersaimedatdevelopingrefrigeratorforaparticularmission.Inspiteofthis
effort,systemswithstableandconsistentperformancearescanty.
Technical data on many mechanical refrigerators for spacecraft application are presented in clause
5.2.1. The heading, Existing Systems, which is used in this clause for consistency with similar
material in the other Parts, is a bit misleading here, since most of the described systems are in the

51
ECSSEHB3101Part14A
5December2011
development phase (even, work on some of them has been interrupted). The mentioned data are
includedtobeusedasanaidinthepreliminarydesignofnewsystems.

5.2.1 Reverse-Brayton cycle


(RotaryReciprocatingSystems)

5.2.1.1 Description
Figure51describestheidealreverseBraytoncycleforasinglestagesystem.

Figure51:ReverseBraytonCycleRefrigerator.FromSherman(1978)[216].

Thethermodynamicprocessesassociatedwiththiscyclearethefollowing(seeFigure51):
atob.Thegasiscompressedwithsomeincreaseinentropy.
btoc.Heatisrejectedtotheambienttemperatureheatsinkinanaftercooler.
ctod.Thehighpressuregasiscooledinthemainheatexchanger.Intherealcyclethepressureindis
slightlylessthanthatinbbecauseofthepressurelossesinthetwoheatexchangers.
dtoe.Expansionwithsomeentropyincreaseintheexpander.
etof.Heatadditionintheheatloadexchanger.
ftoa.Thegasisheatedinthemainheatexchangerandreturnstothecompressorinlet.
Theidealcycleisnotthermodynamicallyreversible.AlthoughBraytoncyclerefrigerationsystemsare
frequentlyusedinplantinstallationsbecauseofthecomponentsimplicityandacceptableefficiency,it
isdifficultforsmallcoolerstoachievehighefficiency.
Miniature refrigerators utilizing rotaryreciprocating machinery are being developed for space
applications by Arthur D. Little (Figure 52). In these machines the pistons are rotated as well as
reciprocated.Thispermitstheuseofportstocontrolthegasflowandclearancesealstolimitleakage.
Electromagnetic actuators drive the pistons. The machinery has relatively few moving parts, all of
whicharecompletelysupportedonselfactinggasbearings.Therearenorubbingorslidingsurfaces
asinconventionalreciprocatingequipment.Therequiredrefrigerationmachineryiscontainedintwo
separateunits:acompressorassemblyandanexpanderpackage.

52
ECSSEHB3101Part14A
5December2011

Figure52:CompressorcrosssectionofADLrotaryreciprocatingrefrigerator.
FromDonabedian(1972)[59].

The spaceborne refrigerator system includes radiators for cooling of the compressed gases and for
rejecting the heat generated by electrical losses in the housing. Power conditioning equipment is
required to convert the output of the basis source of electrical power into an AC of the power
frequency,amplitudeandphase.
ThetestprogramfortheADLrotaryreciprocatingcoolerisnotyetcomplete(Sherman(1978)[216]).

5.2.1.2 Development problems


The rotaryreciprocating refrigerator holds promise of extended life since most of the problems
associated with wear, sealing, and contamination have been essentially eliminated. However, the
complexity of this system represents an inherent development risk since it is a relatively novel
approach.AsummaryofpotentialproblemareasisshowninTable51.

Table51:DevelopmentProblemAreasofBraytonCycleRotaryReciprocating
Refrigerators
Component Problem

CounterflowHeatExchanger Nonmetallicspacermaterialbeingusedincurrentdesignto
reducelongitudinalheatconductioncouldbeasourceof
contamination.Theuseoflaminatedmaterialsisapotential
sourceofleakage.

LinearActuators Outgassingofthecoilscouldcontaminatethepistonbore
assemblies.

RotaryMotors Assessmentoftheperformanceinasubmergedenvironmentis
needed.

GasSpringsandBearing Manufacturingtolerancemustbebuildup.

Linkages(ReedConnectors) Fatigueduetoloadreversalsandendloading.

Traps/Filters Filterfailureduetoimproperassembly.
NOTE FromDonabedian(1972)[59].

53
ECSSEHB3101Part14A
5December2011

5.2.2 Reverse-Brayton and Claude cycle refrigerators


(Turbomachinerysystem)

5.2.2.1 Description
The reverseBrayton cycle, which has been described above, can be approximated by a turbo
refrigerator.Thistypeofcoolerhaslonglifetimepotentialbecauseitshighrotationalspeed(above105
rpm) allows the efficient use of gas bearings. Lubrication of the cycle working fluid alleviates
contaminationandfoulingproblemsinthelowtemperatureregionsofthecycle.
Intheserefrigeratorstheworkinggas,initiallyatambienttemperature,iscompressedandthencooled
back to the ambient temperature in the aftercooler. The high pressure gas is then passed through a
seriesofcounterflowheatexchangersandisexpandedinoneormoreturbineswherethemechanical
energyisextracted.Thegasisthendirectedthroughthecoolingloadheatexchanger.
Astheminimumtemperature,pointeinFigure51,isloweredpartialcondensationoftheworking
fluid may appear. For such low temperatures JouleThomson or throttling expansion (clause 5.3) is
usedinsteadoftheexpansioninaturbine.ThisisthebasicideaoftheClaudecycle.Itcanbeseenby
comparisonofFigure53andFigure55thattheClaudecycleiseffectivelyaJouleThomsoncyclein
whichtheeffectivesinktemperatureisloweredbyaBraytoncyclerefrigerator.

Figure53:ClaudeCycleRefrigerator.FromDonabedian(1972)[59].

5.2.2.2 Development problems


Theturbomachineryrefrigeratorhasafavorablepotentialforachievinglonglifespacebornesystems.
There are currently no identifiable life limiting components. Potential problem areas identified are
showninTable52.
Thepresentstatusofturbomachinerydevelopmentinminiaturerefrigeratorsseemstoindicatethat:
1. It is possible to construct small turbo expanders and compressors suitable for
refrigerationapplication.
2. The reliability and continuous running time for these units can be very good, with
maintenanceintervalsof104hoursorbetterexpected.

54
ECSSEHB3101Part14A
5December2011
3. The heat exchanger and the compressor presently set the limit on size reduction of
turborefrigerators.
4. It is not likely that a miniature turbo machinery refrigerator will be commercially
availableforseveralyears.

Table52:DevelopmentProblemAreasofBraytonandClaudeCycleRefrigerators
Component Problem

CounterflowHeatExchanger Nonmetallicspacermaterialbeingusedincurrentdesignto
reducelongitudinalheatconductioncouldbeasourceof
contamination.Theuseoflaminatedmaterialsisapotential
sourceofleakage.

Compressors Rigiddimensionalaccuracyisrequiredtomaintainproper
clearancesbetweenrotatingassembliesandthrustbearings.

TurboAlternators Particlecontaminationofthegimbaledthrustbearing
assembly.

GasBearing Manufacturingtolerancemustbebuildup.

Traps/Filters Manufacturingandimproperassemblycouldcausefilter
breakthrough.
NOTE FromDonabedian(1972)[59].

5.2.3 Gifford-McMahon/Solvay cycle refrigerators

5.2.3.1 Description
ThebasicrefrigerationcycleusedinthistypeofsystemswasoriginallyconceivedbyE.Solvayin1886
as a basic derivative of the Stirling cycle. A number of modifications have been made by various
researchers such as K. W. Taconis, W. E. Gifford, and H. O. McMahon. Refrigerator units
manufacturedareusuallymarketedusingvariousnames,suchasGiffordMcMahonandSolvay,with
and without the adjective modified. They are basically the same cycle but with different
modificationsintheexpander.
ThebasicexpansionprocessisillustratedinFigure54.Inpositionattheinletvalveisopenandthe
exhaustclosed.Theregeneratorandothervoidvolumesarefilledtothehigherpressure.

55
ECSSEHB3101Part14A
5December2011

Figure54:SolvayCycleRefrigerator.FromDonabedian(1972)[59].

a to b. This piston moves outward and working fluid enters the cylinder after being cooled in the
regenerator.
b to c. The inlet valve is closed and the fluid pressure falls until the piston reaches its outermost
position.
ctod.Theexhaustvalveisopenedandthefluidinthesystemexpandstod.
d to e. The piston moves inward, expelling the cold working fluid from the system. This fluid also
coolstheregenerator.
e to f. The exhaust valve is closed and the piston continues to move until it reaches the innermost
position.
ftoa.Theinletvalveisclosedandthefluidinthesystemiscompressedfromftoa.

5.2.3.2 Development problems


Because of the commercial attractiveness of the GiffordMcMahon/Solvay cycle units, substantial
development effort and production knowledge have been achieved. The primary advantage of this
type of cooler is that the cooling head can be separated from the compressor, and thence it can be
moreeasilyintegratedwiththeload.
Theprimarylimitationisthelowerefficiencyofthesystem,ascomparedwithVuilleumierorStirling
systems,thenceitrequiressignificantlyhigherpowerinputthansystemsbasedonthementionedtwo
cycles for the same cooling load. Nevertheless, the GiffordMcMahon/Solvay cycle units currently
provide the bulk of commercial low temperature cooling and the longest unattended lifetime.
Althoughsignificantimprovementsinpowerrequirements,arenotlikely,substantialmassreduction
inthesystemsmaybeexpectedwiththeuseofcompressorunitswhichareoptimizedforminimum
mass.

56
ECSSEHB3101Part14A
5December2011

5.2.4 Joule-Thomson Closed Cycle Refrigerator

5.2.4.1 Description
ApracticalJouleThomsonrefrigeratorcycleisshowninFigure55.Thiscycleisessentiallyidentical
tothereversedBraytoncycleexceptforonefundamentaldifference:theexpansionprocess,dtoe,is
accomplishedbyexpansionthroughathrottlingvalveratherthanthroughaturbine.

Figure55:JouleThomsonClosedCycleRefrigerator.FromDonabedian(1972)
[59].

In a typical application, the working fluid, gaseous nitrogen, undergoes the following
transformations:
a to b. The fluid is compressed to approximately 1,72x107 Pa in a multistage oillubricated
reciprocatingcompressor.
b to c. Once compressed, the gaseous nitrogen is cooled by ram air or by a fan mounted on the
compressor assembly. The gaseous nitrogen then passes through an absorber/filter component in
order to remove oil vapor and other trace contaminants, which might solidify at cryogenic
temperatures.
ctod.Thepurifiedhighpressurenitrogenenterstheminiatureregenerativeheatexchangerwhereit
iscooledbythereturninglowpressuregas.
d to e. The highpressure gaseous nitrogen is expanded at the exit of the heat exchanger. This
producesatemperaturedropwhichissufficienttopartiallyliquefythegas.
etof.Thelatentheatoftheliquidnitrogenisusedtoprovidethemaincooling.
ftoa.Thelowpressuregas,afterbeingusedforprecoolinginincominggas,returnstothefirststage
ofthecompressortobeginanewcycle.

5.2.4.2 Development problems


The major advantage of JouleThomson closed cycle coolers is that the compressor module can be
located far from the heat sink, and thence the entire cooling system can be packaged into varied
configurations. Other advantages are that no adjustment is required regardless of ambient

57
ECSSEHB3101Part14A
5December2011
temperatures,andthatrapidcooldowncanbeachieved(approximatelyin3to5minutes).Overhaul
timesfrom200to500hoursaretypicalwithmeantimebetweenfailuresof1000to2000hours.
The primary disadvantage of the JouleThomson closed cycle system from the standpoint of space
application is its high power requirements. For this reason the system has not been extensively
developed for spacecraft application. Another disadvantage is that heat rejection is produced
isothermallyonlyattheliquidtemperatureoftherefrigerantbeingused,thatlimitstheflexibilityof
application.

5.2.5 Stirling cycle refrigerators

5.2.5.1 Description
TheStirlingsystempossessesseveraloftheprimaryrequirementsofaspacebornerefrigeratorsystem,
suchaslowpowerconsumptionandsmallsizeandmass.ItsidealefficiencyreachestheCarnotlimit
(seeclause5.2.1).
Heat evolution takes place in the Stirling cycle by alternately compressing and expanding a given
quantityofagasinaclosedcycle.Compressiontakesplaceatambienttemperaturesoastofacilitate
heatrejection;theexpansionisperformedattherequiredcoldtemperature.
Forthesakeofexplanation,theStirlingcyclemaybesplitupintofourstages,asindicatesinFigure
56.IdealcyclediagramsaresketchedinFigure57.

Figure56:StirlingCycleRefrigeratorOperation.FromSherman(1978)[216].

58
ECSSEHB3101Part14A
5December2011

Figure57:StirlingCycleRefrigeratorIdealPressureVolumeandTemperature
EntropyDiagrams.FromSherman(1978)[216].

Inpositionamostoftheworkinggasoccupiesthesocalledambientend,atambienttemperature.
a to b. Isothermal compression. The gas is compressed by inward motion of the compressor piston.
This compression takes place at room temperature. Heat is rejected through the ambient heat
exchanger.
btoc.Constantvolumecooling.Thegasistransferred,throughtheregenerator,fromtheambientto
the cold end, at constant overall volume, by means of the displacer which usually encloses the
regenerator. The gas flowing through the regenerator is cooled to nearly the temperature which
prevailsinthecoldend.Heattransferredtotheregeneratorwillberejectedinstagedtoa.
c to d. Isothermal cooling. The gas is expanded by moving together both the displacer and the
compressorpiston.Thesensor,orwhateverdevice,iscooledthroughthecoolingloadheatexchanger.
d to a. Constant volume heating. The gas is displaced from the cold to the ambient end by the
displacer.Duringthistransfer,theflowinggaslowersthetemperatureoftheregenerator.

5.2.5.2 Development problems


Although small Stirling cycle refrigerators have been employed in aircraft and shortduration space
missionssuchasSkylab(Sherman(1978)[216]),theonlytrulylongdurationdatapresentlyavailable
are provided by four refrigerators which were launched on Feb. 24, 1979 aboard the DOD P781
Satellitetoprovideorbitalcoolingoftwogammarayspectrometers(Sherman(1982)[217]).
The refrigerator was developed by theNorth American Philips Corporation,with the John Hopkins
Applied Physics Laboratory providing the electronic controls and the space qualification program.
Integrationwiththegammarayspectrometersandevaluationoforbitaloperationdatawasmadeby
Lockheed,PaloAltoResearchLaboratories(Naes&Nast(1980)[160]).
A schematic representation of this refrigerator is shown in Figure 58. Performance data are
summarizedinclause5.2.1.

59
ECSSEHB3101Part14A
5December2011

Figure58:SchematicrepresentationofNorthAmericanPhilipsrefrigerator,
showingrhombicdrivemechanism.Thedrivehastwocounterrotating
crankshafts,eachpoweredbyadrivemotor.Byadjustingthemassofthe
reciprocatingmembersofthedriveandbyaddingappropriatecounterweightsto
thecrankshafts,thecenterofthegravityofallthemovingpartscanbekept
stationary.FromBalas,Leffel&Wingate(1978)[16].

A description of the cooler design, qualification tests, performance and lifetime tests on ground has
beenmadebyBalas,Leffel&Wingate(1978)[16].InorbitperformanceshavebeenreportedbyNaes
&Nast(1980)[160].
AsofJuly1980,whenthespacecrafthasbeeninorbitfor500days,thecumulativeoperatingtimeof
eachofthefourcoolerswas2657,5134,9542and6518h.Thelatestdata(Sherman(1982)[217])show
that one of the machines has surpassed 15000 h running time. A detailed report of the inorbit
performanceofthesecoolersisgivenbelow.
A very recent approach to the long lifetime problem consists in driving reciprocating components
directlywithlinearmotors,avoidingcontactbetweenmovingcomponentsandthemachinehousing
orthemotor.Noncontactoperationcanbeachievedbymagneticorgasbearingandclearanceseals.
TheNorthAmericanPhilipssinglestagemagneticbearingrefrigerator,Figure59,hasbeendevised
withthefollowingrequirementsinmind:

60
ECSSEHB3101Part14A
5December2011

Figure59:SchematicrepresentationofNorthAmericanPhilipsMagneticBearing
refrigerator,showingthelinearmotorsforpistonanddisplacerandthemagnetic
bearing.Thedisplacerrodpassesthroughthepiston.FromSherman,Gasser,
Benson&McCormick(1980)[221].

1. Norubbingsurfaces
2. Lineardrivesystemwithoutmechanicallinkages
3. Electroniccontrolofpistonanddisplacermotions
4. Dynamicalbalance
5. Allmetalworkinggaschamber
Performance data of this refrigerator are given in clause 5.2.1. For a description of the design,
definitionstudiesandcomponenttestingseeShermanetal(1980)[221]orthemorerecentpaperby
Daniels,Gasser&Sherman(1982)[49].
Alinearreciprocatingrefrigeratorbasedontheabovephilosophy(norubbingsurfaces,lineardriving)
but using gas bearing has been studied by Energy Research and Generation, Inc, see clause 5.2.1.
ResultsfromthedesignstudyhavebeenreportedbyShermanetal(1980)[221].

5.2.5.3 In orbit performance


FourRhombicDriveStirlingCyclemechanicalrefrigeratorsprovidedorbitalcoolingoftwoidentical
gammaray spectrometer detectors, called Gamma003 and 004. Each detector used two refrigerators
whichwillbeidentifiedinthefollowingasrefrigerators1and2forGamma003,andrefrigerators3
and4forGamma004.
Onesinglerefrigeratorissufficienttocooleachdetector.However,twounitswereplacedinparallel
toincreasetheoverallsystemreliability.Thus,astheperformanceofonerefrigeratordegrades,both
areoperatedtogethertofurtherreducethedetectortemperature.

61
ECSSEHB3101Part14A
5December2011
The entiresystem is enclosed withinavacuum shell. Vacuumismaintained both on ground and in
orbitbya2l.s1vacuumpump.
Figure510showsaschematicofthesystemofrefrigerators1and2.

Figure510:Couplingoftworefrigeratorunitstoprovidecoolingofasingle
detector.ThecompleterefrigeratorcanbeseeninFigure58.Here,onthecontrary,
onlythefirstandsecondstagesofbothrefrigeratorsareshown.FromNaes&Nast
(1980)[160].

Attachedtobothfirstandsecondstagecoldtipsareflexiblemechanicalcoupling,theaimofwhichis
reducing the launch structural loads transmitted to the cold tip by the thermal system as well as
suppressingmicrophonicsfromtherefrigeratortothedetector.Bothflexiblecouplingstothesecond
stage cold tips were instrumented and calibrated to allow for measurement of the cold tip heat
transferrate.
Coolingofthedetectorisachievedviaacoppercoldfingerthecrosssectionofwhichcanbeseenin
thefigure.
A thermal guard shroud surrounds the second stage cold tips ofboth refrigerators and the detector
cold finger attach point. This shroud is thermally coupled to the first stage cold tips, providing an
intermediatetemperatureenvironmenttothesecondstage.
Enclosedbytheshroudisalargethermalcapacitor(nodetailsgiven)designedtoretardthedetector
heatingratetolessthan8k.h1oncetherefrigeratoristurnedoff.Thislargeheatingtimeisneededfor
data collection by the detector with the refrigerators off, to avoid excessive microphonics to the
detector.
Performancerelateddatatakeninorbitwere:Coldtiptemperatureandheattransferrate,crankcase
pressure and working gas pressure, motor speed and motor current, all them related to the
refrigerator itself. In addition, shroud and base plate temperature, and pressure within the vacuum
shellweremeasured.
DatainFigure511correspondtorefrigerator2.Firstandsecondstagetemperaturesweremeasured,
inthelaboratory,vs.theappliedheattransferratetoeachstage.
Figure512givesthetemperaturesofshroudandrefrigeratorcoldtip(secondstage)inGamma004
system.

62
ECSSEHB3101Part14A
5December2011

Figure511:GroundTesttemperatures,ofthefirstandsecondstagevs.Second
stageheattransferrate,Q2,fordifferentvaluesofthefirststageheattransferrate,
Q1,andmotorrpm.Thedatacorrespondtorefrigerator2butaretypicalofthefour
units.FromNaes&Nast(1980)[160]. firststage,Q1=1,5W,1000rpm; first
stage,Q1=1,5W,1150rpm; secondstageQ1=1,5W,1000rpm; secondstageQ
1=1,5W,1150rpm; firststage,Q1=2W,1000rpm; Q1=2W,1000rpm.

Figure512:Inorbittemperature,T,ofseveralcomponentsofGamma004systems
vs.Orbitaltime,t.FromNaes&Nast(1980)[160]. coldtipofrefrigerator3;
coldtipofrefrigerator4; shroud; groundtestvalueofcoldtipofrefrigerator3;
groundtestvalueofshroud.

Althoughinitialtemperaturesagreedwithgroundtestdata,agradualwarmingtrendsoonappeared.
Atapproximately150daysinorbitbothrefrigerators,3and4,weresimultaneouslyoperated.Thisdid
not improve the performance of the system because of helium losses in refrigerator 4, as will be
discussedbelow.
ThebehaviorofGamma003,Figure513,wassomewhatdifferent.Theinitialcoldfingertemperature
greatlyexceededthegroundtestvalueof68K.Nevertheless,thesamewarmingtrendasforGamma
004canbeobserved.After125days,operationwithbothrefrigeratorswasinitiated,whichresultedin
a substantial temperature reduction. This mode of operation was interrupted, because of power
limitations,approximatelyat330days.

63
ECSSEHB3101Part14A
5December2011

Figure513:Inorbittemperature,T,ofseveralcomponentsofGamma003systems
vs.Orbitaltime,t.FromNaes&Nast(1980)[160]. coldtipofrefrigerator2;
coldtipofrefrigerator1; shroud; groundtestvalueofcoldtipofrefrigerator2;
groundtestvalueofshroud.

Finally, Figure 514 shows the heat transfer rate from the detector and from the individual
refrigerators,fortheGamma003system.

Figure514:Inorbitheattransferrates,Q,fromGamma003detectorto
refrigerators1and2,vs.orbitaltime,t.FromNaes&Nast(1980)[160]. detector
heatload.Refrigerator2on; heatloadthroughmeter1,Q1.Refrigerator1off;
heatloadthroughmeter2,Q2.Refrigerator2on; refrigerators1and2on;
refrigerators1and2on; refrigerators1and2on.

The relative cooling load attributable to each refrigerator is measured by the meters placed at the
flexible couplings near the second stage cold tip. The heat transfer rate from the detector, Q3, is
presumablydeducedfromQ1andQ2.Thetotalheatremovedbyrefrigerator2,Q2,isinitially0,6W,
whichisthesumofthedetectorplustheparasiticheattransferratefromtheoffunit1.Thedetector

64
ECSSEHB3101Part14A
5December2011
heatrateissubstantiallyhigherthan0,114Wmeasuredingroundtesting.Thisaccountsforthehigher
initialcoldtiptemperatureshowninFigure513.
Causesofsystemdegradation:
1. 1)Heliumloss.Thefourrefrigeratorswererechargedtothedesignpressureof4,89x105
Pathreemonthspriortolaunch.
Atlaunchingtime,refrigerators1,2and3hadlostpressureatarateof4,5x103Pa.mo1,
within expectations. On the other hand, the pressure in refrigerator 4 had dropped to
approximately3x105Pa.
Groundtestdatashowthatareductioninpressurefrom4,89x105Pato4,17x105Pawill
causeanincreaseinthecoldfingertemperaturefrom75Kto105Katthedesignrateof
0,3W.
Heliumlosscould,thus,explainmostofthelongtermtemperatureincreaseexperienced
bybothdetectors.
Although refrigerator 4 showed abnormal helium losses, the effect is hardly noticed in
Figure 512, since this refrigerator started operation after 250 days in orbit. Notice that
operationofunit2produceslargertemperaturereductions,Figure513,thanthosedueto
refrigerator4inFigure512.
2. 2) Abnormally high detector heat load in system Gamma 003, Figure 513 and Figure
514.
The high heat transfer rates are possibly due to water vapor deposition on the low
emittancegoldsurfaces.
The high vacuum surrounding the Gamma 003 detector could not be maintained after
launch,andthesystemwasopentospace.Asthedetectorwaslaunchedcold(prechilled
withliquidnitrogen),criticalsurfacesbecamecontaminated.Partialdecontaminationwas
achieved,after75daysandalsoafter100daysinorbit,bywarmingupthesystemabove
200K.Thiscausedasignificantreductioninthedetectorheatload,althoughgroundtest
resultscouldnotbereproduced.

Table53:PotentialProblemAreasAssociatedwiththeStirlingRefrigerators
Component Problem PotentialSolutions FurtherProblems

FluidWorking Contamination Carefulpurgingand Solidparticlesobstructing


Spacein outgassing. passages.
General Allmetalworkingspaces.
Drysealsandbearings...

Regenerator Agingand Asabove.


contamination. External(notinthe
Matrixflexureand displacer).
rubbing

Counterflow Manufacturing
HeatExchanger tolerancesmayshiftflow
characteristics(see
ECSSEHB3101Part13
clauses11.4.3and11.5).

Seals Wear... Oillubrication Contaminationoffluid

65
ECSSEHB3101Part14A
5December2011

Component Problem PotentialSolutions FurtherProblems


Hardonhardspring workingspace.
loaded... Solidparticlesobstructing
passages.
Clearance(noncontact
Workingfluidleakage. Manufacturingtolerances.
seals)...
Workingfluidleakage.

Bearings Wear. Oillubrication Contaminationoffluid


Hardonhardbearing. workingspace.
Noncontactoperation(gas Developmentwork
ormagnetic)... required.

Piston Largeaxialtemperature Regeneratorincorporated Matrixflexureandrubbing.


gradients. intopiston.

Driving Vibrations... Counterrotatingmasses... Noperfectbalanceis


Mechanism Rectilineardriving achieved.
Developmentwork
Phasingpistonand required.
Electroniccontrol,when
displacer.... Passiveoractivebalancing
rectilineardrivingisused
required.
Developmentwork
required.

Detector Degradationdueto Flexiblecouplingbetween


machinemicrophonics... coldtipanddetector.
DegradationduetoEMI. BrushlessDCmotors.
Contamination... Outgassing.

5.2.6 Vuilleumier cycle refrigerator

5.2.6.1 Description
This constant volume cycle was patented by Rudolph Vuilleumier in 1918. The interest in the
developmentofaVuilleumiercyclerefrigeratorforspacecraftapplicationwasarousedfairlyrecently
because a refrigerator of this type promises the advantages of longlifetime operation, compactness,
andlowweight.
Figure 515 gives a schematic of the Vuilleumier refrigerator. It consists of a hot cylinder, cold
cylinder,andsump(orambientsection).Theworkinggasismovedfromonesectionofthemachine
totheotherandthroughthetworegeneratorsbythemovementofhotandcolddisplacers.Anelectric
motordrivesthedisplacersthroughthecrankmechanism.

66
ECSSEHB3101Part14A
5December2011

Figure515:SchematicoftheVuilleumierCycleRefrigerator.FromSherman
(1978)[216].

Becausetheregeneratorsaredesignedforlowpressureloss,theVuilleumierrefrigeratorshavelittle
compression or expansion of the working gas by the motion of the displacers. Consequently, the
electricmotorpower,bearingloads,andvibrationsaresmall.
The steady state operation of the Vuilleumier refrigeration cycle can be explained by resorting to
Figure516(aschematicoffourcrankpositionsencounteredduringoperation)andFigure517(the
pressurevolumediagramsforthecoldcylinder,hotcylinderandtotalgasvolume).Itisassumedin
theexplanationthatthepressurelossacrosstheregeneratorsiszero,andthencethepressureinthe
threesectionsarealwaysthesame.Itshouldbenoted,inaddition,thatthefollowingisasimpleview
of the operation of the Vuilleumier refrigerator, in that only the predominant processes for a given
crankpositionarepointedout.AdditionaldetailsontheactualoperationcanbefoundedinSherman
(1971)[218].

Figure516:VuilleumierCycleRefrigerator.FromSherman(1971)[218].

Figure517:PressureVolumeDiagrams,fortheColdCylinder,HotCylinderand
TotalGas,oftheVuilleumierCycleRefrigerator.FromSherman(1971)[218].

67
ECSSEHB3101Part14A
5December2011
Webeginwiththecrankinpositiona.Atthispointthecolddisplacerisatitsmaximumdisplacement
position,andthehotdisplacerisonlyatitshalfmaximumposition.Themeangastemperatureinthe
refrigeratorisrelativelylow,andconsequentlythegaspressureislow.
atob.Boththehotandthecoldcylindervolumesdecrease.Partofthecoldgasisforcedthroughthe
coldregeneratorwhereitisheatedtonearlyTAbeforeenteringtheambientsection.Similarly,partof
the hot gas is forced through the hot regenerator where itis cooled to nearly TA before entering the
ambient section. The mean gas temperature and the gas pressure change very little during this
process. Nevertheless, some gas expansion, with the resulting heat absorption from the load, takes
place.
btoc.Thehotcylindervolumeincreasesandthecoldcylindervolumedecreases.Thecoldgaswhich
is forced through the cold regenerator is heated to nearly TA, while part of the ambient gas, which
flows through the hot regenerator, is heated to nearly TH before entering the hot cylinder. The net
effectofthehotcylindervolumetricincreaseandthecoldcylinderdecreaseisanincreaseinthemean
gastemperatureandthegaspressure(acompression).Thence,heatisrejectedattheambientsection.
c to d. Both the hot and cold cylindervolumes increase. Part ofthe ambientgas moves through the
cold regenerator, losses heat and enters the cold volume at nearly TC. On the hot side, part of the
ambientgasmovesthroughthehotregenerator,absorbsheat,andentersthehotvolumeatnearlyTH.
The system pressure does not greatly change during this process. Nevertheless, there is some gas
compressionwiththecorrespondingheatrejectionattheambientsection.
d to a. The hot cylinder volume decreases forcing the gas through the hot regenerator where it is
cooled to nearly TA, whereas the cold cylinder volume increases forcing the gas through the cold
regeneratorwhereitiscooledtoTC.Theneteffectisthedecreaseofboththemeangastemperature
and the gas pressure (an expansion). This expansion results in heat absorption at the cold and hot
ends.

5.2.6.2 Development problems


The basic problem areas associated with the development of Vuilleumier refrigerators are listed in
Table54below.TheprincipaldegradationmodeoftheVuilleumierrefrigeratorseemstobedifficult
tomaintaintherequiredtemperature.Thisistracedtotwocauses:Contaminationofrefrigerantand
sealwear.

Table54:DevelopmentProblemAreasofVuillemierCycleRefrigerators
Component Problems

Regenerators Degradationofthecharacteristicsduetoagingandcontamination.

CounterflowHeat
Manufacturingtolerancesmayshiftflowcharacteristics
Exchanger

HeatCoils Assemblydefectsandagingmaycauseperformanceanomalies.

DisplacerSeals Wearcausesbothcontaminationandleakage.

CrankshaftSeals Useofrollingsealsmaycauseleakagethroughthesealmaterial.Eithersubstantial
improvementinthesealdesignorworkingfluidresupplydevicesarerequired.

Displacers Largetemperaturegradientalongthelengthofdisplacers.

DriveBearing Requireoillubricationforextensivelife.
NOTE FromDonabedian(1972)[59].

68
ECSSEHB3101Part14A
5December2011

ThepresentstateofseveralVuilleumierrefrigeratorprogramshasbeenreviewedbySherman(1978)
[216].
NASAGoddard Vuilleumiercycle refrigerator program started in 1969. The program lead to a
nominal5W,75Kmachine(builtbyGarrettAiResearch,seeClause5.2.1)thathasmanyinteresting
features, such as solid bearing composed of Boeing compact 6841 running against Inconel 718
(flamesprayed with tungsten carbide in the sump and chromium carbide at the hot end), and
labyrinthsealsforbothhotandcolddisplacers.TestswiththeGoddardVuilleumierrefrigeratormet
thermalgoals with299 W required to achieve7 W of coolingat 75 K. Theunitran6000 h without a
mechanicalfix.
The Air Force has hadan extensiveVuilleumier refrigerator programduringthe past decade. Work
has been concentrated on the development of a threestage machine with 12 W, 10 W, and 0,3 W
coolingloadsat75K,33Kand11,5K,respectively;alifetimegoalof2x104hofunattendedoperation
andamaximumof2700Wofinputpowerwereestablished.Developmentcontractswereawardedto
HughesAircraftandPhilipsLaboratories.
The Hughes Aircraft threestage Vuilleumier refrigerator uses drylubricated (MoS2) ball bearings,
withabearingretainermadeofRoulonA(filledTeflon)with5%MoS2.Thisrefrigeratorhasspring
loaded lip seals on both displacers and flexure pivots at the displacer driverod interfaces. After
severalmechanicalandtestproblems,themachineaccumulatedatotalofabout3000hofoperation,
butthenhadtohaveacrackinthecrankcaserepairedand,atthesametime,manypartsupgraded.
Testingoftherefrigeratorhasnowresumed.
The Philips Vuilleumier refrigerator has unloaded, sleeve type, displacer seals and an oillubricated
rhombic drive with unequal strokes. This wet lubrication (using a rolling socka polyurethane
rubberseal)isthemainfeatureofthePhilipsVuilleumierrefrigerator.Therubbersealisolatestheoil
filledcrankcasefromtheheliumworkingvolumewhilebeingsubjectedtothereciprocatingmotionof
therefrigeratordriveshafts.Theoilisthensuppliedtothedrivemechanismbearingsbygeartypeoil
pumpsdrivenbythemachinecrankshafts.
UnfortunatelythisbasicallysimplePhilipsapproachpresentsseveraldrawbacks.First,becauseofits
strength limitations the pressure differential across the rolling seal is limited to a few atmospheres.
This necessitates the useof an auxiliary oil pump/regulator system. Second,because the helium gas
can penetrate the rollingseal diaphragm, it is necessary to use a heliummakeup supply for the
workingvolumeandaheliumseparatorfortheoilsupply.

5.2.7 Existing systems


Technical data on closed cycle mechanical refrigerators are collected in this clause. These data are
arrangedasshowninTablebelow.
In order to provide a comparative evaluation of the various cycles, the inverse efficiency and the
systemmassperunitofrefrigerationpower,ofseveralrefrigeratortypes,havebeenplottedversusthe
operatingtemperatureinFigure518andFigure519respectively.Theinverseefficiencyisdefinedas
therequiredmechanicalpowerperunitofrefrigerationpower.

69
ECSSEHB3101Part14A
5December2011

Figure518:Inverseefficiency(requiredpowerperunitofrefrigerationpower)1,
vs.operatingtemperature,T,forseveralclosedcyclerefrigerators.
aBraytonrefrigerators(TurbomachinerySystems).
bStirlingrefrigerators.
cVuilleumierrefrigerators.
dGiffordMcMahon/Solvayrefrigerators.
FromDonabedian(1972)[59].
Alsoshownarecurvesforclosedcyclerefrigeratorsoperatingwiththequoted
efficiencies(inpercentagesofCarnot)throughthewholetemperaturerange.From
Haskin&Dexter(1979)[83].
TheCarnotefficiencyforamachineworkingbetweenTCandTHtemperaturesis
givenbyc=1TC/TH.Verylowoperatingtemperaturesresultinareduced
efficiencyforagivencoolingloadandagivencycle.

70
ECSSEHB3101Part14A
5December2011

Figure519:Systemmassperunitofrefrigerationpower,Mp,vs.operating
temperature,Tforseveralclosedcyclerefrigerators.
aGiffordMcMahon/Solvayrefrigerators.
bStirlingrefrigerators.
FromDonabedian(1972)[59].

Theplottedcurvesonlyrepresentgrosstrendssincetheyhavebeendrawnthroughtheclusterofdata
points from the individual units listed in the data tables. Precise correlations are very difficult to
obtain because of the wide scattering and inconsistencies in the data. Furthermore, no curves are
shown for several of the closed cycle systems presented in the tables since the available data are
insufficienttowarrantinclusionofthecurves.
Figure520showsthesystemmassperunitofrefrigerationpower,Mp,forrepresentativeclosedcycle
refrigerating systems, and for passive radiant coolers (see ECSSEHB3101 Part 9, clause 6). Both
coolingtechniquescouldcompete,whenthelifetimemustexceedtwoyears,attemperaturesbetween
80Kand110Kforrefrigeratingpowersorcoolingloadsbelow1W,andforlargercoolingloadsat
highertemperatures.

71
ECSSEHB3101Part14A
5December2011

Figure520:Systemmassperunitofrefrigerationpower(orcoolingload),Mp,for
representativeclosedcyclerefrigeratingsystemsandforpassiveradiantcoolers.
Closedcyclerefrigerators,Q=0,1W. Closedcyclerefrigerators,Q=1W.
Closedcyclerefrigerators,Q=10W. Passiveradiantcoolers;Q=0,1W.
Passiveradiantcoolers;Q=1W. Passiveradiantcoolers;Q=10W.FromHaskin
&Dexter(1979)[83]. Smallesttemperatureattainedbyclosedcyclerefrigerators
inorbit. Smallesttemperatureattainedbypassiveradiantcoolersinorbit.From
Sherman(1982)[217].

72
ECSSEHB3101Part14A
5December2011

ArrangementoftheCompilerDataonClosedCycleRefrigerators

Cycle Manufacturer No.a

Brayton(RotaryReciprocating A.D.Little 3
Systems).

Brayton(TurbomachinerySystems) A.D.Little 1
GarrettAiResearch 1
HymaticEngineeringCo. 1

Claude AirProducts&Chemicals,Inc. 1
GeneralElectricCo. 3

GiffordMcMahon CryogenicTechnology,Inc. 10
Cryomech,Inc. 4

JouleThomson AirProducts&Chemicals,Inc. 2
GarrettAiResearch 6
SantaBarbaraResearchCenter 1

Solvay AirProducts&Chemicals,Inc. 6
Kinergetics 1

Stirling HughesAircraftCo. 3
MalakerCorp. 7
NorthAmericanPhilipsCorp.(USA) 4
PhilipswithAPLJohnsHopkins 1
PhilipsunderNASAGoddard 1
EnergyResearchandGeneration,Inc. 1
PhilipsLaboratories(Netherlands) 1

Taconis BritishOxygenCo.,Ltd. 1

Vuilleuimier GarrettAiResearch 1
HughesAircraftCo. 10
Kinergetics 1
NorthAmericanPhilipsCorp.(USA) 3
RCADefenseElectronicsLaboratory 1

a Numberofrefrigeratorslisted.

ThementionedfigureisasummaryofatradeoffperformedbyHaskin&Dexter(1979)[217]onthe
followingbases:
1. Sourceofthedata.Dataweretakenfromexamplesfoundintheliteratureconsistingof
flightworthyhardware,testmodelsanddesignproposals.
2. Systemmass.Thecryogenicrefrigeratorincludes:
Refrigeratoritself,whichaccountsfor10%to40%oftotalsystemmass.
Heatrejectionradiatorwithassociatedpumps,valves,piping,...

73
ECSSEHB3101Part14A
5December2011
Power supply which, in most cases, consists of solar cell arrays, electrical storage
batteriesandregulationequipment.
Whendatafortheheatrejectionradiatorwereunavailable,anaveragevalueof22W.kg1
heat rejection capability wasassumed.A common penalty of8,8W.kg1 wasappliedin
anycasetoaccountforcurrentsolararrayandbatterypowersupplycapabilitiesinthe
orbitsconsidered.
Thepassiveradiantcoolerincludes:
Radiatorfins.
Heatpipestocoupletheradiatortotheinstrumenttobecooled.
Supports,shrouds,insulation.
When the actual mass of the radiator was unknown an average value of17kg.m2 was
usedtoderivethesystemmass.
3. Orbit.Favorableorbitsforpassiveradiantcoolershavebeenconsidered(geosynchronous
and sun synchronous orbits above 740 km). Less favorable orbits would require larger
radiatorareas.Thechosenorbitsarealsowellsuitedforsolararraypowersupplies.
4. Operating temperatures. Values of Mp in Figure 520 have been plotted vs. the design
temperatureofthecoolingdeviceunderconsideration.
Closedcyclerefrigeratingsystemsareoftenstaged,andthestagesusedtoprovidecoolingatdifferent
temperature levels. The added penalty for secondary cooling at other stages is often a minor
considerationsincestagingisprovidedforotherreasons.
Similarly,whenstagingoftheradiantcoolerisusedtoreducetheheattransfertothesurroundings,
the higher temperature stages can be used to cool optical elements or other detectors significant
penalty.
DatapointsforpassiveradiantcoolersinFigure520arealmostindependentofthecoolingloadQ,
andcanbescaledupordownproportionallytoQ.ThisisnotsoforrefrigeratorswhoseMpdecreases
andwhoseefficiencyincreaseswithincreasedrefrigerationpower.
Figure 520 shows that the closed cycle refrigerator systems exhibit lower weights when cooling at
verylowtemperaturesisrequired,whereasthepassiveradiantcoolershavelowerweightsathigher
temperatures.ThecrossoverpointsforQ=0,1W,1Wand10Wcanbeseeninthefigure.Itcanbe
inferredthattherefrigeratorswillbecompetitiveatstillhighertemperaturesforcoolingloadsabove
10W.
Otherparameterofinterestisthearearequiredperunitrefrigeratingpower.
Forclosedcyclerefrigeratorstheassociatedareaisthesumofthesolarcellarrayandradiatorarea.
Any other exposed area (refrigerator, pump, etc.) was assumed to be negligible. A solar cell array
specificpowerof89W.m2wastakeninanycase.Whennodataontheradiatorsizewereavailablean
appropriateareawasderivedassumingaheatrejectioncapabilityof150W.m2.
For passive radiant coolers the projected area refers to total radiator and shield area refers to total
radiatorandshieldareasasprojectedontoaplanethatisparalleltotheradiatingsurface.
ResultsareshowninFigure521.AlthoughtheyhavebeenobtainedforQ=1W,theQdependenceis
surelynegligible.

74
ECSSEHB3101Part14A
5December2011

Figure521:Systemareaperunitofrefrigerationpower(orcoolingload),Ap/Mp,
forclosedcyclerefrigeratingsystemsandforpassiveradiantcoolers. Closed
cyclerefrigerators,Q=1W. Passiveradiantcoolers;Q=1W.FromHaskin&
Dexter(1979)[83].Althoughtheareas,Ap,havebeencalculatedfor1Wcooling,
theycouldbescaledinapproximatelydirectproportiontothecoolingload.Ap/Q=
7,13x107T4isthebestfitting,bytheleastsquaresmethod,tothedataforpassive
radiantcoolers. Smallesttemperatureattainedbyclosedcyclerefrigeratorsin
orbit. Smallesttemperatureattainedbypassiveradiantcoolersinorbit.From
Sherman(1982)[217].

Finally, additional considerations to be taken into account in the section of the cooling method
employedare,
1. Factors such as reliability and component availability. Closed cycle refrigerators for
greater than one year missions probably will not be available until mid 1980s. On the
otherhand,largecryogenicradiatorsarestillunproven.
2. Costsforrefrigeratorsystemsareprobablyhigherthanforpassiveradiantcoolerswhen
associatedresearchanddevelopmentisconsidered.
3. The competitive position of the refrigerator systems may improve in coming years. For
example,themerecombinationofthesolarcellarrayandtheheatrejectionradiatorinto
onestructurecouldsaveupto20%inmassand30%inrequiredarea.

75
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

BRAYTON(RotaryReciprocating BRAYTON
CYCLE
Systems) (TurbomachinerySystems)

MANUFACTURER A.D.Littlea A.D.Littlea Garrettb

TRADENAME

MODELOR
DesignStudy Prototype
PROGRAM

WORKINGFLUID Helium Helium Nitrogen

OPERATINGTEMP. 15to50 15to30 20to60


3,6 80
RANGE[K] (twostages) (twostages) (twostages)

OPERATING
PRESSURERANGE 0,30to0,73
px105[Pa]

TYPICAL 15Wat50K 15Wat30K 20Wat60K


1Wat3,6K 2Wat80K
REFRIGERATION 1Wat15K 1Wat15K 2Wat20K

SETTINGONTIME
WITHOUTLOAD 360approx.
[min]

TEMP.STABILITY

[K]

EXPANDER[rpm]

POWERINPUT[W] 1300 2160 1760 1310 375

VOLTAGE[V] 115

PHASE/FREQUENCY
3/60
[Hz]

COOLINGMEAN Water

AMBIENTTEMP.[K]

0,305Diameter 0,305Dia.
CRYOSTATSIZE[m]
1,219Length 1,016Length

COMPRESSORSIZE 0,203Dia. 0,216Dia. 0,203Dia.



[m] 0,914Length 0,914Length 0,914Length

REFRIGERATOR 70,75 91,61 83,90 56,23 68,48

76
ECSSEHB3101Part14A
5December2011

MASS[kg]

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME

[h]

AVAILABILITY

TIME[mo]

SYSTEMCOST[$]
a A.D.Little,Inc.,Cambridge,Massachusetts,USA.
b GarrettCorp.,AiResearchManufacturingCo.,Phoenix,Arizona,USA.
NOTE FromDonabedian(1972)[59].

77
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE BRAYTON CLAUDE

MANUFACTURER Hymatica AirProductsb GeneralElectricCo.c

TRADENAME

MODELOR
Prototype E311 ArmyATP USAFADP
PROGRAM

Heliumor
WORKINGFLUID Helium Helium
Neond

OPERATINGTEMP. 5to150 12to60


19to28 3,3to4,5 4,4
RANGE[K] (Threestages) (Twostages)

OPERATING
PRESSURERANGE 1,01to30,4 0,61to20,3
px105[Pa]

200Wat150K
TYPICAL 40Wat60K
0,3Wat28K 1Wat3,9K 2Wat4,4K 40Wat50K
REFRIGERATION 1,5Wat12K
3,5at5K

SETTINGONTIME
420for6,8kg
WITHOUTLOAD 30
ofCopper
[min]

TEMP.STABILITY
+/0,01
[K]

EXPANDER[rpm] 1500 365

POWERINPUT[W] 7000 9000 16000 4000

VOLTAGE[V] 208460

PHASE/FREQUENCY
3/60
[Hz]

COOLINGMEAN Air

AMBIENTTEMP.[K] 241to325

0,508x0,635x
CRYOSTATSIZE[m]
1,346 0,610Dia.e 0,508Dia.e
1,219Length 1,651Length
COMPRESSORSIZE 0,686Dia.

[m] 1,067Length

78
ECSSEHB3101Part14A
5December2011

CRYOSTATMASS
86,17
[kg]
113,38e
COMPRESSORMASS
104,31
[kg]

MEANTIME
BETWEEN 33640
FAILURES[h]

OVERHAULTIME
3000to6000
[h]

AVAILABILITY
6
TIME[mo]

40000to
SYSTEMCOST[$]
100000
a HymaticEngineeringCompany,England.
b AirProductsandChemicals,Inc.,Allentown,Pennsylvania,USA.
c GeneralElectricCo.,Schenectady,NewYork,USA.
d FromStreed,Murphy&Brna(1971)[229].
e Thesevaluesarerelatedtotheentirerefrigerator.
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

79
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE GIFFORDMcMAHON

MANUFACTURER CryogenicTechnologya

TRADENAME Cryodyne

MODELOR
400 0110 1020 350 0120/RC30
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP.
3,8to4,5 6,5to25 12to28 15to28 19to28
RANGE[K]

OPERATING
PRESSURERANGE 1to19,8 5,1to19,2 5,1to18,7 4,6to12,7 8,6to22,3
px105[Pa]

TYPICAL
1,25Wat4,2K 1Wat10K 2Wat13K 2Wat18,5K 1Wat26K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 180 60 30 45 15
[min]

TEMP.STABILITY
+/0,12 +/1 +/1 +/1 +/1,5
[K]

EXPANDER[rpm] 72 150

POWERINPUT[W] 5500 2100 6100 2100 650

VOLTAGE[V] 440 208230 208440 208230 208

PHASE/FREQUENCY
3/60 1/5060 3/5060 1/5060 3/400
[Hz]

COOLINGMEAN Air

AMBIENTTEMP.[K] 268to325 241to325 241to328 219to328

0,457Diameter 0,061x0,127x 0,330x0,203x 0,457x0,127x


CRYOSTATSIZE[m]
1,219Length 0,305 0,508 0,203

COMPRESSORSIZE 0,762x0,813x 0,431x0,381x 0,533x0,660x 0,432x0,381x



[m] 1,372 0,711 1,041 0,711

CRYOSTATMASS
56,69 13,61 14,97 9,98 2,04
[kg]

80
ECSSEHB3101Part14A
5December2011

COMPRESSORMASS
362,8 79,36 192,74 79,36 9,52
[kg]

MEANTIME
BETWEEN 13500 3000
FAILURES[h]

OVERHAULTIME
3000 1000
[h]

AVAILABILITY
6 4 3 12 3
TIME[mo]

SYSTEMCOST[$] 75000 18000 15500 13500 8000


a CryogenicTechnology,Inc.,Waltham,Massachusetts,USA.
NOTE FromDonabedian(1972)[59].

