Thermal Analysis and Correlation of the Mars Odyssey Spacecraft's Solar Array During Aerobraking Operations

11 Pages·2002·1.1 MB·English

by  Dec, John A.

AIAA 2002-4536
Thermal Analysis and Correlation of the
Mars Odyssey Spacecraft's Solar Array
During Aerobraking Operations
John A. Dec, Joseph F. Gasbarre
NASA Langley Research Center
Hampton VA.
Benjamin E. George, 1Lt
United States Air Force, Kirkland AFB
Albuquerque, NM
AIAAJAAS Astrodynamics Conference
5-8 August 2002
Monterey, California
AIAA 2002-4536
THERMAL ANALYSIS AND CORRELATION OF THE MARS ODYSSEY
SPACECRAFT'S SOLAR ARRAY DURING AEROBRAKING OPERATIONS
John A. Dec and Joseph F. Gasbarre
National Aeronautics and Space Administration
Langley Research Center
Hampton, VA 23681-2199
Benjamin E. George, 1Lt
United States Air Force
Kirtland Air Force Base
Albuquerque, NM 87117
ABSTRACT around Mars. To place itself into its 2-hour, 400km
circular, sun-synchronous mapping orbit, Odyssey used
The Mars Odyssey spacecraft made use of multipass the multipass aerobraking technique, which was utilized
aerobraking to gradually reduce its orbit period from a bythe Magellan spacecraft around Venus and the Mars
highly elliptical insertion orbit to its final science orbit. Global Surveyor spacecraft around Mars. Aerobraking
Aerobraking operations provided an opportunity to makes use of atmospheric drag on each orbit pass to
apply advanced thermal analysis techniques to predict gradually reduce a spacecraft's velocity at periapsis,
the temperature of the spacecraft's solar array for each which then reduces the altitude at apoapsis. The
drag pass. Odyssey telemetry data was used to correlate Magellan spacecraft was the first three-axis stabilized
the thermal model. The thermal analysis was tightly spacecraft to perform this type of multipass
coupled to the flight mechanics, aerodynamics, and aerobraking.1 Mars Global Surveyor, the second to use
atmospheric modeling efforts being performed during multipass aerobraking, gradually reduced its orbit from
operations. Specifically, the thermal analysis an elliptic 48-hour period to about a 380km 2-hour
predictions required a calculation ofthe spacecraft's circular orbit.
velocity relative to the atmosphere, a prediction of the
atmospheric density, and aprediction of the heat To control periapsis altitude, Magellan planned on
transfer coefficients due to aerodynamic heating. using thermocouple data to signal the need to perform a
Temperature correlations were performed by comparing periapsis raise maneuver, which would raise the
predicted temperatures of the thermocouples to the periapsis altitude, lower the maximum atmospheric
actual thermocouple readings from the spacecraft. density experienced, and thus lower the aerodynamic
Time histories of the spacecraft relative velocity, heating onthe spacecraft and solar arrays. In the
atmospheric density, and heat transfer coefficients, literature, it was noted that Magellan experienced at
calculated using flight accelerometer and qnaternion least 5thermocouple failures prior to the start of
data, were used to calculate the aerodynamic heating. aerobraking. 2 It isunclear from the available literature
During aerobraking operations, the correlations were if any of these thermocouples were located onthe solar
used to continually update the thermal model, thus arrays and were to be used during operations. It isalso
increasing confidence in the predictions. This paper unclear whether or not there were any thermal models
describes the thermal analysis that was performed and developed and used during the Magellan aerobraking
presents the correlations to the flight data. process to make up for the inoperable thermocouples,
but there are references toheat rate and surface
INTRODUCTION temperature calculations being performed using a direct
simulation Monte Carlo particle method. 3'4 In any
event it isclear that in the early 1990's the limitations
The Mars Odyssey spacecraft was launched on a Delta
of computers would have prohibited the use of a
II launch vehicle from Cape Canaveral Air Force
detailed enough finite element or finite difference
Station on April 7th2001. On October 23rd 2001, after a
model that could have been run in atimely enough
197-day cruise, the spacecraft performed apropulsive
fashion to be used during operations. During the Mars
maneuver to insert itself into an 18.5-hour elliptic orbit
Global Surveyor operations, the heat rate encountered
Copyright©2002bytheAmericanInstituteofAeronanticsandAstronautics,Inc.NocopyrightisassertedintheUnitedStatesunderTitle17,
U.S.Code. TheU.S.Governmenthasaroyalty-freelicensetoexerciseallrightsunderthecopyrightclaimedhereinforGovernmentalpurposes.
