NASA-TM-84631
NASA Technical Memorandum
84631
DESIGN,FABRICATIONAND TEST OF
LIQUIDMETAL HEAT-PIPESANDWICHPANELS
AL BASIULISAND CHARLESJ. CAHARDA
APRI
L 1983
NationalAeronauticsand
Space Administration
Langley ResearchCenter
Hampton, Virginia 23665
19830014270
DESIGN_FABRICATION AND TEST OF LIQUID METAL HEAT-PIPE SANDWICH PANELS
A. Basiulis
HughesAircraft Company
Torrance, California
C. ]. Camarda
NASA Langley ResearchCenter
Hampton,Virginia
Abstract
Integral heat-pipe sandwich panels, which synergistically combine the thermal efficiency of heat pipes and the
structural
efficiency
of honeycomb sandwich panel construction, were fabricated and tested. The designs utilize
two different wickable honeycomb cores, facesheets with
screen mesh sintered to the internal surfaces, and potassium or sodium as the working fluid. Panels were tested by
radiant heating, and the results indicate successful heat
pipe operation at temperatures
of approximately
922K
(1200°F).
These panels, in addition to solving potential
thermal
stress problems
in an Airframe-lntegrated
Scramjet Engine, have potential applications as cold plates
for electronic component cooling, as radiators for space
platforms, and as low distortion, large area structures.
LANGLEYSCRAMJET
CONCEPT
HONEYCOMB
STRUCTU
RE
LARGEAT
THROUGH
HONEYCOMB
Introduction
Design studies of the NASA Langley Airframe-lntegrated Scramjet Enginel have indicated potential thermal
stress problems.
Excessive thermal stresses result from
large transient temperature
gradients across the honeycomb sandwich walls of the engine structure during engine
startup and shutdown.
Conventional
heat-pipe
panel
designs can reduce the thermal
gradients.
However,
inherent in these designs are problems associated with
bonding the heat pipes to the honeycomb panels) the
resultant thermal gradients due to contact resistances, and
the probability of a substantial increase in panel mass. An
alternate solution to these problems is the development of
an integral heat-pipe sandwich panel 2 that synergistically
combines the thermal efficiency
of heat pipes with the
structural efficiency of sandwich construction, with only a
negligible increase in mass. A preliminary evaluation of
such a concept was reported by Peeples. 3
)LING
JACKET
rOPWALL
.SIDEWALL
COOLEDSTRUCTURE
In addition
to the above application,
heat-pipe
sandwich panels have potential as cold plates for electronic
and circuit card cooling) as radiators for space platforms,
and as low distortion) large area structures (e.g., space
antennas).
To verify
the feasibility
of a heat-pipe
sandwich panel, a program was initiated (NASA Contract
NASI-16556) to fabricate
several low mass liquid metal
heat-pipe honeycomb panels,
Fig. I Features of cooled scramjet structure.
This paper describes the thermal environment that led
to the investigation of a heat-pipe sandwich panel) illustrates the preliminary
design considerations and testing,
describes manufacturing and fabrication details, discusses
preliminary performance testing, and comments on poten-
surfaces
exposed to aerodynamic
flow
are cooled
regeneratively by the circulation of hydrogen fuel (prior to
injection) through a cooling jacket. Inconel 718 was chosen
for the honeycomb primary structure) with Hastelloy-X
or
Nickel-200 chosen for the cooling jacket. The honeycomb
front facesheet is 0.15 cm (0.06 in.) thick) the back facesheet is 0.13 cm (0.05 in.) thick, and the honeycomb cell is
a 0.6#-cm (0.25 in.) hexagonal arrangement constructed of
0.008-cm (0.003 in.) foil-gage ribbon.
tial future applications.
Environment
Design of Heat-Pipe
Sandwich
Panels
Temperature gradients through the honeycomb walls
during transient operation (i.e., engine startup or shutdown)
may very well control the structural design. A mission
profile of a research-type vehicle was used by Buchmann1
to predict the thermal/structural response quantities. A
finite-difference analysis model of a section of the
sidewall-topwall (Fig. 2) was used to calculate the transient temperatures shown in Fig. 3. Note from Fig. 3(a)
that at startup, the front facesheet quickly rises to 890K
NASALangleyResearch Center has been involved in a
research program for the development of AirframeIntegrated Scramjet Engine concepts. 1 Results of that
study indicate that an all-honeycomb primary structure,
illustrated in Fig. 1, has less deflection and complexity
than beam and honeycomb combinations of equal mass.