CLOSEDCYCLEREFRIGERATORS

CYCLE GIFFORDMcMAHON

MANUFACTURER CryogenicTechnologya

TRADENAME Cryodyne

MODELOR
0125 077 0120 0277 1020
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP.
19to28 25to150 19to30 40to120 13to20
RANGE[K]

OPERATING
PRESSURERANGE 9,6to18,7 5,1to20,3 9,6to18,9 5,2to19,1
px105[Pa]

TYPICAL
1Wat26K 125Wat77K 1Wat26K 3Wat77K 10Wat20K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 15 30 15 10 30
[min]

TEMP.STABILITY
+/1,5 +/2
[K]

EXPANDER[rpm] 200 150 333 72

81
ECSSEHB3101Part14A
5December2011

POWERINPUT[W] 750 6100 800 525 5600

VOLTAGE[V] 115 206 208 220

PHASE/FREQUENCY
1/5060 3/400 3/60
[Hz]

COOLINGMEAN Air

AMBIENTTEMP.[K] 241to328 241to325 292to328 219to328 269to325

0,152x0,127x 0,419x0,178x 0,038Dia. 0,025Dia.


CRYOSTATSIZE[m]
0,279 0,381 0,263Length 0,127Length

COMPRESSORSIZE 0,381x0,305x 0,686x0,381x 0,267x0,267x 0,254x0,229



[m] 0,229 0,356 0,203 x0,165

CRYOSTATMASS
4,54 12,70 2,27 1,36 14,97
[kg]

COMPRESSORMASS
22,68 52,15 9,07 5,90 192,74
[kg]

MEANTIME
BETWEEN 10000 3000
FAILURES[h]

OVERHAULTIME
3000 1000 500 3000
[h]

AVAILABILITY
4 3
TIME[mo]

SYSTEMCOST[$] 11000 18500


a CryogenicTechnology,Inc.,Waltham,Massachusetts,USA.
NOTE FromDonabedian(1972)[59].

82
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

JOULE
CYCLE GIFFORDMcMAHON
THOMSON

AirProducts
MANUFACTURER CryomechInc.a b

TRADENAME

MODELOR
GB02 GB12 AL01 AL02 J801000
PROGRAM

WORKINGFLUID Helium Nitrogen

OPERATINGTEMP.
7,5to25 9to30 23to80 23to89 77
RANGE[K]

OPERATING
PRESSURERANGE 10,1to24,3
px105[Pa]

TYPICAL
1Wat9,5K 4Wat13K 1Wat25K 10Wat30K 2Wat77K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 25 35 12 25 5
[min]

TEMP.STABILITY
+/0,5
[K]

EXPANDER[rpm] 144 3850c

POWERINPUT[W] 3000 900 3000 600

VOLTAGE[V] 220 110220 220

PHASE/FREQUENCY
1/5060
[Hz]

COOLINGMEAN Air Air

AMBIENTTEMP.[K]

0,127x0,127x 0,127x0,127x 0,064x0,064x 0,127x0,127x 0,127x0,203


CRYOSTATSIZE[m]
0,533 0,610 0,368 0,457 x0,305

COMPRESSORSIZE 0,457x0,737x
0,737x0,483x0,686
[m] 0,686

83
ECSSEHB3101Part14A
5December2011

CRYOSTATMASS
11,34 2,27 11,34
[kg]
8,16d
COMPRESSORMASS
79,36 79,36
[kg]

MEANTIME
BETWEEN 5000
FAILURES[h]

OVERHAULTIME
3000 500
[h]

AVAILABILITY
1,52
TIME[mo]

SYSTEMCOST[$] 13200 8600 10300 9000


a CryomechInc.,Jamesville,NewYork,USA.
b AirProductsandChemicals,Inc.,Allentown,Pennsylvania,USA.
c Thisvalueisrelatedtothecompressor.
d Totalmassoftherefrigerator.
NOTE FromDonabedian(1972)[59].

84
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE JOULETHOMSON

MANUFACTURER Airproductsa GarrettAiResearchb

TRADENAME

MODELOR
J303500 133386 133488 144406 800334
PROGRAM

Nitrogen/
WORKINGFLUID Nitrogen
Helium

OPERATINGTEMP. 23and77(Two
77
RANGE[K] stages)

OPERATING
PRESSURERANGE 1to157 1to178,3 1to111,4
px105[Pa]

TYPICAL 2Wat77K 0,75Wat77K


5Wat77K 3Wat77K 1Wat77K
REFRIGERATION 0,35Wat23K 2,25Wat77Kc

SETTINGONTIME
WITHOUTLOAD 6 12 6,5 8
[min]

TEMP.STABILITY
+/0,5
[K]

EXPANDER[rpm] 3850d None

POWERINPUT[W] 1050 450 650 450 460

VOLTAGE[V] 115208

PHASE/FREQUENCY
3/400
[Hz]

COOLINGMEAN Air RamAir Air RamAir AirFanc

AMBIENTTEMP.[K] 233to329

14,16x103m3 42,48x103 12,74x103 42,48x103


CRYOSTATSIZE[m] (Volume) m3(Volume) m3(Volume) m3(Volume)

COMPRESSORSIZE

[m]

REFRIGERATOR
8,16 11,34 10,20 8,84 10,20
MASS[kg]

MEANTIME
BETWEEN 1000 1000 2000
FAILURES[h]

85
ECSSEHB3101Part14A
5December2011

OVERHAULTIME
500 200 300 400 500
[h]

AVAILABILITY
4 4 6 4
TIME[mo]

SYSTEMCOST[$] 10500 9000 8000 12000


a Airproducts&Chemicals,Inc.,Allentown,Pennsylvania,USA.
b GarrettCorp.AiResearchManufacturingCompany,Phoenix,Arizona,USA.
c FromStreed,Murphy&Brna(1971)[229].
d Thisvalueisrelatedtothecompressor.
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

CLOSEDCYCLEREFRIGERATORS

CYCLE JOULETHOMSON SOLVAY

MANUFACTURER GarrettAiResearcha SantaBarbarab AirProductsc

TRADENAME Displex

MODELOR
800398 800656 I1
PROGRAM

WORKINGFLUID Nitrogen Nitrogen Helium

OPERATINGTEMP.
77 77 30to150 5to300
RANGE[K]

OPERATING
PRESSURERANGE 1to157 1to178,3 10,1to23,3 6,1to20,3
px105[Pa]

TYPICAL
1Wat77K 2,5Wat77K 2Wat77K 20Wat77K 1,5Wat77K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 4 5 16 5
[min]

TEMP.STABILITY
+/0,5 +/2
[K]

EXPANDER[rpm] None 144 385

POWERINPUT[W] 650 530 326 1700 340

VOLTAGE[V] 115208 208460

86
ECSSEHB3101Part14A
5December2011

PHASE/FREQUENCY
3/400 3/60
[Hz]

COOLINGMEAN AirFand Air Air Liquid

AMBIENTTEMP.[K] 233to329 278to316 Upto338

42,48x103m3 37,66x103m3 0,178Dia. 0,041Dia. 0,038x0,025x


CRYOSTATSIZE[m] (Volume) (Volume) 0,318Length 0,203Length 0,025

COMPRESSORSIZE 0,330x0,737 0,127x0,140x



[m] x0,330 0,184

CRYOSTATMASS
4,81 0,23
[kg]
10,43 e 9,07 e 7,26 e

COMPRESSORMASS
59,96 4,76
[kg]

MEANTIME
BETWEEN 1000 2000 5000to10000 3086
FAILURES[h]

OVERHAULTIME 300
100 500 5000
[h] 1000d

AVAILABILITY
4 3 3
TIME[mo]

SYSTEMCOST[$] 10000 7000 10000 5000to10000


a GarrettCorp.,AiResearchManufacturingCompany,Phoenix,Arizona,USA.
b SantaBarbaraResearchCenter,Goleta,California,USA.
c AirProducts&Chemicals,Inc.,Allentown,Pennsylvania,USA.
d FromStreed,Murphy&Brna(1971)[229].
e Totalmassoftherefrigerator.
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

87
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE SOLVAY

MANUFACTURER AirProductsa Kinergeticsb

TRADENAME Displex

MODELOR
CS102 CS202 CS1003 MS1003 SRC07
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP.
30to200 30to300 50to300 30to77 50to77
RANGE[K]

OPERATING
PRESSURERANGE 10,1to25,3
px105[Pa]

TYPICAL
17Wat77K 1Wat17K 1Wat77K 1Wat38K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 20 45 5
[min]

TEMP.STABILITY

[K]

EXPANDER[rpm]

POWERINPUT[W] 1700 1735 400 368 400

VOLTAGE[V] 230 115

PHASE/FREQUENCY
1/60 3/60 3/400
[Hz]

COOLINGMEAN Air

AMBIENTTEMP.[K] 277to317 Upto330

0,102x0,102x 0,013Dia. 0,032Dia.


CRYOSTATSIZE[m]
0,381x0,432x 0,432 0,127Length 0,095Length
0,559c
COMPRESSORSIZE 0,559x0,432x 0,533x0,381x 0,184Dia.

[m] 0,381 0,279 0,216Length

CRYOSTATMASS
1,50 0,23 5,62
[kg]

88
ECSSEHB3101Part14A
5December2011

COMPRESSORMASS
68,03 68,03 27,21 6,35
[kg]

MEANTIME
BETWEEN 3100
FAILURES[h]

OVERHAULTIME
3000 4500 1200
[h]

AVAILABILITY
Immediate
TIME[mo]

SYSTEMCOST[$]
a AirProducts&Chemicals,Inc.,Allentown,Pennsylvania,USA.
b Kinergetics(FormerlytheSubmarineSystemDivisionofSterlingElectronics).Tarzana,California,USA.
c Totalsizeoftherefrigerator.
NOTE FromDonabedian(1972)[59].

CLOSEDCYCLEREFRIGERATORS

CYCLE STIRLING

MANUFACTURER HughesAircraftCo.a MalakerCorp.b

TRADENAME Cryomite

MODELOR
VIIC VIIR
PROGRAM

WORKINGFLUID Helium Helium

OPERATINGTEMP.
25 80 80 25to77 40to125
RANGE[K]

OPERATING
PRESSURERANGE
px105[Pa]

TYPICAL 2Wat25K
0,8Wat25K 1,5Wat80K 2Wat77K 60Wat77K
REFRIGERATION 1Wat22Kc

SETTINGONTIME
WITHOUTLOAD 15 3 5 8 38
[min]

TEMP.STABILITY
+/0,02
[K]

89
ECSSEHB3101Part14A
5December2011

EXPANDER[rpm]

POWERINPUT[W] 620 500 580 480 1220

VOLTAGE[V] 115 208

PHASE/FREQUENCY
3/400 3/400
[Hz]

COOLINGMEAN Air AirorLiquid AirorLiquid Air

AMBIENTTEMP.[K] 241to327 218to344 MILSTD210B

0,216x0,127x 0,211x0,142x 0,211x0,127x 0,122Dia. 0,165Dia.


CRYOSTATSIZE[m]
0,254 0,152 0,152 0,292Length 0,597Length

COMPRESSORSIZE
None None
[m]

CRYOSTATMASS
7,26 4,54 5,90
[kg]
7,03d 18,14d
COMPRESSORMASS

[kg]

MEANTIME
BETWEEN 40000
FAILURES[h]

OVERHAULTIME
500 1000
[h]

AVAILABILITY
Immediate
TIME[mo]

SYSTEMCOST[$] 5195 17500


a HughesAircraftCo.,CulverCity,California,USA.
b MalakerCorp.,HighBridge,NewJersey,USA.Alloperationsonthisfirmwereterminatedin1972.
c FromStreed,Murphy&Brna(1971)[229].
d Totalmassoftherefrigerator.
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

90
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE STIRLING

MANUFACTURER MalakerCorp.a

TRADENAME Cryomite

MODELOR
MarkXX MarkXIVA MarkXV MarkXVI3 MarkXV4
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP.
40to120 45to100 54to100 77to110 54to100
RANGE[K]

OPERATING
PRESSURERANGE
px105[Pa]

TYPICAL
110Wat77K 2,8Wat77K 1Wat77K 8,2Wat77K 1Wat77K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 7,4 7 8 7 8
[min]

TEMP.STABILITY

[K]

EXPANDER[rpm]

POWERINPUT[W] 1990 108 29,5 208 29,5

VOLTAGE[V] 208 24DC

PHASE/FREQUENCY
3/400
[Hz]

Free
COOLINGMEAN AirorLiquid Air
Convection

AMBIENTTEMP.[K] MILSTD210B

0,483x0,457x 0,074Dia. 0,074Dia. 0,089Dia. 0,076Dia.


CRYOSTATSIZE[m]
0,406 0,337Length 0,311Length 0,356Length 0,286Length

COMPRESSORSIZE

[m]

REFRIGERATOR 29,48 2,49 2,27 4,54 2,27

91
ECSSEHB3101Part14A
5December2011

MASS[kg]

MEANTIME
BETWEEN 40000
FAILURES[h]

OVERHAULTIME
1000
[h]

AVAILABILITY
Immediate
TIME[mo]

SYSTEMCOST[$] 24000 9000 8500 9000


a MalakerCorp.,HighBridge,Newjersey,USA.Alloperationsofthisfirmwereterminatedby1972.
NOTE FromDonabedian(1972)[59].

CLOSEDCYCLEREFRIGERATORS

CYCLE STIRLING

PhilipswithAPL
MANUFACTURER Philipsa
JohnsHopkinsa

RhombicDrive
TRADENAME Cryogem MicroCryogem
Cooler

Nuclear
MODELOR Monitoring
42100 P/N460600
PROGRAM ResearchOffice
DOD,ARPA

WORKINGFLUID Helium

OPERATINGTEMP. 64to70and135to
7to300 20to40 40to300 40to80
RANGE[K] 150(Twostages)b

OPERATING
PRESSURERANGE 3,7to6,1 8,6to17,2 4,6to8,6 4,8c
px105[Pa]

TYPICAL 0,5Wat12 1,5Wat140Kb,c


2Wat30K 1,5Wat77K 1Wat50K
REFRIGERATION K 0,3Wat<75K

SETTINGONTIME
WITHOUTLOAD 15 10 3 4
[min]

TEMP.STABILITY

[K]

92
ECSSEHB3101Part14A
5December2011

1000canvary
EXPANDER[rpm] 600 1750 1800
within+/200b

POWERINPUT[W] 700 350 90 120 30b,c

VOLTAGE[V] 320 208 24DC 115 24to30DC

PHASE/FREQUENCY
3/60 3/400 3/400
[Hz]

COOLINGMEAN Water Air AirorLiquid Air

267to322for
storage.b
AMBIENTTEMP.[K] 219to316 218to348 272to318for
conductionheat
rejection.

0,152x0,305 0,152x0,127 0,102x0,102 0,038Dia.


CRYOSTATSIZE[m]
x0,610 x0,279 x0,203 0,267Length 0,154x0,180x
0,307b,c
COMPRESSORSIZE

[m]

5,3without
REFRIGERATOR electronicsb
15,88 5,44 1,36 1,81
MASS[kg] 7,2with
electronicsb,c

MEANTIME
BETWEEN 1500
FAILURES[h]

OVERHAULTIME
600 500 1000 15000c
[h]

Immediate.Four
AVAILABILITY refrigeratorswere
24 Immediate
TIME[mo] launchedonFeb.
24,1979.

SYSTEMCOST[$] 8500 6500


a NorthAmericanPhilipsCorporation,BriarcliffManor,NewJersey,USA.
b FromBalas,Leffel,&Wingate(1978)[16].
c FromSherman(1982)[217].SeealsoNaes&Nast(1980)[160].
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

93
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE STIRLING

PhilipsunderNASA EnergyResearchand
MANUFACTURER Philipsc
Goddarda Generation,Inc.b

TRADENAME MagneticBearingCooler GasBearingCooler

MODELOR
Prototype DesignStudy N20d
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP.
65to300e,f,g 65to300g 12d
RANGE[K]

OPERATING
PRESSURERANGE 16e
px105[Pa]

TYPICAL
5Wat65Ke 11,3Wat65Kg 10Wat20Kd
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 15d
[min]

TEMP.STABILITY

[K]

EXPANDER[rpm] 1650g 1750d

180.g
POWERINPUT[W] 48g 1750d
Cooleralone,96

VOLTAGE[V] 2000VAd

PHASE/FREQUENCY
/28f 3/5060d
[Hz]

COOLINGMEAN Airg Airg AirorLiquidd

AMBIENTTEMP.[K] 300

0,063Diameterh 0,032Diameterg 0,102Diameterd


CRYOSTATSIZE[m]
0,750Length 0,063Length 0,191Length

COMPRESSORSIZE 0,370Diameterh 0,042Diameterg


0,483x0,356x0,330d
[m] 1,050Length 0,237Length

94
ECSSEHB3101Part14A
5December2011

REFRIGERATOR
20withoutelectronicsf 3g 50,79d
MASS[kg]

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME
26000to44000g 26000to44000g 4000d
[h]

AVAILABILITY

TIME[mo]

SYSTEMCOST[$]
a NorthAmericanPhilipsCorporation,BriarcliffManor,NewJersey,USA.
b EnergyResearchandGenerationCorporation,Oakland,California,USA.
c PhilipsResearchLaboratories,Eindhoven,Netherlands.
d FromDonabedian(1972)[59].
e FromSherman(1982)[217].SeealsoNaes&Nast(1980)[160].
f FromDaniels,Gasser&Sherman(1982)[49].
g FromSherman,Gasser,Goldowsky,Benson&McCormick(1980)[221].
h EstimatedbythecompilerafterFig.9ofSherman(1982)[217]andavolumeof5,4x103m3quotedby
Sherman,Gasser,Goldowsky,Benson&McCormick(1980)[221].

CLOSEDCYCLEREFRIGERATORS

CYCLE TACONIS VUILLEUMIER

BritishOxygen
MANUFACTURER a
Garrettb HughesAircraftCo.c

TRADENAME

MODELOR
IR16MKII ICICLE FlightTestModels
PROGRAM

WORKINGFLUID Helium Helium Helium

OPERATINGTEMP.
12to23 75 25to75 70to90 75to90
RANGE[K]

OPERATING
PRESSURERANGE 10,1to20,2 48,1to55,3
px105[Pa]

TYPICAL 2Wat25K
2,5Wat12K 5Wat75K 0,2Wat85K 1,5Wat75K
REFRIGERATION 3Wat75K

95
ECSSEHB3101Part14A
5December2011

SETTINGONTIME
25for0,1kgof
WITHOUTLOAD 40 15 10
Copper
[min]

TEMP.STABILITY[K]

EXPANDER[rpm] 166 400 900

POWERINPUT[W] 2640 3665d 1200 180 200

VOLTAGE[V] 240 115 28DC 28DC

PHASE/FREQUENCY
1/50 3/400
[Hz]

HeatPipeto HeatPipeto
COOLINGMEAN AirandWater Liquid Air
Radiatore Radiator

AMBIENTTEMP.[K] 300Approx. 255to322 394 218to344

0,203Dia. 0,191x0,241x 0,165x0,145


CRYOSTATSIZE[m] Notspecified
0,330Length 0,254 x0,127

COMPRESSORSIZE 0,914x0,533x
None None
[m] 0,686

CRYOSTATMASS Refrigerator:
9,07
[kg] 3,2;Isotope&
Structure:11,1, 8,84f 3,81f 2,61f
Insulation,
COMPRESSORMASS
108,84 activecontrol:
[kg]
8e

MEANTIME
Objective:
BETWEENFAILURES 6000g 3000 1000
5000h
[h]

OVERHAULTIME[h] 17000to43000 1000 3000 1000h

AVAILABILITYTIME

[mo]

SYSTEMCOST[$] 9100

96
ECSSEHB3101Part14A
5December2011
a BritishOxygenCo.LTD.,London,England.
b GarrettCorp.,AiResearchManufacturingCompany,Phoenix,Arizona,USA.
c HughesAircraftCo.,CulverCity,California,USA.
d Powersuppliesareboththermal(350W)andelectrical(15W).200Warefordisplaceraction,theremaining
accountsforinsulationslosses(mainlyisotopehousingandhotheatpipe).Theisotopesource,coupledtothe
refrigeratorviathehotheatpipe,providesthethermalpower(Shelpuk,Crouthamel&Cygnarowicz(1970)
[215].
e FromShelpuk,Crouthamel&Cygnarowicz(1970)[215].
f Totalmassoftherefrigerator.
g FromSherman(1978)[216].
h FromStreed,Murphy&Brna(1971)[229].
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

CLOSEDCYCLEREFRIGERATORS

CYCLE VUILLEUMIER

MANUFACTURER HughesAircraftCo.a

TRADENAME

MODELOR
Prototype X447550 SESP712
PROGRAM

WORKINGFLUID Helium

OPERATINGTEMP. 30and75(Two 15and60


Around77b 15to75 30to75
RANGE[K] stages) (Twostages)

OPERATING
PRESSURERANGE Upto27,4 33,4to41,5
px105[Pa]

TYPICAL 0,5Wat30K 0,15Wat15K


0,6Wat77Kb 0,15Wat15K 0,5Wat30K
REFRIGERATION 6Wat75K 3,5Wat55K

SETTINGONTIME
WITHOUTLOAD 30
[min]

TEMP.STABILITY
+/0,1
[K]

EXPANDER[rpm] 240 260

POWERINPUT[W] 60b 370 550 480 540

VOLTAGE[V] 28DC 2430DC

97
ECSSEHB3101Part14A
5December2011

PHASE/FREQUENCY

[Hz]

COOLINGMEAN Airb liquid

AMBIENTTEMP.[K]

CRYOSTATSIZE[m] 0,267x0,345x0,198

COMPRESSORSIZE
None
[m]

REFRIGERATOR
4,35 27,21
MASS[kg]

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME Objectiveb
Objectiveb10000 4300
[h] 3000

AVAILABILITY

TIME[mo]

SYSTEMCOST[$]
a HughesAircraftCo.,CulverCity,California,USA.
b FromStreed,Murphy&Brna(1971)[229].
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

CLOSEDCYCLEREFRIGERATORS

CYCLE VUILLEUMIER

MANUFACTURER HughesAircraftCo.a

TRADENAME

MODELORPROGRAM Prototypeb HICAPc

WORKINGFLUID Helium

OPERATINGTEMP. Around8,5,11and50(three
50to120b
RANGE[K] stages)c

OPERATINGPRESSURE

RANGEpx105[Pa]

TYPICALREFRIGERATION 0,6Wat77Kb <1Wat12Kc

98
ECSSEHB3101Part14A
5December2011

10Wat33K
17Wat75K

SETTINGONTIME
20b
WITHOUTLOAD[min]

TEMP.STABILITY[K] +/5b

EXPANDER[rpm] Sixvaluesbetween100and300c

200to300for28Vsource.c,d
POWERINPUT[W] 105b
1200to1800for100Vsource.

VOLTAGE[V] 28DCand100DC

PHASE/FREQUENCY[Hz]

COOLINGMEAN Coolanol20c

AMBIENTTEMP.[K] 278to313c

CRYOSTATSIZE[m] 0,889x0,762x0,317c

COMPRESSORSIZE[m] None

68plusc
REFRIGERATORMASS[kg] 2,95b
14,5(electronicinterfaceunit)

MEANTIMEBETWEEN
3000b
FAILURES[h]

OVERHAULTIME[h] 1000b 20000c

AVAILABILITYTIME[mo]

SYSTEMCOST[$]

a HughesAircraftCo.,CulverCity,California,USA.
b FromDonabedian(1972)[59].
c FromDoody(1980)[60].
d Atpresentelectricallypoweredfromresistanceheaters(Doody(1980)[60]).Thermalenergystorageunitsare
beingdeveloped(Richter&Mahefkey(1980)[195]).

99
ECSSEHB3101Part14A
5December2011

CLOSEDCYCLEREFRIGERATORS

CYCLE VUILLEUMIER

MANUFACTURER Kinergeticsa Philipsb RCAc

TRADENAME

MODELOR
Prototype Prototype
PROGRAM

WORKINGFLUID Helium Helium Helium

OPERATINGTEMP.
77 77to200 77 60to100 77
RANGE[K]

OPERATING
PRESSURERANGE 28,4to38,5 30,4(Highpressure)
px105[Pa]

TYPICAL
0,4Wat77K 0,5Wat77K 1wat77K 0,5Wat77K 1,6Wat77K
REFRIGERATION

SETTINGONTIME
WITHOUTLOAD 9 20
[min]

TEMP.STABILITY

[K]

EXPANDER[rpm] 750 600

POWERINPUT[W] 65 70 120 90 Thermal

VOLTAGE[V]

PHASE/FREQUENCY

[Hz]

COOLINGMEAN Air Air Air

AMBIENTTEMP.[K]

0,305x0,203 1,067x0,457
CRYOSTATSIZE[m]
x0,152 x0,457

COMPRESSORSIZE

[m]

REFRIGERATOR
2,72 36,73 6,80 4,67 7,26
MASS[kg]

100
ECSSEHB3101Part14A
5December2011

MEANTIME
BETWEEN 3000
FAILURES[h]

OVERHAULTIME
1000
[h]

AVAILABILITY

TIME[mo]

SYSTEMCOST[$]
a Kinergetics(FormerlytheSubmarineSystemsDivisionofSterlingElectronics),Tarzana,California,USA.
b NorthAmericanPhilipsCorporation,BriarcliffManor,NewJersey,USA.
c RCA,defenseElectronicsLaboratory,Camden,NewJersey,USA.
NOTE FromDonabedian(1972)[59].
80KCOOLER(BAecooler)
Manufacturer:BritishAerospaceSpaceSystems,Ltd.BAehasmanufactured3batchesof6coolers.
Description:Oxfordtypecooler.Singlestage80KsplitStirlingcycle.

Figure522:80Kcoolerschematic.FromJewell(1991)[103].

Mass
Compressor:3,0kg.
Displacer:0,9kg.
Electronics:4,5kg.
Size
Compressor:length0,20m,diameter0,12m.

101
ECSSEHB3101Part14A
5December2011
Displacer:length0,19m,diameter0,075m.
Electronics:0,225mx0,230mx0,150m.
Powerconsumption(typical):30W
Workingfluid:Helium.
Performance(typical):0,8Wcoolingat80K(for30Winputpower).
Performancetesting
Coolerno.6of1stbatch.
Ambienttemperature293K.
Inputpower1535W.

Figure523:Coolerheatliftperformancevs.grosscompressorinputpower.From
Scull&Jewell(1991)[211].

References:Jewell(1988,1991)[104]&[103],Lewis(1988)[133],Scull&Jewell(1991)[211].
20KCOOLER
Manufacturer/Developer: British Aerospace Space Systems, Ltd. (Engineering model). Rutherford
AppletonLaboratory(Developmentmodel).
Description:Twostage20KsplitStirlingcycle(basedonOxfordtypetechnology).

102
ECSSEHB3101Part14A
5December2011

Figure524:20Kcoolerschematic.FromJonesetal.(1991)[110].

Mass
Compressor(each):3,5kg.
Displacer:1,3kg.
Electronics:4,5kg.
Size
Compressor(each):0,13mx0,14mx0,16m.
Displacer:0,075mx0,075mx0,21m.
Electronics:0,251mx0,254mx0,186m.
Powerconsumption
Compressor(each):35W.
Displacer:2W.
Electronics:30W.
Workingfluid:Helium.
Performance
60mWat20K.
300mWat30K.
Performancetesting

103
ECSSEHB3101Part14A
5December2011

Figure525:Heatliftperformanceof:a)developmentmodel;b)engineering
model.FromJonesetal.(1991)[110].

References:Jewell(1991)[103],Jonesetal.(1991)[110].
4KCOOLER
Developer:RutherfordAppletonLaboratory.
Description: Two stage Stirling cooler (20 K cooler above) and a precooler for a 4 K Joule Thomson
stage.Developmentmodel.

Figure526:4Kcoolerlayout.FromBradshaw&Orlowska(1988)[27].

Mass
Size
Workingfluid:Helium.
Performance
Precooler:300mWofcoolingat30Kfor~80Winputpower.
4Kstage:1,5mWat4,17Kfor126Winputpower,withamassflowof3,9mg/s.(Goal:5mWat4K
for<150W)

104
ECSSEHB3101Part14A
5December2011
Performancetesting

Figure527:Coolingpower/massflowvs.precoolertemperature.FromBradshaw
&Orlowska(1991)[28].

NOTE:A2,5Kprototypecoolerpracticallyidenticaltothe4Kcooler(exceptfortheuseofHe3gas)is
underdevelopment.Performancegoal:50Wat2,5K.
References:Bradshaw&Orlowska(1988,1991)[27]&[28],Jewell(1991)[103].

5.3 Open cycle


Opencycle or expendable refrigeration systems of interest for space application include the use of
highpressure gas combined with JouleThomson expansion valve, cryogenic liquids either in the
subcriticalorsupercriticalstate,cryogenicsolidsand,forsomespecialapplications,liquidswhichcan
bestoredatroomtemperature.
The basic attractiveness of these system is based on: simplicity, reliability, relative economy and
negligiblepowerrequirements.
Thebasicdisadvantagesarethelimitedlifeduetoheatleakage,andtherapidincreaseinmassand
volumeforextendedmissions.
In the following pages, main emphasis is placed on highpressure gas JouleThomson cryostats.
Cryostatsforstoredsolid,liquidorgaseouscryogensareshortlydiscussedinclause5.3.1andmore
extensivelyinClause6ofthisPart.

5.3.1 Joule-Thomson open cycle refrigerators

5.3.1.1 Description
The JouleThomson cryogenic cooler is based on the cooling caused by the adiabatic expansion of a
gas.

105
ECSSEHB3101Part14A
5December2011
When a real gas is expanded at constant enthalpy (JouleThomson or throttling process) the total
temperature does not remain constant, opposite to what happens with a perfect gas. Rather, if the
temperatureofthegasisbelowacertaininversiontemperature(whichdependsonpressure)thetotal
temperaturewilldecreasewhenthepressuredecreases,whichwillresultinacoolingofthegas.
Letp0andT0bethepressureandtemperatureinagasafterthrottlingfromgiveninitialconditionspi
p0andTi.WhenT0isplottedvs.p0acurveexhibitingamaximumvalueofT0results.Similarcurves
will be obtained for different initial conditions, see the isenthalps, thin nearly horizontal lines in
Figure 528. The local slope, , of the curves is the JouleThomson (or JouleKelvin) coefficient, =
(T/p)h,wherehindicatesthattheevolutionfromitootakesplaceatconstantenthalpy.
The JouleThomson inversion temperature curve, heavy lines in Figure 528, is the locus of points
where =0.Intheregionwhere >0anisenthalpicexpansionwilldecreasethetotaltemperatureof
thegas,andconverselywhere<0.
Forperfectgases=0everywhere(constantenthalpymeansconstanttemperature).

Figure528:Isenthalpsandinversioncurvefordifferentgasses.aHydrogen.b
Helium.cNitrogen.FromZemansky(1968)[272].Datainb,afterHill&
Loumasmaa(1960)[89],arenolongervalidforabove20K.Upperisenthalpsare
insteadfromAngus&deReuck(1977)[6],pp.64127.Thelocusofthemaximahas
beendrawnbythecompilerasadottedline.

Purely thermodynamic considerations indicate that the expansion should start at the inversion
temperaturecurve.Forcertaingasesthemaximuminversiontemperature(thatcorrespondingtozero
pressure) is below room temperature (see for example Tmax for Hydrogen or Helium in Figure 528)
and, thence, precooling to the indicated temperature is required before JouleThomson expansion.
Manyothergases(forexampleNitrogeninFigure528)havemaximuminversiontemperaturesabove
roomtemperatureandnoprecoolingisrequired.
ValuesofTmaxandpmaxfortypicalgasesaregiveninTable55.

106
ECSSEHB3101Part14A
5December2011

Table55:MaximumInversionTemperatureandPressureofSelectedCryogens
Tmax[K] Pmaxx107[Pa]
CRYOGEN
Calculateda Calculatedb Reportedc Calculateda Reportedc

Ar 775
1018 723 4,50
Argon 789

CO2
1223
Carbon 2053 1500 6,64
1326
Dioxide

H2 189
224 202 1,18 1,67
Hydrogen 239

He 39
35 40 0,20 0,38
Helium 66

N2 615
852 621 3,06 4,05
Nitrogen 620
a CalculatedbythecompilerthroughvanderWaalstwoconstantequation.Tmax=27Tc/4,Pmax=9pc.SeeLoeb
(1961)[135].CriticalpointconditionshavebeentakenfromTable81,Clause8.1.1.
b CalculatedbythecompilerfromthevirialequationofstateandforceconstantsfortheLennardJones
intermolecularpotential.Tmax=6,47/k.Heliumexhibitsquantumdeviationsfromthislaw.SeeHirschfelder,
Curtiss&Bird(1954)[91],Fig.6.63,p.174andTableIA,pp.11101112.
c FromZemansky(1968)[272].

JouleThomsoncryostatsareusedtocoolinfraredsensorsinelectroopticaldevices.Normallysensor
and cryostat are packaged inside a double walled glass dewar. A schematic of the cryostatdewar
systemisillustratedinFigure529.

Figure529:SchematicofatypicalJTcryostatdewarsystem.FromHellwig(1980)
[86].

Gasstoredinthesupplytankflowsthroughtheheatexchanger23whereitiscooledbytheexitgas
56. Between 3 and 4 the gas is expanded at constant enthalpy through the expansion or throttling
valve. The gas becomes partiallyliquefied in the expansion chamber where itis vaporizedagain by
contactingthewarmdetector.Theoutgoingvaporflowsthroughtheheatexchanger56.
In the simpler design the gas is supplied continuously from a high pressure source. More
sophisticated devices use some kind of gas flow regulation. Modern JT cooled cryostats could have
either a rapidstarting capability, an external control system to operate on a selfdemand basis, or
both.

107
ECSSEHB3101Part14A
5December2011
Normalflowcontrolconsists,Figure530,inaneedleactuatedbyapressurizedbellowswhichallows
forwideopenflowuntiltheliquidbuildupintheexpansionchambercondensesthepressurantand
thebellowsisdrawnintothecryostatrestrictingflow.

Figure530:SchematicofaselfdemandflowJTcryostatdewarsystem.FromOren
&Gutfinger(1979)[175].Thesketchofthevariableorificecontrollingdeviceis
fromBuller(1970)[35].

Inseveralinstances(Buller(1970)[35])thebellowsdoesnotrespondtothepresenceoftheliquidat
thedetector,ratheritisthermallyattachedtotheheatexchangersupportasmalldistanceawayfrom
thecoldend.Thebellowsispressurizedsothatwhenitactuatesliquidandvaporcoexistinsideand,
thence,theinternalpressure,bellowslengthandneedlepositiononlydependontemperature.
Bimetallicelementsorelectroniccircuitsarealsousedforflowcontrol.Forinstance,Peterson,Wurtz
&Winner(1976)[180]describesanimpedanceresponsiveelectronicswitchcircuitwhichenergizesthe
windingofanormallyopensolenoidvalvewhenliquidnitrogenwetsanabsorbingpacking.

5.3.1.2 Development problems


ManymanufacturersareinvolvedinanumberofopencycleJouleThomsoncoolingunitsforspecific
applications.
The most common problem of JT cryostats is clogging of the finned tube which constitutes the heat
exchanger. Dust particles and traces of highfreezing point gases should be excluded from the gas
entering the tube. To avoid clogging, a chemical dryer containing an absorbent material which
removestheexcessmoisture,andaporoussinteredmetalfilter,toremoveanyfloatingparticle,canbe
insertedinthehighpressurelineaheadofthefinnedtube.
Thestoredgasmassandvolumearetheprimarypenaltyfactorintheseunits.
General considerations for the design and optimization of highpressure storage vessels are fairly
wellknown.Basically,theserelatetothedesirabilityofminimizingcontainervolumepenaltiesbythe
useoflargestoragepressureswithoutincurringexcessivepressureshellmass.
Gascompressibilityeffectsareofextremeimportanceinpressurevesseldesign.Atpressuresover106
Pagasesbecomelessandlesscompressiblesothatvolumesavingathighpressuresarediminished.It
is therefore of extreme importance to investigate in detail the effect of initial pressure, on both
containermassandvolume,intheoptimizationofstoragevesseldesign.Althoughpmax,Figure528,
couldbeagoodstartingpointintheestimationoftherequiredinitialpressure,reliabledataonTmax
are not easily found. On the other hand, the accuracy of the computed data, based on diluted gas
theories,isathighpressures.

108
ECSSEHB3101Part14A
5December2011

5.3.2 Existing systems


Technicaldataconcerningopencyclerefrigeratorsarecollectedinthisclause.Thesedataarearranged
asshownintheTablebelow.

ArrangementoftheCompiledDataonOpenCycleRefrigerators

Cycle Manufacturer Noa

JouleThomson AirProducts&Chemicals,Inc. 6
HughesAircraftCo. 1
HymaticEngineeringCo. 2
SantaBarbaraResearchCenter 4
a Numberofrefrigeratorslisted.

OPENCYCLEREFRIGERATORS

CYCLE JOULETHOMSON

MANUFACTURER AirProductsa

TRADENAME CryoTip

MODELOR
AC1 AC2 AC2L AC3L AC1101A
PROGRAM

Liquid
Liquid
Nitrogenor Nitrogen,
WORKINGFLUID Nitrogen Nitrogenor Nitrogen
Hydrogen Hydrogenor
Hydrogen
Helium

OPERATING
68to300 16to300 16to300 3,6to300 80to300
TEMP.RANGE[K]

OPERATING
80b
PRESSURE 80b 80b
1to152c
RANGEpx105[Pa]

TYPICAL
7Wat80K 4Wat23K 6Wat22K 0,5Wat4,4K 0,75Wat80K
REFRIGERATION

TEMP.STABILITY
+/0,10
[K]

COOLDOWN
TIME(NOLOAD) 5 10 40
[min]

109
ECSSEHB3101Part14A
5December2011

GASFLOWRATE
0,401(LN2)
ATTYPICAL 1,79(N2) 0,401(LN2)
1,58 0,127(H2) 0,345
REFRIGERATION 0,113(H2) 0,142(H2)
0,354(He)
[kg.h1]

REQUIRED
None 30deg.ofvertical None
ATTITUDE

COOLINGMEAN Nonec

AMBIENTTEMP.

[K]

CRYOSTATSIZE 0,089x0,089 0,089Dia. 0,152Dia. 0,152Dia.



[m] x0,127 0,191Length 0,432Length 0,584Length

CRYOSTATMASS
0,453 0,680 9,98 9,98 0,227
[kg]

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME
1000c
[h]

PRIMARYUSE Laboratory

AVAILABILITY
Immediate
TIME[mo]

SYSTEMCOSTd
1585 2690 3395 4430 2850
[$]
a AirproductsandChemicals,Inc.,Allentown,Pennsylvania,USA.
b Minimumsupplypressure.
c FromStreed,Murphy&Brna(1971)[229].
d Notincludesgasstorage.
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

110
ECSSEHB3101Part14A
5December2011

OPENCYCLEREFRIGERATORS

CYCLE JOULETHOMSON

MANUFACTURER AirProductsa Hughesb HymaticEngineeringc

TRADENAME CryoTip Minicooler

MODELOR
AC2209 JTO001 MAC215 MC8
PROGRAM

Argon,
Nitrogenor
WORKINGFLUID Argon HydrogenorAir Nitrogenor
Hydrogend
Air

OPERATINGTEMP.
87 21,27 80to90
RANGE[K]

OPERATING
187,2(N2)d,e
PRESSURERANGE 116e 120e
118,2(H2)
px105[Pa]

TYPICAL
0,25at22Kd 40Wat87K 2Wat21K 10Wat80K
REFRIGERATION

TEMP.STABILITY
+/0,10
[K]

COOLDOWNTIME 12from245Kd
0,2 3 0,5
(NOLOAD)[min] 23from295K

GASFLOWRATE
ATTYPICAL 0,320to0,355(N2)d 0,055(H2)
0,113 0,128(N2)
REFRIGERATION 0,023to0,026(H2) 0,118(Air)
[kg.h1]

REQUIRED
None None
ATTITUDE

COOLINGMEAN

AMBIENTTEMP.

[K]

0,0076
0,006Diameterd 0,025Diameter
CRYOSTATSIZE[m] Diameter
0,130Length 0,063Length
0,051Length

CRYOSTATMASS
0,068 0,0056f
[kg]

111
ECSSEHB3101Part14A
5December2011

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME

[h]

PRIMARYUSE MarinerMarsd FalconMissile

AVAILABILITY
Immediate Immediate
TIME[mo]

SYSTEMCOST[$] 4300
a AirProductsandChemicals,Inc.,Allentown,Pennsylvania,USA.
b HughesAircraftCo.,CulverCity,California,USA.
c HymaticEngineeringCo.LTD,The,Redditch,Worcs.England.
d FromHughes&Herr(1973)[97].
e Minimumsupplypressure.
f FromAnon(1966)[7].
NOTE DatainthistablearefromDonabedian(1972)[59]unlessotherwisestated.

OPENCYCLEREFRIGERATORS

CYCLE JOULETHOMSON

MANUFACTURER SantaBarbaraa

TRADENAME

MODELOR
9159 41614 42902
PROGRAM

WORKINGFLUID Nitrogen Argon ArgonorNitrogenb Argonc

OPERATING 87to100
77to300 87to300 77to300
TEMP.RANGE[K] c

OPERATING
PRESSURE 50to550d
RANGEpx105[Pa]

TYPICAL
0,2W
REFRIGERATION

TEMP.STABILITY
+/5e
[K]

COOLDOWN 0,033f 0,5c

112
ECSSEHB3101Part14A
5December2011

TIME(NOLOAD)
[min]

GASFLOWRATE 0,022(Ar)at4,15x107
ATTYPICAL 0,294at0,69x10 Pa
7 0,028at5,50x10 7 Pa
0,032c
REFRIGERATION 0,490at1,10x107Pa Pa 0,026(N2)at3,40x107
[kg.h1] Pa

REQUIRED
None
ATTITUDE

COOLINGMEAN

AMBIENTTEMP.
300
[K]

CRYOSTATSIZE 5,28x103Dia.g 5,28x103Dia.g 5,28x103Dia.g



[m] 0,051Length 0,025Length 0,051Length

CRYOSTATMASS

[kg]

MEANTIME
BETWEEN
FAILURES[h]

OVERHAULTIME

[h]

PRIMARYUSE IRsensorscooling

AVAILABILITY

TIME[mo]

SYSTEMCOST[$]
a SantaBarbaraResearchCenter,Goleta,California,USA.
b AdditionaldataonthisrefrigeratoraregiveninOren&Gutfinger(1979)[175].
c FromHellwig(1980)[86].
d Supplypressure.
e 42902isademandflowcryostatwhichcanbeadjustedtooperateeitherwithconstantdetectortemperature
orwithsometemperaturefluctuation(asstated)toobtainmaximumgaseconomy.
f Atmaximumoperatingpressure,beforeactuationofaflowrestrictingneedle.
g Thesevaluesarerelatedtotheheatexchangeronly.
NOTE DatainthistablearefromBuller(1970)[35]unlessotherwisestated.

113
ECSSEHB3101Part14A
5December2011

5.3.3 Stored liquid or solid cryogen open refrigerators

5.3.3.1 Description
Cryogensmaybestored:1)Asliquidinequilibriumwiththeirvapors,subcriticalstorage,2)athigher
pressuresandtemperatures,assupercritical,homogeneousfluids,and3)assolidcryogens.
1. Under normal gravity conditions, the fluid is usually stored in the twophase form
becauseoftheassociatedmasssaving.
Temperatures available using most common liquefied gases, in the subcritical storage
systems,rangefromapproximately4to77K.Byvaryingthepressure,thetemperatureof
aliquefiedgascanbevariedtoprovidecoolingfromthetriplepointtothecriticalpoint.
Aconstanttemperaturecoolingcanbeprovided,withinthisrange,bypressurecontrol.
The cooling systems which use this concept could be of two types, namely: the direct
contact,orintegralcoolers,andtheliquidfeedcoolers.
2. Inthespacebornesystem,theabsenceofgravityoraccelerationorientationimpedesthe
use of the standard twophase systems, since random orientation of the liquid vapor
interface during weightlessness could result in liquid ejection through the venting duct
(seeclause6.4.1).
Spacestoragesystemofcryogenicliquidscanalsobeaccomplishedbypressurizationof
the cryogen to supercritical pressures. Now the absence of gravity or acceleration
orientation forces does not effect the delivery or venting of fluid since the supply and
venting ports are, at all times, in direct communication with a relatively homogeneous
fluid.However,thetotalmassperunitmassoffluidstoredisgreaterwithsupercritical
systemsthanthatencounteredwithlowpressuretwophasesystems.
When extended storage of cryogens is required to provide cooling at a specified low
temperature, new design problems appear. For example: during a constant pressure
withdrawalofasupercriticalfluidfromatank,thetemperatureofthefluidcontinually
rises due to he added input energy from the environment to the container. In order to
maintain a relatively constant temperature for cooling purposes, a decaying tank
pressure must be utilized. Other problems arise when using vapor cooled shields in
subcriticalsystems(seeClause6),sincetheventedfluidmaynotbeavailableforshield
cooling before being used for cooling the load. In such cases a secondary or guard
cryogenmaybeused(seeClause6.4.1).
3. Cryogenscanbestoredinthesolidphase.Coolingsystembasedonthesublimationofa
solid coolant show considerable promise since a number of problems associated with
eithersubcriticalorsupercriticalstorageareavoided.Solidcryogenrefrigeratorsconsist
inasolidifiedgas,aninsulatedcontainer,aventingpathtospace,andaconductionpath
fromthecoolanttothedevicebeingcooled.Theoperatingtemperatureobtainablewith
thesesystemsdependsuponthechoiceofcoolant,thepressuremaintainedinthesystem,
and the heat load. The advantages of the use of a solid cryogen over normal liquid
cryogenshavebeenmentionedintheGeneralIntroductionofClause4ofthisPart.

114
ECSSEHB3101Part14A
5December2011

6
VCS Dewars

6.1 General
ThisClauseconcernsDewarsholdingcryogensusedforcoolingsystemsinthetemperaturerange0
100K.
Theselowtemperaturelevelsaremainlyrequiredbyspaceborneelectronicsystemsoperatingunder
verylownoiseconditions.
Coolingbyliquidorsolidcryogensmaybebasedoneitherofthefollowingideas(seealsoclause5.3).
1. Thecryogenisstoredasasourceoflowtemperaturegases.Theusefulnessoftheseopen
cycle cooling systems, based on stored liquid cryogens, appears to be confined to cases
whererefrigerationisrequiredforrelativelyshortperiod(Breckenridge(1972)[29]).
2. Thecryogenisindirectcontactwiththecomponenttobecooled.Theheatistransferred
from the component through boiling or sublimation of the cryogen at the component
surface. The resulting vapors are, in most cases, ejected outside. These vapors could be
used to cool the insulation surrounding the cryogenic container, as in the case of the
VaporCooledShieldDewars(VCSDewars).Onlywhentheboilingpointofthecoolant
exceeds the environment temperature makes sense to condense the vapors and send
thembacktothecontainertorepeatthecycle.
Forspaceapplicationssolidcryogensofferthefollowingadvantagesover(normal)liquidcryogens:
1. Latentheatofsublimationisgreaterthanlatentheatofvaporization.
2. Liquidposeproblemsofphaseseparationunderreducedgravityconditions.
3. Soliddensityishigherthanliquiddensity.
AsolidcryogencoolerforanIRdetectorisshowninFigure61.

115
ECSSEHB3101Part14A
5December2011

Figure61:Schematicrepresentationofasolidgryogencooler.FromBreckenridge
(1972)[29].

Superfluid liquid helium, on the other hand, has a number of unusual properties which make it
exceptionallysuitabletoprovideverylowtemperatureenvironments.Itisasuperconductorofheat,it
formsathinandhighlyconductivefilminallsurfaces,andithasessentiallyzeroviscosityforcertain
types of flow. Because of these reasons, many VCS Dewars under development will use superfluid
heliumasthecoolingcryogen
The amount of cryogen required to perform a given mission, and thus the mass of the system,
dependsnotonlyonthedetectorcoolingloadandonthelifetime,butalsoontheparasiticheatleaks.
Extremelyefficientthermalinsulationsshouldbedevisedtokeeplowtheseheatleaks.
The VCS Dewars which are being introduced can be also used for long term storage of cryogenic
propellantsinspace,themaindifferencesbetweenbothapplicationsarethetemperaturelevelandthe
geometricscale.
ThebasicideaoftheseDewarsconsistsinusingthecoldnessoftheevaporatingcryogentocoolthe
insulation.Thisisachievedbymeansofanumberofhighlyconductivemetallicshieldsplacedwithin
the insulation perpendicularly to the temperature gradient and set in contact with venting ducts
conductingtheboiloffvapors.
The contact between the metallic shields and the venting ducts could be achieved by either of the
followingprocedures.
1. The shields are soldered around the circumference of metallic discs which have been
previously joined to the venting duct or penetration. This procedure will henceforth be
callednormalattachmentsincetheaxisofthepenetrationisroughlyperpendiculartothe
shield.
2. In the tangential attachment, the vaporventing duct, in the shape of a cooling coil, is
eithersolderedorcementedtothemetallicfoil.Normallytheentranceandtheexitends
ofthecoilareconnectedtotheneckofthecryostat.
Additional details concerning VCS systems are given in Clause 6.4 of this Part. For the present
purposesitissufficienttoenvisageaVCSDewarasitissketchedinFigure62.Thehighlyconductive
shieldsareplacedbetweenthelayersofamultilayerinsulationextendingalongtheentirecontainer.
No precise details concerning the thermal contact between shields and venting tube are required at
thismoment.

116
ECSSEHB3101Part14A
5December2011

Figure62:SketchofatypicalVCSDewar.FromNiendorf&Choksi(1967)[169].