Allotherrightsarereservedbythecopyrightowner.
1
American Institute of Aeronautics and Astronautics
duringprevioudsragpassewsasreconstrucutesdinga
components and provide the structural integrity during
1-dimensiotnhaelrmamlodealtvarioulsocations
launch and throughout aerobraking. Specifically, the
arountdhesolaarrraysT.hismodeulsedthespacecraft
solar array structure isa sandwich construction with a
andsolaarrraythermocoutpelmeperatudraetaasinput
0.190ram facesheet of M55J graphite composite, a
todetermiwnehahteartatethespacecraanftdsolar
19.05mm aluminum honeycomb core, and another
arrayesxperienceAdth. ermamlodetol predicthte
0.190ram M55J composite facesheet.
solaarrraytemperatufroersfuturedragpasseasndto
reconstruthcetsolapranetel mperatufroerspasotrbits
wasnotavailableO.riginalldy,uringoperationthse,
planforMarsOdyssewyastousea1-dimensional
l
thermamlodeslimilatrothatoftheMarsGlobal
SurveymorodelU. nlikeMagellaanndMarsGlobal
SurveyoMr,arsOdyssewyasthefirstmultipass ............................................................................@.....
aerobrakimngissiotnomakeuseofdetaile3d-
dimensionfianliteelemetnhtermamlodedluring
operatiotnospredicthtetemperatufroersfutureorbits
andreconstruthcetsolapranetel mperatufroerspast
orbitsT. hismodeul,sedinadditiotnothe1-
dimensionmaoldepl,rovideddetaile3d-,dimensional
temperatuprreofileosfthesolaarrrayt,ransiepnltotsof .._ f_s_+_i ,_;
maximummateriatelmperatuarensdt,ransiepnltotsof
Figure 1. 3-D geometric representation of the Mars
thermocoutpelmeperaturIetsa.lsoprovidethdeability
Odyssey solar array.
toidentiftyhehottesstpotosnthesolaarrraya,nd
provideadmeantosdevelothpermaliml itsbaseodn The next layer is a0.05 lmm Kapton sheet that is co-
heartateoratmosphedreicnsity. cured to the M55J graphite beneath it. The next layer is
the solar cell layer and is made up of several sublayers:
0.190ram Gallium Arsenide solar cells, 0.152ram cover
SOLAR ARRAY DESCRIPTION
glass, and 0.229mm of adhesives 6. Figure 2 shows a
cross-sectional view of the solar array at a
The function of Mars Odyssey's solar array is two-fold. represenatative location.
First, its primary function isto provide a stable power
source for the spacecraft and scientific instruments
during all phases ofthe mission. Second, while in the
stowed configuration, the solar array provides a large
:........._:. ?
surface area onwhich the aerodynamic forces can act
iii iiii_iiii_!_iiiiiiiii_!_!iiiiiiiii__!iiiiiiiiiii_iiii_!iiiiiiiiii_!_iiiiiiiiii!_!iiiiiiiii__i!._.*ii.i._ii:i_ii,ii__iiii!_iiiiiiiiii!_!_iiiiiiii__!_iiiiiiiiii__iiii
during each drag pass ofaerobraking. Following Mars
,_,_._,_ ........... : _:_,_
orbit insertion the Odyssey spacecraft begins
aerobraking by slicing through the Martian atmosphere
at a relative velocity of about 4.57 km/s. Despite low
atmospheric densities of about 80 kg/km 3,these high !i
velocities, along with the large exposed surface area,
produce significant aerodynamic heating onthe solar
array, and spacecraft. Careful design and construction
of the solar array allows itto withstand this Figure 2. Cross-section of Mars Odyssey solar array
aerodynamic heating. 6 To reinforce certain areas of the array, sections of the
standard 19.05ram thick, 1.0 lb/ft3aluminum
The Mars Odyssey solar array is athree panel, layered honeycomb core were replaced with ahigher density
construction of low density materials. The panels are core of the same thickness. Also, to reinforce the
commonly referred to as the +X, -X, and mid panels. array's hard points, a doubler sandwich structure was
The +X and 32 panels are mirror images of each other, used. The doubler sandwich structure consisted of
while the mid panel has a unique geometry. Figure 1 0.381ram layers of M55J graphite composite on either
shows a 3-dimensional geometric representation of the side of a 18.16lmm aluminum honeycomb core. The
solar array. This geometric representation was doubler core densities varied depending on the location
developed inMSC/PATRAN and was used inthe within the array and the expected loading inthe region
aerobraking thermal analysis. Each panel consists of of concern. Examples of such reinforced hard points on
five layers; the first three make up the structural the array are the hinge mounting points. Overall, the
2
American Institute of Aeronautics and Astronautics
arraydesigmnakeusseoffourdifferendtensity
oriented with its high gain antenna pointed towards
aluminuhmoneycomcobres.