Hence, an all-honeycomb configuration was chosen as the
best structural concept. All internal and external engine
i
TOP- -___S
WALL _IEl'l
• I,_
_
CORNERBRACKtI
TYPICAL
TEMPERATUI_[
-- -- -BA
_I
J_
•
•
•
•
f.J_
:
....:
FRONTFACESHEET--I_
OF COOLINGJACKET i I SIDEI! WALL
II
Fig. 2
If
I_
NODE
honeycomb
core could
be a foil-gage
or
to reduce thermal
gradients
in the woven
faces. mesh
The screen
wiekable
CK FACESHEET
flow.
The tohoneycomb
is notched
at each
end to
aliquid
screen
sintered
foil ribbons;
this allows
face-to-face
OF HONEYCOMB
ZONE
allow intracellular
liquid flow by capillary action.
This
design
allows tothetheentire
to be
screen also
is sintered
internalsurface
faces ofof the
the facing
sandwich
to
wetted by liquid and thus aid in evaporation and also help
1
II
II
Mathematical model for transient thermal anal_Isis
of honeycomb topwall-sidewall corner section. _
allow
intracellular
liquid flow
and perforated
enableof
intracellular
vapor flow.
Although
the primary tomode
heat transfer is in the transverse direction (face to face)
for the present application,
the choice of other design
alternatives
can enable varying degrees of in-plane heat
transfer.
Critical Element Evaluation
To accommodate the heat-pipe sandwich panel requirements, the structure must consist of two facings with
internally wickable faces bonded to a perforated, wickable
honeycomb core material (as shown in Fig. 5). Several
(ll#0°F),
resulting
in a front-to-back
AT of 667K
(1200°F) for a Hastelloy-X
core.
At engine shutdown,
whether caused by normal occurrences or an abnormality
[ MAXIMUM
such as a flameout, the temperature relationships of the
front of the cooling jacket and the back of the honeycomb
are
reversed,
as shown in
Fig.than
3(b).
front-to-back
AT
developed
is somewhat
less
at The
startup--on
the order
of 5561< (1000°F) for the Hastelloy-X core. These thermal
premature fatigue failure.l
Solutions
noted in Ref. l result in concepts that
complex or heavier, or both.
gradients
result
in
excessive
thermal
to this problem
are either more
stresses
_
(HASTELLOY-X)
_
|
F
r
_'"FRONT
v
_
emerged as a solution to the above problem,
The heat-pipe
sandwich panels fabricated
in the past met unique requirements of unilorm temperature
over a large surface area.
The basic idea for the heat-pipe
sandwich panel
_
field,4
uniform have
temperature
over operation
a large area5
(0.5x6
These panels
demonstrated
in a zero-g
meters (1.64x19.7 ft.)) and an isothermal surface6 (within
0.0IK (0.02°F)).
However, all these panels were built by
welding or furnace brazing by highly skilled technicians
and, although they met all the technical requirements, they
were very costly to manufacture.
The primary objective was to fabricate a heat-pipe
l_oneycomb sandwich using a wickable honeycomb core,
appropriate working fluid, and wickable internal faces that
would enhance the transverse heat transfer capability of
ihe honeycomb. During operation, heat would be absorbed
at the heated face by the evaporation of working fluid,
The heated vapor flows, due to a pressure differential,
to
the cooler face, where it condenses and gives up its stored
heat. The cycle is completed with the return flow of liquid
condensate back to the heated face by the capillary
pumping action of the wickable core. A schematic of the
heat-pipe sandwich panel concept is shown in Fig. 5. A
F(N_//
900 |
and
The objective during this program was to design and
fabricate a cost- and mass-effective
sandwich panel using
existing manufacturing
techniques and equipment.