TheadvantageoftheVCSsystemcanbeunderstoodfromFigure63,whichrepresentsthatpartofthe
insulation close to a venting duct in a normal attachment system. The relative amount and the
directionoftheheattransfertoorfromthesystemisindicatedbymeansofarrowsofvaryinglengths.

Figure63:HeattransfermechanismthroughanormalattachmentVCSDewar.
FromNiendorf&Choksi(1967)[169].

Itcanbeseenthattheheat,Q,totheoutershieldisequaltotheheat,Q1,transferredtotheventedgas
through the shield, plus Q2 the heat which continues through the insulation. This mechanism is
repeated at each conductive shield, reducing the heat which would be transferred through a
conventionalmultilayerinsulationsystem.

6.2 Theoretical analysis

6.2.1 Introduction
The following analysis is an extension of the work performed by Paivanas, Roberts & Wang (1965)
[177],andbyNiendorf&Choksi(1967)[169].Heretheeffectofpenetrationontheheatleaksistaken
intoaccountinafairlysimplifiedway.
AtheoreticalmodelwhichisintendedtodescribethebehaviorofaVCSshouldconsiderthefollowing
points:

117
ECSSEHB3101Part14A
5December2011
1. Thermal conductivities of insulation, venting ducts and supports as function of
temperature.
2. Finitenumberofmetallicshieldsattachedtotheventingduct.
3. Finitethermalconductivityalongthemetallicshields.
4. Thermalresistancebetweenthemetallicshieldsandventingduct.
5. Convective heat transfer between gas and venting duct, and between gas and cooled
supports,ifany.
6. degradationoftheinsulationbecauseofthepenetrations.
A theory based on such fine details, however, would greatly hinder the attainment of the insight
requiredtodealeasilywithVCSs.Thus,amodelbasedonthefollowingsimplifyingassumptionswill
beused:
1. Thermalconductivitiesofinsulations,ventingductsandsupportsareuniformandbased
onsomemeantemperature.
2. Thenumberofmetallicshieldsisinfinite.
3. Conductivityalongtheshieldsisinfinite,sothatthereisnotemperaturegradientalong
them.Asaconsequence,aonedimensionalmodelofheattransferthroughtheshieldcan
beapplied,inthismodelthetemperatureisonlyfunctionofthecoordinate,x,acrossthe
insulation.
4. Thereisnothermalcontactresistancebetweentheshieldsandtheventingduct.
5. Thermal conduction through the insulation and through supports occur separately but
simultaneously,withnointerchangeofheatbetweenthem.Thence,isnodegradationof
theinsulationbecauseofthepenetrations.
6. The heat transfer rate between vented vapors and duct walls is infinite, so that the gas
temperatureateachcrosssectionoftheductisthatcorrespondingtotheductwallatthe
samecrosssection.
7. Anareaweightedaverageoftheventingductandtheinsulationthermalconductivityis
representedbyasinglefunction,k(T).
8. Theinsulationiscooledbythevaporsflowingthroughtheventingduct,whereascooling
ofthesupportsiftheyarevaporcooledisprovidedbythevaporflowingthrougheach
ofthem.
9. Thespecificheatofthevaporisconstant.
The validity of the most significant of these simplifying assumptions will be discussed in further
clauses.

6.2.2 The idealized model


Theonedimensionalequationfortheheatbalanceinaninfinitelythinlayeroftheinsulation,asthat
sketchedinFigure64,maybewrittenasfollow:
d dT dT
kA mc p [61]
dx dx dx

withtheboundaryconditions:

118
ECSSEHB3101Part14A
5December2011

x 0 , T TC , x t , T TH [62]

Figure64:Insulationmodelgeometry.

Similar equations and boundary conditions are applicable to each one of the p supports; it is only
requiredtosubstituteksj,Asj,Tsj,msj,andtsj,(j=1,2,.....p)fork,A,T,mandt.
Inaddition,theboiloffrateisrelatedtotheheatinputtothecryogenbymeansoftheexpression:

p

m msj h fg Q1 Qm Qs

j 1 [63]
dT
p
dTs
Q1 kA k sj Asj
dx x 0 j 1 dx x 0

Eq.[61]withboundaryconditions[62]canbeintegrated,whenkisuniform,leadingtothefollowing
equation:

dT TH TC
kA mc p mc [64]
dx x 0 p
t
e kA 1

SubstitutingEq.[64],andsimilarexpressionsforeachsupport,intoEq.[63]oneobtains:


msj
1 0
p
mo S S
1 mS / mo 1 [65]
m e 1 j 1 m
m msj t sj kA
S
e mo mt k sj Asj 1

where 1 is the dimensionless detector heat load, m0 is the characteristic boiloff rate, and S the
sensibility of the cryogen. All these magnitudes have been defined in the List of Symbols. In
particular,m0istheboiloffrateofthesamecryogenstoredinavesselhavingthesamesizeasthat
being considered, but thermally insulated by means of a conventional MLI with the same k and no
penetrations(bystoreditismeantherethatnoelectronicdeviceistobecooled).Sistheratioofthe
maximumcoolingbyvaporconvectiontothatbyphasechange.

119
ECSSEHB3101Part14A
5December2011
Eq. [65] would give, through the ratio m/m0, a measure of the improvement achieved by vapor
cooling. For example, m/m0 < 1 for a VCS storage container, Whereas m/m0 = 1 for a conventionally
insulated storage container. When additional heat loads are taken into account, either because
penetrations or electronic devices, m/m0 can be larger than unity, as will be shown below. For each
value of these additional heat loads, however, m/m0 will be a monotonic function of S. The largest
value of m/m0 will be always that corresponding to S = 0 (no vapor cooling), and the smallest that
correspondingtoS.
Forslightlycooledsupports((msj/m)2<<1)Eq.[65]becomes:

mo S m t p k sj Asj p msj
1
m e
mS / mo
1
1 o
m kA j 1 t sj

j 1 m
0 [66]

FortheconventionalMLI(S=0)theratiom/m0becomes:

t p k sj Asj
m
1
kA j 1 t sj
1
p
[67]
mo S 0 msj
1
j 1 m

Forvaporcoolingwithacryogenofinfinitelylargesensibility:

t p k sj Asj
m
1
kA j 1 t sj
p
[68]
mo S msj
1
j 1 m

Values of the ratio m/m0 for different heat additions to the cryogen, other than those across the
insulation,havebeenplottedagainstthesensibility,S,inFigure65.ItisshowninthisFigurethatthe
largest gain for a given cryogen, i.e., the smallest value of m/m0, is achieved when the heat transfer
acrosstheinsulationisdominant,i.e.,whenthecryostatisexclusivelyusedforstoragepurposes.In
these cases, vapor cooling is clearly advisable for high sensibility cryogens such as hydrogen or
helium.

120
ECSSEHB3101Part14A
5December2011

Figure65:Ratiom/m0againstthecryogensensibility,S,fordifferentvaluesofthe
heatadditionstothecryogenotherthanthoseacrosstheinsulation.Nocooled
supports(msj=0).Calculatedbythecompiler.

Whenotherpathsofheatadditiontothecryogen(suchaspenetrationsorelectronicheatloads)are
present,theimprovementbecauseofvaporcoolingisnotsolarge.ItcanbededucedfromEqs.[67]
and [68] that the ratio of the smallest to the largest evaporation rates tends to unity when the heat
loads other than those across the insulation increase. The reason for this is clear: when the heat
transferthroughtheinsulationisnotdominantnosubstantialgainisachievedbyreducingit.
Table61givesvaluesofthelowerboundsofm/m0forseveraltypicalcryogens.Theadvantageofthe
VCSsystemisnoteworthy,atleastforstorageorshippingcryostats.
Values of the lower bounds of m/m0 for temperature dependent thermal conductivity, k, of the
insulationarealsogiveninTable61.Thesevalueshavebeenobtainedbytheprocedureexplainedin
Clause6.2.3.1.
ThethermophysicaldatawhichhavebeenusedforestimatingSandm/m0aregivenintheupperpart
ofTable61.AmoreextensivecollectionofdatacanbefoundinClause8.1ofthisPart.

121
ECSSEHB3101Part14A
5December2011

Table61:BenefitObtainedfromaVCSSysteminaStorageContainer
CRYOGEN Helium Hydrogen Neon Nitrogen Argon Oxygen Methane

NormalBoilingPoint[K] 4,2 20,39 27,2 77,4 87,4 90,1 111,6

NormalMeltingPoint[K] a 13,98 24,5 63,4 83,6 54,4 90,7

VaporSpecificHeat
5,23 13,88 1,03 1,04 0,52 0,91 2,09
cpx103[J.kg1.K1]

HeatofVaporization
20,52 448,2 89,09 199,3 161,7 213,1 508,2
hfgx103[J.kg1]

HeatofSublimation
508,2 105,6 224,7 185,6 226,8 568,7
hfgx103[J.kg1]

LIQUID TH[K]

300 64,148b 8,659 3,154 1,162 0,684 0,896 0,775

Sensibility,S 200 42,462b 5,562 1,998 0,640 0,362 0,469 0,364

150 31,618b 4,014 1,420 0,379 0,201 0,256 0,158

300 0,065 0,262 0,452 0,664 0,762 0,714 0,740

m/mo(k= k ) 200 0,089 0,338 0,550 0,773 0,853 0,820 0,853

150 0,110 0,402 0,622 0,848 0,911 0,890 0,928

300 0,030 0,182 0,365 0,614 0,726 0,673 0,708


m/mo
200 0,045 0,256 0,475 0,745 0,836 0,800 0,840
kdependsonT
150 0,059 0,322 0,560 0,833 0,904 0,882 0,925

SOLID TH[K]

300 7,812 2,687 1,095 0,606 0,985 0,770

Sensibility,S 200 5,081 1,712 0,632 0,326 0,584 0,402

150 3,715 1,224 0,401 0,186 0,384 0,218

300 0,279 0,486 0,675 0,782 0,696 0,742

m/mo(k= k ) 200 0,355 0,583 0,775 0,866 0,788 0,841

150 0,417 0,653 0,841 0,917 0,846 0,905

300 0,193 0,398 0,622 0,747 0,641 0,704


m/mo
200 0,267 0,508 0,742 0,849 0,753 0,823
kdependsonT
150 0,331 0,591 0,822 0,910 0,825 0,897

122
ECSSEHB3101Part14A
5December2011

a Heliumremainsfluidevenatabsolutezero,providedthatthepressuredoesnotexceedabout2,5x106Pa.
b AccordingtoPaivanasetal.(1965)[177],Heliumsensibilityhasbeenloweredby15%toallowforthe
significantfractionofvaporizedliquidwhichremainsinthetankassaturatedvapor.

6.2.3 Evaluation of the restrictions involved in the idealized


model
The simplifying assumption listed in Clause 6.2.1 will be critically appraised in the following
paragraphs.
Inmanycases,correctivefactors,whichenablethecalculationofthedeviationsfromtheidealmodel,
havebeentabulatedand/orgraphicallyplotted.Inothercasestheproblemismorecomplicatedand
referencestoexperimentsorcomputerizedanalysisbyothersaregiven.
Thecorrectivefactorswillbepresentedas:
i=mi/m,
wheremiisthevalueoftheevaporationmassflowrateincreasedbytheithadverseeffect,andmthe
valueobtainedundertheidealizedmodel.
Obviouslyiislargerthanoneinanycase.Theonlyexceptionappearsinthestudyoftheinfluenceof
thetemperaturedependentthermalconductivityoftheinsulation.
A method to account for the simultaneous departures of the adverse effects from the ideal level is
showninClause6.2.3.5.
Theinformationconcerningeachindividualadverseeffectisarrangedasfollows:First,theanalytical
background is briefly reviewed, as far as possible, then values of i are given; finally,
recommendationsaresuggestedconcerninghowthesystemshouldbedesignedorinstalledtohold
theseadverseaffectsintoanegligiblelevel.
AsummaryoftheseveralcorrectiveeffectsisshowninTable62.

Table62:RelaxationoftheRestrictionsInvolvedintheIdealizedModel
BasicAssumptionof MoreElaborate Corrective
Comments
theIdealModel Model Factor

Temperature Onedimensionalpurely k>1 Itisassumedinanycasethatheat


independentthermal conducivemodel. transferthroughtheinsulationis
conductivity. k=k1T. purelyconductive.Thevalidityof
thisassumptionisevaluatedin
Clause6.2.3.1.1.

Thenumberofvapor Onedimensionalpurely n>1 Optimumpositioningofthe


cooledshieldsis conducivemodelwitha shields,andmisplacingeffectsare
infinitelylarge. finitenumberofvapor alsoconsidered.
cooledshields. Afactornktakingintoaccount
bothnandkeffectsisalsogiven.

Conductancealongthe Axisymmetrical y>1 Threedimensionaleffect.


shieldisinfinitelylarge. configuration,centered Otherthreedimensionaleffects
inaventingduct.Purely arise,whicharecommontoany
conductivemodel.Finite MLIappliedaroundsmall

123
ECSSEHB3101Part14A
5December2011

BasicAssumptionof MoreElaborate Corrective


Comments
theIdealModel Model Factor
conductancealongthe containers.
shield.

Otherthree ValuesofkeffforMLIwrapped
dimensionaleffects. aroundatankofasolidcryogen
VCSDewarare1,3to4,1times
largerthanthosededucedfrom
flatplatecalorimetertechniques.
(Bell,Nast&Wedel(1977)[21]).

Thermalconductance Procedurestoincreasethis
betweenshieldsand contactthermalconductanceare
ventingductis widelyknown.
infinitelylarge.

Convectiveheat Realisticvaluesofthe Nu>1


transferfromventing Nusseltnumberforboth
tubewallstoflowing laminarandturbulent
vaporsisinfinitely flowthroughcylindrical
large. tubes.

Heattransferprocesses Resultsfromacomputerized
throughtheinsulation analysisforaMLIindicatethata
andthrough verysmalldegreeofcoupling
penetrationsdonot takesplace(Bell,Nast&Wedel
interact. (1977)[21]).

The list of simplifying assumptions used in the idealized model is larger than the list of correcting
factorsaccountingfortheinaccuraciesoftheseassumptions.Evenwhenweareunabletoestimatethe
correction,welistthesimplifyingassumptiontoemphasizetheneedforfurtherstudy.

6.2.3.1 Temperature dependence of thermal conductivity


Theproblemdefinedbythedifferentialequation[61]withboundaryconditions[62],concerningthe
heat transfer through the insulation devoided of penetrations, can be solved when the thermal
conductivity,k,isaknownfunctionoftemperature.Thedimensionlessboiloffrateis(Paivanasetal.
(1965)[177]):

m 1 k d
1

mo k 0 1 S
[69]

m ln 1 S
[610]
mo S

couldbeusedtocalculatethesensitivenessoftheresultswhichhavebeenobtainedtotheassumption
ofuniformthermalconductivitythroughouttheinsulation.

124
ECSSEHB3101Part14A
5December2011
6.2.3.1.1 Data on k()
Dataonthetemperaturedependenceofkcannotbeobtainedeasily.Althoughalargeamountoftests
have been performed to measure the effective thermal conductivity of an MLI as a function of the
warm boundary temperature, TH, or of a characteristic temperature defined somehow, it is stressed
that the values from these tests should not be considered as an average thermal conductivity of the
materials constituting the insulation, rather they give, in an expeditious way, the heat flux as a
function of temperature for a very complex transfer process involving conduction, radiation, and
(perhaps)convection.
If the above objection, which is applicable to any Fourier lawtype analysis of the heat transfer
through an MLI, is skipped, one can estimate k(). Representative examples giving the effective
thermalconductivityofseveralMLIsareshowninTable63.
It can de deduced from these examples that the conduction term dominates the heat transfer at the
lowtemperaturesinvolved.

125
ECSSEHB3101Part14A
5December2011

Table63:(Effective)ThermalConductivityofSeveralMLIsvs.Temperature.

SAMPLEDESCRIPTION:ShieldsofAluminium
foil.Spacersofglassfiberpaper.

TYPICALVALUES:N/t=2x103m1to4x103m1.
=88kg.m3.
keff=3,1x105W.m1.K1forTH=294K,TC=20K.

EMPIRICALCORRELATION:
keff=1,854x107T+2,775x1013T3.


REFERENCES:Sampledescriptionandtypical
valuesfromMatsch(1962)[143].Empirical
correlationafterPaivanasetal.(1965)[177].

SAMPLEDESCRIPTION:Shieldsofdouble
aluminizedMylar.Spacersofsilknet.

TYPICALVALUES:N/t=1,5x103m1.
keff=6,6x106W.m1.K1forTH=135K,TC=55K.

EMPIRICALCORRELATION:
keff=6,80x1013(N/t)1,56(TH+TC)/2+
5,40x1010(TH4,67TC4,67)/(THTC)N/t

REFERENCES:Sampledescriptionandmeasured
datafromKelleretal.(1974)[118].Empirical
correlationfromBelletal.(1977)[21].

SAMPLEDESCRIPTION:Shieldsofdouble
aluminizedMylar.SpacersofTissuglas.

TYPICALVALUES:N/t=4,3x103m1.
keff=1,3x105W.m1.K1forTH=296K,TC=54K.

EMPIRICALCORRELATION:
keff=4,65x1019(N/t)2,91(TH+TC)
3,22x1022(N/t)2,91(TH3TC3)/(THTC)+
8,01x1010(TH4,67TC4,67)/(THTC)N/t

REFERENCES:Sampledescriptionandmeasured
datafromKelleretal.(1974)[118].Empirical
correlationfromBelletal.(1977)[21].

126
ECSSEHB3101Part14A
5December2011
6.2.3.1.2 Analytical background
ThethreesetofdatapresentedinTable63indicatethatalinearkvs.Trelationshipcouldrepresent
fairlyaccuratelytheeffectivethermalconductivityofatypicalMLIattemperaturesbelow200K.In
addition,sinceconductioncontrolstheheattransferprocess,itmakessensetointroducetheconcept
ofinsulationlocalthermalconductivity.

Underthevalidityofthisassumption,Eq.[69]with k =>k1(TH+TC)/2yields:

m 2TC ln 1 S T TC 1 ln 1 S
2 H 1 [611]
mo TH TC S TH TC S S

DataonTable61labelledkdependsonThavebeencalculatedbymeansofEq.[611].
Dividing the value of m/m0 given by Eq. [611] by that deduced from Eq. [610] one deduces the
correctivefactor,k,ortheinfluenceofthetemperaturedependentthermalconductivity.

It is understood that the values of k for> both systems under comparison (that with temperature
independent k, and that with k linearly depending on T) should be identical, otherwise the
characteristicboiloffrates,m0,woulddiffer,andtheirratioshouldbetakenintoaccountcalculatek.
ThedatagiveninTable61indicatethatm/m0isfairlysensitivetotheuniformthermalconductivity
assumption,particularlywhenSislarge.
Thecorrectivefactor, k,isshowninFigure66asafunctionofthesensibility,S,ofthecryogen,for
severalvaluesoftheratioTC/TH.
Noticethat k,whichissmallerthanoneinthisparticularcase,doesnotdependonthespecificvalue
ofk1,assumingofcoursethat k is>thesameforbothsystemsundercomparison.
Inordertomagnifythecorrectiveeffect,1hasbeenplottedinlieuofkinFigure66.Thefactthat
k is smaller than one indicates that the variable conductivity insulation is more effective than that
withuniformconductivity,andthisissobecauseoftheeffectofthecoolerlayersoftheinsulation.

127
ECSSEHB3101Part14A
5December2011

Figure66:Correctivefactor,k,forthedependenceofinsulationthermal
conductivity,k,ontemperature,T,againstthesensibility,S,ofthecryogen,for
severalvaluesofthetemperatureratio,TC/TH.Alinearthermalconductivityvs.
temperaturedependencehasbeenassumed.Calculatedbythecompiler.

6.2.3.2 Finite number of cooled shields


Letusconsiderasystemwithafinitenumber,n,ofcooledshields,joinedtotheventingduct,within
an otherwise homogeneous insulation. The aim of this clause is threefold: 1) to optimize shield
positioning, once n is fixed, in order to minimize the heat transfer through the insulation; 2) to
calculate the corrective factor, n, as a function of cryogen properties, boundary temperatures, and
numberofshields,assumingthatthecooledshieldsareplacedintheiroptimumpositions,and3)to
evaluatehowsensitivetheheattransferthroughtheinsulationistoshieldpositioningotherthanthe
optimumone.

6.2.3.2.1 Analytical background


Theheatbalancefortheithcooledshieldcanbewritten,intermsoftheusualdimensionlessvariables,
asfollows(seeFigure67):

S i i 1 i i 1
mn
, i 1,2,, n [612]
mo

InEq.[612]iisthedimensionlessheattransferratethroughtheithinsulationlayer:

Dj
j , j 0,1,2,, n [613]
j
where

128
ECSSEHB3101Part14A
5December2011

k
j 1
Dj

j
k
d [614]

and
j j 1 j , o 0 , n1 1 [615]

Figure67:Insulationmodelwithfinitenumberofshields.

Obviously the sum of thicknesses between shields plus those between the bounding faces of the
insulationandtheneighbouringshieldswillequalthetotalinsulationthickness,t.
Thence
n

j 0
j 1
[616]

Theproblemisnowtominimizetheratiomn/m0,whichappearsinthenequations[612],withthen+2
auxiliary conditions [613] and [616]. Following an idea set forth by Bejan (1975), who dealt with a
similarproblem,themethodofLagrangemultiplierswillbeused.Tothisendwefirstintroduce2n+2
additionalunknowns,i,jand,andthenminimizeanewfunction,F,definedasfollows:

129
ECSSEHB3101Part14A
5December2011

n
m
F i n S i i 1 i i 1
i 1 mo
[617]
n Dj n
j j j 1
j 0 j j 0

with the auxiliary conditions [612], [613] and [616]. In other words, we are to solve the following
5n+4equations:F/i=0;F/j=0;F/j=0;Eq.[612];Eq.[613],andEq.[616]whicharen,n+1,
n+1,n,n+1and1equations,respectivelytodeterminethepropervaluesof:mn/m0,i,j,j,i,jand
. There are 5n+5 unknowns, but one among i, j or can be assumed to be equal to unity.
Eliminationofj,i,jandyields:

i i 1 Di
i
i 1 Di 1 i [618]


and
1 S i 1 i 1 Di
i
1 S i Di i [619]

where,accordingtoEq.[614],Di/i=k(i)/ k .>
1. Whentheinsulationthermalconductivityistemperatureindependent,Di=iqi,thence
Eq.[614]becomes:
i=i1.
Thisequationindicatesthattheminimumheatfluxthroughtheinsulationoccurswhen
the distances between shields, and between the bounding faces of the insulation and
neighbouringshields,areallequal.
CombinationofEqs.[612],[613]and[614]yields:

S
i i 1
m
i 1 i 1 n
mo n 1 [620]

whereastheexpressionrelatingtheboiloffratewiththeheatinputtothecryogen,

mn 1
1 , o 0
mo n 1 [621]

130
ECSSEHB3101Part14A
5December2011
providesuswiththedimensionlesstemperatureforthe1stcooledshield.FromEqs.[620]
and[621],withn=1,onecandeducethetemperaturejump,iqi,intermsofi,mn/m0,
Sandn.Additionoftheseveraltemperaturejumpsgives

1 S n1 1
1

n 1
mn
mo S [622]

The corrective term, n, which gives the influence of the finite number of shields
assumingtemperatureindependentthermalconductivitybecomes,afterrecallingEq.[6
10],:

1 S n1 1
1

n n n 1
m
m ln 1 S [623]

2. Whentheinsulationthermalconductivityisalinearfunctionoftemperature,k=k1T,Eqs.
[624]and[625],below,shouldsubstituteforEqs.[618]and[619].

i 2Ti
[624]
i 1 Ti Ti 1
and
TH TC S Ti TC Ti 21 Ti 2
TH TC S Ti 1 TC 2Ti Ti Ti 1
[625]

ThesetwoequationstogetherwithEq.[616]furnishtheoptimumshieldspacingandthe
correspondingtemperaturefield.
Thecorrectivefactor, nk,forthefinitenumberofshieldswhenkisalinearfunctionofT,
canbeexpressedas:

mnk 1 T12 TC2 S


nk
m 1 TH TC ln 1 S
2 2
[626]

whereT1and 1,whichcorrespondtothefirstcooledshield,canbededucedashasbeen
indicated.Theresultingexpressioniscomplicated,andonlythenumericalresultswillbe
given.

131
ECSSEHB3101Part14A
5December2011
6.2.3.2.2 Some numerical results
Theabovemathematicalbackgroundhasbeenappliedtothecomputationofthefollowingdata.
1. Foruniforminsulationthermalconductivity.
Table64givesnumericalvaluesofthecorrectivefactor, n,ascalculatedbyEq.[623],
fortypicalcryogens,whenTH=300K,200Kor150K.
Figure 68 presents similar data. Now n1 is plotted against the cryogen sensibility, S,
forvariousvaluesofn.
ItcanbededucedfrombothTable64andFigure68thatwhenthesensibility,S,ofthe
cryogen is large, as for liquid helium, a very reduced number, n, of cooled shields will
decreasesubstantiallytheboiloffrate.

132
ECSSEHB3101Part14A
5December2011

Table64:CorrectiveFactor,n,GivingtheInfluenceoftheNumber,n,of
ConductiveShieldsontheBoiloffRateofSeveralCryogens.(k= k ).

Liquid Solid Solid Solid Solid


n SolidOxygen
Helium Hydrogen Neon Nitrogen Argon
a

TH=300K

0 15,370 3,590 2,059 1,481 1,279 1,437


1 3,387 1,809 1,410 1,210 1,128 1,193
2 2,172 1,469 1,253 1,134 1,083 1,124
3 1,763 1,329 1,182 1,098 1,062 1,091
4 1,563 1,253 1,143 1,078 1,049 1,072
5 1,445 1,205 1,117 1,064 1,041 1,059
6 1,368 1,173 1,099 1,055 1,035 1,051
8 1,273 1,131 1,076 1,042 1,027 1,039
10 1,216 1,106 1,062 1,034 1,022 1,032

TH=200K

0 11,266 2,815 1,716 1,290 1,115 1,270


1 2,967 1,624 1,297 1,133 1,074 1,124
2 2,001 1,371 1,186 1,086 1,049 1,081
3 1,663 1,264 1,136 1,064 1,036 1,060
4 1,493 1,204 1,107 1,051 1,029 1,047
5 1,392 1,167 1,088 1,042 1,024 1,039
6 1,325 1,141 1,075 1,036 1,020 1,034
8 1,242 1,107 1,058 10,28 1,016 1,026
10 1,193 1,087 1,047 1,023 1,013 1,021

TH=150K

0 9,079 2,396 1,531 1,189 1,090 1,181


1 2,705 1,511 1,229 1,089 1,044 1,086
2 1,890 1,309 1,146 1,058 1,029 1,056
3 1,595 1,222 1,107 1,043 1,022 1,042
4 1,446 1,172 1,084 1,034 1,017 1,033
5 1,356 1,141 1,070 1,029 1,014 1,028
6 1,296 1,119 1,059 1,024 1,012 1,024
8 1,221 1,091 1,046 1,019 1,010 1,018
10 1,177 1,074 1,037 1,015 1,008 1,015
a Heliumsensibilityhasbeenloweredby15%toallowforthesignificantfractionofvaporizedliquidwhich
remainsinthetankassaturatedvapor.

133
ECSSEHB3101Part14A
5December2011

Figure68:Correctivefactor,n,accountingfortheinfluenceofthefinitenumber,
n,ofshields,vs.thesensibilitySofthecryogen,forseveralvaluesofn.Calculated
bythecompiler.

2. Data for systems whose thermal conductivity is a linear function of temperature are
showninTable65andTable68.
Table65givesthecooledshieldtemperatures,Ti,anddimensionlesspositions, i,which
minimize the boiloff rate for several cryogens; two to seven cooled shields are
considered.
The factor nk appears in Table 68 for the same cryogens, boundary temperatures, and
number of shields as those in Table 64. It is deduced from both tables that the gain
achievedbyanumber,n,ofoptimallyplacedcooledshieldsislargerwhenthethermal
conductivityvarieslinearlywithtemperature.

134
ECSSEHB3101Part14A
5December2011

Table65:CooledShieldTemperatures,Ti,andDimensionlessShieldPositions,i,
WhichMinimizetheBoiloffRateforSeveralCryogens.(k=k1T)

LIQUIDHELIUMa

SOLIDHYDROGEN
Table66
SOLIDNEON

SOLIDOXYGEN

SOLIDNITROGEN

SOLIDARGON
Table67
LIQUIDOXYGEN

SOLIDMETHANE
a Heliumsensibilityhasbeenloweredby15%toallowforthesignificantfractionofvaporizedliquidwhich
remainsinthetankassaturatedvapor.

Table66:CooledShieldTemperatures,Ti,andDimensionlessShieldPositions,
i,WhichMinimizetheBoiloffRateforSeveralCryogens.(k=k1T)
LIQUIDHELIUMa SOLID SOLIDNEON SOLIDOXYGEN
HYDROGEN

Ti[K] i b Ti[K] i Ti[K] i Ti[K] i

H 300 1 300 1 300 1 300 1


2 119,6 0,499 166,9 0,534 192,1 0,553 215,5 0,580
1 32,6 0,180 73 0,199 100,7 0,212 134,4 0,239
C 4,2 0 14 0 24,5 0 54,4 0

H 300 1 300 1 300 1 300 1


3 160,3 0,597 198,9 0,637 219 0,656 237 0,680
2 71,1 0,307 118,3 0,348 146,5 0,369 175,7 0,401
1 23,2 0,114 57,1 0,133 82 0,145 115,4 0,170
C 4,2 0 14 0 24,5 0 54,4 0

H 300 1 300 1 300 1 300 1


4 187,4 0,666 218,9 0,704 235,4 0,721 249,8 0,741
3 105,2 0,405 150,1 0,455 175,8 0,478 200,7 0,509
2 50,3 0,213 93,4 0,252 121,2 0,273 152,3 0,304
1 18,5 0,081 48,2 0,099 71,1 0,109 104 0,131
C 4,2 0 14 0 24,5 0 54,4 0

H 300 1 300 1 300 1 300 1


5 206,2 0,716 232,4 0,751 246,3 0,766 258,3 0,784
4 132,3 0,483 173,1 0,533 196 0,556 217,4 0,584

135
ECSSEHB3101Part14A
5December2011

LIQUIDHELIUMa SOLID SOLIDNEON SOLIDOXYGEN


HYDROGEN

Ti[K] i b Ti[K] i Ti[K] i Ti[K] i


3 77,1 0,298 121,8 0,348 148,9 0,372 176,9 0,404
2 39 0,159 78,3 0,196 105 0,214 136,8 0,243
1 15,7 0,062 42,5 0,078 63,8 0,088 96,3 0,107
C 4,2 0 14 0 24,5 0 54,4 0

H 300 0,753 300 1 300 1 300 1


6 219,9 0,544 242,2 0,785 254,1 0,799 264,4 0,814
5 153,6 0,371 190,2 0,593 210,6 0,615 229,3 0,640
4 100,8 0,232 143,9 0,424 169,4 0,449 194,6 0,479
3 60,5 0,126 103,3 0,279 130,5 0,302 160,1 0,333
2 32 0,050 68,2 0,158 93,7 0,175 125,6 0,202
1 13,9 0 38,6 0,064 58,6 0,073 90,7 0,090
C 4,2 14 0 24,5 0 54,4 0

H 300 1 300 1 300 1 300 1


7 230,2 0,782 249,5 0,811 259,9 0,823 268,9 0,838
6 170,5 0,593 203,3 0,639 221,6 0,660 238,2 0,683
5 120,8 0,431 161,4 0,485 185,1 0,510 207,8 0,539
4 80,7 0,296 123,8 0,349 150,3 0,374 177,6 0,405
3 49,8 0,187 90,3 0,231 117,1 0,253 147,5 0,283
2 27,3 0,103 61 0,132 85,3 0,148 117,3 0,173
1 12,6 0,041 35,6 0,054 54,7 0,062 86,5 0,078
C 4,2 0 14 0 24,5 0 54,4 0

aHeliumsensibilityhasbeenloweredby15%toallowforthesignificantfractionofvaporizedliquidwhich
remainsinthetankassaturatedvapor.
b Forthedefinitionofiseesketch.

Table67:CooledShieldTemperatures,Ti,andDimensionlessShieldPositions,i,
WhichMinimizetheBoiloffRateforSeveralCryogens.(k=k1T)

SOLIDNITROGEN SOLIDARGON LIQUIDOXYGEN SOLIDMETHANE

Ti[K] i b Ti[K] i Ti[K] i Ti[K] i

H 300 1 300 1 300 1 300 1


2 216,6 0,586 228 0,598 226,2 0,602 227,8 0,602
1 138,2 0,247 156,5 0,260 156,5 0,265 158,4 0,265
C 63,4 0 83,6 0 90,1 0 90,7 0

136
ECSSEHB3101Part14A
5December2011

SOLIDNITROGEN SOLIDARGON LIQUIDOXYGEN SOLIDMETHANE

Ti[K] i b Ti[K] i Ti[K] i Ti[K] i

H 300 1 300 1 300 1 300 1


3 237,6 0,684 246,3 0,694 244,6 0,697 246 0,697
2 177,7 0,408 192,9 0,422 191,4 0,426 193,4 0,426
1 120,1 0,177 139,1 0,188 140,1 0,192 141,9 0,192
C 63,4 0 83,6 0 90,1 0 90,7 0

H 300 1 300 1 300 1 300 1


4 250,2 0,745 257,2 0,754 255,7 0,755 256,9 0,756
3 202 0,514 214,6 0,528 212,8 0,531 214,6 0,532
2 155,1 0,311 171,9 0,324 171,1 0,329 173,1 0,329
1 101,9 0,137 128,6 0,147 130,3 0,151 132 0,151
C 63,4 0 83,6 0 90,1 0 90,7 0

H 300 1 300 1 300 1 300 1


5 258,6 0,786 264,5 0,794 263,1 0,795 264,1 0,795
4 218,2 0,589 229,1 0,601 227,2 0,604 228,8 0,604
3 178,8 0,410 193,6 0,424 192,1 0,428 194 0,428
2 140,2 0,250 157,9 0,263 157,7 0,267 159,6 0,267
1 101,9 0,112 121,5 0,120 123,6 0,124 125,3 0,124
C 63,4 0 83,6 0 90,1 0 90,7 0

H 300 1 300 1 300 1 300 1


6 264,6 0,816 269,6 0,823 268,4 0,823 269,3 0,824
5 229,9 0,644 239,3 0,655 237,5 0,657 239 0,658
4 195,9 0,485 209 0,499 207,2 0,502 209,1 0,502
3 162,5 0,340 178,6 0,353 177,4 0,357 179,4 0,357
2 129,5 0,209 147,8 0,221 148,1 0,225 150 0,225
1 96,7 0,095 116,3 0,102 119,1 0,105 120,5 0,105
C 63,4 0 83,6 0 90,1 0 90,7 0

H 300 1 300 1 300 1 300 1


7 269 0,839 273,5 0,845 272,4 0,845 273,1 0,846
6 238,6 0,686 247 0,697 245,3 0,698 246,6 0,699
5 208,8 0,544 220,5 0,557 218,6 0,559 220,4 0,560
4 179,4 0,411 194 0,425 192,4 0,428 194,4 0,429
3 150,4 0,289 167,2 0,302 166,5 0,306 168,5 0,307
2 121,6 0,179 140,2 0,190 141 0,194 142,8 0,194
1 92,7 0,082 112,4 0,088 115,4 0,091 116,9 0,091
C 63,4 0 83,6 0 90,1 0 90,7 0

b Forthedefinitionofiseesketch.

137
ECSSEHB3101Part14A
5December2011

Table68:CorrectiveFactor,nk,GivingtheInfluenceoftheNumber,n,of
ConductiveShieldsontheBoiloffRateofSeveralCryogens.(k=k1T).

Liquid Solid Solid Solid Solid


n SolidNeon
Heliuma Hydrogen Nitrogen Argon Oxygen

TH=300K

0 15,370 3,590 2,059 1,481 1,279 1,437


1 1,737 1,335 1,200 1,132 1,089 1,117
2 0,991 1,030 1,038 1,050 1,039 1,041
3 0,781 0,918 0,972 1,013 1,016 1,007
4 0,688 0,862 0,936 0,993 1,003 0,988
5 0,636 0,827 0,914 0,980 0,994 0,976
6 0,603 0,805 0,899 0,971 0,989 0,968
8 0,564 0,776 0,880 0,959 0,981 0,957
10 0,542 0,760 0,868 0,952 0,976 0,950

TH=200K

0 11,266 2,815 1,716 1,290 1,155 1,270


1 1,651 1,285 1,161 1,095 1,058 1,085
2 1,006 1,043 1,040 1,044 1,030 1,037
3 0,814 0,950 0,982 1,020 1,017 1,015
4 0,726 0,902 0,950 1,007 1,009 1,002
5 0,676 0,872 0,930 0,998 1,004 0,994
6 0,645 0,852 0,916 0,992 1,001 0,988
8 0,607 0,827 0,899 0,984 0,996 0,981
10 0,585 0,812 0,888 0,979 0,993 0,976

TH=150K

0 9,079 2,396 1,531 1,189 1,090 1,181


1 1,596 1,251 1,135 1,070 1,037 1,064
2 1,018 1,051 1,044 1,036 1,021 1,032
3 0,839 0,971 1,004 1,021 1,014 1,016
4 0,775 0,929 0,982 1,011 1,009 1,008
5 0,707 0,903 0,968 1,006 1,006 1,002
6 0,677 0,885 0,958 1,001 1,004 0,998
8 0,639 0,863 0,946 0,996 1,001 0,993
10 0,618 0,849 0,938 0,992 1,000 0,989

a Heliumsensibilityhasbeenloweredby15%toallowforthesignificantfractionofvaporizedliquidwhich
remainsinthetankassaturatedvapor.

138
ECSSEHB3101Part14A
5December2011

6.2.3.2.3 Shield position tolerance


Once the number, n, of shields is fixed, the sensitiveness of the heat transfer to shield misplacing is
considered.
1. In the case of uniform thermal conductivity, the expression relating the dimensionless
boiloffrate,mn/m0,toshieldpositioning,i,canbewrittenforn=2,asfollows:


1 1 21
1
m2
1 S1
22 2 1 31 mo 0
[627]
1 1 1 1 mo mo 1
S m2 m2

where subscripts 1 and 2 denote shield positioning, except for m2/m0 which is the
dimensionlessboiloffrateofthesystemwithtwocooledshields.
RepresentativeresultsaregiveninFigure69.Contoursofconstantvaluesoftheratioof
the flux to the uncooled shield heat flux are plotted as functions of the dimensionless
distances of both shields to the cold boundary of the insulation. Different cryogen
sensibilities have been considered. The numerical values labelling the various contours
indicate the corrective term n/nopt1. Obviously the resulting configuration is
symmetricalabouta45axis,sincetheshieldpositionsareinterchangeable,nevertheless
only contours where 1 < 2 have been represented. This mapping procedure has been
usedbyAtherton&Prentiss(1973)[12].

139
ECSSEHB3101Part14A
5December2011

Figure69:Contoursofconstantvaluesoftheratiooftheheatfluxthroughthe
VCSsystemtotheuncooledshieldheatflux,mappedasfunctionsofthe
dimensionlessdistances,1and2,ofthetwovaporcooledshieldstothecoldface
oftheinsulation,forseveralvaluesofthesensibility,S,ofthecryogen.Uniform
insulationthermalconductivity.Thenumericalvalueslabellingthecontours
correspondston/nopt1.Calculatedbythecompiler.

Itcanbededuced,fromthedistortionwhichthecontoursufferwhenSincreases,thatthe
sensitivenessoftheheatfluxtoshieldmisplacingincreaseswithS,buttheincreasedheat
fluxis,inanycase,small.
Although the calculation of the optimum position for any number of cooled shields
presents no insurmountable difficulties, the graphical representation of the results
becomesincreasinglydifficultwithincreasingnumberofshields.Figure610,borrowed
fromAtherton&Prentiss(1973)[12],presentsdataforthreevaporheliumcooledshields.
The minimum heat flux is obtained with 1 = 0,25, 2 = 0,50, 3 = 0,75. The mapped
contours give the displacement of the first shield, from its optimum position, which
increasestheheatfluxby10%asfunctionsofthepositions, 2and 3,oftheothertwo
shields. Again we note that the heat flux is remarkably insensitive to discrepancies in
shieldpositioning.Forexample,ifthesecondshieldiscorrectlyplacedat 2=0,5andthe

140
ECSSEHB3101Part14A
5December2011
thirdshieldmisplacedat 3=0,9,thefirstshieldcanstillbeplacedanywherewithin 1=
0,250,10withoutincreasingtheheatfluxmorethan10%fromtheminimumvalue.

Figure610:Contoursofdimensionlessdisplacementsofasingleshieldfromits
optimumposition(1=0,25)whichproducea10%increaseintheheatfluxthrough
athreeshieldsystem.Thecontoursaremappedasfunctionsoftheremainingtwo
shieldsdimensionlesspositions.Numericalvaluesareforheliumbetween4Kand
300K.FromAtherton&Prentiss(1973)[12].

2. Whenthethermalconductivityisalinearfunctionoftemperature,theexpressionrelating
the dimensionless boiloff rate to shield positioning becomes very involved. It is,
nevertheless,feasibletoplotFiguresequivalenttoFigure69.Thishasbeendoneandthe
resultsarepresentedinFigure611.ItcanbededucedfromthisFigurethatalthoughthe
tolerancetoshieldmisplacingislessthaninthecaseofuniformthermalconductivityitis
stillfairformanufacturingpurposes.

6.2.3.3 Finite convective heat transfer in the venting duct


Sofarithasbeenassumedthatthevaporbulktemperatureateachcrosssectionoftheventingductis
equal to the duct wall temperature at that cross section. This is tantamount to assume that the
convectiveheattransfercoefficientisinfinitelylarge.
Wewillconsidernowthecasewheretheconvectiveheattransfercoefficientisfinite,sothatthevapor
bulktemperature,Tb,willbedifferentfromtheventingtubewalltemperature,Tw.
Theadverseeffectsassociatedtothefiniteconvectiveheattransferinthevaporsflowingthroughthe
ventingductareparticularlycriticalwhentheshieldsarenormallyattachedtotheduct.Thencethe
contentofthisclauseconcerns,unlessotherwisestated,thenormalattachmentmode.
It is assumed throughout the clause that the duct temperature, Tw, is equal to the insulation
temperature at the corresponding value of x. It is required that: 1) the shield to duct thermal joint
conductance be infinitely large, and 2) some procedure be devised to avoid the longitudinal heat
transferalongthetubewall.Thisprecautionisnotrequiredinthecaseoftangentialattachment.

141
ECSSEHB3101Part14A
5December2011

Figure611:Contoursofconstantvaluesoftheratiooftheheatfluxthroughthe
VCSsystemtotheuncooledshieldheatflux,mappedasfunctionsofthe
dimensionlessdistances,1and2,ofthetwovaporcooledshieldstothecoldface
oftheinsulation,forseveralcryogensintypicalcases.Temperaturedependent
insulationthermalconductivity(k=k1T).Thenumericalvalueslabellingthe
contourscorrespondston/nopt1.Calculatedbythecompiler.

6.2.3.3.1 Analytical background


Theheatbalancethroughtheinsulationcanbewritten,intermsoftheusualdimensionlessvariables
andparameters,asfollows:

d k dT m Nu dTb
S
d k d mo

d [628]

T is used instead of the dimensionless function to introduce more easily the temperature
dependenceofthephysicalmagnitudesinvolved.

142
ECSSEHB3101Part14A
5December2011
Concerningthevaporflowingthroughtheventingduct:

dTb mo t 2 k b
Nu T Tb [629]
d m Nu S A k

NubeingtheNusseltnumber(seeECSSEHB3101Part13Clause6.2and6.3).
Theboundaryconditionsare:

0 , T Tb Tc , 1 , T TH [630]

theboiloffrateisgivenby:

m Nu
TH TC k dT
mo k d 0 [631]

Beforeintegratingtheabovesystem,letusexaminethedependenceontemperatureoftheparameters
involved.
1. In the case of laminar flow through a circular duct with fully developed velocity and
temperature profiles, the Nusselt number, Nu, becomes (Figure 61, ECSSEHB3101
Part13Clause6.3.1.1):

Nu=4,364forconstantheatfluxalongtheduct [632]

Nu=3,66forconstantwalltemperaturealongtheduct [633]

The first value applies to the normal attachment case, and the second to the tangential
attachment.
Forturbulentflow(DittusBoelterformula(seeECSSEHB3101Part13Clause9.2)):

Nu 0,023Re 2 Pr
0, 4
[634]

where the fluid properties are evaluated at the arithmetic mean temperature difference
(see ECSSEHB3101 Part 13 Clause 6.2). Eq. [634] is valid under the following
conditions:
1) Re > 104. For not too high values of the Prandtl number this bound can be lowered
downto4x103(ESDU68006(1968)[65]).
2)0,7<Pr<100.3)t/d60.

143
ECSSEHB3101Part14A
5December2011
It is deduced that Nu depends on temperature at most through the exit temperature
difference, and thence can be considered as a constant parameter when interesting the
abovesystemofequations.
2. The temperature dependence of the fluid physical properties are, for most vapors,
roughlyasfollows:kb ~ b ~T0,7,whereascpispracticallytemperatureindependent.From
thisfactitisdeducedthatSandm0aretemperatureindependent.misalsotemperature
independentbecausetherequirementofmasspreservation.
3. 3)Theinsulationthermalconductivityisassumedtobealinearfunctionoftemperature,
k=k1T.
Introducing the dimensionless temperature, = T/TH, and the new dimensionless
independentvariable =( k /k(T H))(m Nu/m 0) theabovesystem,afterintegrating
onceEq.[628],becomes:

d
1 C S b C [635]
d

2
d b mo 1 0,7
r b b [636]
d m Nu S

whereristhedimensionlessheattransfercoefficient,whichisdefinedas:

t 2 k TH k b TH
r Nu [637]
A k2

Theboundaryconditionsare:

0 , b C
1 C m Nu [638]
, 1
2 mo

Eqs.[635]and[636]withboundaryconditions[638]defineaneigenvalueproblem.The
third boundary condition can be only met when mNu/m takes a definite value which
dependsonS,randC.

6.2.3.3.2 Some numerical results


The above mentioned problem has been solved numerically bya shootingmethod (Nachtsheim &
Swigert(1965)[159]).
Numerical values of the corrective factor Nu = mNu/m have been plotted versus r, for different
cryogens in Figure 612 to Figure 614. The sole parameter which distinguishes these Figures from
each other is the warm temperature, TH. It can be deduced from this Figures that, unless r is
sufficientlylarge, Nuissomewhatlargerthanunity,whichindicatesthatsubstantialadditionalboil
offiscausedbyinadequateconvectiveheattransferintheventingduct.

144
ECSSEHB3101Part14A
5December2011

Figure612:FactorNu,accountingforfiniteconvectiveheattransferintheventing
duct,vs.coefficientr,forseveralcryogens.TH=300K.Calculatedbythecompiler.

EXPLANATION

Key Cryogen(liquid) S Nuforr=0

a Argon 0,684 1,38

b Oxygen 0,896 1,49

c Nitrogen 1,162 1,64

d Neon 3,154 2,77

e Hydrogen 8,659 5,59

f Helium 64,148 33,3

145
ECSSEHB3101Part14A
5December2011

Figure613:FactorNu,accountingforfiniteconvectiveheattransferintheventing
duct,vs.coefficientr,forseveralcryogens.TH=200K.Calculatedbythecompiler.

EXPLANATION

Key Cryogen(liquid) S Nuforr=0

a Argon 0,362 1,20

b Oxygen 0,469 1,25

c Nitrogen 0,640 1,34

d Neon 1,998 2,11

e Hydrogen 5,562 3,92

f Helium 42,462 22,7

146
ECSSEHB3101Part14A
5December2011

Figure614:FactorNu,accountingforfiniteconvectiveheattransferintheventing
duct,vs.coefficientr,forseveralcryogens.TH=150K.Calculatedbythecompiler.

EXPLANATION

Key Cryogen(liquid) S Nuforr=0

c Nitrogen 0,379 1,20

d Neon 1,420 1,79

e Hydrogen 4,014 3,11

f Helium 31,618 16,9

Tolookmorecloselyattheinfluenceofronthetemperatureofthevaporsemergingfromtheventing
duct,Figure615givesTb= bTHagaistTb= TH,forseveralvaluesofthedimensionlessheattransfer
coefficient,r.Thefluidisvaporheliumandthewarmfacetemperature,TH=300K.Itcanbeseenthat
the vapor exit temperature value of Tb corresponding to T = 300K decreases sharply when r is
decreased.

147
ECSSEHB3101Part14A
5December2011

Figure615:Heliumvaporbulktemperature,Tb,vs.insulationtemperature,T,for
differentvaluesofthedimensionlessheattransfercoefficient,r.TH=300K.
Calculatedbythecompiler.

Figure 616 gives the distribution of temperature across the insulation in the same cases as in the
previousFigure.TheintricatenatureoftheinfluenceofronT(),whichhasbeennotedbyTsao(1974)
[242],canbeexplainedasfollows:Whenrissmall,theheliumflowingthroughtheducthasalowand
practically constant temperature, as shown in Figure 615. Hence the thermal conductivity, kb, of
heliumisrelativelylow.Thissmallvalueofkbpreventssufficientheatfrombeingtransferredfromthe
insulationtothegasandkeepstheinsulationhotand,thus,itsthermalconductivitylarge.Thehighk
andthelowkbconditioninducesalargeheatfluxacrosstheinsulationandasmallheatfluxacrossthe
ductwall.Thiseffectstartsatthecoldendandextendstothewarmend.Asrincreasestheinsulation
temperaturedecreasesfirstandthenincreasesagain.Forlargevaluesofrthevaportemperatureand
theinsulationtemperaturearealmostequalandincreaseuniformlyacrosstheinsulation.