Earth and the solar array normal to the sun. In
transitioning to the aerobrake configuration, both heat
Toprotecthtearrayfromaerodynamheicatinga,
rates and view factors to space for the solar array were
0.072ratmhicklayeorfmultilayeinrsulatio(Mn LI)was
calculated as the solar array articulated to its stowed
placeodnthefacesheeextposetodtheaerodynamic
position. As the spacecraft slewed tothe aerobraking
heatinagndarountdheedgeosftheM55Jgraphite
configuration, the solar array's view factors to space
facesheeTth.ebulkoftheMLIwasonthefacesheet
did not change, so view factors did not have to be
surfacaen,dasmapllortionwrappeadrountdheedges recalculated for that maneuver.
andterminateodnthesolacrellside.Thewidthofthe
MLIonthesolacrellsiderangefdrom50 14gram.
Figures 3and 4 show the spacecraft and solar array in
the aerobrake configuration and vacuum phase
ThesolaarrrayhasfivethermocouplTehs.eriesone
respectively. This part of the analysis required detailed
thermocouopnlethemidpaneolnthesolacrellsideof
knowledge ofthe orbit and spacecraft orientation, and
thearrayT. he+XandXpanelesachhaveone
thus was highly dependent on the flight mechanics
thermocouopnlethesolacrellsideandoneonthe"hot"
analysis.
facesheseidte.Ontheengineeridnrgawin6g,tshe
thermocoupalreesdesignatTe1d,T2,T3,T4,andT5.
T1andT4areonthe+/-Xpaneolntheexposed
facesheseidte.T2,T3,andT5areonthesolacrell
sideF;igure1showtshelocationosfthethermocouples
onthearray.
THERMAL MODELING
The thermal analysis of the solar array had three main
components: the view factor and orbital heating
analysis, the aerodynamic heating analysis, and the VelocityVector
computation of solar array temperatures. The
calculations were performed in two flight regimes. One
regime was the vacuum phase where the spacecraft was Figure 3. Mars Odyssey Thermal Desktop model in
in orbit around Mars, but out of the atmosphere. The the aerobrake configuration.
thermal environment in this phase was dominated by
the solar and planetary heating with negligible
aerodynamic heating. The second regime was the
aerobraking phase, or drag pass, where the spacecraft
made its excursion into the atmosphere. The thermal iiiii#i!iiiiiiiiii
environment inthis phase was dominated by the
aerodynamic heating with aerodynamic heating being
roughly 60 times greater than the orbital heating.
Temperatures were calculated for both flight regimes.
Accurate calculation of the solar/planetary flux during
flight, as well as aerodynamic heating during drag
passes, requires that the thermal analysis be tightly
coupled to the flight mechanics, aerodynamics, and
atmospheric analysis. The orbital heating and view
factor analysis, or radiation model, was developed
Figure 4. Mars Odyssey Thermal Desktop model in
using Thermal Desktop, acommercially available
the vacuum phase configuration, view from Earth.
software package. 7 The radiation model was developed
using engineering drawings of the spacecraft and solar The aerodynamic heating analysis consisted of
array. 6 View factor and heat rate calculations were calculating the atmospheric density, the spacecraft
performed for several different solar array velocity relative to the atmosphere, and the heating
configurations and spacecraft orientations. First, coefficient for points spatially across the array. Using
calculations were made with the spacecraft and solar this information, the incident aerodynamic heating was
array in its vacuum phase configuration; the spacecraft calculated using equation 1,
3
American Institute of Aeronautics and Astronautics
and adhesives that were modeled as an averaged mass
spread evenly across the layer. This was an engineering
1 V3
approximation since the wire, solder etc. were not
Qi = _t o" .CH (1)
evenly distributed across the panels. The titanium
hinges and dampers that physically connect the three
where pis the atmospheric density, V is the relative solar panels were also included in the model and
velocity, and CHis the heat transfer coefficient. This provided a thermal link between the three panels. The
calculation was made at 5 second intervals throughout model was highly detailed and included the variations
the aerobraking pass to give atransient representation inthe aluminum honeycomb core density as well as the
ofaerodynamic heating. The flight mechanics team varying thickness of M55J composite doublers. For all
reconstructed the aerobraking pass and provided the materials, properties were included as functions of
relative velocity. The density and the heating temperature; for the aluminum honeycomb core and
coefficients were calculated using a Direct Simulation M55J facesheets, orthotropic material properties were
Monte Carlo (DSMC) particle method which allowed included. The five spacecraft thermocouples were
the density to be calculated using acceleration readings modeled as bar elements which had mass and were
from the flight accelerometer and the heating thermally connected to the spacecraft with acontact
coefficients to be calculated using the density and the resistance. Overall the PATRAN thermal model was a
relative atmospheric velocity. Due to the long run physically accurate 3D representation ofthe solar array.