The
upper and lower ends of the core have flanges that enable
spot welding to the faces. The entire sandwich panel can
be constructed by simultaneously spot welding the core
ribbons to each other and to the faces using the manufacturing technique illustrated in Fig. 4. The spot welds are
so close together that they form an almost continuous
bond. Since the entire panel is spot welded, this eliminates
_he need for bonding and possible materials compatibility
problems.
MAXIMUMAT
700t
_
800I-
I
t
/
_
! /HAST.-X
300_
120
/
I
I
I
60o
I080
TIME (SECONDS)
(a)$TARTUP
MAXIMUM
•.&T (Ni)
900
"
__
\
MAXIMUM
/(HASTELLOY-X)
_T
/
.\.
_.
BACK
o_
- 700
LU
:E 500
_
.,
300 l
0
[
]
]
80
120
TIME (SECONDS)
(b) SHUTDOWN
Fig. 3 Honeycomb temperature
histories at zone I. 1
WHEELS
E
TRODE
STRIP
CORE
RIBBON
_..
1. FLANGE WELD
J
|
.
"-....
ELECTRODE
PANEL
FABRICATION
MACHINE
TIPS
2. NODE WELD
Fig. #
_"_
Honeycomb panel welding machine and manufacturing technique (courtesy of Astech).
Fig. 6
techniques were considered for internal facesheet wicking:
sintering a screen to the facing, spot welding a screen to
the facesheet, and grooving or roughening the facesheet by
grid blasting.
Grooving and roughening were rejected
because of facesheet warping and the poor surface left for
subsequent welding.
Sintering the metal screen to the
facing was chosen as having more structural integrity than
spot welding.
Figure 6 shows a photomicrograph of the
sintered screen facesheet,
Photomicrograph showing diffusion
screen sintered to facesheet.
bonding of
assembly for process testing and preliminary performance
testing; and a machine-assembled resistance-welded
prototype for delivery and final testing.
The proof-pressure
test specimen was assembled,
vacuum leak checked_ pressure tested up to 3.45 MPa
(500 psi), and vacuum leak checked again. During and after
testing, the honeycomb panel assembly retained structural
and vacuum integrity.
A hand-built,
spot-welded core
assembly was fabricated, processed with potassium working
fluid and, after preliminary test at I075K (1475°F), was
delivered to NASA Langley Research Center for further
testing.
Two designs for the honeycomb core were considered:
a foil-gage sintered screen material (shown in Fig. 7) and a
metal screen sintered to foil-gage stainless steel material,
Both designs met structural and wicking requirements, with
the former offering better wicking and the latter providing
a stronger structural design. Figures 8 and 9 show performance limits for a heat-pipe honeycomb sandwich panel
constructed
with Regimesh K material for sodium and
potassium working fluids,
Fabrication
of Test Models
Sandwich panels were fabricated by Astech using an
automated procedure for simultaneously resistance welding
honeycomb ribbons to the facesheets. Completed sandwich
Sample honeycomb ribbons were formed by Astech*
using standard equipment, and test samples were fabricated for evaluation.
Both samples, sintered screen and
screen on foil_ met strength and wicking requirements.
Three different designs of honeycomb sandwich panels
were fabricated:
a resistance-welded
core assembly for
proof-pressure
testing; a handmad% spot-welded
core
* A division of TRE Corporation,
j
,
Santa Ana, CA.
TOP FACESHEET
//
WlCKABLE TO
HONEYCOMB
NOTCHED
ALLOW
LIQUID FLOW, PERFOR
ATED TO ALLOW
VAPOR FLOW
,._,.
SCREEN SIN'I'ERED TO
INTERNAL
FACES TO ALLOW
IN PLANE FLOWOF LIQUID
ALONG FACES
Fig. 5 Heat pipe sandwich panel concept.
Fig 7 Photomicrograph
3
of Regimesh K sintered screen.
I
WICK
OPERATING
10,000----
I
TEMPERATURE.._I
l
DETAIL
_ 10o0
_
if"
Z
<
-
<
00
f-/
ENTRAIN
/
/
/
o.//
r
I
!I
/
U"
/_S°N'C !
!
/"
?
!
/?