148
ECSSEHB3101Part14A
5December2011

Figure616:Temperature,T,acrosstheinsulationfordifferentvaluesofthe
dimensionlessheattransfercoefficientr.Heliumvaporcooling.TH=300K.
Calculatedbythecompiler.

6.2.3.3.3 Reduction of the adverse effects


Toachievelargervaluesofrthedesignerhasthefollowingchoices:
1. Increasing Nu by convective heat transfer enhancement in the duct. Augmentative
convection heat transfer devices are extensively discussed in ECSSEHB3101 Part 13
Clause5.Theapplicationofthesetechniquesheredoesnotlookverypromising.
2. Increasing the local valueof the insulation thickness, t, as sketched in Figure 617. This
local increase can be very effective, since t appears squared in Eq. [640]. It is also
convenientfromthemanufacturingpointofview.

Figure617:SketchofaVCSinsulationinthenearnessoftheventingduct.
Normalattachment.AfterPaivanasetal.(1965)[177].

3. Joiningtheventingduct,intheshapeofaspiral,totheinnersurfaceofatube,filledwith
insulating powder to which the conductive shields are attached. To minimize heat

149
ECSSEHB3101Part14A
5December2011
conductionalongtheenclosingtubethethreadofthespirallingventingductshouldbe
muchlargerthanitsdiameter.
4. Attaching the venting duct tangentially to the shield. This solution allows a drastic
increaseintheductlength.
In the tangential attachment system the venting duct is joined to several metallic shields. After
spirallingaroundoneshield,theductisattached,foranotherspiral,tothenextshield,andsoon.
Toestimatethelengthofventingductwhichisattachedtoeachmetallicshield,oneassumesthatthe
thermaljointconductancebetweentheshieldandtheductisverylarge,andthatthelongitudinalheat
conductionalongtheductisnegligible.Thence,thewalltemperatureofaportionoftheductwillbe
thatoftheshieldtowhichthisportionisattached.
The length of the part of the duct attached to a shield should afford the fluid bulk temperature to
increasefromsomeinitialvalue,Ti(thetemperatureofthenearestinnershield)toavalueclosetothe
ductwalltemperature,Tw.
Therequiredductlength,forthelaminarflowcase,canbefoundinJacob(1958)[101],Vol.1,p.461.
Fig.256.
TheturbulentcaseisconsideredinKays(1966)[117].Thetemperaturefieldisextressedasaninfinite
seriesexpansioninEigenfunctions.ThefiveleadingtermsintheseriesaretabulatedinTable127of
the mentioned source. The Reynolds numbers (and, thence, the mass flow rates) required for fully
developedturbulentflowrates,however,muchtoolargeforthepresentapplication.

6.2.3.4 Finite thermal conductivity of the cooled shields


The assumption of infinitely large thermal conductivity of the conductive shields is justified when
thick shields of highly conductive metals are used. The thickness of the shields. And thus their
thermalconductivityis,however,weightlimitedsothattheassumptioncouldbenolongervalidin
manyinstances.
Theabovementionedassumptionwillbeevaluatedintermsofasmallparameter,,definedas:
=(kx/ky)(R2/t2).
kx and ky, which are assumed to be constant, are the thermal conductivities across and along the
insulation,respectively,whereastandRarethecharacteristiclengths.
To keep the analysis simple enough we will consider the axisymmetrical configuration which is
sketchedinFigure618(b).

150
ECSSEHB3101Part14A
5December2011

Figure618:Sketchoftheinsulationandofthesimplifiedconfigurationsusedto
analyzetheinfluenceofthefinitethermalconductivityoftheshields.(a)
Insulation.(b)Simplifiedconfigurationinthephysicalcoordinatesx,y.(c)
Simplifiedconfigurationinthestretchedcoordinates,,.

6.2.3.4.1 Analytical background


The problem is mathematically defined, in terms of the dimensionless function, , and the
dimensionlessvariables,=x/t;=y/R,asfollows:
Differentialequation:

2 1
0
2
[639]

Boundaryconditions:
0 , 0, 0 [640]

1 , 1, 1 [641]

151
ECSSEHB3101Part14A
5December2011

, 1 2 m y ,
, S [642]
2 mo

,
1 , 0 [643]
1

wheremy,m0andSaredefinedasbefore.Inparticular,m0isgivenby:

mo

k xR 2 1 2 TH TC
[644]
h fg t

Theexpressionrelatingtheheatadditiontothecryogenboiloffratebecomes:


1
my 2
2
d [645]
mo 1 0

The right hand side in Eq. [645] is the dimensionless heat transfer rate across the cold face of the
insulation(=0).

6.2.3.4.2 Orders of magnitude of and


The Dewar sketched in Figure 619 is similar to that flown on the Nimbus F weather satellite
containing solid methane. This Dewar has been quoted repeatedly and is described with detail in
Clause6.4.3.1.

Figure619:SketchofatypicalspaceborneDewar.Allthedimensionsareinmm.

TheMLIisdoubledaluminizedMylarTissuglast=0,025mthick.AccordingtoNast,Barnes&Wedel
(1976)[161]therearetwoblanketswithasinglecooledshieldsandwichedbetweenthem.Thefactthat
intheNimbusFDewarthiscooledshieldisgroundedtoasecondtankcontainingsolidammoniaand
doesnotcompletelysurroundthemethanetankwillbedisregardedhere.
TheeffectivethermalconductivityoftheMLI,deducedfromBell,Nast&Wedel(1977)[21],Fig.2,is:
kx=3,3x105W.m1.K1;TH=296KandTC=50K.

152
ECSSEHB3101Part14A
5December2011
Neitheroftheabovementionedreferencesgivesthecharacteristicsofthecooledshield,thencewewill
assume that this shield iscopper, 3,8x105 m thick. For copper at150 K, which is the ammonia tank
temperature, k = 400 W.m1.K1 (see Figure 855 in this Part). Assuming that the lateral thermal
conductanceisexclusivelyduetothecoppershield,
ky=(0,4x103)(3,8x105)/(5x102)=0,3W.m1.K1.
Ontheotherhand,R,measuredasindicatedinFigure618(a),takesavalueR=0,76m.Thus,
=(kx/ky)(R/t)2=2,5x102.
Thisvalueof justifiesasmallperturbationapproachtotheproblemoflateralheattransfer, being
thesmallparameter.
Noticethat isalsosmall;accordingtoFigure619atypicalvalueis =0,08,andthissimplifiesthe
numericalcomputationsonwhichtheresultspresentedinClause6.2.3.4.4arebased.

6.2.3.4.3 Singular perturbation approach


Sincetheparameterissmall,weseekstraightforwardexpansionofthesolutionintheform:


, i i , [646]
i 0

Substituting the expansion [646] into Eq. [639] with boundary conditions [640], and equating
coefficients of equal powers of leads to the following set of differential equations and boundary
conditions(i=0,1,2,...):

o
0

[647]
2 i 1 1 i
0 for i 1
2

0 , i 0, 0 [648]

1 , o 1, 1 , i 1, 0 for i 1 [649]

i , 1 2 m y i 1 ,
, S [650]
2 mo

i ,
1 , 0 [651]
1

Thedifferentialequations[647]withboundaryconditions[651]suggestseekingforsolutionsofthe
form:

153
ECSSEHB3101Part14A
5December2011

o , Ao [652]

1 d 2 Ao
1 , A1 f1 [653]
2 d 2

1 d 2 A1 1 d 4 Ao
2 , A2 f f 2 [654]
2 d 2 4 d 4
1

wherefi()isthesolutionoftheordinarydifferentialequation

d df1 d df i 1
2 , 2f i for i 1
d d d d
[655]

withtheboundaryconditions
, f i 0 [656]

df i
1 , 0 [657]
d 1

InordertocalculateAi()weenforcethefulfillmentoftheboundaryconditions[650].thisyieldsthe
followingsetofordinarydifferentialequations(noticethat(dfi/d)=12):

d 2 Ao m y dAo
S 0 [658]
d 2
mo d

3
d 2 A1 m y dA1 1 m y
F1 o
dA
S S [659]
d 2
mo d 2 mo d

where
df 2

F1 d2
[660]
1

Thegeneralsolutionsofthedifferentialequation[658]are

154
ECSSEHB3101Part14A
5December2011

my

Ao M o N o e
S
mo [661]

my 3
1 m
A1 M 1 N1e S y F1 Ao M o
S
mo
[662]
2 mo

where Mi and Ni are yet unknown integration constants. In order to calculate these integration
constants, we take into account the boundary conditions [648] and [649]. The zero order solution
results to be that corresponding to an infinitely large thermal conductivity along the shields. The
boundary conditions [648] and [649] cannot be satisfied for any when i 1 because of the
functions which appear in the right hand side of Eqs. [653], [654],... . Thence, we will enforce the
fulfillmentoftheseboundaryconditionsfor=exclusively.Thisgives:

1
No my

[663]
S
e mo
1

3
1 m
N1 S y F1 N o 1 N o [664]
2 mo

andsimilarexpressionsforMithatwedonotreproduceheresincetheoverallheatfluxthrough =0
isofconcernwhereasthetemperaturedistributionisnot.
The mathematical problem defined by means of Eqs. [647] and [648] presents two noteworthy
characteristics:
1. One must proceed tofirst order to find thezero order solution,to second order tofind
the first order, and so on. This greatly complicates the algebra involved. The backward
influenceofthesuccessivedifferentialequations[647]istypicalofellipticalproblemsas
theoneunderconsideration.
2. The functions i(,), i 1, which are usually called outer functions (Nayfeh (1973)
[162]),donotfulfiltheboundaryconditionsatbothendfaces(=0; =1)exceptfor
=.Thereforetheresultingoutersolutionfor,althoughitapproximates(uptotheorder
considered)theexactsolutionas0awayfrombothendfaces,breaksdowninregions
ofnonuniformityclosetothesefaces.Theseregionsarecalledboundarylayersorinner
regions. Within them exact solution changes sharply with in order to retrieve the
boundaryconditions.
Todetermineaninnerexpansionvalidintheboundarylayercloseto =0,thelayerisstretchedbya
factor . The exponent is determined so that both terms in Eq. [639] become of the same order
when both and are of order unity. It is seen that this is achieved when = 1/2. The same
stretchingisapplicabletotheboundarylayercloseto =1.Thethicknessoftheeitherlayerbeingof
order1/2.

155
ECSSEHB3101Part14A
5December2011
Inordertosolvetheinnerproblemnearthecoldendface, =0,weintroducetheinnervariable =
1/2 (which is of order unity throughout the inner layer) and we try a composite expansion of the
form:
j 1

, Outer solution 2
j ,
j 0
inner solution
[665]

The Inner Solution will ensure the fulfillment of the boundary conditions at = 0 and will, in
addition, fade out for distances from the cold end face of order 1/2. The reason for the existence of
terms 1/2,3/2,...,canbemadeapparentwhentheboundarycondition[642]isexpressedintermsof
the inner variables ,. Finally, no zero order terms appear in the solution, since the boundary
conditionat=0issatisfied,tozeroorder,bytheoutersolution.
Theinnerfunctions j(,)willbedefinedbythefollowingsequenceofdifferentialequationsand
boundaryconditions(j=0,1,2,...):

2 j 1 j
0

[666]
2

0 , j 0, i 0, for j 2i 1

j 0, 0
[667]
for j 2i 2 , i 1

, i , 0 [668]

j , 1 2 m y j 1 ,
, S [669]
2 mo

j ,
1 , 0 [670]
1

0(,),whichsatisfieshomogeneousboundaryconditions,vanishidentically.Theproblemfor 1(,)
thenbecomes:

2 1 1 1
0
2 [671]

0 , 1 0, 1 0, [672]

156
ECSSEHB3101Part14A
5December2011

, 1 , 0 [673]

1 ,
, 0 [674]

1 ,
1 , 0 [675]
1

Notice that 1(,) does not contribute to the overall heat flux through = 0, it merely readjusts the
temperature field. This is so since, according to the boundary conditions [673] to [675], no heat is
transferredthroughthreeoftheboundaries,thencetheoverallheatflowthroughthefourthboundary
willbezero.Nevertheless,1iscalculatedinordertoproceedtohigherorderterms.
Solutionstoproblemsofheatflowinsemiinfinitehollowcircularcylinders,whichareverysimilarto
that defined by means of Eq. [671] with boundary conditions [672], are widely known (see f.e.,
Carslaw(1921)[38]pp.127129orBudaketal.(1964)[33]pp.468469).Followingtheusualpractice,
wereachtheexpressionof1(,):

0, C d
1 o n

1 , 2 e n
Co n
Co2 n 2 Co2 n
[676]
n 1

whereC0(n)=Y1(n)J0(n)J1(n)Y0(n),J(n)andY(n)aretheBesselfunctionsoffirstand
second kind respectively (Abramowitz & Stegun (1965) [1]), and n are the roots of the equation
Y1(n)J1(n)J1(n)Y1(n)=0.
Theaboveexpressionfor1(,)becomes,aftersomealgebra:

21 2 d 2 Ao
1 ,
d 2 0

Co n
1
e n

4 n
[677]
n 1
C n
3 2
n o
2

Once1(,)hasbeencalculated,theproblemfor2(,)becomes:

2 2 1 2
0
2
[678]

157
ECSSEHB3101Part14A
5December2011

0 , 2 0, 0 [679]

, 2 , 0 [680]

,
2 ,

2 1 2
2

S
m y d 2 Ao
2 2 mo d 2 0
[681]

e n
4 n
n 1
n3Co2 n
2

2 ,
1 , 0 [682]
1

Noticeagainthatwedonotneedthefullsolutionofthisproblemtocalculatetheheattransferrate
acrossthecoldfaceoftheinsulation(=0),sincetheoverallheattransferratethroughthiscoldface
mustbalancethatthroughthecylinder=.

2 0 d 0 2 d [683]

wheretherighthandsidewillbeevaluatedbyresortingtoboundaryconditions[681].
UsingEqs.[661]and[663]forcalculatingd2A0/d2resultsin:

3
my
S
1
2
2 1 2
2
mo

1
d
2 2
4 n2
[684]
0
my
n 1
n4Co2 n
S
e mo
1 2

Bringing 0and 1oftheoutersolutionplusterm 2oftheinnersolutionintherighthandsideofEq.


[645],

158
ECSSEHB3101Part14A
5December2011

1 o 1
1
d d
0 0
my 2 Outer Solution


mo 1 2 3 / 2 2
1 [685]
d Higher order terms
0
Inner Solution
in Outer variables

Curiously enough the correction to the cryogen boiloff rate because of the first inner term, 2, is of
order.Thiscanbeseenbyrecallingthat=1/2and,thence


1 1
20 d 20 d
3/ 2

[686]

ComingbacktoEq.[685],wededucefromEqs.[652],[655]and[661],

o
1
2 my
1 2 0
d S No [687]
mo

1
1
2 m
2
d S y N 1 [688]
1 0 mo

whereN0isgiveninEq.[663]andN1inEq.[664].
SubstitutionofEqs.[684],[686],and[687]with[663],intoEq.[685]yields:

my

F1 m y e mo
3 S
4 1 2
S
2 mo S mmy 2 2
my
S
e mo 1
1 e o
1 0
3/ 2
[689]
S 1 my
2

2
S

n 1 4C 2 4 n mo
n o n
2

WecanfurtherassumethatintherighthandsideofEq.[689]Smy/m0isequivalenttoln(1+S).Thisis
thezeroorderapproximationanditssubstitutionintofirstordertermswillresultinanerroroforder
2.Then,neglectingothertermsoforder2,wefinallyarriveto(recallEq.[610]:

159
ECSSEHB3101Part14A
5December2011

F1
ln 1 S

4 1 2

2
2 2

y
my
1 1 S


0
3/ 2
[690]
m 4 2 1 S
ln 1 S
n 1 n4 Co2 n 2n

6.2.3.4.4 Some numerical results


Eq.[690]hasbeenusedforcalculatingthecorrectivefactor yasafunctionofthesmallparameter ,
forrepresentativevaluesofthesensibilitySandthedimensionlessouterradius,,oftheventingtube
towhichtheshieldsaresupposedtobeattached.
Thecoefficientofthefirstordertermintheexpansionofyinpowerseriesofhasbeenplottedvs.S,
fortwodifferentvaluesof,inFigure620.

Figure620:Coefficient,(y1)/,ofthefirstordercorrectionaccountingforthe
influenceofthefinitethermalconductivityoftheVCSsonthecryogenboiloff
rate,asafunctionofthecryogensensibility,S,fortwovaluesofthedimensionless
outerradiusoftheventingduct,.Theresultshavebeenobtainedbymeansofa
perturbationschemeinthesmallparameter,,whichmeasuredtheratioofnormal

160
ECSSEHB3101Part14A
5December2011

tolateralheatflux,andarevalidprovidedthattermsoforder3/2canbeneglected.
Calculatedbythecompiler.

ThecorrectionislargerthanthatfromPaivanasetal.(1965)[177],Fig.5,byafactoroftheorderof3,
butthisfactordecreaseswhenincreases.TheaxialsymmetryoftheconfigurationsketchedinFigure
618 allowed us to show forth the influence of the radius which does not appear in the two
dimensionalmodeloftheaboveauthors.

6.2.3.4.5 Reduction of the adverse effects


Itshouldbesaidthatthedesignerhasnotafairchancetoreduce inasystemofgivensize.Notice
thatitismoreeffectivetoreducethesize(R)thantoincreasethenumberofconductiveshields,orthe
thicknessofeachshield(ky).Thisseemstoindicatethatnormalattachmentoftheshields(seeClause
6.1)isbestsuitedtosmallsystems.
Whenthesystemsizeidfixed,twopossibilitiesareopen:
1. To increase the insulation thickness, t, keeping constant the number N/t, or radiation
shieldsperunitthicknessinordernottoincreasekxwhich,looselyspeaking,isinversely
proportionaltoN/t.
2. Toattachtheventingducttangentiallytotheshields.Inthiscasethecharacteristiclength
along the insulation, R, is not of the order of the perimeter of the system, it is, on the
contrary,oftheorderofthepitchofthespiralshapedventingduct.

6.2.3.5 Superposition of adverse effects


The preceding sections dealt with the individual corrections to a simplified model of heat transfer
through a VCS insulation. Now a prediction method is required to account for the simultaneous
departureofseveralofthedegradingeffectsfromtheideallevel.
Lackingofarigoroustheory,theassumptionwillbeintroducedthatthereisnointeractionbetween
theseveralnonidealphenomena.Thus,theboiloffratecouldbeobtainedsimplybymultiplyingthe
idealboiloffratebythecorrectivefactorsi.
Logicallythedesignerwouldtrytokeeptheadverseeffectsasreducedaspossible.Inthatcaseallthe
correctivefactorsaccountingfortheseadverseaffectswouldbeclosetounity.Undertheseconditions
alineartheorywhichisequivalenttotheassumptionoftheinteractionwouldbejustified.
There is one case in which the above assumption can be evaluated. Table 64 gives the corrective
factor, n,whichaccountsfortheinfluenceofthenumber,n,ofshieldsfortemperatureindependent
thermalconductivity,whereasTable68presentssimilardatawhenthethermalconductivitydepends
linearlyontemperature.theratio, nk/n,betweenbothcorrectivefactorsshouldbecloseto k,which
takesintoaccounttheinfluenceoftemperaturedependentthermalconductivityontheboiloffrate.
Tocheckthisconjecture,letusconsiderthevaluesinthefollowingtable,whichhavebeendeduced
fromTable64andTable68.ThedatacorrespondtoheliumwithTH=300K.

n 0 1 2 3 4 5 6 8 10

nk/n 1,0 0,5128 0,4563 0,4430 0,4402 0,4401 0,4408 0,4431 0,4457

161
ECSSEHB3101Part14A
5December2011
InTable61twovaluesofm/m0,aregivenforheliumwithTH=300K.Thefirstonecorrespondtok=
k and>thesecondonetokdependingontemperature.Theratioofthesecondvaluetothefirstoneis
k=0,46.
It is seen that the error incurred in assuming no interactive adverse effects is close to 4% except for
situationscorrespondingtolargedeviationsfromtheidealcase,asinthecasen=0oftheabovetable.

6.3 Supports

6.3.1 Introduction
Supportsarebasiccomponentsofcryogenictankagewhoseaimistransferringinertialloads,dueto
themassesofcryogenandinnervessel,totheoutershell.
Intheabsenceofsupports,theseinertialloadswouldactdirectlyontheinsulation,compressingitand
degradingitscharacteristics,nottomentiontheproblemswhichcouldariseasaresultofbreakageof
venttubesandfillinglines.
Fromthethermalcontrolpointofviewthesupportsposesevereproblemsbecause:
1. theirthermalconductivityislargerthanthatoftheinsulationand,
2. theperformanceofanMLIishighlysensitivetopenetrationsthroughthelayers.
Thesepenetrationslocallydegradetheinsulation,anditsnegativeeffectscanextendoversignificant
areas.
Sincetheheatleaksfromthecryogencouldbegreatlyaffectedbythesupports,considerablecareand
ingenuityshouldbeexercisedtodesignsupportsbeingbothstrong,towithstandthestressestowhich
thetankislikelytobesubjected,andgoodinsulators,tocontributelittletoheatinleaks.
Theconfigurationchosentosupportaninsulatedvesseldependsonthedesignerabilitytotacklethe
specificproblemsofaparticulardesign.Itsmanufacturewillinvolvecraftmanshipandtechnology.
Somecommonlyusedsupportsare:
1. Tensionrodsofhighstrengthmaterial,
2. tensileties,
3. saddlebandsofmetalorplastic,
4. plasticcompressionblocks,
5. multiplecontactsupports(stackeddiscs)and
6. compressiontubes.
InthecaseofspaceborneVCSDewarsthemostcommonlyusedare:
1. Tubesundertensionalflexuralloads,and
2. lowthermalconductancetensileties.
The use of plastic compression blocks has found limited application (see JPLCaltech VCS Dewar,
Clause6.4.3.2).
Stacked discs use the thermal joint conductance between two solids tore duce heat leaks through
them.Thisidea,whichwasfirstsuggestedbyMikesell&Scott(1956)[151]andextensivelydeveloped
by Probert (1967) [186], has been considered for supporting spaceborne cryogenicpropellant

162
ECSSEHB3101Part14A
5December2011
containers (Glaser et al. (1967) [75]). Its use for small containers, however, seems to be ruled out
becauseofthebulkyframeworkrequiredtoholdthediscsinplace.
VaporcoolingofthesupportshasbeenanalyzedbyTsao(1974)[242],andBejan(1975)[20].InTsaos
analysis helium vapors cool the entire length of the support, whereas in the system considered by
Bejan cooling is concentrated at some points along the support (cooling stations) leaving the in
between stretches free of coolant. The idea of combining support and venting tube has been often
suggested(see,forexample,Bennetetal.(1974)[23].
A helpful concept in the first choice of the support material is the Material Figure of Merit or
strengththermalconductivityratio.Compositedesignallowsthecombinationofseveralmaterialsin
themostfavorableway.
ThisClauseconcerningsupportsisarrangedinageneraltospecificorder.Figuresofmeritforseveral
materialsatcryogenictemperaturesaregiveninClause6.3.2.Then,dataonlowthermalfluxtubing
are presented in Clause 6.3.3. Finally, several representative supports are described in full detail.
TensileandflexuralsupportsaredealtwithinClause6.3.4andcompressivesupportsinClause6.3.5.
The information presented here could be complemented with that in Clause 6.5 (Existing Systems).
Duplicityinpresentationisavoidedinanycase.

6.3.2 Support materials


Supportmaterialshaveahighstructuralstrengthaswellaslowthermalconductivity.
Leaving aside the cooled supports, the heat transfer through the support is expressed as (Fouriers
law):
Qs=ksAs(Ts/x),
whereksisameanvalueforthesupportmaterialthermalconductivityinthetemperaturerangeTCto
TH.
Thesupportcrosssectionalareadependsonthemaximumload,P,tobesupported.Thus:
As=P/,
beingthedesignstressatthegiventemperature.
Combinationoftheaboveequationsyields,
Qs=(ks/)P(Ts/x).
P and TS/x do not depend on the material properties, thence the chosen material should have a
strengthconductivityratio, /ks,aslargeaspossibleinordertominimizetheheatleaksthroughthe
support.
Table69givestheratio /ksforseveralmaterialsusedintensilesupports,atcryogenictemperatures.
Itshouldbeemphasizedthat/ksisnottheonlyparameterwhichdefinesthemeritofagivensupport
material.Thematerialheatcapacityandthethermalinertiaoftheselecteddesignoughttobeaslow
aspossibletominimizecryogenboiloffduringthecooldown.

163
ECSSEHB3101Part14A
5December2011

Table69:FigureofMerit,/k,ofSeveralTensileSupportMaterialsatCryogenic
Temperaturesa

Ultimate Figureof
Thermal
Tensile YieldStress, Merit,
Material Conductivity,
Strength, 0,2x106[Pa] (/k)x106
k[W.m1.K1]
ultx106[Pa] [Pa.m.K.W1]

METALLIC

CopperAnnealed 82,7 475,0 0,104

Brass 412,0 100,0 1,65

Aluminium2024 379,0 81,3 2,80

Aluminium7075 482,0 86,5 3,34

StainlessSteel 251,0 10,2 14,8


(annealed)

Titanium 586,0 15,8 22,2

35Ni50Fe14Cr 702,0 12,6 22,2


Alloy

KMonel 689,0 17,1 24,2

HastelloyB 448,0 9,35 28,7

StainlessSteelDrawn 1030,0 8,8 70,2


Wire

TitaniumAlloy(4% 1000,0 5,87 102,0


Al4%Mn)

INORGANIC

68%Fiberglass32% 378,0 0,36 420,0


Resin

Fiberglass 1720,0 0,92 748,0

ORGANIC

NylonHighTenacity 606,0 0,245 247,0

Terylene(high 730,0 0,15 487,0


tenacityPolyester)

164
ECSSEHB3101Part14A
5December2011

Ultimate Figureof
Thermal
Tensile YieldStress, Merit,
Material Conductivity,
Strength, 0,2x106[Pa] (/k)x106
k[W.m1.K1]
ultx106[Pa] [Pa.m.K.W1]

Araldite6060 830,0 0,17 488,0

Fortisan36 1070,0 0,20 535,0


a DatainthisTablecorrespondtotemperaturecloseto100K.
NOTE Designstressistakenas60%ofyieldstressor40%ofultimatetensilestrengthformetallicand
inorganicmaterials,and10%ofultimatetensilestrengthfororganicmaterials.
From:Molnar(1971)[154].

6.3.3 Low thermal conductance tubing


Fiberglasstubingdesignhavebeendevelopedtoreduceboththeheatleaksthroughpenetrationsand
supports,andthesystemmass.
Dataonthethermalconductivityoffiberglassepoxystructuraltubesinthetemperaturerange4K
320 K have been reported by Foster, Naes & Barnes (1975) [70]. Some of these data are collected in
Clause8.4.
Composite tubing incorporatesa thin metal tubingliner overwrapped with fibrous materialusinga
suitablematrix.
Thesetubeswerefirstdevelopedforcryogenictransfersystems(Halletal.(1971)[80]).Themetallic
linerswereintendedtoprovideleakfreeservice,whereasthefiberglasscompositeprovidedstrength
andprotectionfromhandlingdamage.Becausetheoverwrapisagoodthermalinsulatorandtheliner
hasaverysmallwallcrosssectionalarea,longitudinalthermalconductanceisconsiderablyreduced
whencomparedtoanallmetalsupport.
Acompositesupportstrutwithtitaniumendfittingsandaconventionalallmetalsupportareshown
inFigure621.AccordingtoHall&Spond(1977)[81],thermalconductivitytestsperformedwithS
glassandboronepoxycompositesindicatethatcompositetanksupportsresultinareductioninheat
leakof60%overtheconventionalallmetalconfiguration.

165
ECSSEHB3101Part14A
5December2011

Figure621:Cryogenicsupportstubes.a)Composite.b)Allmetal.Alldimensions
areinmm.FromHall&Spond(1977)[81].

A thorough information on the manufacture and testing of thesetubes is given by Hall et al. (1971)
[80].AlesscompletebutmoreaccessiblereferenceisthatbyHall&Spond(1977)[81].Unfortunately
notensileand/orflexuraltestsofcompletetubeassembliesarereported,probablybecauseofthefact
that the main emphasis was placed onthe development of cryogenic transfer lines, nevertheless the
manufacturingdetailsandheattransferdatamaybeofinterest.
Twelvestylesoftubingwithdiameters12,7x103m,50,8x103mand127x103mwerefabricated,and
testedattemperaturesfrom20Kto294Kpressureupto20,7x10Pa.
Metal liners were fabricated by resistance welding, fusion welding or chemically milling a thicker
stockmaterial.
End fittings were joined to the liners by fusion or resistance welding. Explosive forming techniques
providecapabilityforjoiningaluminiumendfittingsdirectlytothinstainlesssteelliners.
Stainlesssteelisrecommendedforliners.321CRES(321StainlessSteel:0,18Cr,0,10Ni,0,04Ti,0,08
C, Fe Balance. CRES: Crucible Electric Steel Co., Homestead, Pa.) exhibits excellent welding
properties, although, after fusion welding, a relatively large heat affected zone appears in the weld
areaand,thence,thestructuralpropertiesarereduced.347CRES(347StainlessSteel:0,18Cr,0,10Ni,
0,08 Nb, 0,08 C, Fe Balance. CRES: Crucible Electric Steel Co., Homestead, Pa.) has been selected
becauseofitsexcellentweldingandformingcharacteristics.
Overwrappingwasperformedbyeithermachinewrappedfilamentwinding,overwrapbybraiding,
orhandtapewrapping.
Therecommendedreinforcementis20endS/HTSglassrovingwithanepoxycompatiblefinish.
Resinmatrixisselectedonthebasisofgoodmechanicalpropertiesatroomtemperatures,reasonably
good performance at cryogenic temperatures, shelf life in prepreg form (in excess of 2 weeks), and
adequatecuringtemperature.Theresin5868Rfulfilstheserequirements,itcuresat422K.
SomeresultsfromthethermaltestsaresummarizedinFigure622toFigure625.Anouterthermal
guardshroudwasplacedtominimizetheradicalheatflux,qm,duringtheperformanceofthetests.On
thecontrary,noinsulationwasplacedinternallyandthetubeswereevacuated.Data,notgivenhere,
indicatethattheeffectofchangesintheinternaltubeemittance, ,isnegligible.Onlydatafortubes
withdiameters12,7x103mand50,8x103mhavebeenreproducedhere.

166
ECSSEHB3101Part14A
5December2011
Although the tubes are too slender for supporting purposes some of the trends in Figure 622 to
Figure 625 are worth being mentioned. It can be seen, for example, that there is a limit in the
reductionofheatfluxtobegainedbyusingthefiberglassoverwraptechnique,andthatthereduction
inlinerthicknessdecreasestheheatfluxtentimesmoreeffectivelythanthereductionintheoverwrap
thickness.

Figure622:Heattransferrate,Qs,throughfiberglassoverwrappedandthrough
allstainlesssteelsupportsvs.supportlength,L,forseveralvaluesoflinerwall
thickness,tl,andoverwrapthickness,to.(a)Innerdiameterofthetube,d=
12,7x103m.(b)d=50,8x103m.FromHalletal.(1971)[80].

167
ECSSEHB3101Part14A
5December2011

Figure623:Heattransferrate,Qs,throughfiberglassoverwrappedsupportsvs.
linerwallthickness,tl,forseveralsupportlengths,Landoverwrapthickness,to=
0,762x103m.Hoopwrapping.(a)Innerdiameterofthetube,d=12,7x103m.(b)d=
50,8x103m.FromHalletal.(1971)[80].

168
ECSSEHB3101Part14A
5December2011

Figure624:Heattransferrate,Qs,throughfiberglassoverwrappedstainlesssteel
supportsvs.overwrapthickness,to,forseveralsupportslengths,L,andlinerwall
thicknesstl=0,51x103m.Hoopwrapping.(a)Innerdiameterofthetube,d=
12,7x103m.(b)d=50,8x103m.FromHalletal.(1971)[80].

169
ECSSEHB3101Part14A
5December2011

Figure625:Heattransferrate,Qs,throughfiberglassoverwrappedstainlesssteel
supportsvs.warmboundarytemperature,TH,forseveralvaluesofthecold
boundarytemperature,TC.Tubelength,L,linerwallthickness,tl,andoverwrap
thickness,to,asindicatedintheinsert.Hoopwrapping.(a)Innerdiameterofthe
tube,d=12,7x103m.(b)d=50,8x103m.FromHalletal.(1971)[80].

6.3.4 Tensile and flexural supports


Tensileflexuralsupportscouldbeofthefollowingtypes:
1. Tubes,mademainlyfromplasticcompositesand/oraluminiumortitanium.
2. Lowthermalconductanceties.
Metallicorplasticsaddlebands,wirecablesorchainsarealsousedfortensileapplications.
Support length should be as large as possible in order to minimize the heat flux through it. If the
available space between the inner and outer vessel is not large enough to accommodate a long
suspensionmember,standoffsareaddedtotheouterand/orinnervesselsasshowninFigure626(b).
Furtherresistancetoheattransfercanbeachievedbyusingspringsorwashersateitherendoratboth
endsofthesupports.

170
ECSSEHB3101Part14A
5December2011

Figure626:Typicalsupportingmethods.Noticehowtherodsshownin(a)are
crossedtominimizetheeffectofthermalcontractionandtoincreasethelengthof
theheatflowpath.In(b),longsuspensionrodsareaccommodatedinstandoffs.
FromBarron(1966)[18].

Figure627showsacloseupofasupportforaDewaroftheliquidheliumcooledIRtelescopebeing
developedbyGermanSpaceAgency(DFVLR)foruseonboardtheShuttle/Spacelab.Thesesupports
areconstructedoffiberglassreinforcedcompositewithcrosssectionsof0,87x104m2(thosesupports
inthebottomofthecontainer)and0,33x104m2(thoseinthetop),andaredesignedtowithstandthe
launch vibration of the vehicle, the minimum frequency being 30 Hz. The heat leaks through all
supportswouldbe150mW,butcanbeconsiderablyreducedbycontactingthesupportstotheVCS.

Figure627:Tensilesupportofaliquidheliumtank.FromLemke,Klipping&
Rmisch(1978)[131].

Inadditiontosupportingtheinnervesselandotheritems,thesupportcabealsousedtorestrainthe
cooled shields of the VCS system so that they do not contact each other bringing forth a thermal
bridge.ThiscanbeachievedbyspacingdiscslockedonthesupportassketchedinFigure628.These
discsprovidespacingbetweentheshieldswithoutimposingaxialloadswhichwouldtendtobuckle
theshields.

171
ECSSEHB3101Part14A
5December2011

Figure628:Spacingdiscs.FromBennettetal.(1974)[23].

Theuseofvariablegeometrysupportingsystemshasbeensetforthrecently.Urbach&Herring(1976)
[243] depict a liquid helium Dewar, developed by Ball Brothers Research Corp. (see Clause 6.4.3.2),
whosesupportingsystemconsistsofsixtitaniumsupportswhichareretractable,plussixfiberglass
supportsowhichtheVCSsareattached.Thetitaniumsupportsareusedduringthehighacceleration
phaseoflaunching.Onceinorbitthesesupportsareretractedintothecryogencontainerwhichrests
onthefiberglasssupports.

6.3.4.2 Support tubes


TheinnervesselofaVCSDewarcanbesupportedbyahollowcantileverbeam.Thisconfiguration
couldpresentseveraladvantages,inadditiontoitsstructuralappropriateness.
1. Thebeamcanbefashionedasanextensionoftheinnervesselnecktube,providingdirect
accesstothecontainerforcryogenfilling.
2. Vaporcoolingofthesupportcanbeeasilyachieved.
3. Ifasinglebeamcanwithstandthedesignloads,itpenetratesthroughtheinsulationonly
once,whereasasystemofmultiplerodswouldpenetraterepeatedly.
4. Finally,thebeamcouldserveasaheatexchangerbetweenthehighlyconductivemetallic
shields and the boiloff vapors. It is, however not always easy to profit from this
advantagewithoutincreasingtheheattransferalongthebeamwalls.
Figure629,whichhasbeenborrowedfromBennettetal.(1974)[23],showsahollowbeamdevisedto
supportaliquidheliumDewarwhichhasbeenstudiedbyInternationalResearch&DevelopmentCo.
Ltd.NewcastleuponTyne.

172
ECSSEHB3101Part14A
5December2011

Figure629:SupporttubeforaliquidheliumDewar.FromBennettetal.(1974)
[23].

Thesupportismadefrom304StainlessSteel.Itis0,3mlong.Itscrosssectionalareais2,09x104m2at
thewarmend,and0,88x104m2atthecoldend.
The supported mass is 390 kg. The support can withstand a steady acceleration of 10 g and 5 g
vibration,bothintheaxialdirection.
Theventingductisthermallyconnectedtotheinnersurfaceofthesupportbyeithersoftsolderingor
brazing.Thusthesupportservesasaheattransfermediumbetweencooledshieldsandventingduct.
Toinsurethattheventingductdoesnotcontributesignificantlytothethermalconductionalongthe
supporttubethethreadofthespirallingventingductshouldbeseveraltimesitsdiameter.
After the cryostat is filled with helium, the support tube is packed with metallized hollow
microspherestoprovideinsulationforboththeprelaunchconditionandthemission.
Theinnervesselisrestrainedfromlateralmovementbythreeradialfiberglassties,each0,5mlong
and0,165x104m2incrosssectionalarea.SpacingdiscsasthosesketchedinFigure628areprovided.
Thecryogentemperatureis2K.Theestimatedheatleakthroughthesupportisnotgiven.

6.3.4.3 Low thermal conductance ties


TwowaysofsupportingcryogeniccontainersbytensiletiesaresketchedinFigure630.Sixtension
members are required in both cases. In the system depicted in Figure 630(a) each set of three
structural members connects to a cantilever that penetrates through the insulation. In the Figure
630(b) each structural member reaches the inner vessel at a different location and, therefore,
penetratestheinsulationseparately.

173
ECSSEHB3101Part14A
5December2011

Figure630:Twowaysofsupportingcryogeniccontainersbymeansoftensileties.
AfterGlaseretal.(1967)[75].

TheliquidheliumcryostatshowninFigure631,beingdevelopedbyESAforuseonboardSpacelab,
hasatanksupportedby24fiberglassstruts.Eachstrutiscomposedofthreepartsconnectedtogether
atthelevelofeachshieldinordertofacilitatetheintegrationandtoachieveagoodthermalcontact.

Figure631:SketchoftheSuperfluidHeliumCryostatforSpaceUse(CRHESUS)
showingthetensiletiesusedforsupportingtheheliumtank.FromLizonTati&
Girard(1978)[134].

TheheatflowbetweenthemainelementsofthecryostatisshowninFigure632.

174
ECSSEHB3101Part14A
5December2011

Figure632:CRHESUSheatflowdiagram.FromLizonTati&Girard(1978)[134].

6.3.5 Compressive supports


There are two main philosophies for the design of compressive supports, namely: supports
manufactured using low thermal conductance materials, and supports based on the high thermal
contactresistancewhichcanbeachievedbetweencontactingmaterialsinvacuum.Onlythefirsttype
ofsupportwillbedescribedhere.

6.3.5.1 Low thermal conductance compressive supports


Mason (1972) [142] briefly describes two configurations that were considered for supporting liquid
helium cryostats. The first design is a titanium strut, and the second a fiberglass pad. The last one
turnsouttobemoreefficient.Eachoneofthesupports(titaniumrodorfiberglasspad)isthermally
connectedtoeachcooledshields,passingthroughit,sothattheinwardflowingheatwouldbelargely
interceptedandcarriedoffbytheoutflowingheliumgas(seeJPLCaltechVCSDewar,Clause6.4.3.2).
Another lowtemperature structural support, which is not particularly designed for spacecraft
although it purports interesting features, is described by Heim & Fast (1973) [85]. This support is
constructedusingthreetubesandfourflangesasitisshowninFigure633.Noticethatadecreasein
the length of the intermediate tube because of thermal contraction will increase the overall column
length.Selectingtubematerialssuchthatthethermalcontractionoftheintermediatetubebeequalto
thesumofthecontractionoftheinnerandtheoutertubes,theoverallcolumnheightwillnotchange
whenthecolumniscooleddowntooperatingconditions.

175
ECSSEHB3101Part14A
5December2011

Figure633:Compositecolumncompressivesupport.FromHeim&Fast(1973)
[85].

Thesupportswereconstructedusingaluminiumfortheintermediatetubeandfiberglassepoxyfor
theinnerandoutertubes.Bothfiberglassepoxytubeshavea1,6x103mwallthicknesssincethiswas
thethinnesttubewallmanufacturedusingnormalshoppractice.Awall3,2x103mthickwaschosen
forthealuminiumtubetoassuregoodtubetoflangewelds.Fouraluminiumflangesweremachined
with 6x103 m deep groves, to accept the epoxy fiberglass tubes, and two of these flanges were
weldedtothealuminiumtubeasshowninFigure633.Priortobondingwithepoxy,allaluminium
bondingsurfaceswerechemicallyetchedwithasodiumdichromatesulphuricacidsolution,whereas
theinnerandouterbondingsurfaceswerelightlysandedandwipedwithcleaningagent.Allflange
jointswerebondedtogetherwithalowtemperatureepoxyandcuredatroomtemperature.
Aftercuring,thetopandbottomflangesweremachinednormallytothecolumncenterline.Several
layersofsuperinsulationwereinsertedbetweenthetubestominimizeheattransfer.
Thespecimenwhichhasbeendescribedwastestedwiththefollowingresults:
Collapseload...........8,53x104N
Heatleak...............13,53x103W
Heightchange................0
ThesecompositecolumnsarebeingusedattheUSNationalAcceleratorLaboratory,Batavia,Illinois,
tosuperconductingmagnetcoils.TheVCSDewarflownontheNimbusFweathersatellitehadsimilar
supports.

6.4 Phase separators

6.4.1 Introduction
This Clausedeals with phase separators for dewars holding cryogens other than superfluid helium.
This means that separation between liquid and vapor cannot be based on the thermomechanical
effect (see Clause 7.1.1.1). The superfluid porous plug, based on this effect, will be introduced in
1

Clause7.4.

176
ECSSEHB3101Part14A
5December2011
Theultimateaimofanyphaseseparatoristoallowventingofvaporonlyforrelievingthepressure
increaseswhichresultfromtheheatinputtothestoredcryogen.Ventingofliquidwouldresultina
lossofcoolantandadecreaseofthecoolingeffectivenessofthesystem.
Ventingofvaporonly,torelievethecontainer,canbeachievedefficientlyundersteadyconditions,in
the terrestrial laboratory, when the necksupport of the dewar is placed vertically upwards or only
slightlytilted.Underreducedgravityconditions,onthecontrary,theliquidvaporinterfacecanshift
easilybytheactionofsmalldisturbingforcesandseepageofrefrigerantliquidwouldresult.
Phase separation is also a problem in liquid propellant tankage, but there the aim is twofold: 1)
Venting of vaporonly, as above, and 2) Feeding the engines with liquid only. Problems associated
withliquidretentiondevicesareoccasionallyconsideredinthisClauseinasmuchasliquidretention
ensuresventingofvapors,butthemass(below100kg),size(below1m)andothercharacteristicsof
thetanksindicatethattheyareusedforcryogeniccoolingpurposes.
Several (mostly desirable) features of separating systems for cryogeniccooling dewars are
summarizedinTable610.

Table610:MainFeaturesofSeparatingSystemsforVCSDewars

Aim Ventingofvaporsonly

Background 1.Workatcryogenictemperatures.
Requirements 2.workunderreducedgravity.
3.Sturdy.canwithstandstructuraleffectsofaccelerations.
4.Satisfactoryperformancecanbepromptlyrecuperatedafter
accelerations.
5.Nomovingparts.
6.Passive.Noextraenergyrequired.
7.Basedonavailabledesigndata.

Criticallity Smallullages.

Alternative 1.Singlebubble.
StartingFluid 2.Manybubbles(Lowqualityfluid).
Configuration 3.Manydroplets(Highqualityfluid).

Suitable 1.Bubblepositioner.
Devices 2.Lowqualityfluidseparator.
3.Highqualityfluidseparator.

Variousdevices,withtheiradvantagesandlimitations,areintroducedinTable611.Mostofthemdo
notcompletelyfulfilthebackgroundrequirementsinTable610.

177
ECSSEHB3101Part14A
5December2011

Table611:PhaseSeparatingSystems
Basic Concept Description Advantages Drawbacks References
Principle

Vehicle Settling Rocketandaventvalve. Simple,widelyused. Singlerestart,shortmissions


acceleration rocket only.

Centrifugal Mechanical Movingblades.Drive Highlyefficientseparation Motorpowerdissipatinginto Mitchelletal.


active stirrers unitinsidetank. forhighqualityfluids thebath.Gyroscopiceffecton (1967)[152],
(nearlyvapors). thespacecraft. Seidel(1982)
[213]

Driveunitoutsidetank. Asabove. Gyroscopiceffect.EMI. Seidel(1982)


Rotationtransferred Motorpowernot [213]
throughamagnetic dissipatingintothebath.
clutch.

Peripheral Outerfluidlayerset Minimizesthemassof Motorpowerdissipatinginto


fan intomotionbya movingparts. thebath.Gyroscopiceffect.
centrifugalfan.

Rotating Wholetankdrivenby Motorpowernot Increasedgyroscopiceffect.


tank anouterunit. dissipatingintothetank.In Rotatingsealingandbearing
steadystate,fluidrotates lubricationproblems.
withoutfriction.

Centrifugal Separator Fluidisforcedtomovein Separationitselfis Apumpisrequiredtodrain Seidel(1982)


passive nozzle. highlycurvedpaths,through passive. theliquid [213]
amultistagenozzle.Liquidis
drainedout.

Cyclon Fluidenterstangentiallya Asabove.Low Asabove.Onlyeffectivewith

178
ECSSEHB3101Part14A
5December2011

vortexcavity.Resultinggas complexity. highqualityfluids.


glowsviaaheatexchanger.
Liquidisdrainedout.

Localvortex Alocalvortexismechanically Favourableliquid Motorinsidethebath.Liquid


generatedatthetankcenter. stirringeffectonthe motioncouldaffectsatellite
Gasinthevortexcoreenters temperaturefield stabilization.
aheatexchangerthrougha
throttlingvalve.

Ultrasonic Droplet Ultrasonicstandingwavesin Simpledesign,low Inputpowerdissipatesintothe


arrester aresonatortubeprevent weight. bath.Dropletsizedependent
liquiddropletsfromentering performance.Lifetime
thetube. problems.

Atomizer Dropletvaporization Asabove. Inputpowerdissipatesintothe


ultrasonicallyenhanced. bath.Lifetimeproblems.

Capillary Total Liquidvaporinterfaceis Completelypassive. Smalltanksonly.Require Mitchelletal.


controlledbysuitablebaffles muchdevelopmentwork. (1967)[152],
orbundleoftubes.See Recoveryafteraccelerations Seidel(1982)
Clauses6.4.3to6.4.7 abovedesignvaluesis [213]
unknown.

Partial Apassagewithporouswalls Heatexchanger Operationisinterruptedwhen


drawstheliquidincontact operatesunder thepressuredifference(inside
withthewallbycapillary constantanddefined tooutsidepassage)exceedthe
action.Thisliquidispumped conditions,provided capillarypressure.
backintothetank.Thevapor thatthesystem Considerabledevelopment
entersasmallheatexchanger. operationisnot work.
interrupted.

Dielectro Total Anonuniformelectricfield Nomovingparts. Requireshighvoltages(~106


phoretic isproducedbyanarrayof V).Electricalinsulation
electrodes.Liquidmoves problems.Electrodesdonot

179
ECSSEHB3101Part14A
5December2011

towardregionsofhighfield withstandaccelerations.
strength.Ventvalveisplaced Recoveryafteraccelerations
intothelowfieldregion. abovedesignvaluesis
unknown.

Partial Fluidisintroducedintothe Asabove.Phasesare Asabove.Efficiencyis


separator.Liquidheld separatedinapartial unknown.
betweenelectrodesasabove. compartment.
Vaporisvented.

Diffusion Separating Amembrane,permeableto Nomovingparts.Low Membranesarefluidspecific. Seidel(1982)


membrane gas,closesthetankoutset. complexity.Separating Lifetimeproblems. [213],Henis&
membranesareused Tripody(1983)
intheindustry. [87]

Thermo Forced Thefluiddrawnfromthetankis Operationdoesnot Ifanelectricallydrivenfan Mitchelletal.


dynamic convection throttledtoalowerpressure dependonliquid isused,heatisdissipating (1967)[152],
andtemperatureandpassed vapordistributionin intothebath. Seidel(1982)
throughaheatexchangerwhere thetank.Proven [213]
itcompletelyvaporizescooling concept.
thebulkfluidwhichisforcedby
afanunitthroughtheotherside
oftheexchanger.Operationis
intermittent.