times needed to perform the DSMC computations, an Compared to the as-built mass of the solar array of 32.3
aerothermodynamic database was developed prior to kg, the mass of the PATRAN thermal model was 33.4
aerobraking. Then for each time step, the density was kg, a difference of only 3.4%.
calculated by interpolating the accelerometer data, and
the heating coefficients across the array were calculated The PATRAN thermal model required input boundary
by interpolating between density and relative velocity. conditions from the two analysis components
The interpolation error onthe heating coefficients was mentioned earlier. The model included radiation to
calculated to be about 2%, which was within the space, incorporating the view factors that were
accuracy ofthe DSMC calculations. The heat transfer calculated from Thermal Desktop. The orbital and
coefficients ranged from apeak value of 0.90 to a low planetary heat fluxes calculated from Thermal Desktop
value of 0,and included the surface accommodation were applied to the surfaces of the model as well. The
coefficient. The model also accounts for reflected heat, aerodynamic heating and reflected heating boundary
which was approximated empirically for this mission conditions were applied to the exposed M55J facesheet
using equation 2 and is a function of the incident heat and MLI surfaces. Heating on the edges of the panels
flux and surface temperature. was included as 5% of the incident heating of the nodes
closest to the edge. Edge radiation was included around
the outer-most edges with a view factor to space of 1.0.
Qref =ltor3Io.ol5-t-O.lIl- 1 Q_i 3 .ILalll Radiation back tothe spacecraft bus was included, with
[_ _ 7pU \300) (2) the spacecraft bus simply modeled as a node with a
constant temperature. Initially this temperature was
approximated using spacecraft temperature data for the
where Twall isthe surface temperature of the solar array.
trajectory correction maneuvers (TCM's), and then was
Equations 1and 2reveal the coupling of the thermal
updated using data from the first few orbit drag passes.
analysis to the flight mechanics, aerodynamics and
atmospheric analysis. These calculations provided an
Steady state temperatures were calculated for the solar
accurate representation of the aerodynamic heating as
array in the vacuum phase prior to the start of the drag
well as reflected heating that was a function of time as
pass. As the spacecraft began to stow the solar array
well as position onthe solar array.
and slew into the aerobrake configuration, atransient
analysis was made to obtain the initial temperatures at
The temperatures for the solar array were calculated
the start of the drag pass. The temperatures obtained
with the commercially available software
prior to the drag pass were primarily dependent onthe
MSC/PATRAN Thermal. s Like the Thermal Desktop
orbital heat rates and view factors calculated from
model, the PATRAN Thermal model was developed
Thermal Desktop, but most significantly, they were
solely using engineering drawings. 6 The model
influenced by the fact that the spacecraft passed into
represented allthree solar panels, and included both
solar occultation and was shaded by Mars onaverage,
facesheet layers, the aluminum honeycomb core, the
for about 4minutes before the drag pass began. This
Kapton, the MLI, and one layer for the solar cells. The
reduced the initial temperatures and, allowed the solar
solar cell layer included the cover glass, wire, solder,
array to absorb more energy and therefore increased the
4
American Institute of Aeronautics and Astronautics
margibnetweethneflighttemperatuarensdthethermal
thermal analysis. Despite the tight operations schedule,
limits.Finallyd,uringthedragpassa,transient
the analysis generated alarge quantity of information
analyswisasperformetoddetermitnheetemperatures,
about the thermal state of the solar array. Figure 5
wherteheaerodynamheicatinwgasdominant.
shows the predicted temperature distribution for orbit
Temperatpurreedictiownseremadperiortoeachdrag
pass 40,just after periapsis, which istypical of the
pasasndtemperatuwreesrecalculateudsingactual
majority of the drag passes encountered. This
spacecrtaefltemettroyreconstrupcatsotrbits.
temperature distribution was calculated using a
predicted density, and velocity profile, and assumed a
Amorecompledteiscussioonnthethermaalnalysis
nominal orientation relative to the atmosphere. Figure
andthemethodolougsyedontheMarsOdyssesyolar
6shows the predicted temperature of the thermocouples
arraycanbefoundinarelatepdublication.