_-3.......
fi
4-
400
-+--
600
800
TEMPERATURE
1000
(OK)
Fig. 8 Performance limits vs. temeprature for
RegimeshK and soduim fluid.
panels were delivered to Hughes for further processing.
Figure 10 shows the completed honeycomb panel.
To
eliminate potential contamination, panels were degreased,
then fired in dry hydrogen at 1173K (1652°F). At this point
the sidewalls
Fig. l0 Completed honeycomb panel prior to processing
and final assembly.
and processing tube were welded in place)
completing the heat pipe assembly.
Figure 11 shows the
complete heat pipe assembly. The panel was then fired in
dry hydrogen at 1173K (1652°F) to remove oxides which
were formed during final assembly. After leak check_ the
panel was placed in a vacuum chamber and heated to
1273K (1832°F) for final cleaning and outgassing.
After
_
OPERATING
TEMPERATURE_I
--
_
WICK
.._.._L_--."
• _"
-•
10.000_---- _r""
_"
-'_'--J_'l ,_
d
/ I
J
1-
1000
ENTRAINMENT
o
/
Z
F-
F-
final processed.
leak check, During
the panel
was chargedtests,
with the
working
fluid
and
preliminary
heat-pipe
panel was isothermal over the active surface but did show
some excess fluid in the processing tube.
Figure 12 shows
the heat-pipe panel during preliminary test.
"
/
100
-r
SONIC
10
I
/
/
1
I
400
I
600
TEMPERATURE
Fig. 9
I
'I
I
I,
800
1000
(°K)
Performance
limits vs. temeprature
Regimesh K and potassium fluid.
for
Fig. l 1 Completed
4
heat pipe assembly
prior to processing.
L_--_
2.54cm
(1.0 in.)
10 16cm (4.0in.) ---.-,_
T
2.54 cm (1.0 in.)
5.08 cm (2.0 in.)
•
1
S
5.08 cm (2.0 in.)
!- +
+
2.54 cm (1.0 in.)
Fig. 12 Heat-pipe
panel during preliminary
testing.
_L
SIDE
VIEW
TOPANDBOTTOMVIEW
Preliminary Radiant Heat Tests
Two prototype
Fig. I# Thermocouple locations.
panels) one empty and the other contain-
ing potassium as the working fluid) were heated simultaneously by radiant heat lamps) as shown in Fig. 13. The
heaters are quartz lamps with a heated length of 6.35 cm
(25 in.) and having a rated power of 2500 watts (2.37 Btu/s)
at 500 volts.
Each lamp bank contains eight lamps. Six
lamp banks were energized for each test.
One of the
panels was located directly under one lamp bank and the
other panel was located the same vertical distance from
the heaters but under another lamp bank. The distance of
the panels from the lamp banks and the voltage to the
lamps were varied. Power was applied as a step voltage
input to the lamps. Power was applied for approximately 5
to 10 minutes and then abruptly shut off. Five thermocouples were located on the top and five on the bottom
surfaces of the panels to measure temperature gradients)
and four thermocouples were located along one side to
study heat pipe startup performance.
Thermocouple locations are shown in Fig. 14. The panels were tested with
and without insulation covering the bottom and sides of the
panels. The insulation prevents heat loss by free convection and simulates the adiabatic boundary conditions described in Ref. I.
The panels were tested nine times) and results of those
tests are summarized in Table I. Comparisons of temperature histories of a heat-pipe and non-heat-pipe sandwich
panel with insulated and uninsulated surfaces are shown in
Fig. 15. For the insulated panel tests shown in Fig. 15(a))
the temperatures of the back face of the non-heat-pipe
panel and the heat-pipe panel temperatures continue to
rise and slowly approach the temperature of the front
facesheet of the non-heat-pipe panel as expected. Results
of the uninsulated panel tests (Fig. 15(b))indicate
that all
temperatures level off and appear to reach a steady-state
condition.
As mentioned in Refs. 8 and 9, during heat-pipe
startup
from
the frozen
state)
a nearly
constant
temperature continuum region propagates from the evaporator to the condensor section of the heat pipe. As shown in
AIR\ COOLANTLINE FOR LAMPS
HEAT LAMPS_
Table I
Summary of radiant heat tests of heat-pipe
sandwich panel
MAX _ T
DIST. FROM
HEATERS
RUN
VOLTAGE
NO
INSULATED
CM (IN.)