Pulsed Fluidisadmitted,bythrottling, Lowcomplexity. Vaporizationtimeistoo


constant toacontrolvolume.Thereitis largetoaccommodate
volume completelyvaporizedwhileit transients.Fluidspecific.
refrigeratesthebulk.

Pulsed Asabove,buttheheat Applicabletoany Lifetime,reliabilityand Seidel(1982)


constant exchangerisatubespirally fluid.Enhances wearingofthevalvesare [213],Mlleret
pressure wouldaroundthetankwall. insulation unknown.Valvesdissipate al.(1983)[157]

180
ECSSEHB3101Part14A
5December2011

Thetubeinletwillbeplaced effectiveness.The heatintothebath.


nearasupportofthetank.A concepthasbeen
tankexitvalveisopenfora tested.
presettimeintervalwhenthe
lowerspecifiedtankpressureis
exceeded,andthenkeptclosed
forapresettime.Thevalveis
permanentlyopeniftheupper
pressurelimitisreached.
Pressurewithinthetubeis
maintainedbyareliefvalve(the
phaseseparatorexitvalve(see
Figure634b)).

Pulsating(Flip Ahotwireanalyzesthefluid Lowcomplexityand Lifetime,reliabilityand Seidel(1982)


Flop) whichenterstheseparator, systemmass.Safe wearingofthevalveare [213]
activatingavalvewithtwoexit operationofthehot unknown.Valveand
ports,sothateitherliquidorgas werehasbeen hotwiredissipateheatinto
isthrottledinathrottling demonstrated. thebath.
cascade.Afterthrottlingthe
fluidentersaheatexchanger.

181
ECSSEHB3101Part14A
5December2011

6.4.2 Thermodynamic vent system


Thissystem,alsoknownasHeatExchangeorThermalConditioningSystemisthesimplestandmost
reliableamongthosebeingconsideredatpresent.TwodifferentversionsaresketchedinFigure634.

Figure634:Schematicofthermodynamicventsystem.a)Forcedconvection.From
Mitchelletal.(1967).b)Pulsedconstantpressure.FromMlleretal.(1983)[157].

The twophase fluid drawn from the container is throttled and passed through a heat exchanger to
vaporize any liquid still remaining in the vent stream, and to cool the bulk tank fluid circulating
throughtheothersideoftheexchanger.
After leaving the heat exchanger, the vent fluid can be either vented directly or further expanded
throughaturbinetosupplypowertoatankmixer.
Since the quality of the inlet stream could vary, a modulating valve is used for controlling the
throttlingprocess.Flowcontroldownstreamoftheheatexchangerismaintainedbyavalvesensing
containerpressure.
Themainadvantageofthesystemisthatitsoperationisindependentoftheliquidvapordistribution
inthecontainerordisturbingaccelerationsonthevehicle.
Bulk mixing is used for temperature uniformization. If the mixer is electrically driven there is an
increaseintheexternalheatinputtothecontainer.Thiscouldbeeliminatedbyuseofaturbinedrive.
Theheatexchangercanbeacoiledtube,plateandfin,oraporousplugclosetoaheatsource.Inthe
socalledinternalconfiguration,theheatexchangerisplacedwithinthecontainer,asinFigure634a.
The cryogenic fluid and vapor are the heatexchanger hotsidefluid. Operation of the system then
reduces the pressure within the container. Other locations of the heat exchanger, such as on the
insulation (Figure 634b), or on the container wall, will result in larger system masses but do not
require recirculation of the fluid for effective cooling. Nevertheless, recirculation is still useful in
containers holding cryogenic liquids for long times, to avoid temperature stratification (Moses &
Gluck(1973)[156])andtoprovideamoreuniformfluid.Capillarypumpinghasbeenconsideredfor
thesepurposes(Blatt&Aydelott(1978)[25]).
The performance ofan internal heat exchangershould not vary significantlywith gravity level. The
mixing process, on the contrary, strongly depends on this level. Aydelott (1976, 1979) [14] & [15]
considered jet stirring in partially filled containers under reduced gravity. The interaction of the jet
with the liquidvapor interface and the resulting bubbles, unsteadiness of the interface, etc... are

182
ECSSEHB3101Part14A
5December2011
complicated phenomena deserving careful watching. The nonsteady effects of superimposed
accelerationshavenotbeenconsideredyet.
A very simple thermodynamic phase separator which has been utilized in Dewars onboard the
CosmosseriesisshowninFigure635.Itconsistsofatubetwistedintoaspiral(theheatexchanger)
withathrottlingdeviceattheinputend.

Figure635:Thermodynamicphaseseparator.FromFradkov&Troitskii(1975)
[71].

Theinnerpressureofthecontainerisp1.Theexitpressurep2(p2<p1)iskeptconstantduringtheflight
bycontrollingthegasexhaustintospace.
Under the action of the pressure difference, p1p2, throttling of a small part of the cryogen from the
container takes place in the throttling device and the fluid temperature within the spiral tube is
lowered. As a result of the heat transfer from the bulk liquid, the liquid within the heat exchanger
evaporatesandonlyvaporreachestheexit.

6.4.3 Capillary barriers


A capillary barrier is a perforated plate which inhibits fluid motion through capillary action and
providesamechanismforviscousenergydissipation.

6.4.3.1 Capillary barriers under static conditions


Letusconsideracapillarybarrierinitiallyplacedonaliquidvaporinterface.
Whengasbeginstopassthroughanysmallopeningwhichisinitiallywet,theinterfaceanchoredto
theedgesoftheopeningbecomescurvedandapressuredifferential(thecapillaryorbubblepressure)
preventsthepassageofthegasintotheliquid.
Figure636illustratesthebehaviorofacapillarybarrierinthepresenceofawettingliquid.

183
ECSSEHB3101Part14A
5December2011

Figure636:Acapillarybarrierinstaticequilibrium.FromMcCarthy(1968)[144].

ThepressureatlevelAintheliquidis
pl=pvlgl,
whereasthepressureinthevapor,thedensityofwhichisnegligible,ispvatanylevel.
The shape of the interface in the level A opening must result in a capillary pressure balancing the
differencepvpl
(1/R1+1/R2)=gl,
where is the liquidvapor surface tension and R1, R2 the two principal radii of curvature of the
interface.
Foracircularopening,diameterd,anawettingliquid,thelargestpressuredifferenceappearswhen
theinterfacebecomeshemispherical,R1=R2=d/2.Thus,
4/dgl
andthecurvatureoftheinterfacedecreasesasldecreases,assketchedinFigure636.
Staticequilibrium,intermsofthestaticBondnumber(seeClause6.4.7.1),requires
Bo=gld/
It is assumed that the liquid wets the barrier (i.e., the liquidsolid contact angle, , is close to zero).
Thus, the barrier offers resistance to vapor only. Conversely, when approaches 180 there is no
resistancetopassageofvapor.
Fluids such as those used for cryogenic cooling and cryogenic propellants have, in general, zero
contactanglesonsteel,aluminium,titanium,andallothermetalsthathavebeeninvestigated,andon
ceramicororganicglassysolids(Neu&Good(1963)[166]).

6.4.3.2 Capillary barriers under dynamic conditions


Let a container with a capillarybarrier partition which suddenly suffers an angular disturbance
aroundanaxisnormaltotheplaneoftheFigure637.

184
ECSSEHB3101Part14A
5December2011

Figure637:Containerwithacapillarybarrierpartition.FromMcCarthy(1968)
[144].(a)Anangularaccelerationappearswhentheinterfaceisformedatthe
barrier.(b)Theconfigurationreachesasteadyangularvelocitybeforeinteraction
oftheinterfacewiththebarrier.SeeTable612forthedefinitionofthe
experimentalconditions.

Table612:ExperimentalConditionsforCapillaryBarrierStabilityStudies(Figure
637).

Parameters (a)Bond (b)WeberNumber Comments


Number ControlledMode
ControlledMode

Geometrical 0,40<l/D<0,85 h/D Noeffectofl/DandL/Dfoundin


Ratios case(a).Forcase(b),seeFigure640
1<L/D<2 L/D andFigure641.

d/D Nodeffectfoundincase(b)even
thoughdvariedbyafactorof3.

lo/D lo/Dsmallenoughforsmall
dissipationbutlargeenoughtoreach
the =>0state.

0,1<Op<0,46 Op Opistheratioofopentototalareaof
thebarrier.Akintoporosity,
seeClause7.4.3.

StaticBond gdD/=0 gD2/=0 Testswereconductedinadrop


Number towerandinaKC135parabolic
flightaircraft.

Rotational LdD/
Bond
Number

Rotational 2L2D/ 2D3/ Thechoiceofthesecondformofthe


Weber WebernumberisbasedonFigure

185
ECSSEHB3101Part14A
5December2011

Parameters (a)Bond (b)WeberNumber Comments


Number ControlledMode
ControlledMode

Number 641.

Reynolds LD/>1 LD/>1 Reynoldsnumberkeptlargeinthe


Number tests.

NOTE FromMcCarthy(1968)[144].

In the case of Figure 637a the critical disturbance corresponds to an acceleration parallel to the
barrier.ThisisthecounterpartofFigure636with L insteadofg.Theresultsofthetestsareshown
inFigure638.Itisseenthatthebarrierislessstableunderdynamicthanunderstaticconditions,>the
criticalBondnumbers being, respectively, less than 1and 4.See alsoFigure639, wherea Reynolds
numberinfluenceappears.

Figure638:Resultsofbarrierdynamicstabilitytests.Bondnumbercontrolled
mode.Testswereinsufficientfordeterminingtheeffectonbarrierstabilityofthe
variousdimensionlessparameters.FromMcCarthy(1968)[144].

186
ECSSEHB3101Part14A
5December2011

Figure639:Resultsofdynamicstabilitytestswithdifferentbarriers.Bond
numbercontrolledmode.Theacceleration,g,isparalleltothebarrier.FromFester
(1973)[67].AReynoldsnumberthroughtheholehasbeenplottedvs.thecritical
Bondnumber.

Normal(tobarrier)accelerations,whichproducetheinstabilityoftheinterface,areusuallynotcritical.
Whentheporesofthecapillarybarrierarecircularofdiameterd,thecriticalBondnumberhappensto
be
Bo=gd2/=3,36
when the contact angle is zero (Fester (1973) [67]). This has been verified by drop tower tests of
capillarybarrierswhichalsoshownthatthecriticalBoforasquareweavescreenisslightlylessthan
thatgivenabove.
Figure637bcorrespondstoacontainerwhichcouldreachasteadyangularvelocitybeforetheliquid
vapor interface approaches the barrier, thence the phenomena of interest occur while there is no
acceleration of the reference frame (other than gravitational). The controlling parameter is not the
Bondnumber(ratioofhydrostatictosurfacetensionforces)buttheWebernumberwhichistheratio
ofdynamicforcesinthefluidtosurfacetensionforces(seeClause6.4.7.1).
Results of several tests are giveninFigure640 where Wec is an analyticallyobtained criticalWeber
number(Figure641).

187
ECSSEHB3101Part14A
5December2011

Figure640:Resultsofbarrierdynamicstabilitytests.Webernumbercontrolled
mode.FromMcCarthy(1968)[144].TheWebernumberinabscissaeisnormalized
withananalyticalcriticalWebernumberWec,whichisgiveninFigure641below.

Figure641:CriticalWebernumber,Wec,asafunctionofgeometry,l/D,and
positionoftheaxisofrotation,L/D.Theseresultshavebeenobtainedbyuseofa
potential(incompressible,inviscid,irrotationalflow)theorywithOp1,although
assumingthatthebarrierinducesacapillarypressuredifference.FromGluck
(1970)[76].

ThedampingeffectofperforatedbarrierscanbeassessedintermsofaWebernumberdefinedas,
We=V2d/,
whereVistheliquidimpingementvelocity.Therelativemeritsofdifferenttypesofbarriersaregiven
inFigure642.

188
ECSSEHB3101Part14A
5December2011

Figure642:Dampingperformanceofselectedbarriers.FromFester(1973)[67].The
dampingcategoriesAtoGareassociatedtotheflowpatternsresultingafter
impingementoftheliquidwiththebarrier,fromorderly(A)toirregular(G).

Capillary barriers are hardly used alone as separating devices. They can, however, enhance the
performanceofotherdevicesasinthecaseofthecompartmentedtankinFigure643.

189
ECSSEHB3101Part14A
5December2011

Figure643:Compartmentedtankdevice.FromFester,Eberhardt&Tegart(1975)
[68].

Inthissystemascreenlinerconcentricwiththetankwallholdstheliquid(orpartofit).Thisliquid
canbeusedeitherforwallcoolingorforfeedingthethrusters.Retentionoftheliquidwithintheliner
depends on the balance of pressure forces (static and dynamic) and surface tension forces. The
additionofbarriers,asshowninFigure643,reducesthehydrostaticheadmakingiteasiertoretain
liquidinahighgenvironment.Obviously,theventingcompartmentshouldbedepletedpriortothe
othercompartments.

6.4.4 Porous media


Conceptswheretheliquidisconfinedtoaporousmediumcanbeappliedtoverysmalltanks,suchas
for microthrusters. Systems based on these concepts have been successfully tested in orbit, for
example:intheAmericansatelliteATSIV(DeBrocketal.(1971))andJapanesesatelliteMST4(Enya
etal.(1981)[64]).
Pointsofconcerninthedesignofsuchsystemsare:

6.4.4.1 Sustained liquid height


Inacapillarytube,theaxisofwhichisparalleltothedirectionofacceleration,themaximumliquid
heightsustainedundernomotionis(Figure644),

Figure644:Sustainedliquidheightinacapillarytube.

190
ECSSEHB3101Part14A
5December2011

l 4Lb cos / d
2
[691]

whereisthesolidliquidvaporcontactangleandLbthesocalledBondlength
Lb=[/(lv)g]1/2
which, for a given g, is a material property (see Clause 6.4.6.2). Usually l >v and the later is
neglected.TheBondlengthappearswheneverhydrostaticandcapillaryforcesbalanceeachother.
Iftheliquidwetsthesurfaceofthecapillarybecomesclosetozeroandlreachesitsmaximumvalue
forgivenLbandd.
Equation[691],above,canbeusedinordinaryfibrouslayers,whenchannelgeometryisnotuniform,
providedthatanequivalentdiameter,DE,isusedford.
DE=[/(1)]do,
beingthevolumeporosity(seeTable746,Clause7.4.3),anddothediameterofthefiber.
Experimentalresultsforseveralfibers(closeto0,94)areshowninFigure645.Liquidwasethanol.
Acapsulecontainingtheliquidandpackedfiberswascentrifuged.Sustainedliquidheightisdeduced
fromliquidmassinsidethecapsulebeforeandaftercentrifugation.Gravitylevelwasaround40go.A
smallamountofliquidwaslostbecauseevaporationintotheatmosphere.

Figure645:Sustainedethanolheight,l,vs.diameteroffiber,do.Gravitylevel40go.
FromEnya,Kisaragi,Ochiai,Sasao&Kuriki(1981)[64].

ThesolidlineinFigure645hasbeendeducedfromEq.[691]with =0andDE=10,2do.Thecorrect
expressionofDEwith=0,94shouldbeDE=15,7do.Thence,experimentalheightsare1,5timeshigher
than those predicted. Reasons for the deviation could be the highly distorted configuration of the
channelsanduncertaintiesinthevaluesofanddo.
Withstandingoftheaccelerationbythefibersisalsoimportant.CeramicfiberKaowool(seeECSSE
HB3101Part7,clause5)exhibitsgoodcharacteristicsbutsmallpiecesoffiberareentrainedintothe
vapor.

191
ECSSEHB3101Part14A
5December2011
Figure646givessimilarresultsforammonia.Theporousmediumwaspackingglasswool,do=106m
and0,95.Packinglengthwas20mmexceptforpointlabelledunderfilledNH3whichwas10mm.

Figure646:Sustainedliquidheight,l,vs.gravitylevel,g/go.Liquidsare:Ammonia
(circle),underfilledAmmonia(square),andethanol(triangle).Matrixisglass
wool,do=106m.SolidlineshavebeendeducedfromEq.[691]with=0andthe
quotedvaluesofd.FromEnya,Kisaragi,Ochiai,SasaoandKuriki(1981)[64].

Accelerationwaschangedfrom10goto150go.
The capsule was hermetically sealed and the volume of the spread liquid was measured through a
scaleintheglasscontainer.
Solidlines,whicharedrawnmerelyforreference,havebeendeducedfromEq.[691]with =0and
thequotedvaluesofd.
Sustained liquid height was twice as high as the theoretical value. This could be attributed to glass
woolgeometryortoaperforatedTeflonplateplacedtopreventammoniaspillage.

6.4.4.2 Boiling within the voids


Boilingwithinthevoidsofaporousmediumisunlikely,butshouldnotberuledoutcompletelyin
practicalinstances.ForareviewofboilingintubesseeCollier(1981)[46].
Boiling cannot occur until the wall temperature exceeds the liquid saturation temperature. The
difference between both temperatures is the degree of superheat. Obviously the local fluid bulk
temperature can be smaller than the saturation temperature, the difference between both is called
degreeofsupercooling.
As the heat flux is increased the first bubbles appear on cavities of the heated wall. If the lowest
temperature on the bubble surface is greater than the bubble equilibrium temperature, the bubbles
willgrow.ThiscriterionisduetoHsu(1962)andissketchedinFigure647.Thebubbleequilibrium
temperaturedependsonthesaturationtemperature,Tsat,saturationpressure,psat,heatofvaporization,
hfg,gasconstantofthevapor,R,andcapillarypressure,/r.

192
ECSSEHB3101Part14A
5December2011

Figure647:Criterionfortheonsetofnucleationinsubcooledboiling.After
Collier(1981)[46].

TheincipienceofnucleateboilingintubeshasbeenconsideredbyDavis&Anderson(1966)[52].They
givethecriticalradius,rc,ofanactivecavityfornucleateboilingas,

rc Bkl / q
1/ 2
[692]

with
B=[2Tsat(1/v1/l)]/hfg
Intheseequationsthesymbolshavetheusualmeanings.
Atlowvaluesofqandofthepressure(whichappearsimplicitlyinB)thecriticalradiiaremuchlarger
thantheradiiofthevoidsanditisdifficulttoreachboiling(Enyaetal.(1981)[64]).Nevertheless,Eq.
[692]isbased,amongothers,onthefollowingtwohypotheses:
1. Bubleradiusismuchsmallerthanthatofthetube,and
2. Thereisasufficientlywide(continuous)rangeofactivecavitysizesavailable.
Whenrcresulttobetoolargeanestimateofthelargestactivecavityontheheatingsurfaceshould
be made. In the case of water reasonable agreement with experimental data was found when a
maximumactivesizeof106mradiuswasused.

6.4.5 Baffled tanks


Several types of internal baffles can be used for ullage bubble positioning, liquid acquisition and
centerofmasscontrolofstoredliquidsunderreducedgravityconditions.Themainemphasisishere
placed in bubble positioning for liquidfree vapor venting, particularly when the ullage volume is
small,insphericaltanksofradiusR.

193
ECSSEHB3101Part14A
5December2011
6.4.5.1 Post
Acentralpost,seeinsertinFigure648,canbeusedtopositionabubblebetweenthetopofthepost
andthetopofatank.Thepostheight,l,requiredtopositionagivenullage,U,isdeducedfromFigure
648 (curve labelled d/R = 0). In this particular instance gravity is nil (spherical bubble), the tank is
spherical,andthediameterofthepostzero.Uisgivenby

U 1 1 / 2 R
3
[693]

Figure648:Postheight,l,requiredtopositionagivenullage,U,underreduced
gravity.SeeClause6.4.5.2forexplanationofcurvesd/R=/0,BandC.

Anullageof0,125(or12,5%)correspondstoasphericalbubbleofdiameterR.Whenthepostisvery
slender (d 0) the location of the bubble is indeterminate if the ullage is smaller than 0,125 (notice
sphereBintheinsertofFigure648).Thus,thepostlengthisusuallysmallerthanthetankradius.
AccordingtoamathematicalanalysisquotedbyTegart&Fester(1975)[232]:
1. Theaxialstabilityoftheconfigurationincreaseswhenthepostdiameterandthetaperto
therootincreases.
2. The interface contacting the top edge of the post is less stable than one intersecting the
postelsewhere.

194
ECSSEHB3101Part14A
5December2011
3. The configuration has little lateral stability. Any lateral disturbance produces an offset
ullagealthoughrecenteringoccurswhentheperturbationdisappears.
Thepostsaresimple,easytodesign,manufactureandinstall,butprovidepoorullagecontrol.

6.4.5.2 Standpipe
Thestandpipeisessentiallyalargediameter,hollowpost.Openingsareprovidedaroundthebaseso
itcanfillwithliquid.Figure649,fromPetrashetal.(1963)[184],showsaconfigurationthebehavior
of which under large ullage conditions has been tested in MA7 spacecraft. Height of standpipe is
determinedinamannersimilartothatusedforthepost,althoughtheresultsareslightlymodifiedby
the larger values of d. For instance, considering a thinwalled standpipe of diameter d, the ullage
fraction,U,sphereAinFigure650,is:

2

U 1 1 / 2 R d / 4 R / 1 1 / 2 R
3
[694]

Figure649:Experimentalglasstankwithastandpipe.FromPetrash,Nussle&
Otto(1963)[184].Allthedimensionsareinmm.

195
ECSSEHB3101Part14A
5December2011

Figure650:Minimumullagecenteringcapabilityofthestandpipe.

which reduces to Eq. [693] above when d << R. Curves of U for different values of d/R have been
showninFigure648.
Thestandpipeswallowsthebubble,sphereBinFigure650,when
Ud3/8R3.
ThecorrespondinglimitingcurvehasbeenlabelledcurveBinFigure648.
Centeringofthebubblecannotbeaccomplishedforullagefractionseitherlessthand3/8R3(sphereB
inFigure650)orlessthan(1/2d/4R)3(sphereCinFigure650).Thelastmentionedlimitationhas
beenlabelledCinFigure648.
Analysisoftheullagecenteringcapabilityofthestandpipeshowedthat(Tegart&Fester(1975)[232]):
1. Atlowullagevolumesthelateralstabilityofthestandpipeismuchlikethatofthepost.
2. Atlargesullagesmostoftheliquidbecomesorientedinsidethestandpipe,Figure651.

Figure651:Liquidacquisitionbythestandpipeforlargeullages.FromPetrash,
Nussle&Otto(1983)[184].

Thestandpipesaresimple,easytodesign,manufactureandinstall.Nevertheless,theirullagecontrol
isstillpoor.

196
ECSSEHB3101Part14A
5December2011
6.4.5.3 Post with fingers
A post with fingers, Figure 652, is slightly better than the post alone. Smaller bubbles (up to 0,05
ullage volumes) can be stabilized when the ullage is initially centered, but other stable bubble
locationsarepossible.Stabilityintheaxialdirectionisthesameasforthepost.

Figure652:Centralpostwiththin,offaxis,posts(fingers).FromTegartetal.
(1972)[233].

The post with fingers device presents minimum loading and handling problems, and slightly
improvedullagecontrol.

6.4.5.4 Vanes
Thisdeviceconsistsofseveralvanesradiatingoutwardfromasmallsupportingpost.
The height of the device depends on the ullage volume as in Figure 648, other variables being the
numberandprofileofthevanes.
The vane profile is so designed that the sole ullage equilibrium configuration is the centered one,
capillaryforcespushtheoffsetbubbletoitsequilibriumposition.
Aninnerlimitinthevaneprofileallowstheundistortedbubbletobetangenttothetankwallandjust
touch the vanes. In order to stabilize the bubble in the centered position the vane profile is kept
outsidetheinnerlimitexceptforthecenteredposition.
The relationship between the vane profile, ullage volume and number of vanes is deduced from
Figure653b
Rmin()=(Rr)cos(/2)[r2(Rr)2sin2(/2)]1/2
where:=(2/n)sin,nnumberofvanes,andr=RU1/3.

197
ECSSEHB3101Part14A
5December2011

Figure653:Criteriatodeducevaneprofilelimits.FromTegartetal.(1972)[233].

Theouterlimitiscontrolledbyaphenomenoncalledbubblebreakup,madeapparentindroptower
tests.Asaconsequenceofthisbreakuptheullagesplitsintoanumberofbubbleswhichfitwithinthe
vanes with no capillary forces acting on them. The experiments indicate that the maximum local
lengthofthevaneisgivenbythetangencycondition,Figure653c
Rmax()=Rcos(/2)/[1+sin(/2)],
whichresultstobeullageindependentsincer=Rmaxtan(/2).
InnerandouterlimitscalculatedintypicalcasesareshowninFigure654.

198
ECSSEHB3101Part14A
5December2011

Figure654:Limitingvaneprofiles,Rmin/RandRmax/Rforn=6,8and12vanes.
Rmin/RhasbeencalculatedforanullageU=0,05.Rmax/Risullageindependent.
AfterTegartetal.(1972)[233].

Inmostcasestheprofileofthevanegoesbetweenthesetwolimits.Nevertheless,insomeinstances
the outer limit is exceeded near the root of the central post, for example when operation at larger
ullagesisrequiredorwhenthepumpingcapabilityofthevanemustbeenhanced.
Therestoringforceofaneffectivevaneprofilecanbeestimatedasfollows:
1st.Thefinitenumberofvanes(6,8,12,...)issubstitutedbyaninfinitenumber(centralaxisymmetric
body)theradiusRoofwhich,measuredfromthetankcenter,increaseswithdisplacementangle,.
2nd. The geometry of the displaced and distorted bubble and the capillary forces acting on it are
simplifiedsothattheforcebecomesproportionaltothedifferenceincurvaturebetweenoppositeends
ofthebubble,asshowninFigure655.

199
ECSSEHB3101Part14A
5December2011

Figure655:Simplifiedbubblegeometry.Thebubbleisheldbytwocontiguous
vanesandshapesupasifitwereheldbytheeffectivevane.FromTegartetal.
(1972)[233].

Thecapillarypressuredifferenceacrossthebubbleis:

1 1 1 1
pb p a
rb rc ra rc [695]

thence,
1 1
K
rb ra [696]

resultstobeproportionaltotherestoringforce.
Thedefinitionofbubbleshapeiscrucial.UnfortunatelythisisnotdiscussedinthereportbyTegartet
al.(1975)[233].
When the ullage is not too large and R Ro() no too small, it can be assumed that the bubble is
stabilizedbetweentwoconcentricsphereswithradiiRoandRrespectively.Thus,thebubblebecomes
axisymmetric, Figure 656, and its volume can be easily computed. The relationship between the
ullage,U,andtheangle,a,whichcharacterizesthebubbleshape,becomes:

1 R a 1 Ro
3 3

U 1 o 1 cos 1
2 R 2 8 R

3 1 Ro / R a [697]
cos a sin
2 4 1 Ro / R 2

200
ECSSEHB3101Part14A
5December2011

Figure656:Theidealdistortedaxisymmetricalbubble.

Curvesofavs.Ro/Rforseveralvaluesoftheullage,U,areshowninFigure657.

201
ECSSEHB3101Part14A
5December2011

Figure657:Angleawhichmeasuresthedistortionofthebubblevs.ratio,Ro/R,of
innerbodyradiustotankradius.Calculatedbythecompiler.

Once Ro() is defined, and a calculated, the curvatures at both ends of the bubble (Figure 655) are
givenby(omittingthecontributionofrc)

1 2 1 2
,
ra a rb a
R Ro R Ro [698]
2 2

andKcanbecalculatedbymeansofEq.[696]above.
Three vane profiles from Tegart et al. (1972) [233] are shown in Figure 658. The dimensionless
restoringforceRKcorrespondingtoeachprofileisalsoshown.

202
ECSSEHB3101Part14A
5December2011

Figure658:Typicaleffectivevaneprofiles,Ro/R,anddimensionlessrestoring
force,RK,vs.displacementangle,.TheFigurehasbeenreplottedbythe
compilerafterarepresentationinpolarcoordinatesbyTegartetal.(1972)[233].

TheresultsinFigure659havebeencalculatedbyuseoftheverysimpletheorywhichhasbeenjust
presented.Theprofilesaregivenbytheequation,

Ro m
1 2 U 1
3

R k [699]

where k and m are two parameters, both close to unity, which individualize the members of the
family. Other profiles could be considered. The forces RK shown in the Figure seem to exhibit the
sametrendsandordersofmagnitudeasthoseinTegartetal.(1972)[233].

203
ECSSEHB3101Part14A
5December2011

Figure659:Typicaleffectivevaneprofiles,Ro/R,anddimensionlessrestoring
force,RK,vs.displacementangle,.Thevaneprofileshavebeencalculatedby
Eq.[699]withtheshownvaluesofkandm.ForceshavebeendeducedfromEqs.
[696]to[698].

3rd.Giventhenumber,n,ofvanes,theactualvaneprofileisdeducedbyintersectionoftheenvelope
ofthebubbleswithcouplesofplanescontainingtheverticalaxisofthetank,forminganangle2/n,
andcontactingsymmetricallyeitherendofthebubble.
Vanes are complex to design and difficult to manufacture and install, they are, however, highly
reliableandprovideagoodullagecontrol.

6.4.5.5 Standpipe with vanes


Thiscombineddeviceisbetterthaneachalone.Vanenumberandprofilearethesameasforthevane
andtheperformanceissimilar.

204
ECSSEHB3101Part14A
5December2011

6.4.6 Empirical data for design

6.4.6.1 Surface tension

6.4.6.2 Bond length


Bond number, Bo, measures the relative importance of gravitational and surface tension forces (see
Clause6.4.7.1).
Bo=gl/(/R)
where isthedensitydifference,ltheextensionoftheinterfaceinthedirectionofgravityaction(or
itsdistancefromtheequipotentialsurfaceonwhichthetwofluidshavethesamepressure), isthe
surfacetensionandRthesmallestradiusofcurvatureoftheinterface.
Usually,theBondnumberiswrittenas
Bo=Rl/(Lb)2
withLb=[/(g)]1/2thesocalledBondlength.WhenRandlarebothsmallerthantheBondlength,
gravityeffectscanbeneglectedandtheconfigurationatrestiscontrolledbysurfacetensionforces.
Thetermbondlengthappears,withacompletelydifferentmeaning,concerninginteratomicdistances
(Pauling&Pauling(1975)).AnyconfusionisavoidedifcapitalBisusedinBondwhenreferringtoLb.
Figure660toFigure662givetheBondlengthsvs.temperatureforthecryogenslistedinTable81,
Clause 8.1.1, at saturation and under normal gravity conditions. Sources of data for drawing the
FiguresareindicatedinTable613.TheParachorapproximation(Quayle(1953))hasbeenusedwhen
availabledataareincomplete(dottedlinesintheFigures).

205
ECSSEHB3101Part14A
5December2011

Figure660:Bondlength,Lb,asafunctionofT,forsaturatedArgon,Methane,
NitrogenandOxygen.

206
ECSSEHB3101Part14A
5December2011

Figure661:Bondlength,Lb,asafunctionofT,forsaturatedEthane,Carbon
DioxideandAmmonia.

207
ECSSEHB3101Part14A
5December2011

Figure662:Bondlength,Lb,asafunctionofT,forsaturatedHydrogen,Helium
andNeon.

208
ECSSEHB3101Part14A
5December2011

Table613:SourcesofDataforCalculatingBondLengths

lv Parachor
Reference Reference Reference
Temp.Range[K] Temp.Range[K]

Ar Vargaftik(1975)[253] Vargaftik(1975)[253] Fromandlvdatain


Argon 84K90K 84K150K therange84K90K

CH4 Vargaftik(1975)[253] Vargaftik(1975)[253] Quayle(1953)[188]


Methane 93K113K 91K190,55K

C2H6 Vargaftik(1975)[253] Vargaftik(1975)[253] Quayle(1953)[188]


Ethane 113K183K 150K305,5K

CO2Carbon Kutateladzeetal.(1966) Kutateladzeetal.(1966)[127] Quayle(1953)[188]


Dioxide [127] Vargaftik(1975)[253]
223K293K 216,55K304,19K

H2Hydrogen Vargaftik(1975)[253] Vargaftik(1975)[253] Fromandlvdatain


Normal 15K32,77K 14K33,23K therange15K25K

H2Hydrogen Vargaftik(1975)[253] Vargaftik(1975)[253] Fromandlvdatain


Para 14K32,98K 14K32,98K therange19K25K

He Johnson(1961)[109] Vargaftik(1975)[253] Valuesofvnotavailable


Helium 2,5K5,2K 2,2K5,18K above5,18K

N2 Vargaftik(1975)[253] Vargaftik(1975)[253] Fromandlvdatain


Nitrogen 68K90K 63,15K126,25K therange70K90K

NH3 ECSSEHB3101Part9 ECSSEHB3101Part9 N.A.


Ammonia Clause6.3 Clause6.3
195,3K404K 195,3404K

Ne Vargaftik(1975)[253] Vargaftik(1975)[253] Fromandlvdatain


Neon 24K28K 25K44,4K therange25K28K

O2 Weast(1976)[260] Vargaftik(1975)[253] Fromandlvdatain


Oxygen 70K90K 54,35K154,77K therange70K90K

209
ECSSEHB3101Part14A
5December2011

6.4.6.3 Wetting
Wettingofthesolidbytheliquidiscrucialtotheperformanceofanyliquidretentionsystem.
Wettabilityismeasuredbytheliquidsolidcontactangle.Whenthecontactangle,,isclosetozero,it
is said that the liquid wets the solid. Unfortunately measurements of the apparent contact angle
hardlyyieldreproduciblevaluesduetomacroscopicdisturbingeffects.
The contact angle depends on the liquidvapor surface tension, and on the characteristics and
cleanlinessofthesurface.
Figure 663 indicates that an almost linear relationship exists between cos and for homologous
seriesofliquidsonthesamesolid.Thecriticalsurfacetensionisthatcorrespondingtocos=1.Ifthe
surfacetensionislessthancritical,theliquidspreadsonthesolidsurface.

Figure663:Relationbetweencontactangle,,andsurfacetension,,forseveral
liquidsonthequotedsurfaces.

Explanation

Key Surface Comments References

Polytetrafluorethylene(TFE) WideVarietyof Fox&Zisman(1950),


liquids. Ellison&Zisman(1954)
T=293K

210
ECSSEHB3101Part14A
5December2011

Key Surface Comments References

Polyvinylchloride(PVC) T=293K Ellison&Zisman(1954)



Polyethylene(PE)

Polyvinylfluoride(PVF,Tedlar)

Sapphire T=293K Fox,Hare&Zisman


(1955)
Marked SS( SSwithfattyacid
line monolayer)
Brass

Quartz,Gold,Silver Enyaetal.(1982)[64]

Lowfreeenergysurfaceshavesmallcriticalsurfacetensions,whereashighfreeenergysurfaces(like
mostmetals)havelargecriticalsurfacetensions,butsincetheyhavelargeadsorptionforces,theyare
pronetocontamination.Theresultingfilmcoatedsurfaceisalowfreeenergysurface.
Normally,surfacetensiondecreaseswithincreasingtemperaturesothatliquidswhichdonotspread
atlowertemperaturescouldspreadonthesamesurfaceathighertemperatures.

6.4.6.4 Compatibility
Compatibility of cryogens with their containers should be carefully considered. Many cryogens,
particularlypropellants,areextremelyreactive.
Theresistanceofmanyalloystotheseliquidsdependsontheformationofaninertcorrosionresistant
barriercoating.Inseveralcasesthemetalscoulddecomposetheliquid.Finally,severalmetaloxidizer
combinationsmayigniteunderimpact,nottomentionthattherearecryogenswhicharehazardous
materialsandrequirecarefulhandling.
Hydrogen embrittlement is considered in Clause 9.3.1 and material compatibility with Oxygen in
Clause1.1.1and9.2.2.CompatibilitydataofmaterialswithAmmonia(NH3)aresummarizedinTable
614,below.

211
ECSSEHB3101Part14A
5December2011

Table614:CompatibilityofMaterialswithAmmonia.Nonmetals.

Material Temperature,[K]a

Gas Liquid

Class1 Class2 Class3 Class4 Class1 Class2 Class3 Class4

OrganicMaterials 297 297


Rubber,Hard
Linings

Rubber,Soft 297 297


Linings

Rubber,Natural 297 HOT

GRS* 297 HOT

Neoprene 297 HOT

ButylRubber 297 HOT

Thiokol COLD

GlassFabricand HOT
Silicone
Elastomer

SiliconeGreases HOT

Haveg41Epon 373

Silicone 297
Elastomer

SiliconeResins 297

Teflon HOT

Cork 297

Vinyl HOT HOT


Copolymers

Phenolics HOT HOT

Furans HOT HOT

Polyethylene HOT HOT

212
ECSSEHB3101Part14A
5December2011

Material Temperature,[K]a

Gas Liquid

Class1 Class2 Class3 Class4 Class1 Class2 Class3 Class4

KelF HOT HOT

Vinylidene COLD COLD


Chloride

SulfurCement COLD COLD

Bitominous COLD COLD


Composition

Polystyrene 297

Polyesters 297

Phenol 297
Formaldehyde

Nonmetals 373
Glass

Stoneware 373

Karbate 1366

Carbon 1366

Graphite 1366

NOTE *GlassFabricandSiliconeRubber.
a Maximumtemperaturepermissibleforagivenmaterialinclasses1to3.Minimumtemperatureatwhicha
givematerialbecomesclass4.


CompatibilityClassificationforNonMetals

Class

1 2 3 4

VolumeChange 0to+25 10to+25 10to+25 <10or>+25


Percent

DurometerReading 3 10 10 <10or>+10
Change

213
ECSSEHB3101Part14A
5December2011

EffectonCryogen None SlightChange ModerateChange Severe

VisualExamination NoChange SlightChange ModerateChange Severelyblistered


orcracked,dissolved

GeneralUsage Satisfactory Satisfactoryfor Satisfactoryfor Unsatisfactory


repeatedshort shorttimeusage
termusage


Table614(Cont.)CompatibilityofMaterialswithAmmonia.Metals

Material Temperature,[K]a

Gas Liquid

Class1 Class2 Class3 Class4 Class1 Class2 Class3 Class4

Aluminium 373 311 353 353 353 353

302Stainless 297 <755


Steel

304Stainless 589 <755


Steel

316Stainless 589 <755


Steel

347Stainless 297
Steel

410Stainless 589 728


Steel

430Stainless 589 <755


Steel

Worthwite 297 <755

Durimet20 297 <755

Carpenter20 589

MildSteel 589 <755 297

CastIron 589 <755 297

SiIron 373 297 297

214
ECSSEHB3101Part14A
5December2011

NiCastIrons, 297 344


LowCu

NiCastIrons, 297 ALL


HighCu

Nickel 533 533 <866 297

Inconel 644 866 866 297

Monel 533 533 <866 297

HastelloyB 589 589 811

HastelloyC 589 589 811

HastelloyD 589 589 811

HastelloyF 589 811

Chlorimet23 297

NickleCopper 297

Copper 297 HIGH LOW

YellowBrass 297 HIGH LOW

RedBrass 297 HIGH LOW

TinBronze 297 HIGH LOW

AlBronze 297 HIGH LOW

SiBronze 297 HIGH LOW

CuNickel 297 HIGH LOW

Gold 373 HIGH

Lead 297 400 297

DowMetalC LOW

DowMetalF1 LOW

DowMetalH LOW

DowMetalJ1 LOW

DowMetalM LOW

215
ECSSEHB3101Part14A
5December2011

Platinum 373 HIGH

IrPlatinum HIGH HIGH

RhPlatinum HIGH HIGH

Silver 297 HOT

AgCu ALL ALL

Titanium 353

Tantalum 373 HIGH 373 HIGH

Zinc 297

Zirconium 353
a Maximumtemperaturepermissibleforagivenmaterialinclasses1to3.Minimumtemperatureatwhicha
givematerialbecomesclass4.


CompatibilityClassificationforMetals

Class Rating PenetrationRate Decompositionof Shock


[mils/yr]b Cryogen Sensitivity

1 Excellent <1 No No

2 Good <5 No No

3 Fair 5to500 Some No

4 Poor 50 Extensive Yes


b 1mil=25,4x106m.
NOTE FromRunck(1965)[203].

6.4.7 Testing
Testofphaseseparationdevicestobeusedonboardspacecraftseemsatfirstglancetodependonthe
ability to reproduce the low gravity levels which prevail in orbital flight. This is a hard task
particularlyfortimesexceedingafewseconds.Nevertheless,engineeringjudgementanddimensional
analysiscanbeprofitablyusedtosimulateinorbitbehaviorthrough1gtests.
1. When fluid motion is forced so that the imposed pressure differences are larger than
hydrostaticpressuredifferencesat1g,theeffectofthegravitylevelisirrelevant.System
behaviorcanbeunderstoodintermsoflaboratoryobservations.
2. Systems which exhibit similar performances working in their normal and in upside
downpositionarenotverysensitivetothegravitylevel.

216
ECSSEHB3101Part14A
5December2011
3. Matters are quite different when imposed pressure differences are low and nonplanar
interfaces exist, as in the case of capillarydominated configurations. In particular,
interfaces are larger under reduced gravity since the weight of the fluid contained by
surfacetensionforcesissmaller.EveninthiscaseusefulresultscanbeobtainedonEarth
byreproducingthedominantdimensionlessparameters.

6.4.7.1 Dimensionless parameters


For purely mechanical problems (no gradients of temperature and/or concentration), and when
interfacesarepresent,therelevantdimensionlessparametersare:
Bondnumber:ratioofgravitytosurfacetensionforces,
Bo=gl/(/R).
Capillarynumber:ratioofviscoustosurfacetensionforces,
Ca=V/.
Webernumber:ratioofinertiatosurfacetensionforces,
We=V2l/.
where V is a characteristic velocity, l and R are characteristic length (usually l is measured in the
directionofgravityactionandRthesmallestradiusofcurvatureoftheinterface), isthedensityof
thedenserliquid,thedensitydifference,thesurfacetensionandthedynamicviscosityofoneof
theliquids(normallythedenserone).
To exactly reproduce all three parameters is usually not easy nor required. In hydrostatics the
controllingparameteristheBondnumber,Bo.
ReducedBondnumberoperationcanbeachievedonEarthbyreducingl(whichisoftenimpractical)
or (Plateau simulation). Another technique consists in using a colloidal suspension of magnetic
particles in the test fluid which is suported by an electromagnetic field, the strength of which is
adjustedtosimulatelowgbodyforcesonthefluid.
In the case of bulk flow, when no interfaces appear, the relevant parameters for purely mechanical
problemsare:
Froudenumber:ratioofinertiatogravityforces,
Fr=V2/gl.
Reynoldsnumber:ratioofinertiatoviscousforces,
Re=Vl/.
Eckertnumber:ratioofpressuretoinertiaforces,
Ec=p/V2,
wherelisacharacteristiclength,pareferencepressureandtheothervariablesaredefinedasabove.
SometimesaGrashofnumber,Gr,isused,Gr=Re/Fr.
Thereproductionofthesenumbersdoesnotpresentagreatdifficultyviaanappropriatechoiceofthe
characteristiclength,l,theworkingfluid, and ,theforcedflowvelocity,V,andthepressurelevel,
p.

217
ECSSEHB3101Part14A
5December2011

6.5 Existing systems

6.5.1 Introduction
Thereareavastvarietyofcryogenicsystems,forterrestrialandspaceapplications,whoseinsulation
is cooled by an evaporating cryogen. Stored cryogens in space have long been used for several
purposes:rocketpropulsion,environmentalcontrolsystems,andfuelcells.Liquidneonwasusedfor
Geminiflights6and7toprovide32Kcoolingforinterferometerspectrometers.Nevertheless,onlyin
recentyearshasdevelopmentproceedonVCSDewarsforlongtermcoolingofdetectorsandoptical
components.
Stored solidcryogen systems offer distinct advantages over stored liquid systems for 10 K to 90 K
coolinginspace(seeClause6.1inthisClause).Liquidhelium(He4)isusedtocoverthe1,8K4,2K
temperaturerange.
InmanyinstancestheVCSsarecooledbythevaporsfromthecryogenstoredintheinnercontainer.
Some other systems, however, use a guard cryogen in addition to the primary cryogen. Figure 664
shows a schematic of a dual stage solidcryogen cooler. The primary cryogen maintains the desired
sensor temperature, while the secondary cryogen, which has a substantially higher heat of
sublimation than the primary cryogen, intercepts heat from the outer shell and provides a low
temperatureenvironmentaroundtheprimarystage.

Figure664:Sketchofadualstagesolidcooler.FromNastetal.(1976)[161].

Figure 665 depicts two configurations of He4 coolers for instruments. That in Figure 665(a) is a
singlestage cooler whose toroidal shape allows the IR telescope to be surrounded by the cooler.
Figure665(b)showsacooleremployingasolidcryogensecondarystage.

218
ECSSEHB3101Part14A
5December2011

Figure665:Liquidhelium(He4)coolers.a)Singlestage.b)Dualstage.From
Sherman(1978)[216].

AcriticalproblemintheperformanceoftheVCSinsulationsystemsistheattachmentofthecooled
shieldstothecoolingduct.Thisattachmentcanbeachievedeithernormallyortangentially.
1. Inthenormalattachmentmodethecooledshieldsreachthecoolingductperpendicularly
toit.Thethermalcontactcanbeachievedeitherjoiningtheshieldstotheventingduct,by
soft soldering or brazing, as in Figure 617, or through heat stations across the piping
penetrations which are thermally bonded to both a support tube and the cooling duct,
Figure666.

Figure666:NormalattachmentoftheVCSstothecoolingductthroughheat
stationis.FromGlaseretal.(1967)[75].

NormalattachmentisusedinsmallsystemshavingseveralVCSs.
2. Inthetangentialattachmentmodetheventingduct,usuallyintheshapeofacoolingcoil,
is soldered or cemented to the metallic shields. The venting duct, which originates

219
ECSSEHB3101Part14A
5December2011
somewhereinthenecksupport,spiralsaroundtheinnermostshield.Afterthisspiral,the
ductisattachedforthenextonearoundthenearestshield.Itmakesanotherspiralbefore
reachingthenextshield,andsoon,Figure667.

Figure667:Tangentialattachmentofthecoolingducttotheshields.Sketchedby
thecompilerafterHopkins&Chronic(1973)[94].

Severalideashavebeensetforthinordertoincreaseboththeavailabilityandtheperformanceofsolid
cryogencoolers.Themultimissioncooler(Sherman(1978)[216])wouldbereusableandcompatible
withavarietyofcryogenstocovera10Kto90Ktemperaturerangewithaonetothreeyearslifetime.
Another idea, which would increase the lifetime of solidcryogen coolers has been considered by
Sherman & Brennan (1976) [219]. It consists in using a heat piperadiator module to cool either the
outershellofthecooleroranintermediateshield.

6.5.2 Data on existing systems


Data on several spaceborne storedcryogen systems (two dual stage solidcryogen coolers and three
liquidheliumcoolers)arepresentedinthefollowingpages.
The information on each cooler is arranged as follows: the left hand side page presents a brief
descriptionandcommentsconcerningthesysteminvolved,whereastherighthandsidepagecontains
asummarytableandasketchofthesystem.
The most extensively documented among the systems presented was flown on Nimbus F
(LOCKHEED VCS DEWAR 1975). In addition to a complete description of the system and its
performance, due to Nast et al. (1976) [161], a theoretical study of the gasdynamic behavior of the
methane venting has been published by Rae & Dunn (1976) [189]. The aim of this study was to
estimate the torques applied to the spacecraft as a result of the vented gases impinging on several
surfacesneartheexitpoint.

6.5.2.1 Lockheed VCS Dewar 1968


DESCRIPTION AND COMMENTS. This IR Detector Cooler, developed by Lockheed Missiles and
SpaceCo.,PaloAlto.Calif.,isdesignedtooperateat50Kfor1year,providingacoolingrateof0,025
W.
Thecoolerconsistsofthreemainparts:
1. Acontainerwithsolidargon,forcoolingtheIRdetectorviaacooperthermallink.