9 and the predicted heat rate as a function of time
centered onperiapsis. A similar transient plot was
generated for each material, where the maximum
DRAG PASS AND TEMPERATURE predicted temperature for any location on that material
istracked. Figure 7shows the maximum material
PREDICTIONS
temperatures as a function of time.
The initial goal of the thermal analysis was to provide
an independent validation and verification of the
analysis already being performed onthe solar array.
The high fidelity nature of this 3-dimensional analysis,
and the speed at which analysis results could be
produced, were compelling evidence to include this
analysis in the trajectory decision making process
during aerobraking operations.
The timeline for spacecraft maneuver decisions dictated
the turnaround time for analysis. Each day the drag
pass temperature predictions aswell as any
reconstructed temperatures had to be completed by 12
noon eastern standard time. For the analysis to be
useful during operations and be included in the
Aerobraking Planning Group's (APG) maneuver Figure 5. Predicted temperature distribution on the
decision-making process, ithad to be generated solar array Nr orbit pass 40.
efficiently. Utilizing a 1.7Ghz dual XEON processor
computer made itpossible forthis highly detailed finite
element analysis torun in about one hour. As the
mission progressed, the duration of each drag pass
increased. This caused the analysis run times to
increase slightly but never caused them to exceed 1
hour 15minutes. Computer speed alone was not the
only means of increasing efficiency. The spacecraft's
configuration in the vacuum phase was always the same
so vacuum phase temperatures remained virtually
constant. Also, the occultation duration changed very
slowly, causing the initial temperatures to change ata
slow, predictable rate. Therefore, to increase analysis
efficiency, the view factor, orbital heating, and vacuum
phase temperature analyses were performed onan as-
needed basis.
Figure 6. Thermocouple temperature predictions,
The thermal analysis was highly dependent on all of the orbit pass 040, peak heat flux = 0.232 W/cm 2.
other analyses being performed. The thermal analysis This plot differs slightly from the plot of the
took the longest to run and was always the last tobe thermocouple temperatures. The obvious difference is
completed. Any problems arising within another thatthe maximum material temperatures are
group's analysis caused delays in the completion of the significantly higher than the predicted thermocouple
5
American Institute of Aeronautics and Astronautics
temperaturIensth.ecaseoftheM55Jgraphitwe,hich
Although the uncertainties inthe density predictions
wasexposetodtheflow,thefiguresshowthatthepeak
were present, they did not impact the prediction ofthe
temperatuwraes85°Chighetrhanthepeakoneitheorf
thermal limit lines, which turned out to be avery useful
thefaceshetheetrmocoupTle1sa,ndT4.
tool. The flight corridor for main phase aerobraking
was chosen based onthe maximum Q dot, which was
the value of the aerodynamic heating that would cause
the solar array to exceed its flight allowable
temperature limit for the structure of 175°C. The upper
flight corridor was reduced by a factor of 2to carry
100% margin with respect to the limit. Figure 8is a
plot of the maximum solar array temperature as a
/
function of Q dot, covering orbit passes 77 through 99.
This plot was updated on aweekly basis by running the
PATRAN thermal model with varying density and
hence heating profiles. Four different density profiles
were used: the low, middle, and upper ends of the flight
j,*' %
corridor, as well as one that which would produce a
-250 -200 -150 -100 -50 0 5O 100 150 200 25O maximum Q dot of 0.8 W/cm 2. A Q dot of 0.8 W/cm 2
Time (s)
was chosen as the upper bounding case because it
Figure 7. Maximum predicted material guaranteed the solar array prediction would exceed the
temperatures, orbit pass 040, peak heat flux = 0.232 flight allowable temperature of 175°C. By using a
W/cm 2. predicted Q dot, or one derived from flight data, the
JPL Navigation team could quickly determine the
This was aresult of having MLI covering the facesheet
maximum predicted temperature ofthe solar array.
inthe areas near those thermocouples. The MLI
provided sufficient shielding to the underlying material 200
from the incident heat flux to prevent the material in 190
1_0
those regions from reaching higher temperatures. 170
160
Throughout the main phase of aerobraking, the 160
magnitude of the temperature difference between the 140
130
material maximum and the thermocouple maximum G 120
0_110
was dependent on the heat flux, and the difference grew 100
as the magnitude ofthe heat flux grew. The material _o
maximum was always higher than what the 7O
60
thermocouples indicated and was always located inthe 60
4O
lower, outboard quadrant of the +X panel. Another SO
minor difference between the thermocouple and 2O
10
maximum material plots is due to the fact that the
location of the maximum material temperature changed
throughout the pass. In Figure 7where there seem to be Figure 8. Temperature as afunction of Q dot and
discontinuities and the temperature seems tojump, the orbit pass.