V
1
NO
10.5
(4.125)
380
2
460
J
t
PANEL
NON-HEAT-
Fig. 13 Sandwich panels in position under radiant
heat lamps.
5
NON
HEAT
PIPE
PIPE
338
(809)
403
(725)
352
474
HEAT
PIPE
891
(1145)
916
NON
HEAT
PIPE
926
(1208)
970
(1287)
250
313
(564)
407
(733)
803
(986)
863
(1093
4
300
343
(617)
467
(841)
856
(1081)
920
(1197)
5
358
353
(635)
487
877)
804
(1167)
984
(1312)
250
319
(575)
467
(840)
842
(1056)
894
(1150
354
490
923
974
(638)
(882)
(1202)
(1293)
535
968
1020
5.1
(2.0)
YES
LEADS
PIPE PANEL
HEAT
(1189)
6
/
THERMOCOUPLE
(OF)
(854)
)
_[.
K
(OF)
(634)
3
HEAT_PIPE
MAX T
K
7
307
8
356
9
463
378
(680)
(963) (1283)(1376
422 587 1073 1078
(750) (1056)(1472)(1480)
NONHEAT
1000 -__+
80C
/
o PANEL
800
THERMOCOUPLE
/
LOCATIONS
v
600
II i /
#
,"
4
/
/
400
_
TOPSURFACE
....
BOTTOMSURFACE
I
100
',.'I/
600
I
til
_
.....
_ -- _
....
400
I
I
300
TIME (SECONDS)
w
500
100
1
z
0.51cm(0.2 in.)
1.02cm(0.4 in.)
1.52cm(0.6in.)
2.03cm(0.8in.)
I
I
300
TIME (SECONDS)
500
(a) WITH INSULATION
Fig. 16 Temperature histories along the side of the
heat-pipe panel, illustrating startup
performance.
1000 --
o_
_-
_
..........
?
H
600
--"
400
]
100
J
I
300
TIME (SECONDS)
I
500
Initial studies indicate the heat-pipe honeycomb sandwich panels can be fabricated.
The technology and commercial equipment are available to construct all-welded
machine-assembled
honeycomb panels.
At present, such
shapes for use in airframe structures.
Calculations and
experiments with subscale test specimens indicate the
feasibility
of full-scale
heat-pipe
sandwich structures.
Potential
applications
for heat-pipe
sandwich panels
include=
alleviating
excessive thermal stresses in jet
engines, cooling electronic components and circuit cards,
limiting
thermal distortions in large structures such as
space antennas, and as radiators for space platforms.
(b) WITHOUTINSULATION
Acknowledgements
Fig. 15 Comparison of temperature histories of a heatwithout
pipe andinsulation.
non-heat-pipe sandwich panel with and
Fig. 15, once this continuum front reaches the back facesheet, the temperature
there rises very rapidly as compared to the back facesheet
of the non-heat-pipe
panel.
The temperature
at which continuum
flow begins and the
rate at which the continuum
front propagates
depend on
sonic flow limit of the vapor.
the working fluid, the temperature,
using a potassium
heat-pipe
The authors wish to acknowledge T.R. Lamp, H. Tanzer
and 3.T. Burdette of Hughes Aircraft Co., who supported
the design, fabrication,
processing and testing of the
honeycomb heat pipe sandwich panels, and T. Bernard of
Astech for his support and for the fabrication
of
honeycomb sandwich segments.
600
--
AT
the heat input and the
sandwich
panel
instead
of a
non-heat-pipe
panel is 27 percent.
It is possible that this
reduction can be increased by using cesium as the working
fluid; this is currently being investigated.
Figure 16 gives
some idea of the rate of continuum region growth. The
results are characteristic of startup of liquid metal heat
pipes as presented in Ref. 9. A typical comparison of
temperature gradients through the depth of the honeycomb
is shown in Fig. 17. As shown, the non-heat-pipe
panel
temperature
gradient
peaks slightly after that of the
run.
heat-pipe panel and is 29 percent higher for this particular
_
_}
o
200
#
I
100
_J
I
300
TIME (SECONDS)
[
500
Fig. 17 Comparison of temperature gradients for a
heat-pipe and non-heat-pipe sandwich panel.