220
ECSSEHB3101Part14A
5December2011
2. Another container with solid carbon dioxide, which is the source of the vapors which
cooltheinsulationsurroundingthesolidargoncontainer.
3. Aliquidnitrogenheatexchangerforthesolidificationofbothcryogensduringthefilling
process.
Theusefulnessofsolidcarbondioxideasasourceofcoolingvaporsarisesfromthefactthatthelatent
heat of sublimation of carbon dioxide is approximately three times that of argon, whereas their
densitiesareverysimilar.
Thesolidargonandsolidcarbondioxidecontainersarefromstainlesssteelsheets0,61x103mthick
(cylindrical surfaces) and 0,91x103 m thick (top and bottom). Both containers were designed to
withstandinternalandexternalpressuredifferencesupto105Pa.Thesolidcarbondioxidecontainer
is placed above the solid argon container, and is thermally connected to a 3,81x105 m thick copper
shield which surrounds the solid argon container. Thence, this solid argon container, which is gold
plated,issubjectedtosolidcarbondioxideradiationtemperature.A5,08x102mthickMLIthermally
insulatesboththecooledcoppershieldandthesolidcarbondioxidecontainer.ThisMLIconsistsin
150 layers of doublealuminized Mylar with Tissuglas spacers. A cage of stainless steel tubing
(1,59x103 mouter diameter and0,254x103 m wall thickness), structurally joined to the solid carbon
dioxide container, supports the cooled copper shield and the MLI in the area surrounding the solid
argoncontainer.Asecondradiationshield,withaconstructionsimilartothatoftheabovementioned
outer shield, is placed amidst the solid argon container and the outer shield. The inner shield is
supportedbyasystemofNylonthreads.
The solid carbon dioxidecontainer is supported by a cantilever beam with a fixed end support and
concentrated load at full span. The beam is made from fiberglass filament wound structure
impregnatedwithanepoxyresin.Itsinternaldiameterisuniform,4,37x102m,whereasitsthickness
varies from 1,14x103 m at its clamped end to 0,305x103 m at its free end. Aluminium flanges are
fastenedtoeachendofthebeamwithepoxyadhesive.Thebeamisboltedtothecoolersupportflange
andtothecarbondioxidecontainerthroughtheseflanges.
Thesolidargoncontainerissupportedfromthebottomofthesolidcarbondioxidecontainerbythree
fiberglassepoxylaminatecolumns,12,7x103mdiameterand0,51x103mwallthickness.Inorderto
increase the thermal path length as far as possible, these columns extend from the bottom of the
carbondioxidecontainertothebottomoftheargoncontainer.
Thebeamsupportingthecarbondioxidecontainerisalsousedtoroutethefourfillandventinglines
(two for the liquid nitrogen heat exchanger, one for the solid carbon dioxide container, and the last
oneforthesolidargoncontainer).Tubingisstainlesssteel,exceptforseveralstraightportions,which
areMylaraimingatreducingheatleaks.
Thecontainers,whosefillingismadewiththecryogensinthegaseousphase,arecooledduringthe
filling process with liquid nitrogen flowing through a heat exchanger fashioned from 0,653x103 m
diameterstainlesssteeltubingsolderedtotheouterwallsofthecontainer.Toprovidegoodthermal
contactbetweentheliquidnitrogenheatexchangerandbothcryogensduringthecooldownprocess
thecontainersarepartiallyfilledwithcooperexpandedfoam.Thesamecopperfoamprovidesgood
thermalcontactduringthemissionbetweensolidargonandcopperthermallinkwiththeIRdetector.
Theargongasisliquifiedandsubsequentlysolidifiedduringcooldown.Argonmustbecontinuously
suppliedinordertomaintainagaspressureinexcessof0,685x10Pa(Argontriplepointpressure).In
thismanner,voidsoccurringbecauseofsolidificationcontractionarebeingcontinuouslyfilled.Solid
carbondioxideisgrowndirectlyfromthevaporphase.Topreventthegrowingsolidcarbondioxide
from blocking the feeding line, the inlet manifold is located near the innermost section of the
container,asdistantaspossiblefromtheliquidnitrogenheatexchanger.
Thesystemhasbeentested.Table615belowcomparespredictedtomeasuredcharacteristics.

221
ECSSEHB3101Part14A
5December2011

Table615:CharacteristicsoftheLockheedVCSDewar

Design MeasuredValue
Value

DetectorOperatingTemperature[K] 50 52

AllowableDetectorCoolingLoad[W] 0,025 0,0176

HeatLeaktoSolidArgon[W] 0,015 0,0286

HeatLeaktoSolidCarbonDioxide[W] 0,076 0,074

TotalMass(OuterVacuumJacketExcluded)[kg] 13,57 15,47

NOTE Reference:Caren&Coston(1968)[36].

MANUFACTURERAND/ORDEVELOPER LockheddMissilesandSpaceCompanya

PRIMARYUSE IRDetectorCooler

CRYOGEN SolidArgonandSolidCO2

NATUREOFPROGRAM ConstructionofaPrototype

CURRENTSTATUSOFPROGRAM PrototypeTestedbyMarch1967

OUTERTEMPERATURE 300 NUMBEROFVCS 1


[K]

OPERATING 50b DETECTORCOOLING 0,025


TEMPERATURE[K] LOAD[W]

OPERATINGPRESSURE 0,015b
20b TOTALHEATLEAKS[W]
[Pa] 0,076c

LIFETIME[h] 8760(1yr) SYSTEM Cylindrical


CONFIGURATION

TOTALSYSTEMMASS 13,57
[kg] SYSTEMDIMENSIONS
[m] 0,235/0,235
6,22
b Length/Diameter
MASSOFCRYOGEN[kg]
3,99c
a PaloAlto.California.USA.
b CorrespondstoSolidArgon.
c CorrespondstoSolidCO2.

222
ECSSEHB3101Part14A
5December2011

6.5.2.2 LOCKHEED VCS DEWAR 1975


DESCRIPTION AND COMMENTS. This dual stage IR detector cooler, developed by Lockheed
Missiles & Space Company, Inc. Palo Alto. Calif., has been employed to cool to 6367 K a trimetal
detector array and the associated focusing optics and filters, and also to cool at 152 K other optical
elements.Thecoolingrateprovidedtothedetectorwas0,043W.Thesystemwasplacedintoorbiton
theNimbusFSpacecraft.
The cooler consists of a primary cryogen (solid methane) and a secondary or guard cryogen (solid
ammonia) whose pressure is maintained below the triple pointin order to prevent liquidformation
and coolant loss in the microgravity environment. The primary cryogen maintains the sensor
temperatureat652K.Thesecondarycryogenprovidesalowtemperatureenvironmentaroundthe
primarystageandcoolstheopticsofthedetector.
Anopencycleliquidnitrogencoolantloopforgroundoperationisusedforcoolingpurposesduring
filling, and for maintaining the cryogens in a nonvented condition during vehicle integration and
checkout.
Thefinalconfigurationofthecoolerisshowninthenextpage.Themaincomponentsare:thedetector
capsuleassembly(DCA),thefiberglasstubesupportingstructure,thetankscontainingtheprimary
and the secondary coolants, a VCS grounded to the ammonia tank, multilayer insulations (MLI)
aroundthetanksandfiberglasssupporttubes,plumbinglines,outershellandmountingplate.Fiber
glass tube 1 connects the main support flange to tube 2 which is connected to the ammonia tank.
Fiberglasstube3connectstheammoniatanktothetube4,whichisconnectedtothemethanetank.
An aluminium tube, thermally grounded to theammonia tank, is used to decrease the heat transfer
betweentubes2and3andtocooltheopticsofthedetector.SlittedMLIisusedbetweentubes:slitting
aims at reducing the lateral heat transfer in the MLI. MLI also insulates the methane tank from the
ammonia tank and itsgrounded shields, andinsulates the ammonia tank andshield from the outer
shell.

223
ECSSEHB3101Part14A
5December2011
Thetotalmassis24kg,including5,4kgofsolidammoniaand6,4kgofsolidmethane(thelaunched
massofsolidmethanewas,however,5,8kg).Thediameterofthecooleris0,356manditslength0,654
m.
The DCA interfaces with the cooler through two shrinkfit connections. The first one connects the
detectorfocalplanetothemethanetank,andthesecondconnectstheDCAthermalguardandoptics
toathermalpathtowardammoniatank.TheDCAcanberemovedeasilyatroomtemperature,while
at operational temperatures the differential contraction on the shrinkfit members provides thermal
contactandstructuralintegrity.
Support tubes were made of 1543/E787 fiberglass. The tanks were made of 6061 aluminium. Vent
linesareofconvolutedTeflon,whichminimizesheatleaksandretainsahighdegreeofflexibilityat
cryogenic temperatures. The methane fill/vent Teflon tube goes from the methane tank through the
MLI, then to the ammonia tank, where it is thermally grounded, and through the ammonia MLI to
space.Theammoniafill/ventTeflontubegoesfromtheammoniatankthroughtheammoniaMLIto
space.Thesensibleheatoftheventgasisusedtoremoveheatfromtheammoniatank.Effectivevapor
coolingofbothventlinesandoftheMLIalsooccurred.
TheMLIofthetanksconsistsofdoublealuminizedMylarTissuglas.Layerdensitywas4300m1and
layerthickness25,4x103m.Theinsulationwasappliedtothetanksbyspirallywrappingthelayers,
goringtheends,andthenformingbuttjointsusingaluminizedMylartape.Thisprocesswasrepeated
tenlayersatatime(upto110layers)withthegorejointsstaggeredtominimizeradiationheattransfer
throughthegaps.
The vacuum shell consists of a 6061 aluminium cylinder with a spun elliptical dome bonded to the
cylinder,andaflangewhichcontainsan0ringandprovidesthemainvacuumseal.
Themountingplateisa6061aluminiumwebstiffenedstructuretowhichtheexteriorplumbing,outer
shell,andcoolerareattached.
The cooler was launched aboard the Nimbus F vehicle on June 12, 1975. At the time of launch the
methane and ammonia were 73 K and 137 K respectively. On the fourth orbit the explosive valve
ventingthemethanetospacewasfired,andtheinitialhighmethaneflowratetemporallyupsetthe
spacecraftintherollaxis,butcontrolwasreestablishedafterashortperiodoftime.
Table616showsthepredictedheatloadsthroughthedifferentcomponents.Thetemperaturehistory
oftheinfrareddetectorsandtheopticsispresentedinFigure668.Thedetectorremainedwithinthe
specific temperature for over six months. The temperature rise near the end of life is believed to be
duetotheincreasingtemperaturegradientbetweenthesmallremainingvolumeofsolidcryogenand
the container walls. The temperature rise occurred after approximately seven months, when the
methanewascompletelydepleted.Theammonialifetimeexceededthemethanebyalargevalue,as
showninFigure668.Theammoniatemperaturewasconstantwithinthedataresolution(0,8K)until
methanewasexhausted,atwhichtimethaammoniareachedanewequilibriumtemperature.

224
ECSSEHB3101Part14A
5December2011

Table616:PredictedHeatLoads

Qx103[W]

AmmoniaGross* MethaneNet

MLI 214 29
FiberglassSupports 36 31
Radiation 41 13
VentGas 27
DexiglasIntermediary 12
SlittedMLI 13 12
LN2andVentLines 16 4

Total 305 89

NOTE *Thenetbalanceontheammonia,whichincludesheatlossestothemethane,iscalculatedtobe0,226
W.
Reference:Nast,Barnes&Wedel(1976)[161].

Figure668:Detector,T1,andoptics,T2,temperaturevs.orbitaltime.

MANUFACTURERAND/ORDEVELOPER LockheedMissilesandSpaceCompanya

PRIMARYUSE IRDetectorCooler

CRYOGEN SolidMethaneandSolidAmmonia

NATUREOFPROGRAM DevelopmentofaFlightModel

CURRENTSTATUSOFPROGRAM FlowninJune1975onNimbusP

OUTERTEMPERATURE 300 NUMBEROFVCS 1


[K]

OPERATING 63,564,5b DETECTORCOOLING 0,043

225
ECSSEHB3101Part14A
5December2011

TEMPERATURE[K] 152,4c LOAD[W]

OPERATINGPRESSURE 0,089b
TOTALHEATLEAKS[W]
[Pa] 0,226c

LIFETIME[h] 5040b(7mo) SYSTEMCONFIGURATION Cylindrical

TOTALSYSTEMMASS 24
[kg] SYSTEMDIMENSIONS[m]
0,654/0,356
Length/Diameter
5,76 b
MASSOFCRYOGEN[kg]
5,35c
a PaloAltoResearchLaboratory,PaloAlto,California.USA.
b CorrespondstoSolidMethane.
c CorrespondstoSolidAmmonia.

6.5.2.3 JPL & Caltech VCS Dewar


DESCRIPTIONANDCOMMENTS.ThisIRDetectorCooleristheresultofacooperativeeffortbyJPL
andCaltechaimingatthedevelopmentofasystemforcoolingto2Kabolometersensor.
The detector is incorporated into the inner container, in good thermal contact with the superfluid
helium,andwellwithinthevaporcooledshields.Inthiswaytheheatinputtothedetectorisgreatly
reduced,theadditionalheatinputthroughtheopeningintheVCSinsulationsystembeingnegligible.
The heat addition to the cryogen is reduced further since the outer container comes only in contact
withthreemainareas(Figure669):thedeepspace,thebus,fromwhichitisinsulatedandshielded,
andthetelescopewhosetemperatureiscloseto100K.Consequently,theoutercontainertemperature
iscloseto100K.
Thermal insulation is provided by a system having three equally spaced vapor cooled shields. The
distancebetweenshields,andbetweentheboundingfacesoftheinsulationandneighbouringshields

226
ECSSEHB3101Part14A
5December2011
areallequalto25,4x103m.Theshieldsprovidebothradiationshieldingandcoolingoftheinsulation
andsupports.Nopaddingbetweenshieldsisusedexceptforsupportingpurposes.

Figure669:JPLCaltechIRdetectorcoolerarrangement.

Theventingduct,whichismadefromstainlesssteeltubing,isthermallyconnectedtoeachshieldin
turn.Ventingductlengthis1,53m,anditswallsurfaceareais0,72x104m2.
Two support designs have been considered.Both are sketchedin the following page,although only
onewillbeusedinthefuturedevelopments.Thetitaniumstrutprovidesamorerigidlinkingbetween
detectorandtelescope,butafiberglasspadlooksmoreefficientand,thence,itisonlyoneanalyzed.
Each support (titanium rod or fiberglass pad) is thermally connected to the various cooled shields
andpassesthroughthemsothattheheatleakthroughthesupportsislargelyinterceptedandcarried
offbytheoutflowingheliumgas.Thetotalcrosssectionalareaoftheeightfiberglasssupportpads,
fourateachendofthecontainer,is0,15m2.
Theheattransferprocesswithinthecryostatiscontrolledbythefollowingremarkablepropertiesof
superfluidheliumII(seeClause7inthisPart):1)afilmwiththicknessoftheorder107mspreadsover
all the inner surfaces, 2) the superfluid component flows readily through this film toward the heat
sources, whereas the normal fluid flows away from the source carrying the heat either through
convectionorthroughevaporation.
Asetofcapillarytubes,5x103mindiameter,runningparalleltotheaxisofthecryostat,havebeen
incorporated into the design. Its aim is twofold: 1) to avoid liquid sloshing during launch and
insertion operations, and 2) to assure that the liquid helium is in thermal contact with the detector
mounting.Thisthermalcontactisachievedthroughsurfacetensionwhich,inordertominimizethe
liquidfreesurfacearea,increasestheareaoftheheliumcapillarytubeinterface,whichresultsinthe
thickeningoftheheliumlayerincontactwiththedetector.
Liquidvaporphaseseparationintheventingductispresumablyachievedbyasuperfluidplug(see
Clause7.4inthisPartand,inparticular,theinformationfromSchotte(1984)[209]).
The flow of boiloff helium is restricted at its exit by a variable orifice valve, controlled so as to
maintaintheinternalpressureat1,63x103Pa.(heliumvaporpressureat1,8K).
Reference:Mason(1972)[142].

227
ECSSEHB3101Part14A
5December2011

MANUFACTURERAND/ORDEVELOPER JPLandCaltecha

PRIMARYUSE IRDetectorCooler

CRYOGEN LiquidHelium

NATUREOFPROGRAM DesignStudyb

CURRENTSTATUSOFPROGRAM Completed

OUTERTEMPERATURE 100 NUMBEROFVCS 3


[K]

OPERATING 2 DETECTORCOOLING 0,035


TEMPERATURE[K] LOAD[W]

OPERATINGPRESSURE 1,67x103 TOTALHEATLEAKS[W]


[Pa]

LIFETIME[h] 4380(6mo) SYSTEM Cylindrical


CONFIGURATION

TOTALSYSTEMMASS 116 SYSTEMDIMENSIONS[m]


[kg] Length/Diameter

MASSOFCRYOGEN[kg] 27,3 0,80/0,65

a JetPropulsionLaboratoryandCaliforniaInstituteofTechnology.Pasadena.California.USA.
b PreliminaryphaseoftheIRASDewardevelopment.IRAS,ajointDutchUKUSsurveysatellite,waslaunched
in1981(Neugebauer(1978)[167]).

228
ECSSEHB3101Part14A
5December2011

6.5.2.4 Garrett VCS Dewar 1975


DESCRIPTION AND COMMENTS: This Dewar was developed for an orbiting highfield
superconductingmagneticspectrometer,withaoneyearmission,whichwasplannedasapartofthe
NASAHighEnergyAstronomicalObservatory(HEAO)seriesofscientificflights.Theseflightswere
initiallyscheduledfor19751977.
The cryogenic cooling system was based on a feasibilitystudy performed by Ball BrothersResearch
Corp.AcontractfordesigningandbuildinganengineeringmodelofthesystemwasawardedinMay
1972 to AiResearch division of the Garrett Corporation in Torrance, California. In January 1973 the
HEAO satellite program was redirected and the superconducting magnet experiment postponed ( a
Dewar, developed by Ball Brothers Research Corp., which will be used on the (HEAO) B and C is
described by Sherman (1978) [216]. This Dewar will contain methane as the primary cryogen and
ammoniaasthesecondary.Thedimensionsare0,76mlongand0,56mindiameter.Theloadedmass
willbe75kg.Coolingload:2x102W).
TheGarrettVCSDewarhasalengthof2,38mandanouterdiameterof1,83m.Itcontainsabout430
kgofliquidheliumwithapproximatelya5%ullage.
Thesupportsystemconsistsofarelativestiffarrayofsixteenfiberglassepoxybandsbetweeneight
equallyspacedattachpointsonacentraloutershellgirthring,andfourattachmentpointsateachend
ofthepressurevessel.Thebandsupportsareroughly0,68mlongwithatotalcrosssectionalareaof
about 8,77x104 m2. The band supports are preloaded in tension to about 0,483x109 Pa to ensure a
relativelyhighstressaftercooldown.WiththeflightpressurevesselandtwomagneticcoilstheDewar
minimum axial natural frequency would be about 20 to 30 Hz, which would keep maximum load
factorbelow10gonaTitanIIIBlaunch.Thesupportsaremechanicallyandthermallyconnectedto
thevaporcooledshieldstointerceptsomeoftheconductionheatwhichwouldotherwiseflowtoward
thepressurevessel.
The insulation system consists of MLIs and to aluminium VCSs 103 m thick. The pressure vessel is
insulatedwithonelayeroflowemittancefoil,whereastheinnerandouterVCSsarewrappedwith40
and 120 layers respectively of 6x106 m doublealuminized Mylar with a 104 m thick Dacron net
spacer at a layer density of about 2600 layer.m1 maximum (see ECCSE3009 Part 7 clause 6.11 ,
DoubleAluminized Mylar, Silk Netting, for further details regarding similar MLI systems).The
insulation thickness is 0,114 m everywhere except at the spatial detector end where the coil is

229
ECSSEHB3101Part14A
5December2011
mounted. There the minimum thicknessis0,076 mon the axis between the two torispherical heads,
whichapproximate2/1ellipsoids.
TheDewarhasa2,5mlongstainlesssteelfillline,a3,6mlongfillventline,anda46mlongshield
heat exchanger line. The innermost vaporcooled shield functions as a boiler, or external
thermodynamicphaseseparator,inaliquidexpulsionmode(seeClause1.1.1).Theshieldlinestartsat
atee,approximately1,5mfromthepressurevessel,inthefillventlineotheboilershield,itisspirally
wrappedovertheshield,andterminatesatthetankpressureregulatoroutsidetheoutershell.
Phase separation is achieved by using the diamagnetism of superfluid liquid helium (helium is
repelledbymagneticfields,buttheforceperunitvolumeontheliquidismuchlargerthantheforce
perunitvolumeonthevapor).Tothisaimthemouthoftheventisplacednearthemagnet,wherethe
product of magnetic field and field gradient is maximum, this results in phase separation which
increasestheoperationallifetimeoftheDewarbyabout25%(ascomparedwithliquidexpulsion).
Allinstrumentationtothecryostatpressurevesseliscontainedineitherthefilllineortheventline,
thuseliminatingtheneedforcoldvacuumtightelectricalfeedthroughs.
Reference:Pope,Smoot,Smith&Taylor(1975)[186].

MANUFACTURERAND/ORDEVELOPER GarrettAiResearcha

PRIMARYUSE SupercoolingMagnetCooling

CRYOGEN LiquidHelium

NATUREOFPROGRAM DevelopmentofanEngineeringModel

CURRENTSTATUSOFPROGRAM Completed

OUTER NUMBEROFVCS 2
TEMPERATURE[K]

OPERATING 4 DETECTORCOOLING
TEMPERATURE[K] LOAD[W]

OPERATING TOTALHEATLEAKS[W]
PRESSURE[Pa]

LIFETIME[h] 8760(1yr) SYSTEM Cylindrical


CONFIGURATION

TOTALSYSTEM 1320b SYSTEMDIMENSIONS[m]


MASS[kg] Length/Diameter

MASSOFCRYOGEN 430
[kg] 2,38/1,83

a DivisionofGarrettCorp.,Torrance,California,USA.
b Magnetsystemmassisalsoincluded.

230
ECSSEHB3101Part14A
5December2011
6.5.2.5 Ball Brothers VCS Dewar 1976
DESCRIPTION AND COMMENTS. Over the years 1974 to 1976 Ball Brothers Research Corporation
hasdesignedandfabricatedalonglifeliquidheliumDewar.Thisdevelopmentprogramwasinitiated
to provide a refrigeration system for future scientific experiments in space such as IR telescope,
superconducting magnets, and precision gyroscopes. The Dewar can contain 86 kg of superfluid
heliuminspaceforamissionof1year.
Analyticalsimulationofthestructuralperformanceproveddifficult,soastructuralmodelwasbuilt
prior to completion of the engineering model by mid1975. Testing with normal liquid helium at
ambient conditions successfully concluded the program. Additional testing with superfluid helium
wasconductedinearly1976.
the pressure vessel (inner container) is 1,02 m long and is made with 5083 aluminium. Two
torisphericalheads,oneateachend,arejoinedbyacylindricalcentersection.Aninstrumentcavity,
0,457mindiameter,isplaceinthecenterofthecontainer.Fivevalvesandtwosuperfluidplugs(see
Clause 7.2) are placed on the cavity. The valves are used to direct and control flow filling, venting,
and/orporousplugoperation.Afterfilling,acrossovervalvemaybeopenedandthecryostatvented
throughboth(fillingandventing)lines,ifrequired.Oneofthesuperfluidplugsisceramicandhasan
average pore size of0,5x106 m; the other plugis nickel, with anominal pore size of2x106 m. Both
plugshavebeentestedinthelaboratoryandweresetupinthecryostatforcomparisontesting.
The insulation system consists of four MLI blankets (6,35x106 m thick Mylar sheets aluminized on
both sides and spaced by Dracon net (see ECSSEHB3101 Part 7 clause 6.1.1, DoubleAluminized
Mylar, Silk Netting, for further details regarding similar systems)) whichare separated by three
aluminium vapor cooled shields. Stainless steel filling and venting ducts are attached to the vapor
cooledshieldsbymetal)clipsandaluminiumfilledpolyurethaneresin.
The inner container, the VCS and the outer shell are structurally linked by a support system. Six
titaniumsupportsareprovidedforgroundandlaunchoperations.Thesesupportsareretractedinto
theinnercontainerfororbitalflight.Inorbit,thesystemconsistsexclusivelyofsixfiberglasssupports
whicharecooledthroughtheVCS.
Theoutercontainerisconstructedof5083aluminium,andconsistsoftwotorisphericalheads,of1,37
mouterdiameter,joinedby0,856mlongcylinder.Acoverplate,0,61mdiameter,inthetopheadis
providedforaccesstotheinnercavity.Externalplumbingisattachedtotheoutercontainer.
Thefirstlifetimetestofthethermalmodelinvolvedapproximatelytwomonthsoftesting,ofwhich
the last three weeks were the formal stabilization time. During the last week, the VCS Dewar
insulation system was stabilized to within 1 K, and the flow rate was stable to less than 3%.
Vacuum pressure was 5x104 Pa and the tank pressure was 105 10 Pa. The environmental
temperaturewas2932K.Theheliummassboiloffratewas2,5x106kg.s1forthelastweekofthe
test. This correspond to an overall heat leak of 0,059 W and secures a storage time of 8110 h. The
calculatedlifetime,basedonaheatleakof0,041Wwas11600h.Aheliumleaklocatedintheinternal
plumbingcouldaccountforthebelowpredictedexperimentalperformance.
Achartofthevariousthermalelementheatleaks,temperatures,andcoolingprovidedbytheeffluent
gaseousheliumispresentedinFigure670.

231
ECSSEHB3101Part14A
5December2011

Figure670:HeatFlowdiagramoftheBallBrothersLiquidheliumDewar.

Reference:Urbach&Herring(1976)[243].

MANUFACTURERAND/ORDEVELOPER BallBrotherResearchCorporationa

PRIMARYUSE LongTermStorage

CRYOGEN LiquidHelium

NATUREOFPROGRAM Experimental

CURRENTSTATUSOFPROGRAM Completed

OUTERTEMPERATURE 300 NUMBEROFVCS 3


[K]

OPERATING 4,2 DETECTORCOOLING


TEMPERATURE[K] LOAD[W]

OPERATINGPRESSURE 105 TOTALHEATLEAKS 0,041


[Pa] [W]

LIFETIME[h] 8760(1yr) SYSTEM Cylindrical


CONFIGURATION

TOTALSYSTEMMASS SYSTEMDIMENSIONS
[kg] [m]
Length/Diameter 1,02/1,37
MASSOFCRYOGEN[kg] 86
a Boulder.Colorado.USA.

232
ECSSEHB3101Part14A
5December2011

233
ECSSEHB3101Part14A
5December2011

7
Superfluid Helium

7.1 Dynamics of superfluids


Attemperturesclosetoabsolutezero,quantumeffectswouldcontrolthepropertiesofafluid.
Heliumremainsfluidatabsolutezero,providedthatthepressuredoesnotexceedabout2,5x106Pa
(25atm).
ThephasediagramforHe4isshowninFigure71.TheregimeconsideredinthisClauseislabelledHe
IIinthediagram.

Figure71:PhasediagramforHe4(nottoscale).FromArp(1970)[10].

Liquid helium undergoes a phase transition at a temperature of about 2,2 K, at the vapor pressure.
Thistransitionisdetectedbytheshapedpeakofthespecificheatvs.temperaturecurve.
At temperatures above the lambda point, helium behaves like a normal viscous fluid. This phase is
calledHeliumI.
Below the lambda point, liquid helium Helium II has a numberof remarkable properties,the most
important of which is superfluidity. Strictly speaking only one isotope of helium, He4, becomes
superfluidbelow2,2K.In1971,however,aphasetransitionwasfoundinthelightisotopeofHelium,
He3(seeclause8.3)atatemperatureof2,7x103K.Onlyafter1974ithasbeenwidelyacceptedthat
thisnewphaseissuperfluid(Mermin&Lee(1976)[149]).

234
ECSSEHB3101Part14A
5December2011
Macroscopic superfluid dynamics is based on the following fundamental result of the microscopic
theory:AttemperaturesotherthanzeroHeIIbehavesasifitwereamixtureoftwodifferentliquids.
Oneoftheseisasuperfluid,whichischaracterizedbyadensity s,zeroentropy,zeroviscosity,and,

wheremotionisinvolved,avelocity u s .Theotherisanormalviscousfluid,whichischaracterizedby

adensity n,and,whereappropriate,avelocity u n .Nofrictionoccursbetweenthesetwopartsofthe

liquid in their relative motion, at least provided that both un and u s are small (see Figure 74 in
clause7.1.2anddiscussionthereof).
TodepicttheHeIImotionasamixtureofnormalandsuperfluidflowsallowsasimpleanalysisofthe
phenomena which occur when quantum effects are important but, like any description of quantum
phenomenainclassicalterms,itshouldbeviewedwithcaution.

7.1.1 Relevant equations of superfluid dynamics


It is not intended to derive here the complete system of differential equations and boundary
conditionswhichdescribemacroscopicallytheflowofHeII,rathertherelevantresults,whichwillbe
usedinthelastpartofthisClause,areoutlined.
Because of the existence of critical velocities (clause 7.1.2) the equations of superfluid dynamics for

helium II are physicallymeaningful only when the velocities u s and un are not too large.
Unfortunatelytherelevantequationscannotbeconsistentlyderivedfromtheconservationlawswhen
higherpowersofthevelocitiesareneglected,thereforeourapproachwillbeasfollows:
1. Thegeneraloutlineofthecalculationsleadingtothegeneralequations,withoutmaking

anyassumptionconcerningthevelocities u s and u n ,willbegiveninthisparagraph.

2. The resulting equations will be particularized to the steady, slow flow case (clause
7.1.1.1).
3. Critical velocities will appear through an heuristic introduction of the frictional effects
(clause7.1.2).
Details concerning the basic equations can be found in Landau & Lifshitz (1959)[129], Khalatnikow
(1965)[122],andRoberts&Donnelly(1974)[196],amongothers.
Ashasbeensaid,HeIImaybethoughtofasamixtureoftwocomponents:thesuperfluidcomponent
(subscripts)andthenormalcomponent(n).
1. Theconservationoftotalmasscanbeexpressedas:


j 0 [71]
t


where = s+ nistheactualdensityofHeII,whilethemassfluxdensityvector, j ,is
givenby

j s u s nun [72]

Foronedimensionalsteadyflow,themassconservationequationmaybewrittenas:

235
ECSSEHB3101Part14A
5December2011

m jAFL [73]

mbeingthemassflowrateandAFLtheductcrosssectionalarea.
2. The equations governing the momentum balance for superfluid and normal fluid are,
respectively:


u s
u s u s p sT U
1
[74]
t


u n m m
u n u n s u s n u n
t n n
[75]
1
p s s T U n u n
n n

wherep,sand narerespectivelythepressure,entropyperunitmass,andnormalfluid
viscosity, while Uis the potentialofany conservative bodyforceper unit volume (e.g.,
gravity)towhichthesystemcouldbesubjected.msandmnaredefined,respectivelyas:

s
ms sus [76]
t

n
mn nun [77]
t

Equation [75] results by substraction of Eq. [74] to the equation governing the overall
momentumbalance(seeRoberts&Donnelly(1974)[196]).
Itisfoundexperimentally(persistenceofcirculationexperiments)thatthecirculationof
the superfluid component of velocity around any superfluid material path remains
constant. This is a consequence of the KelvinHelmholtz theorem of inviscid fluids in a
potential body force field, the only difference being that now the fluid needs not to be
incompressibleorbarotropic.

Constancyofcirculationsuggeststhatthetotalderivativeof u s (lefthandsideofEq.[7
4]) must be equal to the gradient of a scalar,(+U), where will be identified with the
chemical potential per unit mass. This chemical potential is a function of p, T and the

velocitydifference, u n u s .
Neglectinghigherordertermsinthisvelocitydifference,isdefinedas:

n 2
u n u s
1
sT [78]
2

236
ECSSEHB3101Part14A
5December2011
Strictly speaking, n and s are also functions of the velocity difference, but this can be
disregardedwithinthepresentapproximation.

Inaddition,whenthesuperfluidmotionstartsfromtherest,andboth u s and u n areso
smallthatnoquantizedvorticesappearinthesuperfluid(seeclause7.1.2),thesuperfluid

flowisirrotational,thence u s isequaltothegradientofascalar,.Inthatcase,Eq.[74]
becomes,

1
U 0
2
[79]
t 2


The nonlinear momentum terms ms u s and mn u n should appear in Eq. [75] instead of
beingequallydividedbetweenEqs.[74]and[75],asnotedbyLandau,whorecognized
that constancy of circulation of the superfluid component overrules misleading
symmetryarguments.
3. An additional equation is needed for the still undefined T. This equation, which comes
fromthelawofconservationofenergy,isverycomplicatedand,initscompleteform,of
verylimiteduse.
Forconvenience,wewillwritedowntheentropybalanceequation;tothisaimwestart
withthegeneralexpressionforanonreactingbinarymixture(vonKrmn(1955))

Ds T 2
k i si vd i
Dt T i 1

1
k
T T
2

2
2

i si vd i i vdi i
[710]
T T T 1 1


where vd i is the diffusion velocity of species i and the viscous dissipation function
(Rayleigh).
The left hand side of Eq. [710] gives the entropy variation per unit volume (D/Dt
indicates local plus convective derivative). The first term in the right hand side is the
reversible entropy flux through the boundary of the unit volume element. This flux
comesfrom:1)theconductiveheattransfer,2)thediffusionofspecies.Thesecondterm
givestheentropygeneratedintheunitvolumeelementperunittime.Thetermsinvolved
comefrom:1)viscousdissipation,2)thermalconduction,3)diffusion.
Fluxesduetocoupledeffects(Onsager)havebeenneglected.
Equation[710]isnowappliedtothebinarymixtureofthenormalfluid(1with1s1=s!)
andthesuperfluid(2with2s2=0).l2=sT.Thediffusionvelocitiesare,


v d1 u n s u s n u n s u n u s
[711]

237
ECSSEHB3101Part14A
5December2011


v d 2 u s s u s n u n n u n u s
[712]

Thefollowingequationresults

s T

t

j s k

s su n u s
T

1
k
T 2 T su u n s u u sT [713]


s n s n s
T T T

Usuallytheheatfluxvectorisgivenby


q s sT u n u s
[714]

wherethermalconductiontermsareneglected.

Forastaticcounterflow,inwhichthereisnonetmassflow( j s u s nun 0 ),


q sTu n
[715]

In the case of onedimensional steady flow of He II (j = m/AFL), the following equation


results

m
q sTv n 1
vn AFL [716]

mispositivewhenHeIIflowsfromtheheatsourcetotheheatsink.Inthefollowing,v,
vn,vswillbeusedtoidentifycrosssectionalaveragesofthelocalvelocitiesu,unandus,
respectively.

238
ECSSEHB3101Part14A
5December2011
7.1.1.1 Steady slow flow
Equations [74] and [75] can be simplified when / t= 0 and both the superfluid and normal
velocitiesaresosmallthatitisjustifiedtoneglectquadraticterms.
InthatcaseEq.[74]reducestoLondonsequation,

1
p sT U 0
[717]

Equation[75]forthenormalcomponentofvelocitycanbewrittendownas,

1 s
p sT U n u n
n n [718]

CombinationofEqs.[717]and[718]yields
1 n
p U u n
[719]

which describes the creeping laminar motion (Stokes flow) of a viscous liquid (density , dynamic
viscosityn)duetoanappliedpressuregradientandtoapotentialbodyforcefield.
Equations[717]to[718]intermsof,andEq.[78]become,respectively

u U 0 [720]


sT U n u n
n n [721]


and
n
s T u n
[722]

239
ECSSEHB3101Part14A
5December2011
7.1.1.2 Several thermal effects
Themostsignificantphysicalresultswhichcanbededucedfromthelastparagraphs(clause7.1.1and
clause7.1.1.1)are:
1. TheheattransportthroughHeIIisaccomplishedbymotionofthenormalfluid.Eq.[7

14] shows that the vector u n u s points away from the heat source. When there is no
netmassflow of He II(j =0), the warm normalfluidgoesaway from the heat source,
andthecoldsuperfluidtowardtheheatsource.ThismechanismofheattransferinHe
II, called counterflow heat exchange, is far more effective than is heat transfer by
conduction.
2. AccordingtoEq.[717],leavingasidethebodyforces(U=0),apressuregradientwould
resultinanequalsigntemperaturegradient,andconversely.
(a) Letusconsiderapressuregradientthroughasuperleakfromavessel.Asuperleak
isacapillarycontainingjewelersrougeorsomeporousmaterialforpreventingthe
passageofthenormalfluid.
The temperature gradient appears because the superfluid leaving the vessel
transfersnoheatand,thence,thethermalenergyremaininginthevesselbecomes
distributedoverasmallerquantityofHeII(thespecificthermalenergyand,thus,
thetemperaturethatincreases).ThisisthesocalledMechanocaloriceffect.
(b) If a temperature gradient is established by heating the fluid in the vessel, the
normalfluid,impededbyviscosity,willremainmotionlesswhereasthesuperfluid
willflowupthetemperaturegradienttowardtheheatedvessel.Thiswillgiverise
toapressuregradient(Thermomechanicaleffect).
The helium fountain vividly illustrates the thermomechanical effect. When a
superleakcontainingaheaterandsubmergedinaheliumIIbath,isconnectedtoa
capillary(notasuperleak)extendingabovethebath,ajetofheliumemergesfrom
thecapillaryoncetheheaterisswitchedon.

7.1.2 Frictional effects


Frictional effects other than that for the normal fluid have been observed in experiments where
superfluidheliumflowsalongnarrowcapillaries.
TermscorrespondingtothesefrictionaleffectscanbeaddedtoEqs.[717]and[719],integratedalong
thetube,neglectingpotentialbodyforces,muchinthesamewayasBernoulliequationforidealfluids
isgeneralizedtorealfluidsbyaddingthepressuredropduetofriction(seeECSSEHB3101Part13
clause7.2).Theresultingexpressionsarestrictlyvalidonlyforsteadyflow.
Usually two types of frictional forces are introduced. They are defined as forces per unit volume
([Pa.m1]).
1. Fnrepresentstheexchangeofmomentumbetweenthenormalfluidandthewall.
2. Fsn the mutual exchange between normal and superfluid. Several authors (van der
Heijden,vanderBoog&Kramers(1974)[247])postulatetheexistenceofathirdfrictional
force, Fs, which describes (apart from a capillaryexit correction, which will be
introduced below) the loss of momentum of the superfluid not transferred, through
mutual friction, to the normal fluid. The existence of this nonzero Fs is not widely
acceptedhowever(deHaas&vanBeelen(1976)[55],p.143).

240
ECSSEHB3101Part14A
5December2011
Sinceonlytwoindependentequationsareavailable,noexperimentalconfirmationoftheexistenceof
Fscanbeobtained.Thence,differentinterpretationsofsimilarexperimentaldatacanbefoundinthe
literature.
Friction forces are related to vortex generation, motion and decay. Vortices are created in the bulk
fluidbythermalfluctuations.Wallroughnessisprobablyunimportant,atleastresultsobtainedwith
differentmetalandglasscapillariesaresufficientlysimilartosupportthisstatement.
The motion of the vortex line is governed by the velocity at its axis. The superfluid velocity results
fromthesuperpositionofthetransportflowandofthevelocityinducedbyallvorticesandimages.
Equation [720] with U = 0, integrated along the cylindrical tube of length L with mutual and
superfluidfrictioneffectsincluded,becomes,


s Fsn Fs 0 [723]
L

Similarly,fromEq.[719],
p
Fn Fs 0 [724]
L

IntheseequationsindicatesdifferencebetweentwosectionsofthetubeseparatedbyadistanceL,
Fsnisthemutualfrictionforce,FsthesuperfluidfrictionforceandFnthenormalfrictionforce,perunit
volumeinanycase.
Equations[723]and[724]withexperimentallydeterminedvaluesof (=(1/)psT)and p
allowthedeterminationoftworelationsbetweenthethreeforces.
Asystematicsetofexperimentswithsuperfluidheliumflowthroughcapillaries,wherevsandvnwere
independently controlled, was performed by a group of the Kamerlingh Onnes Laboratory, Leiden
University, the Netherlands. The experimental layout basically consists of a closed circuit partially
filledwithliquidhelium,thecapillarybeingapartofthecircuit(Figure72).

241
ECSSEHB3101Part14A
5December2011

Figure72:SchematicoftheapparatususedbytheLeidengrouptoproduce
heliumflowthroughcapillarieswithindependentvariationofsuperfluidand
normalvelocities.a)FromVanderHeijden,VanderBoog&Kramers(1974)[247].
b)FromDeHaas&VanBeelen(1976)[55].

Aheaterplacedupstreamthecapillaryproducesanormaltransportvelocity,vn,throughit.Themass
flux, svs + nvn, can be independently controlled in two alternative ways. Van der Heijden, van der
Boog & Kramers (1974) [247], and IJsselstein, de Goeje & Kramers (1979) [98] used a second heater
whichevaporatestheliquidheliuminthetubedownstreamthecapillary,andaheatexchangerwhere
helium gas condenses restarting the circuit, Figure 72a De Haas & van Beelen (1976) [55], and
Slegtenhorst & van Beelen (1981) [224] used two plungers which move in opposite directions at the
samespeedwithineitherreservoirofthecircuit,Figure72b.
The following combinations of velocities can be obtained, by appropriate inputs to the heaters or
displacementoftheplungers.
(a) Constantnormalvelocity,vn,andvaryingsuperfluidvelocity,vs.
(b) Constantvelocityratio,vn/vs.
(c) Constantmassflux,svs+nvn.
T(aswellasthemeantemperature)ismeasured. isrelatedtothehydrostaticpressuredifference
between the helium in both reservoirs at known temperatures Eq. [720]. p is deduced from these
valuesbyuseofEq.[78],neglectingthevelocitydifferenceterm.
Couplesofvs,vnvaluescorrespondingtoisochemicalpotential(=0),isothermal(T=0)andisobaric
(p=0) flows can be obtained. In addition, flows with = 0 are produced in a circuit containing a
heater, to control vn, and a porous plug (Staas, Taconis & van Alphen (1961) [227], de Haas & van
Beelen(1976)[55]).Thistypeofflowwillbeinterestinginconnectionwiththesuperfluidporousplug
(seeClause7.4).
TheresultsofexperimentsperformedbyvanderHeijdenetal.(1974)[247],willbesummarizedinthe
following.TheseauthorspostulateanFs0.

242
ECSSEHB3101Part14A
5December2011
Intheseexperiments,capillarieswithDE=294x106 m(or95x106 m),L=0,146mwereused(T=1,2
K).SimilarexperimentshavebeenreportedbydeHaasetal.(1976)[55],IJsselsteinetal.(1979)[98]
andSlegtenhorstetal.(1981)[224].
Thescarcityofresultsfromothersources,orunderwidelydifferentgeometricalconditions,impedes
theformulationofcomparisonsbasedondimensionlessparameters.
The flow is called supercritical when 0, otherwise it is called subcritical. This could be
misleadingwhenanFs0ispostulatedsince(Eq.[723]) couldbezerowhenFsnandFs,although
large,balanceeachother(seeFigure76,below).
a. Thenormalcontributiontothefrictionforcesalwaysdeceleratesthenormalflow.Thepressure
dropduetothiscontribution, pn,isdeterminedbyPoiseuilleslaw(seeECSSEHB3101Part
13clause7.2.2)whentheflowofnormalfluidislaminar

n L
pn LFn 32 vn
DE2 [725]

A kinetic energy correction originating from the exit of the capillary should be taken into
account when relating measured pressure drops to friction forces. In the presently discussed
experimentsthiscorrectionisnegligible.
Normalfluidflowresultstobelaminarinsubcriticalflowaswellasinsupercriticalflowwhen
eithervs0orvsvn=0.
VanderHeijdenetal.(1974)[247]assumethatEq.[725]isvalideveninsupercriticalflow.This
assumptionallowsthemtocalculateLFsnandLFsfromEqs.[723]and[724].
b. The superfluid contribution to the friction forces always decelerates the superfluid flow.
Isobaricflowresults,accordingtoEq.[724],whentheeffectsofnormalandsuperfluidfriction
balanceeachother.Forexample,inthepresentexperimentsthishappenswhenthesuperfluid
andnormalvelocitiesare,respectively,vs=0,044m.s1andvn=0,038m.s1.
Thepressuredropduetosuperfluidcontribution,ps,canbeexpressedas:
ps=LFs=LFssvs2,
where LFs accounts for the superfluid friction in the capillary and svs2, with 1, is the
kineticenergycorrection.
Experimentsperformedunderconstantmassfluxconditions(svs+ nvn=Const.)indicatesthe
tendenciessketchedinFigure73.Inparticular,althoughFsincreaseswithvnvsinthevicinity
ofvnvs=0,thistrendceasesnearvnvs=0,02m.s1,wherehappenstobezero.

243
ECSSEHB3101Part14A
5December2011

Figure73:Thesuperfluidfriction,LFs,vs.relativevelocity,vnvs,forvariousruns
withsvs+nvn=Const.FromvanderHejden,vanderBoog&Kramers(1974)[247].

Letvsobethevalueofvscorrespondingtovn=0.
1. For vso < 0,05 m.s1 and Res = svsDE/n < 1100, the following modified Poiseuilles law
holds,

n L
LFs 32 vs [726]
DE2

2. Forvso0,05m.s1,theBlasiusformulaapplies(seeECSSEHB3101Part13clause7.2.2.1),

1/ 4
3 n v s7
LFs 158L 5
[727]
DE

Recallthetransitionalflowinclassicalhydrodynamics(ECSSEHB3101Part13clause
7.2). Here the normal fluid remains laminar, whereas thesuperfluid becomes turbulent.
DeHaas&vanBeelen(1976)[55],however,suggestthatintheBlasiusbranchthewhole
fluidshouldbehaveasan)ordinaryviscousliquid(seealsoOlijhoeketal.(1967)).
c. Themutualfrictioncontributiontothefrictionforcesdeceleratesthesuperfluidaslongasvs<vn
andconversely.
The mutual friction term appears in Figure 74 as a function of vnvs. According to Gorter &
Mellink(1949)[77],Fsnshouldbehaveas|vsvn|n,withn 3,butthefigureindicatesthatthe
exponentniscertainlydifferentfrom3exceptforlargevaluesoftherelativevelocity|vsvn|
(whereFs,ifexistent,wouldbenegligiblecomparedtoFsn,seeFigure75.Wewillcomeback
tothispointverysoon).Nearvnvs=0therelationshipappearstobelinearandexhibitsaslope
depending on vso. Notice the circles in the abscissae axis (zero mutual friction) which
correspond to subcritical strict counterflow. Other circles correspond to highvelocity
(supercritical)strictcounterflow,whereGorter&Mellinkformulaapproximatelyholds.

244
ECSSEHB3101Part14A
5December2011

Figure74:Themutualfriction,LFsn,vs.relativevelocity,vnvs,fromvarious
constantmassfluxruns.FromvanderHejden,vanderBoog&Kramers(1974)
[247].

Figure75:Mutualfrictiontosuperfluidfrictionratio,Fsn/Fs,vs.relativevelocity,
vnvs,fromvariousrunswithvs0andvn0.FromvanderHeijden,vanderBoog
&Kramers(1974)[247].

Figure75indicatesthatasurprisinglysimpledependencybetweenFsn/Fsandvsvnexistsforthe
resultsdeducedfrommanydifferentruns.Similarresults,withT=1,35K,havebeenobtained.
Thesignificanceoftheseresults,basedonasuperfluidfrictionforce,Fs,themereexistenceof
whichisopentoquestion,isbynomeansclear.

245
ECSSEHB3101Part14A
5December2011
IthasbeenalreadyindicatedthatthepartofFigure74whichcorrespondstosufficientlylarge
valuesof|vnvs|isfairlyinsensitivetotheassumptiononFs(beitzeroornot).Asanexample,
thecirclesinFigure74andFigure75(correspondingtovn=0)allowustoestimateLFsnand
LFs,respectively,versusvs.ThekineticenergycorrectioncanbealsoestimatedbyuseofTable
813,clause8.2,sincethetemperatureisgiven.Atvs=0,16m.s1,Fs/Fsnislessthan0,1although
thisratioincreasesforsmallervaluesofvs.Then,thevalueof sinEq.[723]is,intheregion
oflarge|vnvs|,fairlyinsensitivetoFs.ThedepartureofFsnfromtheGorter&Mellinkformula
withn=3,mentionedinconnectionwithFigure74,couldthenbeduetotheassumedexistence
ofFs.
d. Isothermal and iso chemicalpotential curves. Figure 76 is a summary plot drawn with the
values of vs and vn at which either T or become zero. Shaded area, with its ill defined
boundary,correspondstotrulysubcriticalflow(=0).

Figure76:Isothermalandisochemicalpotentialflowsinthevn,vsplane.The
shadedregioncorrespondstosubcriticalflow(=0).FromvanderHeijden,van
derBoog&Kramers(1974)[247].

The curves are for the wide capillary experiments and seem to be temperature independent.
Resultsforthenarrowcapillaryareslightlydiferent.
Anextrapolationofthedatafor T=0suggeststhattheisothermintersectsthevn=0axisata
criticalvelocityvsc=0,01m.s1.Forthenarrowcapillary,vsc=0,03m.s1.

7.1.3 Counterflow heat transfer


WewillconsiderinthefollowingthecounterflowheattransferthroughHeIIfillinglongcylindrical
tubes of circular cross section. The data could be applied to other channels of simple geometry by
assumingthattheDEintheseequationsrepresentsthehydraulicdiameter(fourtimescrosssectional
areadividedbycrosssectionalperimeter).Nevertheless,seriouserrorsintheestimationofthecritical
velocitiescouldappearforcomplex,multiplyconnectedchannels(Chase(1963)[43]).
Gravitationalorotherbodyforcesareneglectedthroughoutthisparagraph.