the maximum temperature shifts to another node in that
particular material layer. Determining the thermal limit of the solar array in
terms ofthe Q dot was obtained simply by finding the
In Figures 6and 7,the heat rate isrepresented by a value of Q dot that corresponded to 175°C. As the
smooth gaussian like profile. The heating is directly mission progressed the correlation of the thermal model
proportional tothe atmospheric density; the density to flight data improved significantly. Since the thermal
predictions, which came from an Odyssey version of model was updated continuously based onthe
Mars GRAM 2000, donot account for atmospheric correlations, confidence in the limit line chart increased
density variations and thus have an average density dramatically. In figure 8there are two limit lines: one
profile as a function of altitude. Large uncertainties in is alimit line based onthe NASA Langley 3-
predicting the atmospheric density from orbit to orbit dimensional finite element thermal model, and the other
made the temperature predictions less valuable than isthe limit line calculated by Lockheed Martin's 1-
expected, but they were still useful in that they gave a dimensional model of the solar array. The Lockheed
3-dimensional picture of the thermal state ofthe array, limit line was more conservative and as such was used
and could identify any thermal anomalies. as the flight maximum.
6
American Institute of Aeronautics and Astronautics
TEMPERATURE CORRELATIONS TO model's peak temperature overshoots the flight data,
FLIGHT DATA providing a more conservative estimate. Third, the
facesheet thermocouples inthe model cooled down at a
slower rate than the flight data. Finally, there was an
For each orbit pass, the solar array temperatures were
initial temperature offset between the flight data and the
reconstructed using telemetry from the spacecraft.
thermal model.
Comparing the temperatures calculated using the flight
data with the spacecraft thermocouple data provided a
means with which the thermal model's accuracy could -_-T1 facesheet X
T2cellX
be assessed. Moreover, since the thermal analysis was -+-T3 cellCenter
_e--T4facesheet +>
highly dependent onthe other analysis being r T5cell+X
performed, the thermal model correlations could also be
used to give an overall assessment of the aerobraking
analysis process and assess the accuracy of the other
analysis models.
The required inputs for the thermal reconstructions
came largely from accelerometer data transmitted from
the spacecraft. Once the data was in hand, the NASA
-210 -160 -110 -60 -10 4O 9O 140 190 24O
Langley accelerometer team processed the data and Time (s)
obtained the atmospheric density and flight velocity. Figure 9. Reconstructed temperatures, orbit pass
The aerodynamic heating was then calculated the same 106, using the initial uncorrelated thermal model.
way as inthe predictions, with the only difference being
that the density and velocity profiles used were "real" The timing difference between the flight data and the
instead ofpredicted. Vectors to the Sun and to Mars model was affected by the way the thermocouples were
could also be obtained from this data and were used in
represented in the thermal model. The thermal model
the orbital heating component of the analysis. The time was created using only drawings ofthe solar array,
ofperiapsis could also be determined from the data and without detailed manufacturing knowledge. The exact
was used as the reference time for the plots. mass of each thermocouple, as well as the amount of
adhesive used tobond them to the solar array, was
For illustration purposes in this paper, orbit 106 is unknown. These parameters had tobe approximated.
shown at various stages in the correlation process. In the thermal model these are simulated by a contact
Orbit 106was chosen because it was the orbit which
resistance to represent the bonded connection between
had the highest atmospheric density and hence the the thermocouples and the array; and the density and
highest heat rate. The Q dot for orbit 106 peaked at cross-sectional area ofthe thermocouple bar element to
0.544 W/cm 2and the solar array reached a maximum
simulate the mass. The mass of the thermocouple
temperature of 136°C. Orbit 106 was reconstructed affects the speed with which it reacts totemperature
using the thermal model that was available at the changes and to some extent the peak temperature that it
beginning of aerobraking, the model that was current at attains. The contact resistance affects the temperature
the time orbit 106 was made, and with the final difference between the bar element and material layer it
correlated model. At the beginning of aerobraking is connected to, and thus affects the peak temperatures
operations, the only data available was from the reached.