References
h
Buchmann, O. A., "Thermal-Structural Design Study
of an1979.
Airframe-Integrated Scramjet,"NASA CR 3141,
Oct.
2.
Feldman, K. T., 3r., "Flat Plate Heat Pipe With
Structural Wicks," U.S. Patent Appl. No. 803,582.
3.
Peeples, M.E., Reeder, 3. C., and Sontage, K.E.,
"ThermostructuralApplicationsof Heat Pipes," NASA
CR 150906,3une 1979.
4.
6.
Fleishman, G.L., Loose 3. P., and Scallon, T., 3r.,
"Vapor Chambers for Atmospheric Cloud Physics
Laboratory," Third International Heat Pipe Conference, Palo Alto, CA, May 22-24, 1978.
7.
Dunn, P. and Reay, D.A.,
Press, 1976.
8.
Cotter, T.P., "Heat Pipe Startup Dynamics," Heat
Pipes, AIAA SelectedReprint Series,Sept. 1973, VoL
XVI, pp. 42-45.
9.
Camarda, C. 3., "Analysis and Radiant Heating Tests
of a Heat-Pipe-Cooled Leading Edge," NASA TN
D-8468, Aug. 1977.
Fleishman, G. L. and Marcus,B. D., "Flat Plate (Vapor
Chamber/Heat Pipes,"AIAA Paper No. 75-7728, AIAA
t0th ThermophysicsConference,May 1975.
5.
Heat Pipe Conference, Palo Alto, CA, May 22-24,
1978.
Basiulis, A. and Formiller D. 3, "Emerging Heat Pipe
Applications," Proceedings of the Third International
Heat Pipes, Pergamon
1. Report No.
2. Government Accession No.
3, Recipient's Catalog No,
NASATM84631
4. Title and Subtitle
5. Report Date
Design,Fabrication
and Test of LiquidMetalHeat-Pipe
Sandwich Panels
April1983
6. PerformingOrganizationCode
506-53-53-07
7. Author(s)
8. Performing Organization Report No.
A1 Basiulis and Charles J. Camarda
10. Work Unit No.
9. Performing Organization Name and Address
NASALangleyResearchCenter
Hampton,VA 23665
11.
Contract
or Grant No.
13. Type of Report and Period Covered
12
Technical Memorandum
.......
t4 Sponsoring Agency Code
Sponsoring Agency Name and Address
National Aeronautics and SpaceAdministration
Washington, DC20546
15. Supplementary Notes
Al Basiulis:HughesAircraftCompany
Presentedat AIAA/ASME3rd JointThermophysics,
Fluids,Plasmaand Heat Transfer
Conference,
June 7-11,1982,St. Louis_Missouri.
16. Abstract
Integral heat-pipe sandwich panels, which synergistically
combinethe thermal
efficiency of heat pipes and the structural efficiency of honeycomb
sandwich
panel construction, were fabricated and tested. The designs utilize two
different wickable honeycomb
cores, facesheets with screen meshsintered to the
internal surfaces, and potassium or sodiumas the working fluid.
Panels were
tested by radiant heating, and the results indicate swccessful heat pipe
operation at temperatures of approximately 922K (1200VF). These panels, in
addition to solving potential thermal stress problems in an Airframe-Integrated
ScramjetEngine,have potentialapplications
as cold platesfor electronic
componentcooling,as radiatorsfor spaceplatforms,and as low distortion,
largeareastructures.
17. Key Words (Sugg_ted
by Author(s))
HeatPipes
SandwichPanels
18. Distribution Statement
ThermalStress
ElevatedTemperature
ScramjetEngine
SubjectCategory34
19. Security Classif,(of this report)
Unclassified
N-305
Unclassified
- Unlimited
20. SecurityClassif.(of this page)
Unclassified
21. No. of Pages
8
22. Price
A02
ForsalebytheNationalTechnicalInformation
Service,Springfield,Virginia22161
View publication stats