246
ECSSEHB3101Part14A
5December2011
Physically,counterflowcorrespondstomotionofwarmnormalfluidawayfromtheheatsourceand
countermotion of cold superfluid toward the heat source. Although both components can be
independently controlled, as said in clause 7.1.2, we will restrict the discussion here to the case in
whichthenetmassfluxiszero(svs+nvn=0).
Threedifferentsteadystateflowregimesappearsequentiallyincounterflowasthevelocitiesvsand
|vsvn| increase (see the review by Arp (1970) [10] and the more recent papers by Childers& Tough
(1976)[44]andLadner&Tough(1979)[128]).Transitionsbetweentheseregimesseemtobeassociated
to the onset of superfluid vorticity (first critical velocity) and to the onset of turbulence in both
componentsoftheHeIIflow(secondcriticalvelocity).
1. Regimeoflowvelocities.Aslongasthevelocitiesvsandvnaresmall,theequationsgiven
inclause7.1.1and7.1.1.1(withU=0)arevalid.Inthecaseofslowflowthroughlong
cylindricaltubesofequivalentdiameterDE,PoiseuilleslawEq.[725]holds.
Assuming counterflow heat exchange, combination of Eqs. [716] withm = 0, and [725]
yields

dp q
32 n 2 [728]
dx sTDE

if,inaddition,Eq.[717]withU=0istakenintoaccount,

dT q
32 2 2n 2 [729]
dx s TDE

It can be deduced from Eq. [729], and from the dependence of n, and s on T, that
q/(dT/dx)variesapproximatelyasT12DE2.
2. Intermediate flow regimes. As vs or |vsvn| increase, quantized vortices appear in the
superfluid because the irrotational superfluid flow pattern becomes energetically
unfavorable.
Thedefinitionofthesuperfluidcriticalvelocity,vsc1,hasbeenthesubjectofcontroversies
inthepast.Evenitsdeterminationonthebasisofexperimentaldatahasnotbeenwithout
ambiguities(seeKeller(1969)[119],p.289andff).
(a) Earlycalculations(Feynman(1955)[69])equatedthekineticenergyoftheflowto
the energy of a quantized vortex formed in a flow passage of diameter DE,
obtainingthefollowingexpressionforthesuperfluidcriticalvelocity,vsc1:

h 1 D
v sc1 ln E [730]
mHe DE 2ao

histhePlancksconstant(2h=(6,625170,00023)x1024J.s),mHetheatomicmassof
helium(mHe=4,0026x(1,659790,00004)x1027kg),andaoanempiricalvortexcore
radius(ao1010m).
(b) Van Alphen et al. (1969) [246], summarizing data from different sources, which
cover seven orders of magnitude in DE, and which were obtained by use of

247
ECSSEHB3101Part14A
5December2011
different experimental techniques, arrived at the following empirical expression
(Leidencriticalvelocity):

vsc1 3,162 10 3 DE1/ 4 [731]

wherevsc1ismeasuredinm.s1andDEinm.AlthoughthedatasupportingEq.[7
31]werealltakenintheneighbourhoodofT=1,4K,VanAlphenetal.pointedout
that this critical velocity is temperature independent in the temperature region
lowerthan5x102Kbelowthepoint.
The differences between the values predicted by Eqs. [730] and [731] are
enormous,especiallyforlargetubes(Figure77).Thereaderinterestedinfurther
detailsconcerningthispointcouldconsultKeller(1969)[119]orArp(1970)[10].In
anycase,itshouldbesaidthatmuchworkneedstobedonebeforepredictingwith
confidencetheonsetofsuperfluidvorticity.

Figure77:Correlationsbetweenthecriticalsuperfluidvelocity,vsc1,andthetube
diameter,DE.Theexperimentaldatahavebeenreplottedbythecompileraftervan
Alphenetal.(1969)[246].Theycorrespondtowidelydifferentflowconditions.*
ClowandReppy,TT50x103K. Fokkens,filmflow. Pellman,superfluid
windtunnel. Chase,heatconductionTT;vn0. VanAlphen,adiabatic
flowrate. VanAlphen,energydissipationtechnique. Kramers,secondsound
attenuationinpuresuperfluidflow. VanAlphen,criticalflowthroughjewellers
rouge. KellerandHammel,isothermalflow. DatafromreviewsofAtkins,and
HammelandKeller.

248
ECSSEHB3101Part14A
5December2011
Althoughitshouldbeemphasizedthatvsc1isnotthatvalueofvsatwhichdeparturesfromthelinear
behaviorappear(Keller(1969)[119],p.289),thevalidityoftheslowflowassumption(clause7.1.1.1)
intheregimeoflowvelocitiescanbeassessedatthisstageoncetheupperboundofvs,andthusofvn,
areknownbymeansofanorderofmagnitudeanalysisofEqs.[74]and[75].Themomentumterms
intheseequationsarenegligiblecomparedtothepressuregradienttermsprovidedthat

v s2 vn2 / L 32 n vn / DE2 ,
where L is the total length of the tube, and Eq. [725] has been used to estimate dp/dx. The above
inequalitiesyieldthefollowingtwoconditions:

vn DE DE
1. 32
n L
2. snottoosmallcomparedton.
vnDE/nisinthisregime,atmost,oforder50,whereasDE/Lcouldbeoftheorderof103
or even smaller; thence, the first condition is fulfilled through the whole region of low
velocities.Thesecondconditionceasestobetrueintheproximityofthepoint.
Nosimpleanalysisoftheflowstructureintheintermediaterangeisavailable.Childers&
Tough (1976) [44] and Ladner & Tough (1979) [128] experimentally investigated the
pressureandtemperaturedropsassociatedwiththethermalcounterflowofHeIIinlong
capillaries.Theymonitoredcontinuouslythepressureandtemperaturedifferencesalong
the tube as the heat flow increased from zero. For low values of q straigth lines
correspondingtoEqs.[728]and[729],respectively,aregenerated(Figure78).

Figure78:Schematicofpressureandtemperaturedropdataasafunctionofheat
flux.

Atagivenql(whichwillcorrespondtovsc1)dp/dxjumpstoaslightlylargervalue,while
dT/dxjumpsmoremarkedly.Thejumpscanbetriggeredbymechanicalvibrationofthe
cryostat or by small but abrupt changes in q. If at this time q is decreased, a hysteresis
loopthroughpointsq2andq3appearsasinFigure78.Theregionbetweenq2andq3(but
notthatpastq3)canbedescribedintermsofamutualfrictionforce(Vinen(1957)[255]).
According to Vinen, mutual friction affects the heat flow through the scattering of the
normal fluid thermal excitation by a tangled mass of quantized vortex lines in the

249
ECSSEHB3101Part14A
5December2011
counterflowingsuperfluid.Thebasicvariableinthetheoryisthelengthperunitvolume,
Lv,ofvortexline.Lv[m2]dependson|vnvs|,TandDE.
Themutualfrictionforceperunitvolume,Fsn,isrelatedtotheforceperunitlength,f,on
asinglevortexlineby
2
Fsn fLv ,
3
wherethefactor2/3takesintoaccountthatonlytheaxialcomponentoffcontributesto
Fsn. For the normal fluid at rest, the force per unit length, f, on a vortex line moving at
velocityvtwillbe
1 s n
f B vt ,
2
directedoppositetovt.Bisadimensionlessparameteroforderonerelatedtothevortex
linerotonscatteringcrosssection. isthestrengthofaquantizedvortexring(=h/mHe,
Eq. (7.14)), vt is the relative velocity between the normal fluid and the vortex lines. For
counterflowVinenassumedvtvs.
OnceFsnisknown,combinationofEqs.[723]and[724],thelastwithFs=0,yields(/L
=d/dx)

dT Fsn 1 dp

dx s s s dx [732]

When Fsn = 0 the relationship between dT/dx and dp/dx for the regime of low velocities
(Eqs.[728]and[729])isrestored.Intheregimeofourpresentconcernthenormalfluid
stillflowslaminarly,thusthesecondtermintherighthandsideofEq.[732]isthepartof
thetemperaturedropwhichcorrespondstotheirrotationalsuperfluidflow,andtherest
givestheadditionaldropwhichwewillcalldT/dx.
The evolution of the length per unit volume, Lv, of vortex line is controlled by two
mutually competing although independent mechanisms: vortex generation and decay.
Bothconcernaninfinitemediumwithnorestrictingboundaries,buttheeffectofthewall
canbetakenintoaccountbyassumingthatitinterfereswiththegenerationmechanism
within a distance = /Lv1/2, being a dimensionless constant, of order one, on which
reststheexistenceofacriticalvelocity.
Figure79 showsa typical sketch ofLvforsteady state (generation balancingdecay). In
additiontoLv=0anoseshapedcurveappearsinthefigure,indicatingthatLvisadouble
valuedfunctionofvsprovidedthatvsDEexceedssomecriticalvalue,whichisgivenby

250
ECSSEHB3101Part14A
5December2011

Figure79:SchematicofLv1/2DEvs.vsDEundersteadystateconditions.From
Childers&Tough(1976)[44].


4 2
vsc DE
n 1 B
,
where l and 2 are two more dimensionless, temperaturedependent, parameters of
orderonewhichaccountforthegeometricalstructureofthevorticity. lisageneration
and2adecayparameter.
Only in the upper part of the curve in Figure 79 Lv increases with vs, consequently the
onlystableallowedvaluesofLvareLv=0andthosedefinedbytheupperpart.
Figure 78 can now be understood in terms of Vinens theory. Increasing the heat flux
fromzero,theregimeinwhichthenormalfluidislaminarandthesuperfluidirrotational
iscrossedover.Whenqisincreasedslowlybeyondq2,curveLv=0wouldcorrespondtoa
metastable equilibrium and a sudden jump to the configuration defined by the upper
partofthecurvewouldresult.Oncevorticityhasbeengenerated,theonlywaytoregain
theconditionLv=0istoreducetheheatfluxbelowq2.
A theory based on so many experimental parameters (l, 2,,B) will inevitably agree
withtheexperimentaldata.Nevertheless,asshownbyChilders&Tough(1976)[44],and
Ladner & Tough (1979) [128], once two combinations of these parameters are deduced
fromexperimentallydeterminedvscanddT/dxforagiventubeatasingletemperature,
the functional form of the mutual friction region is given at all temperatures.
Furthermore,thetemperaturedependenceofl/2,whichinVinensmodelisdetermined
fromexperiments,canbecalculatedbyuseofatheoryduetoSchwarz(1978)[210].The
agreementofthistheorywithvaluesof l/2obtainedbyLadner&Toughinlongglass
tubesofrectangularcrosssectionisfair.
Vinens model does not provide a picture of the onset of vorticity, since the generation
termresultstobezerowhenLv=0.AccordingtoSchwarz(1978)[210],twotypesofonset
behavior are possible. In narrow channels, the rate at which the microscopic vortex
fluctuationsofacriticalsizearethermallynucleatedinthefluidseemstobecontrolling.
Inwidechannelsoneobservesanintermittentbehaviorsimilartothatinclassicalfluids.

251
ECSSEHB3101Part14A
5December2011
3. Regime of large velocities. Mutual friction between both components controls the flow
structureinthisregimewheretheGorter&Mellinkexpressionholds(seeclause7.1.2)

Fsn A s n vn vs
n
[733]

with n 3. In this equation A [m4n.sn2.kg1] is the Gorter & Mellink constant, a slowly
varyingfunctionoftemperatureand(perhaps)ofgeometry.
Assuming that equation [733] is valid in this regime, combination of Eqs. [714] (with
crosssectionalaveragevalues),[732]and[733]yields

n
dT dT 1 dp q
A n
dx dx s dx s s sT [734]

where dT/dx concerns the temperature gradient resulting from mutual friction. Notice
thatthediameter,DE,doesnotappearinEq.[734].
Thephenomenologicalequation[733]alsoapplieswhenvsandvn,althoughpointingin
opposite directions are not related to each other through svs + nvn = 0. Other
mechanismsseemtoberesponsibleforthenonlinearmutualfrictionintubeswithvnvs
orvsvn.
Thesecondcriticalvelocity,boundingfrombelowthisregimeoflargevelocitiescanbe
estimated from experimental results. A critical Reynolds number, Rec, based on the
normalflowvelocity,hasbeenexperimentallyverifiedinsomeexperimentsalthoughit
doesnotappearclearlyinothers(Ladner&Tough(1979)[128]).TherangeofRec(103to3
x 103) is similar to that corresponding to transition to turbulence in classical fluid flow,
althoughacleardependenceontemperaturecanbeobservedinthepresentcase.
Figure710,fromArp(1970)[10],givesthecriticalReynoldsnumbervs.temperaturefor
counterflowheatexchange,asdeducedfrom

vnc 2 DE D
Re E q
n n sT [735]

252
ECSSEHB3101Part14A
5December2011

Figure710:CriticalReynoldsnumberforcounterflowheatexchange,Rec,asa
functionoftemperature,T.FromArp(1970)[10].

The values of vnc2 in the above expression are those found experimentally by Staas,
Taconis&vanAlphen(1961)[227].
Thesameauthorsalsoobservedthattherelationshipbetweenthepressuregradientand
the normal flow velocity is, in this regime, that corresponding to a classical turbulent
flow(seeclause7.1.2andalsoECSSEHB3101Part13clause7.2).
The validity of Figure 710 has been disputed by Slegtenhorst (1981a),who argues that
sincetheflowmesurementsofStaasetal.correspondtoflowwith =0,theycannotbe
automaticallyappliedtocounterflow.
Insum,theexistenceoftwocriticalfluxesisbynomeansclear.Insomecasesqclandqc2
arefairlysharplydefined,whileinothercasestheyapparentlytendtomerge,oroneof
themtendstosmearout.Itappearsthattheexperimentalresolutioninmostoftheearly
experiments on thermal counterflow was such that the entire upper portion of the
hysteresisloopinFigure78appearedasasinglecontinuouscurvewheredT/dxqnwith
n3.
4. SummaryoftheresultsconcerningconterflowheatexchangeinHeII.PlotsofdT/dxvs.q
fortemperaturesintherange1,5K2KaregiveninFigure711.Thefiguresaresimilar
to those by Arp (1970) [10], although they cover a slightly different range of tube
diameters.

253
ECSSEHB3101Part14A
5December2011

Figure711:Diagramswhichrelatethethermalgradient,dT/dx,totheheatflux,q,
incounterflowheatexchange.T=1,5Kto2K.CalculatedbythecompilerafterArp
(1970)[10].

254
ECSSEHB3101Part14A
5December2011
Thecurveslabelledwithtubediameters,DE,correspondtotheregimeoflowvelocitiesandhavebeen
deduced from Eq. [729]. The K, which is also used to identify these lines, is the permeability or
geometricfactorofanequivalentporousmedium(seeclause7.4.2.4).

ThecriticalvaluesofqhavebeendeducedfromEq.[715]with u n calculatedasfollows:

vsclfromEq.[730]withao=1010mandvnclrelatedtovsclthroughtheconditionj=0
vsclfromEq.[731]andvnclasabove;
vnc2fromEq.[735]andFigure710.
The last curve will be used to estimate the heat transfer rate in the intermediate region. Finally, the
solidlinehasbeendeducedfromEq.[734].
The behavior of dT/dx vs. q in the intermediate regime has been conjectured by Arp (1970) [10] as
follows.
1. The temperature gradient required to transfer a given rate, once vscl has been just
exceeded,willbeslightlylargerthanthatpredictedbymeansofEq.[729].
2. Whenthenormalflowvelocityincreasesuptovnc2,dT/dxvs.qpredictedbyEq.[729]is
onetotwoordersofmagnitudelessthanpredictedbyEq.[734].Therealcurveshould
besomewherebetweenthesetwo,andnottooclosetotheupperone.
3. Once Rec has been exceeded, turbulence becomes fully developed as velocities further
increasebyafactoroftwoorless,thenEq.[734]applies.
DottedlinesinFigure711havebeensketchedwiththeseargumentsinmind.
Equation [716], with m = 0, indicates that large values of q can be obtained, without vn exceeding
vnc,when p and, thence, is large enough. That is why pressurized superfluid helium constitutes a
veryappropriatecoolant.
ExperimentaldataforcounterflowheatexchangeinHeIIatatmosphericpressurehavebeenreported
byBonMardion,Claudet&Seyfert(1979)[26].TheirresultsaresummarizedinFigure712,where

X T LT q 3, 4 [736]

L(T)[m]beingthedistancebetweenthechannelsectionattemperatureTandthewarmendatT.q
[W.m2] is the wall mean heat flux between both sections.The curve in Figure 712 could be used in
severalways.

255
ECSSEHB3101Part14A
5December2011

Figure712:TemperatureprofilealongachannelfilledwithHeIIatatmospheric
pressureinconterflowheatexchange.FromBonMardion,Claudet&Seyfert
(1979)[26].

1. Whenq=1W.m2,X(T)isthelengthLgivingthepositionofthecrosssectionofthetube
atwhichthetemperatureisT,measuredfromthatwhereT=T.InthissensetheFigure
712representsthetemperatureprofilealongthetube.
2. Forgivenqandendtemperatures,TiandTo,therequiredchannellengthwillbe

L X Ti X To q 3, 4 [737]

Equation[737]togetherwithFigure712willalso:
3. ProvideqwhenL,TiandToaregiven.
4. ProvideX(To)whenL,qandX(Ti)areknown.
Therangeofparametersexploredexperimentallywas
1,4KTT

256
ECSSEHB3101Part14A
5December2011
2103DE5
1102L100
TheuncertaintyinherentinthecurveofFigure712isoftheorderof3%.

7.1.4 Heat transfer at arbitrary combinations of vn and vs


Among the many experiments performed in wide channels those done on the simple counterflow
configuration have been the most extensive and, to some extent, those giving the least ambiguous
results.
Counterflow could not be representative of the general behavior of He II when both transport
velocitiesareindependentlycontrolled.Thistypeofexperimentsarenoteasilyfoundintheliterature
however.Olijhoeketal.(1973,1974)[173]&[174]reportedresultsoftestswhichcanbeanalyzedin
termsoftheGorter&Mellinkfriction(Eq.[733])althoughvnandvsareindependentfromeachother.
Strictcounterflowseemstoappearevenincomplicatedgeometrieswherevnandvsshouldvaryinan
independentfashion.Inanexperimentalconfiguration(Slegtenhorst(1981b)[223])aclosedcircuitis
formedbytwoidenticalglasscapillaries(1and2)connectedinseriesviaasuperleak(throughwhich
=0)TheremainingpartofthecircuitisaHebaththetemperatureofwhichiskeptconstant.Two
independently controlled heaters at either side of the superleak are used to adjust the normal
velocitiesthrougheithercapillary.
The results from this experiment can be predicted from the flow characteristic of the individual
capillaries.AlinearincreaseofdT2/dxisobservedwhenvn2isslowlyincreasedfromzero,forconstant
vn1.ThevalueofdTl/dx,onitsturn,remainstotallyunaffectedbytheincreasingofvn2.Athighervalues
of vn2 this linear flow state appears to become metastable (recall the discussion in clause 7.1.3, in
connection with Figure 78). At an unpredictable high value ofvn2 a sudden jump in T2 occurs,
accompainedbyasimilarchangeinTl.Thetransitioncanbeiniciatedatsmallervaluesofvn2bythe
applicationofamechanicalshock,butonceithasoccurred,areturntothelinearflowstatethrough
variationofvn2isnotpossible.
Experimental evidence indicates that in the laminar region independent strict counterflows are
establishedinbothcapillariesatleastforhighvaluesofvn.Thiscontradictstheprincipleofconstancy
of the circulation in the superfluid (see clause 7.1.1), as we will see. Mutual friction, after starting,
createsthecirculationwhichispresentintheloopduringtheexperiment.
If the superfluid motion starts from rest (and no mutual friction exists), the circulation in the
superfluidshouldbezeroatanyinstant.Thatis,
vs1vs2=0,
where contributions to circulation other than through the capillaries (which are very slender) have
been neglected. The sign of the circulation is different in both capillaries. On the other hand, mass
conservationinthelooprequires

n vn1 s vs1 n vn 2 s vs 2 2 s vs1 n vn1 vn 2 0 .


Then,conservationofcirculationwouldyield

n
vs1 vs 2 vn1 vn 2
2 s

,
whichdoesnotcorrespondtostrictcounterflowinbothcapillaries(vs1=nvn1/s,vs2=nvn2/s).

257
ECSSEHB3101Part14A
5December2011

7.1.5 Vapor formation


VaporformationwithintheHeIIbulkcouldbeofconcernwhenthepressureandtemperaturewithin
the liquid are close to the saturated liquid line in the absence of thermal flux. This situation could
occur,undernormalgravityconditions,inanopensystemwheretheliquidheliumisinequilibrium
withitsvaporattheliquidsurface.
Local boiling in counterflow heat exchange of He II in a nozzle has been analyzed by Broadwell &
Liepmann(1969)[30].Theirnozzlegeometryinsuredthat|vnvs|wasonlylargenearthethroatwhere
Gorter&Mellinkregimeprevailed.Thecaseofacylindricalchannelunderturbulentconditionshas
beenconsideredbyArp(1970)[10].Asomewhatdifferentsituation,involvinglowheatfluxes,which
couldbesuitabletosuperfluidporousplugs(seeClause7.4),isconsideredhere.
InFigure713HeIIisinequilibriumwithitsvaporataliquidsurfaceplacedabovethetopendofa
slender cylindrical vertical tube. It is assumed that heat is added only to the liquid within the tube
eitherfromthetoporfromthebottom.

Figure713:TubeandHeIIbatharrangement.

Assumingcounterflowheatexchangeatlowvelocities(clause7.1.3)thedistributionofpressureand
temperaturewithinthetubecanbeexpressed,afterEqs.[715],[728]and[729],asfollows:

32q
x
p x p1 g 1 x 2 n dx
DE 0 sT [738]


x
T x T1 2 2 n2 dx
32q
[739]
DE 0 s T
Thesignbeforetheintegralwillbeusedwhentheheatsourceistopplacedinthecylindricaltube.
If (dp/dT)sat is the slope of the vaporliquid equilibrium line, the power input, q*, necessary for local
boilingatlevelx*canbededucedfrom

32q * n
x*
dx g 1 x *
dp DE2 0 sT
[740]
dT sat 32q * n
x*

2
dx
DE 0 s T
2 2

258
ECSSEHB3101Part14A
5December2011
Thisequationisonlyvalidwhenboilingoccurswithinthetube(x*0).
Whentheheatsourceistopplacedandthetemperaturedifferencesaresosmallthattheintroduction
of a mean value of T to evaluate the integrals in Eq. [740] is justified, the following simplified
expressionresults.

x*
1
sTD 2
1 l
q*
dp / dT sat g
E

32 n x*
1 [741]
s l

Usually(dp/dT)sat/s0,1andx*/l<<1,thustheaboveexpressioncanbefurthersimplified.
Equation[741]furnishesanestimateofthefractionx*/l,oftubelengthwhichisfilledwithliquidHe
IIforagivenvalueoftheheatflux,q*,undertheassumptionofonedimensionalflow(planarvapor
liquidinterface).Inrealpracticemattersconcerningboilinginveryslendertubesarenotsosimple,as
wewillseeinthenextparagraph.Seealsoclause7.4.2.4.

7.1.6 Superfluid Helium film


SolidsurfacesincontactwithHeIIarecoveredbyathinfilmofliquid.
Itispresumedthatthisthinfilmwillbethecriticalthermallinkbetweentheheatsource,whichisthe
devicetobecooled,andtheheatsink,whichistheboiloffvapor.Thiswillbeparticularlysoduring
the latter stages of the depletion of the stored Helium II. Thus, the calculation of the heat transfer
throughthisfilm,asafunctionoffilmthickness,wouldbecrucialtoaproperdesign.
Variousdeterminationsofthefilmthicknessforsaturatedlayers(layersinequilibriumwiththebulk
liquid),atrestundernormalgravityconditionshavebeenperformed.Forexample,inatubepartially
filledwithHeIIthethicknessat102mheightabovethemeniscusisbetween3and3,5x108m,and
decreasesasthepower(0,4)oftheheight(Mendelssohn(1960)[148]).
Kontorovich (1956)[125]showed thatwhen asuperfluid filmundergoes potential flow its thickness
shoulddecrease.LetusconsidertheconfigurationshowninFigure714,whereafilmsteadilyflows
downaverticalwalltowardareservoir.

Figure714:Filmandbulkliquidconfiguration

259
ECSSEHB3101Part14A
5December2011
CombinationofEqs.[75],with/t=0,and[78]yields:

1 1 n 2
u s2 U p, T u n u s 0
2 2 [742]


U gz
y3

,
being a constant which specifies the Van der Waals interaction with the wall, and (p,T) the
chemicalpotential,perunitmassofstationaryfluid.
Sincethefilmisextremelythin(seebelow):

1. Wecanneglectthemotionofthenormalcomponent( u n =0).

2. According to the local continuity equation, usy/usz/L, and L being the characteristic
lengthsparalleltotheaxesyandzrespectively(inparticular,isthefilmthickness),and

usy,uszarethecorrespondingcomponentsofthevector u s .Thus,usy(/L)usz<<usz.

3. Sincethemotionofthesuperfluidcomponentisirrotationalthroughoutthefluidregion,
whichincludesapartatrest(seeclause7.1.1),
u sz u sy

y z

andthenthechangewithyofuszisnegligiblecomparedtouszitself,
yusz(/L)usy(/L)2usz<<usz,
thenceweinferthatusz=vs(z)dependsonlyonz.


IntegratingEq.[742],with un =0,overthefilmsurface

S 2
vs gz 3 po , To
2 [743]

where po and To are the pressure and temperature of the reservoir. When the flow is isothermal,
accordingtoEq.[78],

p po
p, T po , To

260
ECSSEHB3101Part14A
5December2011
Neglectingchangesinhydrostaticpressureoftheheliumvapor,pporesultstobethepressurejump
throughthefilmsurfaceatheightz.If,inaddition,thecapillarypressureisnegligible(curvatureof
thefilmsurfacesmallenough),ppo=0and,thence,Eq.[743]becomes:

s 2
vs g 3 0 [744]
2

ThisequationdiffersfromtheclassicalBernoulliequationfornormalliquidsinthe s/factor,which
isduetothefactthatinthiscaseonlythesuperfluidmoves.Ontheotherhand,Eq.[744]indicates
thatshoulddecreasewhenvsincreases(Bernoullithinning).
Now,letmbethemassoffluidtransportedbythefilminunittimeacrossasectionofunitwidth.

m s vs [745]

CombinationofEqs.[744]and[745]yieldsthefollowingexpressiontocalculate:

~ ~
z 3 1 0
~ [746]

where
~
z z / zc , / c ,and
~


m6
zc
8 2 3 s3 g
2 s
and c
m2
arecharacteristiclengthsintheverticalandhorizontaldirection,respectively.
~
z vs.
~


,asgivenbyEq.[746],isrepresentedinFigure715asafullline.

261
ECSSEHB3101Part14A
5December2011

Figure715:Bernoullithinning.ThefulllinecorrespondstoEq.[746].Thedotted
lineistheKontorovich(1956)[125]solution.Neithersolutiongivesthecorrect
transitionofthefilminterfacetothehorizontalfreesurfaceinthereservoir,
becausecapillarypressurehasbeenneglected.Curveslabelledwiththevaluesof
BocorrespondtoEq.[749].

Kontorovich(1956)[125]givesforvs.mtheexpression


1/ 3
m2
1/ 3

1 [747]
gz 6 s gz

whichcanbededucedbyapproximatesolutionofEq.[746]when
~


issmallenough.IndimensionlesscoordinatesEq.[747]becomes

~ 1 1
~ 1 / 3 1 ~ 1 / 3 [748]
z 3z

262
ECSSEHB3101Part14A
5December2011
Equation [748] has been represented in Figure 715 by means of a dotted line. It deviates from the
exactsolutionofEq.[746]nearz=0,andapparentlydeviatesinthewrongdirection(thefreesurface
ofthefilmshouldvarycontinuouslywithz).
Although the corner near the free surface of the reservoir requires a more complete treatment than
above, since most of the already introduced simplifying assumptions become invalid in that region,
themereconsiderationofthecapillarypressureyieldsarealisticshapeofthefreesurfacethroughout.
Whencapillarypressureistakenintoaccount,Eq.[746]becomes

~ ~
~ ~ 3 d 2 / d~
z2
z 3 1
~
Bo ~ 2 3/ 2
d [749]
2 ~
dz

whereBo=gc2/isaBondnumber(seeclause6.4.7.1),whichdependsonthetemperature,themass
flowrateandthegravitylevel.=zc/cisanaspectratiowhich,inadditiontotemperature,strongly
dependsonthemassflowrate.Thisdifferentialequationisintegratedwiththefollowingboundary
conditions:
1. Farfromthereservoir,
~ ~
z , ~z 3 1 0
~

.
2. Atthereservoirfreesurface,
~
z 0 , d / d~
~ z

.
Assuming that Bo is large enough, the right hand side of Eq. [749] is negligible except near
~ ~
z 1 0
~

,
where the curvature of the free surface becomes very large. An approximate solution of Eq. [749],
validwhenBoislarge,canbeobtainedbyuseofthemethodofmatchedasymptoticexpansions(see
VanDyke(1975)[248]).
Letusassume1.Introducinginnervariables,whichareoforderunityintheregionunderstudy

X Bo ~
z

,
~

Y Bo 1
,
Eq.[749]becomes,afterneglectinghigherorderterms,

263
ECSSEHB3101Part14A
5December2011

d 2Y / dX 2
X Y
1 dY / dX

2 3/ 2
[750]

withtheboundaryconditions:

X , X Y 0

,
X 0 , dY / dX

.
This is the wellknown problem of finding a curve the curvature of which is proportional to the
distancetoastraightline,

X Y 0

.
The solution, with our particular boundary conditions, is too complicated (and useless) to be given
here,butitisrepresentedbythecurveslabelledwiththevaluesofBoinFigure715.
Summingup,Kontorovichtheoryseemstocorrectlypredictthethinningofasuperfluidfilmflowing
stationary,butEq.[747]cannotbeusedforaquantitativeassessmentoftheeffect.
Saturatedfilmsofheliumaretypically100atomiclayersthick(1atomiclayer=3,6A=3,6x1010m)
andcanflowwithvelocitiesupto0,5m.s1.Theexpectedchangeinthicknessforsuchafilmcouldbe
about20atomiclayersinsomecases.
ThefirstattempttoexperimentallydetecttheBernoullithinningwasmadebyKeller(1970)[120].In
hisexperimentsthesaturatedfilmofHeIIwascausedtoflow,subcritically,betweentheplatesofa
capacitor.Measuredcapacitanceshouldvaryproportionallytothethicknessofthefilm.Nochangein
filmthicknesswasobservedwithin5Aforallflowvelocitiesuptoabout0,5m.s1.
Many authors followed Keller in the confirmation or disproving of Bernoulli thinning. A list of
attemptsisgiveninTable71.Thelistisbynomeanscompleteandshouldnotbeconsideredasapoll
todiscoverthemostwidelyacceptedopinion.SeealsoareviewbyDeBruynOuboter(1973)[53]of
theworkperformedupto1972.
Up to the moment no thinning has been observed in experiments performed under truly steady
conditions.RecallthatEq.[742]hasbeenobtainedbyassuming/t=0.
VanSpronsenetal.(1973,1974)arguethatcondensationofthevaportendstorestoretheoriginalfilm
thickness.Intheirsetofexperiments,whereheliumfilmcoverstheinsideofaverylongandnarrow
capillarywoundintoaspiral,condensationisavoidedbecausethefilm,whichoscillatesbetweenboth
endreservoirsofthetube,drainsawaythecondensateswhichcannotbeaccumulated.
ExperimentsundernonsteadyconditionshavebeenperformedalsobyWang&Petrac(1975)[258].
These authors, however recognize that, since the characteristic length of their experimental cell (L =
0,025 m) is much smaller than the wavelength of surface waves (third sound, 2,4 m), nonsteady
effectsintroducedbyanysurfacedisturbanceshouldbetakenintoaccount.Table71.

264
ECSSEHB3101Part14A
5December2011

Table71:SeveralAttemptstoExperimentallyVerifytheBernoulliThinning(BT)
Author Keller(1970) VanSpronsen Telschowetal. Wang&Petrac
etal.(1973) (1975)[234] (1975)[258]
[251]

Aimofthe BTdetectionin Avoidanceof Useofmore Nonsteadyand


experiment steadyflow. condensationby sensitive liquidvapor
oscillatingflow. techniques. instabilityeffects.

Temp.[K] 1,11,9 1,21,8 1,51,8 1,12,0

Drivingforce Hydrostatic Hydrostatic Thermal. Thermal.


pressure. pressure.

Typeofflow Subcritical. Critical. Subcritical. Subcritical.

Max.Velocity[m.s1] 0,5 0,3 1. 0,16

Typicallayer 400. 150. 40. 1400.


thickness[A]

Characteristiclength 0,08 200. 0,15 0,025


ofthecell[m]

Characteristiclength 2x105
oftheflow[m] (seetext)

Frequency[Hz] 0 0,002 0 0,025

Measurement Capacitive Liquidlevelin Dopplershiftof Capacitive.


technique theend surfacewaves.
reservoirs.

Accuracy[A] 5. 0,1 0,07 1.

Thinningdetected No Yes No Independentof


BT.

Integrating Eq. [74] along each streamline, without neglecting nonsteady effects, and pursuing the
samereasoningasabovetoarriveatEq.[744],thefollowingequationresults:


x


t 0
u dl s vs2 gz 3 0
2 [751]

where

dl

isthevectorelementofstreamline.

265
ECSSEHB3101Part14A
5December2011
Quasisteady experiments will be performed under conditions such that nonsteady terms be
negligiblecomparedtothekineticenergyterm,i.e.:

Leff s
vs
t 2

,
Leffandtbeing,respectively,acharacteristiclengthandacharacteristictimeofthefluidflow.
Although in the case of Van Spronsen et al. (1974) the characteristic time can be deduced from a
typical frequency of the oscillation, which they quote, the characteristic length is not so clearly
defined.
The oscillation of the film (Atkins oscillation), the aim of which is to avoid the accumulation of
condensates, is induced by moving up and down, opposite to each other, both end reservoirs. In
addition to the characteristic frequency of the oscillation, a characteristic length appears in the
definitionoftheinertialmassoftheoscillatingsystem.
x
A
Leff dx
0
s As

,
whereListhelengthofthecapillary,AthecrosssectionalareaoftheendreservoirsandAsthecross
sectionalareaoftheliquidwithinthetube(theshapeofthecrosssectionsisannularbecausethefilm
isattachedtothewall).
SinceAsissmall,Leffislarge,Leff2x105m,renderingLeff/tthousandtimeslargerthansvs/2.
EvenassumingthatthecharacteristiclengthisL,thetimederivativetermbecomesofthesameorder
asthekineticenergyterm.
Thence, the thinning observed in this experiment could be, partially or totally, due to other than
Bernoullieffect.
Experiments to measure, inter alia, the film thickness under reduced gravity conditions have been
performed(Masonetal.(1976)[141],Yang&Mason(1980)[268]).
These experiments were undertaken to check Eq. [744] which indicates that the thickness of
stationaryfilms(vs=0)shouldvaryas(gz)1/3.
Thefirstexperiments(March1976)wereperformedusinganaircraftwhichprovided45periodsof20
sreducedgravity.
After1976aBlackBrantsoundingrocketwasused.Itprovidedafreefallperiodofabout5minof106
glevel.Theexperiments,togetherwithpreparatorylaboratorytestsonEarth,havebeenreportedby
Yang&Mason(1980)[268].
The apparatus used to measure the film thickness was a quartz crystal microbalance (QCM). If a
properly cut crystal oscillating in a shear mode is loaded with a liquid film on its surface, its
fundamental resonance frequency shifts according to the layer thickness. In this way thicknesses of
filmsseveralhundredsAngstromthickcanbemeasuredwithlessthan2%error.
Thesupportingsurfaceswereorientedoneparalleltothethrustaxisandtwoperpendicularly,within
acellmaintainedatslightlybelowsaturationpressure(Figure716).

266
ECSSEHB3101Part14A
5December2011

Figure716:Cellusedtoperformreducedgravitytest.Thefilmthickness
experimentswereperformedinthelefthandsidecompartment.FromYang&
Mason(1980)[268].

A crucial, and only partially solved problem was keeping the film thickness in the chamber against
temperature variations in the range 1,5 K to 2 K. To this aim, an helium reservoir packed with
aluminiumpowderwasconnectedtothefilmexperimentchamber.Thepowderpackageprovidesan
additionallargesurfaceareaandactsasacapillaryretentiondevicewhichsuppliesheliumtothecell
when the temperature rises, absorbing helium when it decreases. This device, although successful,
greatlycomplicatedtheinterpretationoftheexperimentresults.
Themainconclusionswerethefollowing:
1. The reduced gravity behavior of the helium can be understood in terms of laboratory
observations.
2. Film thicknesses are independent of the position in the chamber and are thicker than
thoseinthesamechamberundernormalgravitybyafactoroftheorderof2.Thickening
of the film, nevertheless, could be partly due to the release of liquid by the powder
package.
Asaconsequenceoffilmthickeningunderreducedgravity,thecriticalvelocitiesandvanderWaals
forcesarereducedand,thence,thefilmsbecamemoresensitivetoheating.
Anothersetofexperimentsonthepropertiesofsuperfluidheliuminzerog(SHFE)hasbeenselected
by NASA for performance on board Spacelab 2. The experimental set includes: measurement of the
fluidthermalanddynamicbehaviorpropertiesofquantizedsurfacewavesinthesuperfluidfilm,and
performanceofaDewarinspace.Afairlydetaileddescriptionofthesystembeingdevelopedforthis
setofexperimentsismadebyUrbanetal.(1978)[244].

7.2 Kapitza conductance


Experiments measuring the thermal conductivity of liquid helium at low temperatures should take
intoaccountthethermalboundaryconductance,hk,firstnoticedinexperimentswithHe4byKapitza
in1941.

267
ECSSEHB3101Part14A
5December2011
Formally,hkisdefinedas:
Q/ A
hk lim
T 0 T

,
whereQ/AistheheatflowperunitareaacrosstheliquidsolidinterfaceofareaA.
hkrangesbetween1W.m2.K1and50x103W.m2.K1at1,9Kforthevarioussolidsinvestigated(1,9K
isanoptimumtemperatureforheattransferthroughthebulkfluid,itis,forexample,theoperating
temperatureofseveralsuperconductingparticleaccelerators).At1,9K,usingthehighestobservedhk
of50x103W.m2.K1onefindsthatthelengthofheliumIIwhichwouldpresentanequivalentthermal
barriertoasmallheatfluxisabout3000km(Snyder(1970)[225]).ThusKapitzaconductancecontrols
the heat transfer from a solid body to helium II at temperatures near the point. At lower
temperaturesofabout0,5K,whereheliumisapoorthermalconductor,thereversemaybetrue.
Kapitza conductance is not associated with the peculiarities of superfluid helium as was shown
experimentallywithHe3abovethetransitiontemperature(see,f.e.,Keller(1969)[119]pp.192ff).
Theheattransferbetweenliquidheliumandasolidispoorlyunderstood.Fromtheexperimentalside,
Kapitza conductance has been measured for only about twodozens solids (no data on structural
materials such as stainless steels or titanium have been found), and order of magnitude differences
have been reported for samples of the same solid. The most apparent explanation is difference in
samplepreparation,whichcanbeduetooneormoreparametersofthefollowinglist(Mittag(1973)
[153]): material impurity density, dislocation density, surface damage, stress, grain size of crystals,
surfaceroughness,impurityandoxidelayersonthesurface.
Thetheoreticalpictureisnotbetter(seethereviewarticlebySnyder(1970)[225]).Heattransferacross
the interfaceis described in terms of the transmission and reflection of acoustic waves (phonons) at
the interface. The Phonon Radiation Limit appears to furnish an upper limit for the Kapitza
conductance but the actual mechanisms of energy exchange at the interface, which determine the
lowerlimitsfortheconductances,arepoorlyunderstood.
Most theoretical work is based in the phonon transmission mechanism put forth by Khalatnikov
(AcousticMismatchTheory)whosuggestedthatbecauseofthemismatchoftheacousticimpedance
between a solid and a liquid, only a small fraction of the available phonons should be transmitted.
Theoreticalexpressionspredictconductances2or3ordersofmagnitudebelowthephononradiation
limit,andabout1to2ordersofmagnitudebelowthevaluesobservedexperimentally(thefiguresare
not so bad for nonmetals). The acoustic mismatch, however, predicts a T3 temperature dependence
fortheKapitzaconductancewhichisapproximatelyobeyedforavarietyofsolids,althoughtheactual
exponentofTrangesfromn=2,3,ton=4,2(smallervalueshavebeenreported,seeTable72toTable
741below).
ManyauthorstriedtoimproveuponKhalatnikovtheory.Forexample,Challisetal.(1961)[39]took
intoaccountthefactthatthelayerofheliumclosetothesolidhasamuchhigherdensitythanthebulk
liquid. They obtained a temperature exponent n = 4,2, and results closer to those observed
experimentallyinthetemperaturerange1,4Kto2,2K.
Athirdtheoreticalapproachintermsofathermalaccommodationcoefficient,,hasbeenknownlong
beforeKapitzaconductancewasdiscovered(seeEstermann(1955)[66]foranapplicationtorarefied
gasflowproblems).Insufficientdataareavailabletoprofitablyusethistheoryinthepresentinstance.
Summarizing,usefulvaluesofKapitzaconductancefordesignpurposescannotbeobtainedfromthe
theoreticalformulaeexistingintheliterature.
Inthefollowingasummaryofreporteddataisgiven.

268
ECSSEHB3101Part14A
5December2011
Samplecharacteristicshavebeenspecifiedasfaraspossible.Nevertheless,therelevanceofanygiven
surface parameter is uncertain because of the lack of theoretical understanding of the interface
conductanceproblem.
Materials in Table 72 to Table 741 have been arranged by ascending Debye temperatures, D.The
Debyecharacteristictemperatureisdefinedas:

h m
D
k
,
wherehisthePlancksconstant,kistheBoltzmannsconstant.k=1,380x1023J.K1and mis2times
theupperfrequencyofthecrystal.
Tablesof DforbothnonmetalsandmetalscanbefoundinZemansky(1968)[272]pp.318and327
respectively.
ThereasonforintroducingtheDebyetemperatureisthat,accordingtoChallis(1962)[39],hkvariesas
D1whenonlythehighesthkvaluesreportedforeachsolidareconsidered.
Anattempthasbeenmadetoencloseinthetablesinformationregardingthemeasurementtechniques
usedtogetthedata.
Basically,thetechniquesconsistinproducinga(small)heatflowacrossthesampletoliquidinterface.
Theresultingtemperaturedifferencebetweenthesampleandtheliquidismeasuredwhenthesystem
reachesthermalequilibrium.Itiscrucialtoavoidheattransferpathsotherthanthroughtheinterface.
Inordertoachievethisahardvacuumismaintainedinsidethecanwhichenclosestheexperimental
cell.
Themaindistinctivefeaturesoftheseveraltechniquesare:theshapeandsizeofthesolid,thehelium
bath in direct contact with the solid, the geometry of the interface with a sharp outline of its
boundaries,attachmentofthesample,thevacuumenvironment,heatingproceduresandtemperature
measurementinbothsolidandliquid.
Thesetofsketchesshownbelowwillbereferredlaterinthetable.Mainadvantagesanddrawbacksof
thedifferenttechniquesarebrieflyindicated.

7.2.1 Measuring methods


A.LongSample
Thesampleismachinedtotheshapeofatopheadedcylinder.Thedischeadingthecylinderissoft
solderedintoathinsheetformingthebaseofasmallheliumcryostat.
Sampleandcryostatareinavacuumandsurroundedbyalowtemperaturevessel.
Heatisappliedtothelowerendofthesamplethroughspoolheaters.
Temperaturesaremeasuredattwopointsalongthesampleandatasinglepointintheheliumbath.
The temperature field in the solid near the interface is certainly not onedimensional and in several
instancesappropriatecorrectionsshouldbeintroduced.
A1.ItisquitesimilartoAalthoughtheinterfaceareaisbetterdefined.
A2. The sample supported between two brass flanges by four stainless steel screws. Stainless steel
springs keep the pressure on an indium 0ring seal between the sample and the top brass flange.
Sampleandflangesarehungfromathinwalledstainlesssteeltubethetopendofwhichisopento
theliquidheliumbath.

269
ECSSEHB3101Part14A
5December2011
Temperatureismeasuredinfourlocations,twointhesampleandtwointheliquidheliumwithinthe
tube.
Heatisappliedthroughastripheaterepoxiedtothelowerendofthesample.
Heatconductioninparallelwiththesampleisnegligible.
A3.Thetubecontainingtheliquidheliumisfastenedtothevacuumchamberwhereasthesampleis
pressedagainstthetubebymeansofaspring.
A4. The metal is cast onto the top of a precoated sapphire rod resulting in a binary sample of
high/lowacousticimpedancematerials.ThetopofthesampleformsthelowerbaseoftheHeIIbath.
Temperaturesaremeasuredatsixpointsinthesampleplusoneadditionalpointintheheliumbath.
Heatisappliedtothelowerpartofthesample.

B.ShortSampleandTube
Athinwalledstainlesssteeltubeissoldered(orglued)tothesampleandattachedtothebottomofa
copper container which is evacuated later on. Several such containers are submerged in a helium II
bath.
Temperatures are measured with thermometers wound on the sample surface. Bath temperature is
closelyregulatedandmeasuredbyathermometer.
Vacuumfaceofthespecimenislocallyheated.
B1.Thesampleissealedtoaglasstube.Sampleandtubearesurroundedbyaglassvacuumjacket.
Stressesintheglassarerelievedbyannealing.
B2.Themercury,liquidatroomtemperature,fillsanylontube,thelowerendofwhichisforcedonto
a cupronickel sleeve. A brass plunger, tipped with a Teflon cap, forms a seal impervious to liquid
mercurywhichisremovedafterthemercuryisfrozen.
Temperatures are measured in four points along the mercury column by use the thermometers
connectedtothemercurybyplatinumwiresthroughholesinthenylontube.
Aheaterisplacedonthetopofthetube.Theheaterensuresthatthesamplewouldfreezefromthe
bottomduringcooldown,andprovidestherequiredheatfluxduringtest.
Heat transfer through the cupronickel sleeve should be accounted for in the normal state tests,
althoughitisquitenegligibleinthesuperconductingstate.

270
ECSSEHB3101Part14A
5December2011
C.LongCylinder
A thinwalled stainless steel tube, which contains the He II, is terminated at one end by the sample
and at the other end by a surface held at the bath temperature and by the helium feeding line. The
sampleissealedtothetubewithanindiumwire.
A coil former, bolted to the stainless steel tube fulfils the twofold role of fixing the sample and of
supportingaheatercoilmountedonthetopofthesample.
TheassemblyisenclosedinanevacuatedcoppercanwithintheHe4bath(fortemperaturesabove1
K).
Temperatures are measured at two points along the stainless steel tube and a single point in the
sample. A cavity is machined in the sample to place a thermometer as close to the interface as
possible.
Differential stresses, partially taken up by the indium seal, could result from cooling to low
temperatures.
Thismethodhasbeenusedabove0,3K(withanappropriateHe3refrigerator).
C1.DiffersfromCinthesamplemounting.Herethecoilformerandthesamplearemadeofthesame
material,sothatnostressesarisethroughcoolingtolowtemperatures.
Theinterfaceareaisnorclearlydefined.

D.Sandwich(ParallelPlateCell)
Thelowacousticimpedancemediumiscastbetweentwometallicplatesfurnishedwithappropriate
insulatingspacers.
Oneoftheplatesisheated,whereastheoppositeremainsinthermalcontactwithaheliumbath.
Temperaturesaremeasuredattheendplates.
D1. When the low acoustic impedance medium freezes at low temperatures, two access ports are
provided to the lower plate. Once the space between plates has been filled through one port, the
secondissealedoff.
Bothplatescanbeindependentlyheated.
Thecellisscrewedtoacopperflangeinthermalcontactwiththeheliumbath.

271
ECSSEHB3101Part14A
5December2011
E.HollowRod
Thesampleismachinedtotheshapeofathickwalledcylinder.
Twochannelsaredrilledoutinthewalltoaccommodatethethermometers.Theheaterisplacedinthe
centralcavity.
Theaspectratiooftheconfigurationislessthan10.
Cavitiesarefilledwithheliumgasatatmosphericpressure.
Theassemblyisloweredintoaheliumbaththetemperatureofwhichiscloselymonitored.
F.ResistanceThermometer
Avarnishcoatedconstantanstrip(theheater)isfastenedtoaphenolicsubstrate.
Theconstantanstripiscoatedinitsturnbyagraphitelayer(thesampleandresistivecomponentof
the thermometer). Potential taps are connected to the graphite layer by means of a silver paint. The
assemblyismountedinaHeIIcryostat.
Thethermalconductivityofthegraphiteshouldbeaccountedforintheheatbalance.

7.2.2 Experimental data


Figures of Kapitza conductance, hk, vs. temperature, T, summarizing the data, are shown in the
following.Figure717concernslowDebyetemperaturemetalscontactingliquidhelium.Figure718
corresponds to copper contacting low acoustic impedance materials (acoustic impedance is density
timessoundvelocity).Figure719isforstructuralmetalsandFigure720fornonmetals,inbothcases
contactingliquidhelium.
Moredetaileddataaregiveninthesubsequenttables.Theyarearrangedasfollows:Metallicmaterials
in contact with liquid helium (Table 72), Nonmetals in contact with liquid helium (Table 732) and
Metalsincontactwithlowacousticimpedancemedia(Table741).

272
ECSSEHB3101Part14A
5December2011

Figure717:Kapitzaconductance,hk,oflowDebyetemperaturemetals,Mercury,
Lead,GoldandSilverincontactwithLiquidHelium,vs.temperature,T.SeeTable
72below.

273
ECSSEHB3101Part14A
5December2011

Figure718:Kapitzaconductance,hk,ofCopperincontactwithvariouslow
acousticimpedancematerialsvs.temperature,T.SeeTable72andTable741
below.Theoreticalresultsarealsoshowninthisfigure.

274
ECSSEHB3101Part14A
5December2011

Figure719:Kapitzaconductance,hk,ofTungsten,Aluminium,Molybdenumand
Beryllium,incontactwithLiquidHelium,vs.temperature,T.SeeTable72below.