TCM's. This data was used as a starting point in
correlating the model, but since the thermal Looking back at Figure 5,the highest temperatures in
environments were drastically different, it was not the panel occur onthe facesheet side, but directly
sufficient to allow full correlation of the model. Thus, behind these high temperature areas are the solar cell
the correlations for the first few aerobraking passes did side thermocouples. In the model, too much heat from
not match the flight data very well. Figure 9shows a the facesheet side was being transferred to the solar cell
transient plot of the flight thermocouple temperatures side and down the "wings" to the areas near the
and the reconstructed thermocouple temperatures for facesheet thermocouples. Differences in the peak
orbit pass 106. temperature magnitudes were primarily affected by two
factors. The first was not including the reflected heat
Looking at the plot, several statements can be made calculation initially, and second was the through panel
about the initial thermal model. First is that the time to
conductance. Early on when the heat fluxes were still
reach the peak temperature in the thermal model lagged low, it was believed that the contribution of the
the flight thermocouple data. Second, the thermal reflected heat was insignificant, but this did not prove
7
American Institute of Aeronautics and Astronautics
tobethecaseI.ncludintghereflectehdearteducethde
flight article was impeding the heat flow to those
facesheseidtetemperatuarnedreducethdeamounotf
thermocouples. There were three possible explanations.
heattransferrteodthesolacrellsideanddowntothe
One was that since in the wing region the facesheet and
wingswherTe1andT4arelocatedA.lsos,light
aluminum honeycomb core are very narrow, there may
adjustmetnottshethrougphaneclonductanwceere
have been amanufacturing flaw, or discontinuity in the
madteohelpalleviatseomeofthistransferrheedaats
material disrupting the heat flow, whereas in the
well.Themidpanethl ermocoupTl3e,,wasnot
thermal model each material layer was continuous.
exposetodtheaerodynamheicatinagndthusits
Second, during thermocouple installation, the
temperatudridenotchangmeuchA. sidefromthe
thermocouples may not have been placed in exactly the
initiatlemperatudriefferencitewasingoodagreement
same location as indicated in the drawings. In the
withtheflightdataT. hefactthatT3matchewdell
thermal model, the thermocouples were placed as
indicatethdatthethrougthhethicknescsonductance
indicated onthe drawings, this would definitely cause
wasconsistewnitththeflightarticles,oadjustmetnots
some disparity. Finally, in refining the model details, it
theconductanocftehepanelwsasminimalI.twasnot
was discovered that there was up to 6.35mm of
altogethienrtuitiveb,utthecoodl ownperformanocfe
syntactic foam packed around the edges of the
thefaceshetheetrmocoupwleassaffectebdythe
aluminum honeycomb core in some regions. Syntactic
assumveadlueoftheeffectiveemissivi(ty_*b)etween
foam has avery low conductivity and would impede
theMLIandtheM55JfacesheWet.hilethea*hadits
heat flow in the regions where it was present. Figure 11
greateesftfecotnthefaceshetheetrmocoupTle1sa,nd
shows the wing region of the +X solar panel without
T4,italsohadaslighitmpacotnthecellside
the Kapton layer. To simulate the effects of the
thermocoupTle2asndT5.Initiallyi,ntheabsencoef
syntactic foam, the thermal conductivity of the nodes in
anytesdt atath, eeffectiveemissiviwtyas the affected areas in the M55J facesheet and aluminum
conservativcehlyoseans0.1.Afterexamininthge
honeycomb core material were reduced and the density
flightdataandrunninsgevercaalsevsaryintghe increased. These modifications were based onthe
effectiveemissivitayn,a*of0.05wasfoundtobest
simple rule of mixtures. Modifying the thermal
matcthhecoodl own.
conductivity and density in this way was done to
expedite the inclusion ofthe syntactic foam and avoid
Usingdatafromorbits005throug0h20a,suitable
addition of new solid geometry and additional finite
correlatiofonrthethermocoumplaesasndcontact
element meshing in the model.
resistanwceasfoundA. lsot,hecalculatiofonrthe
reflectehdeawt asincludeadndthevaluefor_*was
Atthis point the initial temperature error was still
updatedF.igure10showosrbipt ass106withthese
present. For the initial temperature error to be
correlatioanpsplied.
corrected, a temperature correlation in the vacuum
phase up tothe start of the drag pass was required. To
accomplish this, a correlation coupling the Thermal
G
60
._0i..........i.............:.............. .....
-210 -160 dl0 _0 -10 40 90 140 190 240
Time (s)
Figure 10. Reconstructed temperatures, orbit pass
106, using the partially correlated thermal model.
At this point in the correlation, the timing difference
was minimized, the peak temperature overshoot was Figure 11. Close up view of+X panel wing region.
reduced, and the cool down performance had shown
Desktop model and the PATRAN thermal model had to
improvement. The cool down performance and the
be developed. There were three main adjustments
overshoot for T1 and T4 did improve but they were still
made to correlate the initial temperatures. First was the
at unacceptable levels. It was as if something inthe temperature assumed for the spacecraft in the PATRAN
8
American Institute of Aeronautics and Astronautics
modelM. akinagdjustmetnottshespacecraft
Locally, there is less mass, and most significantly,
temperatuarlseohadasignificainmtpacotnthecool
lower solar absorptivity. Lower solar absorptivity in
downrateforeachofthethermocouplTehse.second
these local regions would reduce the amount of solar
andthirdadjustmewntesreintheThermaDlesktop
and planetary heat flux and thus reduce the temperature
modelI.ntheThermDalesktompodethl esolaarrrayis
in these regions as observed in the flight data. The
represenbteyd2-dimensiosnuarlfacewsithtwoactive
difference in local absorptivity was not modeled, but
sidesI.nitiallyitwasfeltthatanysolaorrplanetary
for future missions certainly should be.
fluxthatwasincidenintthethirddimensioonr,edge
on,wouldbeinsignificanTth.isturnedouttobeabad
As mentioned previously, Odyssey encountered the
assumptiaosntherewasasignificaanmt ounotffluxon
highest density of the mission on orbit pass 106, and
theinsideedgeofthewingregionnseaTr 1andT4.
hence this was the orbit pass with the largest
Initiallyt,osimplifytheorbitahleaftluxcalculations
aerodynamic heat flux. The higher heat flux
andreducreuntimes,pecularwityasnotincluded.
intensified the effect of not including the local
Howeveinr,ordetroincreastheefidelityofthe
differences in solar heat flux. Looking at only orbit 106
calculatiointsw,asincludedIn.cludintghespecularity
can be deceiving because the differences between the
provetdobethemissineglemeninttheThermal
flight data and the thermal model are maximized in
Desktompodealndhadthegreateismtpacotnthe
general, the differences throughout the mission were
initiatlemperatucroerrelatioTn.hesaedjustmeansts
not this great. Table 1summarizes the average
wellastheadjustmemntasdeincorrelatintogthedrag
differences inpeak temperatures for each thermocouple
pasdsatawereincludeidnthemodel.
at three different points during the main phase of
aerobraking and gives the overall mission average.
Figure12showosrbipt ass106withallofthe
Taking a closer look at the temperatures over the entire
correlatioandjustmeanptspliedT.hemidpanel
drag pass, a root mean squared (RMS) average for each
thermocoupTl3e,,matchethseflightdataexactlyT.he
thermocouple over the duration of the drag pass was
differencbeetweethnefaceshetheetrmocoup(Tle1s&
computed.
T4)andtheflightdatawasreducetodlessthan10°C
fortheduratioonfthedragpassT. heriesstillsome Table 1. Summary of peak temperature difference
overshoeortrobretweethnecellsidethermocoupTle2s, between the flight data and thermal model
andT5.T3matchinthgeflightdataaswellasitdoeiss
Average Peak Temperature Difference (°C)
anindicatiotnhatoveraltlh,ethrougphanel
Orbits Orbits Orbits Mission
conductanwcaescorrelatveedrywell.Howevethr,e Sensor
locaclonductannceeatrhecellsidethermocoupTle2s 005-019 020-095 096-225 Average
T1 4.1 3.1 2.1 2.90
andT5mayhavebeenslightlydifferenTt.hesolacrell
sidewasmodeleadsasinglec,ontinuoluasyeorf T2 3.2 1.7 4.6 2.97
material. T3 0.1 0.2 0.2 0.16
T4 3.8 3.6 3.3 3.47
60.......................................m............ ............................................................................ T5 6.5 6.9 9.2 7.53
--T1 Flight _-T1 facesheet ×
Table 2 shows the RMS average for each thermocouple
at three points during the main phase of aerobraking.
Table 2. Summary of RMS temperature differences
•..........i...........b.e.tween the flight data and thermal model
RMS Temperature Difference (°C)
Orbits Orbits Orbits Mission
Sensor
005 - 019 020 - 095 096 -225 Average
-8° I ...............i....................,....... , ..... T1 7.2 6.2 4.3 5.90
-210 -180 410 _0 -10 40 _0 140 1_0 240 T2 4.6 4.1 5.4 4.70
Time (s)
T3 2.0 3.1 1.9 2.33
Figure 12. Recontructed temperatures, orbit pass T4 6.9 6.5 5.6 6.33
106, using the best correlated thermal model.
T5 3.9 3.7 5.7 4.43
In reality, the solar array has wires, solder, and some
The RMS average provides a somewhat better means
spots where there are no solar cells. The thermocouples
with which to assess how close the model was coming
T2 and T5 are in regions that do not contain any solar
tothe flight data for the duration of each drag pass and
cells, so they are mounted directly to the Kapton layer.
9
American Institute of Aeronautics and Astronautics