Figure720:Kapitzaconductance,hk,ofNonmetalsincontactwithLiquidhelium
vs.temperatureT.SeeTable732below.

275
ECSSEHB3101Part14A
5December2011

Table72:KapitzaConductance,hk,ofMetalsinContactwithLiquidHelium(HeII)
SolidD[K] References

Mercury72,seeTable73 Neeper,Pearce&Wasilik(1967)[164]

<99,995%pure Challis(1962)[39]

99,995%pure
Lead105,see
Table75 <99,9995%pure

Nodetailsgiven. Challis,Dransfeld&Wilks(1961)[40]

99,9997%pure WeyYeng(1962)[263]

Indium108,seeTable77 Neeper&Dillinger(1964)

Gold164,seeTable79 Johnson&Little(1963)[108]

Tin199,seeTable711 WeyYeng(1962)[263]

Platinum240,seeTable713

Silver225,seeTable715 Alnaimi&vanderSluijs(1974)[4]

Palladium274,seeTable717 Reivari(1969)[192]

Niobium275,seeTable718 Mittag(1973)[153]

99,999%pure Jones&vanderSluijs(1973)[111]

99,9%pure WeyYeng(1962)[263]

OFHC(oxygenfreehigh Johnson&Little(1963)[108]
Copper343, conductivity) Mittag(1973)[153]
seeTable720
Highpurity(nofurther Challis,Dransfeld&Wilks(1961)[40]
detailsgiven)

99,999%pure Challis(1962)[39]
Patullo&vanderSluijs(1983)[178]

Tungsten400,seeTable722 Johnson&Little(1963)[108]

Aluminium428,seeTable724 VanSciver(1978)[249]
Mittag(1973)[153]

Nickel450,seeTable726 WeyYeng(1962)[263]

Molybdenum450,seeTable728 Alnaimi&vanderSluijs(1975)[3]

Beryllium1440,seeTable730

276
ECSSEHB3101Part14A
5December2011

Table73:KapitzaConductance,hk,ofMercury72inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

17000 3,6 SurfaceAx104[m2]:0,4



99,999%pure Measur.Method:A4 Table74
2

20000 3,5 Temp.Range[K]:1,12,1

Table74:SampleDescriptionofMercury72inTable73
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Mercurywasheldbyathin Thecannisterwasplaced 75x104dia. Ca.


wallednylontube,thelower intotheDewarand 5x102height Superconducting
17000 endofwhichwasforcedona pressurereducedto state
cupronickelsleevecapped 2,7x104Pa.Mercury
withasealimperviousto solidifiedfromthebottom
liquidmercury.Thesleeve upward.Oncethecapwas Ca.Normalstate
waswetwithmercuryusing removedthemercury
nitricacidasaflux.The contactedtheliquid
20000 lowerbaseoftheassembly helium.Copper
wassoftsolderedintothe contaminatesthelower
bottomofastainlesssteel partofthemercury
cannister. column.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

277
ECSSEHB3101Part14A
5December2011

Table75:KapitzaConductance,hk,ofLead105inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

<99,995%pure 6600 2,9 Measur.Method:A



Temp.Range[K]:1,12,1
11000 2,6

99,995%pure 4200 4,1


11000 3,1

46000 4,7

32000 3,8

9300 2,0

6500 2,9

<99,9995%pure

4900 2,8
Table76
2

5700 3,0

4500 3,2

4400 3,4

Nodetailsgiven. 15000 3,7 Temp.Range[K]:12,17


2900 30,1 Measur.Method:B


Temp.Range[K]:1,152,16

1900 3,20,1

99,9997%pure 7300 30,1

5600 30,1

7100 2,90,1

278
ECSSEHB3101Part14A
5December2011

Table76:SampleDescriptionofLead105inTable75
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Machinedtosize.Annealed Duringtest,surfacesofthe Ca.


inavacuumfor16hat543K. samplenotcontactingHeII
6600
Normalstate(longitudinal wereinavacuum
fieldof6,4x104A.m1). surroundedbyavesselat
eitherheliumorhydrogen
11000 Sameasabove. temperatures. Ca.

Sameasabove. Ca.Uncorrected
Superconductingstate. forthe
4200 temperature
dropnearthe
interface.

Sameasabove. Ca.
11000 Normalstate(longitudinal
fieldof6,4x104A.m1)

46000 Sameasabove. Ca.

Sameasabove. Ca.
32000 Testmadejustafter Extrapolated
preparation. fromT1,7K.

Sameasabove. Duringtest,surfacesofthe Ca.


9300 Testmade2,3monthsafter samplenotcontactingHeII Extrapolated
preparation. wereinavacuum fromT1,8K.
surroundedbyavesselat
Sameasabove. eitherheliumorhydrogen Ca.
6500 Superconductingstate.2,3 temperatures.
monthsafterpreparation.
Storageopentolaboratory
Sameasabove. atmosphere. Onesingle

2,4monthsafterpreparation. value.

Sameasabove. Ca.
4900 Normalstate(6,4x104A.m1)
18monthsafterpreparation.

Sameasabove. Ca.
5700
21monthsafterpreparation.

Sameasabove. Ca.
4500 Superconductingstate.18
monthsafterpreparation.

Sameasabove. Ca.
4400
21mothsafterpreparation.

279
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Etchedfor3minwithdilute Exposedtoairforafew Ca.


nitricacid. minutesafteretching.
Spaceabovethesurface
15000
wasevacuatedatroom
temperatureandflushed
withcleanheliumgas.

Machinedtosize.Cleaned Thecellcontainingthe 14x103dia. 1stTest


2900
withalcoholtetrachloride. samplewasevacuatedand 2,8x103thick
Mounted.1sttest.Aftertwo placedintoaheliumbath
2ndTest
monthssurfacecleanedinthe (togetherwithseveralother
liquidheliumbath.2ndtest. cells).
Sevendaysatroom Nodetailsgivenonstorage 3rdTest
1900
temperature,cleanedinthe betweentests.
heliumbathsothatfurther
cleaningnolongerchanged 4thTest
7300 hk.3rdtest.Afterfourmonths,
4thtest.

Machinedandcleanedas Thecellcontainingthe 9,2x103dia. 1stTest


5600
above.Mounted.1sttest. samplewasevacuatedand 5,5x103height
Electropolishedremovinga placedintoaheliumbath
7100 layerof34x106m.Washed, (togetherwithseveralother 2ndTest
driedandmounted.2ndtest. cells).
Electropolishedremoving Nodetailsgivenonstorage 3rdTest
102x106m.Sevendaysat betweentests.
roomtemperature.3rdtest.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table77:KapitzaConductance,hk,ofIndium108inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:0,32
Spectroscopically 11000 3
Measur.Method:B Table78
2

pure(99,999%) (2800) (2.7)


Temp.Range[K]:1,02,1

280
ECSSEHB3101Part14A
5December2011

Table78:SampleDescriptionofIndium108inTable77

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

11000 Polishedfaceofasapphire Assemblywithmold Indium: Secondvalue


(2800) substratewasindiumcoated returnedtovacuumand 6,35x103dia. obtainedaftera
inaheliumatmosphere. annealedfortwohoursat 38x103height fewweeks,in
Assemblyplacedinaclose 420K. Sapphire: either(n)or(s)
fittingmoldandbackedina Moldcutaway. 6,35x10 dia. states.
3

vacuumof1,3x103Pa. 32x103height
Additionalindiumwascast
andzonerefinedinthemold.
Freeendfaceofindiumwas
machined.

Table79:KapitzaConductance,hk,ofGold164inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

8800 SurfaceAx104[m2]:1,5

99,99%pure 3 Measur.Method:A3 Table710
HeatFluxQ/A[W.m2]:20490
2

8300
Temp.Range[K]:1,252,10

Table710:SampleDescriptionofGold164inTable79
Storage,
hkat1,9K Sample Sample
Environmentbefore Comments
[W.m2.K1] Conditions Dimensions[m]
Cooldown

8800 2x102dia. Datapointscorrespondto


1,27x104thick twosimilarsamples.

Goldisquiteinertand
8300
formsnostableoxides.

281
ECSSEHB3101Part14A
5December2011

Table711:KapitzaConductance,hk,ofTin199inContactwithLiquidHelium(He
II)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

5200 2,90,1 Measur.Method:B


99,997%pure Temp.Range[K]:1,152,16 Table712
2

5200 2,90,1

Table712:SampleDescriptionofTin199inTable711
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Machinedtosize.Cleanedwith Thecellcontainingthe 10,2x103dia. Bothtests


alcoholandcarbon samplewasevacuatedand 3
12,5x10 furnishedthe
5200 tetrachloride.Mounted.1sttest. placedintoaheliumbath sameresults.
height
Electropolishedremovinga (togetherwithseveralother Transitioninto
layerof2,5x106mthick.2nd cells).Nodetailsonstorage thenormalstate
test. betweentests. (2,4x104A.m1
field)hadno
effectonhk.The
5200
samewith
rotatinghelium
upto400rpm.

Table713:KapitzaConductance,hk,ofPlatinum240inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

Measur.Method:B1
99,9%pure 2000 2,30,1 Table714
2

Temp.Range[K]:1,152,16

282
ECSSEHB3101Part14A
5December2011

Table714:SampleDescriptionofPlatinum240inTable713
Sample
hkat1,9K Storage,Environmentbefore
SampleConditions Dimensions Comments
[W.m2.K1] Cooldown
[m]

Fromafoil.Cleanedwith Sampleandtubewere 7,3x103dia.


alcoholandcarbon surroundedbyaglassvacuum 4
10 thick
2000
tetrachloride.Sealedtoaglass jet.Annealedinanovenat773
tube. Kfor3h.Mounting.

Table715:KapitzaConductance,hk,ofSilver225inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

6100 3,4 Measur.Method:C



Temp.Range[K]:12
6500 3,3

4300 2,1

4900 3,2

99,992%pure 1500 0,80 Table716


2

1900 1,2

1500 0,30

1100 0,70

1000 1,2

283
ECSSEHB3101Part14A
5December2011

Table716:SampleDescriptionofSilver225inTable715
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

6100 Annealedfor8hat900K Undervacuumof1,3x105 Sample1.Test1.Ca


andcooleddownslowly. Pafromfirstmountingto
6500 Electropolishedusingthe endoflasttest.An Sample1.Test2.Ca
followingelectrolytic exceptionwasmade
4300 Sample1.Test3.Ca
solution:35gofsilver betweentests4and5of
cyanide,37gofpotasium sample1,wherethevalve
4900 Sample1.Test4.Ca
cyanideand39gof totheatmospherewas
potasiumcarbonatein1lof openedforashorttime Sample1.Test5.Ca.Hk
1500 water.Currentdensitywas beforevacuumwas decreaseswhenthe
140A.m2.Alayer104m restored. sampleisexposedtoair.
thickwasremoved
1900 Sample1.Test6.Ca

1500 Sample1.Test7.Ca

1100 Sample2.Test1.Ca

1000 Sample2.Test2.Ca
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table717:KapitzaConductance,hk,ofPalladium274inContactwithLiquid
Helium(HeII)
hkat1,9K Temp.
2 1 Others Comments
[W.m .K ] Exponentn

Measur.Method:C Estimatedinthecourseofanexperimenttomeasure
100at1K 3(assumed) Temp.Range[K]:0,1 MssbauerspectraofFe57withCo56sourceina
1 palladiummatrix

284
ECSSEHB3101Part14A
5December2011

Table718:KapitzaConductance,hk,ofNiobium275inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

99,9%pure. Measur.Method:A3
Lessthan500 1800300 3,620,32 HeatFluxQ/A[W.m2]:10400
ppmTa Temp.Range[K]:1,32,1 Table719
2

100ppmTa 4000500 4,650,28

Table719:SampleDescriptionofNiobium275inTable718
Storage, Sample
hkat1,9K
SampleConditions Environmentbefore Dimensions Comments
[W.m2.K1]
Cooldown [m]

Machinedwith0,3x103mcutdepth. 2,5x102dia. Superconducting


1
Fivemonthsinair.Rinsedwith 2,5x102height state(0Am
1800300
toluene,acetoneandethanol.r= field).
1,0030,0022.

Chemicallypolished3minina Vacuumtreatment
solutionof60x103lnitricacid(70%) consistedofheatingfor
and40x103lhydrofluoricacid(40%), 18hinacontainerat
273K293K.Highvacuumandhigh 463Kwithanend
4000500 temperatureannealed.Aftervacuum vacuum105Pa.
treatment,softsolderdepositswere Remainedinthe
removed.Chemicallypolishedfor6 container,under
min,rinsedwithethanol.r=1,0007 vacuum,for6dat293
0,0004. K.Inair15min.

285
ECSSEHB3101Part14A
5December2011

Table720:KapitzaConductance,hk,ofCopper343inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

99,999%pure 8400 6,4 Measur.Method:C,C1 Table721


2

HeatFluxQ/A[W.m2]:16,23,32,50,92
3800 4,3 Temp.Range[K]:1,22,0

22000 4,5

3900 1,6

12500 3,9

7500 3,2

17000 3,5

15000 3,7

22000 4,3

22000 4,3

1100 2,4

1400 3,8

590 1,6

500 1,6

1200 4,2

890 3,0

960 3,0

13000 3,1

7000 6,1

5100 6,6

6100 6,0

2300 2,0

2800 4,2

19000 3,0

5300 5,9

286
ECSSEHB3101Part14A
5December2011

Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

29000 3,7

2700 3,2

16000 2,5

9000 2,2

3100 3,2

Measur.Method:E
99,9%pure 25500 2,60,1
Temp.Range[K]:0,5702,075

6600 3,1 SurfaceAx104[m2]:1,5



Measur.Method:A3
6600 3,1 Temp.Range[K]:1,252,10

OFHC(oxygen 6200 3
freehigh
conductivity) 7600 2,6

7300700 3,340,27 Measur.Method:A3


HeatFluxQ/A[W.m2]:10to400
6700600 4,110,27 Temp.Range[K]:1,32,1

Highpurity(no 2700 2,4 Measur.Method:A



furtherdetails Temp.Range[K]:1,42,17
given) 5500 2,8

99,999%pure 4100 2,6 Measur.Method:A



Temp.Range[K]:1,12,1

3600 2,6

380at1K 3,9 SurfaceAx104[m2]:0,159



Measur.Method:C
HeatFluxQ/A[W.m2]:0,03to0,2at0,35
1000at1K 4,3 K2to25at1,3K
Temp.Range[K]:0,31,3

287
ECSSEHB3101Part14A
5December2011

Table721:SampleDescriptionofCopper343inTable720
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

8400 Machinedtosize. Preservedundera 6x103dia. 0toolangle


Remachinedusingthesame vacuumofabout 2x103height
3800 3 0toolangle
tools,positionedinthesame 1,3x10 Paandexposed
wayat2150revolutionsper totheatmosphere
22000 0toolangle
minute,making4cutsof duringabout20minfor
50,8x106mdeptheach. mounting.
3900 0toolangle

12500 12toolangle

7500 Machinedasabove.Hand
polishedwithincreasingly
17000 finepolishingpaperupto
thehighestgradeavailable.
15000
Electropolishedin30%
orthophosporicacid
22000
solutionsfortimesvarying
22000 between2minand53h.
Washedwithdeionized
wateranddriedunder 1,5x106m
1100 vacuum. electropolishing
depth

1,5x106m
1400 electropolishing
depth

40x106m
590 electropolishing
depth

80x106m
500 electropolishing
depth

120x106m
1200 electropolishing
depth

120x106m
890 electropolishing
depth

Electropolishedasabove 120x106me.d.+
960
with105mthicklayer etched
removed,thensilvercoated
underavacuumof1,3x105 3,5x109msilver
13000 coatingthickness.

288
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]
Paupto60s. Batch2

4,0x109msilver
7000 coatingthickness.
Batch2

5,0x109msilver
5100 coatingthickness.
Batch2

7,0x109msilver
6100 coatingthickness.
Batch1

7,0x109msilver
2300 coatingthickness.
Batch1

8,0x109msilver
2800 coatingthickness.
Batch1

11,0x109msilver
19000 coatingthickness.
Batch2

11,5x109msilver
5300 coatingthickness.
Batch1

15x109msilver
29000 coatingthickness.
Batch2

18x109msilver
2700 coatingthickness.
Batch1

35x109msilver
16000 coatingthickness.
Batch2

78x109msilver
9000 coatingthickness.
Batch2

90x109msilver
3100 coatingthickness.
Batch1

289
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Machinedtosize. Surfacenotincontact Thickwalled Additional


withHeIIwereina vessel, experiments
heliumgasatmosphere 5x103mi.d., indicatedthatthe
25500
atnormalpressure. 50x103mlong pressure
dependenceofhk
isnegligible.

Machined,chemically 3,17x102dia. Ca.


cleanedandrinsedinwater 1,46x102height
andethanol.Mounted.
0,47x104m2of1,64x104m2
6600
interfacecleanedwhile
immersedinHeII.Pieces
resultingfromcutting
remainedintheinterface.

Abovesampleonceexposed Thesystemwasnot Ca.


toatmosphereforaweek. disturbedduring
6600 The2ndtestwascarriedout exposuretoatmosphere.
withthechipsstilllyingon
theinterface.

Machined,cleanedand Ca.
4
rinsedasabove.0,85x10 m
2
6200
interfacecleanedwhile
immersed.

Abovesampleonceexposed Thesystemwasnot Ca.


7600 toatmosphereforaweek. disturbedduring
exposuretoatmosphere.

Machinedtosize,vapor Incontactwithairafter 2,5x102dia.


degreased,rinsedwith preparationforseveral 2
2,5x10 height
7300700
ethanol.R=1,270,1. hoursbeforepumping
down.

Sameasabove.Chemically Incontactwithairafter
polished1minat343K. preparationfor1hour
6700600 Electropolished6minat293 beforepumpingdown.
K.Rinsedinethanol.R=
1,00040,0003.

Polishedwithjeweller s Exposedtoairforafew Ca.Theratioofhk


2700
rouge. minutesafter underthe
preparationwhilebeing saturatedvapor
Etchedfor3minwithdilute mounted.Subsequently pressuretothat
5500 nitricacid. spaceabovethesurface under20x105Pais
wasevacuatedatroom almost

290
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]
temperatureandflushed independentof
withcleanheliumgas pressureabove1,3
beforecoolingdown. K.

Deformedbyhammering Surfacesofthesample Ca.Influenceof


4100 andthenmachined. notcontactingHeII plastic
Mechanicallypolished. wereinavacuumand deformation.
surroundedbyavessel
Abovesampleonce ateitherheliumor
removedfromthe hydrogentemperatures
3600 experimentalcell.Annealed duringtest.
invacuumat898Kfor4h
andthenlightlypolished.

Machined.Annealedunder Mountingtook1530 5,6x103dia. Ca.hk/T3showsa


vacuumat1223Kfor11h. min.Pumpedonfora 3x103height maximumatabout
Cooledover13h. totalof29dupto4x105 Holedrilled, 1K.Thisbehavior
380at1K
Handpolished. Pa.Contamination 2x103dia. seemstobe
Electropolishedremoving carefullyprevented. 2x103height characteristicof
152x106m.Washed. cleansurfaces.

Machined,annealedand Mountingasabove.
handpolishedasabove. Pumpedonforatotalof
1000at1K Electropolishedremoving 15dupto8x102Pa.
61x106m.Washed. Contaminationcarefully
prevented.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table722:KapitzaConductance,hk,ofTungsten400inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:1,5
Measur.Method:A3
99,95%pure 2420 3,5 Table723
HeatFluxQ/A[W.m2]:20to490
2

Temp.Range[K]:1,252,10

291
ECSSEHB3101Part14A
5December2011

Table723:SampleDescriptionofTungsten400inTable722
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

2420 2x102dia. Ca.


4
1,27x10 thick
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table724:KapitzaConductance,hk,ofAluminium428inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:1,48
99,99%pure Measur.Method:A2
7400300 2,8
RRR=50080 HeatFluxQ/A[W.m2]:104to2x104
Temp.Range[K]:1,52,1

6700 2,7 SurfaceAx104[m2]:1,48



Measur.Method:A2
99,999%pure 6400 2,8 HeatFluxQ/A[W.m2]:104to2x104

Temp.Range[K]:1,72,1
RRR=1500
# 6200 2,9
(RRRisthe Table725
2

residual
# 6400 2,8
resistanceratio,
4,2/300) 6700 2,8

6600 2,8

6061Alalloy 40008000 30,4 Measur.Method:A3


HeatFluxQ/A[W.m2]:10to400
Ultrapure
5200300 4,210,28 Temp.Range[K]:1,32,1
RRR=13000

292
ECSSEHB3101Part14A
5December2011

Table725:SampleDescriptionofAluminium428inTable724
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Machinedtosizeand TheentireDewar 3x102dia. Ca.Near


washedwithmethyl containingtheexperimental 11x102height saturatedvapor
7400300 alcohol.Spacingbetween cellwasevacuatedwitha pressure.Bath
machininggrooveswas liquidnitrogentrapped temperature,T=
typically105m.Groove diffusionpumptolessthan 1,85K
depthwasanorderof 1,3x102Paatroom
magnitudesmaller. temperaturebeforecool Ca.Near
down.Duringthetesta saturatedvapor
6700 vacuumof1,3x103Pawas pressure.Bath
maintainedinthevacuum, temperature,T=
container. 2,01K

Ca.Near
saturatedvapor
6400 pressure.Bath
temperature,T=
1,89K

Ca.Near
saturatedvapor
6200 pressure.Bath
temperature,T=
1,79K

Caat2,43x103Pa
pressure
6400 (saturation).Bath
temperature,T=
1,89K

Caat33,4x103Pa
pressure.Bath
6700
temperature,T=
1,89K

Caat84,1x103Pa
pressure.Bath
6600
temperature,T=
1,89K

T6temper.Machinedto Incontactwithairafter 2,5x102dia.


size.Vapordegreased, preparationfor1hbefore 2
2,5x10 height
40008000
thenrinsedwithethanol.r pumpingdown.
=1,0040,002.

Vacuummelted,annealed Incontactwithairfor1h
5200300
for1hinair,cooleddown beforepumpingdown.

293
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]
to293Kin24h.Machined.
SurfacegroundwithNo
500emerypaper,vapor
degreased,chemically
polishedfor3,5minthen
rinsedwithethanol.
r=1,00040,0004
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table726:KapitzaConductance,hk,ofNickel450inContactwithLiquidHelium
(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

4000 2,90,1 Measur.Method:B


99,99%pure Table727
2

Temp.Range[K]:1,152,16

294
ECSSEHB3101Part14A
5December2011

Table727:SampleDescriptionofNickel450inTable726
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

Machinedtosize.Cleanedwith Beforefillingwithliquid 1,4x103dia.


4000 alcoholandcarbon helium,theDewarwasfilled 2,6x103thick
tetrachloride.Mounted. withgaseoushelium.No
detailsaregivenonstorage
Machinedandcleanedas betweentests. 10,2x103dia. Nochanges
above.Electropolished 2,8x103thick wereobservedin
removing14x106m.1sttest. the2ndand3rd
Electropolishedremoving tests.Some
370
56x106m.2ndtest. capillaryeffect
Electropolishedremoving couldarisesince
140x106m.3rdtest. poreswere
observed.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table728:KapitzaConductance,hk,ofMolybdenum450inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

1400100 1,50,1 SurfaceAx104[m2]:0,126



99,9%pure Measur.Method:C Table729
HeatFluxQ/A[W.m2]:12,24
2

97080 0,80,2
Temp.Range[K]:1,21,9

295
ECSSEHB3101Part14A
5December2011

Table729:SampleDescriptionofMolybdenum450inTable728
Storage, Sample
hkat1,9K
SampleConditions Environment Dimensions Comments
[W.m2.K1]
beforeCooldown [m]

Annealedundervacuum1500 Storedundera Ca.Electropolished


1400100 Kfor14h.Electropolishedto vacuumof1,3x103 byerrorusingamild
remove104musingasolution Pa.Exposedtoair steelelectrode.
ofonepartsulfuricacidand forabout20min
sevenpartsmethanol.Stainless duringmounting. Ca.Lowtemperature
steelcathode.Currentdensity: Pumpeddown. exponent.hk/T3vs.T

97080 103A.m2.Washedwith Differentpumping seemstoexhibita


methanol. times. maximumnearT=1
K.Stainlesssteel
electrode.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table730:KapitzaConductance,hk,ofBeryllium1440inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:0,126
Measur.Method:C
99%pure 26040 1,10,5 Table731
HeatFluxQ/A[W.m2]:12,24
2

Temp.Range[K]:1,21,9

296
ECSSEHB3101Part14A
5December2011

Table731:SampleDescriptionofBeryllium1440inTable730
Storage,
Sample
hkat1,9K Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCool
[m]
down

26040 Annealedundervacuum. Ca.Low


Electropolishedtoremove104m temperature
usingasolutionof100parts exponent.No
orthophosphoricacid,3parts pressure
sulfuricacid,3partsglycerol,3 dependenceofhk/T3
partsethanol.Stainlesssteel hasbeenfounded
cathode.1,2x103A.m2.Washedwith between105Paand
methanol. 3x105Pa.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure717,Figure718andFigure719andcanbe
identifiedbythekeygiveninthepreviousentryoftheTable.

Table732:KapitzaConductance,hk,ofNonmetalsinContactwithLiquidHelium
(HeII)
SolidD[K] References

Graphite420,seeTable733 Haben&Frederking(1975)[79]

Quartz(SiO2)470,seeTable735 WeyYeng(1962)[263]
Challis,Dransfeld&Wilks(1961)[40]

Silicon636,seeTable737 Johnson&Little(1963)[108]

LithiumFluoride730,seeTable739

Table733:KapitzaConductance,hk,ofGraphite420inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:0,05
Aquadagcoating
b 1000 2,2 HeatFluxQ/A[W.m2]:5,5to350 Table734
2

Temp.Range[K]:1,42,0
b AquadagisacolloidaldispersionofgraphiteinwaterproducedbyAchesonColloidsCompany,PortHuron,
Michigan,USA.

297
ECSSEHB3101Part14A
5December2011

Table734:SampleDescriptionofGraphite420inTable733
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

1000 Avarnishcoated(GEvarnish Driedinambientairand Constantan: Ca.Thereisa


7031)constantanstrip,fastened mountedintheDewar.Aftera 3,05x102long singlevaluefrom
tothegrooveofaphenolic setofteststhesamplewas 3,84x104wide thesecondtest.
substrate,isovercoatedwith warmedupandexposedto 1,8x105thick hk=1100at
Aquadag.Potentialtapsare ambientconditions.2ndtest. Varnish:5x106 T=1,87K.
connectedtothegraphitelayer to105thick. Applicationsto
bymeansofasilverpaint. Graphite: thermometryin
20x106thick. liquidhelium.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure720andcanbeidentifiedbythekeygiveninthe
previousentryoftheTable.

Table735:KapitzaConductance,hk,ofQuartz(SiO2)470inContactwithLiquid
Helium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

Measur.Method:B
5700 3,60,2
Temp.Range[K]:1,1512,157
SingleCrystal Table736
2

SurfaceAx104[m2]:0,19
3800 3 Measur.Method:A1
Temp.Range[K]:1,42,17

Table736:SampleDescriptionofQuartz(SiO2)470inTable735
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

5700 Surfacenormaltothe Beforefillingwithliquid 12,5x103dia.


principalaxisofthe helium,theDewarwasfilled 10,9x103height
crystalwasgroundand withgaseoushelium.
polished.

3800 Surfacewasground Oncemountedinthe 0,5x102dia. Ca.Quartz


2
experimentalcell,thespace 4,75x10 height exhibitsa
abovethesurfacewas relativelyhigh
evacuatedatroom thermal

298
ECSSEHB3101Part14A
5December2011

Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]
temperatureandflushed conductivityat
severaltimeswithclean low
heliumgasbeforecooling temperatures.
down.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure720andcanbeidentifiedbythekeygiveninthe
previousentryoftheTable.

Table737:KapitzaConductance,hk,ofSilicon636inContactwithLiquidHelium
(HeII)
hkat1,9K Temp. Sample
Sample Key 2 1 Others
[W.m .K ] Exponentn Description

Singlecrystalgrownin 3700 3,2 SurfaceAx104[m2]:1,5 Table738


2

the[1,1,1]direction. Measur.Method:A3
ptyperesistivity>10 HeatFluxQ/A[W.m2]:20to490
m,impurities<1,3x109 4000 4,2 Temp.Range[K]:1,252,10
atom.m3,
dislocation:8x104m2

Sameasaboveexcept
dislocationdensity: 4200 4,2
8x106m2

Table738:SampleDescriptionofSilicon636inTable737
Sample
hkat1,9K Storage,Environment
SampleConditions Dimensions Comments
[W.m2.K1] beforeCooldown
[m]

3700 Interfacialplanenormalto 3,17x102dia. Ca.Studyofthe


thegrowthaxis,groundand 1,46x102height influenceof
polished. dislocation
densityonhk.
4000 Sameasaboveandetched.

4200 Interfacialplanenormalto
thegrowthaxis,ground,
polishedandetched.

299
ECSSEHB3101Part14A
5December2011
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure720andcanbeidentifiedbythekeygiveninthe
previousentryoftheTable.

Table739:KapitzaConductance,hk,ofLithiumFluoride730inContactwith
LiquidHelium(HeII)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

SurfaceAx104[m2]:1,5
99,99%pure Measur.Method:A3
4800 3,8 Table740
HeatFluxQ/A[W.m2]:20to490
2

singlecrystal
Temp.Range[K]:1,252,10

Table740:SampleDescriptionofLithiumFluoride730inTable739
Storage, Sample
hkat1,9K
SampleConditions Environment Dimensions Comments
[W.m2.K1]
beforeCooldown [m]

4800 Theinterfaceisan 2x102dia. Ca.Basicinterest.Noheat


2
[1,0,0]surface. 1,27x10 height transferbychargecarriers
atliquidhelium
temperature.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure720andcanbeidentifiedbythekeygiveninthe
previousentryoftheTable.

Table741:KapitzaConductance,hk,ofMetalsinContactwithLowAcoustiv
ImpedanceMedia(LAIM)
SolidD[K] References

Indium108,seeTable742 Neeper&Dilliger(1964)[163]

Copper343,seeTable744 Reynolds&Anderson(1976,1977)[193]&
[194]
Schmidt(1975)[207]

300
ECSSEHB3101Part14A
5December2011

Table742:KapitzaConductance,hk,ofIndium108inContactwith
LowAcousticImpedanceMedia(LAIM)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

2280(2280) 2,88(2.84)
SurfaceAx104[m2]:0,32
99,99%pure 2230(2240) 2,88(2.86) Measur.Method:A4
Table743
HeatFluxQ/A[W.m2]:6200
2

Indium
2280(2310) 2,81(2.80)
Temp.Range[K]:1,02,1
2070(2080) 2,81(2.78)

Table743:SampleDescriptionofIndium108inTable742
Storage, Sample
hkat1,9K
LAIM SampleConditions Environment Dimensions Comments
[W.m2.K1]
beforeCooldown [m]

Sapphire Polishedfaceofa Assemblywithmold Indium: Sample1.TestA.


cylindricalsapphire returnedtovacuum 6,35x103dia. Superconducting
2280(2280) crystalwasindium andannealedin 38x103height stateor(normal
coatedinahelium vacuumfortwo Sapphire: state,upto
1
atmosphere.Assembly hoursat420K.Mold 6,35x10 dia. 0,6x10 A.m )
3 6

wasplacedinaclose cutaway. 32,103height


fittingmoldandbacked Sample1.TestB.
inavacuumof1,3x103 Superconducting
2230(2240) Pa.Additionalindium stateor(normal
wascastandrefinedby state,upto
themoltenzone 0,6x106A.m1)
techniquewithinthe
Sample2.TestA.
mold.Twonominally
Superconducting
equalsamples(1and2)
2280(2310) stateor(normal
wereprepared.
state,upto
0,6x106A.m1)

Sample2.TestB.
Superconducting
2070(2080) stateor(normal
state,upto
0,6x106A.m1)

301
ECSSEHB3101Part14A
5December2011

Table744:KapitzaConductance,hk,ofCopper343inContactwith
LowAcousticImpedanceMedia(LAIM)
Temp.
hkat1,9K Sample
Sample Key Exponent Others
[W.m2.K1] Description
n

400(at1K) 3,5

340(at1K) 3,6
Copper(no SurfaceAx104[m2]:5,13(eachendplate)
detailsare Measur.Method:D1
300(at1K) 3,4
given) Temp.Range[K]:0,71
Table745
2

310(at1K) 3,5

300(at1K) 3,6

Commercial Measur.Method:D
3000 1,9
Copper Temp.Range[K]:14

Table745:SampleDescriptionofCopper343inTable744
hkat1,9K
LAIM SampleConditions Comments
[W.m2.K1]

400(at1K) Bothendplatesmechanicallypolished. Ca.33x105Papressure

Bothendplatesmechanicallypolished.Then
340(at1K)
vacuumannealedat873K.

Oneendplatemechanicallypolishedand Ca.37x105Papressure
300(at1K) SolidHe4
sandblasted.Thesecondmechanicallypolished
andelectropolished.

Bothmechanicallypolishedand
310(at1K)
electropolished.

300(at1K) Sameasabove.

EpoxyResin:CY221withhardenerHY979 Becauseofhkthethermal
Epoxy (CIBACo.)in10:3ratio. conductivityoffilledepoxies
3000
Resin stronglydependsonthegrain
sizeofthemetallicfiller.
a Cindicatesthatthevaluesofhkandnhavebeencalculatedbythecompilerthroughbestfittingofhk=ATnto
datapoints.ThesedatapointshavebeenreplottedinFigure718andcanbeidentifiedbythekeygiveninthe
previousentryoftheTable.

302
ECSSEHB3101Part14A
5December2011

7.3 Thermo-acoustic oscillations


Thermoacoustic (Taconis) oscillations, which can reach considerable amplitudes, appear in tubes
which are hot at their closed end and cold (liquid helium temperature) at their open end. These
oscillations are accompained by considerable heat flux along the tube into the helium reservoir
increasingtheevaporationrateofliquidheliumtoavalueoftheorderof103largerthanthatwhich
wouldoccurwithoutoscillation.
DampingofacousticwavesinlongtubesduetofrictionisanoldprobleminGasDynamics(seeRott
(1969) [200] and references therein). Early attempts to use these waves in pulsating combustion are
describedinThring(1961)[235].
Oscillationsappearandpersistbecausetheperiodicheatfluxtothegasissuchthattheheataddition
coincideswiththephaseofhighpressure,thatistosay:thegasshiftedtowardthehotclosedendis
bothcompressedandheatedthroughthewalls,whereasinthemotionawayfromthehotclosedend
botheffectsreversetheirsigns.
Firstattemptstoqualitativelycalculatethermallydrivenacousticoscillationsinatubewithastrong
axialtemperaturegradientwerebasedonlinearizedflowequations,accountingforthewallfriction
throughanoscillatingboundarylayer(theStokeslayer),thethickness,(/)1/2,ofwhichwasassumed
tobesmallcomparedwiththetuberadius.andbeingthekinematicviscosityoftheoscillatinggas
andtheangularfrequency,respectively.
Comparisonwithexperimentalresultswasdisappointing.Calculationsonlyprovidedtheasymptote
ofacertainbranchofthestabilitycurveand,intheparticularcaseofhelium,thematerialconstants
aresuchthattheasymptoteliesatpracticallyinfinitehottocoldtemperatureratios.
Rott (1969) [200] showed that a second order boundary layer theory is required to obtain the
asymptoticvalueofthestabilitycurveforlowviscosityandinvestigatedthestabilitylimitsingeneral
(Rott(1973)[201]),calculatinginanotherpaper(Rott(1975)[202])thesecondorderheatflux.
With the aim of simplifying the analysis, Rott assumes that the temperature distribution along the
tubeaxisisastepfunctionwithachangeatpositionloftheaxis.ThelengthofthetubeisL,andthat
ofthewarmpartLl.Thisassumptioncouldbefairlyrealisticwhenthetubeissupportedbyaflange
attachedtothetopoftheheliumbath.
NeutralstabilitycurvesforheliumarepresentedbyRott(1973)[201]asvs.YCwithasaparameter,
where

TH DE / 2 Ll
, YC ,
TC C / 1/ 2 l

.
SubscriptCindicatespropertiesofthelowtemperaturegas.Inthecaseofthedistinctionisessential
sinceinRottsanalysistheviscousregionreachesthecoreofthetubeinthehotpart,butitisverythin
comparedtothetuberadiusinthecoldpart.
Results deduced from Rotts analysis have been experimentally confirmed by Yazaki, Tominaga &
Narahara (1979) [269] as can be seen in Figure 721. These authors used two U tubes of different
internal diameters, and two different warm temperatures. Density (and then ) was changed by
pressurization of the gas, and pressure was measured with transducers placed at the end of the U
tubes.

303
ECSSEHB3101Part14A
5December2011

Figure721:TheneutralstabilitycurveforTaconisoscillationswhen=1. DE=
2,4x103m,TH=288K; DE=2,4x103m,TH=77,3K; DE=4,4x103m,TH=288
K; DE=4,4x103m,TH=77,3KFromYazaki,Tominaga&Narahara(1979)[269].

A method for preventing pressure oscillations in tubes connecting liquid helium reservoirs to room
temperature consists in drilling holes in the tube wall about mid distance between cold and warm
ends. Figure 722 shows a tube with an enlarged warm end sealed with a thin rubber membrane.
When the perforated tube is inserted into a cryostat no oscillations are observed. A second tube
simulatingtheneckofthecryostatisalsoshowninthefigure.

304
ECSSEHB3101Part14A
5December2011

Figure722:DeviceforpreventingTaconisoscillations.Allthedimensionsarein
mm.FromHilal&McIntosh(1976)[88].

With independence of its detrimental effects on the life of the helium contained in the reservoir,
Taconisoscillationisusedforlocatingtheliquidvaporinterfaceinhelium(Gaffney&Clement(1955)
[73]).

7.4 The superfluid plug

7.4.1 Phase separation in superfluid helium


Liquid He II (He4) can be used for cooling research equipment, such as superconducting magnets,
linearaccelerators,spacebornesensors,etc.
Asystemtomaintaincryogenicequipmentattemperaturesbelow2,5KbyimmersioninliquidHeII
wasproposedbySelzer,Fairbank&Everitt(1971)[214].UnderzerogconditionsthecontactofHeII
with the equipment to be cooled is achieved in this system by means of the superfluid film which
coatsalltheinnersurfacesofthecryostat(seeclause7.1.6).
Amongthedifferentproblemswhichthissystemwouldpose,thatofliquidvaporphaseseparationis
thesubjectofthisclause.
Theaimofphaseseparatorsistoavoidtheseepageoftheliquidrefrigerantintothenecksupportof
thecryostat,whoseexitisatambienttemperature,withtheconsequentrapidevaporationoftheliquid
anddecreaseinthecoolingeffectivenessofthesystem.
Areviewofphaseseparatorsnotbasedonsuperfluideffectsandwhichcanbeappliedtoanynormal
(nonsuperfluid)liquidhasbeenmadeinClause6.4.
WhenthecryogenisHeIIphaseseparationmaytakeplacewithinaporousplug.

305
ECSSEHB3101Part14A
5December2011
Even assuming that the total amount of helium entering the plug is liquidHe II, only vapor will
emergefromtheexitfaceoftheplugifconditionsarearrangedsothatthethermomechanicalforces
overcomethemechanicalpressure,i.e.:providedthatthechemicalpotential(Eq.[78])attheexitof
the plug equals that at the entrance. Under such conditions the liquid evaporates inside the plug,
coolingit.
ThesuperfluidporousplugisthenathermalphaseseparatorforHeIIwhichisbasedonthefactthat,
under the two fluid model, the normal component is driven through the plug by the joint effect of
mechanicalandthermomechanicalpressureandiscontainedbyviscousforces(Eq.[718]),whereas
thesuperfluidcomponentiscontainedbythethermomechanicalpressure.Ifagivenmassflowrate
ofliquidHeIIenterstheplug,thesamemassflowrateoffluidwillleaveit,butthisfluidwouldbe
onlyvaporifthermomechanicalforcesarelargeenoughtocontainthesuperfluidcomponent.

7.4.2 Simplified theory of the superfluid plug


The theory presented in this paragraph is that by Selzer, Fairbank & Everitt (1971) [214], with
refinementsbyUrban,Katz&Karr(1975)[245].
TheconfigurationusedtosubstantiatethediscussionissketchedinFigure723.plandp2arethevapor
pressures of liquid helium at temperatures Tl and T2 respectively. In most experiments concerning
superfluidplugsliquidisabsentfromchamber2,hencep2doesnotdependonT2.

Figure723:Superfluidplugarrangement.Theintakefaceoftheplugislocatedat
x=0.

Liquidheliumisindirectcontactwiththeupperfaceoftheplug.
Hydrostaticpressureisincludedsinceitshouldbetakenintoaccountwhenperformingexperiments
intheterrestriallaboratory.
The upper chamber is assimilated to the bath of a Dewar containing He II. The lower chamber is
ventedtoabsolutevacuumthroughalineofknownimpedance.Theimpedance,F,istheratioofthe
massflowratetothepressuredropalongtheline.

7.4.2.1 Simplifying assumptions


1. Theflowthroughtheplugisstationaryandonedimensional.
2. Theslowflowassumption(clause7.1.1.1)holds.
3. Differencesoftemperaturebetweenupperandlowerchamberaresmallcomparedwith
theprevailingtemperature,thuspressure,densityandentropygradientswouldalsobe
small.

306
ECSSEHB3101Part14A
5December2011
7.4.2.2 Basic equations
1. Under simplifying assumptions 1 and 2 above, the linearized momentum equations for
superfluid and normal fluid (Eqs. [74] and [75]) become respectively the Londons
equationforthethermomechanicalpressure(Eq.[717],p.39)andStokesequationfor
creepingflowofaviscousliquid(Eq.[719]).
The first of the lastly mentioned equations can beintegrated between both end faces of
theporousplug,undertheassumptionofconstantands,yielding

p1 gl p 2 s T1 T2 [752]

gravity forces within the plug, the thickness, t, of which is of the order of a few
millimeters,havebeenneglected.
2. Theheattransferrate,Q,requiredtoevaporatethemassflowrate,m,throughtheplugis
givenbyQ=m(hfg+cp(T2Tl)),hfgbeingtheheatofvaporizationoftheliquidandmcp(T2
Tl)thesensibleheat,whichinthepresentcaseisnegligible.
Aheatexchangebalancebetweenthefluidintheplugandtheplugitselfyields,

mh fg hT1 T2 [753]

wherehisthethermalconductanceoftheplugplusfluid.
ThecombinationofEqs.[752]and[753]leadsto

m h / sh fg p1 gl p2 [754]

AsimilarequationhasbeenobtainedbyKarr&Urban(1978,1980)[113]&[114].Eq.[7
54]relatesthemassflowratetothepressuredrop,throughtheimpedanceh/shfg.
Whenthepressuredropintheventinglineisnegligible,thefactorintoparenthesisinthe
righthandsideofEq.[754]issimplythevaporpressureoftheupperchamber,psat(Tl),
plusthehydrostaticpressure, gl,otherwisetheimpedance,F=m/p2,oftheventingline
shouldbetakenintoaccount.Inthatcase,Eq.[754]becomes

p1 gl
m
sh fg 1
[755]
h F

givenbyUrbanetal.(1975)[245].InmostcasesFisafunctionofm,andoftheoutletto
inletpressureratio,p0/p2,oftheventingline(seeclause7.4.2.5).
The hydrostatic pressure term can be neglected, compared with pl,except at very low
temperatures. For example, when l = 0,3 m and under normal gravity conditions, gl is
equal to the He II saturation pressure for T = 1,455 K and smaller than the saturation

307
ECSSEHB3101Part14A
5December2011
pressureforlargervaluesofT.Nevertheless,glisveryseldomnegligiblecomparedwith
plp2.

7.4.2.3 Thermal conductance of the plug plus fluid


Thepredictionofthethermalconductanceofanyliquidfilledporousmediumisnotaneasytask.A
discussionoftheavailablemodelsofthermalconductionthroughporousmediamaybefoundinBear
(1972)[19].
Itisgenerallyagreedthattheupperandlowerlimitsofthethermalconductance,h,aregivenbythe
followingtwoexpressions:
1. Upperlimit.Parallelconductionmodel,

h 1 hs hl [756]

2. Lowerlimit.Seriesconductionmodel,

1
h
1 [757]

hs hl

Intheseexpressions istheplugporosity(volumeofporestobulkvolumeratio),hsisthethermal
conductanceofaplug(ofthesamesizeandmaterial)withnovoids,andhlthethermalconductance
oftheHeIIfillinganotherfictiousplug(samesize)with =1,althoughwiththerightvalueofthe
permeability(seeclause7.4.2.4).ItisindicatedagainthatheattransferthroughHeIIiscontrolledby
motionofthenormalfluid.
Earlydevelopmentsofthesuperfluidplug(Selzeretal.(1971)[214]),whichwerebasedontheparallel
thermalconductionmodel,usedporousplugsofhighthermalconductance.Evenmore,theplugwas
embeddedinalargemetallicblockwiththeaimoffurtherenhancingitsthermalconductance.
More recent experimental evidence (Elsner (1973) [63], Petrac (1975) [181], Urban et al. (1975) [245],
Karr & Urban (1978) [113]) indicates that ceramic plugs of fairly low thermal conductance are also
efficient phase separators.Thence, in the followingwe willrestrict ourselves to the assumption that
theheattransferthroughtheliquidisdominant.
In addition, we will assume that the overall heat conduction takes place according to the parallel
model.Noticethat,takenforgrantedthattheheattransferthroughtheliquidisdominant,thethermal
conductance,h,accordingtoeithermodeldifferinafactor2.Furthermore,dataonporousplugsare
scarceandincomplete.Evenassumingthatthelineartheoryisappropriate,whichisnotalwaysthe
case, as we shall see, the resulting expressions are plagued with uncertainties mainly related to the
poorly known characteristics of the porous media. It is, thus, difficult to obtain from the available
experimentalknowledgeadetailedinformationregardingthethermalconductionmodel.

7.4.2.4 Thermal conductance of the liquid helium


An effective thermal conductivity (or a thermal conductance) of the liquid helium within the plug,
much larger than that resulting from the purely molecular heat transport process, can be deduced
fromtheexpressionoftheheatflow,Eq.[716].

308
ECSSEHB3101Part14A
5December2011
DarcyslawforflowthroughporousmediawillbeusedinsteadofPoiseuillesequation(Eq.[725])to
relatethepressuregradient,dp/dx,tothenormalvelocitycomponent,vn

dp
n vn
dx K [758]

where K [m2] is the permeability, or geometric factor of the porous medium. This parameter
characterizes the ease with which a fluid may be made to flow through the medium by an applied
pressuregradient.ForacirculartubeofdiameterDE,K=DE2/32andPoiseuillesequationisrestored.
Foranannularslitofwidthd,K=d2/12.
TocalculatethethermalconductanceoftheliquidheliumwecombineEq.[716](q=mhfg/AFLandAFL
=AFR)withEqs.[754]and[758].

AFR 2 s 2T 1
h K
t n 1
sT
[759]
h fg

VeryusuallythelastfactorintherighthandsideofEq.[759]isomittedsincesT<<hfg(sT/hfg=3x103
at T = 1,2 K, sT/hfg = 101 at T = 2,1 K). A similar equation, with K = d2/12, has been used by Karr &
Urban(1978,1980)followingRoberts&Donnelly(1974)[196].Noticethattheexpressionwhichresults
oncethelastfactoroftherighthandsideinEq.[759]hasbeenomitted,followsfromEq.[714]under
the counterflow assumption (no net mass flow) although, strictly speaking, this assumption is not
requiredhere.
Oncehhasbeenobtained,themassflowratethroughtheplugcanberelatedtothepressuredropby
useofEq.[754]

sT
m K
AFR
p1 gl p2
t n h fg sT [760]

ThisisthesocalledAllenReekieRulewhichholdsforanyvalueofpl+glp2providedthatthenormal
fluidflowremainslaminar.
TheexperimentalvalidationofEq.[760]willbeattemptedinclause7.4.2.6.
Whether or not liquid and vapor coexist within the plug is a matter of discussion (Schotte (1982)
[209]).
Boilingwithinabundleoffinecapillariesoraporousmediumisaveryunlikelyphenomenonsince
thebubblesaresosmallthatthesuperheatofboilingwouldbeexceedinglylarge(Enyaetal.(1981)
[64]). Nevertheless, situations occur where, due to the liquid cooling by the plug, an additional
fountain pressure appears which pushes upstream the liquidvapor interface until a balance of

309
ECSSEHB3101Part14A
5December2011
fountainpressure,capillarypressureandvanderWaalsforcesisreached.Thiseffecthasbeencalled
choking(Schotte(1982,1984)[208]&[209])seealsoclause7.4.2.7.
The calculation of an effective liquid conductivity when liquid and vapor coexist within the plug is
verycomplicatedbecauseofthefollowingreasons:
1. Thevolumetricliquidcontentcannotbeeasilyestimated,aspointedoutinclause7.1.6.
2. The permeability, K, of the porous medium is a function of that liquid content (Philip
(1970)[185]).
3. Vaporinclusionswillaffecttheheattransfermechanism.
Nomuchattentiontothissituationhasbeenpaidintheliterature,hencefurth